article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
observations of large scale structure ( lss ) @xcite and cosmic microwave background radiation ( cmbr ) @xcite indicate that the universe is highly homogenous and isotropic on large scales . under standard assumptions on the matter
content , in the standard general relativistic cosmological models isotropy is a special feature , requiring a high degree of fine tuning ( a special set of initial conditions ) in order to reproduce the observed universe .
specifically , in the set of spatially homogeneous cosmological models with a fluid which satisfy the energy conditions @xmath2 and @xmath3 , those which approach isotropy at late times is of measure zero @xcite . in these models
the anisotropy dominates the dynamics in the neighborhood of the singularity , while the effect of matter is negligible : thus the singularity can be said to be velocity dominated @xcite . in general anisotropic models do not isotropize sufficiently as they evolve into the future ( models do not evolve towards a friedmann - robertson - walker universe rapidly enough ) .
this problem is known as the isotropy problem and can be solved by inflation .
many investigations on this issue show that a large class of models evolve towards a homogenous and isotropic model when under the influence of inflationary matter .
this means that even models with initially high levels of anisotropy can evolve into the universe we observe today .
a large body of work exists relating specifically to the study of spatially homogenous and anisotropic bianchi cosmological models . in a classical paper ,
wald showed that initially expanding bianchi models with a cosmological constant asymptotically approach a de sitter state ( hence a spatially homogenous and isotropic model ) , except for the bianchi type ix mixmaster models which may re - collapse @xcite . in the case of inhomogeneous models
it is not so clear cut and in order for inflation to begin , sufficiently homogenous initial conditions are required ( see e.g. @xcite ) . as far as singularities in general relativity
are concerned , it is widely accepted @xcite that the belinskii , khalatnikov and lifshitz ( bkl ) @xcite picture is correct : for fluids with linear eos @xmath4 ( @xmath5 ) or scalar fields ( @xmath6 ) and are special in this respect , see @xcite and references therein .
] , when the singularity of generic inhomogeneous anisotropic cosmological models is approached , the dynamics is asymptotically that of the mixmaster model , the singularity is thus velocity dominated and the effect of matter is negligible @xcite .
more recently various theories of gravity that go beyond general relativity have been considered .
for instance , high energy modifications of gravity could come from extra dimensions ( as required in string theory ) . in the brane world
@xcite scenario the extra dimensions produce a term quadratic in the energy density in the 4-dimensional effective energy - momentum tensor , i.e. this quadratic term appears in the 4-d evolution and constraint equations on the brane . under the reasonable assumption of neglecting 5-d weyl tensor contributions in these equations , this quadratic term has the very interesting effect of suppressing anisotropy at early enough times , as the singularity is approached . in the case of a bianchi
i brane - world cosmology containing a scalar field with a large kinetic term the initial expansion is quasi - isotropic @xcite . in this scenario ,
the initial anisotropy provides increased damping in the scalar field equation of motion , thus resulting in an extended period of inflation .
the rapid decay of the anisotropy results in the class of initial conditions allowed by observations to be greatly increased . under the same assumptions ,
bianchi i and bianchi v brane - world cosmological models containing standard cosmological fluids with linear equation of state ( eos ) ( @xmath7 ) also behave in a similar manner @xcite . the same is also true for more general homogeneous models @xcite and even some inhomogeneous exact solutions @xcite .
finally , within the limitations of a perturbative treatment , the quadratic - term - dominated isotropic brane - world models have been shown to be local past attractors in the larger state space of inhomogeneous and anisotropic models @xcite .
more precisely , again assuming that the 5-d weyl tensor contribution to the brane can be neglected , large - scale perturbations of the isotropic models vanish in the past , approaching the singularity .
large scale perturbations @xmath8 are the only type relevant to this kind of analysis for non - inflationary matter , because for any given @xmath9 there is always an early enough time such that @xmath8 is satisfied a given scale much larger than the hubble horizon @xmath10 at a given time will enter the horizon " ( @xmath11 ) at a later time . in practice , neglecting small scales perturbations is equivalent to neglect gradients and thus the laplace operator term in the evolution equations , which then become ordinary differential equations . in the perturbative context , this is analogous to the bkl approximation @xcite . ] .
thus in the brane scenario the observed high isotropy of the universe is built in , i.e. the natural outcome of _ generic initial conditions _
, unlike in general relativity where general cosmological models with a linear eos are highly anisotropic in the past , approaching the singularity _ a - la _ bkl , i.e. with mixmaster dynamics @xcite . it would therefore be interesting to see if the behavior of the anisotropy at the singularity found in the brane scenario is recreated in the general relativistic context if we consider a quadratic term in the eos .
that is , do we have a scenario where , in general relativity , the generic asymptotic attractor in the past is an isotropic model , thanks to a non - linear eos .
this is the question we aim to investigate , for the specific case of the eos @xmath12 .
this is a simplified version of that used in paper i@xcite , where the focus was on using a quadratic eos to represent a dark energy component or unified dark matter .
it is general enough to analyze the high energy dynamics we are interested here , while still allowing for an effective cosmological constant and phantom behavior .
the paper is organized as follows . in section
ii we outline the setup of the models under investigation , defining three classes , one of which represents a fluid with phantom behavior . using dynamical system methods @xcite , in section iii we consider the anisotropic bianchi i and v class of homogenous models containing a perfect fluid with quadratic eos , showing how models with standard non - phantom matter are asymptotic in the past to an isotropic fixed point is . using covariant and gauge invariant variables @xcite , in section iv
we give the linearized evolution and constraint equations for generic inhomogeneous and anisotropic perturbations about a flat friedmann model with a quadratic eos .
we solve the equations in the large scale limit @xmath8 and show that the perturbations vanish as the singularity is approached , indicating that the isotropic model is is the generic past attractor .
we then finish with some concluding remarks and discussion in section v.
we begin with a summary of the most relevant points that can be derived mostly from an analysis of energy conservation alone . more details can be found in section ii in paper i@xcite .
the energy conservation equation for a cosmological model containing a perfect fluid is : @xmath13 where @xmath14 is the energy density ,
@xmath15 the isotropic pressure and @xmath16 is the hubble expansion scalar .
an average scale factor @xmath17 is defined , at least locally , by @xmath18 . through this relation solutions of eq .
( [ conserv ] ) can be found in terms of @xmath17 , but first an eos must be specified .
usually this is taken to be a linear function of the energy density of the form @xmath0 , with @xmath19 . in this case
( [ conserv ] ) gives : @xmath20 in the case of the homogenous and anisotropic bianchi i cosmological models , the amplitude of the shear scales as : @xmath21 this example shows how the shear can decay rapidly in an expanding model , growing larger for smaller @xmath17 . if the dynamics of the model is such that @xmath22 , both @xmath14 and @xmath23 blow up , and the model has a physical singularity . in the case
@xmath24 , the so called phantom " fluid @xcite , the energy density counter intuitively decays ( grows ) in the past ( future ) and therefore any shear component always comes to dominate at the initial big bang " singularity . in the case when @xmath25 the energy density decays during expansion , but at a slower rate than the shear , so that once again the shear component dominates at the singularity . in approaching these singularities matter with linear eos @xmath5
is un - influential , therefore these singularities may be termed velocity dominated @xcite .
it is only in the case when @xmath26 ( super - stiff " fluid ) that the energy density is dominant in the past and therefore the initial singularity is isotropic .
this indicates that as the initial singularity is approached in the standard homogenous model with @xmath5 , any residual anisotropy will generally come to dominate near the singularity .
the anisotropy is a generic feature of these models and the initial singularity is in general anisotropic and velocity dominated .
as discussed earlier , motivated by the results in the brane - world models , we aim to investigate the effects of introducing a non - linear eos on the behavior of anisotropic models .
the eos we propose is quadratic in the energy density ( we neglect the constant pressure term introduced in paper i@xcite as in general this only modifies late time behavior ) : @xmath27 the discrete parameter @xmath28 denotes the sign of the quadratic term , @xmath29 , and @xmath30 sets the energy scale above which this term becomes relevant . the energy conservation equation ( [ conserv ] ) can now be integrated to give the scaling of the energy density as a function of the scale factor ( except for @xmath31 ) .
@xmath32 where @xmath33 , @xmath34 represent the energy density and scale factor at an arbitrary time @xmath35 .
this is valid for all values of @xmath28 , @xmath36 and @xmath37 ( @xmath38 ) .
defining @xmath39 the fluid admits an effective positive cosmological constant point , @xmath40 , only if @xmath41 .
as shown in paper i@xcite , in the case of @xmath42 the fluid generally behaves either in a phantom manner or is asymptotic to a finite energy density in the past , the effective cosmological constant @xmath43 , for all open and flat models . in this scenario
, the initial singularity is going to be dominated by anisotropic expansion as seen in the standard linear eos case , due to the sub - dominance of the energy density at early times ; therefore we do not consider the @xmath42 case further . in the case of @xmath44 , the expression ( [ rhoa ] ) for the energy density @xmath45
is conveniently rewritten in three different ways , defining @xmath46 and assuming @xmath47 .
+ * a : * @xmath48 , @xmath49 , @xmath50 in this case @xmath51 , with @xmath52 .
the past singularity is similar to certain future singularities found in dark energy models @xcite and is referred to as type iii singularities : * for @xmath53 , @xmath54 , @xmath55 and @xmath56 where @xmath57 and @xmath58 are constants with @xmath59 ( we reach @xmath60 at a finite @xmath61 ) .
the main difference is that in our case the singularity occurs in the past . at late times
the fluid behaves in a similar manner to a fluid with a linear eos and @xmath62 . +
* b : * @xmath63 , @xmath64 , @xmath65 in this case @xmath51 , with @xmath66 . as in case * a * , the fluid approaches a type iii singularity in the past , but now approaches an effective cosmological constant at late times .
+ * c : * @xmath63 , @xmath67 , @xmath68 in this case @xmath69 , with @xmath70 .
the fluid behaves in a phantom manner but tends to an effective cosmological constant at late times .
+ in the cases * a * and * b * , as we approach the past singularity the energy density approaches infinity , in fact the singularity occurs at a finite scale factor , that is for the energy density one has @xmath55 as @xmath54
. the cosmological model emerges at a finite scale factor @xmath58 , with infinite energy density and infinite isotropic pressure @xmath15 .
the model emerges from a curvature singularity , infinitely decelerates and asymptotically approaches a standard flat friedmann model with a linear eos ( case * a * ) or a de sitter model ( case * b * ) . in the case * c * , the fluid behaves in a phantom manner and approaches @xmath71 at late times . in this case
the energy density is sub - dominant at early times and the anisotropy will dominate as we will show in the proceeding sections . in cases * a * and * b * , we have the possibility that the energy density could be the dominant component in the past , forcing the singularity to be isotropic .
this would result in isotropic expansion at early times as seen in the brane - world model , with the fluid also approaching a non super - stiff eos at late times .
we will now proceed with the analysis of anisotropic cosmological models containing a fluid with a quadratic eos .
we now study the dynamics of homogeneous and anisotropic bianchi cosmological models .
in particular we investigate the bianchi i and v classes of models that contain the flat and open friedmann models .
the matter content will be modeled by a non - tilted ( that is the fluid flow remains orthogonal to the group orbits ) perfect fluid with a quadratic eos .
we make use of the orthonormal frame formalism which results in a system of first - order evolution equations . we have adopted a notation consistent with that used in @xcite ( greek indices run from @xmath72 and latin indices run from @xmath73 ) .
the orthonormal frame formalism is a 1 + 3 decomposition of the einstein field equations into evolution and constraint equations .
the decomposition is carried out relative to the time - like vector field @xmath74 of a given orthonormal frame @xmath75 . in the case of models which admit an isometry group ,
@xmath76 is chosen to be the normal ( @xmath77 ) to the space - like group orbits .
the bianchi models are such cosmological models , they admit a three dimensional group of isometries which act simply and transitively on the space - like hyper - surfaces ( surfaces of homogeneity ) . the orthonormal frame @xmath78 ,
is such that : @xmath79 the dynamics can then be described in terms of the spatial commutation functions , the kinematical quantities and the matter variables .
the spatial commutation functions , @xmath80 , are defined by the commutation relations between the spatial basis vectors such that : @xmath81=\gamma^{\mu } { } _ { { \alpha}{\beta}}{\textbf{e}}_{\mu}.\ ] ] the spatial commutation functions can be decomposed into a @xmath82-index symmetric object and a @xmath83-index object which determine the connections on the space - like 3-surfaces .
@xmath84 the kinematical quantities include the hubble scalar @xmath16 , the shear tensor @xmath85 and the angular velocity @xmath86 , relative to a fermi - propagated spatial frame , of the spatial frame @xmath87 along the time - like vector @xmath88 .
the relevant matter variables are the energy density @xmath14 and the isotropic pressure @xmath15 , while the energy flux and the anisotropic pressure are set to zero as we are considering a perfect fluid .
the decomposed einstein field equations for a general bianchi model containing a perfect fluid then take the form : @xmath89 the trace - free spatial ricci tensor ( @xmath90 ) and the spatial curvature scalar ( @xmath91 ) can be given in terms of the spatial commutation functions : @xmath92 the quantity @xmath93 , can be expressed in terms of the spatial commutation functions and has the form : @xmath94 the jacobi identity for vector fields when applied to the spatial frame vectors @xmath95 , in conjunction with the einstein field equations give a set of evolution and constraints equations : @xmath96 the evolution equation for the energy density results from the contracted bianchi identities , or more directly from the conservation equations , and has the form ( [ conserv ] ) .
the latter with the above system of equations reduce to the evolution and constraint equations for given non - tilted perfect fluid bianchi model by making further restrictions on the spatial commutation functions ( @xmath97,@xmath98 ) and kinematical quantities ( @xmath99,@xmath86 ) .
the bianchi i and v models are considered separately in the following sub - sections .
the bianchi i models are a subclass of the bianchi class a models ( @xmath100 ) .
these models are homogeneous and anisotropic cosmological models containing the flat friedmann ( f ) model . in the case of bianchi i
, the group invariant orthonormal frame @xmath101 can be further specialized such that : @xmath102 this in turn means that the three curvature of the spatial hyper - surfaces ( [ curv1 ] , [ curv2 ] ) is zero : @xmath103 the einstein field equations ( [ efe2 ] , [ efe4 ] ) and the jacobi identity ( [ evo1 ] ) imply : @xmath104 the analysis of the system of equations can be simplified by introducing the following set of normalized dimensionless variables : @xmath105 @xmath106 where @xmath107 .
in addition , it is customary to introduce @xmath108 as a dynamically dimensionless measure of anisotropy : we can refer to a cosmological model as isotropic in some limit if , in that limit , @xmath109 . in particular , if there is a singularity for some @xmath110 ( possibly zero ) , we refer to the singularity as isotropic if @xmath109 in the limit @xmath111 .
the above normalization to @xmath112 is useful because the energy scale @xmath113 itself disappears from the dynamical system : in terms of the dimensionless variables above , the complete system of equations for bianchi i models is explicitly made scale invariant , and is : @xmath114 the primes denote differentiation with respect to @xmath115 , the normalized time variable .
the variable @xmath116 is the normalized energy density , @xmath117 is the normalized hubble scalar and @xmath118 is the normalized shear .
the effective cosmological constant point is now given by @xmath119 .
we only consider the region of the state space for which the energy density remains positive ( @xmath120 ) and models which are expanding ( @xmath121 ) .
the evolution equation for the shear can be integrated to give : @xmath122 the shear tends to grow in the past and rapidly decays at late times .
using eqs .
( [ rho1 ] ) , ( [ rho2 ] ) , ( [ rho3 ] ) and ( [ shearevo ] ) we can see how the shear scales with respect to the energy density ( by substituting for the scale factor in each case ) for each of the three sub - cases of the fluid .
+ case * a : * @xmath48 , @xmath49 , @xmath123 in this case , at early times the energy density diverges , @xmath124 and the shear goes to a constant value , @xmath125 . the initial singularity is dominated by the energy density and is forced to be isotropic , as @xmath126 . at late times ,
@xmath127 and @xmath128 .
* b : * @xmath63 , @xmath64 , @xmath129 in this case , at early times the energy density diverges , @xmath124 ; the shear goes to a constant , @xmath125 .
the initial singularity is again dominated by the energy density and is forced to be isotropic , with @xmath109 . at late times
the energy density approaches a constant value , @xmath130 and @xmath128 .
+ case * c : * @xmath63 , @xmath67 , @xmath131 in the final case , the fluid behaves in a phantom manner . at early times @xmath132 and @xmath133 .
the initial singularity is now dominated by the shear and is anisotropic . at late times
the energy density approaches a constant value , @xmath130 and @xmath128 .
+ in cases * a * and * b * we expect the generic past attractor to be an isotropic singularity , while at late times we expect either a minkowski or de sitter model to be the attractor . in case * c * , the fluid behaves in a phantom manner and we expect models to be asymptotic to a anisotropic singularity in the past and a de sitter model in the future .
we now carry out a dynamical systems analysis of the above system .
the system can be reduced from a 3d system to a 2d system by substituting for a variable using the generalized friedmann equation ( [ genfried ] ) .
we choose to remove the normalized hubble scalar @xmath117 , assuming expansion , @xmath134 .
the resulting reduced planar dynamical system is then : @xmath135 the above system of equations is of the form @xmath136 . since this system is autonomous , trajectories in state space connect the fixed / equilibrium points of the system ( @xmath137 ) , which satisfy the system of equations @xmath138 .
the stability nature of the fixed points can be found by carrying out a linear stability analysis , see section ii.b in paper i@xcite for a summary , and e.g. @xcite for more details .
the fixed points of the reduced planar system ( [ xprimei])-([zprimei ] ) , the corresponding eigenvalues and there existence conditions ( the energy density remains positive ( @xmath120 ) and all three variables are real ( @xmath139 ) ) are given in table [ tab1 ] .
0.47@ccccc + name & @xmath116 , @xmath118 & eigenvalues & existence + + m & ( @xmath140 , @xmath140 ) & @xmath140 , @xmath140 & @xmath141 + + ds & ( @xmath142 , @xmath140 ) & @xmath143 , @xmath144 & @xmath145 + + the first point represents the vacuum static minkowski model .
eigenvalues of the linearized system are zero , hence the linearization theorem @xcite does nt apply at this point and we resort to numerical analysis to determine the stability character . in the case
@xmath146 , this is the only finite valued fixed point present .
this is clearly seen in fig .
[ fig1](a ) , corresponding to case * a * , where the black lines represent separatrices , the dots represent fixed points and the thin grey lines are example trajectories .
the minkowski point m is clearly the future attractor for all trajectories in the region .
the behavior in the past is not clear from this diagram ( but trajectories do appear to evolve to a constant @xmath118 ) .
the horizontal black line represents the bianchi i kasner vacuum models , with line element : @xmath147 the parameters @xmath148 satisfy the following constraints : @xmath149 and the average scale factor for expanding models of this type is given by : @xmath150 although these models are asymptotic to minkowski at late times , as seen in fig .
[ fig1](a ) , they are always anisotropic , in the sense that @xmath151 at all times . the kasner singularity is a paradigmatic , characteristic of many models ; the more general bianchi ix singularity is approximated by a series of jumps from a kasner phase to another , see e.g. @xcite .
the vertical black line represents the expanding flat friedmann model with a quadratic eos . in the case of a spatially flat model as in this case
, the friedmann equation can be integrated to give the dependence of the scale factor on time : @xmath152^{3/2 } } \times \left\{\sqrt{(a / a_{\star})^{3(\alpha+1 ) } -1}\right.\nonumber\\ \nonumber\\ & & \left .
-\arctan\left(\sqrt{{(a / a_{\star})^{3(\alpha+1 ) } -1}}\right ) \right\}\end{aligned}\ ] ] the scale factor starts at finite size @xmath58 ( we have fixed constants so that this corresponds to @xmath153 ) , expands and asymptotically approaches the behavior of a model with the standard linear eos , @xmath154 for @xmath145 , fig .
[ fig1](b ) , there is a second fixed point ds representing the generalized flat expanding de sitter model .
this point has attractor stability character , i.e. in this case the generalized de sitter model is the generic future attractor . the new horizontal black line in fig .
[ fig1](b ) is a separatrix representing models with @xmath155 ; it is the dividing line between the phantom ( @xmath156 ) and standard behavior ( @xmath157 ) .
the trajectories below this line represent phantom models ( corresponding to case * c * ) , that is the energy density increases as the model expands . the models in this region appear to be dominated by shear in the past ( @xmath127 , @xmath158 as @xmath159 ) and asymptotically approach a de sitter model in the future ( @xmath160 , @xmath128 as @xmath161 ) . the scale factor for an expanding flat de sitter model
has the form : @xmath162 the trajectories above the separatrix ( corresponding to case * b * ) also asymptotically approach the generalized de sitter model in the future .
the behavior in the past is not clear from the state space in fig .
[ fig1 ] . in order to analyze the dynamics at infinity
, we introduce compactified poincar variables @xmath163 ; these can be expressed in terms of the standard variables @xmath164 as : @xmath165 the relationship can be inverted to give : @xmath166 the new variables allow one to analyze the dynamics at infinity more easily , as @xmath167 .
the evolution equations for the new variables are then given by : @xmath168 where the @xmath169 represent the evolution equations for the original variables .
we now study the compactified and reduced system ( @xmath170 , @xmath171 ) ; the fixed points of the system are given in table [ tab2 ] .
the first two fixed points ( m and ds ) represent the minkowski and generalized de sitter points as in the non - compactified case and retain the same stability character .
the third point ( is ) represents the isotropic singularity , which is matter dominated ( @xmath172 , or @xmath173 ) .
this fixed point always has repeller stability and is the generic past attractor . the final fixed point ( ks )
represents the anisotropic kasner singularity ( @xmath174 , or @xmath133 ) .
this fixed point has saddle stability
. the full state space for the system can now be reconstructed , and is depicted in fig.[fig2 ] , where as before the black lines represent separatrices , the dots represent fixed points and the thin grey lines are example trajectories .
0.47@cccc + name & @xmath170 , @xmath171 & existence & stability ( @xmath175 ) +
+ m & ( @xmath140 , @xmath140 ) & @xmath141 & saddle , attractor + + ds & @xmath176 & @xmath145 & attractor , * + + is & ( @xmath83 , @xmath140 ) & @xmath141 & repeller , repeller + + ks & ( @xmath140 , @xmath83 ) & @xmath141 & saddle , saddle + + in the case @xmath146 , fig . [ fig2](a ) ( this corresponds to case * a * , i.e. this is the compactified version of fig .
[ fig1](a ) ) , there are three fixed points , m , is and ks .
the horizontal black line @xmath177 represents the anisotropic kasner model .
the vertical black line @xmath178 represents the flat friedmann model containing a fluid with a quadratic eos .
the minkowski point m is the future attractor for all trajectories in the region .
the isotropic singularity is is the generic past attractor and the anisotropic kasner singularity ks has saddle stability character .
the set of initial conditions for which anisotropy dominates in the past are special , corresponding to vacuum models @xmath177 , as opposed to being generic as in the case of the _ linear _ eos . in the case
@xmath145 , fig .
[ fig2](b ) , there are four fixed points , m , ds , is and ks .
the state space is split into two regions by the separatrix joining the points ds and ks and given by @xmath179 , where : @xmath180 and @xmath181 . in the region above the separatrix ( @xmath182 , corresponding to case * b * ) the fluid behaves in a standard manner .
the is fixed point has repeller stability character and all trajectories are asymptotic to it in the past .
the ds fixed point has attractor stability and is the generic future attractor .
the ks fixed point has saddle stability in this region . in the region below the separatrix ( @xmath183
, corresponding to case * c * ) the fluid behaves in a phantom manner .
the ks fixed point has repeller stability character and the ds fixed point is an attractor .
the m fixed point has saddle stability character . in general , phantom trajectories in this region
are asymptotic to the kasner singularity in the past and the generalized de sitter model in the future .
the models of type bianchi v are a subclass of the bianchi class b models ( @xmath184 ) .
these are homogeneous and anisotropic cosmological models containing both the flat friedmann ( f ) model and the open friedmann models ( the milne model for @xmath185 ) . in the case of bianchi
v models the group invariant orthonormal frame @xmath186 can again be specialized further :
@xmath187 this in turn means that the spatial curvature scalar is : @xmath188 where the evolution equation for @xmath189 is : @xmath190 the einstein field equation ( [ efe4 ] ) implies : @xmath191 we will continue to use the normalized variables introduced in the previous subsection , @xmath116 , @xmath117 , @xmath118 and @xmath115 .
additionally we introduce the new normalized curvature variable @xmath192 : @xmath193 the generalized friedmann equation is now : @xmath194 as with the previous case , this constraint equation can be used to reduce the dimension of the dynamical system : again we chose to remove @xmath117 .
the resulting set of evolution equation is then : @xmath195 the fixed points of the new reduced system , the corresponding eigenvalues and their existence conditions ( the energy density remains positive , @xmath120 , and all four variables are real , @xmath196 ) are given in table [ tab3 ] .
the fixed points are identical to those found in the bianchi i case . as with the bianchi i case , in order to analyze the dynamics at infinity , we introduce the poincar variables @xmath163 .
we then study the compactified and reduced system ( @xmath170 , @xmath171 , @xmath197 ) , with the fixed points given in table [ tab4 ] .
0.47@ccccc + name & @xmath116 , @xmath118 , @xmath192 & eigenvalues & existence + + m & ( @xmath140 , @xmath140 , @xmath140 ) & @xmath140 , @xmath140 , @xmath140 & @xmath141 + + ds & ( @xmath142 , @xmath140 , @xmath140 ) & @xmath143 , @xmath144 , @xmath198 & @xmath145 + + 0.47@cccc + name & @xmath170 , @xmath171 , @xmath197 & existence & stability ( @xmath175 ) +
+ m & ( @xmath140 , @xmath140 , @xmath140 ) & @xmath141 & saddle , attractor + + ds & @xmath199 & @xmath145 & attractor , * + + cs & ( @xmath140 , @xmath140 , @xmath83 ) & @xmath141 & saddle , saddle + + is & ( @xmath83 , @xmath140 , @xmath140 ) & @xmath141 & repeller , repeller + + ks & ( @xmath140 , @xmath83 , @xmath140 ) & @xmath141 & saddle , saddle + + the first two fixed points m and ds represent the minkowski and generalized de sitter points as in the bianchi i case and retain the same stability character .
the third point ( cs ) is the curvature singularity ) as well as the coordinate singularity of the milne model , the line @xmath200 in the same figure . ] , which is curvature dominated ( @xmath201 , or @xmath202 ) .
the fourth point ( is ) represents the isotropic singularity and the final fixed point ( ks ) represents the anisotropic kasner singularity .
we now discuss the state space diagrams for this system .
the case @xmath146 is shown in fig . [ fig3](a ) ( this corresponds to case * a * ) . the @xmath203 plane ( @xmath204 ) is the invariant manifold which represents the reduced and compactified bianchi i state space , which is a manifold of the larger bianchi v state space .
the @xmath205 plane ( @xmath177 ) is the vacuum invariant manifold and the @xmath206 plane ( @xmath178 ) is the isotropic invariant manifold .
the minkowski point m is the future attractor for all trajectories in the region .
the isotropic singularity ( is ) is the generic past attractor , while the anisotropic kasner singularity ( ks ) and curvature singularity ( cs ) have saddle stability character . as in the bianchi i case , in general the past singularity is isotropic . in the case @xmath145 ,
depicted in fig .
[ fig3](b ) , there are five fixed points , m , ds , cs , is and ks .
the state space is split into two regions by the separatrix manifold joining the points ds , cs and ks .
this manifold is given by @xmath179 , where : @xmath207 and @xmath181 . in the @xmath205 plane ( @xmath177 , vacuum invariant manifold )
the dynamics are unaffected by the change in @xmath37 . the @xmath206 plane ( @xmath178 , isotropic invariant manifold )
is divided into two regions by the separatrix . in the region above the separatrix ( @xmath182 , corresponding to case * b * ) the fluid behaves in a standard manner .
the is fixed point has repeller stability character and all trajectories are asymptotic to it in the past .
the ds fixed point is the generic future attractor .
the cs and ks fixed points have saddle stability character in this region . in the region below the separatrix ( @xmath183
, corresponding to case * c * ) the fluid behaves in a phantom manner .
the ks fixed point is a repeller and the ds fixed point the attractor .
the m and cs fixed points have saddle stability character . in general , trajectories are asymptotic to the kasner singularity in the past and the generalized de sitter model in the future .
we have seen in the previous section that generic trajectories of bianchi i and v anisotropic models with standard ( non - phantom ) matter of type * a * and * b * asymptotically approach an isotropic singularity in the past , with the hubble normalized shear vanishing , @xmath208 .
since the bianchi i and v models are rather special and homogeneous , we now want to see if this behavior is restricted to these models only , or if instead is stable against small generic inhomogeneuos perturbations .
that is , we want to see if appropriately defined dimensionless quantities measuring the deviation from homogeneity in a covariant gauge invariant way decay to zero when the singularity is approached .
since we are only interested in the neighborhood of the singularity , we only need to consider perturbations of the spatially flat ( @xmath209 ) homogenous and isotropic cosmological models ( the flat friedmann models , i.e. the vertical @xmath170 axis in figs .
[ fig2 ] and [ fig3 ] ) with quadratic eos @xmath210 given this eos , @xmath211 is always satisfied for high enough @xmath14 , whatever @xmath37 is , hence at high enough energies the background evolution is non - inflationary , @xmath212 .
therefore we only need to consider large scale perturbations , i.e. scales much larger than the hubble horizon @xmath213 , because for any given perturbation scale @xmath9 there is always a time @xmath214 ( the horizon crossing time ) such that for @xmath215 we have @xmath10 . for simplicity
we now focus on models of type * a * only , with @xmath146 ; it can be shown that similar results can be obtained in case * b * , @xmath145 .
for this case we re - write eq.([rho1 ] ) as @xmath216 where @xmath217 is the finite value of the scale factor such that the energy density has a pole like singularity .
@xmath218 is a constant given by ( [ aa ] ) and is expressed in terms of @xmath219 , @xmath220 , @xmath36 and @xmath37 .
given that the background is a spatially flat friedmann model the friedmann equation ( @xmath221 ) gives directly the hubble expansion scalar @xmath16 , while @xmath220 can be fixed arbitrarily , a freedom that we will use later .
the sound speed for the fluid is given by : @xmath222 the more commonly used eos parameter @xmath223 is : @xmath224 the sound speed and eos variable are then related by the following relationship : @xmath225 thus in case * a * our parameter @xmath37 could be replaced by the combination @xmath226 of the more commonly used parameters @xmath223 and @xmath227 .
we use the covariant 1 + 3 approach @xcite to study inhomogeneous and anisotropic gauge - invariant perturbations . in this approach ,
the information relating to inhomogeneous and anisotropic perturbations is contained in a set of ( 1 + 3 ) covariant variables .
these variables are defined with respect to a preferred time - like observer congruence @xmath228 .
we use appropriate dimensionless density , expansion and curvature gradients which describe the scalar and vector parts of the corresponding perturbations : @xmath229 the dimensionless expansion normalized quantities for the shear @xmath230 , the vorticity @xmath231 , the electric @xmath232 and magnetic @xmath233 parts of the weyl tensor are : @xmath234 where @xmath16 is the hubble scalar , @xmath235 . here
@xmath236 is the usual vorticity vector . in order to calculate all the relevant tensor perturbations it is useful to define a set of variables corresponding to the curls of the standard quantities .
the curl of a given quantity will be denoted by an over - bar . @xmath237
the harmonics defined in @xcite are used to expand the above tensors @xmath238 in terms of scalar ( s ) , vector ( v ) and tensor ( t ) harmonics @xmath239 .
this results in a covariant and gauge invariant splitting of the evolution and constraint equations for the above quantities into three sets of evolution and constraint equations for scalar , vector and tensor modes .
the scalar , vector and tensor quantities can be split such that : @xmath240 as said above , we will carry out our analysis in the long - wavelength limit @xmath10 , or @xmath241 ( @xmath242 ) .
this is analogous to dropping the laplacian terms in the evolution and constraint equations . [
cols="^,^,^,^,^",options="header " , ] the linearized evolution equations for the tensor components of the perturbations in the large - scale limit are : @xmath243 the constraint equations for the tensor variables are @xmath244 the perturbation modes for each of the tensor quantities can be found in table [ table3 ] .
the constants of integration ( @xmath245 and @xmath246 ) correspond to the four independent modes . the non - zero constants @xmath247 , @xmath248 , @xmath249 and @xmath250 can all be expressed in terms of the constants of integration and the eos parameters .
the perturbations all decay at the singularity and evolve as there linear eos counterparts at late times .
in this paper we have investigated the effects of a quadratic eos in homogenous and inhomogeneous anisotropic cosmological models in general relativity , in order to understand if in this context the quadratic eos can isotropize the universe at early times , when the initial singularity is approached .
the first motivation for this work is that in the context of brane world cosmological models the term quadratic in the energy density that appears in the effective 4-dimensional equations of motion on the brane dominates at early times , driving the evolution to an initial isotropic singularity under reasonable assumptions .
therefore , under these assumptions , in the brane world scenario we do not have the classical isotropy problem : the isotropy we observe today is built in , that is it follows from _ generic initial conditions_. this is at odds with the classical and now widely accepted results in general relativity @xcite , where general cosmological models with a linear eos are highly anisotropic in the past , approaching the singularity _ a - la _ bkl , i.e. with mixmaster dynamics , so that in general one needs _ special initial conditions _ to have isotropy at late times , even to initiate an inflationary phase @xcite .
secondly , we also wanted to study the effects at early times and at high energies of the same type of eos used in paper i@xcite , relaxing the symmetries assumed there .
the focus in paper i@xcite was on the possible use of a quadratic eos as an effective way of representing a dark energy component or unified dark matter in the context of homogeneous isotropic models of relevance for the universe at late times .
here we have used the simplified eos @xmath251 , which still allows for an effective cosmological constant and phantom behavior , and is general enough to analyze the dynamics at high energies . in summary ,
our aim here was to investigate if the behavior of the anisotropy at the singularity found in the brane scenario is recreated in the general relativistic context if we consider a quadratic term in the eos .
that is to say , in the language of dynamical system theory , do we have a scenario where the generic asymptotic past attractor in general relativity is an isotropic model , thanks to a non - linear ( specifically , quadratic ) eos . to this end
, we have first summarized in section [ section2 ] the classification of the energy density evolution in three classes , * a * , * b * and * c * , following from the energy conservation for the eos @xmath12 ( a broader classification can be found in section ii of paper i@xcite ) . in section [ section3 ] , using dynamical system methods @xcite , we have studied the bianchi i and v cosmological models with a fluid with this eos .
we first considered the case when @xmath146 ( case * a * ) .
the physically interesting quantity @xmath252 , the hubble normalized dimensionless measure of shear , describes the anisotropy . in the expanding bianchi i and v models this quantity always decreases at late times , as in the case with a _ linear _ eos , approaching a minkowski model .
however , we can now have stages at early times in which @xmath252 grows , vanishing at the initial singularity , and generic trajectories in the compactified state space asymptotically approach in the past the isotropic fixed point is , as depicted in figs . [ fig2 ] and [ fig3 ] .
this behavior is similar to the quasi - isotropic expansion seen in the anisotropic brane - world models , but occurs for different reasons .
the quasi - isotropic expansion is caused by the fact that the eos becomes super - stiff " at early times but decreases to finite values at late times . at the initial singularity ( which is of type iii ,
see @xcite ) the anisotropy is always sub - dominant and we recover the standard _ linear _ eos behavior at late times . in the case
@xmath145 , the state space can be divided in two regions . in the standard fluid region ( @xmath182 ,
case * b * ) the behavior at early times is similar to the @xmath146 case , with generic trajectories in the compactified state space approaching the isotropic fixed point is of figs .
[ fig2 ] and [ fig3 ] . at late times
instead trajectories asymptotically approach generalized expanding de sitter models . finally , in the phantom fluid region ( @xmath183 , case * c * ) the anisotropy dominates at early times and the past singularity is anisotropic .
trajectories are asymptotic to the expanding de sitter model at late times .
the submanifold @xmath253 in the state space for the bianchi models represents homogeneous isotropic spatially flat friedmann - robertson - walker models , with is as past attractor . in order to see
if is can be regarded as the past attractor of a larger class of general models with no symmetries , we have studied in section [ section4 ] , using covariant and gauge invariant variables @xcite , the first order inhomogeneous anisotropic perturbations of these flat isotropic models , focusing on those of class * a*. the linearized equations have been solved in the large scale limit @xmath254 , neglecting the laplacian operators terms . as explained in section [ intro ] , only perturbations larger then the hubble horizon @xmath213 are relevant in our analysis .
the main results of this section are given in table s [ table1 ] , [ table2 ] and [ table3 ] .
the tables contain the form of the solutions and asymptotic behavior of the physically relevant quantities for both early ( @xmath255 ) and late times ( @xmath256 ) .
the quantities have been harmonically decomposed into the standard scalar , vector and tensor components .
these include the expansion normalized vorticity , shear , and electric and magnetic contributions of the weyl tensor .
in addition , we consider the gradients of the energy density , expansion and three - curvature . as expected , at late times the perturbations evolve as if the fluid obeyed a _ linear _ eos , i.e. the standard behavior is recovered in this limit .
the evolution of the perturbations at early times is our main result .
indeed , we find that all perturbations tend to zero near the singularity .
thus a perturbed quasi - isotropic model with generic initial condition evolves from the isotropic past attractor is .
this indicates that in general relativity with matter described by a non - linear eos there are general cosmological models , corresponding to a non zero measure subset of all possible generic initial conditions @xcite , that isotropize as the initial singularity is approached , with is as the past attractor . as for the brane models
, this would be at odds with the standard bkl results @xcite .
singularities are commonly regarded as signaling the failure of classical gravitational theories , requiring a quantum theory of gravity to deal with trans - plankian energies .
even in effective theories , perhaps cosmology does nt require them , as in the pre - big - bang scenario @xcite . in any case , it is important to understand which is the typical classical dynamical behavior in the neighborhood of singularities , as this possibly signals what the generic outcome of a quantum theory should be at low ( sub - plankian ) energies . in the bkl picture for fluids with linear eos and other simple type of matter @xcite
this behavior is mixmaster and the effects of matter become negligible close to the singularity : we may say that effectively in this limit the space - time still tells matter how to move , but matter is no longer able to tell the space - time how to curve .
perhaps this behavior should be regarded as pathological , and peculiar of the energy - momentum tensor of linear eos matter ( or other simple energy - momentum tensor ) in general relativity . in practice we know close to nothing about matter fields at the highest ( close to plankian ) cosmological energies , and these may require a non - linear effective eos description in general , or maybe einstein equations with an effective non - linear eos are the einstein frame version of some effective low energy ( sub - plankian ) theory . in both cases , our analysis of einstein equations with a fluid with a quadratic eos indicates that the typical initial state is isotropic .
the singularity is matter dominated , and of a type that could admit an extension .
interestingly , the eos considered here are a subclass of those used in paper i@xcite to model a dark energy component or unified dark matter , allowing for effective cosmological constants .
kna is supported by pparc ( uk ) .
mb is partly supported by a visiting grant by miur ( italy ) .
the authors would like to thank chris clarkson , roy maartens and david wands for useful comments and discussions .
r. scranton _ et al .
_ [ sdss collaboration ] , arxiv : astro - ph/0307335 ; m. tegmark _ et al . _
[ sdss collaboration ] , phys .
d * 69 * , 103501 ( 2004 ) [ arxiv : astro - ph/0310723 ] . c. l. bennett _ et al .
_ , astrophys . j. suppl .
* 148 * , 1 ( 2003 ) [ arxiv : astro - ph/0302207 ] ; l. page _ et al .
_ , astrophys .
j. suppl .
* 148 * , 233 ( 2003 ) [ arxiv : astro - ph/0302220 ] .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) [ arxiv : astro - ph/0302209 ] .
d. n. spergel _ et al . _
[ wmap collaboration ] , [ arxiv : astro - ph/0603449 ] . for reviews on singularities in general relativity and other theories of gravity
see the articles of rendall and nicolai in _ 100 years of relativity .
space - time structure : einstein and beyond _ , ed . by a.
ahtekar ( world scientific , 2005 ) .
t. shiromizu , k. i. maeda and m. sasaki , phys .
d * 62 * , 024012 ( 2000 ) [ arxiv : gr - qc/9910076 ]
. h. a. bridgman , k. a. malik and d. wands , phys .
d * 65 * , 043502 ( 2002 ) [ arxiv : astro - ph/0107245 ] .
d. langlois , astrophys .
space sci . * 283 * , 469 ( 2003 ) [ arxiv : astro - ph/0301022 ] .
r. maartens , living rev .
* 7 * , 7 ( 2004 ) [ arxiv : gr - qc/0312059 ] .
r. maartens , v. sahni and t. d. saini , phys .
d * 63 * , 063509 ( 2001 ) [ arxiv : gr - qc/0011105 ] .
a. campos and c. f. sopuerta , phys .
d * 63 * , 104012 ( 2001 ) [ arxiv : hep - th/0101060 ] .
a. a. coley , phys .
d * 66 * , 023512 ( 2002 ) [ arxiv : hep - th/0110049 ] .
r. j. van den hoogen , a. a. coley and y. he , phys .
d * 68 * , 023502 ( 2003 ) [ arxiv : gr - qc/0212094 ] .
a. a. coley , y. he and w. c. lim , class .
* 21 * , 1311 ( 2004 ) [ arxiv : gr - qc/0312075 ] .
p. dunsby , n. goheer , m. bruni and a. coley , phys .
d * 69 * , 101303 ( 2004 ) [ arxiv : hep - th/0312174 ] .
n. goheer , p. k. s. dunsby , a. coley and m. bruni , phys .
d * 70 * , 123517 ( 2004 ) [ arxiv : hep - th/0408092 ] .
s. nojiri , s. d. odintsov and s. tsujikawa , phys .
d * 71 * , 063004 ( 2005 ) [ arxiv : hep - th/0501025 ] .
j. d. barrow , class .
* 21 * , 5619 ( 2004 ) [ arxiv : gr - qc/0409062 ] ; j. d. barrow , class .
quant . grav .
* 21 * , l79 ( 2004 ) [ arxiv : gr - qc/0403084 ] . | in standard general - relativistic cosmology , for fluids with a linear equation of state ( eos ) @xmath0 or scalar fields , the high isotropy of the universe requires special initial conditions : singularities are velocity dominated and anisotropic in general . in brane world
effective 4-dimensional cosmological models an effective term , quadratic in the energy density , appears in the evolution equations , which has been shown to be responsible for the suppression of anisotropy and inhomogeneities at the singularity under reasonable assumptions .
thus in the brane world isotropy is generically built in , and singularities are matter dominated .
there is no reason why the effective eos of matter should be linear at the highest energies , and an effective non - linear eos may describe dark energy or unified dark matter ( paper i@xcite ) . in view of this , here we investigate the effects of a quadratic eos in homogenous and inhomogeneous anisotropic cosmological models in general relativity , in order to understand if in this standard context the quadratic eos can isotropize the universe at early times . with respect to paper i@xcite , here we use the simplified eos @xmath1 , which still allows for an effective cosmological constant and phantom behavior , and is general enough to analyze the dynamics at high energies .
we first study homogenous and anisotropic bianchi i and v models , focusing on singularities . using dynamical systems methods , we find the fixed points of the system and study their stability .
we find that models with standard non - phantom behavior are in general asymptotic in the past to an isotropic fixed point is , i.e. in these models even an arbitrarily large anisotropy is suppressed in the past : the singularity is matter dominated . using covariant and gauge invariant variables
, we then study linear anisotropic and inhomogeneous perturbations about the homogenous and isotropic spatially flat models with a quadratic eos .
we find that , in the large scale limit , all perturbations decay asymptotically in the past , indicating that the isotropic fixed point is is the general asymptotic past attractor for non phantom inhomogeneous models with a quadratic eos . |
there has been a lot of interest recently in identifying an infrared fixed point , and the possibility of an associated conformal window , for @xmath3 gauge theories as a function of the number of fermion flavours @xmath2@xcite , motivated by the pioneering work of banks and zaks@xcite . for a valuable review ,
see ref .
much of the interest stems from the longstanding suggestion that a ( nearly ) conformal gauge theory could play a key role in electroweak symmetry breaking beyond the standard model of particle physics@xcite . for fermions in the fundamental representation ,
the prediction obtained from a detailed study using the wilson fermion action@xcite differs from subsequent studies that use other actions@xcite . according to ref .
@xcite it is generally believed that confinement exists in the strong coupling limit for any @xmath2 ; the disagreement with data from the wilson action@xcite suggests that the unphysical wilson term has a dramatic effect . in the present work
, we revisit the strong coupling limit of the wilson action with the goal of providing some additional insight into this issue . an @xmath3 lattice gauge theory with fundamental wilson fermions
is expected to have a phase where flavour and parity are spontaneously broken , known as the aoki phase@xcite .
the expected phase diagram ( for example figure 3 of ref .
@xcite ) is sketched in the plane spanned by the two relevant parameters : the gauge coupling and the fermion mass .
the weak coupling part of the diagram shows `` fingers '' of aoki phase separated by regions of symmetric phase , but ongoing research by various groups is refining this picture for specific discretizations and lattice improvement scenarios@xcite .
the strong coupling part of the original expected phase diagram is simpler : the aoki phase exists for quark masses of smaller magnitude and the symmetric phase exists for quark masses of larger magnitude . in the extreme strong coupling limit ( @xmath4 ) , the phase boundary
is predicted to be at a specific critical value of the hopping parameter , @xmath5 recall that the hopping parameter @xmath6 is related to the dimensionless bare fermion mass @xmath7 as follows : @xmath8 in addition , the dimensionless pion mass in the strong coupling limit is predicted to be a rather simple function of the hopping parameter@xcite , @xmath9 this expression for the pion mass was derived long ago for quenched configurations ( i.e. @xmath10 inside the configurations)@xcite , and initial tests of its validity for @xmath11 were presented in some of the direct numerical simulations within ref .
a central result of our work is the observation of systematic deviations from eq .
( [ eq : pion ] ) in exploratory simulations with multiple wilson fermions at @xmath12 .
for this study we use the standard plaquette gauge action and the standard wilson fermion action for each of the @xmath2 degenerate fermions , @xmath13 \,,\end{aligned}\ ] ] where @xmath14 with @xmath15 the gauge coupling , and where @xmath16 is the @xmath3-valued link variable
. then the partition function @xmath17 is @xmath18 \prod_{f=1}^{n_f}[d { \bar\psi}^f(x)][d \psi^f(x)]\exp(-s ) \nonumber \\ & = & \int [ d u_\mu(x ) ] \left(\det(d_w^{\dagger}d_w)\right)^{\frac{n_f}{2 } } \exp(-s_g ) \,,\end{aligned}\ ] ] where @xmath19 is the kernel of the single - fermion action @xmath20 .
simulations are performed with a standard hybrid monte carlo ( hmc ) . for independent confirmation , simulations for the heavier fermions
are repeated using a polynomial hybrid monte carlo ( phmc ) . in each molecular dynamics
( md ) evolution , the number of steps per trajectory @xmath21 is tuned to beyond @xmath22 acceptance for the metropolis test .
the step size @xmath23 is defined by @xmath24 .
our simulations focus primarily on @xmath12 , but some studies at @xmath25 will also be reported .
lattice sizes include @xmath26 , @xmath27 , @xmath28 and @xmath29 .
with zero temperature simulations in mind , we use periodic boundary conditions and define the euclidean time dimension to be greater than or equal to every spatial dimension . for each chosen set of parameters ,
50 to 100 configurations were collected after thermalization , with 4 or 5 trajectories between saved configurations .
our primary observables are the average plaquette @xmath30 , the pseudoscalar meson mass @xmath31 , and the axial - ward - takahashi - identity quark mass @xmath32 , all defined in standard fashion : @xmath33 where @xmath34 denotes the total number of lattice sites , @xmath35 is the symmetric temporal derivative operator , and the flavour non - singlet bilinear operators ( with flavour index omitted ) are @xmath36
for a first indication of the difference between quenched and dynamical simulations , consider the average plaquette for various numbers of flavours , as shown in fig .
[ fig : plaqnc2b0 ] . in @xmath0 gauge theory with @xmath2 flavours at @xmath12 .
, width=453 ] an abrupt transition is clearly seen for @xmath37 , with a two - state signal defining the transition region in each case .
the data at @xmath38 hint at a transition but are not conclusive , though other observables to be discussed below will confirm a @xmath38 transition . at @xmath39
the transition is either suppressed or absent .
this behaviour implies the existence of a first order transition for sufficiently large @xmath2 , and was already reported for the @xmath1 case in ref .
@xcite and for @xmath0 with two flavours of adjoint - representation fermions in refs .
our results show that the first order transition moves to larger values of @xmath40 as @xmath2 is increased .
all of our observed transitions are in the range @xmath41 .
[ fig : mpib0 ] shows @xmath42 and @xmath32 over a wide range of @xmath40 . notice that @xmath42 and @xmath32 become insensitive to @xmath2 as @xmath40 increases .
this is no surprise because sufficiently heavy quarks have a minimal effect on the vacuum structure of the theory , so meson and quark masses for any @xmath11 approach their quenched ( @xmath10 ) values as @xmath40 grows . and
@xmath32 as functions of @xmath40 in @xmath0 gauge theory with @xmath2 flavours at @xmath12 . ,
width=453 ] but for a smaller range of @xmath40 . , width=453 ] a close - up of the small @xmath40 region
is displayed in fig .
[ fig : mpib0big ] . the same transition that was observed in the elementary plaquette is clearly seen in @xmath42 and @xmath32 as well . from these data
it is clear that the transition exists at least for @xmath43 and the two - state signals imply that it is first order at least for @xmath37 .
the transition moves to larger @xmath40 as @xmath2 is increased and we speculate that the transition will approach @xmath44 as @xmath45 , i.e.the gauge theory with infinitely many fermions will become the free theory . and @xmath32 as a function of @xmath40 in @xmath0 gauge theory for @xmath46 and 12 at @xmath12 .
the prediction from eq .
( [ eq : pion ] ) is shown as a solid blue curve that reaches @xmath47 at @xmath48 .
quadratic fits to lattice data for @xmath42 are shown as a pair of solid curves ( @xmath46 is black and @xmath49 is orange ) .
, width=453 ] it should be noted that fig .
[ fig : plaqnc2b0 ] is obtained for all four of our lattice volumes : no volume dependence is observed .
this was expected in advance because the average plaquette is a short distance quantity .
we have similarly confirmed volume independence for @xmath42 and @xmath32 over our range of volumes , and examples are presented in fig .
[ fig : mpib0nf6 ] .
in particular , the transition shows no evidence of vanishing in the infinite volume limit , and it shows no evidence of moving to @xmath50 in the infinite volume limit .
besides the transition itself , figs .
[ fig : mpib0big ] and [ fig : mpib0nf6 ] show an interesting curvature at values of @xmath40 above the transition , where the slope deviates more and more from the quenched slope when approaching the transition point . if someone wanted to extrapolate to @xmath47 using only data above the transition , then the critical hopping parameter so obtained , let s name it @xmath51 , would be smaller than @xmath52 . of particular interest is the fact that even the data above the transition in figs .
[ fig : mpib0big ] and [ fig : mpib0nf6 ] do not follow eq .
( [ eq : pion ] ) and do not arrive at @xmath53 despite eq .
( [ eq : standardkappa ] ) . to ascertain the properties of the transition and of the regions it separates ,
a range of other observables should be discussed .
that discussion goes beyond the scope of the present work , but brief comments and selected plots can be found in appendix a.
qualitatively the situation for @xmath1 is similar to that for @xmath0 , but our quantitative results indicate that deviations from eq .
( [ eq : pion ] ) are smaller for @xmath1 .
perhaps a large-@xmath54 suppression mechanism is at work .
data for the average plaquette , pseudoscalar meson mass , and axial - ward - takahashi - identity quark mass are plotted in figs .
[ fig : plaqnc3b0 ] and [ fig : nc3mpib0 ] . notice that @xmath42 is consistent with eq .
( [ eq : pion ] ) for @xmath10 , but the @xmath11 data display bending that is very similar to the @xmath0 case but of smaller magnitude .
the search for a possible transition is more expensive for @xmath1 than for @xmath0 because we must work at larger @xmath6 .
a transition is found to exist for @xmath55 but we can not say whether it is abrupt . recall that ref .
@xcite found the transition to occur only for @xmath56 ; they rely to some extent on counting the conjugate gradient iterations required during thermalization to determine the phase for @xmath46 at @xmath57 .
( we comment briefly in appendix b. ) our results at @xmath58 in our case are generated from the last configuration of @xmath59 at @xmath58 .
we use the hmc algorithm with a time step of @xmath60 , while ref .
@xcite uses the approximate @xmath61 algorithm with @xmath62 . ]
do show hints of a transition for @xmath46 . in @xmath1 gauge theory with @xmath2 flavours at @xmath12 .
, width=453 ] and @xmath32 as functions of @xmath40 in @xmath1 gauge theory with @xmath2 flavours at @xmath12 .
the prediction from eq .
( [ eq : pion ] ) is shown as a solid blue curve reaching @xmath47 at @xmath48 .
, width=453 ]
simulations on small space - time lattices in the strong coupling limit display the phase transition already discussed in ref .
@xcite , but the present work makes the additional observation that the dependence of the pseudoscalar meson mass on the hopping parameter deviates from the expected formula , eq .
( [ eq : pion ] ) , as seen in figs .
[ fig : mpib0nf6 ] and [ fig : nc3mpib0 ] . these deviations , which appear as bending in plots of meson ( and quark ) mass versus hopping parameter ,
are easily seen for @xmath63 and are smaller but still observed for @xmath64 . the formula in eq .
( [ eq : pion ] ) was derived through a large-@xmath54 expansion@xcite , and the present data for @xmath65 suggest a rather rapid approach to eq .
( [ eq : pion ] ) as @xmath54 is increased .
( [ eq : pion ] ) is also obtained for any @xmath54 with @xmath10 . because of the phase transition there is generally no value of the hopping parameter at which the pseudoscalar meson becomes massless , but extrapolation from above the phase transition ( i.e. from smaller hopping parameters ) does allow the definition of an effective critical hopping parameter , @xmath66 .
this extrapolated value is found to obey @xmath67 as a consequence of the observed deviation away from eq .
( [ eq : pion ] ) .
the authors of ref .
@xcite performed a detailed study of @xmath1 gauge theory with @xmath2 wilson fermions and from that study they proposed a phase structure with regions of confinement and deconfinement for certain values of @xmath2 .
they also include a brief appendix ( in the most recent publication in @xcite ) devoted to the @xmath0 theory , where @xmath68 is assumed and simulations at that hopping parameter are used as the basis for conclusions about a confinement / deconfinement transition .
the results of ref .
@xcite differ from those obtained using other actions@xcite , and one might ask whether assumptions about @xmath68 might play a role .
however , eq . ( [ ourshift ] ) confirms that for any @xmath2 @xmath69 where @xmath70 is the location of the phase transition ; this fact is sufficient to leave the conclusions of ref .
@xcite intact .
therefore the source of discrepancies between the phase structure obtained from the wilson action versus other actions must be sought elsewhere .
it is worth noting that phenomena in the strong coupling limit are not typical of results at weaker couplings .
for example the average plaquette , meson mass , and quark mass for the @xmath0 theory at @xmath25 are plotted in figs .
[ fig : plaqb2 ] and [ fig : mpib2 ] . instead of an abrupt transition that is essentially independent of lattice volume
, we now see smoothly continuous functions and the meson mass has a significant volume dependence .
detailed studies at @xmath71 are left for future work . in @xmath0 gauge theory with @xmath2 flavours at @xmath25 on @xmath27 lattices .
, width=453 ] and @xmath32 as functions of @xmath40 in @xmath0 gauge theory with @xmath2 flavours at @xmath25 .
lattices are @xmath27 except for one point labelled as @xmath72 .
, width=453 ]
a polyakov loop is the trace of a product of links in a straight line along a lattice axis which , via periodic boundary conditions , closes upon itself .
polyakov loops are often used as an indicator for confinement .
the absolute value of a polyakov loop is plotted in fig .
[ fig : polyakov ] at @xmath12 on @xmath26 lattices . on @xmath2 flavours
is expressed through the difference @xmath73 .
this difference is plotted as a function of @xmath40 in @xmath0 gauge theory on @xmath26 lattices at @xmath12 with @xmath46 and 8 .
, width=453 ] for each of the two cases @xmath46 and @xmath59 , the graph shows the polyakov loops in both the shorter ( @xmath74 ) and longer ( @xmath75 ) lattice directions .
the shorter polyakov loops are numerically large and dependent on the hopping parameter , with a two - state signal at the transition for large enough @xmath2 .
the longer polyakov loops are consistent with zero for all values of @xmath6 .
we refrain from drawing conclusions about confinement from this brief consideration of polyakov loops . for discussions of chiral symmetry and its possible breaking ,
natural quantities include the chiral condensate @xmath76 and the pseudoscalar density @xmath77 .
the low - lying eigenvalues of the wilson - dirac operator relate to these observables through the spectral representation of the quark propagator . for instance , @xmath78 and @xmath79 , where @xmath80 and @xmath81 are an eigenvalue and eigenvector respectively of the hermitian wilson - dirac operator @xmath82 . the upper panel of fig .
[ fig : propnorm ] is the lowest eigenvalue , @xmath83 , defined by @xmath84 in the eigenvalue equation , @xmath85 , and obtained by the ritz functional method .
this behaviour hints at the dependence @xmath86 . at least for @xmath43 ,
two phases are found and in both phases the eigenvalues are not particularly small . because wilson fermions break chiral symmetry explicitly , the chiral condensate is not adequate and should be replaced by the subtracted chiral condensate defined by @xmath87 where @xmath88 .
the fermion mass @xmath32 itself is discussed in the main body of this article , and it is most naturally defined only for @xmath40 values above the transition , so here we focus on the observation of @xmath89 .
the lower panel of fig .
[ fig : propnorm ] displays @xmath90 for @xmath0 gauge theory at strong coupling for various @xmath2 .
indications of a @xmath91 divergence in the confinement phase ( i.e. @xmath40 above the transition ) are visible , as expected . for both observables in fig .
[ fig : propnorm ] , the transition is visible when @xmath43 .
gauge theory with @xmath2 flavours at @xmath12 on a @xmath26 lattice .
upper panel : @xmath84 , where @xmath92 is the lowest eigenvalue of @xmath93 , as a function of @xmath40 .
lower panel : @xmath94 as a function of @xmath40 .
, width=453 ] gauge theory with @xmath2 flavours at @xmath12 on a @xmath26 lattice .
upper panel : @xmath84 , where @xmath92 is the lowest eigenvalue of @xmath93 , as a function of @xmath40 .
lower panel : @xmath94 as a function of @xmath40 .
, width=455 ]
the hmc algorithm uses a number of trajectories , traditionally distinguished by an integer @xmath95 .
each trajectory uses some number of molecular dynamics steps , @xmath21 , and each of those uses some number of conjugate gradient ( cg ) iterations , @xmath96 with @xmath97 , to attain the imposed precision .
the average number of cg iterations per md step is therefore @xmath98 .
the left panel of fig .
[ fig : cgsu2 ] shows this quantity for sequential trajectories during thermalization for @xmath0 with @xmath46 , @xmath12 , and @xmath99 on a @xmath27 lattice with precision defined by @xmath100 . for reference ,
the average plaquette is also plotted for each trajectory .
the hot start requires more trajectories before reaching thermalization , and some of those trajectories needed many cg iterations .
one could possibly be tempted to abort such a calculation before reaching equilibrium and assume that the increasing @xmath101 would never return to a smaller value at larger @xmath95 .
the right panel of fig .
[ fig : cgsu2 ] shows @xmath102 for @xmath1 at @xmath46 , @xmath12 , and @xmath103 on a @xmath26 lattice with @xmath100 and @xmath104 .
when configuration generation starts from the last configuration of @xmath59 at @xmath103 , the hmc becomes stable at @xmath105 acceptance and @xmath106 . on the other hand ,
the hot and cold starts are not accepted by the hmc s metropolis test with @xmath107 . in both cases trajectories in about 300 runs .
] of fig .
[ fig : cgsu2 ] , the large value of @xmath102 does not imply the existence of a massless mode . versus trajectory index @xmath95 in @xmath0 gauge theory on @xmath27 lattice at @xmath12 and @xmath99 with @xmath46 .
right panel : @xmath102 versus @xmath95 in @xmath1 gauge theory with @xmath46 on @xmath26 lattice at @xmath12 , @xmath57 and @xmath104 .
, title="fig:",width=275 ] versus trajectory index @xmath95 in @xmath0 gauge theory on @xmath27 lattice at @xmath12 and @xmath99 with @xmath46 .
right panel : @xmath102 versus @xmath95 in @xmath1 gauge theory with @xmath46 on @xmath26 lattice at @xmath12 , @xmath57 and @xmath104 .
, title="fig:",width=294 ]
this work was supported in part by the natural sciences and engineering research council of canada , and was made possible by the facilities of the shared hierarchical academic research network ( sharcnet : www.sharcnet.ca ) .
thanks the university of regina for hospitality during her stay there , and we acknowledge use of regina s vxrack computer cluster . 99 y. iwasaki , k. kanaya , s. sakai and t. yoshie , phys . rev .
lett . * 69 * , 21 ( 1992 ) and y. iwasaki , k. kanaya , s. kaya , s. sakai and t. yoshie , phys .
d * 54 * , 7010 ( 1996 ) [ arxiv : hep - lat/9605030 ] and y. iwasaki , k. kanaya , s. kaya , s. sakai and t. yoshie , phys .
d * 69 * , 014507 ( 2004 ) [ arxiv : hep - lat/0309159 ] .
s. catterall and f. sannino , phys . rev .
d * 76 * , 034504 ( 2007 ) [ arxiv:0705.1664 [ hep - lat ] ] and s. catterall , j. giedt , f. sannino and j. schneible , jhep * 0811 * , 009 ( 2008 ) [ arxiv:0807.0792 [ hep - lat ] ]
. t. appelquist , g. t. fleming and e. t. neil , phys .
lett . * 100 * , 171607 ( 2008 ) [ erratum - ibid . *
102 * , 149902 ( 2009 ) ] [ arxiv:0712.0609 [ hep - ph ] ] and phys .
d * 79 * , 076010 ( 2009 ) [ arxiv:0901.3766 [ hep - ph ] ] . y. shamir , b. svetitsky and t. degrand , phys . rev .
d * 78 * , 031502 ( 2008 ) [ arxiv:0803.1707 [ hep - lat ] ] and phys .
d * 79 * , 034501 ( 2009 ) [ arxiv:0812.1427 [ hep - lat ] ] .
a. deuzeman , m. p. lombardo and e. pallante , phys .
b * 670 * , 41 ( 2008 ) [ arxiv:0804.2905 [ hep - lat ] ] and arxiv:0904.4662 [ hep - ph ] . l. del debbio , a. patella and c. pica , arxiv:0805.2058 [ hep - lat ] and arxiv:0812.0570 [ hep - lat ] .
z. fodor , k. holland , j. kuti , d. nogradi and c. schroeder , arxiv:0809.4888 [ hep - lat ] and arxiv:0809.4890 [ hep - lat ] and arxiv:0907.4562 [ hep - lat ] .
a. j. hietanen , j. rantaharju , k. rummukainen and k. tuominen , jhep * 0905 * , 025 ( 2009 ) [ arxiv:0812.1467 [ hep - lat ] ] and a. j. hietanen , k. rummukainen and k. tuominen , arxiv:0904.0864 [ hep - lat ] .
a. hasenfratz , arxiv:0907.0919 [ hep - lat ] .
t. banks and a. zaks , nucl .
b * 196 * , 189 ( 1982 ) .
g. t. fleming , pos * lattice2008 * , 021 ( 2008 ) [ arxiv:0812.2035 [ hep - lat ] ] .
b. holdom , phys .
d * 24 * , 1441 ( 1981 ) and phys .
b * 150 * , 301 ( 1985 ) .
k. yamawaki , m. bando and k .- i .
matumoto , phys .
lett . *
56 * , 1335 ( 1986 ) .
t. appelquist and l. c. r. wijewardhana , phys .
d * 35 * , 774 ( 1987 ) and phys .
d * 36 * , 568 ( 1987 ) .
t. degrand and a. hasenfratz , arxiv:0906.1976 [ hep - lat ] .
s. aoki , phys .
d * 30 * , 2653 ( 1984 ) and s. aoki , phys .
* 57 * , 3136 ( 1986 ) .
s. aoki , prog .
suppl . * 122 * , 179 ( 1996 ) [ arxiv : hep - lat/9509008 ] .
f. farchioni _ et al .
_ , eur .
j. c * 39 * , 421 ( 2005 ) [ arxiv : hep - lat/0406039 ] . c. michael and c. urbach [ etm collaboration ] , pos * lat2007 * ( 2007 ) 122 [ arxiv:0709.4564 [ hep - lat ] ] .
boucaud _ et al . _
[ etm collaboration ] , comput .
commun .
* 179 * , 695 ( 2008 ) [ arxiv:0803.0224 [ hep - lat ] ] .
e. m. ilgenfritz , w. kerler , m. muller - preussker , a. sternbeck and h. stuben , phys .
d * 69 * , 074511 ( 2004 ) [ arxiv : hep - lat/0309057 ] .
f. farchioni _ et al .
_ , pos * lat2005 * , 033 ( 2006 ) [ arxiv : hep - lat/0509036 ] . v. azcoiti , g. di carlo and a. vaquero , phys . rev .
d * 79 * , 014509 ( 2009 ) [ arxiv:0809.2972 [ hep - lat ] ] .
s. r. sharpe , phys .
d * 79 * , 054503 ( 2009 ) [ arxiv:0811.0409 [ hep - lat ] ] .
k. g. wilson , in _ new phenomena in subnuclear physics _ ,
proceedings of the 1975 international school of subnuclear physics , erice , 1975 , edited by a. zichichi ( plenum , new york , 1977 ) .
n. kawamoto , nucl .
b * 190 * , 617 ( 1981 ) . | we explore aspects of the phase structure of @xmath0 and @xmath1 lattice gauge theories at strong coupling with many flavours @xmath2 of wilson fermions in the fundamental representation .
the pseudoscalar meson mass as a function of hopping parameter is observed to deviate from the expected analytic dependence , at least for sufficiently large @xmath2 .
implications of this effect are discussed , including the relevance to recent searches for an infrared fixed point .
kei - ichi nagai , georgina carrillo - ruiz , gergana koleva , randy lewis _ department of physics and astronomy , york university , toronto , canada , m3j 1p3 _ pacs numbers : 11.5.15.ha |
cardiac interbeat intervals normally fluctuate in a complex manner.@xmath6 recent studies reveal that under normal conditions , beat - to - beat fluctuations in the heart rate may display extended correlations of the type typically exhibited by dynamical systems far from equilibrium .
it has been argued,@xmath7 for example , that the various stages of sleep may be characterized by long - range correlations of heart rates separated by a large number of beats .
the interbeat fluctuations in the heart rates belong to a much broader class of many natural , as well as man - made , phenomena that are characterized by a degree of stochasticity .
turbulent flows , fluctuations in the stock market prices , seismic recordings , the internet traffic , and pressure fluctuations in packed - bed chemical reactors are example of time - dependent stochastic phenomena , while the surface roughness of many materials@xmath8 are examples of such phenomena that are length scale - dependent .
the focus of the present paper is on the intriguing statistical properties of interbeat interval sequences , the analysis of which has attracted the attention of researchers from different disciplines.@xmath9 analysis of heartbeat fluctuations focused initially on short - time oscillations associated with breathing , blood pressure and neuroautonomic control.@xmath10 studies of longer heartbeat records , however , revealed @xmath11like behavior.@xmath12 recent analysis of very long time series indicates that under healthy conditions , interbeat intervals may exhibit power - law anticorrelations,@xmath13 follow universal scaling in their distributions,@xmath14 and are characterized by a broad multifractal spectrum.@xmath15 such scaling features change with the disease and advanced age.@xmath16 the possible existence of scale - invariant properties in the seemingly noisy heartbeat fluctuations is may be attributed to highly complex , nonlinear mechanisms of physiological control,@xmath17 as it is known that circadian rhythms are associated with periodic changes in key physiological processes.@xmath18 in figure 1 samples of interbeats fluctuations of healthy subjects and those with congestive heart failure ( chf ) are shown .
= 9truecm recently , friedrich and peinke were able@xmath19 to derive a fokker - planck ( fp ) equation for describing the evolution of the probability distribution function of stochastic properties of turbulent free jets , in terms of the relevant length scale .
they pointed out that the conditional probability density of the _ increments _ of a stochastic field ( for example , the increments in the velocity field in turbulent flow ) satisfies the chapman - kolmogorov ( ck ) equation , even though the the velocity field itself contains long - range , nondecaying correlations .
as is well - known , satisfying the ck equation is a necessary condition for any fluctuating data to be a markovian process over the relevant length ( or time ) scales.@xmath20 hence , one has a way of analyzing stochastic phenomena in terms of the corresponding fp and ck equations . in this paper
the method proposed by friedrich and peinke is used to compute the kramers - moyal ( km ) coefficients for the _ increments _ of interbeat intervals fluctuatations , @xmath21 . here
, @xmath22 is the interbeat increments which , for all the samples , is defined as , @xmath23 , where @xmath24 is the standard deviations of the increments in the interbeats data .
it is shown that the first and second km coefficients representing , respectively , the drift and diffusion coefficients in the fp equation , have well - defined values , while the third- and fourth - order km coefficients are small .
therefore , a fp evolution equation@xmath20 is developed for the probability density function ( pdf ) @xmath25 which , in turn , is used to gain information on changing the shape of pdf as a function of the time scale @xmath26 ( see also ref . [ 37 ] for another interesting and carefully - analyzed example of the application of the ck equation to stochastic phenomena ) .
the plan of this paper is as follows . in section 2
we describe the friedrich - peinke method in terms of a km expansion and the fp equation .
we then apply the method in section 3 to the analysis of the increments in the interbeat fluctuations .
a complete characterization of the statistical properties of the interbeat fluctuation requires evaluation of the joint pdfs , @xmath27 , for an arbitrary @xmath28 , the number of data points .
if the phenomenon is a markov process , an important simplification arises in that , the @xmath28-point joint pdf @xmath29 is generated by the product of the conditional probabilities @xmath30 , for @xmath31 .
thus , as the first step of analyzing a stochastic time series , we check whether the increments in the data follow a markov chain .
as mentioned above , a necessary condition for a stochastic phenomenon to be a markov process is that the ck equation,@xmath19
@xmath32 should hold for any value of @xmath33 , in the interval @xmath34 therefore , we check the validity of the ck equation for describing the data using many values of the @xmath1 triplets , by comparing the directly - evaluated conditional probability distributions @xmath35 with those calculated according to right - hand side of eq .
( 1 ) . in fig .
2 , the directly - computed pdf is compared with the one obtained from eq . ( 1 ) . allowing for a statistical error of the order of the square root of the number of events in each bin , we find that the pdfs are statistically identical . = 9truecm it is well - known that the ck equation yields an evolution equation for the distribution function @xmath25 across the scales @xmath36 .
the ck equation , when formulated in differential form , yields a master equation , which takes on the form of a fp equation:@xmath20 @xmath37p(\delta x,\tau)\;.\end{aligned}\ ] ] the drift and diffusion coefficients , @xmath38 and @xmath39 , are estimated directly from the data and the moments @xmath40 of the conditional probability distributions : @xmath41 @xmath42 the coefficients @xmath43 are known as the kramers - moyal ( km ) coefficients . as an application of the method , we analyzed both daytime ( 12:00 pm to 18:00 pm ) and nighttime ( 12:00 am to 6:00 am ) heartbeat time series of healthy subjects , and the daytime records of patients with chf .
our data base includes 10 healthy subjects ( 7 females and 3 males with ages between 20 and 50 , and an average age of 34.3 years ) , and 12 subjects with chf , with 3 females and 9 males with ages between 22 and 71 , and an average age of 60.8 years ) .
the resulting drift and diffusion coefficients , @xmath44 and @xmath45 , are displayed in figures 3 and 4 .
it turns out that the drift coefficient @xmath44 is a linear function of @xmath22 , whereas the diffusivity @xmath45 is quadratic in @xmath22 .
estimates of these coefficients are less accurate for large values of @xmath22 and , thus , the uncertainties increase . using the data set for the healthy subjects we find that , = 9truecm = 9truecm @xmath46 whereas for the patients with chf we obtain , @xmath47 we also computed the _
average _ of the coefficients @xmath44 and @xmath45 for the entire set of the healthy subjects , as well as those with chf .
according to the pawulas theorem,@xmath48 the km expansion is truncated after the second term , provided that the fourth - order coefficient @xmath49 vanishes .
for the data that we analyze the coefficient @xmath50 is about @xmath51 for the healthy subjects , and about @xmath52 for those with chf .
equations ( 5 ) and ( 6 ) state that the drift coefficients for the healthy subjects and those with chf have the same order of magnitude , whereas the diffusion coefficients for given @xmath36 and @xmath22 are different by about one order of magnitude .
this points to a relatively simple way of distinguishing the two classes of the subjects .
moreover , the @xmath36-dependence of the diffusion coefficient for the healthy subjects is stronger than that of those with chf ( in the sense that the numerical coefficients of the @xmath53 are larger for the healthy subjects ) .
these are shown in figures 3 and 4 .
the strong @xmath54dependence of the diffusion coefficient @xmath45 for the healthy subjects indicates that the nature of the pdf of their increments @xmath22 for given @xmath36 , i.e. , @xmath25 , is intermittent , and that its shape should change strongly with @xmath36 .
however , for the subjects with chf the pdf is not so sensitive to the change of the time scale @xmath36 , hence indicating that the increments fluctuations for the subjects with chf is _ not _ intermittent .
these results are completely compatible with the recent discoveries that the interbeat fluctuations for healthy subjects and those with chf have fractal and multifractal properties , respectively.@xmath15
we have shown that the probability density of the interbeat interval increments satisfies a fokker - planck equation , which encodes the markovian nature of the increments fluctuations .
we have been able to compute reliably the first two kramers - moyal coefficients for the stochastic processes @xmath22 - the drift and diffusion coefficients in the fp representation - and , using the polynomial ansatz,@xmath19 obtain simple expressions for them in terms of @xmath22 and the time scale @xmath36 .
we have shown that the drift and diffusion coefficients of the increments in the interbeat fluctuations of healthy subjects and patients with chf have different behavior , when analyzed by the method we use in this paper .
hence , they help one to distinguish the two groups of the subjects . moreover
, one can obtain the form of the path probability functional of the increments in the interbeat intervals in the time scale , which naturally encodes the scale dependence of the probability density .
this , in turn , provides a clear physical picture of the intermittent nature of interbeat intervals fluctuations .
let us emphasize that the previous analysis@xmath6 of the data that we consider in this paper indicated that there may be long - range correlations in the data which might be characterized by self - affine fractal distributions , such as the fractional brownian motion or other types of stochastic processes that give rise to such correlations
. in that method one distinguishes healthy subjects from those with chf in terms of the _ type _ of the correlations that might exist in the data .
for example , if the data follow a fractional brownian motion , then the corresponding hurst exponent @xmath55 is used to distinguish the two classes of the subjects , as @xmath56 ( @xmath57 ) indicates negative ( positive ) correlations in the data , while @xmath58 indicates that the increments in the data follow brownian motion .
the method proposed in the present paper is different from such analyses in that , the _ increments _ in the data are analyzed in terms of markov processes .
this is _ not _ in contradiction with the previous analyses .
our analysis does indicate the existence of correlations in the increments , but , as is well - known in the theory of markov processes , such correlations , though extended , eventually decay .
we distinguish the healthy subjects from those with chf in terms of the _ differences _ between the drift and diffusion coefficients of the fokker - plank equation that we construct for the incremental data which , in our view , provides a clearer and more physical way of understanding the differences between the two groups of the subjects than the previous method .
lipsitz , j. mietus , g.b .
moody , and a.l .
goldberger , _ circulation _ * 81 * , 1803 ( 1990 ) ; d.t .
kaplan , _ et al .
_ , _ biophys
. j. _ * 59 * , 945 ( 1991 ) ; n. iyengar , _ et al .
_ , _ am . j. physiol . _ * 271 * , r1078 ( 1996 ) .
peng , s.v .
buldyrev , s. havlin , m. simons , h.e .
stanley , and a.l .
goldberger , _ phys .
e _ * 49 * 1685 ( 1994 ) ; c .- k .
peng , j.m .
hausdorff , and a.l .
goldberger , in _ nonlinear dynamics , self - organization , and biomedicine _ , edited by j. walleczek ( cambridge university press , cambridge , 1999 ) . | statistical properties of interbeat intervals cascade are evaluated by considering the joint probability distribution @xmath0 for two interbeat increments @xmath1 and @xmath2 of different time scales @xmath3 and @xmath4 .
we present evidence that the conditional probability distribution @xmath5 may obey a chapman - kolmogorov equation .
the corresponding kramers - moyal ( km ) coefficients are evaluated .
it is shown that while the first and second km coefficients , i.e. , the drift and diffusion coefficients , take on well - defined and significant values , the higher - order coefficients in the km expansion are very small . as a result
, the joint probability distributions of the increments in the interbeat intervals obey a fokker - planck equation .
the method provides a novel technique for distinguishing the two classes of subjects in terms of the drift and diffusion coefficients , which behave differently for two classes of the subjects , namely , healthy subjects and those with congestive heart failure .
= 2 17 cm 112 cm = 2 ] |
the relative roles of neutrino heating and bipolar mhd jets as the underlying mechanism behind core - collapse supernovae ( sne ) are controversial @xcite . however , despite current uncertainties about the specific engine that drives core - collapse explosions , a variety of observations and hydrodynamic modeling make a compelling case that high - mass sne are intrinsically aspherical events .
observations of extragalactic core - collapse sne show increasing late - time linear polarization levels suggesting that the innermost layers driving the sn expansion are aspherical @xcite . in the case of sn 1987a ,
spectropolarimetric observations and the early detection of gamma - rays and hard x - rays indicating the transport of freshly synthesized @xmath17ni from the core to the h - rich envelope lead to a model of envelope asymmetries and ejecta fragmentation @xcite .
recent _ hubble space telescope _
( _ hst _ ) images of sn 1987a reveal an elongated , axially symmetric remnant @xcite .
state - of - the - art numerical simulations also show that spherically symmetric core - collapses do not yield successful explosions @xcite .
this has lead some modelers to investigate the effects of rapid rotation and magnetic fields leading to magnetorotational jet models @xcite .
others have investigated asymmetric neutrino - driven models in order to generate aspherical sometimes even jet - like sn explosions @xcite .
asymmetries in either the neutrino heating or mhd jets might also explain pulsar ` kick ' velocities @xcite .
observations of ` long - duration ' gamma - ray bursts ( grbs ) suggest highly aspherical core - collapse explosions . in the popular ` collapsar ' grb model
, a high mass core - collapse sn creates a black hole generating bipolar relativistic jets @xcite .
these jets excite an external shock from which energetic electrons , created through interaction with the ambient medium , radiate synchrotron emission giving rise to broadband afterglows .
direct connections between sne and grbs include the observational coincidences between grbs and sn 1998bw and sn 2003dh , ` bumps ' in some grb afterglow light - curves consistent with underlying sn explosions @xcite , and indications that grbs are related to star forming regions @xcite . compared to sn observations , evidence for aspherical sn explosions based on supernova remnant observations is much less clear .
one of the most often cited asymmetrically expanding snrs is that of cassiopeia a ( cas a ) , currently the youngest known ( @xmath18300 yr ) galactic , core - collapse remnant . on even the earliest photographic plate images , a ` flare ' or ` jet ' of knots and filaments could be seen extending out along the northeast limb about @xmath19 from the remnant center at a pa @xmath18 70@xmath20 @xcite . whereas the main @xmath82@xmath21 radius emission ring of ejecta expands at velocities @xmath22 km s@xmath2 , debris in the ne jet have estimated velocities more than twice as large , up to 14000 km s@xmath2 @xcite . however , both the main shell s so - called ` fast - moving knots ' ( fmks ) and ne jet ejecta knots have similar emission line spectra ; namely , strong lines of [ s ii ] @xmath236716,6731 , suggesting similar chemical make - ups . a southwest ` counterjet ' of high - velocity , [ s ii ] emitting ejecta was recently discovered @xcite and confirmed in x - rays and the infrared @xcite . while the presence of a jet and counterjet in this high - mass progenitor snr might indicate an aspherical sn expansion , the nature of these jet features is controversial .
proposed explanations include uneven ejecta decelerations due to local ism density variations and cavities @xcite , bipolar mhd jets @xcite , or mildly asymmetrical neutrino - driven explosions @xcite .
complicating matters , the ne and sw jets are not the only outlying ejecta around cas a , with several dozen high - velocity , nitrogen - rich knots located in other regions of the remnant @xcite . in this paper ,
we report results from a deep _ hst _ imaging survey of the cas a snr .
these images reveal a large population of high - velocity knots of ejecta with an asymmetric distribution .
a complete catalog of outer ejecta knots will be presented in a separate paper @xcite . here
we address the overall spatial distribution and expansion velocity of the remnant s outer ejecta and discuss limits on the dynamical age of the remnant .
the observations and knot flux measurement procedures are described in @xmath242 and @xmath243 , with the results presented and discussed in @xmath244 and @xmath245 , respectively .
high resolution _ hst _ images of the cas a remnant were obtained on 46 march 2004 and 45 december 2004 using the wide field channel ( wfc ) of the advanced camera for surveys ( acs ; @xcite ) .
the acs / wfc consists of two @xmath25 ccds with an average image pixel scale of @xmath26 providing a @xmath27 field of view .
four 2-point line dithered images were taken in each of the four acs / wfc sloan digital sky survey ( sdss ) filters , namely f450w , f625w , f775w , and f850lp ( i.e. , sdss g , r , i , and z ) , at each target position to permit cosmic ray removal , coverage of the @xmath28 interchip gap , and to minimize saturation effects of bright stars in the target fields .
total integration times in the f450w , f625w , f775w , and f850lp filters were 2000 s , 2400 s , 2000 s , and 2000 s , respectively .
standard pipeline iraf / stsdas data reduction was done including debiasing , flat - fielding , geometric distortion corrections , photometric calibrations , and cosmic ray and hot pixel removal .
the stsdas _ drizzle _ task was used to combine exposures in each filter . due to significant reddening toward cas a ( a@xmath29 = @xmath30 mag ; @xcite ) , [ o iii ] @xmath234959,5007 line emission was too weak to be detected for most outlying knots and we have not included f450w images in our analysis .
we measured outer knot acs / wfc fluxes from the three remaining sdss filter image sets using the automated source extraction software package sextractor @xcite . in cases where the sextractor program failed to return a reasonable flux or failed to return a flux at all ,
the knot fluxes were calculated by hand . in all cases ,
the fluxes were calculated using 5 pixel apertures .
background estimates were performed by sextractor using a @xmath31 pixel rectangular annulus about the isophotal limits of the object .
when fluxes were calculated manually , background estimation was performed by calculating the total @xmath32 pixel aperture flux in at least five positions near the object ( avoiding other sources ) and then subtracting the mean computed `` background '' flux from the total object pixel sum .
most knots whose fluxes required manual computation were located near a bright background source or very close to another ejecta knot .
high - velocity , outer ejecta knots were identified through proper motion measurements on the march and december 2004 acs / wfc images @xcite .
ejecta knots were defined as being high - velocity if their epoch 2004.3 radial distance exceeded 100@xmath33 from the remnant s center of expansion ( v@xmath34 5500 km s@xmath2 ) and they showed a proper motion @xmath5
@xmath35 yr@xmath2 ( see @xcite ) .
following @xcite and @xcite , we used the filter fluxes to bin outlying ejecta knots into three emission classes ; namely , strong [ n ii ] @xmath236548,6583 emission knots , strong [ o ii ] @xmath237319,7330 emission knots , and strong [ s ii ] @xmath236716,6731 fmk - like knots .
flux ratio criteria between these three classes were chosen to segregate knots with similar ratios seen for main shell or outer ejecta knots with existing spectroscopic data .
for example , outlying ejecta knots whose @xmath36 spectra show largely just [ n ii ] @xmath236548,6583 emission like those discussed by @xcite exhibit f625w / f775w and f625w / f850lp ratios more than an order of magnitude larger than [ s ii ] bright knots .
in similar fashion , the newly discovered outlying o - rich knots which show a 6000 - 7500 spectrum with strong [ o i ] and [ o ii ] line emissions @xcite , exhibit the o / s sensitive f775w / f850lp ratio several times greater than even strong [ o ii ] @xmath237319,7330 emission main shell knots . specifically , we chose a flux ratio for f775w/(f625w + f850lp ) @xmath5 1.0 to separate out the [ o ii ] strong fmks from the [ s ii ] strong fmks ; that is , those knots with stronger [ o ii ] @xmath237319,7330 emission detected via the f775w filter than the combined strength of [ n ii ] , [ s iii ] and [ s ii ] emissions detected in the f625w and f850lp filters .
similarly , knots with strong [ n ii ] emissions were selected via f625w/(f775w + f850lp ) @xmath5 1.0 , thereby selecting those knots where the combined flux of [ o i ] @xmath236300,6364 , [ s ii ] @xmath236716,6731 , and [ n ii ] @xmath236548,6583 emissions was greater than the sum of f775w flux , due mostly to [ o ii ] , and f850lp flux sensitive to the near - infrared [ s iii ] and [ s ii ] emissions . since [ o i ] @xmath236300,6364 flux rarely , if ever , exceeds the [ o ii ] @xmath237319,7330 flux , and the observed [ s ii ] @xmath236716,6731 emission is unlikely to ever exceed the combined flux of [ s iii ] @xmath239069,9531 plus [ s ii ] @xmath231028710370 line emissions @xcite , then any knot for which the f625w/(f775w + f850lp ) @xmath5 1.0 requires the presence of significant [ n ii ] @xmath236548,6583 emission .
examination of march and december 2004 acs / wfc images revealed a total of 1825 high proper motion ( @xmath37 yr@xmath2 ) ejecta knots around the cas a remnant @xcite .
this is a much larger population of high - velocity ejecta knots than previously identified or suspected from ground - based images @xcite .
a total of @xmath38 strong [ n ii ] emission , @xmath39 strong [ o ii ] emission , and @xmath40 fmk - like knots were identified .
although nearly half of the 1825 cataloged outer knots were found in the ne jet , high - velocity outlying ejecta were identified in many other regions around the remnant .
outlying optical ejecta knots range from 105@xmath33 to 300@xmath33 in radial distance from the center of expansion , placing them in projection close to or outside the remnant s @xmath8 6000 km s@xmath2 forward shock front as determined by the remnant s outermost x - ray emission @xcite .
the location and distribution of these knots with respect to the remnant s outer x - ray emission filaments associated with the forward shock front can be seen in figure 1 .
here we show the locations of all 1825 cataloged outer knots projected onto the 1 msec _
acis image ( epoch 2004.3 ; @xcite ) .
knot positions are marked with open circles color coded either red , green , or blue to indicate those knots with spectra dominated by strong [ n ii ] @xmath236548,6583 , [ o ii ] @xmath237319,7330 , or [ s ii ] @xmath236716,6731 line emissions , respectively . the transverse expansion velocities of cas a s outermost ejecta ( fig .
1 ) show a strongly aspherical structure due principally to the ne and sw jets which appear as distinct and roughly opposing features ; ne jet : p.a .
= @xmath41 , sw jet : p.a .
= @xmath7 .
the jets have similar maximum radial distances and proper motion derived expansion velocities ; namely @xmath42 and @xmath43 km s@xmath2 .
they also contain the remnant s highest velocity [ s ii ] emitting ejecta ( the blue open circles in fig . 1 ) .
this is in contrast to other regions where the nitrogen - rich ejecta ( open red circles ) represent the remnant s highest velocity debris ( @xmath9 km s@xmath2 ) followed by the o - rich ejecta , and then the s - rich knots @xcite . examination of the _ hst _ images also revealed a lack of outlying ejecta knots along the remnant s northern and southern regions .
not a single ejecta knot could be found along a 55@xmath44 wide region in position angle along the remnant s southern limb ( i.e. , pa = @xmath45 ) and in a narrower 15@xmath44 position angle zone along the north ( pa = @xmath46 ) .
the northern gap would actually be larger if one only considered knots with radial distances greater than @xmath47 in which case the northern gap size grows to @xmath48 in position angle ( pa = @xmath49 ) .
we note that , unlike for the northern limb region , acs images did not completely cover the whole southern limb region imaging only out to a distance of @xmath47 due south of the center of expansion .
however , inspection of ground - based images covering regions farther to the south ( @xmath50 ) revealed no bright farther outlying knots .
thus , despite missing acs / wfc coverage along the south , the lack of any knots at smaller distances like those seen along the east and west edges of the southern gap ( r = @xmath51 ) indicate a strong likelihood that there are simply no bright , high - velocity ejecta knots in either of the nnw and sse directions .
this ne sw jet / counterjet and nnw sse gap asymmetry in the distribution of the remnant s outer optical ejecta knots can be clearly seen in the proper motion plots of figure 2 . in the top panel ,
we plot the extrapolated 320 yr proper motion paths for the 1825 identified outer knots based on measured march to december 2004 positions .
the central white circle has a radius of @xmath52 centered on the remnant s estimated center of expansion ( coe ; @xmath53(j2000 ) = @xmath54 , @xmath55(j2000 ) = @xmath56 ; @xcite ) .
although errors in derived knot proper motions lead to some knot trajectories missing the expansion center by substantial distances filling in somewhat the northern and southern gaps , a strongly aspherical distribution of ejecta jets and gaps is readily apparent .
when the proper motion extrapolations are replaced with predicted knot proper motions based solely on march 2004 knot positions and the coe , a structure of opposing jet features and north and south gaps appears even more striking ( fig . 2 , bottom panel ) .
the circle in the figure marks a radial distance of @xmath57 from the coe corresponding to a proper motion of @xmath58 yr@xmath2 for an age of 320 years and an implied @xmath810,000 km s@xmath2 transverse velocity assuming a remnant distance of 3.4 kpc .
as can be seen from the figure , this circle encompasses nearly all the remnant s n - rich outer ejecta away from the jet regions .
peak ne and sw jet optical knot transverse velocities lie at pa = @xmath59 and at pa = @xmath60 , respectively , i.e. , @xmath8180@xmath44 apart .
these position angles are close to those seen also in the x - rays and infrared for the ne and sw jets .
that is , from the 1 ms _ chandra _ @xcite and 24 micron _
images @xcite , we find pa = 63@xmath4465@xmath44 for the center of the ne jet and pa = 246@xmath44248@xmath44 for the center of the sw jet , some @xmath61 apart . while the significance of the northern and southern gaps in the distribution of the outer knots is uncertain ( see below ) , they are situated roughly orthogonal to this ne sw jet alignment line
. the middles of the southern and northern outer knot gaps lie at pa = @xmath62 and @xmath63 respectively , some 107@xmath44 and 99@xmath44 off from a 63@xmath44243@xmath44 alignment line .
lastly , extrapolated late 17th century knot positions relative to the remnant s estimated coe indicates some evidence for non - uniform deceleration around the remnant .
figure 3 ( left panel ) shows the estimated positions for all 1825 outer knots for the @xcite estimated convergent year 1671 .
the estimated coe is marked by the circle ( radius = @xmath52 ) .
here , one sees that the scatter of points for the outer knots is significantly away from the @xcite center , centered instead @xmath64 to the east .
( note : a check using the same 17 outer knots employed by @xcite using the new acs images together with archival images agree with their derived coe . )
one possible cause for this shift eastward is a greater deceleration of knots along different regions around the remnant , particularly along the eastern limb , giving rise to broader and slower average expansion velocities for a given radial distance and look - back time .
a different shift off from the coe is seen when plotting just northern and southern knots ( fig . 3 ; right panel ) . for 416 north and south knots ,
the shift from the coe shows an even greater displacement , namely @xmath65 to the southeast .
although optical debris may constitute only a small fraction of the total ejected mass , cas a s fastest moving material is perhaps best studied optically . for example
, optical emission from the ne and sw jets can be traced about @xmath66 farther out than in x - rays or infrared @xcite and only a handful of outer optical ejecta knots around the rest of the remnant are detected in even the deepest radio or x - ray images .
fragmentation of sn ejecta into dense clumps may come about by rayleigh - taylor instabilities brought on by the deceleration of the ejecta by the ambient medium @xcite , passage of the reverse shock @xcite , or inside the progenitor during the sn explosion @xcite .
optical emission will arise when @xmath18100 km s@xmath2 internal shocks are formed in the clumps driven by the high stagnation pressure behind the clump s bow shock .
@xcite studied similarity solutions for sn blast waves driven by clumpy ejecta and found that such clumps will eventually move ahead of a remnant s forward shock front .
he concluded that this process may be a generic feature of snrs with clumped ejecta .
the presence of numerous dense and relatively cool optically emitting , high - velocity ejecta knots out ahead of cas a s forward shock front supports hamilton s basic results .
namely , we find a significant number of optical outer ejecta knots lying ( in projection ) outside the faint x - ray emitting filaments marking the current location of the remnant s forward blast wave @xcite .
asymmetries in cas a s outermost debris may offer clues as to the nature of the sn explosion engine .
what one observes at a particular epoch , however , may not necessarily reflect the sn s true explosion dynamics .
ejecta velocities will be modified ( to varying degrees ) by passage through the ambient medium as ejecta knots are only made optically visible due to this interaction , and more knots will be visible in regions where there is more circumstellar material .
the detected distribution can also change with time since some 1020% of outer optical knots vary significantly in brightness over just a few years due presumably to variable interaction with a clumpy csm @xcite .
in addition , for a knot to be visible it must be sufficiently large and dense to generate detectable forbidden line emission .
it must also survive disruption via mass ablation from kelvin - helmhotz instabilities during passage through the reverse shock , forward shock , and the surrounding ambient medium .
consequently , what one sees can be a distorted and biased view of the true distribution of a remnant s outer debris .
nonetheless , even with all these caveats , the asymmetry and non - uniformity of the remnant s outermost optical ejecta as seen in the plots of figure 2 is striking .
the ne and sw jets extend out to nearly @xmath67 , more than twice the radius of the remnant s main emission shell , and lie on virtually opposite sides of the remnant s center of expansion .
the jets also appear to lie roughly in the plane of the sky to within @xmath68 @xcite .
this combination of an opposing , sulfur - rich jet / counterjet structure with similar maximum expansion velocities plus gaps of high - velocity ejecta aligned roughly perpendicular is suggestive of an aspherical , bipolar expansion . from 2-d models , @xcite calculated that a progenitor s axisymmetric wind structure , where the highest density lay in the equatorial plane , could generate bipolar protrusions of sn ejecta extending @xmath69 times the radius of the main shell .
the relative scale of such expansion asymmetry is not unlike that seen in the cas a jets and only a mild asymmetry ( @xmath70/@xmath71 @xmath18 2 ) is needed to form obvious protrusions . while significant heavy element material can end up at much larger radial distances than elsewhere , the original progenitor chemical layering should largely persist within the protrusion .
that is , the highest velocity material should mainly exhibit the chemical abundances of the progenitor s outermost layers , namely n , he , and o. however , just the opposite is observed .
the farther out one goes in either of cas a s jets , the weaker oxygen emission lines become relative to those of sulfur @xcite . the fact that one sees largely undecelerated ejecta throughout the remnant with @xmath72 ( see @xcite ) indicates that the observed compositional structure is set early in the explosion .
in addition , outside of the jet regions , the s - rich material has a much more limited velocity range .
all this suggests that the s - rich ne jet , rather than reflecting some sort of turbulent mixing region or a rapid expansion into surrounding ism / csm due to a lower surrounding density , more likely represents a true stream of high - speed ejecta formed when underlying material was ejected up through the star s outer layers .
other observational evidence in favor of an ejection of underlying material comes from the location of the so - called ` mixed ejecta knots ' and from energy estimates in the jet region .
mixed ejecta knots show both nitrogen and sulfur overabundances , which could have resulted from turbulent mixing of h and n - rich layers with underlying s and o - rich layers , are only seen in the jet regions @xcite .
the acs / wfc images show that many of the mixed knots are not the result of line - of - sight superposition of knots with different chemical make - up but appear to be ejecta with mixed chemical properties @xcite .
in addition , from an analysis of _ chandra _ x - ray data , @xcite found a shallower outer envelope near the base of the ne jet which they interpreted as indicating more of the initial explosion s energy being directed in this polar direction as opposed to equatorial . on the other hand , however , while the prominent shell rupture near the base of the ne jet seen in x - rays and radio images might be seen as supporting evidence for this picture @xcite , there is no obvious main shell rupture near the sw counterjet .
although our _ hst _ survey of the highest - velocity ejecta in cas a makes a compelling case for a high - velocity bipolar expansion , what the jets and gaps are telling us about the underlying core - collapse mechanism is unclear .
currently , there is no consensus as to the details of the explosion process for core - collapse sne except that models with completely spherical neutrino heating mechanisms fail to yield successful explosions @xcite .
this has led to considerable effort to understanding the effects of rotation on neutrino heating and the importance of magnetic fields in the proto - neutron star . around the rotation axis of a collapsing star
, accretion flow will be non - spherical and accelerated thereby lowering the needed critical neutrino luminosity @xcite .
the degree of neutrino re - vitalization of the shock front created by the core - bounce may also be affected by convection in the proto - neutron star through the mechanical outward transfer of neutrinos .
investigations into the effects of rotation on the anisotropic neutrino emission from the proto - neutron star indicate a weakening of the core bounce that seeds the neutrino - drive convection , with angular momentum tending to stabilize the core , constraining convection to the polar regions @xcite .
models of anisotropic neutrino radiation indicate more powerful explosions are generated , which then lead to a prolate expansion geometry @xcite . @xcite
found the bipolar expansion was not strongly collimated ( @xmath73 ) unless the rotation rate was large . for cas a , @xcite favored a rotation enhanced , neutrino - driven model in which the x - ray observed fe - rich se and nw regions , and not the ne and sw jet / counterjet , mark the progenitor s rotation axis . instead of driving the explosion , the ne sw jets would have formed following the neutrino - driven main explosion via an under - energetic jet - like ejection created by an mhd jet or proto - neutron star wind ( possibly associated with accretion by the ns ; @xcite ) emerging into an already expanding debris field . but
this model requires a nearly 90@xmath44 post - explosion precession of the ns s rotation axis from nw se to ne
sw @xcite , the cause of which is left unexplained . on the other hand , models of magnetorotationally induced jets capable of generating sn explosions have also been invoked to explain the cas a jets @xcite . in this view
, rotation leads to magnetic field amplification thereby generating non - relativistic axial jets of mhd energy @xmath74 erg which then initiate a bipolar supernova explosion .
this is somewhat analogous to the narrow relativistic jets proposed in the collapsar model as the central engine for grbs .
@xcite questioned the accuracy of anisotropic jet explosion simulations and @xcite argue that if the cas a jets were driven by outflowing core material they should be fe - rich instead of the s and si rich material observed .
if cas a s ne and sw jets were driven by ni - rich material , the energy deposited by radioactive @xmath17ni decay might have created hot , low density ni - rich bubbles .
this could make fe - rich material in the jets today too cool for detection in x - rays while also being too diffuse and low density to detect as optically bright knots .
furthermore , while both the ne and sw jets can be traced farthest out in the optical , optical studies have never detected appreciable fe - rich material anywhere in cas a. optical fe line emissions are weak and hard to detect in even the brightest main shell knots and there is no optical fe - rich material seen corresponding to the fe - rich se and n regions observed in the x - rays @xcite .
morphologically , both mhd jet and neutrino - driven expansion models produce aspherical ` jets ' with axial expansion ratios around 2 like those seen for cas a @xcite .
although much of the ne jet s central and farthest extending line of optical filaments lie virtually in the plane of the sky @xcite , knots lying at a projected radial distance of 150170 arcsec out from the coe show radial velocities from @xmath75 to @xmath76 km s@xmath2 , indicating an expansion cone about @xmath77 wide @xcite .
this means that the ne jet is more like a fan of several streams of ejecta knots ( rather than a single narrow jet ) , about as deep as it appears wide , i.e. , about @xmath78 .
this is consistent with their appearance on x - ray and infrared images which show both jets as two or three ejecta streams rather than one single narrowly focused line of emission .
for example , there are three main optical lines of knots in the ne jet which correlate with x - ray and ir emission fingers . although the sw jet is optically much less well - defined , the locations of the outermost [ s ii ] emitting ejecta correlate reasonably well with the x - ray and ir emission ` fingers ' . beside the jets , a possible additional clue as to the nature of the central core - collapse sn engine
may be the absence of high - speed ejecta along the north and south limbs .
not all proposed core - collapse models show a pronounced decrease in outermost expansion velocity suggested by the nearly opposing ejecta gaps revealed in the _ hst _ data ( fig . 2 ) .
however , similar gaps in the distribution of slower moving ejecta are not found in the remnant s main emission ring of reverse shocked debris based on radio , x - ray , and optical maps .
while the remnant s forward shock shows no decrease in these directions as one might expect if the expansion velocity was significantly lower in these northern and southern gap regions @xcite , a gap in the forward shock front would be rapidly filled by the blast wave advancing in from the sides .
thus , while the north and south ejecta gaps are interesting especially given their positions relative to the jets , their meaning is presently unclear .
nevertheless , if these gaps are truly found to be areas devoid of high - speed ejecta seen elsewhere around the remnant and orthogonal to the ne / sw jets they may provide some insight for testing core - collapse models . first - light images of cas a taken by the _
chandra _ x - ray observatory revealed a central x - ray point source ( xps ) in the remnant @xcite .
although its nature is uncertain , this object is likely to be a neutron star but not a pulsar due to a lack of radio pulsations , no detected x - ray or radio plerion , and an x - ray spectrum too steep for an ordinary pulsar @xcite .
it has been suggested that it may be a younger and less luminous example of a subclass of neutron stars known as anomalous x - ray pulsars ( axps ) and soft gamma - ray repeaters ( sgrs ) @xcite .
x - ray emission bursts from axps and sgrs together with their spin - down rates have been explained by a magnetar model in which a neutron star has a much higher surface magnetic field of @xmath79 g than ordinary pulsars @xcite . except for a lack of pulsations , the general properties of the cas a xps and other x - ray emitting but radio quiet compact central objects in fairly young snrs are not all that dissimilar from axps and sgrs @xcite .
core - collapse asymmetries in the sn explosion might help explain the inferred high space velocities ( @xmath80 km s@xmath2 or more ) for neutron stars and radio pulsars @xcite .
although 3d core - collapse models suggest neutrino asymmetries as well as disruption of binaries by symmetric explosions may be insufficient to generate the wide range of observed ` kick ' pulsar velocities @xcite , mhd driven explosions creating unbalanced jets just might @xcite . however , the motion of the remnant s xps is far from being in alignment with cas a s ne sw jets . using the initial _ chandra _ derived position along with subsequent positional measurements using archival _
rosat _ and _ einstein _ image data @xcite , @xcite estimated a @xmath81 displacement of the xps from their derived center of expansion . assuming a common origin for the xps and the expanding ejecta , they estimated a transverse velocity of @xmath82330 km s@xmath2 for a distance of 3.4 kpc . adopting an updated position of the xps ( 2004 epoch ; @xcite ) , the displacement of the xps from the @xcite remnant expansion center is @xmath83 with an implied motion in a southeasterly direction ( position angle = @xmath84 ;
see fig . 4 , left panel ) . assuming a current remnant age @xmath8 325 yr ( see discussion in @xmath245.4 )
, we find a slightly higher transverse velocity of around 350 km s@xmath2 for a distance of 3.4 kpc .
the apparent southernly direction of motion for the xps is roughly orthogonal to the jet - counterjet alignment line , making an asymmetric jet - induced ` kick ' explanation problematic .
interestingly however , the projected motion of the xps is toward the middle of the broad southern gap in the distribution of the outer ejecta knots ( fig . 4 ; right panel ) .
if the distribution of the optically emitting outer ejecta knots is giving us a true picture of the asymmetry in the cas a supernova expansion , then the neutron star s preference to move in a direction lacking in high velocity material may indicate a natal kick aligned with the progenitor s slowest velocity expansion , possibly the progenitor s equatorial region .
@xcite and @xcite have proposed that magnetars could receive natal kicks in part due to their intense magnetic fields which could lead to anisotropic neutrino emissions . however , unlike the case seen here for cas a , these kicks would be in the direction of the rotation axis which the ne sw jets would seem to mark . while some misalignments between the progenitor s spin axis and a neutron star s velocity vector have been reported @xcite , including the high velocity pulsar b1508 + 55 ( v@xmath85 @xmath8 1100 km s@xmath2 ; @xcite )
, there is not strong observational evidence for general misalignments @xcite .
moreover , although some binary disruption type models have been proposed to explain natal kicks perpendicular to the spin axis ( e.g. , @xcite ) , no clear picture has emerged on how such a misalignment would be produced or even whether such models apply to the case of cas a. although only nine months separated the two sets of acs / wfc images , the large number of outer knots identified ( 1825 ) permitted us to estimate the dynamical age of the cas a snr . assuming the coe derived by @xcite and no knot deceleration since the time of the explosion , we show in figure 5 ( left panel ) the estimated arrival date of the 1825 cataloged outlying ejecta knots to within the minimum least squared distance to the coe plotted versus cataloged knot identification number @xcite . in this figure ,
symbol size is inversely proportional to estimated proper motion uncertainty .
knot catalog ids are in order of increasing position angle with the ne jet knots ( i d numbers @xmath18 100 950 ) making up a substantial fraction ( @xmath1845% ) of the 1825 cataloged knots .
the dispersion in estimated knot convergent dates is not uniform as a function of position angle .
for example , for the middle section of the ne jet where many of the fastest ejecta are found ( i.e. , knot ids @xmath86 ) , a decrease in the range of estimated knot arrival dates can be seen in figure 5 ( left panel ) .
this decrease reflects more accurate proper motion values due mainly to the larger radial distances for jet knots from the coe and consequently larger proper motion values leading to smaller percentage measurement errors .
the average arrival date for the 1825 outer knots with undecelerated extrapolated arrival dates between 1580 and 1750 is 1662@xmath8727 yr .
this is consistent with that estimated by @xcite who found an undecelerated convergent date of [email protected] based on a sample of 17 especially long - lasting knots for which archival imaging data were available covering a time span of up to 50 years .
because the catalog of 1825 outer knots covered a wide range of sizes and brightnesses leading to a range of proper motion measurement errors , we examined a much smaller , hand - selected sample of 72 knots which are : i ) relatively bright , and ii ) compact in size or unresolved in the acs / wfc images .
this sample included the 17 outer knots used by @xcite for their estimated convergent date .
the results for this smaller sample , shown in figure 5 ( right panel ) , indicate a convergent date @xmath88 , in excellent agreement with the thorstensen et al . 1671 date , shown here as a vertical dashed line .
while ejecta knots must undergo shock heating and hence some deceleration in order to be optically visible , the least decelerated ones offer stronger upper limits to the remnant s age .
as seen in figure 5 , the region near the top of the left - hand plot , i.e. , knot ids from @xmath89 corresponding to position angles @xmath90 respectively , show displacements toward a convergent date later than 1671 , namely @xmath91 .
these 126 knots are located along the remnant s northwest limb and can be seen in the bottom panel of figure 2 as cluster of nw knots .
if we knew the degree of deceleration these northwestern limb knots might have experienced over the @xmath18300 yr age of the remnant , it would give us a better estimate of the remnant s age and therefore the actual cas a sn explosion date .
for example , if knot decelerations were significant then one might be able to rule out on dynamical grounds a proposed but controversial sighting of the cas a sn by rev .
john flamsteed in august 1680 @xcite
. a knot will undergo deceleration both from the direct interaction with local gas as well as from the internal shock driven into the knot that gives rise to the optical emission observed @xcite .
if treated as a dense undistorted clump , a knot s deceleration due to drag from its interaction with the ambient medium depends on knot velocity and mass , the density of the local medium , and the cross - sectional area of the knot s bow shock which for hypersonic conditions is approximately equal to the knot itself @xcite .
the timescale for knot deceleration ( drag ) is given by @xmath92 @xmath18 @xmath93/@xmath94 where @xmath95 is the density contrast between the knot and the ambient medium ( i.e. , @xmath96/@xmath97 ) , @xmath98 is the knot s radius , and @xmath94 is the knot s velocity @xcite .
based on our acs imaging data , typical outer knots have velocities of 10,000 km s@xmath2 and sizes r @xmath99 corresponding to 0.002 pc at 3.4 kpc .
high - velocity outlying [ s ii ] emitting knot electron densities lie between 2000 16,000 @xmath100 with typical values between 4000 10,000 @xmath100 @xcite . the ambient density around cas
a is not well determined but is estimated to range between 0.4 3.7 @xmath100 @xcite .
choosing a @xmath95 = 10@xmath101 and an outer ejecta knot velocity of 10,000 km s@xmath2 leads to @xmath92 @xmath18 2000 yr , suggesting that outer knot deceleration due to drag may be fairly small at present .
this conclusion is consistent with a lack of detectable knot deceleration over the last 50 yr and a velocity change of @xmath102 5% over 300 yr @xcite .
while model simulations suggest that shocked clumps become flattened ( i.e. , laterally spread ) which would increase their cross section and hence enhance the deceleration , radiative cooling might partially counteract this effect leading to clump breakup into a cluster of smaller , denser knots .
cloud - ism interaction models suggest knot disruption might also occur due to rayleigh - taylor and kelvin - helmholtz instabilities , again resulting in the generation of smaller , dense knot fragments @xcite .
the timescale for initial knot breakup under these conditions , @xmath103 , is uncertain but is likely to be a few ` cloud crushing times ' @xcite or @xmath103 @xmath18 4 @xmath104/@xmath94 . for the knot numbers assumed above
, the disruption timescale is @xmath18 50 yr , meaning the remnant s highest - velocity ejecta clumps might well have already undergone breakup into denser knots .
thus , both radiative cooling effects and dynamical instabilities might help explain the observed small - scale clustering of some of the remnant s outer ejecta knots ( see @xcite ) .
knot deceleration from internal shock passage will cause the largest error in convergent date if it has occurred recently , and given the @xmath105 year lifetimes of most optical knots in cas a @xcite , this is probably the case .
the effective shock deceleration is equal to the internal knot shock velocity for post - shock emission @xcite .
the shock speed must be at least 30 @xmath106 to produce significant [ n ii ] , [ s ii ] and [ o ii ] emissions ( e.g .. @xcite ) . based on the blair et al .
( 2000 ) shock models , shock speeds above about 500 @xmath107 will ionize the oxygen to the helium - like ionization state and higher , reducing the cooling rate by an order of magnitude .
if the outer portions of the snr are roughly in pressure equilibrium , the density is proportional to v@xmath108 , further reducing the cooling rate behind fast shocks .
consequently , shocks above about 500 @xmath106 will have cooling times longer than the age of cas a , and will appear as x - ray rather than optical emission features .
thus the observed knots have been decelerated by 30 to 500 @xmath107 , or 0.3 to 5% of their @xmath1810,000 km s@xmath2 apparent speeds .
this corresponds to an error in convergence time of 0.003 to 0.05 times 320 yr , or 1 to 15 yr , well within the 19 yr uncertainty in the convergence time of the least decelerated northern limb knots .
we conclude that our convergence times estimates are not likely to be significantly affected by knot deceleration . therefore , while the seemingly less decelerated knots located along the remnant s northern limb suggest an explosion date somewhat later than 1670 , overall the measurements are still consistent with a possible sighting of the cas a sn in 1680 .
follow - up acs imaging obtain in a few years may be able to settle this issue more definitely through a firmer estimate of the date for the cas a supernova outburst .
this work was supported by nasa through grants go-8281 , go-9238 , go-9890 , and go-10286 to raf and jm from the space telescope science institute , which is operated by the association of universities for research in astronomy .
rac is supported by nsf grant ast-0307366 and clg is supported through uk pparc grant ppa / g / s/2003/00040 .
akiyama , s. , wheeler , j. c. , meier , d. l. , & lichtenstadt , i. 2003 , , 584 , 954 arnett , w. d. , fryxell , b. a. , & muller , e. 1989 , , 341 , l63 arras , p. , & lai , d. 1999 , , 519 , 745 aschenbach , b. 1999 , iau circ .
7249 ashworth , w. b. 1980 , j. hist .
, 11 , 1 bertin , e. , & arnouts , s. 1996 , , 117 , 393 blair , w. p. , et al .
2000 , , 538 , l61 blondin , j. m. , lundqvist , p. , & chevalier , r. a. 1996 , , 472 , 257 blondin , m. j. , featherstone , n. , borkowski , j. k. , & reynolds , p. s. 2001 , two years of science with chandra symposium , washington , dc , september 2001 , p166 braun , r. 1987 , , 171 , 233 brisken , w. f. , fruchter , a. s. , goss , w. m. , herrnstein , r. m. , & thorsett , s. e. 2003 , , 126 , 3090 buras , r. , rampp , m. , janka , h .-
t . , & kifonidis , k. 2003 , physical review letters , 90 , 241101 burrows , a. , hayes , j. , & fryxell , b. a. 1995 , , 450 , 830 burrows , a. , ott , c. d. , & meakin , c. 2004 ,
cosmic explosions in three dimensions , 209 burrows , a. , walder , r. , ott , c. d. , & livne , e. 2005 , asp conf .
ser . 332 : the fate of the most massive stars , 332 , 358 bykov , a. m. 2003 , , 410 , l5 chakrabarty , d. , pivovaroff , m. j. , hernquist , l. e. , heyl , j. s. , & narayan , r. 2001 , 548 , 800 chatterjee , s. , et al .
2005 , , 630 , l61 chevalier , r. a. 1975 , , 200 , 698 chevalier , r. a. , & klein , r. i. 1978 , , 219 , 994 chevalier , r. a. , & soker , n. 1989 , , 341 , 867 cid - fernandes , r. , plewa , t. , rozyczka , m. , franco , j. , terlevich , r. , tenorio - tagle , g. , & miller , w. 1996 , , 283 , 419 colpi , m. , & wasserman , i. 2002 , , 581 , 1271 cordes , j. m. , & chernoff , d. f. 1998 , , 505 , 315 delaney , t. , & rudnick , l. 2003 , , 589 , 818 deshpande , a. a. , ramachandran , r. , & radhakrishnan , v. 1999 , , 351 , 195 dopita , m. a. , & sutherland , r. s. 1995 , , 455 , 468 duncan , r. c. , & thompson , c. 1992 , , 392 , l9 fesen , r. a. 2001 , , 133 , 161 fesen , r. a. , & becker , r. h. 1991 , , 371 , 621 fesen , r. a. , becker , r. h. , & blair , w. p. 1987
, , 313 , 378 fesen , r. a. , & gunderson , k. s. 1996 , , 470 , 967 fesen , r. a. , hammell , m. c. , morse , j. , chevalier , r. a. , borkowski , k. j. , dopita , m. a. , gerardy , c. l. , lawrence , s. s. , raymond , j. c. , & van den bergh , s. 2006 , , 636 , 859 fesen , r. a. , pavlov , g. g. , & sanwal , d. 2006 , , 636 , 848 ford , h. c. , et al .
1998 , proc .
spie , 3356 , 234 fruchter , a. s. , et al . 1999 , , 519 , l13 fryer ,
c. l. , & heger , a. 2000 , , 541 , 1033 fryer , c. l. 2004 , , 601 , l175 galama , t. j. , et al .
1998 , , 395 , 670 gotthelf , e. v. , koralesky , b. , rudnick , l. , jones , t. w. , hwang , u. , & petre , r. 2001 , , 552 , l39 gull , s. f. 1975 , , 171 , 263 hamilton , a. j. s. 1985 , , 291 , 523 hammell , m. c. , & fesen , r. a. 2006 , in preparation hartigan , p. , morse , j. a. , & raymond , j. 1994 , , 436 , 125 herant , m. , & woosley , s. e. 1994 , , 425 , 814 hjorth , j. , et al .
2003 , , 423 , 847 h " oflich , p. , wheeler , j. c. , & wang , l. 1999 , , 521 , 179 h " oflich , p. , khokhlov , a. , & wang , l. 2001 , aip conf .
proc . 586 : 20th texas symposium on relativistic astrophysics , 586 , 459 hughes , a. , & bailes , m. 1999 , , 522 , 504 hurford , a. p. , & fesen , r. a. 1996 , , 469 , 246 hwang , u. , et al .
2004 , , 615 , l117 janka , h .-
t . , buras , r. , kifonidis , k. , plewa , t. , & rampp , m. 2003 , from twilight to highlight : the physics of supernovae , 39 janka , h. - . , scheck , l. , kifonidis , k. , mueller , e. , & plewa , t. 2005 , asp conf .
ser . 332 : the fate of the most massive stars , 332 , 372 jones , t. w. , kang , h. , & tregillis , i. l. 1994 , , 432 , 194 jun , b .- i . , jones , t. w. , & norman , m. l. 1996 , , 468 , l59 kamper , k. & van den bergh , s. 1976 , , 32 , 351 kawabata , k. s. , et al .
2003 , , 593 , l19 khokhlov , a. m. , h " oflich , p. a. , oran , e. s. , wheeler , j. c. , wang , l. , & chtchelkanova , a. y. 1999 , , 524 , l107 kifonidis , k. , plewa , t. , janka , h .-
t . , & m " uller , e. 2003 , , 408 , 621 klein , r. i. , mckee , c. f. , & colella , p. 1994 , in evolution of the interstellar medium , ed . l. blitz ( berkeley : asp ) , 117 klein , r. i. , mckee , c. f. , & colella , p. 1994
, , 420 , 213 kotake , k. , yamada , s. , & sato , k. 2005 , , 618 , 474 krause , o. et al .
2005 , science , 308 , 1604 laming , j. m. , & hwang , u. 2003 , , 597 , 347 leonard , d. c. , filippenko , a. v. , barth , a. j. , & matheson , t. 2000 , , 536 , 239 leonard , d. c. , filippenko , a. v. , ardila , d. r. , & brotherton , m. s. 2001 , , 553 , 861 leonard , d. c. , & filippenko , a. v. 2001 , , 113 , 920 liebend " orfer , m. , mezzacappa , a. , thielemann , f .- k . , messer , o. e. , hix , w. r. , & bruenn , s. w. 2001 , , 63 , 103004 liebend " orfer , m. , rampp , m. , janka , h .-
t . , & mezzacappa , a. 2005 , , 620 , 840 lyne , a. g. , & lorimer , d. r. 1994 , , 369 , 127 macfadyen , a. i. , & woosley , s. e. 1999 , , 524 , 262 macfadyen , a. i. , woosley , s. e. , & heger , a. 2001 , , 550 , 410 madokoro , h. , shimizu , t. , & motizuki , y. 2003 , , 592 , 1035 madokoro , h. , shimizu , t. m. & motizuki , y. 2004 , , 56 , 663 malesani , d. , et al .
2004 , , 609 , l5 matheson , t. , et al .
2003 , , 599 , 394 mazzali , p. a. , et al .
2005 , science , 308 , 1284 mereghetti , s. , tiengo , a. , israel , g. l. 2002 , , 569 , 275 minkowski , r. 1968 , in `` stars and stellar systems '' , 7 , 623 paczynski , b. 1998 , , 494 , l45 pavlov , g. g. , & zavlin , v. e. 1999 , iau circ .
7270 pavlov , g. g. , zavlin , v. e. , aschenbach , b. , trmper , j. , & sanwal , d. 2000 , , 531 , l53 pavlov , g. g. , sanwal , d. , garmire , g. p. , & zavlin , v. e. 2002 , in neutron stars in supernova remnants , ed .
p. o. slane & b. m. gaensler , asp conf .
271 ( san francisco : asp ) , 247 pavlov , g. g. , sanwal , d. , & teter , m. a. 2004 , in young neutron stars and their environments , iau symp . 218 , ed . f. camilo & b. m. gaensler ( san francisco : asp ) , 239 pavlovsky , c. , et al .
`` acs instrument handbook '' , version 5.0 , ( baltimore : stsci ) rampp , m. , & janka , h .-
2000 , , 539 , l33 reed , j. e. , hester , j. j. , fabian , a. c. , & winkler , p. f. 1995 , , 440 , 706 reynoso , e. m. , & goss , w. m. 2002 , , 575 , 871 rothschild , r. e. & lingenfelter , r. e. 2003 , , 582 , 257 scheck , l. , plewa , t. , janka , h .-
t . , kifonidis , k. , m "
uller , e. 2004 , physical review letters , 92 , 011103 shimizu , t. m. , ebisuzaki , t. , sato , k. , & yamada , s. 2001 , , 552 , 756 stanek , k. z. , et al .
2003 , , 591 , l17 stephenson , f. r. , & green , d. a. 2003 , ` the historical supernova remnants ' , cambridge university press , cambridge uk symbalisty , e. m. d. 1984 , , 285 , 729 tananbaum , h. 1999 , , 7246 thompson , c. & duncan , r. c. 1996 , , 473 , 322 thompson , t. a. , burrows , a. , & pinto , p. a. 2003 , , 592 , 434 thorstensen , j. r. , fesen , r. a. , & van den bergh , s. 2001 , , 122 , 297 takiwaki , t. , kotake , k. , nagataki , s. , & sato , k. 2004 , , 616 , 1086 trammell , s. r. , hines , d. c. , & wheeler , j. c. 1993 , , 414 , l21 van den bergh , s. 1971 , , 165 , 457 van den bergh , s. , & dodd , w. w. 1970 , , 162 , 485 walder , r. , burrows , a. , ott , c. d. , livne , e. , lichtenstadt , i. , & jarrah , m. 2005 , , 626 , 317 wang , l. , wheeler , j. c. , li , z. , clocchiatti , a. 1996 , , 467 , 435 wang , l. , howell , d. a. , h " oflich , p. , & wheeler , j. c. 2001 , , 550 , 1030 wang , l. , et al . 2002 , , 579 , 671 wex , n. , kalogera , v. , & kramer , m. 2000 , , 528 , 401 wheeler , j. c. 2003 , am j. phys .
, 71 , 11 wheeler , j. c. , yi , i. , h " oflich , p. , & wang , l. 2000 , , 537 , 810 wheeler , j. c. , meier , d. l. , & wilson , j. r. 2002 , , 568 , 807 wilson , j. r. , mathews , g. j. , & dalhed , h. e. 2005 , , 628 , 335 winkler , p. f. , roberts , p. f. , kirshner , r. p. 1991 , in `` supernovae : the tenth santa cruz summer workshop in astronomy and astrophysics '' , ed .
woosley ( springer - verlag : new york ) , p 652 woosley , s. e. 1993 , , 405 , 273 yamasaki , t. , & yamada , s. 2005 , , 623 , 1000 | _ hst _ images of the young supernova remnant cas a are used to explore the expansion and spatial distribution of its highest velocity debris .
acs / wfc images taken in march and december @xmath0 with sloan f625w , f775w , and f850lp filters were used to identify 1825 high - velocity , outlying ejecta knots through measured proper motions of @xmath1 yr@xmath2 corresponding to v@xmath3 km s@xmath2 assuming d = 3.4 kpc .
the distribution of derived transverse expansion velocities for these ejecta knots shows a striking bipolar asymmetry with the highest velocity knots ( v@xmath4 @xmath5 10500 km s@xmath2 ) confined to nearly opposing northeast and southwest ` jets ' at p.a .
= @xmath6 and @xmath7 , respectively .
the jets have about the same maximum expansion velocity of @xmath814000 km s@xmath2 and appear kinematically and chemically distinct in that they are the remnant s only s - rich ejecta with expansion velocities above the @xmath9 km s@xmath2 exhibited by outer nitrogen - rich ejecta which otherwise represent the remnant s highest velocity debris .
in addition , we find significant gaps in the spatial distribution of outlying ejecta in directions which are approximately perpendicular to the jets ( p.a .
= @xmath10 and @xmath11 ) .
the remnant s central x - ray point source lies some @xmath12 to the southeast of the estimated expansion center ( pa = @xmath13 ) indicating a projected motion toward the middle of the broad southern ejecta knot gap .
extrapolations of measured nine month proper motions for all 1825 outer ejecta knots and a selected subsample of 72 bright and compact knots suggest explosion dates ( assuming no knot deceleration ) of @xmath14 and @xmath15 , respectively .
we find some evidence for non - uniform deceleration in different directions around the remnant and find 126 knots located along the northwestern limb among the least decelerated ejecta suggesting a convergence date of @xmath16
. a remnant age of around 325 yr would imply a @xmath8 350 km s@xmath2 transverse velocity for the central x - ray point source . |
in the view of the so - called agn unified model @xcite a super - massive black hole ( smbh ) harbored at the center of the agn powers the nuclear activity .
the region where the activity takes place can be observed from different viewing angles . therefore depending on the orientation of the agn
the observer s line of sight intercepts different amounts of the optically thick gas dust structure ( torus ) that surrounds the smbh . the nuclear radiation at optical / uv and x - ray wavelengths is efficiently absorbed by the torus .
the amount of obscuring matter ( n@xmath3 column density associated to the torus ) can be best inferred by x - ray spectra of the agns .
x - ray surveys are therefore powerful tools for agn population studies .
the bias of x - ray surveys strongly depends on the column density associated to the sources and the survey sensitivity : the larger the column density and the worse the flux sensitivity , the better the low absorbed sources are selected .
such selection effect is negligible for unabsorbed sources ( exhibiting n@xmath3 @xmath4 10@xmath5 @xmath6 ) while it affects the absorbed sources ( exhibiting n@xmath3 @xmath7 10@xmath5 @xmath6 ) and it is magnified for sources with column densities n@xmath3 @xmath8 1.5 @xmath9 10@xmath10 @xmath6 .
this latter value corresponds to the inverse of the thompson cross - section ( @xmath11 ) and the optical depth unity for compton scattering .
absorbed sources affected by such high column densities are defined as `` compton - thick '' .
this plays an important role in nowadays most sensitive agn x - ray surveys that are performed by _ xmm - newton _ and _ chandra _ in the energy range @xmath00.5 - 10 kev @xcite . at these energies
less than a mere 10% of the nuclear radiation is energetic enough to pierce through the absorbing compton - thick torus @xcite . on the other hand
the efficiently absorbed optical / uv radiation heats the dust of the obscuring medium , that is expected to waste the absorbed radiation in form of ir emission .
indeed , an ir excess due to warm dust heated by obscured agns has been found @xcite .
infrared power - law selected samples in _ chandra _ deep fields are promising agn
candidates @xcite .
the drawback of the ir selection is that the majority of the detected sources are not agns .
furthermore this approach seems to sample best the sources within redshift 13 @xcite .
this is the same redshift range in which _ chandra _ and _ xmm - newton _ are preferentially selecting most agns in their deep surveys @xcite .
instead the redshift space at z @xmath4 0.4 is so far poorly explored despite extensive studies @xcite .
the low - redshift ( z @xmath4 0.4 ) universe is best fathomed at hard x ray energies ( @xmath7 15 kev ) . with the advent of the integral @xcite and the _ swift _ @xcite missions , the selection of local agns through their hard x
ray ( @xmath715 kev ) emission has proven to be an extremely powerful technique over the last few years .
integral and _ swift _ carry coded - mask telescopes on board , namely the imager on board the integral satellite ( ibis : * ? ? ?
* ) and the burst alert telescope ( bat : * ? ? ?
* ) respectively .
ibis has two detector layers .
one of which is the integral soft gamma - ray imager ( isgri : * ? ? ?
ibis / isgri and bat have two major advantages : 1 ) they have a huge field of view , hence allowing to sample an adequate number of agns at low - redshift 2 ) they operate at energies above 15 kev , hence allowing detecting the photons having enough penetrating power to pierce efficiently even through the compton - thick torus .
a further and major advantage in sampling photons above 15 kev from agns comes from the emitting source itself .
indeed , a broad continuum bump , so - called `` compton - reflection bump '' , peaking at energies between 20 - 30 kev is produced by reflection of the primary nuclear radiation on the inner side of the obscuring gas @xcite .
this spectral component has been found to be dominant in nearby heavily obscured agns @xcite .
the compton - reflection component also plays an important role in reproducing the shape and intensity of the cxb @xcite , that peaks at 30 kev ( for most recent measurements see : * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
estimates based on observations with pds @xcite on board the _
beppo_sax @xcite satellite predict that compton - thick agns should dominate over unobscured agns in the local universe @xcite .
this makes ibis / isgri and bat well suited instruments for detecting obscured agns in the local universe .
ibis / isgri and bat both represent a major improvement for the imaging of the sky above 15 kev .
however coded - mask detectors suffer from heavy systematic effects ( errors ) preventing them from reaching their theoretical limiting sensitivity @xcite .
furthermore by design they block @xmath050% of the incident photons causing an increase of the statistical noise .
these are the reasons that make the detection of extragalactic sources , that are mostly faint , still challenging to undertake . here
we describe an alternative approach which has been developed ad hoc to improve the sensitivity of extragalactic hard x
ray surveys by using ibis / isgri and bat . in this paper
we show that _ swift_/bat and integral / ibis observations can be merged to obtain a more sensitive survey that is able to sample limiting fluxes of @xmath03.3 @xmath9 10@xmath12 erg @xmath6 s@xmath13 in the 18 - 55 kev energy range .
we call this the _ six _ survey , that stands for _ swift_integral hard x - ray survey .
the _ six _ survey extends over a wide sky area of 6200 deg@xmath1 and it is used to obtain a small and persistent sample of faint sources .
this enables the construction of the number density ( log @xmath14log @xmath15 ) as well as developing the x - ray luminosity function ( xlf ) for agns .
in addition we estimate the contribution of this sample of agns to the intensity of the unresolved fraction of the cxb . throughout this paper
we adopt the cosmological parameters of : h@xmath16 = 70 h@xmath17 km s@xmath13 mpc@xmath13 , @xmath18 = 0.73 .
ibis / isgri and bat are both coded mask instruments .
their performances are considered as a milestone for the sky imaging at hard x - ray energies .
they pose themselves as excellent instruments for population studies of faint hard x - ray sources shedding continuously light on the properties of the local agn population @xcite .
currently the two instruments show a difference in the extragalactic sky survey .
this is mainly due to the different pointing strategies adopted for the satellites .
_ swift _ is quasi - randomly pointing the sky , while integral performs targeted observations and long exposures on the galactic plane .
the authors of the 4@xmath19 ibis / isgri catalog @xcite conclude that the non detection of bat detected sources by ibis / isgri is just due to the low exposure of those sources in the isgri detector .
indeed , at comparable exposure time on the extragalactic sky the bat sample @xcite contains @xmath070% extra - galactic sources , the ibis / isgri sample contains @xmath035% @xcite .
table [ tab : bat - isgri ] summarizes the in flight performances of the two instruments . for ibis
/ isgri the flux sensitivity in the 2040 kev energy range at 4.8@xmath20 is computed over 90% of the extragalactic sky @xcite .
the bat flux sensitivity is obtained over the entire extragalactic sky in the 1530 kev energy range at 4.8@xmath20 @xcite .
+ .comparison of bat and ibis / isgri in - flight performances.[tab : bat - isgri ] [ cols="<,^,^,^,^,^",options="header " , ] @xmath21 in units of 10@xmath22 erg@xmath13 s mpc@xmath23 + @xmath24 in units of 10@xmath25 erg s@xmath13
to capitalize on the advantages of selecting local agns at hard x - ray energies we have combined the observations of ibis / isgri and bat .
this greatly enhances the exposure time improving the _ six _ sensitivity as @xmath26 .
moreover , the systematic uncertainties are minimized because the systematic errors of both instruments are uncorrelated .
this reduces the uncertainties by a covariance term leading to a limiting flux sensitivity of 3.3 @xmath9 10@xmath12 erg @xmath6 s@xmath13 in the 1855 kev energy range .
+ bat contributes to the _ six _ survey with a very uniform exposure .
the exposure of ibis / isgri is less uniform ( see figure [ fig : skycov ] ) .
this is explained by the different fovs of both coded masks instruments and by the different pointing strategies that are adopted by the _ swift _ and the integral missions . integral performs a dithering that leads to a large exposure time at the center of the surveyed sky area as shown in figure [ fig : isgri - nep - exp ] .
the center of this sky area corresponds roughly to the value of the ordinate in figure [ fig : skycov ] where the ibis / isgri sky coverage outperforms that of bat .
ibis / isgri is therefore contributing to the _ six _ survey with its long exposures on limited sky areas . from observations at soft x
rays we have inferred the @xmath27-value of 64 out of our 74 agns .
the level of this incompleteness will change as some of these sources are followed up by _ chandra _ ( pi e. bottacini cxc ao-13 ) and xrt .
the fraction of obscured agns as function of luminosity is consistent with a previous work @xcite .
also the drop at low luminosities @xmath010@xmath28 erg s@xmath13 is reproduced in our work .
the relation is represented by an anti correlation shown in figure [ fig : lum - nh ] revealing that the @xmath27 function is luminosity dependent .
in contrast to the simple agn unified scheme @xcite , where agns have the same geometrical structure irrespective of luminosity and redshift , this result shows that the opening angle of the torus surrounding the smbh is larger for more luminous agns .
therefore , high
luminosity agns must be able to clean out their environments .
the physical process responsible for this could be the stronger radiation pressure from the smbh of the more luminous agns . in this scenario the circumnuclear environment is exposed to higher pressure that causes greater mass outflow .
indeed , the existence of such mass outflows is supported by recent observations @xcite .
if accompanied by sufficient mechanical energy transfer , these outflows can provide the coupling of the smbh and host galaxy co - evolution @xcite .
however , it is ambitious to draw conclusions from the observational fact of the anti correlation .
how the covering factor of the torus relates to the luminosity depends on the interplay of the smbh gravity with the gas , radiation , and magnetic fields accompanied by wind outflows .
the thermal pressure in the gas is able to explain the low - velocity outflow as x - ray warm absorbers @xcite . instead , the uv radiatively - driven wind shields itself from x - rays from the central engine @xcite .
on the other hand , a natural shielding is obtained in the magneto - centrifugal outflow scenario @xcite .
finally , in hydrodynamical simulations @xcite high - density gas arises naturally that is able to efficiently absorb x - ray radiation .
all these processes show promising results in numerical simulations .
+ a possible dependence ( decrease ) of the fraction of absorbed agns as function of redshift is marginally detected in this work ( see figure [ fig : lum - z ] ) . even though within a larger redshift ( z @xmath4 3 ) range , a similar relation is found by @xcite .
instead a strong increase of the fraction of obscured agn within 0 @xmath4 z @xmath4 2 is found by @xcite and @xcite .
this dependency on the distance at high redshift can be due to the higher gas content in high redshift ( z @xmath0 1.5 ) galaxies @xcite . in the local universe
our results show a mild decay of the fraction of obscured agns with redshift .
+ the fit to our flux
number density function is consistent with the euclidean model .
this is in good agreement with previous measurements by @xcite , @xcite , @xcite , and @xcite .
we extend the result towards lower fluxes reaching a flux limit of a few 10@xmath12 @xmath6 s@xmath13 . as a result of the increase of the number density of sources ,
the _ six _ source sample improves the number density of sources contributing to the cxb by more than a factor @xmath0 2 compared to the fraction derived from the sample in @xcite .
the contribution of the latter sample to the cxb is obtained by integrating the emission of the agn over the entire extragalactic sky of 30000 deg@xmath29 .
even though in our work we survey @xmath01/5 of the extragalactic sky , our surveyed sky area is particularly suited because of absence of strong sources .
we can estimate the number of seyfert like agn in a deep nustar @xcite survey .
nustar is a nasa mission operating at energies in the range 380 kev and scheduled to be launched in june 2012 .
it is the first high energy x - ray mission using focusing optics .
its relatively small fov ( @xmath0 13 @xmath9 13 arcmin squared ) allows surveying small areas compared to coded mask detectors .
if the luminosity function derived here does not evolve strongly in either normalization and/or slope the estimated number of agn per square degree at energies @xmath030 kev is @xmath0100 for a limiting flux of 10@xmath30 erg @xmath6 s@xmath13
. this will be quite possible to be performed by nustar since it will have an angular resolution @xmath045 arc sec and source confusion will not be a problem .
nustar will perform very sensitive and beam like extragalactic surveys tiling not more than a few deg@xmath1 of sky area .
detailed predictions @xcite show , that independent of nustar s survey strategy ( corner shift or half shift ) and the cxb models , agns will be sampled most efficiently at redshift z @xmath7 0.5 .
figure [ fig : z - lum ] displays the redshift - luminosity plane of seyfert - like agns sampled by bat ( gray circles in background from * ? ? ?
* ) , _ six _ ( solid circles in foreground ) , and nustar .
nustar data are taken from predictions in @xcite and adapted to the energy range 1855 kev . ]
as the nustar survey fields narrow , the redshift distribution of agns is shifted to higher redshift .
this is shown in figure [ fig : z - lum ] where squares are from a shallow survey , crosses from the cosmos survey ( 2 deg@xmath1 ) , and asterisks from the ecdf - s survey ( 0.25 deg@xmath1 ) assuming an exposure of 6.2 msec .
nustar s surveys will not compete with the _ six _ survey , but rather they will be complementary .
indeed , if applied to the whole sky the _ six _ survey will fill the redshift and luminosity gap between the current surveys of ibis / isgri and bat alone and the nustar surveys . the absorbed _ six _ sources at low redshift
are easy follow up targets for nustar .
+ our small sample used for the xlf does not permit detecting an evolution in either luminosity or redshift .
therefore , our data are modeled best by a non - evolving xlf plotted in figure [ fig : lum - function ] .
+ we use the 199 bat - selected seyfert - like agns of @xcite and later used in @xcite as a control sample in order to evaluate whether the _ six _ sample differs in some properties . for both samples
we have derived the redshift distribution excluding radio - loud agns and quasars .
the result is plotted in figure [ fig : redshift - distribution ] where the solid line refers to the _ six _ sample and the dashed line represents the sample of @xcite .
both samples were split into the same bin size . about 57% of the bat sources
are detected within z = 00.025 . in the same redshift bin
the _ six _ survey samples less than 45% of its sources . ]
the _ six _ survey is more sensitive compared to the bat survey .
therefore it samples the sources systematically at higher redshift .
sources at higher redshift are also the most luminous ones .
this is marginally reflected in the xlf ( see figure [ fig : lum - function ] ) for luminosities above @xmath31 .
the best - fit of the xlf adapted from the bat survey ( dashed line ) has a slightly steeper slope at high luminosities . for luminosities below @xmath32 a substantial contribution from the host galaxy to the total luminosity
is expected . therefore , the data point at lowest luminosities in not properly modeled by the _
six _ xlf .
+ finally , we verify whether the control samples and the _ six _ sample are drawn from the same source population .
a kolmogorov - smirnov test applied to the two samples with respect to redshift and luminosity does not allow us to reject the null hypothesis that the two samples are obtained from the same source population .
therefore , the _ six _ is sampling sources at higher redshift from the same source population that is sampled by bat alone .
better statistics for the _ six _ xlf can rule out the high - end slope
. this can be achieved by applying the survey to the whole sky .
a new approach has been developed to survey the sky at hard x - ray energies ( 1855 kev energy band ) by combining the observations of _ swift_/bat and integral / ibis resulting in the _ six _ survey .
first we have performed the independent surveys for both instruments .
then we have resampled , cross - calibrated and merged them . as a result of combining the observations from two different telescopes ,
statistical and systematic uncertainties caused by the high background level of their coded - mask detectors are minimized . in turn the _ six _
survey is more sensitive , like a survey from a virtual new hard x - ray mission .
+ we applied the survey method to 6200 deg@xmath1 of sky ( @xmath0 20% of the entire extragalactic sky ) sampling 113 sources that are : 74 seyfert - like agns , 12 blazars , 5 galaxies , 2 clusters of galaxies , 3 galactic sources , 3 previously detect x - ray sources , and 14 unidentified sources ( of which 7 are newly detected without any counterpart and 7 are of uncertain association ) .
no false detections due to statistical or systematic fluctuations are expected .
the sources are identified through their soft x - ray counterparts and with _ chandra _
follow up observations ( cxc ao-12 ) .
unidentified sources are being followed - up in _ chandra _ ao-13 . among the agn sample
only two sources are compton
thick , accounting for @xmath03% of the entire sample .
+ the number density of our identified sources is @xmath04 times greater than in our control sample of @xcite . even though this represents only a minor fraction of the cxb , the sensitivity improvement with respect to previous measurements is better than a factor of 2 .
+ the fraction of absorbed agn decreases with increasing luminosities .
although the redshift dependence is marginally significant , we find a mild decrease of the fraction of obscured agns with increasing redshift .
these results require that the covering factor of the torus surrounding the smbh changes at least with luminosity .
+ only robustly identified agns were used in our xlf .
the data are well represented by a double power - law model and do not show any evolution in density or in luminosity .
the non - evolving xlf model fits our data best .
+ based on our results we predict the number density of @xmath0100 seyfert like agns that the upcoming nustar mission can detect in 1 deg@xmath1 of surveyed sky area at a limiting flux of @xmath33 .
+ we thank the _ integral _ and the _ swift _ teams for their observations and support .
this research used the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , data obtained from the high energy astrophysics science archive research center ( heasarc ) provided by nasa s goddard space flight center , and the simbad astronomical database , which is operated by the centre de donnes astronomiques de strasbourg .
we also made use of the topcat tool @xcite .
this research has made use of data obtained from the chandra data archive and the chandra source catalog , and software provided by the chandra x - ray center ( cxc ) in the application packages ciao , chips , and sherpa .
this research is based on observations with _ integral _ , an esa project with instruments and science data center funded by esa member states ( especially the pi countries : denmark , france , germany , italy , switzerland , spain ) , and poland , and with the participation of russia and the usa .
we acknowledge g. m. madejski for very helpful suggestions .
we thank s. digel for comments on part of the paper .
d. r. ballantyne is acknowledged for providing predictions on nustar data .
e. b. expresses gratitude to w. collmar who made it possible to have access to computing facilities .
the authors acknowledge the comments of the anonymous referee , which improved this paper .
this research was started under grant dlr fkz 50 or 0405 .
e.b . acknowledges support through sao grants go1 - 12144x and go2 - 13139x .
+ _ facilities : _ , .
133.8152 & 64.4063 & mcg+11 - 11 - 032 & 7.97 & 6.44 & agn & sy2 & 0.03 & 43.40 & 23.01 [ 1 ] + 148.9211 & 69.6846 & m 82 & 7.09 & 4.90 & galaxy interacting & galaxy interacting & 0.0006 & 39.97 & + 149.0083 & 69.0811 & m 81 & 10.1 & 7.34 & agn & liner & 0.0001 & 38.64 & 20.53 [ 1 ] + 150.4723 & 55.6943 & 4c 55.19 & 13.9 & 11.4 & agn & sy2 & 0.004 & 41.64 & 24.7 [ 2 ] + 161.0304 & 70.4238 & mcg+12 - 10 - 067 & 6.31 & 5.67 & agn & sy2 & 0.03 & 43.20 & 23.29 [ 1 ] + 163.2331 & 10.6582 & & 11.6 & 6.67 & & & & & + 166.1311 & 38.2082 & mrk 421 & 101 .
& 78.5 & blazar & bl lac & 0.02 & 44.32 & + 166.4922 & 58.9187 & 1rxs j110537.4 + 585128 & 5.66 & 5.04 & agn & qso & 0.19 & 44.77 & 21.20 [ 1 ] + 166.6683 & 72.5697 & ngc 3516 & 57.8 & 46.0 & agn & sy1.5 & 0.00 & 42.99 & 21.21 [ 1 ] + 168.9601 & 54.4453 & & 6.31 & 4.80 & & & & & + 171.3510 & 54.3702 & mrk 0040 & 10.7 & 9.16 & agn & sy1 & 0.02 & 43.01 & 20.90 [ 2 ] + 172.5265 & -14.817 & om -146 & 17.4 & 8.96 & blazar & fsrq & 1.18 & 47.14 & + 173.1405 & 52.9792 & ngc 3718 & 7.10 & 6.89 & agn & liner & 0.003 & 41.23 & 20.00 [ 2 ] + 173.2427 & 10.2765 & 2masx j11324928 + 1017473 & 7.57 & 5.99 & agn & sy1 & 0.04 & 43.52 & 21.46 [ 1 ] + 174.2112 & 67.6433 & rbs 1004 & 5.50 & 5.42 & blazar & bl lac & 0.13 & 44.42 & + 174.8026 & 59.2078 & rbs 1011 & 11.2 & 9.43 & agn & sy1.5 & 0.06 & 43.98 & 19.58 [ 3 ] + 175.5590 & 10.3157 & ngc 3822 & 5.85 & 5.60 & agn & sy1 & 0.02 & 42.88 & 20.00 [ 1 ] + 175.9280 & 71.6968 & do dra & 15.2 & 10.6 & cv & v * do dra & & & + 176.3579 & 58.9892 & mcg+10 - 17 - 061 & 8.39 & 6.15 & galaxy & galaxy & 0.01 & 42.30 & + 176.4277 & -18.436 & rbs 1030 & 31.5 & 15.4 & agn & sy1 & 0.03 & 43.89 & 20.54 [ 1 ] + 177.0379 & 9.00302 & 2masx j11475508 + 0902284 & 8.07 & 5.24 & agn & sy ?
& 0.06 & 43.96 & 21.00 [ 1 ] + 178.0421 & -11.374 & rbs 1044 & 10.3 & 5.52 & agn & sy1 & 0.04 & 43.76 & + 178.4084 & 49.5092 & rbs 1046 & 4.13 & 4.82 & blazar & fsrq & 0.33 & 45.18 & + 179.5145 & 55.4250 & ngc 3998 & 9.91 & 7.43 & agn & liner & 0.003 & 41.46 & 20.09 [ 1 ] + 179.6989 & 42.5570 & ic 751 & 4.36 & 5.90 & agn & sy2 & 0.03 & 43.00 & + 180.1644 & -1.1668 & 2qz j120045.2 - 011041 & 10.1 & 6.73 & agn & sy1 & 0.37 & 45.67 & + 180.2348 & 6.81394 & cgcg
041 - 020 & 11.8 & 9.36 & agn & sy2 & 0.03 & 43.54 & 22.83 [ 3 ] + 180.2905 & -3.6918 & mrk 1310 & 5.67 & 5.34 & agn & sy1 & 0.01 & 42.69 & 20.72 [ 1 ] + 180.7713 & 44.5210 & ngc 4051 & 25.4 & 23.5 & agn & sy1.5 & 0.002 & 41.41 & 20.47 [ 1 ] + 181.5795 & 52.7170 & ngc 4102 & 10.8 & 10.4 & agn & liner & 0.002 & 41.29 & 20.94 [ 1 ] + 182.2920 & 47.0460 & mrk 0198 & 8.73 & 8.30 & agn & sy2 & 0.02 & 43.07 & 22.80 [ 1 ] + 182.3597 & 43.6981 & ngc 4138 & 11.3 & 11.2 & agn & sy1.9 & 0.002 & 41.33 & 22.90 [ 3 ] + 182.6350 & 39.4063 & ngc 4151 & 239 . & 222 . &
agn & sy1 & 0.003 & 42.74 & 22.50 [ 3 ] + 182.6950 & 38.3332 & kug 1208 + 386 & 11.8 & 6.88 & agn & sy1 & 0.02 & 43.14 & 22.53 [ 2 ] + 183.0916 & -7.5984 & & 9.56 & 6.69 & & & & & + 183.2574 & 7.03921 & 2mass j12124981 + 0659451 & 7.99 & 6.37 & agn & qso & 0.20 & 45.00 & + 184.2881 & 7.17577 & ngc 4235 & 12.3 & 9.99 & agn & sy1 & 0.007 & 42.21 & 21.16 [ 1 ] + 184.7378 & 47.2874 & ngc 4258 & 11.3 & 10.7 & agn & liner & 0.001 & 40.77 & 22.91 [ 4 ] + 185.5332 & 75.3006 & mrk 205 & 11.9 & 8.60 & agn & sy1 & 0.07 & 44.15 & 20.88 [ 5 ] + 185.6009 & 4.20212 & 4c 04.42 & 10.5 & 9.81 & blazar & fsrq & 0.96 & 46.70 & + 185.8403 & 2.68141 & mrk 50 & 12.8 & 8.75 & agn & sy1 & 0.02 & 43.19 & 20.92 [ 1 ] + 186.4482 & 12.6643 & ngc 4388 & 119 . &
94.0 & agn & sy2 & 0.008 & 43.27 & 23.63 [ 3 ] + 187.2817 & 2.04932 & 3c 273 & 172 .
blazar & fsrq & 0.15 & 46.07 & + 188.6686 & 52.6444 & & 6.42 & 5.11 & & & & & + 189.6855 & 9.46984 & 2masx j12384342 + 0927362 & 6.61 & 5.66 & agn & sy2 & 0.08 & 44.05 & + 189.7676 & -16.184 & igr j12391 - 1612 & 16.8 & 10.9 & agn & sy2 & 0.03 & 43.72 & 22.48 [ 1 ] + 189.9053 & -5.3471 & ngc 4593 & 42.2 & 38.1 & agn & sy1 & 0.009 & 42.88 & 20.30 [ 3 ] + 191.7000 & 54.5375 & ngc 4686 & 9.72 & 9.24 & agn & sy2 & 0.01 & 42.78 & 23.84 [ 1 ] + 192.8286 & -11.722 & & 9.22 & 5.81 & & & & & + 193.0649 & -13.419 & ngc 4748 & 9.34 & 6.18 & agn & sy1 & 0.01 & 42.59 & 20.77 [ 1 ] + 193.9939 & 4.33340 & & 5.84 & 5.21 & & & & & + 194.0513 & -5.7909 & 3c 279 & 12.4 & 10.3 & blazar & fsrq & 0.53 & 46.15 & + 195.9917 & 53.7738 & igr j13038 + 5348 & 16.4 & 14.9 & agn & sy1 & 0.02 & 43.52 & 20.81 [ 1 ] + 196.0532 & -5.5644 & ngc 4941 & 9.11 & 6.45 & agn & sy2 & 0.003 & 41.43 & 22.95 [ 1 ] + 196.0877 & -10.309 & ngc 4939 & 12.0 & 8.75 & agn & sy2 & 0.01 & 42.46 & + 196.9632 & -2.0556 & & 7.63 & 7.48 & & & & & + 197.2785 & 11.6407 & ngc 4992 & 20.9 & 17.6 & agn & sy2 & 0.02 & 43.48 & 23.74 [ 6 ] + 198.2996 & -11.127 & rbs 1233 & 8.95 & 6.65 & agn & sy1 & 0.03 & 43.38 & 20.74 [ 1 ] + 198.8493 & 44.4093 & mrk 248 & 11.3 & 11.4 & agn & sy2 & 0.03 & 43.51 & 22.81 [ 2 ] + 199.7661 & -9.3549 & & 6.85 & 5.89 & & & & & + 200.2616 & 8.96113 & ngc5100 & 6.13 & 5.29 & galaxy group & galaxy group & 0.03 & 43.15 & + 200.6131 & -16.733 & mcg-03 - 34 - 063 & 21.3 & 11.0 & agn & sy2 & 0.01 & 43.12 & 23.59 [ 3 ] + 202.1334 & -1.5129 & & 6.34 & 5.36 & & & & & + 203.6978 & -23.425 & eso 509 - 66 & 13.2 & 5.77 & agn & sy2 & 0.04 & 43.79 & 23.05 [ 1 ] + 203.9518 & 3.01973 & ngc 5231 & 8.80 & 5.29 & agn & & 0.02 & 42.97 & 22.23 [ 1 ] + 204.3828 & -13.032 & qso b1334 - 127 & 7.91 & 4.83 & blazar & bl lac & 0.53 & 45.95 & + 204.5677 & 4.54615 & ngc 5252 & 52.1 & 43.0 & agn & sy2 & 0.02 & 43.76 & 22.34 [ 8 ] + 205.0111 & 55.8247 & & 8.60 & 5.48 & & & & & + 205.3109 & -14.660 & rbs 1303 & 13.0 & 6.54 & agn & sy1 & 0.04 & 43.72 & 21.23 [ 1 ] + 206.3981 & 41.6618 & ngc 5290 & 9.73 & 5.77 & galaxy group & galaxy group & 0.00 & 42.20 & + 208.0023 & -18.300 & & 16.9 & 6.12 & & & & & + 208.3393 & 69.3013 & mrk 279 & 21.0 & 16.3 & agn & sy1 & 0.03 & 43.65 & 20.53 [ 3 ] + 208.4456 & -11.406 & & 7.69 & 4.80 & & & & & + 209.0421 & 38.5687 & mrk 0464 & 10.2 & 8.91 & agn & sy1 & 0.05 & 43.79 & 20.00 [ 2 ] + 213.3866 & -3.2043 & ngc 5506 & 131 . &
79.0 & agn & sy1.9 & 0.00 & 43.03 & 22.53 [ 3 ] + 215.4348 & 47.7875 & rbs 1378 & 10.4 & 9.36 & agn & sy1 & 0.07 & 44.11 & 21.26 [ 2 ] + 216.5659 & 37.8241 & abell 1914 & 5.11 & 5.16 & galaxy cluster & galaxy cluster & 0.17 & 44.61 & + 217.2120 & 42.6515 & h 1426 + 428 & 13.0 & 11.6 & blazar & bl lac & 0.12 & 44.75 & + 217.3623 & 1.31454 & mrk 1383 & 12.2 & 6.97 & agn & sy1 & 0.08 & 44.35 & 20.00 [ 2 ] + 218.4907 & 5.47754 & ngc 5674 & 8.13 & 5.55 & agn & sy1 & 0.02 & 43.06 & + 218.8075 & 48.6441 & ngc 5683 & 6.77 & 5.09 & agn & sy1 & 0.04 & 43.42 & + 219.1769 & 58.7837 & mrk 817 & 15.5 & 11.9 & agn & sy1 & 0.03 & 43.54 & 23.49 [ 2 ] + 219.3705 & 58.9051 & & 8.57 & 7.18 & & & & & + 220.2547 & 53.4781 & mrk 477 & 9.68 & 7.34 & agn & sy2 & 0.03 & 43.51 & 24.00 [ 7 ] + 220.6846 & -17.225 & ngc 5728 & 42.2 & 17.6 & agn & sy2 & 0.01 & 42.92 & 24.14 [ 9 ] + 228.8484 & 42.0416 & ngc 5899 & 8.82 & 6.56 & agn & sy2 & 0.01 & 42.15 & 23.12 [ 2 ] + 229.8817 & 65.6329 & mcg+11 - 19 - 006 & 7.33 & 4.91 & agn & sy2 & 0.04 & 43.52 & 21.38 [ 1 ] + 234.0596 & 57.8813 & mrk 290 & 13.0 & 9.86 & agn & sy1 & 0.02 & 43.41 & 20.40 [ 3 ] + 236.6410 & 69.4657 & 2masx j15462424 + 6929102 & 5.66 & 5.17 & galaxy & galaxy & & & + 243.5509 & 65.6973 & mrk 876 & 8.32 & 6.02 & agn & sy1 & 0.11 & 44.49 & 19.06 [ 1 ] + 245.0576 & 81.0390 & mcg+14 - 08 - 004 & 10.4 & 8.79 & agn & sy2 & 0.02 & 43.13 & 22.88 [ 1 ] + 247.1031 & 51.7736 & mrk 1498 & 23.9 & 16.5 & agn & sy1.9 & 0.05 & 44.24 & 23.26 [ 3 ] + 253.1046 & 55.9419 & mcg+09 - 28 - 001 & 5.49 & 4.83 & agn & sy2 & 0.02 & 43.02 & 22.66 [ 1 ] + 259.9293 & 48.9820 & arp 102b & 10.7 & 5.47 & blazar & fsrq & 0.02 & 43.18 & + 260.5280 & 43.2471 & first j172201.9 + 431523 & 9.15 & 5.05 & agn & sy1 & 0.13 & 44.67 & 21.12 [ 1 ] + 270.0433 & 66.5841 & ngc 6552 & 4.98 & 4.96 & agn & sy2 & 0.02 & 42.90 & + 274.0939 & 49.8605 & am her & 35.0 & 23.1 & cv & cv & & & + 275.5807 & 64.3694 & 1es 1821 + 643 & 7.00 & 9.06 & agn & sy1 & 0.29 & 45.29 & 20.00 [ 2 ] + 277.5011 & 48.7652 & 3c 380 & 8.34 & 5.58 & blazar & & 0.69 & 46.24 & + 280.6068 & 79.7714 & 3c 390.3 & 39.8 & 35.7 & agn & sy1 & 0.05 & 44.47 & 21.03 [ 3 ] + 281.3777 & 72.1933 & 2masx j18452628 + 7211008 & 3.97 & 5.22 & agn & sy2 & 0.04 & 43.29 & 22.82 [ 1 ] + 290.2788 & 43.9628 & aco 2319 & 24.2 & 16.6 & galaxy cluster & galaxy cluster & 0.05 & 44.25 & + 291.1899 & 50.2373 & ch cyg & 11.8 & 7.33 & symbiotic star & symbiotic star & & & + 291.2818 & 50.6906 & 2e 1923.7 + 5037 & 7.98 & 4.92 & x - ray source & x - ray source & & & + 291.7341 & 41.5856 & 1rxs j192630.6 + 413314 & 10.8 & 5.53 & x - ray source & x - ray source & & & + 292.1210 & 73.9471 & 1es 1928 + 73.8 & 5.32 & 4.98 & agn & sy1 & 0.03 & 43.20 & 20.93 [ 1 ] + 296.8676 & 44.8043 & cxou j194719.3 + 444942 & 13.8 & 9.04 & agn & sy2 & 0.05 & 43.96 & + 299.7378 & 40.8185 & cxo j195857.9 + 404856 & 38.2 & 31.4 & x - ray source & x - ray source & & & + 300.0223 & 65.1576 & 1es 1959 + 650 & 15.5 & 13.3 & blazar & bl lac & 0.04 & 43.92 & + 310.7268 & 75.1344 & 4c 74.26 & 25.0 & 19.9 & agn & sy1 & 0.10 & 44.83 & 21.22 [ 1 ] + 311.7313 & 65.2038 & & 5.67 & 4.81 & & & & & + 318.6292 & 82.0649 & s5 2116 + 81 & 19.2 & 13.1 & agn & sy1 & 0.08 & 44.54 & 21.03 [ 1 ] + 337.3319 & 66.7884 & igr j22292 + 6647 & 11.2 & 6.12 & agn & sy ? & 0.11 & 44.56 & 21.77 [ 1 ] + , f. , costa , e. , dal fiume , d. , feroci , m. , nicastro , l. , orlandini , m. , palazzi , e. , & zavattini , g. 1997 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
3114 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
o. h. siegmund & m. a. gummin , 206215 , j. , markwardt , c. b. , skinner , g. k. , baumgardner , w. h. , & swift bat survey team .
2009 , in bulletin of the american astronomical society , vol .
41 , bulletin of the american astronomical society , 269+ | the super - massive black - holes in the centers of active galactic nuclei ( agn ) are surrounded by obscuring matter that can block the nuclear radiation . depending on the amount of blocked radiation
, the flux from the agn can be too faint to be detected by currently flying hard x - ray ( above 15 kev ) missions . at these energies
only @xmath01% of the intensity of the cosmic x - ray background ( cxb ) can be resolved into point - like sources that are agns . in this work
we address the question of the undetected sources contributing to the cxb with a very sensitive and new hard x ray survey : the _ six _ survey that is obtained with the new approach of combining the _ swift_/bat and integral / ibis x ray observations .
we merge the observations of both missions .
this enhances the exposure time and reduces systematic uncertainties . as a result
we obtain a new survey over a wide sky area of 6200 deg@xmath1 that is more sensitive than the surveys of _ swift_/bat or integral / ibis alone .
our sample comprises 113 sources : 86 agns ( seyfert - like and blazars ) , 5 galaxies , 2 clusters of galaxies , 3 galactic sources , 3 previously detected unidentified x - ray sources , and 14 unidentified sources .
the scientific outcome from the study of the sample has been properly addressed to study the evolution of agns at redshift below 0.4 .
we do not find any evolution using the 1/v@xmath2 method .
our sample of faint sources are suitable targets for the new generation hard x - ray telescopes with focusing techniques . |
modern physical models of eclipsing binaries and transiting exoplanets ( eclipsing systems = es ) ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) are able to simulate their light curves with impressive fidelity .
however , a solution of the reverse problem , to derive all important parameters of the model from observational data , is seldom unique .
if we have observational data of moderate or poor quality , we are often forced to diminish the number of free parameters and thus use simplified ( and frequently physically inconsistent ) models . a decision which parameters to fix or leave floating is usually nontrivial . nevertheless , for several common practical tasks , such as : 1 ) determination of the mid - eclipse times , 2 ) light curve fitting , 3 ) description and classification of es light curve shapes for the purpose of the current and future surveys like asas or gaia , and 4 ) fine period analysis , we do not need to know the detailed physics of the system .
a good approximation of the shape of observed light curves is usually sufficient in such cases .
we present here a few - parameter general phenomenological model of eclipsing system light curves in the form of the special , analytic , periodic functions that are able to fit an overwhelming majority of the curves with an accuracy better than 1% . for the sake of simplicity
, we will limit our considerations only to systems with approximately circular orbits that represent more than 80% of observed eclipsing systems .
the model function of a monochromatic light curve ( expressed in magnitudes ) of eclipsing systems @xmath0 , can be assumed as the sum of three particular functions : @xmath1 where @xmath2 describes the mutual eclipses of the components , @xmath3 models the proximity effects , while @xmath4 approximates the oconnell effect ( irrespectively of its physical cause ) .
the normally occurring features of all eclipsing system light curves are two more or less symmetrical depressions caused by mutual eclipses of gravitationally bound stellar or planetary components
. the model function of eclipses @xmath5 can be approximated by the sum of two symmetrical analytic functions of the phase @xmath6 .
we assume that the primary eclipse is centered at the phase @xmath7 , while the secondary eclipse ( if one is present ) is centered at phase @xmath8 , @xmath9}{\left\{1\!-\!{\left\{1\!-\!\exp{\left[1\!-\ ! \cosh{\left(\frac{\varphi_k}{d_k}\right)}\right]}\right\}}^{\mathit{\gamma}_{\!k}}\right\}};\\ & \varphi_k={\left[\varphi-(k-1)/2\right ] } - \mathrm{round}{\left[\varphi-(k-1)/2\right]},\nonumber\end{aligned}\ ] ] where the summation is over the number of eclipses , @xmath10 : @xmath11 or @xmath12 ( the common situation for exoplanet transits ) , @xmath13 are the auxiliary phases defined in the interval @xmath14 , @xmath15 are the central depths of eclipses in mag , @xmath16 is the parameter expressing half - widths of eclipses , @xmath17 are the correcting parameters , @xmath18 parameterize the kurtosis of individual eclipses .
the fit of a transit of the exoplanet tres2b ( @xmath12 ) observed by _ kepler _
( see fig.[fig1 ] ) @xcite utilizes five parameters : @xmath19 .
however , the full description of the model monochromatic light curves with two eclipses ( @xmath11 ) would generally require nine free parameters : @xmath20 , and @xmath21 .
nevertheless , the most common situations would permit a reduction to 5 parameters , @xmath22 , because it is usually acceptable to assume that @xmath23 , @xmath24 , @xmath25 , @xmath26 . a detailed discussion of the properties of the eclipse model function @xmath5 and its possible simplifications are given in @xcite .
light variations of eclipsing binaries caused by eclipses are usually modified by proximity and oconnell effects that may be modelled by a sum of cosine and sine harmonic polynomials : @xmath27};\quad f_{\rm c}(\varphi)= \sum_{k = n_{\mathrm p}+n_{\mathrm e}+1}^{n_{\mathrm c}+ n_{\mathrm p}+n_{\mathrm e } } a_k\,\sin(2\,\pi\,\varphi),\ ] ] where @xmath28 , @xmath29 is the number of terms in @xmath30 : @xmath31 , @xmath32 if the oconnell asymmetry is not present then @xmath33 , else @xmath34 ( for more information see * ? ? ? * ) . the uncommon @xmath35 light curve of eclipsing binary v477lyr ( see fig.[477lyr ] ) is determined by 8 parameters : @xmath36 , @xmath37 .
the parameters of the above defined model functions , especially the amplitudes @xmath38 , and parameters @xmath39 , @xmath40 and @xmath41 are generally wavelength dependent .
it follows both from the theory and experience that most of the wavelength dependencies can be well approximated by the low order polynomials of the quantity @xmath42 ; @xmath43 , where @xmath44 is the effective wavelength of corresponding passband , @xmath45 is an arbitrarily selected central wavelength of the data set . then : @xmath46 where @xmath47 , @xmath48 or @xmath49 , are the numbers of degrees of freedom of the corresponding parameters of the model .
the standard set of the monochromatic light curve model parameters of eclipsing binaries ( see sec.[monoch ] ) : @xmath50 can be considered as the special case of the multicolour decomposition eq.[multi ] : @xmath51 .
all needed parameters and their uncertainties can be determined using the standard non - linear weighted least square technique ( see in * ? ? ?
the outlined phenomenological modeling of eclipse - binary and planetary - transit light curves has an advantage of high efficiency and simplicity and thus can be used for a wide range of applications . | we present a new few - parameter phenomenological model of light curves of eclipsing binaries and stars with transiting planets that is able to fit the observed light curves with the accuracy better than 1% of their amplitudes .
the model can be used namely for appropriate descriptions of light curve shapes , classification , mid - eclipse time determination , and fine period analyses . |
let @xmath4 be a global field , @xmath3 its ring of adles and @xmath0 a reductive group defined over @xmath4 .
the theory of automorphic forms involves the study of spaces of functions on @xmath7 as representations of @xmath8 .
the functions involved are often required to be right invariant under certain large compact subgroups @xmath6 of @xmath8 because ( among other reasons ) the double coset space @xmath9 admits nice interpretations .
for example , the classical study of the upper half plane modulo the action of arithmetic subgroups of the real special linear group is a special case of the above when @xmath4 is the field of rational numbers ( see e.g. , ( @xcite , 1 ) .
another special case , which corresponds to taking @xmath4 to be a field of rational functions in one variable and @xmath0 to be @xmath10 is discussed by weil in @xcite .
when @xmath4 is a function field , harder describes a fundamental domain for the action of @xmath11 on @xmath8 in ( @xcite , 1 ) using results from @xcite and @xcite .
this is an analogue of the seigel domain described by godement in @xcite for @xmath12 .
proposition [ prop : omega ] in this article is analogous to these results and the proof proceeds along the lines of @xcite .
harder s description of the fundamental domain is a very basic result in the theory of automorphic forms over function fields ( see e.g. , @xcite , 9 and appendix e ) . from now on let @xmath0 be a split reductive group defined over a finite field @xmath1 with @xmath13 elements . fix a borel subgroup @xmath14 defined over @xmath1 with unipotent radical @xmath15 , and a maximal @xmath1-split torus @xmath16 contained in @xmath14 .
set @xmath2 . for a valuation @xmath17 of @xmath4
, we denote the corresponding local field by @xmath18 and its ring of integers by @xmath19 .
for each @xmath17 , fix a uniformizing element @xmath20 .
in particular , fix @xmath21 as a uniformizing element at the place @xmath22 whose local field is @xmath23 .
let @xmath6 be the maximal compact subgroup @xmath24 of @xmath8 .
this article concerns the double coset space @xmath25 which may be interpreted as the set of isomorphism classes of principal @xmath0-bundles on the projective line . in @xcite
, grothendieck proves that when @xmath0 is a complex reductive group any holomorphic @xmath0-bundle over the complex projective line admits a reduction of structure group to a maximal torus .
( in fact this result has been attributed to dedekind and weber for @xmath26 by geyer ( @xcite , 6 ) who deduces it from a statement in ( @xcite , 22 ) . ) in our adlic setting , this should correspond to the assertion that every double coset has a representative in @xmath27 .
let @xmath28 denote the lattice @xmath29 of algebraic co - characters of @xmath16 .
given @xmath30 , and a valuation @xmath17 denote by @xmath31 the element @xmath32 . recall that @xmath30 is called _ antidominant _ if @xmath33 for each simple root @xmath34 ( see [ section : fundamental_representations ] ) .
precisely stated , the main result of this article is the following : [ theorem : global_reduction ] every double coset in @xmath35 has a unique representative of the form @xmath36 , where @xmath37 is antidominant . in [ sec : global_proof ] , we will deduce theorem [ theorem : global_reduction ] from the following local result which is proved in
[ sec : local_proof ] .
let @xmath38 be the local field @xmath39 of laurent series in @xmath40 with coefficients in @xmath1 .
it contains , as its ring of integers , the discrete valuation ring @xmath41 $ ] , and as a discrete subring , the polynomial ring @xmath42 $ ] .
let @xmath43 .
[ theorem : local ] every double coset in @xmath44 has a unique representative of the form @xmath45 where @xmath37 is antidominant .
the main results proved in this article should be known to the experts , but we have not found them in the literature beyond the case of @xmath10 , for which theorem [ theorem : local ] is proved in ( @xcite , 3 ) .
the results proved in this paper have played an important role in the author s work @xcite , as well as in the work of other authors on @xmath46 @xcite .
let @xmath47 be a vector space defined over @xmath1 .
let @xmath48 be a basis of the free @xmath49-module @xmath50 ( so that @xmath50 is isomorphic to the free @xmath49-module generated by the @xmath51s ) . given a vector @xmath52 , we may write @xmath53 , uniquely , with @xmath54.define @xmath55 [ lemma : canonicity ] if @xmath56 then @xmath57 .
let @xmath58 be the matrix of @xmath0 with respect to the basis chosen above .
let @xmath59 . if @xmath60 , then @xmath61 and @xmath62 & \leq & \sup\limits_{1\leq j \leq n}\ \sup\limits_{1\leq i \leq
n } |x_i g_{ij}| & \mbox { ( ultrametric inequality)}\\[1pc ] & \leq & \sup\limits_{1\leq j \leq n}\ \sup\limits_{1\leq i \leq n } |x_i| & \mbox { ( since } g_{ij } \in
{ \mathbf{o}}\mbox{)}\\[1pc ] & = & \| \mathbf{x } \| . \end{array}\ ] ] hence @xmath63 we may apply the same reasoning to @xmath64 to show that @xmath65 therefore , @xmath66 [ cor : canonicity ] @xmath67 the norm @xmath68 is independent of our choice of basis of @xmath50 . the coordinates of a vector with respect to two different bases differ by a matrix with entries in @xmath49 .
the argument in the proof of lemma [ lemma : canonicity ] shows that the norms with respect to two different bases are equal .
[ lemma : ultrametric ] the norm @xmath68 satisfies the ultrametric triangle inequality , i.e. , for vectors @xmath69 , @xmath70 in @xmath71 @xmath72 write @xmath73 and @xmath74 . then @xmath75 & \leq \sup\{\sup\{|x_1|,|y_1|\ } , \ldots , \sup\{|x_n|,|y_n|\}\}\\[.3pc ] & = \sup\{|x_1|,|y_1| , \ldots , |x_n|,|y_n|\}\\[.3pc ] & = \sup\ { \|\mathbf{x}\|,\|\mathbf{y}\| \}.\end{aligned}\ ] ] for a scalar @xmath76 and a vector @xmath77 @xmath78 [ lemma : bound ] if @xmath79 then there is a constant @xmath80 such that for any vector @xmath77 @xmath81 suppose that @xmath82 has matrix @xmath58 , and @xmath69 has coordinates @xmath83 with respect to the basis @xmath84
. then @xmath85 therefore , let @xmath86 [ lemma : boundone ] if @xmath87 is a non - zero vector then @xmath88 . by corollary [ cor : canonicity ]
, we may assume that the elements @xmath51 of a basis used to define @xmath68 lie in @xmath89 .
then at least one coordinate of @xmath69 is non - zero in @xmath90 . but
any non - zero element in @xmath90 has norm at least one .
therefore , @xmath88 .
[ prop : lower_bound]@xmath67 for any non - zero vector @xmath91 and any @xmath92 there is a positive constant @xmath93 such that for all @xmath94 @xmath95 consequently , for any subset @xmath96 of @xmath97 the set @xmath98 has a positive minimal element .
applying lemma [ lemma : bound ] to @xmath64 , and lemma [ lemma : boundone ] to @xmath99 ( which lies in @xmath100 ) , we have @xmath101 the second part of the assertion follows by noting that the values taken by the norm @xmath102 are of the form @xmath103 , where @xmath104 is an integer .
let @xmath105 be the simple roots with respect to @xmath14 in the root system @xmath106 of @xmath0 with respect to @xmath16 .
let @xmath107 be the weyl group of @xmath0 with respect to @xmath16 .
to each simple root @xmath34 , we associate an element @xmath108 of order two in @xmath109 in the usual way . given a subset @xmath110 of @xmath111 , let @xmath112 denote the subgroup of @xmath109 generated by @xmath113 , and let @xmath114 denote the parabolic subgroup @xmath115 of @xmath0 containing @xmath14 .
this group has a levi decomposition @xmath116 where @xmath117 is a reductive group of rank @xmath118 and @xmath119 is the unipotent radical of @xmath114 .
@xmath120 is a borel subgroup for @xmath117 containing the split torus @xmath16 .
the set of simple roots of @xmath117 with respect to @xmath120 is @xmath121 .
denote by @xmath122 ( resp .
, @xmath123 , @xmath124 ) the parabolic subgroup ( resp . , levi subgroup , unipotent subgroup ) corresponding to the set @xmath125 .
these are the maximal proper parabolic subgroups of @xmath0 containing @xmath14 .
* @xcite * [ theorem : godement ] there exist irreducible finite dimensional representations @xmath126 of @xmath127 vectors @xmath128 that are unique up to scaling , and characters @xmath129 for @xmath130 all defined over @xmath131 such that 1 .
[ item : delta]_@xmath122 is the stabilizer of the line generated by @xmath132 and @xmath133 for each @xmath134 for @xmath130 . _ 2 .
[ item : mu]_the restrictions @xmath135 to @xmath16 of @xmath136s are antidominant weights of @xmath16 with respect to @xmath137 which generate @xmath138 as a vector space over the rational numbers .
[ lemma : lifting_characters ] let @xmath139 be a levi subgroup of @xmath0 associated to a parabolic subgroup @xmath140 containing @xmath14 . then there is a canonical surjection @xmath141 1 if @xmath142 is a parabolic subgroup of @xmath0 containing @xmath14 and contained in @xmath143 then @xmath144 is a levi subgroup for @xmath139 corresponding to the parabolic subgroup @xmath145 of @xmath146 and @xmath147 .
given @xmath148 , we may use the iwasawa decomposition to write @xmath149 , where @xmath150 , @xmath151 and @xmath152 . moreover , if @xmath153 is another such decomposition , then , setting @xmath154 and @xmath155 , @xmath156 on the other hand , @xmath157 since @xmath139 normalizes @xmath158 , @xmath159 , and hence , setting @xmath160 , @xmath161 therefore @xmath162 , so that @xmath163 .
this shows that @xmath164 induces a well defined map @xmath165 .
it is clear that this map is surjective . to see that @xmath147 , note that we may write @xmath166 with @xmath167 , @xmath168 and @xmath169 .
but @xmath170 , so we may write @xmath171 , where @xmath172 and @xmath173 . therefore , we see that @xmath174 .
in the sequel we denote @xmath175 simply by @xmath176 .
define @xmath177 [ prop : omega]@xmath67 @xmath178 .
@xmath67 _ the rank one case _
( following @xcite ) : here @xmath0 has one simple root @xmath179 , and one fundamental representation @xmath180 and a vector @xmath181 such that for any element @xmath182 in the parabolic subgroup @xmath183 , where @xmath15 is the unipotent radical of @xmath14 , @xmath184 where the character @xmath185 ( defined over @xmath1 ) restricts to an anti - dominant weight @xmath186 on the maximal split torus @xmath16 . let @xmath187 .
we wish to show that @xmath188 . to this end , by proposition [ prop : lower_bound ] , and by replacing @xmath82 , if necessary by an appropriate element of @xmath189 , we may assume that @xmath82 has the property that @xmath190 write @xmath191 , where @xmath192 , @xmath193 and @xmath169 . by theorem [ theorem : godement ] and lemma [ lemma : canonicity ] ,
@xmath194 fix an isomorphism @xmath195 defined over @xmath1 , and let @xmath196 be such that @xmath197 .
choose @xmath198 in the nontrivial @xmath199-coset of @xmath200 .
note that if @xmath201 , then @xmath202 , therefore , using proposition [ prop : lower_bound ] , @xmath203 here @xmath204 , and its image is the root subgroup for @xmath205 .
the element @xmath206 lies in the derived group of @xmath0 which is isomorphic to either @xmath207 or @xmath208 in the rank one case . when the derived group of @xmath0 is isomorphic to @xmath207 , we may take @xmath209 to be the right action of @xmath207 on the space of @xmath210-matrices by right multiplication .
one may take the torus @xmath16 to consist of diagonal matrices in @xmath207 , @xmath14 the upper triangular matrices in @xmath207 and @xmath211 to be the vector @xmath212 . calculating with matrices
, one may verify that @xmath213 therefore , @xmath214 choose @xmath96 in @xmath90 such that @xmath215 . then @xmath216 .
suppose that @xmath217 .
then @xmath218 .
this is impossible , since @xmath219 .
it follows that @xmath220 .
therefore , ( [ eq : ineq ] ) can hold only if @xmath221 , which is the same as @xmath222 .
this completes the proof of proposition [ prop : omega ] when the derived group of @xmath0 is isomorphic to @xmath207 .
when the derived group of @xmath0 is isomorphic to @xmath208 , then @xmath0 is the product of its centre with @xmath208 .
therefore , the assertion of proposition [ prop : omega ] for @xmath0 follows from that for @xmath208 .
however , the assertion for @xmath208 follows easily from that for @xmath223 .
the derived group of @xmath223 is @xmath207 , hence the proposition holds for @xmath223 by the argument in the previous paragraph , completing the proof of proposition [ prop : omega ] in the rank one case .
_ the general case _
: let @xmath0 be a group of rank @xmath224 , and @xmath187 . by modifying @xmath82 on
the left by an element of @xmath225 , we may , for the purposes of this proof , assume , using the second assertion of proposition [ prop : lower_bound ] , that @xmath226 note that if @xmath227 , then @xmath228 .
since @xmath229^\times$ ] , @xmath230
. therefore , @xmath231 .
we may use the second assertion of proposition [ prop : lower_bound ] again , to assume , for the purposes of this proof , that @xmath232 while preserving ( [ eq : first_inequality ] ) . continuing in this manner
, we may assume that @xmath233 for @xmath234 .
therefore , it suffices to prove the following : [ lemma : otherthing ] if an element @xmath148 satisfies the inequalities @xmath235 for each integer @xmath236 then @xmath237 .
the proof of proposition [ prop : omega ] in the rank one case shows that lemma [ lemma : otherthing ] is true when @xmath0 is of semisimple rank one .
we prove it in general assuming the validity of theorem [ theorem : local ] in the rank one case .
suppose that @xmath82 satisfies the inequalities ( [ eq : jth_inequality ] ) for each @xmath238 .
write @xmath239 , with @xmath240 and @xmath152
. then @xmath241 can be written as @xmath242 , where @xmath243 and @xmath244 . since @xmath245 fixes @xmath132 ,
the inequalities ( [ eq : jth_inequality ] ) imply that @xmath246 from the rank one case , @xmath247 for some @xmath248 , @xmath249 and @xmath250 such that @xmath251 .
@xmath252 maps @xmath132 into @xmath100 . from lemma [ lemma : fundamental_inequality ]
it follows that @xmath253 equation ( [ eq : new ] ) implies that the above must be an equality .
this forces @xmath254 , and hence also @xmath255 .
write @xmath256 with @xmath257 and @xmath258 .
then viewing @xmath34 as a rational character of @xmath259 that is trivial on @xmath260 , we have @xmath261 repeating this argument for each @xmath262 completes the proof of lemma [ lemma : otherthing ] .
in order to prove the existence part of theorem [ theorem : local ] , it suffices to show that every element @xmath82 in @xmath263 may be written as @xmath264 , where @xmath265 , @xmath30 is antidominant and @xmath152 . to this end
, we may assume ( using the iwasawa decomposition ) that we are given @xmath237 , with @xmath266 , with @xmath257 and @xmath258 .
since @xmath82 , and hence @xmath267 , is in @xmath263 , @xmath268 , so that @xmath269 , for @xmath130 . for each root @xmath270 , let @xmath271 denote the corresponding root subgroup .
fix an isomorphism @xmath272 defined over @xmath1 . then for @xmath273 , we have @xmath274 therefore , if we write @xmath275 , where @xmath276 and @xmath277 , then @xmath278 given two positive roots @xmath279 and @xmath280 , the commutator @xmath281 $ ] is contained in the product of root subgroups @xmath282 where the @xmath283 are roots which can be written as positive linear combinations of @xmath279 and @xmath280 and are distinct from either @xmath279 or @xmath280 .
moreover , we may enumerate the positive roots as @xmath284 so that if @xmath285 , then @xmath286 can not be written as a sum of @xmath287 and any other positive roots .
write @xmath288 as @xmath289 .
then @xmath290 if we write @xmath291 , where @xmath292 $ ] and @xmath277 , then @xmath293 since @xmath294 , @xmath295 , and the image of @xmath296 normalizes all the subsequent root subgroups whose elements appear in the above expression , we may assume for the purpose of proving theorem [ theorem : local ] , that @xmath297 for @xmath298 .
we may continue in this manner to reduce @xmath299 to @xmath267 .
it is then easy to see ( using the decomposition @xmath300 ) that @xmath267 may be replaced by @xmath301 for @xmath250 . since @xmath251 , it follows that @xmath302 is antidominant , proving the existence part of theorem [ theorem : local ] .
we now prove the uniqueness part of theorem [ theorem : local ] . in order to do this , it suffices to show that if @xmath302 and @xmath303 are two dominant co - weights , and @xmath304 for some @xmath305 and @xmath152 , then @xmath306 . since the weights @xmath307 corresponding to the fundamental representations in theorem [ theorem : godement ] generate the vector space @xmath138 , it suffices to show that @xmath308 for each @xmath262 . in order to do this , we need the following : [ lemma : fundamental_inequality ] for any non - zero vector @xmath309 and any antidominant co - weight @xmath310 @xmath311 since @xmath16 is an @xmath1-split torus and @xmath312 is defined over @xmath1 , @xmath47 has a decomposition ( over @xmath1 ) into root subspaces @xmath313 where @xmath16 acts on @xmath314 by the character @xmath315 .
it is easy to see that @xmath135 is the lowest weight of @xmath16 occurring in @xmath126 , so that @xmath316 for any weight @xmath317 of @xmath16 occurring in @xmath126 and any antidominant co - weight @xmath318 . given any vector @xmath319 , we may write @xmath320 where @xmath321 and @xmath322 for each @xmath104 and the @xmath323s are not necessarily distinct . thus @xmath324 & = \sup_j\ { | \lambda_j(\pi^\mu)x_j| \}\\[.3pc ] & = \sup_j\ { q^{-\langle \lambda_j , \mu \rangle}|x_j|\}\\[.3pc ] & \geq q^{-\langle \mu_i,\mu \rangle } \sup_j\{|x_j|\}\\[.3pc ] & = \| \mathbf{v}_i \rho_i(\pi^\mu)\|\ ; \|\mathbf{v}\|.\end{aligned}\ ] ] since @xmath325 , this completes the proof of lemma [ lemma : fundamental_inequality ] .
lemma [ lemma : fundamental_inequality ] allows us to compare @xmath326 and @xmath327 : @xmath328 & \leq \frac{\|\mathbf{v}_i\rho_i(\gamma \pi^\eta)\|}{\|\mathbf{v}_i\rho_i(\gamma)\|}\\[.3pc ] & \leq \frac{\| \mathbf{v}_i\rho_i(\gamma \pi^\eta)\|}{\|\mathbf{v}_i\|}\\[.3pc ] & = \frac{\|\mathbf{v}_i\rho_1(\pi^\nu)\|}{\|\mathbf{v}_i\|}\\[.3pc ] & = q^{-\langle \mu_i , \nu \rangle}.\end{aligned}\ ] ] the first inequality is lemma [ lemma : fundamental_inequality ] applied to @xmath329 .
the second inequality follows from lemma [ lemma : boundone ] with @xmath330 .
interchanging the roles of @xmath302 and @xmath303 in the above arguments shows that @xmath331 for each @xmath262 .
this completes the proof of the uniqueness part of the assertion of theorem [ theorem : local ] .
if @xmath332 is an element of @xmath8 then , since @xmath333 for all but finitely many places @xmath17 of @xmath4 , we may assume , for the purpose of proving theorem [ theorem : global_reduction ] that @xmath82 is a finite product @xmath334 , with @xmath335 and @xmath336 , @xmath337 , for @xmath338 . by theorem [ theorem : local ] , there is a decomposition @xmath339 where @xmath340)$ ] , @xmath341 , and @xmath342 .
now @xmath343 and @xmath344 are both contained in @xmath11 and in @xmath345 for all @xmath346 .
therefore , by multiplying @xmath82 on the left by @xmath347 we get an element of the subset @xmath348 of @xmath8 . we have now reduced @xmath82 to an element with non - trivial entries only at most @xmath349 places and @xmath22 .
we may continue in this manner until the entries at all places except @xmath22 are trivial .
finally , the use of theorem [ theorem : local ] to @xmath350 gives us a representative each double coset of type asserted by theorem [ theorem : global_reduction ] .
the uniqueness part of the theorem follows from the corresponding assertion in the local situation , because two elements @xmath82 and @xmath351 of @xmath352 lie in the same double coset if and only if @xmath353 , with @xmath354)$ ] and @xmath355 .
we thank robert kottwitz for his guidance and r narasimhan and dipendra prasad for some historical and bibliographical information .
we thank the university of chicago and the centre de recherches mathmatiques for their support during the preparation of this article .
chevalley c , les poids dominants , in : sminaire c. chevalley , 19561958 . classification des groupes de lie algbriques , _ secrtariat mathmatique _ ( paris : 11 rue pierre curie ) ( 1958 ) pp .
16 - 01 to 16 - 09 laumon g , _ cohomology of drinfeld modular varieties .
( cambridge : cambridge university press ) ( 1997 ) ; automorphic forms , trace formulas and langlands correspondence , with an appendix by jean - loup waldspurger piatetski - shapiro i i , classical and adelic automorphic forms , an introduction , in : automorphic forms , representations and @xmath139-functions ( _ proc .
pure math . _
( corvallis , ore . : oregon state univ . ) ( 1977 ) ) part 1 , pp .
185 188 ; _ am .
: providence ) ( 1979 ) weil a , on the analogue of the modular group in characteristic @xmath182 , in : functional analysis and related fields , _ proc .
conf . for m. stone _ , ( chicago , ill . : university of chicago ) ( 1968 ) pp .
211223 ( new york : springer ) ( 1970 ) | let @xmath0 be a split reductive group over a finite field @xmath1 .
let @xmath2 and let @xmath3 denote the adles of @xmath4 .
we show that every double coset in @xmath5 has a representative in a maximal split torus of @xmath0 . here
@xmath6 is the set of integral adlic points of @xmath0 .
when @xmath0 ranges over general linear groups this is equivalent to the assertion that any algebraic vector bundle over the projective line is isomorphic to a direct sum of line bundles .
= msam5 at 10pt # 1artpath#1 |
the correlation function is one of the most frequently used statistics in cosmology .
conventionally the correlation function and the power spectrum are considered together to characterise the basic clustering properties of galaxies or of clusters of galaxies .
they complement each other in the sense that the power spectrum describes the structure in fourier space and the correlation function in real space .
intuitively , it is clear that , on small scales comparable to the size of systems of galaxies , the galaxy correlation function should characterise the distribution of galaxies in systems of galaxies . on these scales
the galaxy correlation function is large this is the manifestation of the clustering of galaxies . with increasing scale
the clustering gets weaker and at some scale the correlation function approaches zero . traditionally the correlation function is expressed in log log units and is calculated only up to the zero crossing separation .
this separation is sometimes used as a parameter characterising the transition from clustering to homogeneity .
the correlation function of clusters of galaxies has a similar behaviour , only it is shifted towards larger scales .
the clustering scale is here about a factor of 4 times larger than that for galaxies . on large scales
the correlation function of galaxies depends primarily on the distribution of systems of galaxies .
thus , properties of the distribution of these systems influence the behaviour of the correlation function on large scales .
similarly , the correlation function of clusters of galaxies on large scales depends on the distribution of systems of clusters , i.e. superclusters of galaxies .
einasto et al .
( 1997a , b , hereafter e97 and paper ii ) determined the correlation function of clusters belonging to rich superclusters using data in a distance interval @xmath0 @xmath1 mpc .
the main finding of this study was the detection of a series of almost regularly spaced maxima and minima .
this property of the correlation function was called _
oscillations_. the period of oscillations was found to be @xmath2 @xmath1 mpc .
this behaviour of the correlation function emphasises the presence of certain regularities in the distribution of clusters of galaxies .
a regular network of clusters of approximately the same scale was actually noticed by tully et al .
( 1992 ) and einasto et al .
( 1994 , 1997c , hereafter paper i ) .
these considerations have motivated us to perform an analysis of the correlation function as a geometrical statistic . such analysis helps us to understand the relation between the geometry of the distribution of clusters and superclusters and the correlation function . we shall use simple geometrical models of the particle distribution and calculate the correlation function for these models with the aim to find the relation between the correlation function and the geometry of the cluster distribution .
geometric information contained in the correlation function has been studied so far only for relatively small separations .
bahcall , henriksen , & smith ( 1989 ) used a model in which galaxies were located on the surfaces of spherical shells and clusters of galaxies resided at the intersections of the shells .
several models are discussed by einasto ( 1992 ) with the aim of giving a geometrical interpretation for the galaxy correlation function on small scales .
it was also noted that if the systems of galaxies are located in a regular grid then the correlation function has a secondary maximum at a scale which corresponds to the grid length ( step ) .
in contrast to previous studies we focus our attention in the present paper on large scales . on these scales the errors of the correlation function are of the same order as the function itself , thus a careful error analysis is crucial
. additionally we found for our models the power spectrum and the distribution of void radii in order to compare the relationship of these statistics to sample distribution geometry .
the paper is organised as follows .
we start the paper with a short description of the basic formulae and methods we use in calculations ( section 2 )
. then we use analytic models for the spectra and calculate respective correlation functions to get a general picture of the behaviour of the correlation function on large scales ( section 3 ) . thereafter we describe geometrical models of the distribution of clusters used to investigate properties of the correlation function ( section 4 ) .
the discussion of the correlation function for these models , the comparison of this function with the power spectrum and void radius statistics , and the discussion of the influence of the sample shape follow ( section 5 ) .
next we discuss errors of the correlation function ( section 6 ) .
finally we discuss geometric properties of the correlation function on small scales ( section 7 ) .
the paper ends with the main conclusions .
the two - point correlation function @xmath3 is defined as the excess over poisson of the joint probability of finding objects in two volume elements separated by @xmath4 and averaged over a very large volume ( peebles 1980 ) .
it is possible , in practice , to determine this function only for a limited volume , which is called the estimate of the correlation function .
we shall use the term `` correlation function '' for its estimate , and calculate it using the classical formula : @xmath5 where @xmath6 is the number of pairs of galaxies ( or clusters of galaxies ) in the range of distances @xmath7 , @xmath8 is the bin size , @xmath9 is the respective number of pairs in a poisson sample of points , @xmath10 and @xmath11 are the mean number densities of clusters in respective samples , and brackets @xmath12 mean ensemble average .
the summation is over the whole volume under study , and it is assumed that the galaxy and poisson samples have identical shape , volume and selection function .
we define our model samples in cubic volumes of side - length @xmath13 , and calculate the correlation function for the whole distance interval from @xmath14 to @xmath15 , dividing this interval linearly into 100 equal bins . for a given sample shape and selection function
we calculate @xmath16 as a mean value of ten realizations of random samples of 2000 clusters with the same selection function as the model sample . in this paper
we use only simple selection functions : in addition to cubical samples we consider double - conical samples ( see below ) . to suppress random noise we smooth the correlation function with a gaussian window . after smoothing
the poisson error is virtually absent and the residual error is determined by the cosmic variance of the geometry of the particle distribution , i.e. by the variation of the correlation function found for different volumes of space .
we calculate also the power spectrum , @xmath17 , for our models . for this purpose we can use the relation between the power spectrum and the correlation function , as they form a pair of fourier transformations : @xmath18 @xmath19 here the wavenumber @xmath20 is measured in units of @xmath21 , and is related to the wavelength , @xmath22 . to find the power spectrum from the correlation function we could use the eq .
however , it is more convenient to use direct methods which are similar to the calculation of the spectrum of @xmath23-body models of structure evolution , calculating first the distribution of clusters in the fourier space and then averaging over various wavenumber directions .
the distribution of clusters in @xmath20-space is found with the fast fourier transform using @xmath24 cells , which allows us to find 32 modes in the spectrum .
this method assumes that the whole space is filled with similar periodic cubes .
in addition to geometric models where the distribution of clusters in real space is given initially , we also find for comparison the relation between the correlation function and power spectrum analytically for some simple power spectrum models . in this case
we use the eq .
following eq . ( 2 ) we have calculated the correlation function for several power spectra . in general , for any given power spectrum the correlation function can not be calculated analytically . however , we can determine @xmath3 using a simple ansatz for @xmath17 : @xmath25 where @xmath26 is the power index on large wavelengths ( harrison - zeldovich spectrum ) , the negative @xmath27 is the index on small wavelengths , @xmath28 is the transition wavenumber , and @xmath29 is the normalisation amplitude . from observations @xmath27
is expected to be about @xmath30 .
thus , we have to integrate @xmath31 where @xmath32 , and @xmath33 depends on the place @xmath28 of the maximum of the power spectrum . the solution of the integral yields @xmath34 \bigr . \nonumber\\ & \bigl .
+ y^{-3/2 } [ \sqrt{\pi/2 } - \sqrt{2\pi } \hbox{s } ( \sqrt{2y/\pi } ) ] \bigr\ } , \end{aligned}\ ] ] where s denotes fresnel s sine - integral . obviously , the correlation function @xmath3 is oscillating
. the amplitude scales with @xmath35 and the radial dependence with @xmath36 .
the integrals in eq.(5 ) ( the power spectrum left and right of the maximum ) lead to approximately the same contribution in @xmath37 , however with opposite signs .
the resulting oscillations are rapidly damped ( @xmath38 ) . as a second model we consider a smooth transition between positive and negative spectral indices ( peacock & west 1992 , einasto & gramann 1993 ) @xmath39
the third model is the standard cold dark matter model ( scdm ) with the transfer function of bond and efstathiou ( 1984 ) , where @xmath40 .
the transition between positive and negative spectral indices is slower than in the second model .
the maximum of the first two models is chosen to be at @xmath41 mpc in accordance with the scdm model .
the amplitude is taken in accordance with cobe data by adopting @xmath42 ( white , scott & silk 1994 ) .
= 10.5 cm a rapid and precise method to calculate numerically the integral ( 2 ) is given by press et al .
the dispersion of density fluctuations is equal to the value of the correlation function at zero separation , @xmath43 in all cases of practical interest the spectral index is @xmath44 ; i.e. the integral ( 8) is not convergent .
since the dispersion of density fluctuations has a finite value , we conclude that on small scales the real spectral index must decrease so that the spectrum becomes steeper .
the physical reason for this behaviour is galaxy formation ; i.e. the power spectrum arises from objects of finite size .
this effect can be formally described by an exponential cutoff of the spectrum with an effective wavenumber @xmath45 @xmath46 mpc@xmath47 corresponding to a scale @xmath48 @xmath1 mpc .
this parameter does not change the correlation function on scales of interest for the present study .
the spectra and respective correlation functions are shown in figure 1 . a smooth transition between positive and negative spectral indices
can be approximated by a number of sections with constant spectral index @xmath49 .
the index changes at wave - numbers @xmath50 .
then the correlation function is a superposition of given analytic periodic functions with different periods @xmath51 , so the resulting correlation function does not oscillate ( gottlber 1996 ) .
it is evident that the amplitude of the correlation function is proportional to the amplitude of the power spectrum .
furthermore , in the first example with an analytical solution , the scaling of the correlation function with @xmath28 is obvious . in general
the power spectrum can be written as @xmath52 where @xmath29 is the normalisation constant , @xmath53 defines the place of the maximum ( @xmath54 in the ansatz ( 4 ) and @xmath55 for the cdm model , and @xmath56 is the transfer function .
thus the correlation function is a function of @xmath57 , @xmath58 where @xmath59 all correlation functions scale in the same manner as the analytic solution ( 6 ) .
let us finally consider a correction @xmath60 to the power spectrum @xmath61 which leads to @xmath62 .
obviously , this correction leads to superimposed oscillations of the correlation function if it has the form of the power spectrum eq .
( 4 ) ( see section 4.3 ) .
the addition of such a correction with a relative power of about 25 % leads already to significant oscillations in the correlation function .
in this section we describe and analyse the correlation functions and the power spectra for a number of geometrical toy models of the distribution of clusters of galaxies .
models are geometrical in nature ; i.e. in these models not the power spectrum is given initially but the distribution of clusters .
models can be divided into three basic types : random models , regular models , and mixed models . in all models
we explicitly generate clusters that in most models are clustered into superclusters , and designate the models correspondingly by capital letters cl or sc .
the following small letters in the model designation indicate the character of distribution ; there follow two digits which indicate the dispersion of the location of superclusters around the mean position .
these designations shall be described in detail below . for each model
we generate ten random realisations , calculate the mean smoothed correlation function , and estimate the parameters of the correlation function as described below .
we find also the local rms deviations of individual correlation functions from the mean value .
these deviations define the error corridor for the correlation function .
= 10.5 cm random models are designated by `` ran '' in the tables and figures .
we have generated random cluster , supercluster , and void models .
_ random cluster _
samples are generated with the goal of checking the formulae for the calculation of errors .
the number of clusters was taken to be from several hundred to several thousand .
_ random superclusters . _ in this model supercluster centres are positioned randomly in rectangular coordinates .
the multiplicity of the supercluster ( the number of clusters in supercluster @xmath63 ) was chosen randomly in agreement with the multiplicity distribution of real superclusters , as derived in paper i. the multiplicity distribution of real superclusters is from 1 ( isolated clusters like the virgo cluster ) to 32 ( the shapley supercluster ) .
for simplicity only superclusters of richness @xmath64 and 32 are generated .
member clusters of superclusters of richness 2 and higher are located around the centre of the supercluster randomly with an isothermal radial distribution .
actual superclusters are elongated and we have checked the influence of the elongation on the correlation function . in a model with elongated superclusters
negative sections of the correlation function have smaller amplitudes . positions of maxima and minima are not influenced .
the radius of superclusters was chosen as a slow function of the supercluster richness , @xmath65 .
the parameter @xmath66 determines the cluster correlation radius ; we used a value 15 @xmath1 mpc .
these parameters determine the cluster correlation function and the power spectrum on small scales ; values chosen yield good approximations to observed functions . in a cubic volume of size 700 @xmath1 mpc 1200 superclusters are generated , the total number of clusters being a random number around 3000 . in a sample of randomly located superclusters
some .superclusters are located closely together and merge . to find the actual multiplicity distribution of superclusters
a clustering algorithm was applied , as for real cluster samples using a neighbourhood radius 24 @xmath1 mpc . for details of supercluster definition
see paper i. _ random voids _ of diameter 100 @xmath1 mpc are distributed randomly and clusters of galaxies are located randomly outside voids , i.e. no superclusters are generated .
this diameter is equal to the mean diameter of voids in samples of rich clusters of galaxies ( einasto et al .
1994 , paper i ) .
_ the voronoi tessellation _ model is a variant of the random void model . in this case
centres of voids have also been located randomly , but in contrast to the previous model clusters are not distributed randomly outside voids . instead ,
void centres are considered as seeds , and clusters of galaxies are located in corners of a structure which is formed by expanding volumes of seeds .
superclusters in this model form by clustering of clusters if their mutual distances are smaller than the neighbourhood radius 24 @xmath1 mpc used in the definition of superclusters .
the voronoi model has been investigated by a number of authors ( van de weygaert & icke 1989 ; ikeuchi & turner 1991 ; williams , peacock & heavens 1991 ; subbarao & szalay 1992 ; van de weygaert 1994 ) .
we designate the model by `` vor '' .
it has one free parameter , the number density of void centres .
this determines the mean separation of superclusters across voids which is taken as the characteristic length of the model .
we generated ten models of box size 700 @xmath1 mpc with 431 void centres , in which case the characteristic scale is 115 @xmath1 mpc , a value observed in our study of superclusters and voids in paper i , and in the study of the cluster correlation function in paper ii .
results for the correlation functions and power spectra of random models are shown in figure 2 .
all random models have a correlation function with no sign of oscillations . the correlation function of the random supercluster model approaches zero on large scales immediately after the initial maximum .
the correlation function of the voronoi model is different .
it has a minimum at a separation between @xmath67 and @xmath68 @xmath1 mpc and one secondary maximum at @xmath69 @xmath1 mpc .
we discuss differences between the random supercluster and voronoi models later .
the analysis of the correlation function of real cluster samples has shown that it depends strongly on the richness of superclusters ( see paper ii ) .
we have thus made separate analyses of the correlation function for our random models for clusters located in rich and poor superclusters .
we shall discuss this problem in more detail below ( section 7 ) . in these models clusters or superclusters
are located , with a certain degree of fuzziness , on the corners or on the edges between corners ( rods ) of a regular rectangular three - dimensional grid .
the fuzziness of the model was realized by adding to positions of clusters or superclusters random shifts around the exact position on the corner or the edge of the grid .
depending on the degree of the fuzziness we can speak of strictly regular models with no fuzziness or of quasi - regular models with variable fuzziness ( see also einasto 1992 ) .
_ clusters on regular rods .
_ in this model clusters of galaxies are located randomly along rods , regularly spaced in a rectangular grid with a step @xmath70 @xmath1 mpc .
rods are randomly positioned around the exact values defined by the regular grid with a scatter @xmath71 @xmath1 mpc , which is equal to the scatter of mean diameters of voids defined by rich clusters of galaxies ( einasto et al .
1994 , paper i ) . in our naming convention ,
the model is designated by `` rod '' .
rods are cylindrical , and the diameter of cylinders was taken equal to be the mean minor diameter of superclusters , @xmath72 @xmath1 mpc ( einasto et al . 1994 , jaaniste et al .
rods cross at the corners of the grid , and at these points density enhancements form which can be considered as superclusters . _
superclusters on regular rods .
_ in this model instead of clusters , superclusters are located randomly along regularly spaced rods with step @xmath73 @xmath1 mpc . _ superclusters in corners of a regular grid .
_ in this case superclusters are located only in corners of a cubic lattice with constant grid step . in our naming convention , this model is designated as `` cor '' .
clusters of galaxies in superclusters are generated as in the random supercluster model .
random shifts to supercluster positions around corners are added with different values of the mean shift . _
superclusters in a regular net .
_ this model is a superposition of the model with superclusters located in corners of the net , and additional random clusters located along regular rods ; this model is designated as `` net '' .
clusters in rods represent isolated clusters and cluster filaments which are concentrated mostly between superclusters . here , too
, variable random shifts to mean positions in the regular net can be added .
the correlation functions and power spectra for regular models are shown in figure 3 .
the extreme case of regularity is a model with superclusters of equal multiplicity which are positioned exactly in corners of the regular grid .
the only free parameter it has is the step size of the grid .
the correlation function and the spectrum consist of a number of isolated peaks corresponding to mutual distances of all possible pairs of distances of corners of the grid .
the first peak of the correlation function is located in a separation equal to grid step , the next one to the shortest diagonal , and so on .
there exists no regular oscillations of the correlation function of the type observed in analytical models with sharp maximum in the spectrum .
_ this example shows that a strictly regular structure can not produce regular oscillations in the correlation function_.
= 10.5 cm if we add some flexibility to the regular corner model by adding random shifts to supercluster positions then neighbouring peaks of the correlation function join to form broader maxima .
adding clusters on rods makes the correlation function even more smooth .
one such example is shown in upper panels of figure 3 .
if we make the geometry of the model even more random and allow a random position of superclusters on rods we get a correlation with even more regular oscillations , as seen in the lower panels of figure 3 .
this example shows that in order to get a correlation function with regular oscillations there must be a lot of freedom in the supercluster positions .
these models are regular only in a restricted sense ( approximately equal step between rods ) .
with changing random shifts the amplitude of oscillations also changes .
it is larger the more accurately superclusters are positioned in corners or rods of the regular net .
the scatter of superclusters around the mean position used in our toy models is approximately equal to the scatter of diameters of voids between real superclusters .
thus our models should have correlation functions with an amplitude of oscillations which is close to the amplitude of oscillations of a population of real clusters , if clusters form a regular network of superclusters and voids with this scatter .
the analysis made in paper ii shows that this is indeed the case .
the analysis of the correlation function of real cluster samples has shown that clusters in high- and low - density environment have rather different correlation functions .
clusters in high - density environment ( the members of very rich superclusters ) have an oscillating correlation function whereas clusters in low - density regions ( members of poor superclusters and isolated clusters ) have an almost zero correlation at large separations ( paper ii ) . to model this behaviour of the correlation function we have constructed a mixed model .
the model has two populations of clusters : rich superclusters are located randomly in regular rods and form the quasi - regular population , poor superclusters are located randomly throughout the whole volume of the model .
for simplicity we use here randomly located superclusters as a low - density cluster population , taking into account the fact that the correlation function is not sensitive to details of the distribution of clusters in low density environments ( real clusters of the low density population lie near void walls , and void interiors are empty of rich clusters , see paper i ) .
one third of all clusters belong to the quasi - regular population , two thirds to the random one .
the correlation function and spectrum for this model is presented in figure 4 .
the correlation function is oscillating but with a very low amplitude .
the whole observed cluster sample has a similar correlation function , and the amplitudes of oscillations are also similar ( paper ii ) .
this example shows that our mixed model is a good approximation to the real cluster sample .
the mixed model is compared with a @xmath23-body model of structure evolution based on the double power - law spectrum with a sharp transition between short and long wavelength spectral indexes , eq .
the model was calculated similarly to the cdm model and was also used to estimate the cosmic variance of the correlation function ( see section 6 ) .
the model is designated as dps.6 .
the oscillatory behaviour of the double power - law model is less regular ; the first minimum of the correlation function has much larger amplitude .
amplitudes of maxima are approximately as high as in the case of the mixed model , the mean period of oscillations is also similar , but the scatter of individual periods is larger . to we shall use the following parameters to characterise oscillations of the correlation function : the mean separation of the first minimum from zero , @xmath74 ; the position and amplitude of the first secondary maximum , @xmath75 and @xmath76 ; differences between the subsequent maxima , @xmath77 ; and the mean value of differences , derived from the first 4 differences between maxima and from the first 4 differences between minima , @xmath78 .
scaling parameters can be used to derive the value of the true period of oscillations , @xmath79 , which we take to be equal to the grid step of the net of the quasi - regular model , and to the wavelength of the maximum of the spectrum , @xmath80 .
values of these parameters for models with at least one secondary minimum and maximum are given in table 1 .
they are derived from the mean smoothed correlation function of the particular model , calculated from ten realizations . to avoid
the decrease by smoothing the amplitude has been estimated from unsmoothed data .
this table shows that these parameters vary only within rather narrow limits .
the position of the first secondary maximum is almost the same for all geometrical models with identical grid step ( and period @xmath79 ) .
differences between maxima and minima are larger , but not much .
these differences are rather systematic : in all our geometric toy models @xmath81 , and @xmath82 .
the characteristic scale of the supercluster - void network , @xmath79 , can be calculated from the above parameters using the mean values of the following parameters : @xmath83 @xmath84 @xmath85 and @xmath86 as we see the most accurate determination of @xmath79 comes from the last equation .
in this section we continue the analysis of the correlation function and the power spectrum for different cluster distributions .
our main goal is to find which geometric properties of the distribution of superclusters can be detected on the basis of the correlation function and the power spectrum alone .
we investigate the influence of the shape of the volume occupied with clusters .
the available observed samples are generally not cubical in shape , and the shape of sample can , in principle , influence statistical properties of the samples .
the correlation functions with their error corridors and the respective power spectra with their error corridors are plotted for several models in figs 2 4 . on small scales
all correlation functions have a large maximum which reflects the fact that our models have clustering properties . on large scales
the correlation functions are different .
they have one of three principal forms : they become flat on large scales directly from the maximum on small scales ( random supercluster model ) ; have a minimum on intermediate scales followed by one secondary maximum , and then smoothly become flat ( voronoi model ) ; or obey an oscillatory behaviour with alternating secondary maxima and minima of decreasing amplitude ( models with a built - in regular structure of the type of a rectangular lattice )
. geometric interpretation of the correlation function of the first type is simple : here superclusters are distributed randomly and peaks from numbers of objects at various separations cancel each other out .
superclusters separated by large voids produce a correlation function with a minimum , whose location corresponds to the mean separation between superclusters and voids .
the minimum is followed by a secondary maximum corresponding to the distance between superclusters across voids . on still larger scales
there are no regularities in the supercluster - void network ( voids are spaced randomly ) , and the correlation function on very large scales is close to zero since superclusters at various separations cancel each other out as in the previous case .
this distribution is realized in the voronoi model .
the interpretation of the oscillating correlation function is also straightforward . in all models which produce an oscillating correlation function
the distribution of clusters on large scales is quasi - regular in the sense that high - density regions form a fairly regular network with an approximately constant grid length . in order to get regular oscillations
the geometric structure must not be too regular .
an example of very regular structure with superclusters located only in corners of the grid shows that the correlation function has a number of small maxima and minima without any regular oscillations .
quasi - regular oscillations are generated in all models with certain regularity and certain irregularity : the supercluster - void network must have a constant overall scale but individual superclusters must be located flexibly within this network .
the scale of the model is determined by the period of oscillations . to check the correspondence of the model to reality one can use small deviations of the correlation function from strict regularity expressed through parameters @xmath87 and @xmath88 . as we see ( paper ii ) , deviations of the observed correlation function from strict regularity are what is expected for models with constant scale and a certain scatter in individual superclusters positions .
this coincidence can not be accidental , since a visual inspection of the distribution of real superclusters also shows the presence of such a network of superclusters and voids ( paper i ) .
the basic difference between the regular and voronoi models lies in the correlation of positions of high - density regions on large scales : in the regular model positions are correlated and in the voronoi model not .
real cluster samples are limited to a certain maximal distance , and clusters in the galactic equatorial zone are not seen due to obscuration .
thus real observational samples generally have a double conical shape . to investigate the influence of such a sample shape we selected from cubical cluster samples double conical subsamples .
these subsamples have a conical height equal to half the size of the total sample , and the equatorial belt ( galactic latitudes less than @xmath89 ) is excluded .
the volume of these subsamples is about one - quarter the volume of the whole cube , and they contain about one - quarter the number of clusters than is in the original sample . in figure 5
we compare correlation functions of two models the regular net model and the random supercluster model calculated for the whole cubical volume , and for a double conical volume .
the first model is regular and the corresponding correlation function is oscillating ; the second model is not .
this figure shows that the oscillating properties of the sample are preserved in the smaller double conical sample .
only the error corridor is larger in the smaller samples , as expected .
this similarity of the oscillating properties is due to the fact that the small samples are large enough to contain in sufficient quantity the structural elements responsible for oscillations .
= 5.5 cm spectra for models are plotted in figures 2 4 .
power spectra have three regions , corresponding to inhomogeneities with high , medium , and low spatial frequency ( or small , medium , and long wavelengths ) .
this division is given by the maximum of the spectrum .
our calculations show that on small wavelengths ( large wavenumber @xmath20 ) spectra are rather similar .
this result is expected as perturbations on these wavelengths are determined by the internal distribution of clusters within superclusters , and superclusters in our models are generated in a similar way .
calculations by frisch et al .
( 1995 ) show that inhomogeneities on very large wavelengths do not influence the distribution of clusters in models , instead superclusters are modulated , i.e. in some regions they are very massive and in others poor .
regularity of the structure is given by power on intermediate wavelength .
analytic calculations confirm this result that models with only slightly different power spectra in the medium wavelength region may have very different correlation functions .
the mixed model with quasi - regular network of rich superclusters and voids has a spectrum which differs from the spectrum of the random supercluster model only by about 25 % near the maximum .
differences between power spectra of the voronoi and regular rod models are also small whereas correlation functions are very different .
such small differences can result from different geometries of the supercluster population .
these examples show that the correlation function is much more sensitive to the presence of a small regularity in the distribution of clusters than it is to the spectrum itself .
one feature of the correlation function is not yet explained .
why does the voronoi model show a minimum and maximum in the correlation function and the random supercluster model does not ?
one possible reason could be a difference in the distribution of voids . in the random supercluster model voids
can be very small , which is not the case for the voronoi model . to check this possibility we extracted voids from our model cluster samples and calculated the integrated distribution of void radii .
here we used procedures described by einasto , einasto & gramann ( 1989 ) and einasto et al .
results for the void radii are shown in figure 6 .
our calculations show that for most of our models the distribution of void radii is very similar .
a factor which influences the distribution of void radii is the presence of weak filaments between superclusters .
if clusters in high - density regions are distributed similarly in different models then their correlation functions also are similar , as they are , for instance , for models sc.cor.25 and sc.net.20 .
void diameters in these models are , however , completely different , as seen in figure 6 . in models with no filaments between superclusters ( sc.cor.25 )
voids are much larger their mean diameter corresponds to the diagonal of the net whereas in all models where rods between corners are also populated , void diameters are determined by distances between these rods , i.e. by the scale of the net .
we conclude that voids are determined by properties of small filaments and the correlation function by the distribution of clusters in high - density regions ( see also paper i ) .
-2 mm = 6.5 cm
the errors in the two- and three - point correlations are discussed by mo , jing & brner ( 1992 ) .
errors in the correlation function depend on high - order correlations , which are different for different geometrical distributions of clusters .
the difficulty of the determination of errors lies in the fact that they depend on the distribution of clusters in the ensemble , and there is no unique comparison ensemble of samples available .
thus parameters describing the errors of the correlation function must be determined for all different geometrical distributions of clusters . in the calculation of the correlation function
the number of particles in the poisson sample is taken as very large . in this case
we can ignore errors of the poisson sample and calculate the error of the correlation function from the error of the number of pairs @xmath90 .
mo et al . give a formula to calculate the error of @xmath90 through moments of the two - point and three - point correlation function .
the first terms have the form : @xmath91 here @xmath23 is the total number of clusters in the sample , and @xmath92 is a constant which depends on the high - order correlation functions . the first term of the eq .
( 16 ) is the poisson error , due to random errors in sampling of galaxies ; the second term is the cosmic error or variance , due to variations in the distribution of clusters in different parts of the universe . for our models
we have calculated errors in the correlation function from the scatter of different realizations of a particular model , and the eq .
( 16 ) was used only as a convenient parametrisation to determine the value of the parameter @xmath92 . as we see below this formula
is quite accurate and gives results in good agreement with direct estimates . to see better the dependence of errors on the number of clusters in the sample we find approximate formulae for random and cosmic errors of the correlation function on large scales . on these scales the correlation function @xmath93 , and thus @xmath94 .
we get for the random error in the correlation function : @xmath95 and for the cosmic variance : @xmath96 the number of object pairs at separation @xmath4 is proportional to the number of clusters squared , @xmath97 , thus @xmath98 .
random errors depend on the bin length @xmath8 and on the separation @xmath4 . to reduce random errors
we have smoothed the correlation function with a gaussian kernel of dispersion @xmath99 , which can be varied within certain limits . to the cosmic variance is independent of the bin size and scale .
it depends only on the geometry of the distribution of systems of clusters in the sample .
the constant @xmath92 increases with the size of the irregularities in this distribution .
the influence of random and cosmic errors is illustrated in figure 7 . in the left panel
we compare the effect of random and cosmic errors for samples of random clusters .
the scatter of individual values in the unsmoothed correlation function ( circles ) shows the effect of random errors , the overall deviation of open and filled circles from the mean correlation function ( solid curve ) is a visual demonstration of the effect of the cosmic variance .
the influence of the smoothing length is studied in more detail in the accompanying paper devoted to the analysis of real cluster samples ( section 4.4 in paper ii ) .
the number of clusters in observed samples is in some cases only several hundreds , thus random errors are larger . for observed samples a smoothing length about 15 @xmath1 mpc reduces the random noise in practice but does not influence parameters of the correlation function .
our models are designed for comparison with observations , thus we use a smoothing length 15 @xmath1 mpc in most model samples too . in the right panel of figure 7
we compare the error corridors found from the scatter of individual realizations of the random supercluster model and calculated from the eq .
( 16 ) or ( 18 ) ( both formula give practically identical results on large scales ) .
we see that the widths of both error corridors are approximately the same .
this shows that the eq . ( 18 ) is a good approximation of the cosmic variance . to derive the possible cosmic variance we have used not only geometrical toy models but also several @xmath23-body simulations of structure formation and evolution .
these simulations were performed with @xmath100 particles and @xmath101 grid cells using a particle - mesh code .
the size of the computational cube was taken to be @xmath102 @xmath1 mpc @xmath103 , where @xmath104 is the characteristic scale of the supercluster - void network .
clusters of galaxies were selected by a procedure similar to the `` friends - of - friends '' algorithm .
the lower limit of the cluster masses in the simulations is determined by the number of clusters in the sample .
the cube was divided into 8 subcubes of half the size of the original cube ; perturbations within subcubes can be considered as independent of each other .
the correlation function of clusters of galaxies was calculated for all subcubes , and the cosmic variance was found from the scatter of results of individual subcubes .
the power spectrum for the sample of clusters of galaxies was calculated as for all other cluster samples separately for all subcubes , and then the mean value and its scatter were found . in simulations we used a cdm spectrum with the standard transfer function ( bond & efstathiou 1984 ) .
this model is cobe normalized in amplitude at long wavelengths , and it has one free parameter , @xmath105 , depending on the total density of the model @xmath106 and the hubble constant @xmath46 . the model is designated as `` cdm.61 '' , where 6 designates the size of the computational box in units of the scale of the supercluster - void network , and 1 is for the standard high - density model ( @xmath107 ) . to estimate the value of the parameter @xmath92 from the cosmic variance eq .
( 16 ) we found the mean @xmath108 rms deviations of individual values from different realizations of the correlation function for the scale range @xmath109 to @xmath110 .
( here @xmath13 is the size of the computational box . ) in this scale range rms deviations are approximately constant .
the parameter @xmath92 is given in table 2 .
the parameter @xmath92 depends primarily on the regularity of the distribution on large scales . if there are no superclusters present ( random cluster and clusters in rods models ) , the parameter is less than 1 . in the voronoi model and regular net model without shifts
the error parameter has a value @xmath111 . in models with superclusters which have more freedom in their position ( say , to move along rods ) the parameter
has a larger value @xmath112 .
the cdm.61 model also has a value of the parameter @xmath92 close to this . in this case , however , the amplitude of oscillations is smaller than for the sample of clusters in high - density regions .
a net model ( superclusters in corners with additional clusters on rods ) with scatter 15 @xmath1 mpc around the mean position is probably closest to the real case . to determine from a sample of objects the presence of oscillations of the correlation function
, the cosmic error must be smaller than the amplitude of oscillations .
if we adopt @xmath113 and assume that oscillations have an amplitude 0.28 ( paper ii ) , then the sample must have at least 30 clusters .
if we have a mixed population with an expected amplitude of oscillation 0.05 , then such oscillations can be detected with confidence if the sample has at least @xmath114 clusters .
we see that to detect oscillating properties it is useful first to isolate the population with highest level of oscillating nature .
our main emphasis was the study of properties of the correlation function on large scales .
our models contain , however , information also on the correlation function on small scales .
thus it is of interest to have a look on small scales , too .
as it is well known , on small scales the correlation function can be approximated by a power law .
this law can be interpreted in terms of the fractal distribution of galaxies .
the fractal dimension is related to the power index ( szalay and schramm 1985 ) .
the fractal description is the principal geometric interpretation of the correlation function on small scales , and describes the shape of systems of galaxies . on small scales spheroidal systems of galaxies
dominate , whereas on larger scales the main structural elements are filaments of galaxies ( einasto 1992 ) .
it is well known that clusters of galaxies have much larger value of @xmath115 than galaxies ( bahcall & soneira 1983 , klypin & kopylov 1983 ) .
this result is usually interpreted as an indication of the _ stronger clustering _ of clusters of galaxies . however , einasto et al .
( 1994 ) and jaaniste et al .
( 1997 ) showed that the mean minor diameter of superclusters is 20 @xmath1 mpc .
we have calculated the correlation length separately for clusters located in rich and poor superclusters , both in geometrical models and cdm simulations . as in the real case , supercluster richness was determined by the clustering analysis .
our calculations show that the correlation length is very different for these populations , about 17 @xmath1 mpc for clusters in poor superclusters and about 45 @xmath1 mpc for clusters in rich superclusters .
similar results come from the analysis of real clusters in paper ii .
thus , a correct geometric interpretation of the correlation length of clusters of galaxies is to say that systems of clusters of galaxies ( superclusters ) are _ larger _ than systems of galaxies ( clusters ) since the parameter @xmath115 depends primarily on the size of the respective systems .
our study shows that the correlation function has the following basic properties . on large scales
the correlation function of clusters of galaxies depends on the geometry of the distribution of superclusters .
if superclusters are located in a quasiregular net ( supercluster - void network ) then the correlation function has an oscillatory behaviour .
the period of oscillations is equal to the length of the step size of the net .
the amplitude of oscillations depends on the scatter of distances between superclusters in the net and on the presence of a more smoothly distributed population of clusters .
parameters of oscillation and the scale of the net can be determined from observations if the number of clusters is sufficient to get an error corridor of the correlation function which is smaller than the amplitude of oscillations . for a quasiregular population
this minimal number of objects is about 30 , and for a heterogeneous population it is about 1000 , depending on the fraction of the quasiregular component in the whole population of clusters .
if superclusters are located in the corners of a random cellular void network ( voronoi model ) then the correlation function has a minimum , and its location corresponds to the mean separation between superclusters and voids .
there follows a secondary maximum corresponding to the clustering of clusters on opposite sides of voids .
if the distribution of superclusters is random then the correlation function is featureless and becomes flat after the initial highly positive value .
oscillating properties of the correlation function are related to the shape of the power spectrum near the maximum of the spectrum .
oscillations occur only in the case when the spectrum has a sharp maximum and sudden transition of the spectral index from a positive value on large wavelengths and a negative one on short wavelengths .
thus we see that the correlation function is indeed a useful statistic of the geometry of large scale structures .
observationally , we have employed the large scale correlation function of a sample of abell clusters ( e97 , paper ii ) ; we have exploited these results in a study of the correlation function of las campanas redshift survey galaxies ( tucker et al .
1995 , 1997 ) .
this work was supported by estonian science foundation grant 182 and international science foundation grant llf100 .
je and as were supported in potsdam by the deutsche forschungsgemeinschaft ; as was supported by the russian foundation for basic research grant 96 - 02 - 17591 .
we thank r. van de weygaert for his voronoi tessellation model program , and k .- h .
bhning for his help to prepare figures .
bahcall , n.a .
, & soneira , r.m . , 1983 , apj 270 , 20 bahcall , n.a . , henriksen , m.j . , & smith , t.e , 1989 , apj 346 , l45 bond , j.r . , & efstathiou , g. , 1984 , apj 285 , l45 einasto , j. , einasto , m. , gottlber , s. , mller , saar , v. , starobinsky , a.a . , tago , e. , tucker , d. , andernach , h. , & frisch , p. , 1997a
, nature 385 , 139 ( e97 ) einasto , j. , einasto , m. , frisch , p. , gottlber , s. , mller , saar , v. , starobinsky , a.a .
, tago , e. , tucker , d. , & andernach , h.,1997b , mnras ( paper ii ) einasto , j. , einasto , m. , & gramann , m. , 1989 .
mnras 238 , 155 einasto , j. , einasto , m. , gramann , m. , & saar , e. , 1991 .
mnras 248 , 593 einasto , j. , & gramann , m. 1993 , apj 407 , 443 einasto , m. , 1992 , mnras 258 , 571 einasto , m. , einasto , j. , tago , e. , dalton , g. , & andernach , h. , 1994 , mnras 269 , 301 einasto , m. , tago , e. , jaaniste , j. , einasto , j. , & andernach , h. , 1997c , a&a ( in press , paper i , astro - ph/9610088 ) frisch , p. , einasto , j. , einasto , m. , freudling , w. , fricke , k.j . , gramann , m. , saar , v. , & toomet , o. , 1995 , a&a 296 , 611 gottlber s. , 1996 , mapping , measuring and , modelling the universe , eds .
coles p. , martinez v.j .
, pons - borderia m .- j . , astr . soc . of the pacific conference
series vol.94 , 301 ikeuchi , s. , & turner , e. l. , 1991 , mnras 250 , 519 jaaniste , j. , tago , e. , einasto , m. , einasto , j. , & andernach , h. , 1997 , ( in preparation ) klypin , a.a . , & kopylov , a.i . , 1983 ,
astron . lett . 9 , 41 mo , h.j . ,
jing , y.p . , & brner , g. , 1992 , apj 392 , 452 peacock , j.a . , & west , m.j . , 1992 .
mnras 259 , 494 peebles , p.j.e . , 1980 , the large scale structure of the universe , princeton univ . press press , w. h. , teukolsky , s. a. , vetterling , w. t. & flannery , b. p. , 1992
, numerical recipes , cambridge univ .
press subbarao , m. u. , & szalay , a. s. , 1992 , apj 391 , 483 szalay , a.s . , & schramm , d.n .
, 1985 , nature 314 , 718 tucker , d.l . , mller , v. , gottlber , s. , oemler , a. jr . ,
kirshner , r.p . ,
lin , h. , shectman , s.a . , landy , s.d . , & schechter , p.l . , 1995 ,
27 , 1365 tucker , d.l . , oemler , a. jr . , kirschner , r.p . ,
lin , h. , shectman , s.a . , landy , s.d . ,
schechter , p.l .
, mller , v. , gottlber , s. , & einasto , j. , 1997 , mnras ( in press ) tully , r. b. , scaramella , r. , vettolani , g. , & zamorani , g. 1992 , apj 388 , 9 van de weygaert , r. 1994 , a&a 283 , 361 van de weygaert , r. , & icke , v. 1989 , a&a 213 , 1 white , m. , scott , d. , & silk , j. 1994 , ara&a 32 , 319 williams , b. g. , peacock , j. a. & heavens , a. f. , 1991 , mnras 252 , 43p | we investigate properties of the correlation function of clusters of galaxies using geometrical models .
we show that the correlation function contains useful information on the geometry of the distribution of clusters . on small scales
the correlation function depends on the shape and the size of superclusters . on large scales
it describes the geometry of the distribution of superclusters .
if superclusters are distributed randomly then the correlation function on large scales is featureless .
if superclusters have a quasi - regular distribution then this regularity can be detected and measured by the correlation function .
superclusters of galaxies separated by large voids produce a correlation function with a minimum which corresponds to the mean separation between centres of superclusters and voids , followed by a secondary maximum corresponding to the distance between superclusters across voids .
if superclusters and voids have a tendency to form a regular lattice then the correlation function on large scales has quasi - regularly spaced maxima and minima of decaying amplitude ; i.e. , it is oscillating .
the period of oscillations is equal to the step size of the grid of the lattice .
we also calculate the power spectrum and the void diameter distribution for our models and compare the geometrical information of the correlation function with other statistics .
we find that geometric properties ( the regularity of the distribution of clusters on large scales ) are better quantified by the correlation function .
we also analyse errors in the correlation function and the power spectrum by generating random realizations of models and finding the scatter of these realizations .
cosmology ; clustering
large - scale structure of the universe methods ; numerical |
the halo model is a useful language for discussing how galaxy clustering depends on galaxy type : galaxy bias ( see cooray & sheth 2002 for a review ) . to date , the halo model has been used to provide a useful framework for modeling the luminosity dependence of galaxy clustering .
the main goal of this paper is to extend the halo model description of galaxy luminosities to include colors .
this is an important step towards the ultimate goal of providing a description of how the properties of a galaxy , its morphology and spectral energy distribution , are correlated with those of its neighbors .
the hope is that , by relating such correlations between galaxies to the properties of their parent dark matter halos , the halo model will provide a useful guide in the study of galaxy formation .
the halo model description of the luminosity dependence of clustering is usually done in three rather different ways , which have come to be known as the ` halo occupation distribution ' ( hod ; jing , mo & brner 1998 ; benson et al .
2000 ; seljak 2000 ; scoccimarro et al . 2001 ; berlind & weinberg 2002 ; zehavi et al .
2005 ) the ` conditional luminosity function ' ( clf ; peacock & smith 2000 ; yang et al .
2003 ; cooray 2006 ; van den bosch et al .
2007a ) , and the ` subhalo abundance matching ' ( sham ; klypin et al . 1999
; kravtsov et al . 2004 ; vale & ostriker 2006 ; conroy , wechsler & kravtsov 2006 ) methods .
the hod approach uses the abundance and spatial distribution of a given galaxy population ( typically , just the two - point clustering statistics ) to determine how the number of galaxies depends on the mass of the parent halo .
this is done by studying a sequence of volume limited galaxy catalogs , each containing galaxies more luminous than some threshold luminosity .
the clf method attempts , instead , to match the observed luminosity function by specifying how the luminosity distribution in halos changes as a function of halo mass .
one can infer the clf from the hod approach , and vice - versa , so the question arises as to which is the more efficient description . for a given catalog , the hod method requires the fitting of just two free parameters , so it is relatively straightforward .
the clf method requires many more parameters to be fit simultaneously , but uses fewer volume limited catalogs .
shams first identify the subhalos within virialized halos in simulations , and then use subhalo properties to match the subhalo abundances to the observed distribution of luminosities .
once this has been done , clfs or hods can be measured in the simulations . in sph and semi - analytic galaxy formation models ,
central and satellite galaxies are rather different populations ( _ e.g. _ , kauffmann et al .
1999 ; sheth & diaferio 2001 ; guzik & seljak 2002 ; benson et al . 2003 ; sheth 2005 ; zheng et al .
2005 ) . and
so too , in the hod and clf approaches to the halo model , the central galaxy in a halo is assumed to be very different all the others , which are called satellites .
for example , the clf approach must provide a description of how the central and satellite luminosity functions vary as a function of halo mass .
the hod - based analyses predict that the satellite galaxy luminosity function should be approximately independent of halo mass , and hence of group and/or cluster properties ( skibba et al .
skibba et al .
( 2007 ) present evidence from the sdss in support of this prediction .
more recent analysis of a rather different group catalog has confirmed this finding ( hansen et al .
skibba et al .
argued that this independence can reduce the required number of free parameters in clf - based analyses .
one of the goals of the present work is to show that the hod - based approach also provides a rather simple way to understand how galaxy clustering depends on color .
in essence , it provides a simple algorithm for specifying how the joint clf ( _ i.e. _ , the luminosity distribution in two different bands ) varies with halo mass . in principle , this can be done by splitting the sample up into small bins of luminosity _ and _ color , and studying how the clustering signal in each bin changes .
zehavi et al .
( 2005 ) describe a first attempt at this for each bin in luminosity , they use two bins in color : ` red ' or ` blue ' .
( croton et al .
2007 also study the difference in clustering strengths of red and blue galaxies .
they use related statistics , but do not attempt a halo - model description of their measurements . ) as sample sizes increase , it will become possible to split the sample into many more color bins . however , even for this simplest case , zehavi et al . were led to a rather more complex parametrization of the hod than was necessary for the luminosities they caution that , as a result , there are more degeneracies amongst their parameter choices , and so the constraints on the hods they obtain are considerably weaker than for luminosities alone . while such a brute force approach to determining
the hod is certainly possible , we argue below that there may be some merit to recasting the problem as one in which the physics and statistics are more closely related . in essence , our approach exploits the fact that , to a good approximation , galaxies appear to be bimodal in their properties ( _ e.g. _ , blanton et al .
2003 ) . in the present context
, we are interested in the fact that the distribution of colors at fixed luminosity is bimodal ( _ e.g. _ , baldry et al .
2004 ; willmer et al .
our approach is to couple this bimodality with the centre - satellite split in the halo model .
this paper is organized as follows .
section [ model ] describes our approach : it shows the correlation between color and luminosity in the sdss sample , and then describes a model for the luminosities and colors of centrals and satellites which is designed to reproduce this bimodality .
section [ twotests ] describes how to use our model to generate mock catalogs which have the correct luminosity dependence of clustering and the observed color - magnitude relation , as well as how to incorporate our approach into a halo model description of the color - mark two - point correlation function .
section [ compare ] provides a comparison of our model predictions with measurements from the sdss .
these include the clustering signal from ` red ' and ` blue ' galaxies ( defined as being redder or bluer than a critical luminosity dependent color ) and the clustering signal when galaxies are weighted by color the color - mark correlation function .
a final section summarizes our findings . throughout , the restframe magnitudes we quote
are associated with sdss filters shifted to @xmath0 ; the absolute magnitude of the sun in this @xmath1-band filter is 4.76 ( blanton et al . 2003 ) . where necessary , we assume a flat background cosmological model in which @xmath2 , the cosmological constant is @xmath3 , and @xmath4 .
we write the hubble constant as @xmath5 km s@xmath6 mpc@xmath6 .
in addition , we always use ` log ' for the 10-based logarithm and ` ln ' for the natural logarithm .
color in the sdss .
smooth curves show that , at fixed luminosity , the distribution is well modeled by the sum of two gaussian components . ]
baldry et al .
( 2004 ) report that the distribution of rest - frame @xmath7 color at fixed @xmath1-magnitude can be well - modeled as the sum of two gaussian components .
the same is true of the distribution of rest - frame @xmath8 color ( _ e.g. _ , blanton et al .
2005 ) ; we call these the red and blue components of the distribution @xmath9 .
the mean and rms values of these components depend on luminosity .
this dependence is quite well described by simple power laws : @xmath10 the fraction of objects in the blue component decreases with increasing luminosity : @xmath11 and drops toward zero at the bright end .
figure [ bimodal ] shows this bimodality , and the two gaussian component fits which are based on these expressions .
our model of the bimodality , which motivates an algorithm for constructing mock catalogs , and which our halo model calculation requires , uses the red and blue sequences given by equations ( [ reds ] ) and ( [ blues ] ) .
these sequences are also shown in a color - magnitude diagram , figure [ mr195cmd ] , along with the color - magnitude contours of one of the volume - limited sdss catalogs used in section [ compare ] . volume - limited sdss catalog .
solid lines show the mean values of the red and blue sequences ( equations [ reds ] and [ blues ] ) ; dashed line shows the satellite sequence ( equation [ csatseq ] ) , and dotted line shows equation ( [ colorcut ] ) which some authors use to divide the population into red and blue . ]
however , it is common to make a cruder approximation to this bimodality , by simply labeling galaxies as ` red ' if they are redder than @xmath12 and calling them ` blue ' otherwise ( _ e.g. _ , zehavi et al .
2005 ; blanton & berlind 2007 ) .
( the recent analysis of satellite galaxy colors by van den bosch et al .
2007b used a stellar mass - based split , which translates into a similar color cut as the one above , although their cut is slightly steeper with respect to @xmath1-band luminosity . ) in what follows , we will only use this sharp threshold when comparing our results to previous work .
the sdss colors ( and magnitudes ) have measurement errors which contribute to the rms of the red and blue sequences , especially at faint magnitudes .
however , the uncertainties in the @xmath8 galaxy colors in the sdss are typically less than 0.02 mags , so they are unlikely to significantly affect the constraints on the model .
since the measurement errors almost certainly do not correlate with environment , they are not expected to bias the measured color mark correlation functions shown in section [ compare ] ; they will , however , increase the error bars on the clustering signal .
we note , however , that there is an important systematic problem with the colors for which we do not correct : namely , a dusty spiral will appear redder if viewed edge - on rather than face - on .
in fact , a significant fraction of the objects called ` red ' are not the early - types which one typically associates with the ` red sequence ' ( bernardi et al .
mitchell et al .
( 2005 ) estimate that this fraction is of order 40% ( also see maller et al .
2008 ) . since this systematic also affects the luminosities , for which no halo - model analysis to date has yet made a correction , we have not done so here either . to illustrate our approach we will begin with an extreme assumption .
suppose that : ( i ) the bimodal color distribution is independent of halo mass ( by which we mean that the distribution of color at fixed luminosity is independent of halo mass ; the distribution of luminosities , of course , does depend on halo mass ) , and that ( ii ) satellites are drawn from the red part of the bimodal color distribution _ no _ satellites come from the blue sequence .
later in this paper , we will find it necessary to relax the second assumption , but the data does not yet require us to give up the first .
we think assumption ( ii ) is a useful extreme which helps bring into focus the key points of the approach .
given the constraints from the color distribution as a function of luminosity ( section [ sdss ] ) and from luminosity - dependent clustering ( appendix [ zzhod ] ) , these two assumptions allow one to model the halo mass dependence of the colors of both centrals and satellites , and in general to build a model of how galaxy clustering depends on color , without any additional free parameters .
for example , these assumptions imply that the mean satellite color is @xmath13 whereas the first equality is the definition , the final expression shows how one might estimate the left - hand side from a knowledge of the luminosity distribution in halos of mass @xmath14 and the mean color at given luminosity in such halos .
if the distribution of satellite colors at fixed satellite luminosity is independent of halo mass ( this is not unreasonable , given that the distribution of luminosities themselves is approximately independent of halo mass ; see skibba et al .
2006 , 2007 ; hansen et al .
2008 ) , then this becomes @xmath15 thus , given @xmath14 , we integrate over the distribution of satellite luminosities , weighting by @xmath16 .
our simplest model ( assumption ii ) uses equation ( [ reds ] ) , the color magnitude relation along the red sequence , for @xmath16 .
we will show later that setting @xmath17 instead , which is bluer at faint luminosities ( see figure [ mr195cmd ] ) , provides substantially better agreement with the observations .
this is best thought of as a model in which satellites are drawn from the red sequence with probability @xmath18 and from the blue sequence with probability @xmath19 these expressions imply that , for sdss @xmath8 colors , @xmath20 at @xmath21 , and it drops to zero at @xmath22 .
since the fraction of galaxies that are satellites has a similar dependence on luminosity ( we provide explicit hod - derived expressions for this later ) , this model says that although almost sixty percent of the galaxies with @xmath23 are from the blue sequence ( c.f .
equation [ fblue ] ) , slightly less than twenty percent of the galaxies with @xmath21 are blue satellites : only a third of the faint blue galaxies are satellites , the others are centrals . allowing for blue - sequence satellites modifies the discussion below trivially .
it is worth reiterating that , in this model , satellite colors only depend on halo mass because satellite luminosities do .
since @xmath24 depends only weakly on @xmath14 ( skibba et al . 2007 ) , we expect @xmath25 to also depend only weakly on @xmath14 . in practice
, we do not evaluate the integral in equation ( [ cmsatint ] ) as written .
rather , we use a variation of the trick we used in skibba et al .
namely , for some function @xmath26 of @xmath27 , @xmath28 skibba et al . studied the case where @xmath29 , so the inner integral gave @xmath30 . here , we wish to set @xmath26 to be that function of @xmath27 which , when integrated over @xmath27 , yields @xmath31 .
thus , @xmath32 where we have defined @xmath33 if color and luminosity are in magnitudes ( _ i.e. _ , we work in logarithmic rather than linear variables ) then the integral is simpler : @xmath34 where @xmath35 , and @xmath36 is the slope of the relation showing how the mean satellite color changes with magnitude .
that is , @xmath37 or @xmath38 if satellites are drawn from the red sequence ( c.f .
equation [ reds ] ) or from equation ( [ csatseq ] ) .
obtaining an expression for the typical color associated with the central galaxies of @xmath14 halos is more complicated .
although the bimodal distribution of color at fixed luminosity can be thought of as arising from a mix of objects which lie along a blue or a red sequence , in what follows , it will be more useful to think in terms of the central - satellite split . in this case , @xmath39 making @xmath40 .
\label{ccen}\ ] ] if , as we assumed for the satellites , the distribution of central galaxy colors at fixed luminosity is independent of halo mass ( the results of berlind et al .
2005 support this assumption ) , then the mean color as a function of halo mass is simply @xmath41 if there is no scatter between central galaxy luminosity and halo mass ( _ e.g. _ , zehavi et al .
2005 ) . if there is scatter ( _ e.g. _ , zheng et al .
2007 ) , then @xmath42 now , by hypothesis , @xmath43 is given by equation ( [ reds ] ) ( or equation [ csatseq ] ) , whereas @xmath44 is simply the mean color of all galaxies as a function of luminosity .
thus , both these quantities are observables , or are constrained by observables , for the satellites ( skibba 2008 ) ; the only unknown is @xmath45 .
since both numbers are counted in the same volume , this is the same as the ratio of the number densities : @xmath46 .
we discuss how this ratio is determined by the luminosity - based hod in appendix [ zzhod ] .
it is worth noting that the quantity in square brackets in equation ( [ ccen ] ) is negative .
this means that , in general , the colors of central galaxies are _ bluer _ than the average for their luminosities .
although this seems counter to intuition
one is used to thinking of central galaxies as being red it is , in fact , sensible .
essentially , the paradox is resolved when one realizes that the satellites actually inhabit more massive halos than do centrals of the same luminosity .
it may help to note that this effect is most pronounced at low @xmath27 , where the mean color is significantly bluer than the red sequence , and the number of satellites can be large .
low luminosity galaxies that are centrals are hosted by low mass halos , whereas satellite galaxies of similar luminosity are more likely to reside in groups or clusters , so their parent halos are more massive .
thus , our model has placed blue central galaxies in low mass halos and red satellite galaxies in massive halos . at higher luminosities , @xmath44 approaches that of the red sequence . in this limit ,
the term in square brackets becomes small , as does the number of satellites , so the colors of central galaxies tend to @xmath44 : that is , our model places luminous central galaxies on the red sequence .
we now describe two ways to test our model of the bimodality .
the first is numerical we provide an algorithm for constructing mock catalogs which are consistent with our model .
the model can be tested by performing the same analysis on the mocks that was performed on real data .
this is particularly useful for analyses which are somewhat involved or contrived , so that an analytic description is difficult .
the second is analytic we show how our model can be implemented to provide a halo model description of mark correlations when the mark is color . skibba ( 2008 ) describes the result of a third test : a direct measurement of central and satellite colors in group catalogs .
the analysis above shows that one can generate a mock galaxy catalog in two steps : first generate luminosities , and then use them to generate colors .
note that the method used for generating luminosities is _ not _ important : the luminosities could have come from an hod analysis , a clf analysis , or they may be based on a sham .
our algorithm for generating luminosities comes from skibba et al .
briefly , we specify a minimum luminosity @xmath47 which is smaller than the minimum luminosity we wish to study .
we then select the subset of halos in the simulation which have @xmath48 .
each halo is assigned a central galaxy with luminosity given by inverting the relation between halo mass and luminosity ( equation [ lcen ] ) .
we specify the number of satellites the halo contains by choosing an integer from a poisson distribution with mean @xmath49 .
the luminosity of each satellite galaxy is specified by generating a random number @xmath50 distributed uniformly between 0 and 1 , and finding that @xmath27 for which @xmath51 .
this ensures that the satellites have the correct luminosity distribution .
we could assign colors to each of the satellites by drawing a gaussian random number with mean and rms given by inserting the satellite luminosity in equation ( [ reds ] ) for the red sequence .
however , as we show in section [ compare ] , this results in a correlation between color and environment that is too strong compared to the data . instead , we want the satellites to have colors which are bluer than the red sequence at faint luminosities , as specified by equation ( [ csatseq ] ) . to implement this in our mock catalog ,
we draw a uniformly distributed random number @xmath52 .
the satellite is drawn from the red sequence ( a gaussian with mean and rms given by equation [ reds ] ) if @xmath53 , where @xmath54 is given by equation ( [ predsatl ] ) and from the blue sequence ( gaussian with mean and rms from equation [ blues ] ) otherwise .
note that only the luminosity matters for determining the color ; the halo mass plays no additional role .
the colors for central galaxies can also be drawn from either the red or blue sequence . to determine which
, we draw another uniformly distributed random number @xmath55 . if @xmath56 , where @xmath27 is the central object s luminosity , then the object is assigned to the red sequence , so we draw a gaussian with luminosity - dependent mean and rms given by equation ( [ reds ] ) .
else , it is blue , and we use equation ( [ blues ] ) instead . equations ( [ fblue ] ) and ( [ fcen ] ) show that this assigns all central galaxies fainter than @xmath57 to the blue sequence .
finally , we place the central galaxy at the center of its halo , and distribute the satellites around it so that they follow an nfw profile ( see scoccimarro & sheth 2002 for how this can be done efficiently ) .
the resulting mock galaxy catalog has been constructed to have the correct luminosity function as well as the correct luminosity dependence of the galaxy two - point correlation function .
in addition , colors in this catalog are assigned in accordance with the model described previously : satellite and central galaxy colors are assigned such that the galaxy population as a whole has the correct color - luminosity distribution .
color for central galaxies ( red histogram ) , satellite galaxies ( blue dashed histogram ) , and all galaxies ( centrals+satellites ; black dotted histogram ) in a mock catalog with @xmath58 .
the distributions are shown for four intervals in log halo mass , indicated in square brackets in each panel . ]
our model makes a prediction for how the bimodality in color differs for central and satellite galaxies . in figure
[ mockgmrbimodal ] , we show the color distribution as a function of halo mass of central and satellite galaxies in a mock catalog with @xmath58 .
we normalize the central and satellite galaxy distributions by the total number of galaxies in each bin ; consequently , the lower mass halos are dominated by central galaxies , while satellites contribute most of the galaxies in massive halos .
first , note that the satellite distribution is almost the same in each panel : this is a consequence of our assumption that the distribution of satellite colors at fixed luminosity is independent of halo mass ( _ i.e. _ , @xmath59 , and the fact that satellite luminosities are approximately independent of mass as well .
on the other hand , the centrals have a more bimodal distribution in low - mass halos , while in massive halos most of them are on the red sequence .
second , it is interesting that the blue and red modes of the central galaxy bimodal color distribution are closer together than those of the satellite color distribution , such that the blue bump of the centrals tends to peak at the minimum in the satellite distribution .
this halo - model based prescription for making mock catalogs uses three simplifying assumptions which are worth discussing explicitly .
first , although we assume halos are spherical and smooth , the density run of satellites around halo centers is almost certainly neither . generating triaxial distributions is straightforward once prescriptions for how the triaxiality depends on halo mass and how it correlates with environment are available .
once these are known , they can be incorporated into the analytic halo - model description ( smith , watts & sheth 2006 ) .
similarly , parametrizations of halo substructure can also be incorporated into the description ( sheth & jain 2003 ) .
of course , both these types of correlations can be included in the mock catalog directly from a simulation if one simply selects the appropriate number of particles from the halo itself , rather than generating the profile shape synthetically .
this is costly because now one needs the full particle distribution , rather than just the halo catalog , to generate the mock but note that it is not a problem of principle .
second , note that the number of galaxies in a halo , the spatial distribution of galaxies within a halo , and the assignment of luminosities all depend only on halo mass .
none of these depend on the surrounding large - scale structure .
therefore , the mock catalog includes only those environmental effects which arise from the environmental dependence of halo abundances .
this point was made by skibba et al .
( 2006 ) ; it is also true of our prescription for including colors .
third , halos of the same mass will have had a variety of formation histories .
some will have assembled their mass and their galaxy populations more recently than others .
recent assembly means less time for dynamical friction , and , possibly , a younger stellar population .
so , at fixed halo mass , one might expect to find a correlation between the age of a halo and the galaxy population within it .
in particular , the number of galaxies in a halo , their luminosities and their colors may all be correlated with the formation history .
our halo model description ( and associated mock catalog ) ignores all such correlations . to see this clearly ,
note that we assign luminosities and colors to the galaxies in a halo without regard for the number of galaxies in it .
had we used a sham to assign luminosities , then some of correlation between formation history and the galaxy population will have been included .
if one is already carrying along the particle distribution from the simulation to construct the mock , then the next level of complication is to also include additional information about the merger history in the simulation , for use when making the mock .
we also assign colors to satellite galaxies without explicit consideration of the color of the central galaxy , and we make no effort to incorporate color gradients within a halo into our model .
this is mainly because the two - point statistics we study in this paper , weighted or unweighted , are known to be not very sensitive to gradients ( see sheth et al .
2001 , scranton 2002 , sheth 2005 and skibba 2008 for more discussion and simple prescriptions for incorporating color gradients . ) these are all interesting problems for the future ( and they are almost certainly not independent problems ! ) , but the measurements described in the next section do not require these refinements .
mark correlations are an efficient way to quantify the correlation between the properties of galaxies and their environment ( sheth , connolly & skibba 2005 ) .
the two - point mark correlation function is simply @xmath60 where @xmath61 is the traditional two - point correlation function and @xmath62 is the same sum over galaxy pairs separated by @xmath1 , but now each member of the pair is weighted by the ratio of its mark to the mean mark of all the galaxies in the catalog ( _ e.g. _ , stoyan & stoyan 1994 ; beisbart & kerscher 2000 ) .
in effect , the denominator divides - out the contribution to the weighted correlation function which comes from the spatial contribution of the points , leaving only the contribution from the fluctuations of the marks . in models where a galaxy s properties correlate with environment only
because they correlate with host halo mass , but halo abundances correlate with environment , it is relatively straightforward to write down a halo model of mark correlations ( sheth 2005 ) .
since our model of central and satellite colors is precisely of this form , we can build a halo model of color - mark correlations .
appendix b provides a detailed description of how this is done . in principle ,
comparison of this prediction with measurements in the sdss dataset allow a test of our approach .
color ( bottom ) mark correlation functions in a real - space mock catalog in which @xmath58 .
solid curves show the halo model predictions . ] before performing this test with data , figure [ gmrmock ] shows a comparison with measurements in the mock catalog described in the previous section .
the halo population is from the vls simulation ( yoshida et al .
2001 ) , and the mock galaxies have @xmath58 . luminosities and colors were assigned as described above : the top panel shows @xmath63 as a function of real - space separation when luminosity is the mark ; the bottom panel has @xmath8 as the mark .
solid curves show the halo model prediction , computed by inserting the mass dependence of the mean marks for centrals and satellites into the mark correlation formalism of appendix [ colormcf ] .
the luminosity and color mark correlations are significantly above unity , which clearly shows that in denser environments we expect the luminosities of galaxies to be brighter ( top panel ) and the colors to be redder ( bottom panel ) .
the mark correlations also clearly show the transition from the 1-halo term to the 2-halo term at @xmath64 , which is the virial radius of the most massive halos at @xmath65 .
the transition is more pronounced than in the traditional unmarked correlation function @xmath61 .
there is reasonably good agreement between the halo model calculation and the mocks for both the luminosity and color mark correlation functions ; the unmarked correlation functions @xmath61 agree extremely well , so they are not shown .
both panels in the figure show a similar but small discrepancy at similar scales , approximately where the 1 halo-2-halo term transition occurs .
although statistically significant , this discrepancy is small compared to the significance with which the signal itself differs from unity : the halo model calculations are qualitatively , if not quantitatively , correct across a wide range in scales .
the agreement between the model and the mocks is encouraging ; it suggests that much of the environmental dependence of galaxy color arises from the environmental dependence of host halo mass .
in this section we compare color mark projected correlation functions predicted by the halo model to measurements in the sdss ( york et al .
we use two volume - limited large - scale structure samples built from the nyu value - added galaxy catalog ( blanton et al .
2005@xmath66 ) from sdss dr4plus , which is a subset of sdss data release 5 ( adelman - mccarthy et al .
we @xmath67-correct the magnitudes to @xmath0 using the ` kcorrect v4_1 ` code of blanton & roweis ( 2007 ) ; the magnitudes are also corrected for passive evolution .
our fainter catalog has limits @xmath68 , @xmath69 ; it consists of 78356 galaxies with mean density @xmath70 .
our brighter catalog has @xmath71 and @xmath72 , and contains 73468 galaxies with mean density @xmath73 .
these luminosity thresholds approximately correspond to @xmath74 and @xmath75 , where @xmath76 is the break in the schechter function fit to the @xmath1-band luminosity function ( blanton et al .
2003 ) .
colors in the @xmath77 volume - limited catalog . ] for the measured correlation functions and jack - knife errors , which require random catalogs and jack - knife sub - catalogs , we use the hierarchical pixel scheme sdsspix , which characterizes the survey geometry , including edges and holes from missing fields and areas near bright stars .
this same scheme has been used for other clustering analyses ( scranton et al .
2005 , hansen et al .
2007 ) and for lensing analyses ( sheldon et al . 2007 ) .
figure [ markdist ] shows the distribution of @xmath8 colors in our fainter ( @xmath77 ) catalog .
the distributions of petrosian and model colors are similar , although the model colors are slightly redder .
the mean petrosian color is 0.796 , whereas the mean model color is 0.825 .
this is not unexpected galaxies have color gradients , and model colors measure the color on smaller scales .
these mean values are 0.850 and 0.885 in the brighter catalog ( @xmath58 ) .
the blue fractions of the petrosian colors of the fainter and brighter catalogs are , respectively , 44% and 37% using the fixed color - magnitude cut ( equation [ colorcut ] ) and 47% and 43% using the double - gaussian model ( equations [ reds]-[fblue ] ) , and they are @xmath78 lower for the model colors .
we now present our color mark correlation functions . in practice , in order to obviate redshift - space calculations in the halo model and redshift distortions in the data , we use the projected two - point correlation function @xmath79 where @xmath80 , @xmath81 and @xmath82 are the galaxy separations perpendicular and parallel to the line of sight , and we integrate up to line - of - sight separations of @xmath83 .
we estimate @xmath84 using the landy & szalay ( 1993 ) estimator @xmath85 where @xmath86 , @xmath87 , and @xmath88 are the normalized counts of data - data , data - random , and random - random pairs at each separation bin .
we then define the marked projected correlation function @xmath89 which makes @xmath90 on scales larger than a few mpc . for the sdss measurements , we used random catalogs with 10 times as many points as in the data ; the error bars show the variance of the measurements of 30 jack - knife sub - catalogs .
figures [ mcf195 ] and [ mcf205 ] compare the color marked correlation functions for the @xmath77 and @xmath58 catalogs with our predictions . the solid and open points show the measurements for petrosian and model colors .
the color mark signals in the bottom panels are stronger for petrosian colors , at the @xmath91 level , for both luminosity thresholds @xmath77 and @xmath58 . evidently , the environmental dependence of petrosian colors is stronger than that of model colors .
however , this is probably due to the fact that the red and blue peaks are slightly more displaced from one another for petrosian rather than model colors .
the correlation function of galaxies split by color is the measurement that has traditionally been used to show the environmental dependence of color ( _ e.g. _ , zehavi et al .
2005 ; tinker et al .
the top panel in figure [ mcf205 ] shows such measurements for galaxies redder and bluer than the color cut given by equation ( [ colorcut ] ) .
open squares and triangles are for measurements in the sdss and in a mock catalog constructed as described in the previous section .
( the sdss galaxies were split by their petrosian colors ; the measurement is virtually the same when they are split by model colors . )
the mock catalog is at @xmath92 , whereas the sdss measurements ( and corresponding theory curves ) are at @xmath93 .
therefore , to compare the clustering of red and blue galaxies in our mock with the measurements , we measure the ratio of @xmath94 to @xmath95 , and @xmath96 to @xmath95 in our @xmath92 mock .
we then assume that this ratio would be the same at @xmath0 as it is at @xmath92 ; the triangles show the result of applying this ratio to @xmath95 at @xmath0 ( _ i.e. _ , the filled circles ) they represent how the clustering of red and blue galaxies differ from the full sample in our mock .
the agreement between the clustering of the mock galaxies and sdss galaxies is very good , indicating that our model reproduces these traditional measurements of color dependent clustering very well .
color mark correlation function for @xmath77 .
points show sdss measurements for petrosian ( solid points ) and model colors ( open points ) , with jack - knife errors .
solid curves show the halo - model prediction when satellite galaxies can be drawn from either the red or the blue sequences ( equations [ csatseq][pbluesatl ] ) ; dashed curve shows the prediction if satellites are drawn from the red sequence only ( equation [ reds ] ) . ]
color mark correlation function , like figure [ mcf195 ] , but for @xmath58 . in the upper panel , the correlation functions for galaxies redder and bluer than the color cut ( equation [ colorcut ] ) are also shown , for the sdss galaxies ( open squares ) and mock catalog galaxies ( open triangles ) .
for clarity , error bars are only shown for the full sdss catalog . ]
because mark statistics do not require binning of the dataset into coarse bins in color , or coarse bins in density , the mark correlation functions shown in the lower panels of the figures contain significantly more information about environmental correlations than more traditional measures .
they allow the mark to take a continuous range in values , and they yield a clear , quantitative estimate of the correlation between the mark and the environment at a given scale .
mark statistics are also sensitive to the distribution of the marks : for example , for the fainter luminosity threshold ( @xmath77 ) , the color marks have a wider distribution than for the brighter threshold , so some galaxies have colors farther from the mean mark . because these outliers also tend to be in more extreme environments ,
the result is a stronger mark correlation ( _ e.g. _ , @xmath97 _ vs. _ 1.15 ) .
notice also that the mark correlation functions are more curved for the fainter luminosity threshold , with a more distinct transition between the 1-halo and 2-halo terms .
this is because there are more satellite galaxies in the fainter sample , and the mark clustering is more sensitive to their spatial distribution within halos .
note that the model in which no satellites come from the blue sequence ( dashed curve ) produces too strong a signal : galaxies in dense environments are too red . since most of these are satellites , this model places too many red satellites in massive halos .
the difference between the two models is greater for the faint luminosity threshold simply because there are more faint satellites , more of whose colors should be drawn from the blue sequence .
the model in which satellites come from a mix of the two sequences , though they are increasingly red at large luminosities ( equation [ csatseq][pbluesatl ] ) is in good agreement with the measurements on all scales where the statistic is reliably measured .
this suggests that this model of the colors of central and satellite galaxies is a reasonable one .
the good agreement between our model and the data also indicates that the correlation between halo mass and environment is the primary driver of the environmental dependence of galaxy color .
we have developed and tested a simple model for several observed correlations between color and environment on scales of @xmath98 .
our model is built upon the model of luminosity mark clustering of skibba et al .
( 2006 ) , in which the luminosity - dependent halo occupation distribution was constrained by the observed luminosity - dependent correlation functions and galaxy number densities in the sdss .
the model presented here has added constraints from the bimodal distribution of the colors of sdss galaxies as a function of luminosity .
we make two assumptions : ( i ) that the bimodality of the color distribution at fixed luminosity is independent of halo mass , and ( ii ) that satellite galaxies tend to follow a particular sequence in the color - magnitude diagram , one that approaches the red sequence with increasing luminosity ( equation [ csatseq ] ) .
alternatively , this assumption can be phrased as specifying how the fraction of satellites which are drawn from the red and blue sequences depends on luminosity ( equation [ pbluesatl ] ) .
one virtue of our model is the ease with which it allows one to include color information into mock catalogs .
adding colors to a code which successfully reproduces luminosity dependent clustering requires just four simple lines of code two for centrals and two for satellites ( section [ mock ] ) .
this is far more efficient than ` brute - force ' approaches which are based on fitting hods to fine bins in @xmath27 _ and _ color , or others which are based on using observed correlations between color and local density . since bimodality
is also observed at @xmath99 , it would be interesting to see if our approach is similarly successful at interpreting the measurements of coil et al .
( 2008 ) in the deep2 sample .
realistic colors are necessary for providing realistic training sets for galaxy group- and cluster - finding algorithms , and a number of groups are currently developing such mock catalogs .
so we think it is worth emphasizing that our approach can be applied to _ any _ mock catalog which produces the correct luminosity - dependence of clustering .
thus , although we phrased our discussion in terms of an hod - based mock , mocks based on clfs or shams could also use our method for generating colors .
in particular , cluster - finding algorithms that exploit information about brightest cluster galaxies ( bcgs ) , or galaxies positions from the red sequence , or galaxies redshift - distorted positions , or the multiplicity function or total luminosity or stellar mass of groups , could all be tested with mock catalogs constructed with the approach described in this paper .
we will be happy to provide our mock catalogs to those interested , upon request .
more generally , we feel that the simplicity of our approach makes it an attractive way to begin to include the entire sed into the halo model description , and hence into mock catalogs .
specifically , starting from our successful model for adding @xmath8 given @xmath27 , the next step might be to add , say , @xmath7 , given @xmath8 and @xmath27 again assuming that the distribution @xmath100 is independent of halo mass .
this is also attractive because we have shown that such an approach is easily described using the language of the halo model section [ halomark ] provides a halo - model description of the color - mark correlation function .
this facilitates the use of mark statistics in testing our hypothesis that the bimodal color distribution is independent of halo mass .
comparison of our mark correlation measurements with measurements in our mock catalogs and with our halo model calculations ( figures [ mcf195 ] and [ mcf205 ] ) suggest that if the bimodal color distribution is independent of halo mass , then at least some of the noncentral / satellite galaxies in a halo must be drawn from the blue sequence this fraction of blue satellites must be larger at low luminosities .
this is one of the key results of our paper . if satellites lie on the red sequence because their star formation has been quenched by processes such as ` strangulation ' ( _ e.g. _ , weinmann et al .
2006 ) , then our results suggest that quenching is still on - going at lower luminosities .
such processes are expected to modify the colors and star formation rates of satellite galaxies , but not their morphologies ; we investigate this further in a subsequent paper by measuring morphology mark correlations in the sdss galaxy zoo catalog .
we caution , however , that we , like all previous halo model analyses , have ignored the fact that inclination can affect the observed galaxy properties luminosities and colors in the present context .
corrections for inclination - related effects are available in the literature ( giovanelli et al .
1995 ; tully et al .
1998 ; sheth et al . 2003 ) , and they are not negligible .
recent work on this by maller et al .
( 2008 ) , which appeared while our work was being refereed , provides relatively straightforward corrections which may be reasonably accurate . for this reason
, our work should be viewed as attempting a halo - model description of the observed colors , rather than providing a truly physical picture of the intrinsic ( face - on ? ) colors . of course
, if the luminosities and colors had been corrected for inclination effects , we expect our analysis to also yield results which are closer to the true physical picture .
but because we have not yet included these corrections , we believe that statements about the physics of ` strangulation ' , especially at low luminosities , are premature .
we expect our model to be in good agreement with the findings of zehavi et al .
( 2005 ) , who analyzed volume - limited sdss samples after dividing galaxies into two bins in color .
they used a slightly redder color cut than did we to produce the measurements shown in the top panel of figure [ mcf205 ] .
they found that the fraction of central galaxies which lay blueward of this cut increased as @xmath27 decreased ; that there were no faint blue satellites ; and that , although there are blue satellites at intermediate and high @xmath27 , they were about a factor of five less common than red satellites in halos of the same mass .
our model is in qualitative agreement , with the mean central and satellite galaxy colors increasing with both luminosity and halo mass .
zehavi et al .
inferred from their results that the majority of bright galaxies are red centrals of massive halos , and that faint red galaxies are predominantly satellites in massive halos .
this is consistent with swanson et al .
( 2007 ) , who found that both luminous and faint red galaxies are more strongly clustered than moderately bright red galaxies .
we reach a similar conclusion , although not all faint red galaxies are satellites in massive systems : some are centrals in underdense environments .
we also expect our model to be in qualitative agreement with the findings of blanton & berlind ( 2007 ) .
these authors defined blue galaxies as those lying blueward of @xmath101 ( our equation [ colorcut ] ) .
they then found that the color magnitude relation for galaxies in luminous groups tended to have @xmath102 decreasing with group luminosity , but that the red and blue sequences were otherwise approximately independent of group luminosity .
they phrased their findings as showing that the color magnitude relation depends on group luminosity , presumably because they wished to draw attention to the dependence of @xmath102 on group luminosity . in light of the discussion above
, we think this is slightly misleading .
the red and blue sequences in our model are _ independent _ of group properties by construction . in our model ,
the decrease of the blue fraction in luminous groups is simply a consequence of the assumption that satellites tend to be drawn from the red rather than the blue sequence .
this happens because more luminous groups will tend to have more satellites _ and _ redder centrals ( because central galaxy luminosity increases with halo mass which is , in turn , strongly correlated with total luminosity , and luminous galaxies are red ) . since our model has mainly red satellites , the red fraction is larger in more luminous groups .
skibba ( 2008 ) describes the results of a direct comparison of our model predictions with the colors of centrals and satellites in group catalogs . in our model ,
_ all _ environmental correlations arise from the fact that massive halos tend to reside in denser environments ( mo & white 1996 ; sheth & tormen 2002 ) .
recent studies of the environmental dependence of halo assembly have shown that halo properties such as formation time and concentration are correlated with the environment at fixed halo mass ( sheth & tormen 2004 ; gao , springel , white 2005 ; wechsler et al .
2006 ; croton , gao , white 2007 ; wetzel et al .
2007 , keselman & nusser 2007 , zu et al . 2007 ) .
they have found that at fixed mass , halos in dense environments form at slightly earlier times than halos in less dense environments .
the success of our model suggests that such ` assembly bias ' effects are not the primary drivers of the environmental dependence of galaxy colors in the real universe , thus extending previous conclusions about the insignificance of assembly bias on galaxy luminosities ( skibba et al . 2006 ; abbas & sheth 2006 , 2007 ; blanton & berlind 2007 ; tinker et al .
2007 ) , at least for the relatively bright galaxies in the sdss .
further tests , such as analyses of luminosity and color mark statistics of catalogs constructed from semi - analytic models with known assembly bias , would shed more light on these issues , and are the subject of a subsequent paper .
our model does not include the galactic ` conformity ' reported by weinmann et al .
( 2006 ) , in which bluer centrals are likely to be surrounded by bluer satellites , at fixed halo mass . including this effect is the subject of work in progress .
the main quantitative predictions of our model , such as the _ mean _ central and satellite colors as a function of mass , and the correlations between color and environment , are not expected to be significantly affected by this phenomenon , however .
our model also does not include color gradients within halos it has long been known that satellite galaxies near halo centers tend to be redder than in the outskirts . in this case , satellite color marks depend on both the host halo mass and on their distance from the halo center .
halo model analyses show that this should only matter on small scales ( see discussion of fig . 4 in sheth et al .
2001 ; scranton 2002 ) ; for galaxy populations with many satellite galaxies , the 1-halo term of the color mark signal is expected to be slightly higher ( sheth 2005 ) .
skibba ( 2008 ) incorporates this effect , and does find such an increase at small scales .
finally , it is worth emphasizing that mark statistics are sensitive indicators of the correlations between galaxy properties and the environment , and as such are powerful tools for constraining galaxy formation models .
an analysis of marked correlation with star formation rate marks in the sdss and the millennium simulation is the subject of work in progress .
the halo - model description of marked statistics , based on the luminosity dependence of galaxy clustering , also has many applications . in a forthcoming paper ( skibba & sheth 2008 ) , we present a model of stellar mass mark correlations and analyze them with sdss measurements analogous to the color mark correlations presented here .
we thank xi kang and frank van den bosch for valuable discussions , and the aspen center for physics for hospitality in the summer of 2007 , where some of this work was completed .
we also thank tim mckay , the referee , for comments that helped to improve the quality of the paper .
this work was supported by nasa - atp nag-13720 and the nsf under grant ast-0520647 , and by hst - ar-10646 .
we thank jeffrey gardner , andrew connolly , and cameron mcbride for assistance with their ntropy code , which was used to measure all of the correlation functions presented here .
@xmath103tropy was funded by the nasa advanced information systems research program grant nng05ga60 g .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpa ) , the max - planck - institute for astrophysics ( mpia ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . 99 abbas u. , sheth r. k. , 2006 , mnras , 372 , 1749 abbas u. , sheth r. k. , 2007 , mnras , 378 , 641 adelman - mccarthy j. k. et al . , 2007 , apjs , 172 , 634 baldry i. k. , glazebrook k. , brinkmann j. , ivezi . , lupton r. h. , nichol r. c. , szalay a. s. , 2004 , apj , 600 , 681 beisbart c. , kerscher m. , 2000 , apj , 545 , 6 benson a. j. , bower r. g. , frenk c. s. , lacey c. g. , baugh c. m. , cole s. , 2003 , apj , 599 , 38 berlind a. a. , weinberg d. h. , 2002 , apj , 575 , 587 berlind a. a. , blanton m. r. , hogg d. w. , weinberg d. h. , dav r. , eisenstein d. j. , katz n. , 2005 , apj , 629 , 625 berlind a. a. , kazin e. , blanton m. r. , pueblas s. , scoccimarro r. , hogg d. w. , 2006b , astro - ph/0610524 , apj submitted bernardi m. , et al .
2003 , aj , 125 , 1817 blanton , m. r. et al .
, 2003 , apj , 592 , 819 blanton m. r. , eisenstein d. , hogg d. w. , schlegel d. j. , brinkmann j. , 2005 , apj , 629 , 143 blanton m. r. et al .
, 2005b , aj , 129 , 2562 blanton m. r. , berlind a. a. , 2007 , apj , 664 , 791 blanton m. r. , roweis s. , 2007 , aj , 133 , 734 coil a. et al .
2008 , apj , 672 , 153 conroy c. , wechsler r. h. , kravtsov a. v. , 2006 , apj , 647 , 201 cooray a. , sheth r. , 2002 , phys .
, 372 , 1 cooray a. , 2006 , mnras , 365 , 842 croton d. j. , norberg p. , gaztanaga e. , baugh c. m. , 2007 , mnras , 379 , 1562 croton d. j. , gao l. , white s. d. m. , 2007 , mnras , 374 , 1303 diaferio a. , kauffmann g. , colberg j. , white s. d. m. , 1999 , mnras , 307 , 537 gunn j. e. et al . , 1998 ,
aj , 116 , 3040 gunn j. e. et al . , 2006 , aj , 131 , 2332 guzik j. , seljak u. , 2002 , mnras , 335 , 311 hansen s. m. , sheldon e. s. , wechsler r. h. , koester b. p. , 2008
, astro - ph/0710.3780 jing y. , mo h. j. , brner g. , 1998 , apj , 494 , 1 kang x. & van den bosch f. c. , 2008 , apjl , 676 , 101 kauffmann g. , colberg j. m. , diaferio a. , white s. d. m. , 1999 , mnras , 303 , 188 keselman j. a. , nusser a. , 2007 , mnras , 382 , 1853 klypin a. , gottlber s. , kravtsov a. v. , khokhlov a. m. , 1999 , apj , 516 , 530 kravtsov a. v. , berlind a. , wechsler r. h. , klypin a. , gottlber s. , allgood b. , primack j. r. , 2004 , apj , 609 35 landy s. d. , szalay a. s. , 1993 , apj , 412 , 64 mitchell j. , keeton c. , frieman j. , sheth r. k. , 2005 , apj , 622 , 81 navarro j. f. , frenk c. s. , white s. d. m. , 1997 , apj , 490 , 493 park c. , choi y. y. , vogeley m. s. , gott j. r. , blanton m. r. , 2007 , apj , 658 , 898 peacock j. a. , smith r. e. , 2000 , mnras , 318 , 1144 scoccimarro r. , sheth r. k. , hui l. , jain b. , 2001 , apj , 546 , 20 scoccimarro r. , sheth r. k. , 2002 , mnras , 329 , 629 scranton r. , 2002 , mnras , 332 , 697 scranton r. et al . , 2005 , apj , 633 , 589 seljak u. , 2000 , mnras , 318 , 203 sheldon e. et al . , 2007 , astro - ph/0709.1153 sheth r. k. , diaferio a. , hui l. , scoccimarro r. , 2001 , mnras , 326 , 463 sheth r. k. , tormen g. , 1999 , mnras , 308 , 119 sheth r. k. , jain b. , 2003 , mnras , 345 , 529 sheth r. k. , 2005 , mnras , 364 , 796 sheth r. k. , connolly a. j. , skibba r. , 2005 , astro - ph/0511773 sheth r. k. , jimenez r. , panter b. , heavens a. f. , 2006 , astro - ph/0604581 skibba r. a. , sheth r. k. , connolly a. j. , scranton r. , 2006 , mnras , 369 , 68 skibba r. a. , sheth r. k. , martino m. c. , 2007 , mnras , 382 , 1940 skibba r. a. , 2008 , astro - ph/0805.1233 , mnras , submitted skibba r. a. , sheth r. k. , 2008 , in prep .
smith r. e. , watts p. i. r. , sheth r. k. , 2006 , mnras , 365 , 214 stoyan d. , stoyan h. , 1994 , fractals , random shapes , and point fields .
wiley , chichester swanson m. e. c. , tegmark m. , blanton m. , zehavi i. , 2007 , astro - ph/0702584 tinker j. l. , conroy c. , norberg p. , patiri s. g. , weinberg d. h. , warren m. s. , 2007 , astro - ph/0707.3445 vale a. , ostriker j. p. , 2006 ,
mnras , 371 , 1173 van den bosch f. c. , yang x. , mo h. j. , weinmann s. m. , macci a. v. , more s. , cacciato m. , skibba r. , kang x. , 2007a , mnras , 376 , 841 van den bosch f. c. , aquino d. , yang x. , mo h. j. , pasquali a. , mcintosh d. h. , weinmann s. m. , kang x. , 2007b , astro - ph/0710.3164 wang y. , yang x. , mo h. j. , van den bosch f. c. , 2007 , apj , 664 , 608 wechsler r. h. , zentner a. r. , bullock j. s. , kravtsov a. v. , allgood b. , 2006 , apj , 652 , 71 weinmann s. m. , van den bosch f. c. , yang x. , mo h. j. , 2006 , mnras , 366 , 2 yang x. , mo h. j. , van den bosch f. c. , 2003 , mnras , 339 , 1057 yang x. , mo h. j. , van den bosch f. c. , 2008 , apj , 676 , 248 york d. g. et al . , 2000 , aj , 120 , 1579 yoshida n. , sheth r. k. , diaferio a. , 2001 , mnras , 328 , 669 zehavi i. et al . , 2005 , apj , 630 , 1 zheng z. et al . , 2005 , apj , 633 , 791 zheng z. , coil a. l. , zehavi i. , 2007 , apj , 667 , 760 zu y. , zheng z. , zhu g. t. , jing y. p. , 2007
, astro - ph/0712.3570
the main text outlines our model ; actual implementation of it depends on the form of the luminosity - based hod .
these are of two types either the relation between halo mass and central galaxy luminosity is monotonic and deterministic , or there is some scatter .
we use the parametrization of zehavi et al .
( 2005 ) to illustrate the former case , and that of zheng et al .
( 2007 ) to illustrate the latter .
the results described in the main text are not particularly sensitive to this choice , although the plots we are based on hods in which there is scatter .
to evaluate @xmath105 , suppose that the relation between halo mass and central luminosity is deterministic ( _ i.e. _ , there is no scatter around the @xmath106 relation ) .
then @xmath107 where halo model interpretations of sdss galaxy clustering suggest that @xmath108 ( zehavi et al .
2005 ; skibba et al .
2006 ) , and the halo mass function @xmath109 is described in sheth & tormen ( 1999 ) . the number density of satellite galaxies which have luminosity @xmath27 , whatever the mass of the parent halo , is given by differentiating @xmath110 with respect to @xmath27 .
zehavi et al .
( 2005 ) show that @xmath111 where @xmath112 is given by the expression above , and @xmath113 is a weakly increasing function of @xmath27 .
thus , @xmath114^{-\alpha_l } \nonumber\\ & & \
\times\ \int_{m_l}^\infty dm\ , { dn\over dm}\ , \left({m/23\over 10^{12}h^{-1}m_{\odot}}\right)^{\alpha_l},\end{aligned}\ ] ] so @xmath115 and the fraction of objects which are centrals is @xmath116 to see what these expressions imply , suppose that @xmath117 were independent of @xmath27 .
then @xmath118 and @xmath119 in this case , the mean satellite color is independent of halo mass .
if @xmath120 ( not far off from its actual value ) and @xmath121 for some fiducial value of @xmath122 ( halos more massive than @xmath123 are indeed exponentially rare ) , then @xmath124 making @xmath125 .
this ratio decreases as @xmath112 increases as @xmath27 increases , the ratio of satellites to centrals decreases , and the fraction of centrals increases .
in the analyses which follow , we use the actual halo model values of these quantities rather than these approximations .
a reasonable fit to the actual halo model values is given by @xmath126\right]\ ] ] this ratio tends to 0.7 at small luminosities , making the fraction of galaxies which are centrals at @xmath127 about 3/5 ( _ cf .
_ equation [ fcen ] ) , consistent with the satellite fraction @xmath128 of van den bosch et al .
( 2007a ) .
zheng et al . ( 2007 ) allow for stochasticity in the relation between halo mass and central galaxy luminosity .
they assume that @xmath129 ^ 2}{2\sigma_{\mathrm{log}l}^2}\biggr ] , \ ] ] and then set @xmath130 \label{ncenm}\ ] ] and @xmath131 the poisson model for satellite counts sets @xmath132 their table 1 shows how all of the parameters in this hod vary with sdss @xmath1-band luminosity .
we have found that these scalings with @xmath133 are well approximated by @xmath134 as in zehavi et al .
( 2005 ) , the value of @xmath135 , which determines the critical mass above which halos typically host at least one satellite galaxy , is approximately independent of luminosity , while the @xmath136 slope @xmath137 , which characterizes the mass dependence of the efficiency of galaxy formation , increases with luminosity .
the two new hod parameters are @xmath138 and @xmath139 .
they are not constrained well and their uncertainties are large ( see zheng et al . for details ) , but our correlation functions and color mark correlation functions are not very sensitive to their exact values . for the two luminosity thresholds discussed in the main text , @xmath77 and @xmath58 , the parameters above are
@xmath140 and @xmath141 , and the effective value of @xmath142 , approximately independent of luminosity , is similar to the factor of 23 in the zehavi et al .
hod . for our purposes ,
the main difference with this hod model is the scatter in luminosity at fixed mass .
we first discuss how to construct a mock catalog that includes this scatter .
we then explain how our model of the color mark is modified . to account for the scatter between @xmath143 and @xmath144 in the mock catalogs
, we do not simply select the subset of halos in the simulation which have @xmath145 , as we do in the case of a sharp threshold ( e.g. section [ mock ] ) .
instead , we generate uniformly distributed random numbers @xmath146 between 0 and 1 for each halo of mass @xmath147 .
then we keep the halo if @xmath148 ( equation [ ncenm ] ) . as a result ,
only half of the halos with @xmath149 are kept , as are quite a few halos with @xmath150 .
larger values of @xmath138 increase the range of halo masses around @xmath151 and increase the total number of halos because the abundance of halos increases with decreasing mass .
our halo model of the color mark is also modified by the scatter between luminosity and mass , and hence @xmath152 is no longer a step function .
the central galaxy color mark , described in section [ model ] , is slightly more complicated .
the mean central galaxy color as a function of luminosity @xmath153 ( equation [ ccen ] ) depends on the number density of central galaxies as a function of luminosity @xmath154 ( equation [ ncenl ] ) , which now includes an integral : @xmath155 then the central galaxy color mark , which is used in the color mark correlation functions , is also an integral ( equation [ ccenm ] ) : @xmath156 the model of the color mark correlation functions , described in appendix [ colormcf ] , is also modified .
however , we reiterate that , in general , the correlation functions and color mark correlation functions are not sensitive to the exact amount of scatter in mass at fixed luminosity .
we perform our halo model calculations in fourier space .
the two - point correlation function is the fourier transform of the power spectrum @xmath157 in the halo model , @xmath158 is written as the sum of two terms : one that arises from galaxies within the same halo and dominates on small scales ( the 1-halo term ) , and the other from galaxies in different halos which dominates on larger scales ( the 2-halo term ) .
that is , @xmath159 where , @xmath160 , \label{pk1h } \\
p_{2h}(k ) & = & \biggl [ \int dm \,\frac{dn(m)}{dm}\ , \langle n_{\mathrm { cen}}|m\rangle \label{pk2h } \\ & \phantom{= } & \;\ ; \times \ , \frac{1\,+\ , \langle n_{\mathrm { sat}}|m\rangle\ , u_{\mathrm { gal}}(k|m)}{\bar n_{\mathrm { gal}}}\,b(m ) \biggr]^2 \ , p_{\mathrm { lin}}(k ) , \nonumber\end{aligned}\ ] ] where the number density of galaxies @xmath161 is ( _ cf_. , eq . [ nsatden ] ) @xmath162 \label{ngal}\ ] ] and @xmath163 is the fourier transform of the galaxy density profile .
it is standard to assume this has the same form as the dark matter , so we use the form for @xmath146 given by scoccimarro et al.(2001 ) .
the distribution @xmath164 is expected to be well - approximated by a poisson distribution ( _ e.g. _ , kravtsov et al .
2004 ; yang et al . 2008 ) , so we set @xmath165 .
the two parts of the 1-halo term in equation ( [ pk1h ] ) can be thought of as the ` center - satellite term ' and the ` satellite - satellite term ' . to describe the effect of weighting each galaxy
, we use @xmath166 to denote the fourier transform of the weighted correlation function . like the power spectrum
, we write this as the sum of 1- and 2-halo terms : @xmath167 .
since central and satellite galaxies have different properties , we weight central and satellite galaxies separately by their mean mass - dependent marks : @xmath168 and @xmath25 ( section [ model ] ) .
following sheth ( 2005 ) , we write @xmath169 , \label { wk1h}\\ \frac{w_{2h}(k)}{p_\mathrm{lin}(k ) } & = & \biggl [ \int dm \,\frac{dn(m)}{dm}\ , \langle n_{\mathrm { cen}}|m\rangle\,b(m ) \label{wk2h } \\ & & \frac{c_{\mathrm { cen}}(m)\,+\,\langle n_{\mathrm { sat}}|m\rangle\ , \langle c_{\mathrm { sat}}|m , l_{\mathrm { min}}\rangle\ , u_{\mathrm { gal}}(k|m ) } { \bar n_{\mathrm { gal}}\,\bar c } \biggr]^2 \ , , \nonumber\end{aligned}\ ] ] where we normalize by the mean color mark @xmath170 | successful halo - model descriptions of the luminosity dependence of clustering distinguish between the central galaxy in a halo and all the others ( satellites ) . to include colors ,
we provide a prescription for how the color - magnitude relation of centrals and satellites depends on halo mass .
this follows from two assumptions : ( i ) the bimodality of the color distribution at fixed luminosity is independent of halo mass , and ( ii ) the fraction of satellite galaxies which populate the red sequence increases with luminosity .
we show that these two assumptions allow one to build a model of how galaxy clustering depends on color without any additional free parameters than those required to model the luminosity dependence of galaxy clustering .
we then show that the resulting model is in good agreement with the distribution and clustering of colors in the sdss , both by comparing the predicted correlation functions of red and blue galaxies with measurements , and by comparing the predicted color mark correlation function with the measured one .
mark correlation functions are powerful tools for identifying and quantifying correlations between galaxy properties and their environments : our results indicate that the correlation between halo mass and environment is the primary driver for correlations between galaxy colors and the environment ; additional correlations associated with halo ` assembly bias ' are relatively small .
our approach shows explicitly how to construct mock catalogs which include both luminosities _ and _ colors thus providing realistic training sets for , e.g. , galaxy cluster finding algorithms .
our prescription is the first step towards incorporating the entire spectral energy distribution into the halo model approach .
[ firstpage ] methods : analytical - methods : statistical - galaxies : formation - galaxies : evolution - galaxies : clustering - galaxies : halos - dark matter - large scale structure of the universe |
the orion - kl region is a very dynamic system with complicated molecular distributions , complex velocity structure , and extreme temperature and density gradients .
orion - kl has traditionally been viewed as a source that displays a high degree of nitrogen - oxygen chemical differentiation " where oxygen - bearing molecules like methyl formate [ hcooch@xmath3 and nitrogen - bearing molecules like ethyl cyanide [ c@xmath1h@xmath2cn ] are spatially distinct .
it has been suggested that this chemical differentiation " can be driven by the molecular processes at work in these regions @xcite , with gas - phase reactions potentially accounting for the spatial separation of methyl formate and formic acid [ hcooh ] emission peaks in this source .
however , recent high - resolution observations of molecular emission in this region @xcite have called into question the traditional view of chemical differentiation " in this source . the most recent set of high spatial - resolution images presented by @xcite ( hereafter paper i ) indicate that while the emission peaks for o - bearing and n - bearing molecules are spatially - separated in orion - kl , the bulk of the complex molecular emission is co - spatial ; and that any spatial differences are limited to smaller , localized regions .
it is unclear from these observations whether chemical or physical processes shape these molecular distributions .
while gas - phase chemical processing is a likely factor , it is also possible that varying physical conditions , including temperature , density , and kinematics , could influence the molecular spatial distributions in these regions .
however , only a limited amount of high spatial resolution information is available for this source , making definitive conclusions regarding the relative influence of these factors extremely difficult to determine .
high resolution observations that probe the physical influence on the chemistry are necessary to disentangle and interpret these complicated processes . here
we extend the high spatial resolution observations of orion - kl reported in paper i to examine the physical conditions of this region . using observations of methanol [ ch@xmath4oh ]
, we probe the temperature and velocity structure of this region .
we also use the velocity information gained from these observations to develop a kinematic model for orion - kl , and examine the influence of these kinematics on the molecular distributions in this source .
we then compare this information with the molecular emission images reported in paper i. from these results , we draw conclusions as to the mechanisms that are shaping the spatial distributions for each molecule observed .
the observations , results , and discussion are presented below .
the observations were conducted with the combined array for research in millimeter wave astronomy ( carma ) , and the details of the observations are presented in paper i. in addition to those observations , we conducted observations in the carma c configuration in 2010 april .
these observations were comprised of three 4.5 hour tracks .
the observations have a phase center of @xmath5(j2000 ) = @xmath6 and @xmath7(j2000 ) = @xmath8 and have a synthesized beam of @xmath9 .
the @xmath10 coverage of the observations gives projected baselines of 3.9 - 91.0 k@xmath0 ( 13 - 304 m ) .
the correlator was configured to have six 31 mhz windows and six 125 mhz windows , all with channel spacings @xmath11450 khz .
the antenna - based gain calibration was done by self - calibrating on the sio maser in source i at 86.243 ghz .
the solution was then bootstrapped to the other windows .
phase offsets between each band and the sio band were calculated and removed by using observations of 0423 - 013 .
uranus was used to calibrate the absolute amplitudes of 0423 - 013 and these are accurate to within @xmath1120% .
the internal noise source was used to correct the passbands of each 31 mhz window , and observations of 0423 - 013 were used to correct the passbands of the 125 mhz windows .
the data were calibrated , continuum subtracted , and imaged using the miriad software package @xcite .
note that the flux of the point source bn ( see @xcite ) is similar across all array configurations , indicating that the accuracy of the absolute flux calibration between the different tracks is better than 95% . in order to characterize the temperature profile of this region , 13 methanol transitions corresponding to 11 separate spectral lines were observed simultaneously .
methanol is widespread throughout the orion - kl region , and therefore is an ideal tracer for the physical properties of the region .
these 13 transitions span 2.3 orders of magnitude in upper state energy , thus providing a broad range of samples .
calibration uncertainties between the lines were minimized by conducting all methanol observations simultaneously , leaving only the true emission differences as differentiating factors .
table [ tab : methanol ] lists the observed methanol transitions , and includes the rest frequencies , quantum numbers , symmetry label , upper state energy , and linestrength information for each transition .
lrrr 19@xmath12 - 19@xmath13 & 87,066.819 ( 10 ) & 467.7 & 0.20 + 18@xmath14 - 18@xmath15 & 87,094.825 ( 5300 ) & 1308.0 & 8.98 + 13@xmath16 - 13@xmath17 & 84,930.9460 ( 10 ) & 1123.3 & 4.01 + 2@xmath18 - 1@xmath19 & 96,739.363 ( 5 ) & 12.5 & 1.21 + 2@xmath20 - 1@xmath21 a+ & 96,741.377 ( 5 ) & 7.0 & 1.62 + 2@xmath20 - 1@xmath21 & 96,744.549 ( 5 ) & 20.1 & 1.62 + 2@xmath22 - 1@xmath23 & 96,755.507 ( 5 ) & 28.0 & 1.24 + 8@xmath24 - 9@xmath25 & 94,541.806 ( 5 ) & 131.3 & 2.24 + 2@xmath26 - 1@xmath22 a+ & 95,914.310 ( 5 ) & 21.4 & 1.21 + 19@xmath27 - 20@xmath28 a+ & 94,815.075 ( 50 ) & 684.8 & 4.55 + 19@xmath29 - 20@xmath30 & 94,815.075 ( 50 ) & 684.8 & 4.55 + 24@xmath31 - 23@xmath32 & 98,030.396 ( 118 ) & 888.5 & 6.44 + 24@xmath33 - 23@xmath34 & 98,030.396 ( 118 ) &
888.5 & 6.44 + [ tab : methanol ]
the results of these observations were used to determine the effects of temperature , continuum column density , and kinematics on the molecular emission in the orion - kl region .
a temperature map of the region was constructed based on the methanol observations .
the molecular emission maps presented in paper i were then compared to this temperature map .
the velocity information from the methanol observations was also used to examine the kinematics of the region .
the results for each of these aspects of the observations are presented below . -9@xmath25
ch@xmath4oh maps for @xmath35 of 2.5 - 13 km s@xmath36 .
contours are @xmath37 , @xmath38 , @xmath39 , ...
( @xmath40 = 194.5 mjy / beam ) .
the synthesized beam is in the lower left corner of panel ( i).[fig : mtoh - all ] ] maps of ch@xmath4oh emission are shown in figure [ fig : mtoh - all ] for the j=8@xmath24 - 9@xmath25 transition in @xmath35 increments of 2.5 - 13 km s@xmath36 . assuming that the lines are optically thin and that the methanol is in local thermodynamic equilibrium , the total column density of ch@xmath4oh can be found from the relationships @xmath41 and @xmath42 where @xmath43 is the integrated line intensity in jy / beam km / s
, @xmath44 is the rotational - vibrational partition function , @xmath45 is the upper state energy of the transition in k , @xmath46 is the rotational temperature in k , @xmath47 is the opacity correction factor ( see @xcite ) , @xmath48 is the beam filling factor ( see @xcite ) , @xmath49 and @xmath50 are the fwhm gaussian synthesized beam axes in arcseconds , @xmath51 is the product of the line strength and the square of the relevant dipole moment in d@xmath52 , @xmath53 is the transition frequency in ghz , and @xmath54/@xmath55 is the upper state column density divided by its statistical weight ( 2@xmath56 + 1 ) . a boltzmann diagram was used to calculate the rotational temperature of the ch@xmath4oh transitions . using the above equations , a plot of the natural log of equation ( [ eqn : nta ] ) versus @xmath45 was constructed .
a weighted least squares fit to this plot yields the rotational temperature , which is equal to the negative inverse of the slope of the line .
the use of this method does have some pitfalls which are described in detail by @xcite .
the most significant factor to consider when using this analysis method is that the @xmath45 values cover a sufficient range to ensure a reliable temperature determination ; this is not a concern in this work because of the large sample of lines observed .
also , the results of a boltzmann analysis are more robust with a larger sample set . therefore , in this work , the rotational temperature was only calculated in regions where at least 6 of the 11 lines were detected above the 3@xmath40 level .
figure [ fig : temperature ] shows the calculated rotational temperature for the orion - kl region at rest velocities of 2.5 - 11.5 km s@xmath36 .
figure [ fig : uncert ] shows the uncertainty associated with these results .
oh observations .
the bar on the right denotes the temperature scale in k. regions which are white ( black in online edition ) did not have enough emission to make an accurate temperature determination.[fig : temperature ] ] ] from these figures it can be seen that the ch@xmath4oh temperature measurements are not reliable toward the hot core , where the observational data were not of sufficient quality to enable a reliable temperature determination .
this is because the lower @xmath45 transitions become highly self - absorbed in this dense region .
thus , while ch@xmath4oh appears to be a good tracer for temperature in most regions , it fails in the regions with highest methanol column density . in order to get an overall temperature view of the region ,
the maps from this work were combined with the nh@xmath4 temperature maps of @xcite .
the maps were combined in image space using the miriad maths task @xcite , and the combined map is presented in figure [ fig : comb - t ] . there is little spatial overlap between the two sets of images except in the hot core region ; where there was spatial overlap , the results agree to within the uncertainties .
the peak temperature value determined was 415 k. @xcite also measured the temperature of the region with multiple ch@xmath4oh transitions at higher frequencies .
this work more successfully determined the temperature in the hot core region , but suffered from opacity and double emission peak issues in some of the more diffuse regions .
the general temperature gradient observed by @xcite and that found in this work are similar .
the results presented here provide a more comprehensive overview of the temperature of orion - kl .
oh and nh@xmath4 observations [ nh@xmath4 from @xcite ] .
the bar on the right denotes the temperature scale in k. regions which are white ( dark blue in online edition ) did not display enough emission intensity to make an accurate temperature determination.[fig : comb - t ] ] as with paper i , all conclusions regarding hcooch@xmath4 apply equally to dimethyl ether [ ( ch@xmath4)@xmath1o ] , as they are co - spatial on all observed scales .
figure [ fig : mef - t ] shows the naturally weighted hcooch@xmath4 moment map ( contours from paper i ) overlayed on the temperature map from above . from this map
we can see that most of the hcooch@xmath4 emission arises from cooler , less dense regions such as irc5 and irc6 .
there is some emission arising from the warmest regions ( hot core and irc7 ) ; however , the peaks of this emission are not coincident with the temperature peaks .
there is no temperature information corresponding to the most extended parts of the hcooch@xmath4 emission near the compact ridge .
this is likely due to these regions having low density , and the emission being extended and therefore resolved out in the ch@xmath4oh observations . ]
contours overlayed on the color scale temperature profile of orion - kl .
the contours are 3@xmath40 , 6@xmath40 , 9@xmath40 , ...
( @xmath40=4.1 mjy / beam km s@xmath36 ) .
the moment spans the entirety of the hcooch@xmath4 emission in velocity space ( 1.9 - 11.8 km s@xmath36 ) .
the color bar on the right is the key for the temperature scale in k. [ fig : mef - t ] ] as with paper i , all conclusions regarding c@xmath1h@xmath2cn apply equally to vinyl cyanide [ c@xmath1h@xmath4cn ] and methyl cyanide [ ch@xmath4cn ] , as they are all co - spatial on all observed scales
. figure [ fig : etcn - t ] shows the naturally weighted c@xmath1h@xmath2cn moment map ( contours from paper i ) overlayed on the temperature map . unlike hcooch@xmath4 ,
c@xmath1h@xmath2cn peaks in the warmest , densest regions ( hot core , sma 1 , and irc7 ) .
the c@xmath1h@xmath2cn peak at @xmath1112 km s@xmath36 ( see figure 10 of paper i ) is associated with the cooler region near irc6 .
h@xmath2cn ] contours overlayed on the color scale temperature profile of orion - kl .
the contours are 3@xmath40 , 6@xmath40 , 9@xmath40 , ...
( @xmath40=4.1 mjy / beam km s@xmath36 ) .
the moment spans the entirety of the c@xmath1h@xmath2cn emission in velocity space ( -12.8 - 18.4 km s@xmath36 ) .
the color bar on the right is the key for the temperature scale in k.[fig : etcn - t ] ] figure [ fig : ace - t ] shows the naturally weighted ( ch@xmath4)@xmath1co moment map ( contours from paper i ) overlayed on the temperature map .
( ch@xmath4)@xmath1co appears to have some characteristics in common with both hcooch@xmath4 and c@xmath1h@xmath2cn , as it has strong peaks near the warmest regions , but also has some extended structure in the cooler regions . )
@xmath1co ] contours overlayed on the color scale temperature profile of orion - kl .
the contours are 3@xmath40 , 6@xmath40 , 9@xmath40 , ...
( @xmath40=8.1 mjy / beam km s@xmath36 ) .
the moment spans the entirety of the ( ch@xmath4)@xmath1co emission in velocity space ( 2.4 - 8.7 km s@xmath36 ) .
the color bar on the right is the key for the temperature scale in k.[fig : ace - t ] ] figure [ fig : fa - t ] shows the hcooh moment map ( contours , from paper i ) overlayed on the temperature map .
hcooh has a unique distribution with regard to the temperature map compared to the other complex molecules observed .
it has one weak peak near the warmer regions , but the bulk of the emission is coming from the cooler , more extended regions .
full temperature measurements are not available for much of the hcooh distribution because large structures were resolved out in the ch@xmath4oh observations .
contours overlayed on the color scale temperature profile of orion - kl .
the contours are 3@xmath40 , 6@xmath40 , 9@xmath40 , ...
( @xmath40=8.1 mjy / beam km s@xmath36 ) .
the moment spans the entirety of the hcooh emission in velocity space ( 0.2 - 8.1 km s@xmath36 ) . the color bar on the right is the key for the temperature scale in k.[fig : fa - t ] ] figure [ fig : sio - t ] shows the moment map sio contours overlayed on the temperature map .
the sio distribution peaks adjacent to , but not co - spatial with , the higher temperature and continuum column density regions .
@xcite conclude that the main heating source in the orion - kl region is shock heating .
thus it is quite likely that the sio is tracing the shocks that heat the region as they interact with the dense ambient material . ) .
the color bar on the right is the key for the temperature scale in k.[fig : sio - t ] ] these new observations enable discussion of the kinematics of the orion - kl region .
as sio is a well - known shock tracer , we compare the distribution of each molecule with that of sio to gain insight into the physical conditions for the regions where the different molecules are detected .
figure [ fig : mef - sio ] shows the naturally weighted moment map of hcooch@xmath4 emission overlayed on the moment map of sio emission .
the hcooch@xmath4 emission appears to be coming from the edges of the dense sio outflow . in these regions
, the outflow is impacting the ambient dust and gas , and is likely releasing the hcooch@xmath4 molecules from icy grains and into the gas phase . ]
emission ( black contours ) overlayed on sio emission ( grey contours , red contours in online edition).[fig : mef - sio ] ] figure [ fig : etcn - comb ] shows the c@xmath1h@xmath2cn moment map overlayed with the sio outflow moment map . from this map
, it appears that the c@xmath1h@xmath2cn emission is coming from the regions where the outflow is impacting the ambient dusty material to the northeast of source i and near irc7 , releasing the c@xmath1h@xmath2cn into the gas phase .
the lack of detection of c@xmath1h@xmath2cn to the west / northwest of the outflow ( mirroring what is seen to the east / northeast ) can be explained by a lack of dusty material in this direction ( see section [ sec : motion ] ) .
this hypothesis is supported by the fact that irc3 appears to be a reflection of irc2 , indicating little - to - no dust in the intervening regions @xcite .
h@xmath2cn ] emission ( black contours ) overlayed on sio emission ( grey contours , red contours in online edition).[fig : etcn - comb ] ] figure [ fig : ace - comb ] shows the moment map of ( ch@xmath4)@xmath1co emission overlayed on the sio outflow moment map .
although the ( ch@xmath4)@xmath1co image has a much lower spatial resolution than c@xmath1h@xmath2cn and hcooch@xmath4 , it is still clear that it is coming from the regions in the center of the outflow , corresponding to the densest regions where nearly all of the molecular emission features observed in these studies overlap . however , ( ch@xmath4)@xmath1co is not present in all of the dense regions , while c@xmath1h@xmath2cn is present in all of these regions .
( ch@xmath4)@xmath1co does not appear to have any significant emission to the northeast of source i. ) @xmath1co ] emission ( black contours ) overlayed on sio emission ( grey contours , red contours in online edition).[fig : ace - comb ] ] figure [ fig : fa - sio ] shows the hcooh moment map overlayed on the sio outflow moment map . while the hcooh emission appears near the edges of the outflow
, its emission is notably farther away from the center of the outflow than the emission regions corresponding to the other molecules .
these regions traced by the hcooh emission may still contain material from the outflow , but at a low enough density as to be undetected in these observations .
this indicates that hcooh may be easily liberated from grain surfaces by shocks , but is then quickly processed into other molecules , as indicated by the lack of hcooh emission from behind the initial shock front .
this is chemically consistent with the results of recent astrochemical models @xcite and with recent observational studies @xcite , which suggest that hcooh is highly reactive in the gas phase .
emission ( black contours ) overlayed on sio emission ( grey contours , red contours in online edition).[fig : fa - sio ] ] comparison of molecular distributions with continuum sources can also give insight into the physical conditions of the different regions .
in this section we compare the molecular distributions of the observed species with that of the @xmath0 = 3 mm continuum .
the continuum maps are from @xcite .
figure [ fig : mef - contin ] shows the naturally weighted hcooch@xmath4 contours overlayed on the gray scale @xmath0 = 3 mm continuum .
this map shows that the hcooch@xmath4 emission peaks are closely associated with the weaker continuum peaks , but not the strongest ones .
contours overlayed on the naturally weighted @xmath0 = 3 mm continuum gray scale.[fig : mef - contin ] ] figure [ fig : etcn - contin ] shows the naturally weighted c@xmath1h@xmath2cn contours overlayed on the gray scale @xmath0 = 3 mm continuum .
this maps shows that the c@xmath1h@xmath2cn peaks are nearby to strong continuum sources .
contours overlayed on the naturally weighted @xmath0 = 3 mm continuum gray scale.[fig : etcn - contin ] ] figure [ fig : ace - contin ] shows the naturally weighted ( ch@xmath4)@xmath1co contours overlayed on the gray scale @xmath0 = 3 mm continuum .
this map shows that ( ch@xmath4)@xmath1co peaks near the continuum , but not with the continuum peak . )
@xmath1co contours overlayed on the naturally weighted @xmath0 = 3 mm continuum gray scale.[fig : ace - contin ] ] figure [ fig : fa - contin ] shows the hcooh contours overlayed on the grayscale @xmath0 = 3 mm continuum .
the map shows that there are a few weak hcooh peaks associated with the more extended continuum , southwest of source i and near irc3
. however the bulk of the hcooh emission is not associated with any detected continuum emission .
this is not an issue of resolution , as some of the continuum data were obtained at the same time as the hcooh data .
= 3 mm continuum gray scale.[fig : fa - contin ] ] it is apparent from previous observations that the source i and bn sources within the orion - kl region had some type of dynamic interaction @xmath11500 years ago and are moving apart at @xmath110.02 ( @xmath1150 km s@xmath36 ) per year @xcite .
there are also two known outflows in the orion - kl complex , both centered near source i. the first , the high velocity " ( 30 - 200 km s@xmath36 ) outflow , is traced by co , and possibly sio , and has a very large opening angle ( @xmath111 rad ) in the northwest / southeast directions @xcite . the second , the low velocity " outflow , is perpendicular to the high velocity outflow and is traced by sio @xcite .
the actual velocity of the low velocity outflow is not well - constrained .
@xcite concluded the outflow velocity was @xmath1118 km s@xmath36 , however @xcite concluded that the outflow velocity must be much larger else there would be a trail " from the sio outflow tracing the proper motion of source i. @xcite also suggested that the two outflows may be related as the sio outflow may be precessing rather quickly , and the high velocity outflow may be from an era when the sio outflow was oriented nearly perpendicular to its current position .
we hypothesize that there is yet another interpretation of these high velocity clumps .
these clumps might have had close encounters with regions of higher density / mass that may have altered their trajectory , both within the plane of the sky and in / out of the plane of the sky .
figures [ fig : kin ] and [ fig : kin - cont ] show a potential view of the general kinematics of the orion - kl region .
the gray scale is a moment map of the sio outflow [ from figure 3 of @xcite ] , and the black contours are a combined moment map of the c@xmath1h@xmath2cn and hcooch@xmath4 naturally weighted data ( figure [ fig : kin ] ) and the @xmath0 = 3 mm continuum ( figure [ fig : kin - cont ] ) .
the letters ( a e ) indicate strong molecular peaks while the letters ( @xmath5 @xmath57 ) indicate strong continuum peaks ( excluding source i and bn ) . the blue arrow and dashed lines
indicate the most probable path of source i after its interaction with bn ( see @xcite ) , and the blue star indicates the most probable location of this interaction .
the solid red lines indicate potential paths taken by the sio outflow .
the broken red line indicates a potential path for a high velocity clump of gas after leaving source i. labels 1 - 4 indicate high velocity clumps ( hvc ) .
the outflowing gas in the hvc may encounter temperature / density enhancements as it interacts with the denser ambient clumps .
this may be evident in our temperature map of the region ( see figure [ fig : comb - t ] ) .
we see no evidence of an enhancement for hvc 1 - 3 , however the temperature maps do not cover most of the path covered by these hvc . for hvc 4
, we do see a temperature enhancement leading from the hot core region southwest to the edge of the temperature map .
the bend " in this enhancement may be due to the transverse motion of source i over time and may be tracing the history of travel for hvc 4 . in figure
[ fig : kin ] the northeast lobe of the sio outflow seems to be impeded by the material near clump a , while much of the material seems to be escaping " to the east of clump a. additionally , hvc1 appears to have broken through the region between clumps a & b. clump c appears to be an impediment to the outflow in the southwest while hvc4 seems to have passed the clump to the west . from figure [ fig : kin - cont ] we can see that the mass / density of clump @xmath58 may be impeding the outflow while hvc4 passes to the west .
h@xmath2cn and hcooch@xmath4 naturally weighted data .
the letters ( a - e ) indicate the main molecular peaks .
the thin black ( blue in online edition ) arrow and dashed lines indicate the most probable path of source i after its interaction with bn ( see @xcite ) , and the star indicates the most probable location of this interaction .
the solid black ( red in online edition ) lines indicate potential paths taken by the sio outflow , and labels 1 - 4 indicated high velocity clumps for reference in the text .
the broken black ( red in online edition ) line indicates a potential path for a high velocity clump of gas after leaving source i.[fig : kin ] ] = 3 mm continuum from @xcite .
the greek letters ( @xmath5 @xmath57 ) indicate the stronger continuum peaks ( excluding source i and bn ) .
the thin black ( blue in online edition ) arrow and dashed lines indicate the most probable path of source i after its interaction with bn ( see @xcite ) , and the star indicates the most probable location of this interaction .
the solid black ( red in online edition ) lines indicate paths were the sio outflow appears to have broken through the surrounding material , and labels 1 - 4 indicate high velocity clumps for reference in the text .
the broken black ( red in online edition ) line indicates a potential path for a high velocity clump of gas after leaving source i.[fig : kin - cont ] ] from these results , it is clear that the dynamics and chemistry of the orion - kl region are very complex .
comparison of the molecular emission features with the kinematic tracers allows conclusions to be drawn about the effects of physical conditions on the chemistry of this region .
here we overview the major conclusions for each molecule studied .
c@xmath1h@xmath2cn is present at highest abundance in the regions of highest temperature , along the denser edges of shocks , and near the higher column density dusty regions .
the shocks are the most likely mechanism for releasing the c@xmath1h@xmath2cn from grain surfaces , as they provide more thermal and kinetic energy than any other source in the region .
these results also indicate that c@xmath1h@xmath2cn must also forming through gas - phase processing of grain mantle material in the warmer regions of the source .
otherwise , c@xmath1h@xmath2cn would be destroyed in these warmer regions after being liberated from icy grains by shocks .
in contrast is the emission from the high velocity component near irc3/6/20 . here
the temperature is lower , and this region is on the outermost shock region in an area of lower continuum column density .
there is also little continuum emission here .
these results suggest that there are several conditions under which c@xmath1h@xmath2cn can form and be released into the gas phase .
this area may be indirectly heated by ir radiation from irc2 @xcite .
thus , it seems that c@xmath1h@xmath2cn can be present at a range of physical conditions , including very dense and hot regions and less dense and cooler regions .
regardless , a heating source seems to be required for its formation or liberation .
the hcooch@xmath4 emission appears to be coming from two distinct regions .
the first is quite extended , near the location of the compact ridge .
this region is not associated with any strong continuum emission .
this region is on the extreme leading edge of the outflow from source i , indicating that the hcooch@xmath4 is being liberated from grain surfaces .
this emission region is likely the primary contributor of flux in single dish observations , as it would be least affected by beam dilution . by comparing the compact hcooch@xmath4 emission with the different tracers
, it appears that the vast majority of this emission comes from regions on the outer , lower column density portions of the sio outflow ; lower temperature ( @xmath59150 k ) areas ; and areas of lower dust concentration .
there is some notable emission from the hottest / densest parts of orion - kl , but these regions coincide with the edge of the hcooch@xmath4 emission .
it appears that hcooch@xmath4 is released into the gas phase by the initial expanding shock from the outflow , and is destroyed by the arrival of the higher density , higher temperature components of the outflow .
hcooch@xmath4 is likely processed through additional gas - phase reactions once it is liberated from the grain mantle material .
unlike the case of c@xmath1h@xmath2cn emission peaking in the warmer regions , there are no signs of hcooch@xmath4 being formed in the warmer regions of the source , as its emission peaks are offset from these warmer regions and instead coincide with the leading edges of the shocks .
while there are no corresponding high - resolution images for ( ch@xmath4)@xmath1co , the existing images reveal that it does not have a distribution that matches either c@xmath1h@xmath2cn or hcooch@xmath4 .
instead , acetone appears to peak in the region of most significant overlap between hcooch@xmath4 and c@xmath1h@xmath2cn ( as was originally reported by @xcite ) .
the primary peak is located in the warm transition region between the hot and dense part of the outflow , and the cooler outer edges .
however , there is notable emission from both the hottest and densest parts of the cloud , as well as the cooler and less dense regions .
this indicates that ( ch@xmath4)@xmath1co is present in a wide variety of physical conditions , and its emission may therefore be shaped primarily by chemical processes rather than the physical conditions of the cloud .
hcooh has a unique distribution relative to the physical tracers when compared to the other molecules in this study .
its strongest emission peak is located on the extreme outer edge of the sio outflow , where there is no detectable continuum emission .
there are a few , weaker hcooh peaks in the warmer denser regions , but overall it appears that hcooh is easily released by the shocks from the outflow , and then easily destroyed in the higher density and warmer regions .
this is consistent with chemical models that predict that formic acid is readily processed through gas - phase reactions @xcite .
we have presented carma observations of several molecular species , including ethyl cyanide , methyl formate , acetone , and formic acid , toward orion - kl .
these observations cover over an order of magnitude in spatial resolution , yielding excellent sensitivity to both small- and large - scale structures .
we have compared these results with tracers of physical conditions ( temperature , continuum column density , and dynamics ) .
overall , the results presented here indicate that the distributions of complex molecules trace the dynamics of the region , and that in some cases varying physical conditions can affect the morphology of molecular emission .
these results indicate that all of the observed species can exist in a range of temperature and density conditions ; hcooh appears to have the narrowest range of stability in terms of temperature and density , while c@xmath1h@xmath2cn can exist in the widest range of conditions .
furthermore , it appears that chemical processing may produce ( ch@xmath4)@xmath1co under a variety of physical conditions , while hcooh is readily destroyed by gas - phase processing in warmer , dense regions .
lastly , it appears that the traditional view of nitrogen - oxygen chemical differentiation " in orion - kl , where different chemistries lead to spatially - distinct emission regions for these two classes of molecules , does not give the complete picture of this region .
hcooch@xmath4 and c@xmath1h@xmath2cn are found to be primarily co - spatial in this source .
instead of being shaped by chemical processing , local physical conditions like temperature and density most dramatically shape the relative distributions of these particular molecules .
these results reveal that astrochemical modeling of sources must include the effects of physical conditions and differing spatial distributions before the observations of a given source can be compared directly to the model .
sophisticated hydrodynamic models will be required to fully interpret the results presented here .
we thank an anonymous referee for helpful comments .
we thank c. goddi for supplying an ammonia temperature map of orion - kl .
this work was partially funded by nsf grant ast-0540459 and the university of illinois . s.l.w.w .
acknowledges start - up research support from emory university .
support for carma construction was derived from the states of illinois , california , and maryland , the gordon and betty moore foundation , the kenneth t. and eileen l. norris foundation , the associates of the california institute of technology , and the national science foundation .
ongoing carma development and operations are supported by the national science foundation under a cooperative agreement , and by the carma partner universities .
beuther , h. , zhang , q. , greenhill , l. j. , et al .
2005 , , 632 , 355 blake , g. a. , sutton , e. c. , masson , c. r. , & phillips , t. g. 1987 , , 315 , 621 chernin , l. m. , & wright , m. c. h. 1996 , , 467 , 676 friedel , d. n. & snyder , l. e. 2008 , , 672 , 962 friedel , d. n. , & widicus weaver , s. l. 2011 , , 742 , 64 garrod , r. t. , widicus weaver , s. l. , & herbst , e. 2008 , , 682 , 283 goddi , c. , humphreys , e. m. l. , greenhill , l. j. , chandler , c. j. , & matthews , l. d. 2011a , , 728 , 15 goddi , c. , greenhill , l. j. , humphreys , e. m. l. , chandler , c. j. , & matthews , l. d. 2011b , , 739 , l13 goldsmith , p. f. , & langer , w. d. 1999 , , 517 , 209 greenhill , l. j. , gwinn , c. r. , schwartz , c. , moran , j. m. , & diamond , p. j. 1998 , , 396 , 650 greenhill , l. j. , reid , m. j. , chandler , c. j. , diamond , p. j. , & elitzur , m. 2004 , star formation at high angular resolution , 221 , 155 laas , j. c. , garrod , r. t. , herbst , e. , & widicus weaver , s. l. 2011 , , 728 , 71 lee , k. i. , looney , l. w. , klein , r. , & wang , s. 2011 , , 415 , 2790 menten , k. m. , reid , m. j. , forbrich , j. , & brunthaler , a. 2007 , , 474 , 515 neill , j. l. , steber , a. l. , muckle , m. t. , et al .
2011 , journal of physical chemistry a , 115 , 6472 plambeck , r. l. , wright , m. c. h. , mundy , l. g. , & looney , l. w. 1995 , , 455 , l189 plambeck , r. l. , et al .
2009 , , 704 , l25 quan , d. , & herbst , e. 2007 , , 474 , 521 sault , r. j. , teuben , p. j. , & wright , m. c. h. 1995 , asp conf .
ser . 77 : astronomical data analysis software and systems iv , 77 , 433 simpson , j. p. , colgan , s. w. j. , erickson , e. f. , burton , m. g. , & schultz , a. s. b. 2006 , , 642 , 339 snyder , l. e. , et al .
2005 , , 619 , 914 ulich , b. l. , & haas , r. w. 1976 , , 30 , 247 widicus weaver , s. l. & friedel , d. n. 2012 zapata , l. a. , schmid - burgk , j. , ho , p. t. p. , rodrguez , l. f. , & menten , k. m. 2009 , , 704 , l45 | it has recently been suggested that chemical processing can shape the spatial distributions of complex molecules in the orion - kl region and lead to the nitrogen - oxygen `` chemical differentiation '' seen in previous observations of this source .
orion - kl is a very dynamic region , and it is therefore also possible that physical conditions can shape the molecular distributions in this source . only high spatial resolution observations can provide the information needed to disentangle these effects . here
we present millimeter imaging studies of orion - kl at various beam sizes using the combined array for research in millimeter - wave astronomy ( carma ) .
we compare molecular images with high spatial resolution images that trace the temperature , continuum column density , and kinematics of the source in order to investigate the effects of physical conditions on molecular distributions .
these observations were conducted at @xmath0 = 3 mm and included transitions of ethyl cyanide [ c@xmath1h@xmath2cn ] , methyl formate [ hcooch@xmath3 , formic acid [ hcooh ] , acetone [ ( ch@xmath4)@xmath1co ] , sio , and methanol [ ch@xmath4oh ] .
we find differences in the molecular distributions as a function of each of these factors .
these results indicate that acetone may be produced by chemical processing and is robust to large changes in physical conditions , while formic acid is readily destroyed by gas - phase processing in warm and dense regions .
we also find that while the spatial distributions of ethyl cyanide and methyl formate are not distinct as is suggested by the concept of `` chemical differentiation '' , local physical conditions shape the small - scale emission structure for these species . |
due to the extensive studies on the physical properties of the cuprates , many constraints on the pairing mechanism of their high - temperature superconductivity ( htsc ) have been accumulated .
there is no doubt that the strong coulomb interaction and the resultant strong electron correlations play crucial roles .
this effect is believed to be described by single - band hubbard or @xmath3-@xmath4 models in 2d , and the magnetic mechanism for superconductivity has been proposed with the focus on the short range antiferromagnetism or spin singlet formation.@xcite these models have achieved great success in explaining many of the physical properties , such as the pseudogap , generalized magnetic susceptibility observed by neutron scattering , and the single particle green s function found in angle - resolved photoemission spectra ( arpes )
. however , these models are not successful in explaining the variation of in superconducting transition temperature t@xmath0 from material to material and other material dependent properties .
for example , the famous t - linear resistivity within the plane is universally observed among various cuprates , while t@xmath0 s differ by two orders of magnitude.@xcite it is surprising that the resistivity , which is one of the most representative physical observables of the electronic states in solids , is irrelevant to t@xmath0 .
the 2d hubbard and @xmath3-@xmath4 models contain only a few parameters , such as hopping parameters @xmath3 , @xmath5 , @xmath6 and interactions @xmath7 and @xmath4 .
one possibility is that the range of the hopping and magnitude of @xmath5 and @xmath6 are key factors determining t@xmath0 , which is determined by the structure and chemical composition perpendicular to the cuo@xmath8 plane.@xcite however , studies on the @xmath3-@xmath4 model have found that finite @xmath5 suppresses superconducting correlations.@xcite recent investigations of the single - band hubbard model using cluster dynamical mean field theory calculations@xcite do not show increased tendencies towards pairing for larger @xmath5 but variational studies do,@xcite although the latter may be less controlled . from a structural point of view , the only known empirical rule of t@xmath0 is that it increases at optimal doping as the number of cuo@xmath8 layers @xmath9 is increased for @xmath10 .
anderson noticed this @xmath9-dependence at a early stage , and proposed the interlayer mechanism of superconductivity.@xcite the idea is that the single - particle interlayer hopping is suppressed by strong correlations within the layer , while the two - particle hopping is not .
the onset of the latter below t@xmath0 leads to the condensation energy of superconductivity .
experimentally it is found that the @xmath1-axis electron hopping is actually suppressed and there is no coherent band formation perpendicular to the plane.@xcite there is no plasmon observed in the normal state , and below t@xmath0 the josephson plasmon appears in the low energy region ( @xmath11 mev).@xcite the idea of an interlayer mechanism has also been criticized in light of the @xmath1-axis oscillator strength , and has subsequently been abandoned.@xcite increasing evidence for the importance of out - of - plane effects continues to accrue , which points to limitations of intrinsic planar models for the cuprates .
for example , the correlation of out - of - phase oxygen dopant ions in bi@xmath8sr@xmath8cacu@xmath8o@xmath12 ( bi-2212 ) with features in the tunneling density of states @xcite has been interpreted in terms of a local increase of the superconducting pair potential.@xcite in addition , the rapid suppression of t@xmath0 with out - of - plane cation dopants compared to in - plane dopants is surprising given that the former do not appreciably affect in - plane resistivities.@xcite t@xmath0 has also been empirically correlated with the madelung energy difference between apical and planar oxygen atoms.@xcite this has been recently supported by studies on ba@xmath8ca@xmath13cu@xmath14o@xmath15f@xmath8 , a compound which has vastly different transition temperatures by exchanging f with o at the apical site.@xcite arpes studies have inferred a pairing gap which is a factor of 2 larger on the bonding band in comparison to the anti - bonding band,@xcite a trend consistent with other cuprates.@xcite since the fermi surface of the bonding band lies far away from either the antiferromagnetic reciprocal lattice zone boundary , or the van hove points , linking the pairing mechanism with a purely electronic mechanism is not straightforward . as discussed in refs . and , the madelung energy difference and @xmath5 are directly linked .
_ pointed out that the maximal t@xmath0 in each family of cuprate materials scales with the next nearest hopping @xmath5.@xcite the energy of the cu 4@xmath16 orbital relative to the @xmath17-@xmath18 band largely determines @xmath5 as the hybridization with the planar oxygen orbitals allows electrons to more effectively hop between 2@xmath19 orbitals .
strong apical 2@xmath20 - 4@xmath16 hybridization , determined by the madelung energy difference , raises the energy of the 4@xmath16 orbital relative to the @xmath17 band , reducing the effective hopping @xmath5 .
thus , the further away the apical oxygen is located from the cuo@xmath8 plane the larger @xmath5.@xcite similar empirical relations between t@xmath0 and structural details can also be found .
for example , there is an optimal distance between the apical oxygen site and the mirror plane of the unit cell for which t@xmath0 takes on its maximum value , as shown in fig . [
fig : tc_vs_d ] .
however , despite these observations , a direct connection between changes in the bandstructure and the pairing mechanism is lacking and to date , the connection between t@xmath0 and these @xmath1-axis effects remains an empirical intrigue lacking a firm microscopic understanding .
( color online ) the t@xmath0 at optimal doping for many cuprate superconductors plotted as a function of the distance between the apical oxygen site and the mirror plane located at the center of the unit cell . the hg- ( red ) circles , tl- ( blue ) squares and bi - families ( green ) triangles are shown as well as some lsco and lbco systems open ( black ) . ]
a phonon mechanism has also been pursued since the discovery of htsc.@xcite the original intuition by bednorz and mller was that the jahn - teller ( jt ) effect with high - frequency oxygen phonon modes leads to htsc.@xcite however , the degeneracy of the two @xmath21 orbitals is lifted considerably ( @xmath22 ev ) and the jt effect is now believed to be ineffective .
furthermore , there are several experiments suggesting a minor role played by phonons .
these include the small isotope effect on t@xmath0 at optimal doping , @xcite the absence of the phonon effect on the temperature dependence of the resistivity,@xcite and the absence of the phonon bottle - neck effect.@xcite this conclusion was also supported by early density functional calculations which did not give appreciably high values of couplings and did not support a high t@xmath0 in a bcs picture.@xcite the high - t@xmath0 cuprates , as strongly correlated systems , have attracted a great deal of theoretical interest .
@xcite part of this interest has been focused on examining the renormalization of electron - phonon ( el - ph ) interactions by strong electron - electron ( el - el ) correlations .
@xcite for example , studies of the @xmath23 model incorporating an el - ph interaction indicate that polaron crossover may occur at a weaker el - ph coupling strength than in the case of a pure el - ph coupling model .
@xcite quantum monte carlo treatments of the single - band hubbard - holstein model have shown that the renormalized el - ph vertex develops a strong forward scattering peak with no substantial suppression of the el - ph vertex when the hubbard u is large @xcite with similar results obtained for the @xmath3-@xmath4 model .
@xcite however , using slave - boson approaches , the hubbard model plus el - ph interaction was found to possess no significant forward scattering peak together with an overall suppression of the el - ph vertex at low temperature .
@xcite this same study observed an enhancement in the el - ph vertex for small @xmath24 scattering at high temperature , which was linked to phase separation .
cumulant expansion techniques @xcite have also found an enhancement of the el - ph vertex for small @xmath24 , again interpreted in terms of incipient phase separation when approaching a critical value of @xmath7 .
the single - band hubbard - holstein model has also been studied within dynamical mean field theory ( dmft ) . in one study@xcite the effect of the el - ph interaction was to stabilize the insulating state in the vicinity of the density - driven mott transition . in a paramagnetic dmft study @xcite
the el - ph interaction was found to have little effect on the low energy physics produced by the hubbard interaction while modifying the spectral weight associated with the upper and lower hubbard bands at higher energy . from these results
it was concluded that the primary effect of the el - ph interaction was to reduce the effective value of @xmath7 .
later work@xcite considered antiferromagnetic solutions in the presence of el - ph coupling and observed different behavior with strong polaronic effects . in this case , the critical coupling for polaron formation was observed to shift to larger values as the system was doped away from half - filling .
the dichotomy between the paramagnetic and antiferromagnetic treatments@xcite indicates the possible importance of the magnetic order in considering the el - ph interaction in correlated systems . finally
, a dmft study invoking a lang - firsov transformation for the lattice degrees of freedom has found evidence for a competition between the mott insulating ( metallic ) and bipolaronic insulating phases near ( away from ) half - filling.@xcite using the dynamical cluster approximation , an extension of dmft , examinations of the el - ph interaction within small hubbard clusters @xcite find an overall suppression of @xmath25-wave superconducting t@xmath0 with increasing el - ph coupling .
this occurs despite an increase in the apparent pairing correlations within the @xmath26 channel .
the reduction in t@xmath0 was attributed to polaron formation , which reduces quasi - particle weight at the fermi level and suppresses t@xmath0 through the loss of carrier mobility .
the enhancement of pairing correlations reported in ref .
indicates that the bare el - ph vertex has been renormalized in favour of @xmath25-wave pairing , consistent with the observations of ref . .
furthermore , exact diagonalization ( ed ) studies on the @xmath3-@xmath4 model , which include el - ph coupling to buckling and breathing vibrations , also show that the former enhance @xmath25-wave pairing while the latter suppress it.@xcite while these results provide no definitive interpretation of the effect of el - ph coupling in strongly correlated systems , there is strong evidence that whatever impact strong correlations may have , the el - ph interaction may still play a significant role that should not be overlooked in these systems . from an experimental front
, the role of phonons in htsc has become more prominent in recent years . while long - studied from raman , infrared , and neutron measurements , recent arpes and scanning tunneling microscopy ( stm ) measurements on several cuprates has reinvigorated the exploration of the role of phonons on htsc.@xcite experiments on bi-2212,@xcite la@xmath27sr@xmath28cuo@xmath14 ( lsco),@xcite bi@xmath8sr@xmath8cuo@xmath29 ( bi-2201 ) , @xcite ba@xmath8ca@xmath13cu@xmath14o@xmath15f@xmath8,@xcite and tl families tl@xmath8ba@xmath8cacu@xmath8o@xmath15 ( tl-2212 ) , tlba@xmath8ca@xmath8cu@xmath13o@xmath30 ( tl-1223 ) , and tl@xmath8ba@xmath8cuo@xmath31 ( tl-2201)@xcite have revealed kinks in the energy dispersion of these materials .
these kinks have been interpreted as hubbard renormalizations,@xcite coupling to the neutron resonance and/or spin continuum,@xcite and engelsberg - schrieffer renormalizations@xcite due to coupling of electrons to a collection of optical phonons .
these phonons include the out - of - phase @xmath1-axis oxygen buckling modes and the in - plane cu - o bond - stretching modes .
the dispersion kink observed in the nodal region , ( @xmath32 ) - ( @xmath33 ) , and the peak - dip - hump structure observed in the anti - nodal region , @xmath34-(@xmath33 ) , clearly shows that the electrons are interacting with bosons of a well defined energy @xmath35 and @xmath36 mev , respectively .
for the nodal region , it has been convincingly argued that this structure is due to the oxygen bond - stretching phonon as the kink is observed independent of superconductivity and of the presence / absence of the spin resonance peak . as for the anti - nodal region
, it has been claimed that the kink appears only below t@xmath0 , and hence it is attributed to the spin resonance mode at 41 mev.@xcite however , an extensive study conducted more recently has found the evidence for the kink structure in the normal state and over a wide range of momentum space.@xcite further , contrasting single and multilayer cuprate `` kinks '' , and materials known to have a neutron resonance , also indicates that the observed renormalizations are most likely due to optical phonons , although this is still controversial.@xcite it is well known that the @xmath1-axis phonons show some of the most dramatic lineshape changes with doping and temperature compared to any phonons observed via neutron@xcite and raman@xcite scattering .
for example , the apical phonon frequency shifts by as much as 20 @xmath37 with doping and temperature in a number of compounds : la@xmath27sr@xmath28cuo@xmath38 , hgba@xmath8ca@xmath39cu@xmath40o@xmath41 ( @xmath42 ) and bi@xmath8sr@xmath8ca@xmath8cu@xmath13o@xmath43.@xcite moreover , recent arpes data on bi-2201 have shown kinks in the energy range of the @xmath1-axis phonons which are weaker in overdoped compounds in comparison to optimal doped compounds.@xcite this has also been interpreted in terms of increased screening of the el - ph interaction with increasing hole concentrations .
in addition , the anomaly of the raman @xmath44-polarized mode due to the onset of superconductivity is observed in three- and four - layer compounds.@xcite this has been successfully analyzed in terms of the internal electric field produced by the interlayer josephson plasmon and its coupling to the phonon .
this means that the system behaves as an ionic crystal along the @xmath1-axis in the normal state , and suddenly turns into a superconductor .
while the role of the neutron resonance and phonons remains controversial , it is of relevant interest whether these signatures in arpes may be used as an angle - resolved analogy to the tunnelling ripples in conventional superconductors,@xcite thus providing information on the pairing mechanism in the cuprates . in order to connect el - ph coupling to a possible pairing mechanism
a systematic study of coupling across families of cuprate materials is desirable . in ref .
arpes observed renormalizations of the band were interpreted as due to the @xmath2 branch for anti - nodal electrons and the bond - stretching branch for nodal electrons .
while the latter coupling is of a deformation type , the coupling constructed for the @xmath2 branch involves a charge - transfer between planar oxygens due to a modulation of the electrostatic or madelung energies of the planar oxygen sites .
a local crystal field , generated by a mirror plane symmetry breaking , allows for a coupling at first order in atomic displacements.@xcite ( we note here that the @xmath44/@xmath2 nomenclature only holds for raman @xmath45 momentum transfers .
however , throughout this work we denote the entire out - of - phase branch as @xmath2 " and the entire in - phase branch as @xmath46 . ) since the cuprates are poor conductors along the @xmath1-axis , the electrostatic interaction can be thought to be largely unscreened .
calculations based on ewald s method have been performed on yba@xmath8cu@xmath13o@xmath47 ( ybco)@xcite and large crystal fields have been obtained and the resulting coupling matches well with the coupling determined from fano lineshape analysis of raman data.@xcite recently , the issue of whether the el - ph coupling in the cuprates is strong enough to explain the observed band renormalizations has been revisited via density functional ( lda ) calculations , @xcite updating previous estimates.@xcite while many efforts have been made to extract bosonic coupling from arpes renormalizations in the cuprates , there is no widely accepted way to uniquely determine the strength of the coupling at kink energies , and thus reliable comparisons of calculations with experiment must be viewed with some caution.@xcite while lda calculations have provided remarkably good agreement with phonon dispersions , there are a number of facets of lda calculations which may only provide part of the story of el - ph coupling .
as lda calculations overestimate the itinerancy of the electrons , they describe the cuprates as good metals even at half - filling .
moreover , the obtained interlayer transfer integral and hence the @xmath1-axis plasmon frequency is both coherent and much larger than the experimental observation . both of these factors serve to overestimate the screening ability of the cuprates , especially in the underdoped region , and
thus underestimates the strength of the el - ph interaction .
this may be one of the reasons why lda predicts smaller linewidths for the half - breathing oxygen bond - stretching modes and the apical oxygen modes , sometimes by more than one order of magnitude.@xcite it is therefore not clear if these findings indicate that the el - ph coupling is small or that dft - based approaches alone are inadequate for describing the physics of the cuprates .
as has been found in stm experiments,@xcite a nano - scale inhomogeneous structure on a length scale of 15 @xmath48 exists universally in bi-2212 and ybco
. this length scale can not be larger than the screening length and we can conclude that the screening length within the cuo@xmath8 plane is not shorter than 15 @xmath48 , much longer than the thomas - fermi screening length of the typical metal ( @xmath49 ) . furthermore , charge transfer between the layers is almost prohibited . as a result of the transfer integral between
the layer is proportional to @xmath50 ^ 2 $ ] , the opening of the pseudogap in the @xmath51 and @xmath52 regions strongly suppresses the interlayer hopping . considering these discrepancies between lda and experiments
, one can imagine that the el - ph coupling is in reality much stronger than lda predicts .
focusing on the buckling modes , theoretical consideration have been limited to the two - dimensional plane and the inter - layer coulomb interaction has been neglected .
this is usually justified in the metal since the screening length is much shorter than the interlayer distance .
this is not the case in the cuprates . in this paper
we provide a comprehensive and self - contained story on el - ph coupling to oxygen phonons as a function of doping across the cuprate families .
we formulate a theory for el - ph coupling in the cuprates taking into account the local environment around the cuo@xmath8 planes , the poor screening of charge fluctuations out of the plane , doping - dependent band character variations , and structural differences across the cuprate superconductors .
the organization of this paper is as follows . in section
ii we present a general discussion of el - ph coupling to @xmath1-axis oxygen phonons . after summarizing prior work on the out - of - plane planar oxygen modes ,
we then provide a derivation of the coupling to modes involving @xmath1-axis apical oxygen motion . in section iii
we then discuss the anisotropy of the bare couplings and examine the total strength of the bare el - ph coupling and its contribution to the single - particle self - energy and @xmath25-wave anomalous self - energy . in section
iv we develop the formalism for poor screening and examine its implications for the anisotropy and overall magnitude of the renormalized el - ph vertices . due to the poor @xmath1-axis conductivity we find that the @xmath1-axis phonons can not be effectively screened for small in - plane momentum transfer @xmath53 .
this effect becomes more pronounced as the crystal becomes more ionic and screening becomes increasingly inoperable in the underdoped side of the phase diagram .
however , in the case of the @xmath2 modes the coupling is anomalously anti - screened producing an enhancement of the coupling in the anti - nodal region . in terms the projected @xmath25-wave couplings , the small @xmath53 behavior of the screened vertices produces an enhancement in the total phonon contribution to pairing .
therefore , the combined effects of poor screening reduces the total el - ph coupling and enhances the @xmath25-wave projected coupling with doping .
this has important implications for the doping dependence of the el - ph self - energies probed by arpes as well as any contribution to pairing mediated by phonons . in section
v we turn to materials trends and a systematic examination of the madelung potential and crystal field strengths across the bi- , tl- and hg families of cuprates is presented . here , using an ionic point charge model and the ewald summation technique , we identify systematic trends in the strength of the crystal fields which mirror trends in the material s t@xmath0 at optimal doping . through this observation
we link the structure and chemical composition to the strength of the coupling to the @xmath1-axis modes and discuss how this can be used to understand the large variations in t@xmath0 observed across the cuprates .
we also present considerations for doping - induced changes to the value of the crystal field in bi-2212 . in light of these findings
, section vi presents a simple two - channel model for pairing in the cuprates , which includes a dominant , @xmath25-wave pairing , high - energy bosonic mode and a weaker phonon mode . using this model
, we demonstrate that phonons can provide a sizeable enhancement to t@xmath0 which is in excess of the t@xmath0 that would be obtained from phonons alone . furthermore ,
due to the dominant bosonic mode , the resulting value of the isotope exponent @xmath54 is small ( @xmath55 ) despite the large enhancement of t@xmath0 ( @xmath56 k ) .
this calculation , in combination with the materials and doping dependent trends identified in the previous sections , shows that a phonon assisted pairing model provides a natural framework for understanding trends observed for t@xmath0 across the cuprates .
finally , in section vii we conclude by summarizing our findings and discuss open questions concerning el - ph coupling in strongly correlated systems .
in addition to the treatment outline above , in the appendix we explore how el - ph coupling to @xmath1-axis modes is modified by strong correlations in the half - filled parent insulators using exact diagonalization of small multi - band hubbard clusters .
specifically we address how el - ph coupling modifies the properties of the zhang - rice singlet ( zrs ) . here
we find that static lattice displacements have a strong influence on the zrs hoppings , energy and antiferromagnetic exchange energy j. these results have a direct impact on the use of down - folded models such as the @xmath3-@xmath4 model as they indicate that the effects of the el - ph coupling can not be simply cast as modulations of a single parameter such as @xmath3 , @xmath4 , or the energy of the zrs .
in this section we present a review of some generic considerations for electrons coupling to oxygen motions in- and out- of the cuo@xmath8 plane . since many of the derivations have appeared before , we can be brief , with the main aim to generalize previous results to a five band model in order to include off - axis orbitals and apical oxygen phonon modes .
we begin by considering an ideal cuo@xmath8 plane isolated from its environment .
since hopping integrals are modulated to second order in atomic displacements along the @xmath1-axis , the el - ph coupling due to this mechanism is weak .
however , if the same plane is placed in an asymmetric electrostatic environment a local crystal field , which breaks mirror plane symmetry , provides a coupling linear in displacement .
the plane must then spontaneously buckle in a pattern where the oxygen ( copper ) atoms are displaced away from ( towards ) the mirror plane . in a three band
model,@xcite the local field coupling was used to construct a charge - transfer el - ph vertex for coupling to the raman active out - of - phase and in - phase @xmath1-axis oxygen vibrations .
the in - phase phonon modulates charge transfer between planar oxygen and copper orbitals while the out - of - phase phonon modulates charge transfer between only the planar oxygen orbitals .
a single - band el - ph coupling was obtained : @xmath57,\ ] ] where @xmath58 ( @xmath59 ) creates ( annihilates ) an electron in the partially filled antibonding band with momentum @xmath60 , energy @xmath61 , and spin @xmath62 , and @xmath63 ( @xmath64 ) creates ( annihilates ) a phonon of energy @xmath65 and wavevector @xmath24 . considering a modulation of the electrostatic coupling of the charge density at the oxygen sites coupled to the local on - site potential @xmath66 ,
the hamiltonian is of the form @xcite @xmath67,\ ] ] where @xmath68 is the oxygen displacement vector in the unit cell at lattice site * n * , @xmath69 is the electron charge and @xmath70 ( @xmath71 ) is the creates ( annihilates ) an electron at site @xmath72 , which can include both planar and apical oxygen orbitals .
this coupling mechanism differs from the deformation coupling considered in ref . .
expanding for small displacements @xmath73 where @xmath74 includes the madelung contribution to the site energies and the term linear in @xmath75 generates the el - ph interaction @xmath76 @xmath77 is the local crystal field at the oxygen site provided this field is finite , which occurs at locations of broken mirror symmetry in the unit cell . in order to derive the form of the coupling @xmath78 eq .
( [ eq : el - ph ] ) must be rewritten in the form of eq .
( [ eq : interaction ] ) . to do so ,
the el - ph hamiltonian is fourier transformed to momentum space and the oxygen operators are replaced by band representation operators @xmath79 . here , @xmath80 is the oxygen ( @xmath81 for planar oxygen and @xmath82 for apical oxygen ) eigenfunction for the pd-@xmath18 band , which is obtained from a tight - binding model for the cuo@xmath8 plane .
( color online ) the five - band model used to derive the form of the el - ph couplings @xmath78 . ] prior work focusing on the @xmath44 , @xmath2 and breathing branches made use of a three - band model.@xcite in order to extend these works to include the apical oxygen modes this model must be extended to a five - band model as shown in fig .
[ fig:5band ] .
the basis set of this model contains a 4@xmath16 ( @xmath83 ) and 3@xmath26 ( @xmath84,@xmath85 ) orbital on each copper site @xmath86 , two planar oxygen 2@xmath19 orbitals ( @xmath87,@xmath88 ) with @xmath81 , and one apical oxygen 2@xmath20 orbital ( @xmath89,@xmath90 ) . here
, we neglect the in - plane o 2@xmath91 orbitals oriented perpendicular to the cu - o bonds .
these orbitals form weaker @xmath17-@xmath92 bonds with the lower energy cu @xmath93 orbitals and do not contribute heavily to the character of the band crossing the fermi level.@xcite site energies are denoted by @xmath94 , respectively . defining canonical fermions@xcite @xmath54 , @xmath95 from combinations of the planar oxygen orbitals via a wannier transformation
@xmath96 where @xmath97 and @xmath98 , the hamiltonian can be written as @xmath99 : @xmath100 \\ \nonumber & + & 2\bigg [ t_{pd}\mu_{{\textbf{k}},\sigma}d^\dagger_{bk,\sigma}\alpha_{{\textbf{k}},\sigma } + t_{pp}\chi_{\textbf{k}}\alpha^\dagger_{{\textbf{k}},\sigma}\beta_{{\textbf{k}},\sigma } \\
\nonumber & + & t_{ps}\kappa_{{\textbf{k}}}s^\dagger_{{\textbf{k}},\sigma}\alpha_{{\textbf{k}},\sigma } - t_{ps}\lambda_{\textbf{k}}s^\dagger_{{\textbf{k}},\sigma}\beta_{{\textbf{k}},\sigma } + \frac{t^\prime_{ps}}{2}s^\dagger_{{\textbf{k}},\sigma}a_{{\textbf{k}},\sigma } \\ & -&t^\prime_{pp}\kappa_{\textbf{k}}a^\dagger_{{\textbf{k}},\sigma}\alpha_{{\textbf{k}},\sigma } + t^\prime_{pp}\lambda_{\textbf{k}}a^\dagger_{{\textbf{k}},\sigma } \beta_{{\textbf{k}},\sigma } + h.c . \bigg]\end{aligned}\ ] ] with @xmath101 containing the site energies and @xmath102 denoting the hermitian conjugate .
finally , the basis functions are defined as : @xmath103 in the limit where the apical and copper 4@xmath16 orbitals are removed ( @xmath104 , etc . )
there expressions recover prior work carried out using a three - band model.@xcite for planar oxygen vibrations the el - ph coupling is given by ( @xmath105 ) @xmath106\end{aligned}\ ] ] where the minus ( plus ) sign is for coupling to the @xmath2 ( @xmath44 ) branch . here
@xmath107 denotes the oxygen mass , @xmath108 is the @xmath1-axis component of the local field at the planar oxygen site , @xmath109 $ ] is the phonon eigenvector normalization , and @xmath110 denote the assumed dispersionless frequencies of the @xmath2 and @xmath44 branches , respectively . the motion of the heavier cu atoms has been neglected in this treatment
( [ eq : b1 g ] ) is a generic expression for the @xmath44 and @xmath2 vertices , independent of the underlying tightbinding model used to determine the band eigenfunctions @xmath111 . in the three - band model
, the band eigenfunctions are defined as@xcite @xmath112 \\ \phi_{cu}({\textbf{k}})&= & \frac{1}{a({\textbf{k}})}[\epsilon^2({\textbf{k } } ) - t^{\prime2}({\textbf{k } } ) ] , \end{aligned}\ ] ] where @xmath113 , @xmath114 , the normalization is @xmath115 ^ 2 + [ \epsilon({\textbf{k}})t_x({\textbf{k } } ) - t^\prime({\textbf{k}})t_y({\textbf{k}})]^2 \\ & + & [ \epsilon({\textbf{k}})t_y({\textbf{k } } ) - t^\prime({\textbf{k}})t_x({\textbf{k}})]^2,\end{aligned}\ ] ] and @xmath116 is the bare dispersion , given in ref . . from eq .
( [ eq : b1 g ] ) it can be seen that the symmetry of the phonon is implanted into the el - ph coupling to provide substantial momentum anisotropy . for the @xmath2 phonon branch , @xmath78 changes sign for @xmath117 , while for the @xmath44 phonon branch it does not . as a result
the coupling involves all fermionic states for the @xmath44 branch , while for the @xmath2 branch the antinodal states along the brillouin zone ( bz ) axes are weighted heavily and nodal states along the zone diagonal are projected away .
since apical phonons show some of the strongest renormalizations in la- and hg - cuprates,@xcite and since the apical oxygen atoms do not lie in a mirror plane symmetry even in single layer cuprates , they are included as an extension of our previous work .
early on the apical phonon was thought to be quite anharmonic and related to the jahn - teller mechanism in yba@xmath8cu@xmath13o@xmath47,@xcite although many confusing results were found.@xcite more recently the coupling is thought to be electrostatic@xcite in nature and in some treatments weakly momentum dependent.@xcite it should be emphasized that formally a holstein ( momentum independent ) coupling in any model with long - range coulomb interactions will be screened out by backflow due to charge conservation and therefore the coupling is expected to be very small .
this will be discussed in section iv . here
instead we place focus on a strongly momentum dependent coupling arising from charge transfer mechanisms between apical and planar oxygen orbitals , similar to the electronic pathways involved in @xmath1-axis tunnelling.@xcite an apical orbital displacement modulates the madelung energy as in eq .
( [ eq : el - ph ] ) and the resulting el - ph coupling is of the form of eq .
( [ eq : interaction ] ) , with @xmath118 here , @xmath119 is the @xmath1-axis component of the eigenvector for the apical branch coming from the atomic displacement in eq .
( [ eq : el - ph ] ) , @xmath120 , and @xmath121 is the apical eigenfunction , obtained from diagonalizing eq .
( [ eq:5band ] ) . for our purpose
however , we are primarily interested in the leading order momentum dependence of the coupling @xmath78 . starting from eq .
( [ eq:5band ] ) , the lwdin down - folding procedure@xcite is applied to determine the apical character of the resulting partially filled band crossing the fermi level .
the resulting form for the apical eigenfunction @xmath121 is then @xmath122 with @xmath123 defined in eq .
( [ eq : basis ] ) .
finally , for simplicity we neglect the momentum dependence of the apical phonon s eigenvector and set @xmath124 .
more complicated models in which the apical phonon involves the motion of the in - plane oxygens can be treated accordingly .
the local crystal field @xmath125 at the apical oxygen site modulates a charge transfer between the apical oxygen and the planar orbitals .
this mechanism of charge transfer is analogous to the charge transfer mechanism yielding bi - layer splitting.@xcite from eqs .
( [ eq:5band ] ) and ( [ eq : gapex0 ] ) , the resulting momentum dependence of the coupling via this transfer , @xmath126 [ \cos(p_xa ) - \cos(p_ya)]$ ] with @xmath127 , is strongest for anti - nodal electrons , and has a form factor similar to @xmath1-axis hopping @xmath128.@xcite although this coupling has an anisotropy similar to that of the coupling to the @xmath2 branch , it does not contribute to @xmath25-wave pairing due to its phase and momentum dependence at large @xmath129 . for completeness
, we also consider the coupling to the planar cu - o bond stretching modes , the so - called breathing modes , within the framework of the three - band model , as derived in refs . , and .
the bond - stretching modes couple to electrons via both a direct modulation of the hopping integral @xmath130 as well as electrostatic changes in the madelung energies as the orbitals are displaced .
@xcite as done in ref .
we consider only the overlap modulation .
the derivation is briefly sketched here . to obtain the form of the el - ph coupling the overlap integral @xmath130
is taken to be site dependent @xmath131 .
it is then assumed that the cu and o atomic displacements , @xmath132 and @xmath133 , about their equilibrium positions , @xmath134 and @xmath135 , where @xmath136 are basis vectors for the cuo@xmath8 plane , are small . the overlap integral is expanded and only the first order term is retained @xmath137
the modulation of the hopping integrals provides the el - ph coupling hamiltonian @xmath138 \left[d^\dagger_{n,\sigma}p_{n,\sigma,\delta } + h.c . \right],\ ] ] with @xmath139 denoting the phase of the cu - o overlap .
following ref . , we neglect the cu vibration and set @xmath140 , where @xmath141 and @xmath142 is a scalar function that depends on the equilibrium cu - o distance .
the el - ph coupling hamiltonian can then be simplified to @xmath143,\ ] ] as obtained in ref . .
here @xmath144 denotes the displacement of the oxygen atom @xmath145 along the cu - o bond . by introducing the fourier
transform the el - ph coupling may be re - written as : @xmath146(b^\dagger_{\textbf{q}}+ b_{-{\textbf{q}}}).\ ] ] here , @xmath147 is the component of the phonon eigenvector for oxygen @xmath145 parallel to the cu - o bond .
finally , the electronic eigenfunctions of the planar o @xmath148 and cu 3d@xmath149 orbitals @xmath150 for the pd-@xmath18 band , with operators @xmath1 , @xmath151 , are introduced .
the resulting hamiltonian reduces to the form of eq .
( [ eq : el - ph ] ) with @xmath152\ ] ] here @xmath127 , @xmath153 .
we note that generally @xmath154 and thus @xmath155 . since typically @xmath156,@xcite an estimate for the strength of the coupling is obtained from @xmath157 ev / for @xmath158 ev . for @xmath159 mev , this gives @xmath160 mev .
in the previous section it was shown how the explicit form for el - ph coupling to oxygen modes is determined by the nature of the charge transfer modulated by the lattice displacement , the local environment surrounding the cuo@xmath8 plane , as well as the orbital content of a single downfolded band crossing the fermi level .
the relevant parameters - magnitude of orbital hybridization , local crystal field , the charge - transfer energy , the shape of the fermi surface , and the density of states at the fermi level - all contribute in setting the overall magnitude of the coupling as well as the full fermionic * k * and bosonic * q * momentum dependence of the coupling @xmath161 .
the band character enters through the band eigenvectors @xmath162 which further depend on the complexity of the unit cell . at the bz center
the wavefunctions are atomic and the character of the band is unique .
large momentum variations of the band character then occur for increasing momentum , and a very strong momentum dependence of the overall el - ph coupling can occur .
this strong momentum dependence has indeed been observed in recent lda treatments @xcite .
we remark that we are first interested in the magnitude and anisotropy of the bare couplings in the absence of charge screening in order to determine possible discrepancies with lda treatments , which treat correlations on the mean field level and give 3d metallic screening . in order to estimate general tendencies , momentum dependencies as well as magnitudes , in this section we explore some simplifications .
we begin by assuming that @xmath163 is much greater than any relevant energy scale in the system , keeping in mind that the charge transfer energy @xmath164 ev is much reduced from its bare value @xmath165 ev when treating correlations in mean field approaches like lda.@xcite in this limit the @xmath162 functions can be represented as @xmath166 and @xmath167 to the same order , the denominators in these expressions are constant over constant energy contours .
therefore , since we will restrict ourselves largely to the fermi surface , we represent the band functions @xmath168 , @xmath169 and @xmath170 , with @xmath171/2 $ ] and the coefficients determined by @xmath172 , @xmath173 and @xmath174 , respectively . here
@xmath175 denotes a fermi surface average : @xmath176 . in this way
, the overall coupling anisotropy can be simplified without loss of generality . as a consequence ,
the fermionic momentum dependence of the coupling to the breathing modes disappears : @xmath177 substituting the phonon eigenvectors , the coupling to the breathing modes becomes @xmath178 this form has also been obtained in a @xmath3-@xmath4 approach @xcite however , in this case , the oxygen and copper character have been explicitly retained through @xmath179 and @xmath180 , respectively.@xcite the el - ph vertex for the @xmath44 and @xmath2 modes can be likewise simplified @xmath181 \end{aligned}\ ] ] these expressions recover the raman form factors in the limit @xmath182 for each mode and they obey the symmetry conditions for momentum reflections about @xmath183 as discussed previously .
the fermionic momentum dependence can not be neglected in either of these expressions , where in particular , a strong fermionic momentum dependence of the coupling to the @xmath2 modes occur , preferentially weighting anti - nodal states with small momentum transfers .
lastly , the momentum structure of the apical coupling simplifies considerably in the same manner : @xmath184[\cos(p_xa ) - \cos(p_ya)]/4.\ ] ] once again , a substantial fermionic momentum dependence emerges from the @xmath1-axis charge transfer pathways and the apical character of the band . before proceeding further a few comments are in order . in using eqs.([eq : gbr])-([eq : gapex ] ) we have simplified the el - ph couplings while explicitly retaining the role of the band character in determining the overall strength of the couplings . since the eigenfunctions enter to the fourth power for @xmath185 , the total coupling strengths determined in this approach may change considerably when adjusting multi - band parameters . however , this method has the advantage that the materials dependence of the coupling , parameterized by @xmath179 , @xmath186 and @xmath180 , can be calculated using a variety of methods such as exact diagonalization , quantum monte carlo or lda .
we also emphasize that the total coupling strengths ( calculated in the next section ) that we obtain using this formalism are similar to those obtained from lda treatments , even though the latter includes the effects of screening .
@xcite ( color online ) plots of the el - ph coupling constant @xmath187 for fermionic momentum on the fermi surface as a function of transferred momentum @xmath188 ( a1-a4 ) and @xmath189 ( b1-b4 ) , respectively .
a1,b1 ( a2,b2 ) plot coupling to the @xmath44 ( @xmath2 ) branches , respectively , a3 , b3 plot coupling to the breathing branch and a4,b4 , plots coupling to the apical branch .
the colors denote angles from the corner of the bz ( shown in the inset of fig .
[ fig : lambda_fs ] ) given by solid black - @xmath190 , solid red - @xmath191 , solid green - @xmath192 , solid blue - @xmath183 ( nodal ) , dashed green - @xmath193 , dashed red - @xmath194 and dashed black - @xmath195 ( anti - nodal ) . ]
this approach also allows for the use of a renormalized bandstructure , while retaining the explicit band character of the original five - band model .
this is important since the overall strength of the el - ph couplings scales with the density of states at the fermi level @xmath196 ( see eq .
( [ eq : lambdafs ] ) ) ; narrow bandwidth systems will exhibit larger coupling in comparison to large bandwidth systems with the same vertex @xmath78 . with an appropriate choice in parameters , the five - band model given in section [ sectiontwo ]
reasonably reproduces the bandwidth ( and @xmath196 ) determined by lda calculations.@xcite however , as has been noted,@xcite lda over predicts the total bandwidth ( and consequently @xmath196 is under predicted ) in comparison with experiment .
therefore , we expect that the total couplings will be underestimated using the five - band model with parameters chosen to match lda .
a simple rescaling of the five - band model bandstructure in conjunction with the full form of the @xmath162-functions is insufficient to correct this since this procedure would produce incorrect values for the @xmath162-functions and therefore generate errors in the character of the band as a function of @xmath60 .
however , the use of the couplings defined by eqs .
( [ eq : gbr])-([eq : gapex ] ) allows us to resolve this issue . here
, the correct band character is captured by calculating @xmath179 , @xmath180 and @xmath186 using the five - band model but the fermi surface and bandstructure are obtained from elsewhere in order to better match experiment . in this work
, we adopt a five - parameter tightbinding model for bi-2212 derived from fits to arpes data.@xcite this approach allows us to capture the increased value of @xmath196 while simultaneously retaining estimates for the correct band character .
we also note that the specific shape of the fermi surface is not crucial to the overall anisotropy of the couplings . to visualize the momentum dependence of the coupling in more detail , we plot in fig .
[ fig : gkq ] @xmath187 , given by eqs .
( [ eq : gbr ] ) - ( [ eq : gapex ] ) , as a function of transferred momenta @xmath24 along two directions as indicated .
the dependency on transferred momenta arises from the nature of the charge - transfer coupling of the different modes .
the @xmath1-axis modes ( figs [ fig : gkq]a1 - 2,4 and [ fig : gkq]b1 - 2,4 ) , being electrostatic in nature , translate into stronger coupling for small momentum transfers , while the deformation - type coupling of the breathing branches gives stronger coupling at large @xmath24 , and vanishes in the limit @xmath182 ( figs .
[ fig : gkq]a3 and [ fig : gkq]b3 ) .
apart from the breathing modes , an appreciable fermionic wavevector dependence of the couplings to the @xmath1-axis modes is found as a consequence of the character of the underlying atomic vibrations .
for the case of the @xmath44 and apical modes , for momentum transfers along the zone diagonal ( figs .
[ fig : gkq]b1 and [ fig : gkq]b4 , respectively ) , the fermionic dependence is symmetric with respect to reflections about @xmath197 while the @xmath2 coupling ( fig .
[ fig : gkq]b2 ) changes sign .
momentum transfers along the zone face ( figs .
[ fig : gkq]a1 , [ fig : gkq]a2 , [ fig : gkq]a4 ) do not obey any set selection rule , although the symmetric- or anti - symmetric - like character of the coupling is evident .
finally , the strong momentum dependence of the charge transfer along the @xmath1-axis dictates that the apical coupling vanishes for any fermion momentum along the zone diagonal ( fig .
[ fig : gkq]b4 ) .
the strong dependence of @xmath78 on both @xmath60 and @xmath24 leads to anisotropic coupling between electrons and phonons . as the most relevant scattering processes involve those states near the fermi level , the explicit momentum dependence as seen in arpes is most clearly envisioned by calculating @xmath198-dependent self - energies , in terms of el - ph coupling @xmath199 .
we take the modes to be dispersionless @xmath200 , which couple to electrons via the fock piece of the electron self - energy @xmath201 in lowest order@xcite : @xmath202 with @xmath203 , @xmath204 is the chemical potential , @xmath205 is the number of momentum points and @xmath206 is the density of states at the fermi level .
the delta function restricts the sum to initial and final fermion states @xmath60 , @xmath207 , that lie on the fermi surface , with scattering between them governed by the transferred phonon momentum @xmath24 .
the resulting @xmath208 for the four modes are shown in fig .
[ fig : lambda_fs ] . the fermi surface contour and bandstructure are again determined from a five - parameter tightbinding bandstructure.@xcite the el - ph coupling vertices are evaluated using eqs .
( [ eq : gbr ] ) - ( [ eq : gapex ] ) , along with the conventional parameter set ( in ev ) : @xmath158 , @xmath209 , @xmath210 , @xmath211 , @xmath212 , @xmath213 , @xmath214 , @xmath215 .
@xcite for these parameters we obtain @xmath216 , @xmath217 and @xmath218 .
additionally , estimates of the local field strength at the planar and apical oxygen sites for bi-2212 are used : @xmath219 , @xmath220 ev / , respectively .
a systematic derivations of these e field values across the cuprate families is given in section [ madelung ] .
( color online ) plots of @xmath208 for momentum points along the fermi surface .
inset : the definition of the angle @xmath221 used in figs .
[ fig : gkq ] and [ fig : lambda_fs ] .
the parameters used are defined in the text . ]
a strongly varying @xmath222 is obtained for the @xmath1-axis modes , largely weighting anti - nodal fermion states , arising from both the fermion dependence of the bare couplings as well as the small momentum transfers connecting anti - nodal points on the fermi surface , as noted in prior treatments @xcite as well as more recent lda- projected studies.@xcite the anisotropy is particularly strong for the apical coupling because of the strong fermionic momentum dependence of @xmath78 . on the other hand ,
coupling to the breathing modes has a much weaker anisotropy along the fermi surface , due to large momentum transfers ( @xmath223 ) connecting anti - nodal portions of the fermi surface , as well as ( @xmath224 ) and ( @xmath225 ) transfers connecting nodal points .
inclusion of a variation of the copper character across the fermi surface yields a more anisotropic coupling.@xcite however , the character of the anisotropy is not as strong as that observed in ref . , where the breathing branches couple more strongly to anti - nodal states .
the full coupling @xmath226 , renormalizing the single particle self - energy , is given by a sum over all modes averaged over the fermi surface , @xmath227 , and can be visualized most easily as an average over the curves shown in fig .
[ fig : lambda_fs ] .
the renormalization @xmath228 for a @xmath26-wave superconductor gives the el - ph contribution to the anomalous self - energy and is given by a @xmath25-wave projected average @xmath229 where @xmath230 and @xmath25-wave basis function @xmath231/2 $ ] , as before .
a positive ( negative ) @xmath228 denotes an attractive ( repulsive ) pair interaction in the @xmath26 channel .
the @xmath199 values obtained for the four phonon branches considered in this work are given in table [ tbl : lambda ] .
as noted before , the @xmath44 and @xmath2 modes enhance @xmath25-wave pairing , the breathing modes suppress it and the apex modes gives no contribution .
we note that these values for the @xmath226 are similar to those obtained from recent lda results for yba@xmath8cu@xmath13o@xmath47@xcite but are smaller than those calculated for doped cacuo@xmath8.@xcite we again remark that important consequences of screening are not considered here , which can affect both the magnitudes and anisotropies of the couplings as will be discussed in sec .
[ sec : screening ] .
.[tbl : lambda ] tabulated values of @xmath232 for the four phonon branches considered in this work . [ cols="^,^,^",options="header " , ] it is important to note that in arpes there is a strong kinematic constraint that also governs which fermion states can couple to each mode . due to energy conservation of scattering from dispersionless phonons ,
the normal state arpes spectra may show mixing of electron and phonon states if the band energy along a particular cut crosses the phonon mode energy .
this gives rise to a kink in the dispersion , with trailing intensity observed asymptotically to the phonon mode energy .
otherwise , if the band along a particular cut lies above the mode energy , the spectra may show level repulsion in the form of a flattened band bottom along that cut .
( color online ) a contour plot of the band structure of optimal doped bi-2212 obtained from the five - parameter tightbinding model of ref . .
the outermost contour ( solid black ) corresponds to the fs .
the remaining contours correspond to @xmath233 for the four phonon branches considered in this work .
reading from the fs to the @xmath234-point , the contours correspond to the @xmath2 , @xmath44 , breathing and apical branches , respectively . ] in the cuprates , the shallow band near the anti - nodal parts of the bz limits the couplings of those states to only modes with energies less than @xmath235 .
thus , quite generally , we would expect coupling to the modes to disappear from the spectra when @xmath236 along the entire cut .
the contours @xmath237 , which define this kinematic constraint , are plotted in fig .
[ fig : bandcontour ] at optimal doping .
the fermi level is also shown for reference and corresponds to the outermost contour ( solid black ) . in the case of the apical mode , although the coupling may be largest for @xmath238 , a kink effect is prevented kinematically . moreover
, the breathing modes will be observable only for near - nodal cuts due to the form of the coupling as well as the energies of the modes which are large when compared to the @xmath44/@xmath2 modes and the saddle point energy @xmath235 .
in fact , for optimally doped bi-2212 , the @xmath2 modes lie right at @xmath235 , and can be seen most clearly in the anti - nodal directions , although its coupling should be felt across the entire fermi surface as shown in fig .
[ fig : lambda_fs ] . in the superconducting state , the energy contours shift by the maximum value of the energy gap and are set by @xmath239 , and the kinematic range shrinks for coupling to the higher energy modes.@xcite
screening - responsible for reducing the net el - ph coupling in conventional good metals - must be included for all @xmath1-axis phonons . in the usual treatment of screening,@xcite
the screened el - ph coupling @xmath240 is @xmath241 where @xmath242 is the 3d coulomb interaction and @xmath243 is the frequency - dependent polarizability calculated with vertices @xmath244 , @xmath245 , respectively .
since the plasmon frequency @xmath246 is usually much larger than the phonon frequencies in 3d metals , in the limit @xmath182 the effective coupling is @xmath247 with @xmath248 denoting the average value of @xmath78 over the fermi surface .
thus the bare el - ph coupling is normally well screened .
turning now to the cuprates , from the point of view of coulomb interactions , the materials are 3d with @xmath249 and @xmath250 .
however , due to the largely incoherent @xmath1-axis transport observed across the phase diagram ( apart from the overdoped side ) , the polarizability is largely determined by the planar conduction electrons and @xmath251 . as a result ,
important changes to the effects of screening occur for small @xmath252 .
a qualitative feel for the effects of poor screening can be obtained by considering a coupling which is independent of fermion momentum @xmath60 and in the limit of small in - plane momentum transfers , where @xmath253 .
this limit is most relevant for raman active phonons and @xmath254 , with @xmath255 and @xmath256 , with @xmath257 where @xmath258 is the unit cell volume and @xmath259 is the fermi distribution .
the screened el - ph interaction is then @xmath260 with @xmath261 .
thus @xmath1-axis raman - active phonons in the cuprates , with couplings strongest for small momentum transfers @xmath24 and @xmath262 , should survive the effects of screening .
to evaluate the polarizabilities in the cuprates , we use the standard lindhard expression @xmath264 with all wavevectors being two - dimensional .
this form of the polarizability has been considered in previous works focusing on the role of screening and reduced dimensionality in the doping - dependent softening of the in - plane bond - stretching modes.@xcite for the coulomb interaction @xmath265 , more appropriately for the cuprates , we consider a layered system with charge density essentially confined to 2d layers , coulombically coupled across the planes:@xcite @xmath266 where @xmath267 , @xmath268 with @xmath269 the in - plane , out - of - plane dielectric constants , respectively .
the @xmath1-axis lattice constant sets the in - plane momentum scale beyond which the planes become effectively uncoupled , and leads to an interpolation between the 2d and 3d form for the coulomb interaction where @xmath270 provides the @xmath271 dispersion of the interaction .
( color online ) ( a ) plots of the real ( black ) and imaginary ( @xmath272 , red ) parts of the charge susceptibility @xmath273 and ( b)-(f ) the renormalized el - ph coupling @xmath274 in arb . units for @xmath275 and @xmath276 for different values of @xmath271 as indicated by the legend . for ( b)-(f )
the black dashed line corresponds to the bare coupling @xmath78 .
a momentum - independent bare coupling is considered in ( b ) , followed by the @xmath44 , @xmath2 , apical and breathing modes in ( c)-(f ) , respectively . in the case of the breathing modes a small @xmath277 component
has been added to visualize the effects of screening at small @xmath53 .
the @xmath278 axis in all plots is on a logarithmic scale to highlight the small @xmath53 region . ]
the real and imaginary parts of the pure charge polarizability @xmath273 are plotted in fig .
[ fig : pol]a . here , @xmath279 mev has been included to smooth singularities , @xmath116 is taken for optimal doping in bi-2212 , @xcite @xmath280 , @xmath281 , and in - plane@xcite and out - of - plane @xcite conductivities yield @xmath282 for the high- frequency dielectric constants , which determine the plasma frequency . for this choice in parameters we obtain @xmath283 ev at optimal doping in agreement with ref . .
in order to calculate the screened el - ph vertex the static dielectric constants must be used in eq .
( [ eq : vq ] ) . here , we take @xmath284 , which is obtained from optical measurements , @xcite and set @xmath285 .
this choice in ratio results in a factor two reduction in the breathing vertex at @xmath286 which is consistent with ref . .
additionally , we have set @xmath287 mev in fig . [
fig : pol]a .
the real part of @xmath273 rises as @xmath288 before abruptly falling as the imaginary part rises when the condition @xmath289 is satisfied .
these kinematic constraints give similar forms for the mixed polarizabilities @xmath290 , although the detailed @xmath129 dependence may be slightly different . in fig .
[ fig : pol]b - f several renormalized ( screened ) el - ph vertices @xmath291 are plotted for each of the modes considered in this work . here
@xmath292 and the momentum transfer is taken along the zone face @xmath276 and @xmath240 is plotted for several values of @xmath271 as indicated . to illustrate the effects of poor screening of the long - range coulomb interaction , in fig .
[ fig : pol]b , a momentum - independent bare coupling @xmath293 ( arb .
units ) is also shown . for a momentum - independent vertex @xmath294 . in this case , for @xmath295 and @xmath296 screening is perfect , leaving only a small el - ph interaction at large @xmath53 .
this can be seen as the black line , which corresponds to @xmath296 in figs .
[ fig : pol]b - f . for finite @xmath271 a small cone of wavevectors satisfies @xmath297 where screening is inoperable and the renormalized vertex recovers the bare value .
this enhancement of the small @xmath129 coupling occurs over a small range of @xmath129-vectors until @xmath298 satisfy @xmath299 , where a logarithmic divergence occurs which is cut - off by damping . for increasing @xmath271 this condition moves to a progressively larger @xmath53 and a window of momentum points
opens where the vertex can be well - represented by its bare value .
likewise , fig .
[ fig : pol]c - f plot the renormalized couplings for the @xmath44 , @xmath2 , apical and breathing modes , respectively .
for large in - plane momentum transfers @xmath53 the @xmath1-axis couplings are largely unaffected by screening due to the fall - off of the polarizability at large @xmath53 .
however , for the breathing branch , where the bare coupling weights large momentum transfers , the growth of the mixed polarizability leads to an overall suppression of the screened vertex at large momentum transfers .
for small in - plane momentum transfers - relevant for @xmath1-axis raman active phonons considered here - screening is ineffective and the coupling is anomalously enhanced over large @xmath129 coupling .
again , for the breathing branch , only small effects are noticed at small @xmath53 due to the nature of the deformation - type coupling for these modes . since we have neglected any fermionic dependence on the bare breathing couplings ,
the mixed polarizability is equal to the pure charge polarizability in this case .
( color online ) plots of the momentum dependent coupling @xmath300 , eq .
( [ eq : lambda_q ] ) , for @xmath301 @xmath44 ( a ) , @xmath2 ( b ) , apical ( c ) and bond - stretching ( d ) modes .
the coupling are plotted on a logarithmic scale for @xmath53 to highlight the region of poor screening . ] using eq .
( [ eq : gbar ] ) the momentum - dependent el - ph coupling strength @xmath302 @xmath303 is plotted in fig .
[ fig : lambda_q2d ] as a function of @xmath53 ( on a log scale to highlight the small @xmath24 behavior ) for @xmath301 @xmath44 , @xmath2 , apical and bond - stretching modes shown in panels fig .
[ fig : lambda_q2d](a)-(d ) , respectively .
one can clearly see that when summed over all out - of - plane momentum transfers @xmath271 , the net coupling is on the order of the bare coupling for small @xmath53 , while the coupling at large @xmath24 is suppressed .
the fermionic momentum dependence of the bare vertices for the @xmath2 case noticeably alters @xmath304 for momentum transfers along the bz diagonal where it is largely projected out .
the @xmath44 and apical @xmath304 are quite similar even though the anisotropy of the bare interaction is substantially different and weights different regions of the bz .
for all the @xmath1-axis modes , coupling is small for @xmath305 .
since we have neglected hubbard short - range coulomb repulsion , which further suppresses coupling at large momentum transfers,@xcite raman @xmath1-axis modes are thus not expected to appear in transport measurements .
likewise , the breathing modes are altered at small @xmath53 , but the overall coupling does not give much weight for these transfers . since the apical and @xmath2 modes strongly favor coupling to the anti - nodal fermions , we may conclude that the underlying density of states , combined with the fermionic momentum dependence , strongly influences the magnitude of the couplings at small @xmath53 ; the phonon and eigenvectors control the actual angular dependence while the nature of the coupling - electrostatic or deformation - controls the @xmath306-dependence .
( color online ) plots of the @xmath25-wave coupling @xmath307 , eq .
( [ eq : lambda_q_d ] ) , for @xmath301 @xmath44 ( a ) , @xmath2 ( b ) , apical ( c ) and bond - stretching ( d ) modes .
the coupling are plotted on a logarithmic scale for @xmath53 to highlight the region of poor screening . ]
this @xmath24-dependence strongly affects the @xmath25-wave projected pair interaction as well . in fig .
[ fig : lambda_d_q2d ] we plot the @xmath25-wave projected coupling @xmath308 for the same modes considered in fig .
[ fig : lambda_q2d ] . as with fig .
[ fig : lambda_q2d ] , @xmath307 has been plotted on a log scale in order to highlight behavior at small @xmath53 .
one can see straightforwardly that the phonons that strongly favor small @xmath24 scattering largely promote @xmath25-wave pairing ( the @xmath1-axis modes ) while phonons that favor large @xmath24-scattering , such as the bond - stretching modes , are detrimental .
furthermore , for the bond - stretching modes , while a large region of @xmath129 space supports pairing , the large weight near @xmath309 dominates the coupling .
the consequence of poor screening for finite @xmath271 transfers , which accentuates small @xmath129 couplings , is to enhance the overall @xmath25-wave coupling compared to @xmath296 as well as standard debye screening ( @xmath310 ) for @xmath1-axis modes , and diminishes the repulsive part for the bond - stretching modes . to visualize the effect of screening on the fermionic dependence of the coupling , we plot in fig .
[ fig : lambda_fs_screened ] the screened @xmath311 , derived from eq .
( [ eq : lambdafs ] ) where @xmath312 has been replaced with @xmath240 from eq .
( [ eq : gbar ] ) . here , to mimic the effect of doping in the cuprates , we have varied the plasma frequency to smaller values in accordance with experiment , reflecting the increased insulating behavior and ionicity along the @xmath1-axis that occurs with underdoping in the phase diagram of hole- doped cuprates .
the values for the total coupling @xmath313 and @xmath228 defined in eq .
( [ eq : lambdad ] ) are shown in fig .
[ fig : lambda_total ] .
( color online ) plots of the screened e - ph @xmath314 for fermions at the fermi level as a function of @xmath221 for several values of the plasma frequency @xmath246 .
the value of @xmath246 varies from 0.289 ev ( black , solid ) to 0.914 ev ( purple , dashed ) .
the black arrows indicate the direction of increasing @xmath246 .
the parameters used to generate this figure are defined in the text . ]
screening causes several noticeable effects compared to the unscreened case shown in fig , [ fig : lambda_fs ] . in order to mimic the effects of doping the screened el - ph vertex for the four branches
is shown for @xmath246 ranging from 290 to 914 mev .
( the black arrows indicate the direction of increasing @xmath246 . ) as the plasma frequency is increased the @xmath44 , apical and breathing branches can be screened more effectively and the overall vertex is lowered around the fermi surface . in the case of the @xmath44 and apical @xmath1-axis branches ,
the largest effect occurs in the anti - nodal region where the bare vertex is largest . in the case of the @xmath2 branch
there is an anomalous anti - screening which occurs and the @xmath2 vertex in the antinodal region is enhanced with increasing @xmath246 .
this non - intuitive result is due to the out - of - phase oscillations of the oxygen modes since the only the difference between the bare vertices for the @xmath44 and @xmath2 branches is the phase of the phonon eigenvectors . from these results
it is evident that the self - energy due to coupling to @xmath1-axis phonons should redistribute weight around the fermi surface as the number of doped holes varies .
thus , we infer that in the cuprates a window in @xmath129-space at small @xmath53 occurs in which the el - ph interaction can still be quite large and avoid screening .
this would not occur if the material was fully conducting along the @xmath1-axis and this effect increases with underdoping as @xmath246 is reduced with decreasing hole concentration .
in contrast , since poor screening does not affect large momentum transfers , the strength of the breathing coupling dramatically increases for smaller @xmath246 .
this implies that the electron self - energy contribution from the breathing branches should grow with underdoping .
once again , however , we remark that this finding must be viewed with caution since large @xmath24-behavior is governed strongly by coulomb interactions . finally , in fig . [
fig : lambda_total ] we plot the total coupling @xmath232 resulting from the screened el - ph vertex as a function of @xmath246 . following the trends in fig .
[ fig : lambda_fs_screened ] , the total coupling in both channels decreases with increasing @xmath246 for the in - plane breathing branch , as well as the apical and @xmath44 branches , while the coupling to the @xmath2 branch is enhanced . in the case of the @xmath25-wave coupling , the attractive interaction of the @xmath44 branch is largely cancelled by the repulsive interaction of the breathing branch .
therefore , the total attractive interaction for @xmath25-wave pairing , which is primarily provided by the @xmath2 branch , is expected to be enhanced with progressive overdoping .
( color online ) plots of @xmath226 ( left ) and @xmath228 ( right ) as a function of the plasma frequency @xmath246 .
the parameters taken are defined in the text . ]
the transition from insulating to more metallic behavior , reflected in @xmath246 , is one aspect of the doping dependence expected for el - ph coupling .
however , this is not the only change expected to occur with doping . as a sample
is doped away from half - filling the character of the @xmath17-@xmath18 band is expected to change as spectral weight redistributes itself within the quasiparticle and lower and upper hubbard bands.@xcite this will be reflected in the parameters @xmath179 , @xmath186 and @xmath180 . since these parameters enter the el - ph vertices in the fourth power , doping - induced changes in the bandstructure can have a large impact on the strength of the el - ph coupling .
the value of the local crystal fields are also expected to vary with doping .
this was shown in ref . , which examined the role of an interstitial oxygen dopant in the sro / bio layers of bi-2212 and found that the dopant locally enhanced @xmath108 at the planar oxygen site by up to a factor of five .
the structural details of the crystal can also induce changes to the character of the band at fermi level ( @xmath179 , @xmath180 and @xmath186 ) , through variations in the madelung energies of the atoms and the degree of hybridization between the cuo@xmath8 plane and off - plane atoms such as the apical oxygen .
due to these considerations the overall strength of the el - ph coupling can have a complex dependence on both the carrier concentration of the cuo@xmath8 plane as well as the composition and structure of the material . in the next section
we examine the materials dependence of the el - ph interaction by conducting a systematic examination of the role of structure in determining the strength of the interaction across various families of high - t@xmath0 cuprates .
the environment around the cuo@xmath8 plane influences electron dynamics in the plane via the madelung energies and local crystal fields . as a result
, local symmetry breaking plays an important role in lattice dynamics .
for example , x - ray measurements on yba@xmath8cu@xmath13o@xmath47 @xcite have shown that the cuo@xmath8 planes are statically buckled , and a linear el - ph coupling results from the breaking of local mirror plane symmetry .
lda calculations for the infinite layer material cacuo@xmath8 also found evidence for static buckling of the plane due to the steric interactions among the oxygen 2@xmath20 orbitals which produces a substantial linear @xmath2 el - ph coupling.@xcite among the different contributions to local symmetry breaking , this section is devoted to an investigation of the magnitude of the madelung energies and local fields across families of the cuprates as a mechanism of local symmetry breaking .
we remark that lda investigations of el - ph coupling are limited mostly to stoichiometric ybco , with virtual crystal extensions to lsco and cacuo .
lda bandstructure , being metallic in nature , will tend to screen charge variations whereas those charge variations , particularly off the cuo@xmath8 plane , may be unscreened and result in strong local modifications of band parameters triggering nano - scale inhomogeneity as well as strong variations of el - ph coupling .
this was demonstrated in ref .
, where it was shown that the presence of an interstitial oxygen dopant s unscreened charge in the sro / bio layers produced large local enhancements in the crystal field strength at the planar oxygen sites .
this resulted in a local increase in the strength of the coupling to the @xmath44 and @xmath2 branches , which in turn reduced the energy gap and broadened the spectral features in the hole addition / removal spectrum as well as locally increased @xmath4 via the gain in lattice energy . here , we focus our calculations in the ionic limit , where the ions are represented as point charges , and evaluate the electrostatic sums using ewald s method .
we show that the madelung potential landscape is sensitive to the details of the unit cell , producing material - dependent variations in the field strength at the planar and apical oxygen sites . for the field at the oxygen site of the outermost cuo@xmath8 plane , which determines the coupling to the @xmath44 and @xmath2 modes
, this variation mirrors the observed variations in t@xmath0 suggesting a link between these quantities .
the use of formal valences in the electrostatic calculations reflects the stociometric compounds and therefore the estimates obtained from their use corresponds to the parent compounds .
however , in light of the findings of ref . , we also address doping - induced changes in the crystal field strength in bi-2212 in order to assess how these fields are expected to vary with doping .
finally , the variations in @xmath232 that arise as a function of material due to the electrostatic considerations presented here are also discussed . here
we focus on the @xmath2 mode , which has the largest contribution to @xmath25-wave pairing and take in into account variations in the local field , relative shifts in site energies , and changes in the atomic character of the band , all of which are directly affected by the details of the crystal structure .
( color online ) the hgba@xmath8ca@xmath39cu@xmath40o@xmath41 ( @xmath315 ) unit cell showing the locations of the mirror plane ( yellow , mp ) , low symmetry breaking plane ( green , lsb ) and high symmetry breaking planes ( red , hsb ) .
the degree of symmetry breaking establishes the strength of the electric field which couples to the @xmath1-axis phonons . ] a structural view of the lattice across the hg family , as shown in fig .
[ fig : structure ] . starting with @xmath316 shows that the coupling to planar oxygen @xmath1-axis phonons must be weak , while couplings involving the apical oxygen phonon can be strong due to the asymmetry along the @xmath1-axis at the apical site ( cuo@xmath8 on one side and hgo on the other ) .
for @xmath317 the outer layer does not lie in a mirror - plane and the coupling strength depends on the local asymmetry ( in this case bao on one side , ca on the other ) .
however , the innermost plane may lie in a mirror plane when @xmath9 is odd , while the intermediate planes will generally have a weaker degree of symmetry breaking in comparison to the outermost plane .
thus , in the context of el - ph coupling , the most strongly coupled modes will be those involving displacements of the apical and planar oxygen atoms in the outermost layers . in hg compounds
the local asymmetry around the outermost plane is due to ca and bao structures on opposite sides of the cuo@xmath8 plane . in ybco
it is due to ca and sro structures .
such changes in composition will affect the magnitude of the crystal field and these differences can be quantified by examining the madelung potential for the apical and planar oxygen sites for the hg , tl and bi cuprate families .
the electric field strength can be regarded as the first derivative of the madelung potential , which consists of a contribution of all electrostatic interactions between ions in the solid . for a test charge @xmath318 , located at position @xmath319 ,
the madelung potential @xmath320 is defined as : @xmath321 where @xmath322 is the charge density . in the ionic
limit the charge density is comprised of point charges @xmath323 , where the sum @xmath324 is taken over the atoms of the crystal located at @xmath325 .
( color online ) the madelung potential at the cuo@xmath8 planar oxygen site as function of distance from the plane of mirror symmetry : ( a ) hgba@xmath326ca@xmath39cu@xmath327o@xmath328 ( @xmath9 = 1 - 6 , red circles ) .
( b ) tl@xmath326ba@xmath326ca@xmath39cu@xmath327o@xmath328 , ( @xmath9 = 1 - 4 , blue squares ) .
( c ) bi@xmath326sr@xmath326ca@xmath39cu@xmath327o@xmath328 ( @xmath9 = 1 - 3 , green triangles ) .
( d ) an overlay of the hg , tl and bi plots rescaled by the lattice spacing , @xmath244 .
the black line is a fitted curve of the form @xmath329 . ]
there are several common methods used to calculate the madelung potential .
a general and powerful method was given by ewald,@xcite which is used here to determine the madelung energies for the planar and apical oxygen sites in the outermost cuo@xmath326 planes in the unit cell for bi ( n=1 - 3 ) , tl ( n=1 - 4 ) and hg ( n=1 - 6 ) cuprates .
the atomic positions are obtained from the known structural data given in refs . .
in all cases formal valence charges for the ions have been assumed .
it should be noted that for the materials considered herein , the surfaces are non - polar , as opposed to yba@xmath8cu@xmath13o@xmath47 ( ybco).@xcite therefore the surface termination layer is not critical for our calculations . in the case of ybco , assuming that the surface is terminated between the bao and cuo sublayers , a local field between 0.8 - 2.1 ev / was obtained , arising from the asymmetry condition of y@xmath330 and ba@xmath331 on either side of the cuo@xmath8 plane.@xcite this polar surface contribution overwhelms the periodic contribution from ewald s method . without the polar contribution , the coupling for example to the @xmath2 mode is underestimated when compared to that obtained from periodic lda methods . in ref .
it was shown how deviations away from formal valences can give much different values for the local crystal fields .
we have also found that the local electric fields are more sensitive than the madelung energy to small structural distortions and deviations away from formal valences .
therefore we conclude that while the overall field values should be viewed as being approximate , we assume that the variations across the different materials can be qualitatively compared .
( color online ) the madelung energy of the apical oxygen site as a function of distance from the plane of mirror symmetry : ( a ) hgba@xmath326ca@xmath39cu@xmath327o@xmath328 ( @xmath9 = 1 - 6 , red circles ) .
( b ) tl@xmath326ba@xmath326ca@xmath39cu@xmath327o@xmath328 , ( @xmath9 = 1 - 4 , blue squares ) .
( c ) bi@xmath326sr@xmath326ca@xmath39cu@xmath327o@xmath328 ( @xmath9 = 1 - 3 , green triangles ) .
( d ) an overlay of the hg , tl and bi plots in addition of points obtained for lsco and lbco ( black open symbols ) .
the dashed and solid black lines are fitted curves of the form @xmath329 .
the dashed curve is obtained when all of the data is included while the solid curve is obtained when the la data points are excluded . ]
the madelung energies of the planar and apical oxygen sites are plotted in figs .
[ fig : madelungb1 g ] and [ fig : madelungapex ] , respectively , and plotted as a function of the distance between outermost cuo@xmath8 plane ( or apical oxygen ) and the mirror plane located at the center of the unit cell .
the calculated apical madelung energies are in agreement with those obtained in ref . for similar structural data , while the madelung energies for the planar sites are , to the best of our knowledge , presented here for the first time .
one can see that the madelung energies for both the planar and apical sites increase when the outermost cuo@xmath8 plane lies further from the center of the unit cell .
this occurs both when the number of planes per unit cell is increased as well as when the layers are spaced closer to the charge reservoir due to the larger spacer ions between the cuo@xmath8 layers .
scaling the madelung energies by the in - plane lattice constant @xmath244 collapses the data for each cuprate family onto a single curve , as shown in subpanel ( d ) of figs .
[ fig : madelungb1 g ] and [ fig : madelungapex ] . by symmetry
, the electric field falls to zero at a mirror plane implying that @xmath332 must have zero slope at these points . if @xmath333 is measured relative to the mirror plane , such as those indicated in fig . [
fig : structure ] , then the simplest function that satisfies these boundary conditions is @xmath334 , and serves as an overall guide for describing the behavior of the data . for the planar oxygen sites we obtain @xmath335 , @xmath336 while for the apical oxygen sites @xmath337 , @xmath338 ( @xmath339 , @xmath340 when the points for the la family are excluded ) , in units of @xmath341 . fig .
[ fig : efield ] shows the local @xmath342 fields at the apical and planar oxygen sites for the outermost cuo@xmath8 plane in the unit cell for a number of cuprate families having different numbers of layers . for the single layer cuprates
the cuo@xmath8 plane lies in a mirror plane located at the center of the unit cell resulting in a zero field at the planar oxygen sites .
thus there is no el - ph first order coupling to the @xmath44 and @xmath2 branches in the single layer materials .
this is no longer the case for @xmath317 .
examining the largest data set , that of the hg family ( @xmath343 ) , the @xmath342-field at the planar oxygen site rises from @xmath316 and peaks for @xmath344 where the location of the outer plane of the cell is @xmath345 .
as @xmath9 is increased beyond this point the magnitude of the crystal field begins to decrease .
the data for the bi- and tl - families are consistent with this trend - generally the local fields rise and follow the associated madelung energy as the outermost plane moves further from the center of the cell , and falls once the outermost plane moves closer to the edges of the unit cell where the madelung energy profile flattens . since the apical oxygen does not lie in a mirror plane for any cuprate , a finite field is allowed even for @xmath346 .
behaviour similar to the planar oxygen site is also seen with a peak field strength generally occurring for @xmath344 and decreasing beyond this point .
( color online ) ( a1)-(a3 ) the magnitude of the electric field at the apical oxygen site of the hg , tl and bi families of cuprates , respectively .
the data is plotted as a function of the apex distance from the plane of mirror symmetry located at the center of the cuo@xmath8 planes .
( b1)-(b3 ) the magnitude of the electric field at the planar oxygen site of the outermost cuo@xmath8 plane for the set of same compounds .
( c1)-(c3 ) the strength of the electric field scaled by the unit cell lattice parameters for the planar and apical oxygen fields , respectively .
the dashed lines represents a fit @xmath347 , where a = 7.1 and 83.9 ev for the planar and apical sites , respectively .
] the scatter in values is largely due to uncertainties in structural data as well as the oversimplified use of formal valences for all materials .
however , the results are similar among the different cuprates if the position of the outermost plane relative to the center of the cell is considered . figs .
[ fig : efield]c1 and [ fig : efield]c2 show that the local field may be collapsed onto a sinusoidal curve as a function of @xmath1-axis distance when rescaled to account for the unit cell volume variations .
the presented @xmath342-field values are sensitive to the structural data , particularly for the planar oxygen sites where the degree of static cu - o bond buckling is relatively uncertain .
since a static buckling further breaks mirror plane symmetry across the cuo@xmath8 plane , small variations in the degree of buckling can have a large impact on the field values obtained while only having a minor effect on the madelung energies .
for example , neglecting buckling on the order of @xmath348 in the @xmath349 layer materials results in @xmath342-field variations on the order of 20 - 40 percent for the planar oxygen sites , while the variations in the corresponding madelung energies vary only by a few percent .
we have found that both the madelung energies and local fields for the apical oxygen are not as sensitive to the degree of static buckling in the cuo@xmath8 plane .
we now comment on the @xmath342-field values reported here in the context of previous works involving coupling to the @xmath1-axis modes . due to the fact that ba@xmath331 and sr@xmath331 lie on either side of the cuo@xmath8 plane in bi-2212 , previous expectations were that the planar field in this system would be smaller than that found in ybco and work by one of the authors on the raman derived @xmath2 phonon lineshape confirmed these expectations.@xcite since the t@xmath0 s of these two materials are roughly equivalent
, it was concluded that the el - ph coupling was irrelevant to the pairing mechanism .
however , a re - examination of the data set , in view of the greater sample inhomogeneity and impurity effects in bi-2212 compared to ybco , indicates empirically that the field may not be as small as previously considered .
using the theory of ref . , the fano line - shape of the @xmath2 phonon in optimally doped bi-2212@xcite can be fit equivalently with a value of the coupling , parameterized by @xmath199 , which is an order of magnitude different if one assumes a large , intrinsic broadening of the phonon line . since the values of @xmath199 extracted from fitting raman lineshapes
are sensitive to disorder for values of @xmath199 in the range of those found in yba@xmath8ca@xmath13o@xmath350 , a direct comparison with other compounds should again be considered qualitatively , with a view towards materials of comparable quality .
our ewald calculations confirm that the field at the planar oxygen site for bi-2212 ( @xmath351 ev / ) is similar to yba@xmath8cu@xmath13o@xmath350 .
we show in the next section that this value is likely underestimated once considerations for doping are taken into account .
we now briefly comment on the doping dependence of the local crystal fields . in discussing the trends in @xmath108 across the cuprates
we have assumed formal valences for each lattice site , which is appropriate for undoped parent compounds .
however , as was shown in ref . , the doping process can significantly affect the strength of these fields resulting in a substantial local increase in @xmath199 s . from a symmetry point of view
, the movement of charge from the cuo@xmath8 plane to the charge reservoir layers with increased doping will enhance local symmetry breaking across plane . to model this effect in bi-2212 we repeated the calculation for @xmath108 , but this time with the charge of the cuo@xmath8 planes uniformly raised by @xmath352 , corresponding to optimal doping . in order to maintain charge neutrality the charge of the bio and sro layers , where the o dopants are known to sit,@xcite
were uniformly reduced by the appropriate amount . using this crude model
we find the @xmath108 field at the planar site is raised to @xmath353 ev / while the field at the apical site is reduced to @xmath354 ev / , in agreement with the general trends previously reported.@xcite in section iii we were interested in determining the values of @xmath232 at optimal doping and therefore these are the values of the crystal field strength which were used in section [ section : lambda ] to obtain the unscreened @xmath355 and total @xmath232 s . throughout this paper
we have discussed a number of aspects of el - ph coupling in the cuprates which are affected by doping the system .
the local @xmath342-field strength , the ionicity of the compound , characterized by @xmath246 , and the orbital character of the band at the fermi level all modify the el - ph couplings and are all affected by doping .
therefore , el - ph coupling in the cuprates is expected to exhibit a strong doping dependence .
we also note that many of these parameters are also expected to vary between materials comprising the various families and we turn our attention to these trends in the following section .
we end this section by examining the systematic variation in the el - ph coupling strength expected from material to material .
since our goal is to understand the materials variations in t@xmath0 we focus our attention on coupling to the @xmath2 branch , which provides the largest contribution to pairing and whose coupling is the strongest after the effects of screening are taken into account . as discussed previously
, there is a empirical relationship between t@xmath0 and the distance of the apical oxygen from the cuo@xmath8 plane and this relationship this has generally been tied to the effective increase in the next nearest neighbour hopping @xmath5 through hybridization effects associated with the cu 4@xmath16 orbital .
these hybridization effects will also alter the character of the @xmath17-@xmath18 band crossing the fermi level and can therefore affect the overall strength of the coupling to all of the phonon branches considered in this work . for the @xmath2 branch ,
such changes are reflected in @xmath179 and , since this enters the total coupling as the fourth power , changes in hybridization can have a large affect on the total coupling strength .
( color online ) ( a ) the planar oxygen character at the fermi surface @xmath356 as a function of the material s t@xmath0 .
( b ) the corresponding local @xmath342-field at the planar oxygen site of the outermost cuo@xmath8 plane in the parent compound .
( c ) the value of @xmath226 for the @xmath2 modes of the outermost plane for each material . ] in fig .
[ fig : oxygencharacter]a the systematic variation of @xmath179 is presented as a function of @xmath357 for the same materials considered in the fig .
[ fig : efield ] . to estimate @xmath179 we consider the same five - band model used in section [ section : lambda ] , but assign the site energies based on the madelung calculations presented here and the energy level scheme of ref . .
using hole language , we set @xmath358 , @xmath359 , @xmath360 and @xmath361 . here , @xmath362 is the charge transfer energy , which is related to the difference in madelung energies between the cu and planar o sites @xmath363 and is given by @xmath364 where @xmath365 and @xmath366 are the second ionization and second electron affinity energies for the cu and o sites , respectively .
the factor of @xmath367 represents the contribution of the coulomb interaction between the introduced electron - hole pair . following ref . , we take the dielectric constant @xmath368 and @xmath369 ev . in determining @xmath370 , @xmath371 denotes the madelung energy difference between the apical and planar oxygen sites .
in addition to varying the site energies between materials , the values of the hoppings @xmath372 and @xmath373 are also adjusted using the prescription of ref . .
this allows us to account for changes in hybridization due to the different apical distance given in the experimental structural data .
furthermore , we neglect any variation of the @xmath244 and @xmath245 lattice constants and hold the in - plane hybridizations fixed throughout .
finally , correlations are handled at the mean field level .
the overall effect of the mean field corrections is to shift the site energies @xmath374 and @xmath375 is fixed to maintain the difference @xmath376 ev . the hubbard repulsions are set to canonical values with ( in ev ) @xmath377 , @xmath378 and @xmath379 . in eq .
( [ eq : meanfield ] ) , @xmath380 denotes the total number operator for orbital @xmath54 and site @xmath324 and the paramagnetic solution for @xmath381 has been assumed .
the results for @xmath179 , @xmath108 and @xmath232 are shown in fig .
[ fig : oxygencharacter ] .
the e - fields presented here are the same as those used in fig .
[ fig : efield ] , and were obtained by assigning formal valences for each ion .
these field values reflect best the crystal field of the undoped parent compounds and therefore fig .
[ fig : oxygencharacter ] does not include the effects of doping .
as we have already seen , even in this simple model , the redistribution of charge from the charge reservoir to the cuo@xmath8 plane can alter the crystal field values resulting in large changes in the value of @xmath232 . in the case of bi-2212
this produced an order of magnitude change in the value of @xmath232 for the factor 4 change in @xmath108 in the doped lattice .
furthermore , the ionic point charge model neglects any covalent nature of the bonds which will smooth the charge distributions .
therefore , the field strengths in real materials may be quite different from the values reported here , especially in the doped systems .
however , by using the same method for each material we can obtain valuable comparative information about the crystal field variations between materials . as shown in fig .
[ fig : oxygencharacter]a , there is a correlation between the material s
maximum t@xmath0 at optimal doping ( apical distance ) and the planar oxygen character @xmath179 .
again , this can be understood in terms of the increase in the effective planar o - o hopping _ via _ the cu 4@xmath16 as the apical oxygen is further removed from the cuo@xmath8 plane . a similar correlation , shown in fig .
[ fig : oxygencharacter]b , occurs between @xmath108 at the planar oxygen site and t@xmath0 for the reasons previously discussed . since the total couplings @xmath232 for the @xmath2 branch are proportional to the product @xmath382 a clear correlation between the material s t@xmath0 and the @xmath2 coupling naturally arises .
this is shown in fig .
[ fig : oxygencharacter]c . 0.5 cm the double square well model for @xmath383 and @xmath384.,title="fig : " ] the correlation between t@xmath0 and the strength of the coupling to the @xmath2 branch , which provides the largest contribution to @xmath25-wave pairing , provides a natural framework for thinking about materials variations in t@xmath0 .
although it is clear that the overall strength of the el - ph interaction is too small to support htsc on its own , @xcite phonons are not precluded from contributing to pairing , enhancing the pairing correlations from another dominant interaction .
such a possibility has been considered in previous works that have examined the contribution of el - ph coupling in conjunction with antiferromagnetic spin fluctuations.@xcite in such a scenario , a baseline t@xmath0 is set by the dominant pairing mechanism , typically identified with antiferromagnetic spin fluctuations .
such an interaction is likely to be governed by the properties of the cuo@xmath8 plane , such as the charge transfer energy @xmath362 or antiferromagnetic exchange energy @xmath4 , and therefore the strength of this interaction is likely independent of the material .
t@xmath0 is then further enhanced by the weaker contribution from the el - ph interaction which , as we have shown , is strongly materials dependent .
this picture naturally explains the variations in t@xmath0 that are observed across materials and provides a microscopic connection between the structure and composition of the unit cell and the t@xmath0 of the material . in the next section
we continue examine such a multi - channel model and present simplified considerations for t@xmath0 in order to assess the degree to which phonons can enhance a dominant interaction at high energy . as we will show , moderate values of @xmath232 , such as those given in fig .
[ fig : oxygencharacter]c , can lead to sizeable enhancement to t@xmath0 which is greater than the t@xmath0 that would result from phonons alone .
phonons are therefore capable of providing a sufficiently large enhancement to t@xmath0 despite the relatively low value of their total coupling @xmath232 .
in this section we turn our attention to considerations for t@xmath0 by considering contributions to pairing from multiple channels .
t@xmath0 , one of the most difficult quantity to calculate correctly , varies across the cuprate families by two orders of magnitude even though the same cuo@xmath8 building block is present .
however , as we have demonstrated , the total el - ph coupling for the @xmath2 branch shows a strong materials dependence which is empirically correlated with the observed value of t@xmath0 at optimal doping .
therefore it is of interest to explore how el - ph interactions may provide a material dependence to t@xmath0 arising from the material conditions along the @xmath1-axis and poor screening . to calculate t@xmath0 we consider a multi - channel boson exchange model for pairing , which is a straightforward generalization of the eliashberg equations .
a similar approach has recently been used@xcite to account for the qualitative structures in the phonon - modulated density of states of the cuprates as observed by stm.@xcite implicit in this calculation is the assumption that the dominant pairing channel can be described in terms of a boson exchange mechanism and such a scenario is the simplest one which can be adopted in examining the idea of phonon - enhanced htsc .
however , such a model can be perhaps be justified based on several recent works .
for example , studies of the penetration depth in electron - doped cuprates using a model with coexisting afm and sc order has produced good agreement with experiment indicating the importance of afm in these systems.@xcite furthermore , a dmft study of optical conductivity in a moderately correlated single - band hubbard model,@xcite has concluded that afm with moderate correlations is the correct picture for describing the underdoped cuprates rather than strongly correlated mott physics .
finally , a study of the @xmath3-@xmath4 and hubbard models has shown that the @xmath25-wave pairing in these models have a significant contribution from a retard component with an energy set by a few times @xmath4 or a fraction of @xmath7 in the @xmath3-@xmath4 and hubbard models , respectively.@xcite this result indicates that the pairing interaction in these models is largely retarded in nature with an energy scale set by the dynamics of the magnetic excitation spectrum .
these theoretical studies lend some support to the spin - fluctuation mediated picture adopted here although the details may differ .
nevertheless , this model provides a simple means to evaluate the role of phonons in modifying t@xmath0 in a multi - channel model . for this calculation
we assume that each channel can be described in terms of a boson exchange mechanism .
individual channels are indexed by @xmath385 and the el - boson contributions to the mass renormalization @xmath384 and anomalous self - energy @xmath386 are parameterized by @xmath387 and @xmath388 , respectively . making the standard approximations of a structureless band and cylindrical fermi surface ,
the multi - channel eliashberg equations are @xmath389&= & \sum_{\nu } \frac{\lambda_{\nu , z}\omega_\nu}{2 } \int_0^\infty d\omega^\prime \left\langle \frac{\omega^\prime } { \sqrt{\omega^{\prime2 } - \delta^2(\omega)\cos^2(2\theta)}}\right\rangle k_{+}^\nu(\omega^\prime,\omega)\\ \delta(\omega)z(\omega)&= & \sum_{\nu } \frac{\lambda_{\nu,\phi}\omega_\nu}{2 } \int_0^\infty d\omega^\prime \left\langle \frac{\delta(\omega^\prime)\cos^2(2\theta)}{\sqrt{\omega^{\prime2 } - \delta^2(\omega^\prime)\cos^2(2\theta)}}\right\rangle k_{-}^\nu(\omega^\prime,\omega).\end{aligned}\ ] ] here we have assumed a separable form for the anomalous self - energy @xmath390 and @xmath391 is the complex gap function . the kernels in eq .
( [ eq : eliashberg ] ) are given by @xmath392 \bigg [ \frac{1}{\omega+\omega_\nu+\omega^\prime}\pm \frac{1}{\omega - \omega_\mu - \omega^\prime } \bigg ] + [ f(\omega^\prime)+n(\omega_\nu ) ] \bigg [ \frac{1}{\omega-\omega_\nu+\omega^\prime}\pm \frac{1}{\omega+\omega_\nu-\omega^\prime } \bigg]\ ] ] where @xmath259 and @xmath9 are the fermi and bose occupation factors , respectively .
following ref .
we adopt a double square - well model for the frequency dependence of @xmath383 and @xmath384 , as shown in fig .
[ fig : doublewell ] . here
, we identify @xmath393 with the @xmath394 mev @xmath2 phonon branch , and @xmath395 is identified with antiferromagnetic spin fluctuations with @xmath396 mev , reflecting the top of the magnon band in the case of a long - range antiferromagent .
next , we follow a standard set of approximations@xcite and obtain the coupled equations @xmath397 for @xmath398 and @xmath399 for @xmath400 . here
@xmath401 and @xmath402 .
( [ eq : gap1 ] ) and ( [ eq : gap2 ] ) are then solved for the non - trivial solution for @xmath403 and the resulting expression for t@xmath0 is @xmath404 where @xmath405 this expression recovers the mcmillan result ( we have neglected @xmath406 ) for the single well case in either limit when @xmath407,@xmath408 or @xmath409,@xmath410 are set to zero .
other expressions have been obtained for t@xmath0 using different forms of the two - well model and neglecting the difference in contributions from each of the momentum channels ( i.e. @xmath411 ) while taking the spin - fluctuation contribution to be repulsive.@xcite we refer the reader to ref . for a more thorough discussion of the various proposed expressions .
the transition temperature t@xmath0 as a function of el - ph coupling in the @xmath25-wave channel @xmath408 , calculated in the two - square well model , defined in the text .
the dashed line is t@xmath0 expected for the phonon alone . ] in order to assess the phonon contribution to t@xmath0 , we now assume that spin - fluctuations provide the dominant source for @xmath25-wave pairing and parameterize this mode with @xmath412 .
this choice results in a baseline t@xmath413 k in the absence of phonons . for the phonon mode
we assume @xmath414 mev and hold @xmath415 . in fig .
[ fig : tc]a the resulting t@xmath0 as a function of el - ph coupling is presented . for reference ,
the value of t@xmath0 obtained from phonons alone is also shown , multiplied by a factor of 10 .
it is important to note that even small el - ph coupling can enhance t@xmath0 considerably over the single - channel spin fluctuation model .
furthermore , the degree of enhancement depends on the ratio of @xmath416 assumed for the spin fluctuations . in fig .
[ fig : tc]b we plot t@xmath0 for the two - mode model while varying ratios of @xmath416 . in each case
we have adjusted @xmath410 such that the baseline of @xmath417 k is maintained for @xmath418 . in general ,
as @xmath416 increases in the dominant channel , the overall gain in t@xmath0 mediated by phonons also increases but with diminishing returns for larger @xmath416 . an isotropic repulsion in the spin fluctuation channel
is also expected to produce a similar effect .
we also note that our results are limited by the questionable applicability of migdal - eliashberg theory in the cuprates and therefore t@xmath0 itself is a poor metric of a theory to be compared to experiment .
nevertheless , as this enhancement is generic to multi - channel couplings , it may serve as a guide for understanding the material dependence of t@xmath0 itself . in fig .
[ fig : alpha ] the isotope exponent @xmath419 , where @xmath420 is the mass of the oxygen ion , is plotted using the same parameters as in fig .
[ fig : tc ] with @xmath421 , and @xmath422 defined by eq .
( [ eq : nu ] ) , reducing the overall magnitude of @xmath54 from the value for phonons alone .
a similar reduction in @xmath54 and enhancement of t@xmath0 has been reported in ref . when the phonon contribution to pairing was taken to be attractive in the @xmath26 channel .
[ fig : alpha ] shows that the overall isotope exponent is small even though the phonon mediated enhancement to t@xmath0 is sizeable and demonstrates that small isotope exponents are possible despite large enhancements in t@xmath0 due to the el - ph interaction .
some degree of caution is warranted in making a direct comparison to experiment where the isotope exponent is small at optimal doping and increases with underdoping,@xcite while our results show that the overall contribution of el - ph coupling to @xmath25-wave pairing is expected to increase with doping due to screening effects .
therefore one might expect that the isotope exponent should increase with doping , contrary to what has been observed experimentally .
however , doping dependent changes are also likely to occur for a spin - fluctuation mediated pairing mechanism . since it is the relative contributions from each channel that sets the value of @xmath54 a direct calculation of @xmath54 expected experimentally is not possible until the details of the dominate pairing interaction are understood .
nevertheless , these calculations show that phonons with relatively low el - ph coupling strengths can , in principle , play a substantial role in determining t@xmath0 while producing small signatures in traditional metrics such as the isotope exponent .
the isotope exponent @xmath54 as a function of el - ph coupling @xmath408 for the same parameter set used in fig .
[ fig : tc ] . ]
we end with a few comments about these results in relation to prior works .
as we have already noted , the specific results presented are limited by the applicability of eliashberg theory and the fermi liquid description of the cuprates .
such a picture may not be completely applicable , especially in the underdoped region of the phase diagram where the role of correlations becomes more prominent .
as the role of correlations and its interplay with the el - ph interaction is better understood the details of the picture presented here may change .
however , the phonon - mediated enhancement of t@xmath0 is expected remain as a general phenomenon .
we also note that the inclusion of correlations may further enhance the phonon contribution to t@xmath0 through possible enhancements of the el - ph vertex or proximity to afm .
for example , recent dmft work examining phonon - mediated pairing in the fullerides@xcite has found that a phonon - mediated pairing mechanism , which produces an exponentially small t@xmath0 when considered on its own , will be significantly enhanced in proximity to a metal - insulator transition .
the mechanism via which this occurs is a density of states enhancement due to a narrowing of the quasi - particle band and mott physics .
this is a different origin than the enhancement reported here where we are concerned with the overall effects of coherence and band width of the paired electrons .
some aspects of the mott physics is included in our considerations in the form of the breakdown of metallic screening along the @xmath1-axis ( a coulomb effect ) however , we have not included the breakdown of carrier properties in the cuo@xmath8 plane .
this is an intriguing possibility but it is beyond the scope of our paper and we leave it for future work .
finally , it has recently shown that high values of t@xmath0 are also obtainable from el - ph coupling without invoking a second dominant pairing interaction by departing from the conventional fermi liquid framework . in ref .
it was shown that t@xmath0 s on the order of 100 k are attainable for @xmath232 values similar to those obtained here , provided the normal state behaves as a non - fermi - liquid quantum critical metal .
although the simple approach outlined here does not hold in this scenario , much of the doping and materials - dependent variations in coupling still do .
although there appears to be a great deal of universality in the hole - doped cuprates , large changes in t@xmath0 can be achieved through changes in the chemical composition and structure of the unit cell . as we have shown here
, these changes are reflected in a number of properties including the orbital character of the band crossing the fermi level , the strength of local electric fields arising from structural - induced symmetry breaking , doping dependent changes in the underlying band structure , and ionicity of the crystal governing its ability to screen @xmath1-axis perturbations .
given the sensitivity of t@xmath0 to the structural details of the crystal it is clear that the underlying mechanism(s ) for htsc must incorporate these elements in some way . in this work
we have examined a number of aspects of coupling to oxygen modes and demonstrated that the overall coupling to these modes is quite sensitive to these factors .
therefore , the inclusion of el - ph coupling to these modes provides a natural means of linking the electronic properties of the cuo@xmath8 plane to the structural elements of the material lying off plane and the carrier concentration .
this picture calls for oxygen phonons to play a role in htsc . in extending the previous works by some of the authors on el - ph coupling in the cuprates , we have formulated a theory for poor screening in these materials . due to the quasi-2d nature of the cuprates with poor conductivity along the @xmath1-axis
, we have shown that a window of small @xmath24 opens in which screening is inoperable .
this results in an overall enhancement of the total @xmath25-wave projected coupling which enhances the phonon s ability to mediate @xmath25-wave pairing . with progressive doping , the total phonon contribution to @xmath228 is enhanced while the total contribution from @xmath226 , which suppresses @xmath25-wave pairing , is screened away . as a result , doping the system away from half - filling results in an increased contribution to @xmath25-wave superconductivity mediated by el - ph coupling . a systematic examination of the variation in el - ph coupling across the cuprate families was also performed . using experimental structural data we determined the materials variation in the total electric field strength and orbital character of the @xmath17-@xmath18 band at the fermi level and the expected variations in el - ph coupling strength to the @xmath2 branch . a direct correlation between the strength of the coupling and the optimal t@xmath0 for each material was observed . as a result , the materials variation in t@xmath0 can be naturally accounted for through the inclusion of el - ph coupling to the presently accepted intrinsic planar models . to be clear
, we do not suggest that el - ph coupling alone accounts for htsc as both the bare and screened values for the el - ph coupling strength are insufficient to produce large t@xmath0 s on their own .
instead , we propose that the phonons work in conjunction with the presently unidentified dominant mechanism for htsc .
the simplified multi - channel model for t@xmath0 presented here has demonstrated that moderate el - ph coupling , when combined with spin fluctuations , can produce enhancements to the total t@xmath0 in excess of the contribution produced by phonons alone .
furthermore , this enhancement is expected to be a generic phenomena and independent of the presently second mechanism .
therefore , the inclusion of the el - ph interaction can account for the large variations in t@xmath0 observed across the cuprate as a function of material and doping . there are some open issues regarding el - ph coupling in the cuprates .
first of all , our examination of the field strength as a function of material neglected any doping - induced changes to the field strengths .
our simplified treatment of the bi - layer system bi-2212 showed that these effects can be substantial .
therefore , this problem requires a systematic examination of the doping process across the families of cuprates , perhaps with more sophisticated approaches .
this issue is complicated in the multi - layer systems with @xmath423 where there is the possibility for inequivalent dopings for the cuo@xmath8 planes.@xcite this can further modify charge distributions altering the strength of the crystal fields in each plane.@xcite any systematic treatment of the doping induced fields should take this into account .
there is also the additional question of the role of covalency in determining the overall field strength .
finally , we have investigated the role of el - ph coupling in the cuprates , taking into account the long range coulomb interaction and its role in screening .
however , the role of the short - range coulomb interaction and its affect on the el - ph interaction remains an open and intriguing question .
we believe that the findings presented in this paper serve to further highlight the need for progress in this area .
the authors would like to acknowledge useful discussions with a. balatsky , o. gunnarsson , g. sangiovanni , i. mazin , r. hackl and d. j. scalapino .
t. p. d. would like to acknowledge support from nserc , onr grant n00015 - 05 - 1 - 0127 and the a. von humboldt foundation . the stanford work is also supported by doe contract de - ac03 - 765f00515 , nsf grant dmr-0304981 and onr grant n00014 - 01 - 1 - 0048 .
s. j. would like to acknowledge support from nserc and sharcnet .
n. n. is supported by the grant - in - aids for scientific research ( no .
17105002 , 19019004 , 19048008 , 19048015 , and 21244053 ) from the ministry of education , culture , sports , science and technology of japan , and also by funding program for world - leading innovative r&d on science and technology ( first program)
throughout this work the role of electronic correlations was neglected in deriving the anisotropic el - ph couplings arising from different charge transfer processes . in this appendix
an analysis via a different route is undertaken by examining directly the role played by correlations .
previous studies on hubbard clusters have illustrated the usefulness of exact diagonalization for exploring low energy excitations , such as the zhang - rice singlet ( zrs),@xcite as a function of hole number .
in particular , these studies have demonstrated a strong dependence of the zrs energy and mobility , as well as the magnetic exchange @xmath4 , on the planar charge transfer energy @xmath424 .
@xcite ref . pointed out that the stability of the zrs increases if the apical atom lies further away from the cuo@xmath8 plane , which reduces the mixing of the zrs with other low energy states of the cluster .
since the phonons under consideration directly modulate @xmath362 and the apical madelung energy , strong modifications should be seen in the zrs and magnetic exchange . the cell perturbation approach in ref . , which included in - plane oxygen breathing and buckling modes , showed that an effective simple hubbard - like hamiltonian can be deduced with renormalized parameters .
( color online ) cu@xmath8o@xmath30 and cu@xmath8o@xmath425 clusters used to compute the modifications of the effective zhang - rice singlet hopping @xmath3 and @xmath5 and magnetic coupling @xmath4 and @xmath426 . ] here , exact diagonalization studies of multi - band hubbard clusters are used to investigate how the low energy sector is modified by local , static atomic displacements in a pattern given by various phonon eigenvectors .
the modifications of the zrs and @xmath4 due to el - ph coupling receive close attention in order to derive renormalized parameters for single - band hubbard to @xmath3-@xmath4 model hamiltonians .
the clusters show in fig .
[ fig : cluster ] contain two sets of orbitals at the cu site : 3@xmath26 and 4@xmath16 . by symmetry , the hopping amplitude from the apical oxygen site to the cu 3@xmath26 , @xmath427 vanishes , while @xmath428 is non - zero .
thus , a general hubbard model can be written as @xmath429 where the kinetic piece is @xmath430 \\\nonumber & + & t_{pp}\sum_{i,\delta,\delta^\prime,\sigma } q^\prime_{\delta,\delta^\prime } p^\dagger_{i,\delta,\sigma}p_{i,\delta^\prime,\sigma } \\\nonumber & + & t_{ap}\sum_{i,\delta,\sigma } q^{\prime\prime}_{\delta}[a^\dagger_{i,\sigma}p_{i,\delta,\sigma } + h.c . ]
\\\nonumber & - & t_{as}\sum_{i,\sigma}[a^\dagger_{i,\sigma}s_{i,\sigma } + h.c .
] \\ & - & t_{ps}\sum_{i,\sigma}q^{\prime\prime}_{\delta}[s^\dagger_{i,\sigma}p_{i,\delta,\sigma } + h.c.]\end{aligned}\ ] ] with phase factors @xmath431 , @xmath432 , @xmath433 , and the potential piece
is given by @xmath434 here , @xmath244 , @xmath435 , @xmath25 , @xmath16 ( @xmath436,@xmath437 , @xmath438 , @xmath439 ) annihilate ( create ) a hole on apical oxygen , in - plane oxygen ( @xmath81 ) , copper @xmath25- and copper @xmath16-orbital , respectively , @xmath440 is the displacement of a given ion and @xmath441 is the associated local electronic field .
this type of coupling is different than that considered in previous studies , where the buckling modes were represented as an electrostatic modulation of the cu site energies.@xcite since the 4@xmath16 orbitals are rather extended we neglect the on - site coulomb repulsion as well as inter - orbital interactions with the 3@xmath26 orbitals .
there are a total of 13 and 14 orbitals for the cu@xmath8o@xmath30 and cu@xmath8o@xmath425 clusters , respectively . in both cases ,
the clusters contain seven holes in order to investigate the formation and delocalization of a zrs to a nearest neighbour copper - oxide plaquette ( 4 holes being almost exclusively located in the cu 4@xmath16 orbitals ) .
exact diagonalizations in the @xmath442 sector , which contains the ground state , are performed since the hamiltonian conserves @xmath443 .
parameters similar to those of ref . for the hubbard model are used , with values representative of the electric field for the planar and apical oxygen sites in bilayer cuprates ( in ev and ev / for the fields ) : @xmath444 the diagonalization of the cu@xmath8o@xmath30 cluster yields the effective hopping parameter @xmath3 of the zrs as well as the magnetic exchange @xmath4 : the splitting between the groundstate and first excited state gives @xmath445 , whereas the singlet - triplet splitting of the 6-particle problem gives @xmath4 .
the parameters @xmath5 and @xmath426 are analogously defined from the cu@xmath8o@xmath425 cluster . in the absence of el - ph coupling
we obtain the following values : @xmath446 ev , @xmath447 ev , @xmath448 ev , and @xmath426 = 15 mev .
( color online ) the effect of different modes on various zhang - rice parameters for the cu@xmath8o@xmath30 cluster : plots ( a ) and ( b ) show the effect of a given mode on the ground - state energy and on the effective zrs hopping @xmath3 with @xmath449 and @xmath450 .
( a ) the effect of the @xmath44 modes .
( b ) the effect of the @xmath2 mode .
( c ) the effect of the apical mode on @xmath445 . ] fig .
[ fig : edresults ] displays the effect of different phonon modes on the ground state energy , @xmath3 and @xmath4 as a function of static displacement @xmath145 .
although only static atomic displacements have been investigated in configurations corresponding to the phonon eigenvectors , the differences between the relevant time scales associated with the electron hopping and atomic motion justify such an analysis for the purpose of viewing a snap - shot of the modifications of the zrs parameters .
( we also note that ref .
considered a similar calculation with fully quantized atomic motion for the @xmath2 and @xmath44 phonons at @xmath45 and reported similar results for @xmath4 . )
one clearly sees , for small displacements , the main effect on @xmath3 is due to the @xmath44 and @xmath2 modes . for the @xmath2 mode ,
the effective hopping of the zrs is modulated to first order in the displacement , while the energy is only modified at second order , in agreement with perturbative analysis . for the @xmath44 mode
the first - order correction to the energy does not vanish and both @xmath451 and @xmath452 .
the magnitude of the modulation in @xmath3 is comparable to the @xmath2 mode with the same local field .
the effect of the apical mode on @xmath3 is very weak for @xmath453 .
this is not surprising since the zrs has little apical character if the orbital lies away from the cuo@xmath8 plane . for @xmath454 the hybridization of the apical-@xmath91 and cu-@xmath455 orbitals destabilizes the zrs and thus the effective hopping @xmath3 drops as in an earlier work.@xcite also , both the ground state energy @xmath342 and the effective exchange @xmath4 are not affected much by this mode .
( color online ) variation of the effective zhang - rice hopping @xmath5 and magnetic exchange @xmath426 to next - nearest neighbor .
the calculation performed for the cu@xmath8o@xmath425 cluster for @xmath44- , @xmath2- and apex - modes . ]
it is clear , based on the calculations performed , that the three modes modify the effective hopping of the zrs .
however , the mechanism by which the apical mode affects the hopping is very different since the formation of the zrs no longer takes place in the cuo@xmath8 plane if the displacement of the apical oxygen is towards the plane .
the effective next - nearest neighbour quantities @xmath5 and @xmath426 ( see fig .
[ fig : edresults2 ] ) are affected less by the two planar modes . summarized in figs .
[ fig : edresults ] and [ fig : edresults2 ] , our results indicate that for both the @xmath44 and @xmath2 phonons , @xmath456 as well as @xmath457 corrections have a linear dependence on displacement of about the same magnitude . for the apical phonon
the magnitude is different due to the destabilization of the zrs .
the on - site energy is modified mainly by the @xmath44 ep - ph coupling ; which can be quantified by looking at the corresponding modification of the ground - state energy .
modifications to parameters @xmath458 and @xmath459 indicate again that the dominant contribution comes from the @xmath44 mode . however , this conclusion is based on a naive cluster approach and should be corroborated by other means of investigation . in terms of treatments of effective single - band models , such as hubbard or @xmath3-@xmath4 , incorporating phonons
, we note that our results indicate that the effective couplings for these modes are neither purely holstein - like nor can they be treated as bond phonons simply modifying the effective zrs hopping @xmath3 : the effective parameters , such as @xmath3 and @xmath4 , and as a consequence , the mott - hubbard effective interaction @xmath7 in a single - band approach , are all modified .
the actual modification of the effective @xmath7 for static displacements corresponding to these phonon modes requires an investigation of different three - band clusters containing cu @xmath460 , @xmath461 and @xmath462 and oxygen ligand configurations .
this points out the complications involved in ascribing real phonon modes to any corresponding model phonons in a single - band approach , and illustrates the important interplay between el - ph coupling and electronic correlations .
99 see , e.g. , d. j. scalapino , _ handbook of high - temperature superconductivity _ , ed
. j. r. schrieffer , springer ( 2006 ) ; arxiv : cond - mat/0610710 . m. gurvitch and a. t. fiory , phys .
* 59 * , 1337 ( 1987 ) ; s. martin , a. t. fiory , r. m. fleming , l. f. schneemeyer and j. v. waszczak , phys . rev .
b * 41 * , 846 ( 1990 ) .
s. komiya , y. ando , x. f. sun and a. n. lavrov , phys .
b * 65 * , 214535 ( 2002 ) .
e. pavarini , i. dasgupta , t. saha - dasgupta , o. jepsen and o. k. andersen , phys .
lett . * 87 * , 047003 ( 2001 ) .
l. spanu , m. lugas , f. becca and s. sorella , phys .
b * 77 * , 024510 ( 2008 ) .
e. khatami , a. macridin , and m. jarrell , phys .
b * 78 * , 060502(r ) ( 2008 ) . c. t. shih , t. k. lee , r. eder , c .- y . mou and y. c. chen , phys .
lett . * 92 * , 227002 ( 2004 ) .
p. w. anderson , science * 268 * , 1154 ( 1995 ) ; s. chakravarty , h .- y .
kee and e. abrahams , phys .
lett . * 82 * , 2366 ( 1999 ) .
d. van der marel and j. h. kim , j. phys .
56 * , 1825 ( 1995 ) .
a. a. tsvetkov , d. van der marel , k. a. moler , j. r. kirtley , j. l. de boer , a. meetsma , z. f. ren , n. koleshnikov , d. dulic , a. damascelli , m. grininger , j. schtzmann , j. w. van der eb , h. s. somal and j. h. wang , nature * 395 * , 360 ( 1998 ) .
k. mcelroy , j. lee , j. a. slezak , d .- h .
lee , h. eisaki , s. uchida and j. c. davis , science * 309 * , 1048 ( 2005 ) . t. s. nunner , b. m. andersen , a. melikyan and p. j. hirschfeld , phys .
lett . * 95 * , 177003 ( 2005 ) ; t. s. nunner , w. chen , b. m. andersen , a. melikyan and p. j. hirschfeld , phys .
b * 73 * , 104511 ( 2006 ) ; j .- x .
zhu , k. mcelroy , j. lee , t. p. devereaux , q. si , j. c. davis , a. v. balatsky , phys .
* 97 * , 177001 ( 2006 ) ; r. jamei , j. robertson , e - a .
kim , a. fang , a. kapitulnik and s. a. kivelson , phys .
b * 74 * , 174521 ( 2006 ) .
k. fujita , t. noda , k. m. kojima , h. eisaki and s. uchida , phys .
95 * , 097006 ( 2005 ) ; h. eisaki , n. kaneko , d. l. feng , a. damascelli , p. k. mang , k. m. shen , z .- x .
shen and m. greven , phys .
b * 69 * , 064512 ( 2004 ) .
w. b. gao , q. q. liu , l. x. yang , y. yu , f. y li , c. q. jin , s. uchida , phys .
b * 80 * , 094523 ( 2009 ) .
y. ohta , t. tohyama and s. maekawa , phys .
b * 43 * , 2968 ( 1991 ) ; c. di castro , l. f. feiner and m. grilli , phys . rev . lett . *
66 * , 3209 ( 1991 ) ; j. h. jefferson , h. eskes , and l. f. feiner , phys . rev .
b * 45 * , 7959 ( 1992 ) ; l. f. feiner , m. grilli and c. di castro , phys .
b * 45 * , 10647 ( 1992 ) ; r. raimondi , j. h. jefferson and l. f. feiner , phys .
b * 53 * , 8774 ( 1996 ) .
a. iyo , y. tanaka , m. tokumoto and h. ihara , physica c * 366 * , 43 ( 2001 ) ; a. iyo , m. hirai , k. tokiwa , t. watanabe and y. tanaka , supercond .
* 17 * , 143 ( 2004 ) . y. chen , a. iyo , w. yang , x. zhou , d. lu , h. eisaki , t. p. devereaux , z. hussain and z .- x .
shen , phys .
97 * , 236401 ( 2006 ) ; w. xie , o. jepsen , o. k. andersen , y. chen and z .- x .
shen , phys .
* 98 * , 047001 ( 2007 ) .
a. damascelli , z. hussain and z .- x .
shen , rev .
phys . * 75 * , 473 ( 2003 ) .
see , e.g. , a. s. alexandrov and n. f. mott , _ high temperature superconductors and other superfluids _ , taylor and francis , 1994 . j. g. bednorz and k. a. m " uller , z. phys .
b * 64 * , 189 ( 1986 )
. b. batlogg , r. j. cava , a. jayaraman , r. b. van dover , g. a. kourouklis , s. sunshine , d. w. murphy , l. w. rupp , h. s. chen , a. white , k. t. short , a. m. mujsce and e. a. rietman , phys .
* 58 * , 2333 ( 1987 ) ; l. c. bourne , m. f. crommie , a. zettl , h. c. zur loye , s. w. keller , k. l. leary , a. m. stacy , k. j. chang and m. l. cohen , phys .
lett . * 58 * , 2337 ( 1987 ) ; p. j. yvon , r. b. schwarz , c. b. pierce , l. bernardez , a. conners , and r. meisenheimer , phys .
b * 39 * , 6690 , ( 1989 ) .
l. perfetti , p. a. loukakos , m. lisowski , u. bovensiepen , h. eisaki and m. wolf , phys .
lett . * 99 * , 197001 ( 2007 ) .
s. y. savrasov and o. k. andersen , phys .
lett . * 77 * , 4430 ( 1996 ) ; s. y. savrasov , d. y. savrasov and o. k. andersen , phys . rev
. lett . * 72 * , 372 ( 1994 ) .
o. k. andersen , s. y. savrasov , o. jepsen and a. i. liechtenstein , journ .
* 105 * , 285 ( 1996 ) ; o. jepsen , o. k. andersen , i. dasgupta and s. savrasov , j. phys . chem .
solids * 59 * , 1718 ( 1998 ) . t. sakai , d. poilblanc and d. j. scalapino , phys .
b * 55 * , 8445 ( 1997 ) .
m. capone , g. sangiovanni , c. castellani , c. di castro and m. grilli , phys .
lett . * 92 * , 106401 ( 2004 ) . b. j. alder , k. j. runge and r. t. scalettar , phys . rev
. lett . * 79 * , 3022 ( 1997 ) .
g. sangiovanni , m. capone , c. castellani and m. grilli , phys .
lett . * 94 * , 026401 ( 2005 ) .
g. sangiovanni , o. gunnarsson , e. koch , c. castellani and m. capone , phys .
lett . * 97 * , 046404 ( 2006 ) .
p. werner and a. j. millis , phys .
* 99 * , 146404 ( 2007 ) .
o. r " osch and o. gunnarsson , phys .
lett . * 92 * , 146403 ( 2004 ) ; a. s. mishchenko and n. nagaosa , phys .
lett . * 93 * , 036402 ( 2004 ) ; o. gunnarsson and o. rsch , phys .
b * 73 * , 174521 ( 2006 ) ; p.
prelovek , r. zeyher and p. horsch , phys .
lett . * 96 * , 086402 ( 2006 ) . _
_ j. phys . : condens .
matter * 20 * , 043201 ( 2008 ) ; a. s. mishchenko , adv .
2010 * , 306106 ( 2010 ) ; o. rsch and o. gunnarsson , phys .
b * 70 * , 224518 ( 2004 ) .
j. bona , s. maekawa , t. tohyama and p. prelovek , phys .
b * 77 * , 054519 ( 2008 ) .
z. b. huang , w. hanke , e. arrigoni and d. j. scalapino , phys .
b * 68 * , 220507(r ) ( 2003 ) .
m. l. kuli and r. zeyher , phys .
b * 49 * , 4395 ( 1994 ) ; r. zeyher and m. l. kuli , _ ibid . _ * 53 * , 2850 ( 1996 ) ; s. ishihara and n. nagaosa , phys . rev .
b * 69 * , 144520 ( 2004 ) . e. koch and r. zeyher , phys .
b * 70 * , 094510 ( 2004 ) .
r. citro , s. cojocaru and m. marinaro , phys .
b * 72 * , 115108 ( 2005 ) .
a. macridin , b. moritz , m. jarrell and t. maier , phys .
lett . * 97 * , 056402 ( 2006 ) ; a. macridin , b. moritz , m. jarrell and t. maier , arxiv : cont - mat/0611067 ( 2006 ) .
v. i. anisimov , m. a. korotin , j. zaanen and o. k. andersen , * 68 * , 345 ( 1992 ) . j. zaanen and p. b. littlewood , phys .
b * 50 * , 7222 ( 1994 ) . for a review , see e.g. , t. cuk , d. h. lu , z .- j .
zhou , z .- x .
shen , t. p. devereaux and n. nagaosa , phys .
( b ) * 242 * , 11 ( 2005 ) .
a. lanzara , p. v. bogdanov , x. j. zhou , s. a. kellar , d. l. feng , e. d. lu , t. yoshida , h. eisaki , a. fujimori , k. kishio , j .-
shimoyama , t. noda , s. uchida , z. hussain , and z .- x .
shen , nature ( london ) * 442 * , 510 ( 2001 ) .
t. cuk , f. baumberger , d. h. lu , n. ingle , x. j. zhou , h. eisaki , n. kaneko , z. hussain , t. p. devereaux , n. nagaosa and z .- x .
shen , phys .
* 93 * , 117003 ( 2004 ) .
j. lee , k. fujita , k. mcelroy , j. a. slezak , m. wang , t. aiura , h. bando , m. ishikado , t. masui , j .- x .
zhu , a. v. balatsky , h. eisaki , s. uchida and j. c. davis , nature ( london ) * 442 * , 546 ( 2006 ) . w. s. lee , w. meevasana , s. johnston , d. h. lu , i. m. vishik , r. g. moore , h. eisaki , n. kaneko , t. p. devereaux and z .- x .
shen , phys .
b * 77 * , 140504(r ) ( 2008 ) .
x. j. zhou , j. shi , t. yoshida , t. cuk , w. l. yang , v. brouet , j. nakamura , n. mannella , s. komiya , y. ando , f. zhou , w. x. ti , j. w. xiong , z. x. zhao , t. sasagawa , t. kakeshita , h. eisaki , s. uchida , a. fujimori , z. zhang , e. w. plummer , r. b. laughlin , z. hussain and z .- x .
shen , phys .
* 95 * , 117001 ( 2005 ) .
w. meevasana , n. j. c. ingle , d. h. lu , j. r. shi , f. baumberger , k. m. shen , w. s. lee , t. cuk , h. eisaki , t. p. devereaux , n. nagaosa , j. zaanen and z .- x .
shen , phys .
* 96 * , 157003 ( 2006 ) .
y. chen , a. iyo , w. yang , a. ino , m. arita , s. johnston , h. eisaki , h. namatame , m. taniguchi , t. p. devereaux , z. hussain , and z .- x .
shen , phys .
036403 ( 2009 ) . w. s. lee , k. tanaka , i. m. vishik , d. h. lu , r. g. moore , h. eisaki , a. iyo , t. p. devereaux , and z. x. shen , phys .
lett . * 103 * , 067003 ( 2009 ) . k. byczuk , m. kollar , k. held , y .- f .
yang , i. a. nekrasov , th .
pruschke and d. vollhardt , nature physics * 3 * , 168 ( 2007 ) .
m. eschrig , adv .
physics * 55 * , 47 ( 2006 ) , and references therein .
s. engelsberg and j. r. schrieffer , phys .
rev . * 131 * , 993 ( 1963 ) .
a. kaminski , m. randeria , j. c. campuzano , m. r. norman , h. fretwell , j. mesot , t. sato , t. takahashi , and k. kadowaki , phys .
lett . * 86 * , 1070 ( 2001 ) ; t.
k. kim , a. a. kordyuk , s. v. borisenko , a. koitzsch , m. knupfer , h. berger and j. fink , phys .
lett . * 91 * , 167002 ( 2003 ) ; m. r. norman , h. ding , j. c. campuzano , t. takeuchi , m. randeria , t. yokoya , t. takahashi , t. mochiku and k. kadowaki , phys .
lett . * 79 * , 3506 ( 1997 ) ; s. v. borisenko , a. a. kordyuk , v. zabolotnyy , j. geck , d. inosov , a. koitzsch , j. fink , m. knupfer , b. bu chner , v. hinkov , c. t. lin , b. keimer , t. wolf , s. g. chiuzbian , l. patthey and r. follath , phys .
lett . * 96 * , 117004 ( 2006 ) .
s. johnston , w. s. lee , y. chen , e. a. nowadnick , b. moritz , z .- x .
shen and t. p. devereaux , advances in condensed matter physics .
* 2010 * , 968304 ( 2010 ) .
w. s. lee , s. johnston , t. p. devereaux and z .- x .
shen , phys .
b * 75 * , 195116 ( 2007 ) .
l. pintschovius , phys .
( b ) * 242 * , 30 ( 2005 ) .
m. c. krantz , c. thomsen , hj .
mattausch and m. cardona , phys .
b * 50 * , 1165 ( 1994 ) ; i .- s .
yang , h .-
lee , n. h. hur and j. yu , phys .
b * 52 * , 15078 ( 1995 ) ; x. zhou , m. cardona , d. colson and v. viallet , phys . rev .
b * 55 * 12770 ( 1997 ) ; v. g. hadjiev , x. zhou , t. strohm , m. cardona , q. m. lin and c. w. chu , phys .
b * 58 * , 1043 ( 1998 ) ; m. limonov , s. lee , s. tajima and a. yamanaka , phys . rev .
b * 66 * , 054509 ( 2002 ) .
d. munzar and m. cardona , phys .
lett . * 90 * , 077001 ( 2003 ) .
w. l. mcmillan , phys .
b * 167 * , 331 ( 1968 ) .
t. p. devereaux , t. cuk , z .- x .
shen and n. nagaosa , phys .
* 93 * , 117004 ( 2004 ) t. p. devereaux , a. virosztek and a. zawadowski , phys .
b * 51 * , 505 ( 1995 ) ; * 59 * , 14618 ( 1999 ) ; s. barii and i batisti , europhys . lett . *
8 * , 765 ( 1989 ) .
j. li and j. ladik , sol .
state comm .
* 95 * , 35 ( 1995 ) .
r. heid , k .-
bohnen , r. zeyher and d. manske , phys .
lett . * 100 * , 137001 ( 2008 ) .
bohnen , r. heid and m. krauss , europhys . lett . * 64 * , 104 ( 2003 ) .
f. giustino , m. l. cohen and s. g. louie , nature * 452 * , 975 ( 2008 ) . c. n. veenstra , g. l. goodvin , m. berciu , and a. damascelli , arxiv:1003.0141v1 .
d. reznik , g. sangiovanni , o. gunnarsson and t. p. devereaux , nature * 455 * , e6 ( 2008 ) . b. s. shastry , phys . rev . lett . * 63 * , 1288 ( 1989 ) .
j. mustre de leon , s. d. conradson , i. batistic , and a. r. bishop , phys .
lett . * 65 * , 1675 ( 1990 ) .
a. i. liechtenstein , i. i. mazin , o. k. anderson , o. jepsen , phil . mag .
b * 70 * , 643 ( 1994 ) .
j. zhong and h .- b .
sch " uttler , phys .
69 * , 1600 ( 1992 ) ; s. koval and a. greco , sol
. state commun .
* 90 * , 17 ( 1994 ) .
o. rsch , o. gunnarsson , x. j. zhou , t. yoshida , t. sasagawa , a. fujimori , z. hussain , z .- x .
shen , and s. uchida , phys .
lett . * 95 * , 227002 ( 2005 ) .
o. k. andersen , a. i. leichtenstein , o. jepsen and f. paulson , j. phys .
solids * 56 * , 153 ( 1995 ) .
l " owdin , j. chem . phys . * 19 * , 1396 ( 1951 ) .
we note that in ref .
all energies were referenced to the chemical potential in the eigenfunctions @xmath162 .
this overestimated the fermionic momentum dependence for coupling to the breathing modes , as pointed out in ref .
this also led to an overestimation of the magnitude of the coupling to the @xmath2 modes whena value of the electric field @xmath463 ev / was used .
we show in section [ madelung ] that this field value is likely an underestimation , and thus comparable @xmath199 values are obtained here and in ref . .
r. s. markiewicz , s. sahrakorpi , m. lindroos , h. lin and a. bansil , phys .
b * 72 * , 054519 ( 2005 ) .
m. r. norman , m. randeria , h. ding and j. c. campuzano , phys .
b * 52 * , 615 ( 1995 )
. g. d. mahan , _ many - particle physics _
, second edition , plenum press , new york ( 1990 ) . to estimate the magnitude of the apical coupling we have determined , for bi-2212 , the madelung energy difference @xmath464 ev , the dielectric constant @xmath368 , giving @xmath465 ev .
mean field corrections to the site energies then shift these values .
this is derived in section vi .
a. w. sandvik , d. j. scalapino , and n. e. bickers , phys .
b * 69 * , 094523 ( 2004 ) .
a. a. abrikosov and v. m. genkin , zh .
* 65 * , 842 ( 1973 ) [ sov .
jetp , * 38 * , 417 ( 1974 ) ] .
c. falter , m. klenner and q. chen , phys .
b * 48 * , 16690 ( 1993 ) ; c. falter , m. klenner , g. a. hoffmann and q. chen , phys . rev .
b * 55 * , 3308 ( 1997 ) ; t. bauer and c. falter , phys . rev .
b * 80 * , 094525 ( 2009 ) c. falter , m. klenner and g. a. hoffmann , phys . rev .
b * 57 * , 14444 ( 1998 ) . s. i. mukhin and a. mesaros and j. zaanen and f. v. kusmartsev , phys . rev .
b * 76 * , 174521 ( 2007 ) .
a. l. fetter , ann .
( n.y . ) * 88 * , 1 ( 1974 ) .
d. n. aristov and g. khaliullin , phys .
b * 74 * , 045124 ( 2006 ) .
m. a. quijada , d. b. tanner , r. j. kelley , m. onellion , h. berger and g. margaritondo , phys .
b * 60 * , 14917 ( 1999 ) .
n. n. kovaleva , a. v. boris , t. holden , c. ulrich , b. liang , c. t. lin , b. keimer , c. bernhard , j. l. tallon , d. munzar and a. m. stoneham , phys .
b * 69 * , 0.54511 ( 2004 ) .
b. moritz , f. schmitt , w. meevasana , s. johnston , e. m. motoyama , m. graven , d. h. lu , c. kim , r. t. scalettar , z .- x .
shen and t. p. devereaux , new journ .
phys . * 11 * , 093020 ( 2009 ) .
s. johnston , f. vernay and t. p. devereaux , europhys .
lett . , * 86 * , 37007 ( 2009 ) ; k. foyevtsova , r. valenti and p. j. hirschfeld * 79 * , 144424 ( 2009 ) . j. d. jorgensen , m. a. beno , d. g. hinks , l. soderholm , k. j. volin , r. l. hitterman , j. d. grace , i. k. schuller , c. u. segre , k. zhang , and m. s. kleefisch , phys . rev .
b * 36 * , 3608 ( 1987 ) ; e. kaldis , j. rhler , e. liarokapis , n. poulakis , k. conder and p. w. loeffen , phys . rev
. lett . * 79 * , 4894 ( 1997 ) ; o. chmaissem , j. d. jorgensen , s. short , a. knizhnik , y. eckstein and h. shaked , nature ( london ) bf 397 , 45 ( 1999 ) .
p. ewald , ann .
* 369 * , 253 ( 1921 ) .
j. li and j. ladik , sol .
comm . * 95 * , 35 ( 1995 ) ; o.i .
lyakhovitskaya , yu.z .
nozik , kristallografiya * 32 * , 498 ( 1990 ) .
m. francois , k. yvon , p. fischer , and m. decroux , solid state commun . * 63 * , 35 ( 1987 ) .
j. d. jorgensen , h. -b .
sch " uttler , d. g. hinks , d. w. capone ii , k. zhang , m. b. brodsky , and d. j. scalapino , phys .
lett . * 58 * , 1024 ( 1987 ) . c. c. torardi , m. a. subramanian , j. c. calabrese , j. gopalakrishnan , e. m. mccarron , k. j. morrissey , t. r. askew , r. b. flippen , u. chowdhry , and a. w. sleight , phys rev b * 38 * , 225 ( 1988 ) . n. n. kovaleva , a. v. boris , t. holden , c. ulrich , b. liang , c. t. lin , b. keimer , c. bernhard , j. l. tallon , d. munzar , and a. m. stoneham , phys rev b * 69 * , 054511 ( 2004 ) . t. watanabe , t. fujii and a. matsuda , jour .
. jpn . * 72 * , 2924 ( 2003 ) .
q. huang , j. w. lynn , q. xiong and c. w. chu , phys .
b * 52 * , 462 ( 1995 ) . p.g .
radaelle , j.l .
wagner , b.a .
hunter , m.a .
beno , g.s .
knapp , j.d .
jorgensen and d.g .
hinks , physica c * 216 * , 29 ( 1993 ) .
a. r. armstrong , w. i. f. david , i. gameson , p. p. edwards , j. j. capponi , p. bordet , and m. marezio , phys .
b * 52 * , 15551 ( 1995 ) .
m. paranthaman , b.c .
chakoumakos , journ .
. state chem .
* 122 * , 221 ( 1996 ) .
q. huang , o. chmaissen , j. j. capponi , c. chaillout , m. marezio , j. l. tholence , and a. santoro , physica c * 227 * , 1 ( 1994 ) . y. shimakawa , y. kubo , t. manako , y. nakabayashi , and h. igarashi , physica c * 156 * , 97 ( 1988 ) .
d. e. cox , c. c. torardi , m. a. subramanian , j. gopalakrishnan , and a. w. sleight , phys .
b * 38 * , 6624 ( 1988 ) . y
tang , b .- x .
lin , d .- y .
zhou , w .- j .
zhu , f .- x .
chew , x .-
zheng , y .- f .
zheng , n. li , k .- x .
chen and g. lu , science in china series a , * 22 * , 409 ( 1990 ) .
t. p. devereaux , a. virosztek , a. zawadowski , m. opel , p. f. mller , c. hoffmann , r. philipp , r. nemetschek , r. hackl , h. berger , l. forr , a. erb and e. walker , solid state comm . * 108 * , 407 ( 1998 ) ; m. opel , r. hackl , t. p. devereaux , a. virosztek , a. zawadowski , a. erb , e. walker , h. berger and l. forr ' o , phys .
b * 60 * , 9836 ( 1999 ) .
w. a. harrison , _ elementary electronics structure _ , world scientific ( 2004 ) .
y. he , t. s. nunner , p. j. hirschfeld and h .-
cheng , phys .
lett . * 96 * , 197002 ( 2006 ) .
t. s. nunner , j. schmalian and k. h. bennemann , phys .
b * 59 * , 8859 ( 1999 ) .
yunkyu bang , phys .
b * 78 * , 075116 ( 2008 ) .
s. johnston and t. p. devereaux , phys .
b * 81 * , 214512 ( 2010 ) .
t. das , r. s. markiewicz and a. bansil , phys .
lett . * 98 * , 197004 ( 2007 ) .
a. comanac , l. de medici , m. capone and a. j. millis , nature physics * 4 * , 287 ( 2008 ) .
t. a. maier , d. poilblanc and d. j. scalapino , phys .
lett . * 100 * , 237001 ( 2008 ) .
j. p. carbotte , rev .
phys . * 62 * , 1027 ( 1990 ) .
s. v. vonsovsky , yu . a. izyumov and e. z. kurmaev , _ superconductivity of transition metals , their alloys and compounds _ , ( springer - verlag , berlin , 1982 ) .
p. j. williams and j. p. carbotte , phys .
b * 43 * , 7960 ( 1991 ) ; e. j. nicol and j. p. carbotte , phys . rev .
b * 44 * , 12511 ( 1991 ) .
o. v. dolgov , i. i. mazin , a. a. golubov , s. y. savrasov and e. g. maksimov , phys .
lett . * 95 * , 257003 ( 2005 ) .
m. capone , m. fabrizio , c. castellani and e. tosatti , science * 296 * , 2364 ( 2002 )
. m. capone , m. fabrizio , c. castellani and e. tosatti , rev .
phys . * 81 * , 943 ( 2009 ) .
she and j. zaanen , phys .
b * 80 * , 184518 ( 2009 ) . f. c. zhang and t. m. rice , phys .
b * 37 * , 3759 ( 1988 ) .
h. eskes and j. h. jefferson , phys .
b * 48 * , 9788 ( 1993 ) .
p. piekarz , j. konior and j. h. jefferson , phys .
b * 59 * , 14697 ( 1999 ) . | the large variations of t@xmath0 across the cuprate families is one of the major unsolved puzzles in condensed matter physics , and is poorly understood . although there appears to be a great deal of universality in the cuprates , several orders of magnitude changes in t@xmath0 can be achieved through changes in the chemical composition and structure of the unit cell . in this paper
we formulate a systematic examination of the variations in electron - phonon coupling to oxygen phonons in the cuprates , incorporating a number of effects arising from several aspects of chemical composition and doping across cuprate families .
it is argued that the electron - phonon coupling is a very sensitive probe of the material - dependent variations of chemical structure , affecting the orbital character of the band crossing the fermi level , the strength of local electric fields arising from structural - induced symmetry breaking , doping dependent changes in the underlying band structure , and ionicity of the crystal governing the ability of the material to screen @xmath1-axis perturbations . using electrostatic ewald calculations and known experimental structural data , we establish a connection between the material s maximal t@xmath0 at optimal doping and the strength of coupling to @xmath1-axis modes .
we demonstrate that materials with the largest coupling to the out - of - phase bond - buckling ( @xmath2 " ) oxygen phonon branch also have the largest t@xmath0 s . in light of this observation we present model t@xmath0 calculations using a two - well model where phonons work in conjunction with a dominant pairing interaction , presumably due to spin fluctuations , indicating how phonons can generate sizeable enhancements to t@xmath0 despite the relatively small coupling strengths .
combined , these results can provide a natural framework for understanding the doping and material dependence of t@xmath0 across the cuprates . |
in the following , we describe the capabilities of ultraviolet coronagraphic spectroscopy to address the sep / flare / cme and solar wind problems , and we provide a brief description of a science payload that is capable of carrying out the required observations .
uvcs / soho observations have provided new insights into the roles of shock waves , reconnection , and magnetic helicity in cme eruptions ( raymond 2002 ) .
the key parameters in theories of particle acceleration by shocks are the pre - shock plasma conditions ( including seed particle population ) , the shock speed , and the angle between the magnetic field and the shock motion .
uvcs has observed cme - driven shocks through their effect on the widths of uv spectral lines ( raymond et al .
2000 ; mancuso et al .
these observations provide information about the compression ratio in the shock , a crucial parameter for predicting sep spectra , and information about the thermal equilibration among electrons , protons and heavier ions .
electron heating is relatively modest , and the line widths of oxygen and silicon ions imply temperatures far higher than the proton temperatures , a potentially important consideration for models of sep composition . in some events , the shock compression ratio can be determined from type ii burst band splitting ( e.g. , vrnak et al .
2002 ) , but not all radio bursts contain enough detail for this diagnostic to be useful . for a much larger fraction of events , uv spectroscopy can be used to determine the compression ratio via two independent techniques : ( 1 ) measuring pre- to post - shock temperature ratios ( i.e. , both @xmath3 and @xmath4 ) with resonant and thomson - scattered h i ly@xmath5 line widths , then using adiabatic theory to compute the density ratio ; ( 2 ) measuring ion temperatures of several species having different charges and masses and applying collisionless theories of multi - ion shock heating ( e.g. , lee et al .
1987 ; lee & wu 2000 ) to compute the compression ratio that is most consistent with the observations ( for both methods , see mancuso et al .
2002 ) .
uvcs routinely obtains the densities , ionization states and elemental abundances in the pre - cme corona ( e.g. , raymond et al .
the densities obtained by uvcs can be combined with type ii radio burst drift rates to obtain shock speeds .
upper limits on the coronal alfvn speed above active regions were inferred from the derived shock speeds by requiring that the disturbances propagate at least as fast as the local characteristic speed ( mancuso et al .
2003 ) . for a subset of events , the resulting shock speeds are in much better agreement with lasco cme expansion rates than shock speeds based upon average coronal density profiles , although there are some uncertainties in the measured drift rates ( mancuso & raymond 2004 ) .
the alfvn and shock speeds can be inferred from detection of the shock arrival at different heights as determined by the timing of the increase in line widths of uv emission lines ( e.g. , ciaravella et al .
the angle between the shock front and the magnetic field requires the pre - shock field direction , which can be determined from streamer morphology .
severe elemental depletions are often observed in the closed field portions of streamers ( uzzo et al .
2004 ) , providing an additional indicator of field topology .
another parameter potentially vital to the efficiency of shock acceleration is the density of suprathermal seed particles ( e.g. , desai et al .
while uvcs was not able to detect such particles , the improved sensitivity and instrumental profile characterization of next - generation instruments will make it possible to determine suprathermal particle densities out to @xmath66 times the mean proton thermal speed ( cranmer 1998 ) .
although uvcs has proven the feasibility of detecting and characterizing cme shocks for several representative events , next - generation instrumentation can provide more extensive diagnostics and more complete spatial and temporal coverage .
the empirical characterization of the coronal shock conditions and the ambient solar wind properties can then be used as inputs to : ( 1 ) 3d mhd models of the shock propagation through the heliosphere , and ( 2 ) multi - scale models of the sep acceleration , transport , and energy spectrum synthesis ( zank et al .
2000 ; li et al . 2003 ; rice et al .
such constraints need to be applied in order to model specific events , such as the massive solar storms of oct
sep acceleration and transport models can be tested for specific events by using spectroscopy and other remote - sensing diagnostics to constrain the initial parameters of the shock in the corona , then comparing with observed sep energy spectra from , e.g. , the inner heliospheric sentinels .
iterative testing and refinement will ultimately result in a comprehensive validation of a predictive sep model .
models of cmes rely heavily on reconnection in current sheets , either trailing beneath the ejected magnetic flux rope or creating the flux rope in the first place ( see klimchuk 2001 ) .
reconnection dumps large amounts of energy in the lower atmosphere of the sun , creating intense heating , which accounts for the traditional flare ribbons and loops and for the current sheet containing hot plasma ( forbes et al .
1989 ; vestka & cliver 1992 ; forbes & acton 1996 ; priest & forbes 2002 ) .
uvcs observations made it possible for the first time to allow us to carry out diagnostics of the plasma inside the current sheet ( ciaravella et al .
2000 ; ko et al . 2003 ) .
a narrow feature was seen in [ fe xviii ] emission in the space between the post - flare loops and the cme core , indicating electron temperatures near @xmath7 k. significant progress in studying the current sheet and the process of magnetic reconnection in the current sheet was made recently when the uv spectral data of the plasma inside the current sheet during two events on january 8 , 2002 and on november 18 , 2003 were obtained and analyzed ( ko et al .
2003 ; lin et al .
both events developed a fast cme , a growing flare loop system , and a long current sheet that connects cme and flare loops .
in one of these events , the pattern of the reconnection inflow near the current sheet was well recorded in h i ly@xmath5
. this allowed us to deduce the speed of the reconnection inflow directly and even to estimate the thickness of the current sheet . also , it is possible to determine the magnetic field strength by using the observed speeds , densities , and temperatures to compute the kinetic and thermal energy densities in the reconnection region , then assume that this energy comes from the annihilation of an equal amount of magnetic energy just outside the current sheet ( ko et al .
all of the parameters described above are required in order to put empirical constraints on the _ reconnection rate _ and electric field strength in the current sheet .
such progress has very important theoretical consequences : for example , we are able to deduce the electrical resistivity ( conductivity ) in the current sheet , or in the reconnection region .
results obtained from uvcs and other remote - sensing instruments can provide the value of the electrical conductivity of the plasma inside the current sheet in an ongoing eruption for the first time since the impetus of applying reconnection theory to solar eruptions began six decades ago ( giovanelli 1946 ; and also see priest & forbes 2000 ) . with the knowledge of the dynamical process inside the current sheet , we are further able to investigate the particle acceleration taking place in the current sheet
. a strong electric field is induced by magnetic reconnection in the current sheet .
for a typical event , the electric field strength reaches about 5 v / cm ( e.g. , wang et al . 2003 ; qiu et al .
2004 for observations , and martens & kuin 1989 ; forbes & lin 2000 ; lin 2002 for theories ) .
an extremely high value of 50 v / cm was also reported ( xu et al .
2004 ) . in principle , such a strong electric field is able to accelerate any charged particles .
the current sheet is an assembly of waves and electric field , and accelerations can occur in various ways ( see also miller & roberts 1995 ; litvinenko 2000 ) .
we do not yet know exactly what happens in a real eruptive process , but the observations that a series of bright blobs flow successively out of the current sheet ( e.g. , ko et al .
2003 ; lin et al . 2005 ) replicate one of the main characteristics of magnetic reconnection inside the turbulent current sheet : the turbulent eddies or small magnetic islands inside the current sheet tend to merge into bigger ones before they leave the current sheet ( ambrosiano et al .
1988 ) .
further progress in understanding the above processes occurring in the current sheet depends on the accurate measurement of the thickness of the current sheet , plasma parameters in the current sheet ( including electron and ion velocity distributions and densities ) , the speeds of reconnection inflow / outflow near the current sheet , as well as electric and magnetic fields in and around the current sheet .
uv coronagraphic spectroscopy is uniquely suited to these requirements .
uvcs has led to fundamentally new views of the _ acceleration region _ of the solar wind . by measuring emission lines
formed both by collisional excitation and by the resonant scattering of solar - disk photons , uv spectroscopy provides a multi - faceted characterization of the kinetic properties of atoms , ions , and electrons ( e.g. , withbroe et al . 1982 ; cranmer 2002a ) .
the doppler - broadened shapes of emission lines are direct probes of line - of - sight ( los ) particle velocity distributions ( i.e. , essentially providing @xmath8 when the off - limb magnetic field is @xmath6radial ) , and red / blue doppler shifts reveal bulk flows along the los .
integrated intensities of resonantly scattered lines can be used to constrain the solar wind velocity and other details about the velocity distribution in the radial direction ( e.g. , @xmath9 and @xmath10 ) ; this is the so - called `` doppler dimming / pumping '' diagnostic ( e.g. , noci et al .
intensities of collisionally dominated lines especially when combined into an emission measure distribution can constrain electron temperatures , densities , and elemental abundances in the coronal plasma .
even departures from maxwellian and bi - maxwellian velocity distributions are detectable with spectroscopic measurements having sufficient sensitivity and spectral resolution ( e.g. , cranmer 2001 ) . in the high - speed solar wind , uvcs measured outflow speeds that were found to become supersonic much closer to the sun than previously believed . in coronal holes ,
heavy ions ( e.g. , o@xmath11 ) were found to flow faster , to be heated hundreds of times more strongly than protons and electrons , and to have anisotropic temperatures with @xmath12 ( kohl et al .
1997 , 1998 , 1999 ; cranmer et al .
these unexpected results have rekindled theoretical efforts to understand the heating and acceleration of the fast wind in the extended corona ( e.g. , tu & marsch 1997 ; leer et al .
1998 ; axford et al . 1999
; hollweg 1999 ; hollweg & isenberg 2002 ; marsch 2004 ) .
the slow solar wind was found to flow mostly along the outer edges of bright streamers , near locations with measured abundance patterns matching those of the _ in situ _ slow wind ( strachan et al .
2002 ; raymond et al .
1997 ) . the closed - field `` core '' regions of streamers , though , exhibit heavy element abundances only 3 to 30% of those seen at 1 au , indicating gravitational settling ( e.g. , raymond 1999 ; vsquez & raymond 2005 ) .
uvcs observed the transition from a high - density collision - dominated plasma at low heights in streamers to a low - density collisionless plasma at large heights , the latter exhibiting high ion temperatures and anisotropies that suggest similar physics as in the fast wind ( ko et al .
2002 ; frazin et al .
2003 ) . if the kinetic properties of _ additional ions _ were to be measured in the extended corona ( i.e. , a wider sampling of charge / mass combinations ) we could much better constrain the specific kinds of waves that are present as well as the specific collisionless damping modes ( e.g. , cranmer 2002b ) .
measuring the _ electron temperature _ above @xmath61.5 @xmath1 ( never done directly before ) would finally allow us to determine the bulk - plasma heating rate in different solar wind structures , thus putting the firmest ever constraints on models of why the slow [ fast ] speed wind is slow [ fast ] ( e.g. , suess et al . 1999 ; endeve et al . 2004 ; cranmer , these proceedings ) . measuring _
non - maxwellian velocity distributions _ of electrons and positive ions would allow us to test specific models of mhd turbulence , cyclotron resonance , and velocity filtration .
new capabilities such as these would be enabled by greater photon sensitivity , an expanded wavelength range , and the use of measurements that heretofore have only been utilized in a testing capacity ( e.g. , thomson - scattered h i ly@xmath5 to obtain @xmath4 ; the hanle effect to obtain constraints on the magnetic field ) .
these would then allow the relative contributions of different physical processes to the heating and acceleration of all solar wind plasma components to be determined directly .
a mission capable of carrying out the required observations would include two instrument units : a large - aperture ultraviolet coronagraph spectrometer ( see fig . 1 ) and a large - aperture visible light coronagraph .
a suitable design was developed during a midex feasibility study for a mission called asce .
remote external occulters supported by a deployable boom provide much larger unvignetted apertures than are possible with conventional coronagraphs .
these instruments provide major improvements in sensitivity , stray light rejection , spatial resolution , minimum observable height and ultraviolet wavelength range .
new spectroscopic diagnostics for the electron velocity distribution , magnetic field , and parameters for a broad range of newly observable ions are implemented by these instruments .
unprecedented cadences are possible . | we summarize the unique capabilities of uv coronagraphic spectroscopy for determining the detailed plasma properties ( e.g. , density , temperature , outflow speed , composition ) of the source regions of both transient phenomena such as cmes , flares , and solar energetic particles ( seps ) and more time - steady solar wind streams .
uvcs / soho observations have provided the first detailed diagnostics of the physical conditions of cme plasma in the extended corona .
it provided new insights into the roles of shock waves , reconnection , and magnetic helicity in cme eruptions .
we summarize past observations and discuss the diagnostic potential of uv coronagraphic spectroscopy for characterizing two possible sites of sep production : cme shocks and reconnection current sheets .
uvcs / soho has also led to fundamentally new views of the acceleration region of the solar wind .
understanding the physical processes in this region , which ranges from the low corona ( @xmath01.5 @xmath1 ) past the sonic points ( @xmath2 ) , is key to linking the results of solar imaging to _ in situ _ particle and field detection . despite the advances that have resulted from uvcs / soho , more advanced instrumentation could determine properties of additional ions with a wider sampling of charge / mass combinations .
this would provide much better constraints on the specific kinds of waves that are present as well as the specific collisionless damping modes .
electron temperatures and departures from maxwellian velocity distributions could also be measured .
the instrumentation capable of making the above observations will be described .
[ 2005/06/01 1.0 ( src ) ] to be published in the proceedings of _ solar wind 11/soho16 : connecting sun and heliosphere , _
june 1317 , 2005 , whistler , canada , esa sp592 . |
the search for supersymmetry ( susy ) @xcite is one of the main issues in the experimental programs at lep2 and tevatron
. it will play an even more important rle at the future colliders lhc , @xmath3 linear colliders with an energy range up to 2 tev , and @xmath0 colliders with an energy range up to 4 tev .
tevatron and lhc are well designed to discover susy . at these machines even some precision measurements are possible @xcite . however , for a precise determination of the underlying susy parameters lepton colliders will be necessary . owing to their good energy resolution @xmath0
colliders are well suited for this purpose @xcite .
the most exciting feature is the possibility of producing higgs bosons in the s
channel @xcite .
in this paper we study the production of third generation sfermions in @xmath0 annihilation , paying particular attention to the energy range near the higgs boson resonances .
our framework is the minimal supersymmetric standard model ( mssm ) @xcite .
the mssm implies the existence of five physical higgs bosons : two scalars @xmath4 , one pseudoscalar @xmath5 , and two charged ones @xmath6 @xcite . every standard model ( sm ) fermion has two supersymmetric partners , one for each chirality state denoted by @xmath7 and @xmath8 .
the sfermions of the third generation are particularly interesting @xcite because their phenomenology is different compared to that of the sfermions of the first and second generation .
the reasons for this are the mixing between @xmath7 and @xmath8 and the large yukawa couplings .
in particular , the top quark and the stops give substantial contributions to higgs boson masses due to radiative corrections ( see e.g. @xcite ) .
moreover , the top and stop contributions to the renormalization group equations play an essential rle in inducing electroweak symmetry breaking , when the higgs parameters evolve from the gut scale to the electroweak scale @xcite .
therefore , the couplings of the stops to the neutral higgs bosons are of special interest . also the tau and bottom yukawa couplings can be large if @xmath9 is large , giving rise to similar effects in the phenomenology of sbottoms and staus .
the paper is organized in the following way : in sec . ii we present the underlying parameters and give the formulae for the production cross sections . in sec .
iii we discuss numerical results for sfermion pair production . in sec .
iv we summarize the main results . in the appendix
we give the higgs sfermion couplings .
in this section we present the formulae for sfermion mixing and the production cross sections of sfermions at a @xmath0 collider .
the main parameters for the following discussion are @xmath10 , @xmath11 , @xmath9 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 , and @xmath19 @xcite .
@xmath10 is the mass of the pseudoscalar higgs boson , @xmath11 is the higgs mixing parameter in the superpotential , and @xmath20 , where @xmath21 denotes the vacuum expectation value of the higgs doublet @xmath22 .
@xmath12 , @xmath13 , @xmath14 , @xmath15 and @xmath16 are soft susy breaking masses for the sfermions , @xmath23 , @xmath24 , and @xmath19 are trilinear higgs sfermion parameters .
the mass matrix for sfermions in the @xmath25 basis has the following form : @xmath26 with @xmath27 and @xmath28 the mass eigenstates @xmath29 and @xmath30 are related to @xmath31 and @xmath32 by : @xmath33 with the eigenvalues @xmath34 the mixing angle @xmath35 is given by @xmath36 the mass of the tau
sneutrino is given by @xmath37 figure [ feynman ] shows the feynman graphs for the processes @xmath38 @xmath39 .
the total and the differential cross sections read up to @xmath40 : @xmath41 \ , , \label{sigtot } \\
\frac{d \ , \sigma(\mu^+ \mu^- \to \tilde f_i \bar{\tilde f_j } ) } { d \ , \cos \vartheta } q & = & c_{ij } \left [ \frac{\lambda_{ij}}{s^2 } t_{vv } \ , \sin^2 \vartheta + \frac{m_{\tilde f_i}^2-m_{\tilde f_j}^2}{s } t^a_{vh } + \frac{\lambda^{1/2}_{ij}}{s } t^b_{vh } \cos \vartheta + t_{hh } \right ] \label{sigdiff}\end{aligned}\ ] ] with the kinematic function @xmath42 , @xmath43 the scattering angle of @xmath44 , and @xmath45 where @xmath46 is a colour factor which is 3 for squarks and 1 for sleptons .
@xmath47 denotes the contribution from @xmath48 and @xmath49 exchange , @xmath50 the interference terms between gauge and higgs bosons , and @xmath51 the contribution stemming from the exchange of higgs bosons .
the pure gauge boson contribution , the first term of eqs .
( [ sigtot ] ) and ( [ sigdiff ] ) , is the same as for @xmath52 @xcite .
notice that the gauge boson term shows a @xmath53 dependence whereas the terms proportional to @xmath51 and @xmath54 are independent of @xmath43 .
the term proportional to @xmath55 shows a @xmath56 dependence giving rise to a forward backward asymmetry .
however , @xmath55 is proportional to @xmath57 ( see the following equations ) and , therefore , is rather small .
we obtain : + 1 ) @xmath58 @xmath59 + \frac{a^2_{ii } ( v_\mu^2+a_\mu^2 ) \ , s^2 } { 256 \ , \sin^4 \theta_w \cos^4 \theta_w } | d ( z ) |^2 \ , , \nonumber \\
% t_{hh } & = & \ , h_\mu^2\ , \frac{s}{2 e^2 \sin^2 \theta_w } \ , |g^{h^0}_{ii } \sin \alpha d(h^0 ) - g^{h^0}_{ii } \cos \alpha d(h^0 ) |^2 \ , , \nonumber \\
% t^a_{vh } & = & 0 \ , , \nonumber \\
% t^b_{vh } & = & h_\mu m_\mu \ , \frac{2 \sqrt{2}}{e \sin \theta_w } \big [ e_f re [ g^{h^0}_{ii } \sin \alpha d(h^0 ) - g^{h^0}_{ii } \cos \alpha d(h^0 ) ] \nonumber \\ & & - \frac{v_\mu a_{ii } \ , s}{16 \sin^2 \theta_w \cos^2 \theta_w } \ , re[d^*(z ) ( g^{h^0}_{ii } \sin \alpha d(h^0 ) - g^{h^0}_{ii } \cos \alpha d(h^0 ) ) ] \big ] \ , , \label{tii}\end{aligned}\ ] ] 2 ) @xmath60 @xmath61 \ , , \nonumber
\\ % t^a_{vh } & = & - h_\mu m_\mu \
, \frac { \sqrt{2 } \ , a_{ij } \ , a_\mu \sin \beta \ , \
, \,s } { 8 e \sin^3\theta_w \cos^2 \theta_w } g^{a^0}_{12 } \ , re [ d^*(z ) d(a^0 ) ] \ , , \nonumber \\
% t^b_{vh } & = & -h_\mu m_\mu\ , \frac{\sqrt{2 } \ , v_\mu a_{ij } \
, s}{8 e \sin^4 \theta_w \cos^2 \theta_w } re[d^*(z ) ( g^{h^0}_{ij } \sin \alpha d(h^0 ) - g^{h^0}_{ij } \cos \alpha d(h^0 ) ) ] , \label{tij}\end{aligned}\ ] ] where @xmath62 is the yukawa coupling of the muon , @xmath63 , @xmath64 ; @xmath65 is the charge of the sfermions ( @xmath66 ) @xmath67 and @xmath68 are the vector and axial vector couplings of the muon to the @xmath69 boson , @xmath70 , @xmath71 , and @xmath72 are the corresponding couplings @xmath73 : @xmath74 where @xmath75 is the third component of the weak isospin of the fermion @xmath76 .
@xmath77 , @xmath78 , and @xmath79 are the couplings of the sfermions @xmath44 and @xmath80 ( i , j = 1 , 2 ) to the light , the heavy , and the pseudoscalar higgs boson , respectively .
these couplings depend on @xmath81 , @xmath11 , @xmath9 , @xmath82 , @xmath83 , where @xmath84 is the mixing angle of @xmath85 and @xmath86 .
the explicit form of these couplings is given in the appendix .
in this section we present the numerical results for sfermion production in the various channels . in this analysis
we have taken : @xmath87 , @xmath88 , @xmath89 gev , @xmath90 gev , @xmath91 gev , and @xmath92 mev . for the calculation of the cross section
we need the total decay widths of the higgs bosons , where we calculate all possible two body decay modes that are allowed at tree level @xcite . for this calculation
we fix @xmath93 gev and use the gut relation @xmath94 , where @xmath95 and @xmath96 are the @xmath97 and @xmath98 gaugino masses , respectively . as usual
we have included radiative corrections in the calculation of @xmath99 , @xmath100 , and @xmath83 using @xcite . in fig .
[ stopsqrts ] we show the total cross sections for stop production as a function of @xmath101 for @xmath102 gev , @xmath103 , @xmath104 gev , @xmath105 gev , @xmath106 , and @xmath107 gev . for the total widths of the higgs bosons we take @xmath108 gev , @xmath109 gev , @xmath110 , @xmath111 gev , @xmath112 gev , and @xmath113 gev .
the full lines show the total cross sections and the dashed lines the gauge boson contributions .
the latter ones are identical with the cross sections of @xmath114 . for @xmath115 production
the peak results from the @xmath86 exchange leading to an enhancement of @xmath116 fb compared to the gauge boson contribution . for @xmath117 production
the peak is an overlap of the @xmath86 and @xmath5 resonances because @xmath118 and the widths of @xmath5 and @xmath86 are of the order of several gev ( see e.g. @xcite ) .
note that the higgs boson contribution is much larger than the gauge boson contribution .
we have found that the forward
backward asymmetry @xmath119 is @xmath120 at its maximum .
therefore , a rather high luminosity would be needed to measure it . in fig .
[ stopcos ] we show the production cross sections for @xmath121 and @xmath122 ( without including the charge conjugate state ) as a function of @xmath123 .
we have chosen the following procedure for the calculation of the parameters : we fix @xmath104 gev , @xmath105 gev , @xmath124 gev , and @xmath11 , @xmath9 , @xmath10 , @xmath16 , @xmath15 , @xmath19 as above .
we calculate @xmath13 , @xmath14 , and @xmath18 from the stop masses and mixing angle .
we take @xmath125 ( we have checked , that our results are not sensitive to this assumption ) .
these parameters are then used for the calculation of @xmath126 , @xmath127 , @xmath128 , @xmath99 , @xmath100 , @xmath83 , @xmath129 , and @xmath130 .
we have chosen this procedure to minimize the dependence of the physical higgs quantities on @xmath123 . in fig .
[ stopcos]a we show the total cross section @xmath131 , the higgs boson contribution , and the gauge boson contribution for @xmath132 gev .
the higgs boson contribution depends on the sign of @xmath123 .
this leads to a dependence of the total cross section on the sign of @xmath123 contrary to the case of an @xmath3 collider , where the cross section depends only on @xmath133 @xcite .
note that the higgs boson contribution can be larger than the gauge boson contribution .
this is in particular the case for large mixing in the stop sector because the higgs boson couples more strongly to the left
right combination of the stops than to the left left or
right right combinations .
the higgs boson contribution vanishes for @xmath134 because the corresponding coupling is zero for the parameters chosen .
one can disentangle the higgs boson contribution from the gauge boson part by measuring the differential cross section . as can be seen in eq .
( [ sigdiff ] ) the higgs boson contribution of the differential cross section does not depend on @xmath43 whereas the gauge boson contribution shows a @xmath53 shape .
we can safely neglect the interference terms between gauge and higgs bosons because they are rather small as we have seen above . in fig .
[ stopcos]b the total cross section is shown for various values of @xmath101 between 444 gev and 454 gev . for larger values of @xmath101
the cross section is decreasing because @xmath100 varies between 452.8 gev ( @xmath135 ) and 454.2 gev ( @xmath136 ) .
the @xmath123 dependence of @xmath100 is also the reason for @xmath137 if @xmath138 . in fig .
[ stopcos]c we show @xmath139 , the higgs boson contribution and the gauge boson contribution as a function of @xmath123 .
an interesting feature here is that for @xmath140 only the higgs bosons contribute .
moreover , the contribution of the higgs bosons @xmath141 and @xmath142 is generally larger than that of the gauge boson .
the total cross section again depends on the sign of @xmath123 . in fig .
[ stopcos]d we show the total cross section for various values of @xmath101 .
the shift of the peak is due the dependence of @xmath100 and @xmath83 on @xmath18 which is calculated from @xmath123 .
note that the coupling @xmath143 is large in the range where the coupling @xmath144 is small and vice versa .
the minima near @xmath145 are due to the vanishing of the coupling @xmath143 . in fig .
[ sbotsqrts ] the cross sections for sbottom production are shown as a function of @xmath101 for @xmath146 gev , @xmath147 gev , @xmath148 , @xmath149 gev , @xmath150 gev , @xmath151 , @xmath152 gev , @xmath153 , and @xmath107 gev .
@xmath16 , @xmath15 , and @xmath19 are taken as above .
it is interesting that @xmath154 is @xmath155 times larger than @xmath156 at @xmath2 .
this has two implications : first , one gets a cross section that is large enough to be measured even with an integrated luminosity of 10 fb@xmath157 .
second , the @xmath158 cross section is even larger than the @xmath159 production cross section .
note that we only show the cross section for @xmath158 whereas in the experiment one can measure the cross section of @xmath160 . in fig .
[ sbotcos ] we show the cross sections for sbottom production as a function of @xmath128 for @xmath146 gev , @xmath147 gev , @xmath124 gev , @xmath149 gev , @xmath150 gev , @xmath153 , and @xmath107 gev .
@xmath16 , @xmath15 , and @xmath19 are taken as above .
we have calculated the other parameters in the following way : from @xmath126 , @xmath127 , @xmath128 , and @xmath17 we get @xmath13 , @xmath12 , and @xmath11 .
we then take @xmath13 , @xmath161 , and @xmath162 to calculate @xmath14 , @xmath123 , and @xmath18 .
there are similarities to the stop case : for @xmath159 production the higgs boson contribution can be larger than the gauge boson part ( fig .
[ sbotcos]a ) . in the case of @xmath163 production ,
the contribution of the higgs bosons is much larger than that of the gauge boson ( fig . [
sbotcos]c ) .
the main difference compared to the stop case is that the asymmetry in the sign of @xmath128 is much more pronounced in the @xmath163 channel than in the @xmath159 channel as a consequence of the corresponding couplings ( fig .
[ sbotcos]b and d ) .
the peak at @xmath164 in fig .
[ sbotcos]d results not only from the couplings but also from the fact that the total decay width of @xmath5 has a minimum there . in fig .
[ stausqrts ] the stau production cross sections are shown as a function of @xmath101 for @xmath165 gev , @xmath166 gev , @xmath167 , @xmath168 gev , @xmath169 gev , @xmath170 gev , @xmath171 gev , @xmath102 gev , @xmath153 , and @xmath172 gev .
the parameters are chosen such that the higgs boson can not decay into squarks or the top quark .
therefore , the total decay widths of @xmath86 and @xmath5 are one order of magnitude smaller than in the previous examples .
this leads to the large peaks at @xmath173 .
moreover , the decay widths are so small that one can see two peaks in the case of @xmath174 production
. it should be possible to observe both peaks in a real experiment because of the good energy resolution of a @xmath0 collider @xcite . in fig .
[ staucos ] the cross sections for stau production are presented as a function of @xmath175 for @xmath165 gev , @xmath166 gev , @xmath113 gev .
@xmath9 , @xmath10 , and the squark parameters are taken as above .
we have calculated @xmath16 , @xmath15 and @xmath11 from @xmath176 and @xmath175 . with these and the other parameters we have calculated @xmath177 , @xmath100 , @xmath83 , @xmath129 , and @xmath130 .
note that the masses , mixing angle , and decay widths of the higgs bosons depend indirectly on @xmath175 due to the induced change in @xmath11 .
this fact leads to the observed shifts of the maximal cross section with @xmath101 in fig .
[ staucos]b and d. in @xmath174 production the higgs boson contribution is much larger than the gauge boson contribution ( fig.[staucos]c ) similar to the squark production .
this is particularly important , because in this case the production cross section is most likely too small to be seen at an @xmath3 collider @xcite .
the cross sections depend strongly on the sign of @xmath175 in both channels ( fig .
[ staucos]a and c ) . note also that , for the parameters chosen in the higgs couplings to the staus , the gauge couplings are of the same size as the yukawa couplings ( eqs.([higgsstaua])([higgsstaub ] ) ) : in fig . [ sneutsqrts ] we show the total cross sections for sneutrino production as a function of @xmath101 for the same parameters as in fig .
[ stausqrts ] ( implying @xmath178 gev ) .
the large peak at @xmath1 is due to the small total decay width of @xmath86 ( @xmath179 gev ) and due to the coupling @xmath180 which is a gauge coupling ( see eq .
( [ coupsneut ] ) ) stemming from a d term .
we have studied sfermion pair production in @xmath0 annihilation focusing on the impact of the higgs boson resonances in these processes .
we have seen that the production cross sections can be considerably enhanced at these resonances for all sfermions of the third generation .
the most important results are : first , the production cross sections depend on the sign of @xmath82 .
second , the higgs boson contributions dominate the production cross section of @xmath181 .
we have seen that the cross sections can be large enough to be studied at a @xmath0 collider even if the corresponding cross sections are too small to be measured at an @xmath3 collider .
third , the higgs boson contribution can even be larger than the gauge boson contributions in the @xmath182 channel . from these facts
we conclude that a @xmath0 collider is an excellent machine for obtaining important information on the @xmath183 and @xmath184 couplings .
we are very grateful to m. carena and s. protopopescu for their kind invitation to the workshop on physics at the first muon collider and the front end of a muon collider
( fnal , batavia , illinois , usa , november 6 9 , 1997 ) where this study was initiated .
this work was supported by the `` fonds zur frderung der wissenschaftlichen forschung '' of austria , project no .
in this section we list the couplings of the neutral higgs bosons to sfermions .
we concentrate here on the couplings of @xmath86 and @xmath5 which are important for our investigation .
one can get the couplings of @xmath85 from those of @xmath86 by the replacements : @xmath185 and @xmath186 .
+ higgs couplings to stops : @xmath187 higgs couplings to sbottoms : @xmath188 higgs coupling to the tau sneutrino : @xmath189 higgs couplings to staus : @xmath190 99 h. p. nilles , phys . rep . * 110 * , 1 ( 1984 ) .
h. e. haber and g. l. kane , phys
. rep .
* 117 * , 75 ( 1985 ) .
i. hinchliffe et al . , phys .
d * 55 * , 5520 ( 1997 ) ; f. paige , hep - ph/9801395 , to appear in the proceedings of the workshop on physics at the first muon collider and the front end of a muon collider , batavia , il , november 69 , 1997 .
j. f. gunion , hep - ph/9802258 , to be published in the proceedings of workshop on physics at the first muon collider and at the front end of the muon collider , batavia , il , 69 nov 1997 .
j. f. gunion et al . , proceedings of the 1996 dpf / dpb summer study on high - energy physics , snowmass , colorado , june 25
july 12 , 1996 , eds .
d. g. cassel , l. trindle gennari , r. h. siemann , vol . 2 , p. 541 .
v. barger , m. s. berger , j. f. gunion , and t. han , phys .
* 286 * , 1 ( 1997 ) ; h. e. haber , talk given at the workshop on physics at the first muon collider and at the front end of the muon collider
, batavia , il , 69 nov 1997 , to be published in the proceedings . j. f. gunion , h. e. haber , nucl . phys . * 272 * , 1 ( 1986 ) . j. f. gunion , h. e. haber , g. l. kane , and s. dawson , _ the higgs hunter s guide _ , addison - wesley , 1990 , and references therein .
j. ellis and s. rudaz , phys . lett .
b * 128 * , 248 ( 1983 ) ; g. altarelli and r. rckl , phys .
b * 144 * , 126 ( 1984 ) .
a. bartl , w. majerotto , and w. porod , z. phys .
c * 64 * , 499 ( 1994 ) ; a. bartl et al .
, z. phys .
c * 73 * , 469 ( 1997 ) ; a. bartl et al .
, z. phys .
c * 76 * , 549 ( 1997 ) .
e. accomando et al .
( ecfa / desy lc physics working group ) , desy-97 - 100 , ( 1997 ) , submitted to phys .
j. ellis , g. ridolfi , and f. zwirner , phys .
b * 257 * , 83 ( 1991 ) ; phys .
b * 262 * , 477 ( 1991 ) .
h. e. haber and r. hempfling , phys .
lett . * 66 * , 1815 ( 1991 ) .
l. e. ibez and g. g. ross , phys .
b * 110 * , 215 ( 1982 ) .
h. eberl , a. bartl , w. majerotto , nucl .
* b472 * , 481 ( 1996 ) .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto , and y. yamada , phys .
b * 389 * , 538 ( 1996 ) .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto , and y. yamada , phys .
b * 402 * , 303 ( 1997 ) .
a. djouadi , j. kalinowski , p. ohmann , and p. m. zerwas , z. phys .
c * 74 * , 93 ( 1997 ) .
( 320,200 ) ( 30,0 ) ( 10,185 ) ( 45,172 ) ( 45,112 ) ( 80,150 ) ( 135,172 ) ( 133,110 ) ( 45,70 ) ( 45,10 ) ( 80,46 ) ( 135,70 ) ( 133,8 ) ( 222,185 ) ( 257,172 ) ( 257,112 ) ( 300,150 ) ( 347,172 ) ( 345,110 ) ( 257,70 ) ( 257,10 ) ( 278,45 ) ( 347,70 ) ( 345,8 ) | we discuss pair production of stops , sbottoms , staus and tau sneutrinos at a @xmath0 collider .
we present the formulae for the production cross sections and perform a detailed numerical analysis within the minimal supersymmetric standard model .
in particular , we consider sfermion production near @xmath1 and @xmath2 .
uwthph-1998 - 10 + hephy - pub 689/98 + hep - ph/9805248 * sfermion pair production at @xmath0 colliders * + |
the so - called hegselmann - krause bounded confidence model ( hk - model for brevity ) , introduced in @xcite , is one of the most popular mathematical models for the dynamics of opinion formation in groups of interacting agents . in its classical formulation
, we have a finite number @xmath0 of agents , indexed by the integers @xmath1 .
time is measured discretely and the opinion of agent @xmath2 at time @xmath3 is represented by a real number @xmath4 .
there is a fixed parameter @xmath5 such that the dynamics are given by @xmath6 where @xmath7 . in words , at each time step an agent takes account of his so - called _
@xmath8-neighbourhood _ , consisting of those other agents
whose opinions currently lie within distance @xmath8 of his own .
he updates his opinion to the average of those held by the members of his @xmath8-neighbourhood , including himself . in purely formal terms
, the model makes sense if opinions @xmath4 are assumed to come from a set @xmath9 which has enough structure so that it is possible to make sense of the command to + _ ` move to the average of a finite collection of points within distance @xmath8 of your present location'_. + + we shall refer to any rigorous formulation of this procedure as a _ hk - update rule_. the simplest generalisation would be to @xmath10 for any @xmath11 and with the euclidean metric .
this is natural if we imagine that there are @xmath12 ` issues ' on which agents have opinions , and they are only willing to compromise with those whose opinions on _ every _ issue are sufficiently close to their own .
there are rigorous results in the literature concerning the hk - dynamics in euclidean space of arbitrary dimension , see for example @xcite .
note , though , that it is not clear what is the most natural formulation of the hk - dynamics in @xmath10 for @xmath13 .
for example , the @xmath14-metric may better capture the notion that an agent will only compromise with those whose opinions on all issues are close to their own indeed , the @xmath15-metric seems better suited when considering the hk - dynamics as a rendezvous procedure , as in for example @xcite . however , it is also plausible that the hk - update rule is too strict a criterion for compromise .
at least one group of authors @xcite has suggested a more general model in which there is a second parameter @xmath16 such that one agent will compromise with another whose opinions on at least @xmath17 of the @xmath12 issues are within a fixed distance of his own .
note that , in this generality , we are no longer dealing with a hk - update rule , as formulated above , since if @xmath18 then the ` distance ' between agents does not satisfy the triangle inequality .
if we want to stick with the hk - update rule , then perhaps the simplest and most natural example to consider after euclidean space is a circle .
let @xmath19 denote a circle of perimeter @xmath20 , equipped with its natural metric which we will denote by @xmath21 .
if @xmath22 then the ` average location of those in your @xmath8-neighbourhood ' is not well - defined , unless there is a priori agreement on an orientation of the circle and an origin , i.e. : on a continuous bijection from @xmath19 to @xmath23 . in any case , if such a priori agreement existed , then the dynamics would become trivial as any configuration would collapse immediately to a complete consensus .
the circle is also uninteresting if @xmath24 , since then there must be a pair of consecutive agents at distance greater than @xmath8 apart , so the configuration evolves as if it were living on the real line , with these two agents as the extreme opinions .
hence , when studying the circle , we may assume that @xmath25 .
indeed , without loss of generality we can fix @xmath26 and have @xmath27 $ ] as the only variable parameter .
there are at least two possible motivations for considering the hk - dynamics on a circle : a. firstly , it is possible to think of ` real - life ' situations where it makes more sense to consider opinions as lying on a circle instead of the real line .
for example , the subject under debate may be the choice of a time of day or date on which to hold some event .
b. secondly , the circle is perhaps the simplest example of a space on which the hk - dynamics behave differently from in euclidean space in a fundamental sense which we now describe .
it is a well - known fact that , in euclidean space , any configuration of opinions obeying the dynamics in ( [ eq : update ] ) will freeze in finite time , that is , there will be some @xmath28 such that @xmath29 for all @xmath30 and @xmath31 . in fact , it is known that the time taken to freeze is bounded above by a universal polynomial function of the number @xmath0 of agents . in one dimension ,
the best bound to date is @xmath32 , proven independently in @xcite and @xcite . in higher dimensions , the best published bound @xcite is @xmath33 , but a recent preprint @xcite improves this to @xmath34 .
moreover , in a frozen configuration in @xmath10 , agents must either _ agree _ , that is @xmath35 , or be beyond each other s influence , that is @xmath36 . on the other hand ,
if on the circle we place @xmath37 agents at angles @xmath38 , @xmath39 , then we have a frozen configuration in which no proper subset of the agents is isolated from its complement , but nor are any two agents in agreement
. there also exist frozen configurations with these properties in which agents are not equally spaced .
it is easy to see that any such configuration must include at least @xmath40 agents , and that the only example with @xmath40 agents , modulo translations , is that in figure [ fig : stable ] below .
furthermore , on the circle there are configurations which never freeze .
for example , if we perturb the configuration in figure [ fig : stable ] by a sufficiently small amount then we will obtain one starting from which we never freeze , since the updated configurations will converge back to that in figure [ fig : stable ] , possibly translated , at an exponential rate .
see remark [ rem:4points ] below for a proof of this last statement . before proceeding ,
let us formally define the term _ convergence _ for a finite set of agents obeying the hk - dynamics on a circle : [ defi : converge ] let @xmath41 , @xmath42 .
let @xmath43 , the latter equipped with the product topology .
suppose that , for all @xmath44 , the vectors @xmath45 are defined inductively by the hk - update rule , with @xmath46 .
if , for some @xmath47 , we have @xmath48 in the product topology , then the sequence of configurations @xmath49 is said to _ converge _ to @xmath50 .
it is common to abuse terminology slightly and say that the initial configuration @xmath51 converges to @xmath50 . in @xcite
it is stated as an open problem for @xmath19 whether any initial configuration will converge .
the only other explicit mention of this problem that we could find is in @xcite and , as far as we know , it has remained open up to now .
our first main result gives a positive answer : [ thm : converge ] for any @xmath41 , @xmath42 and @xmath52 , the sequence @xmath53 defined by the hk - update rule converges .
the proof of this result will be given in the next section .
it requires only moderate variations on ideas already presented in the literature when studying the hk - model in euclidean space .
two important and well - known concepts are involved .
let @xmath54 .
firstly , we define the _ energy _ of the configuration @xmath55 as @xmath56 secondly , we define the _ influence graph _ @xmath57 corresponding to @xmath55 to be the undirected graph whose vertices are @xmath58 and where an edge is placed between @xmath2 and @xmath59 if and only if @xmath60 . given an orientation of the circle , we can also define a _ directed influence graph _ @xmath61 by choosing the oriented pair @xmath62 if and only if the shortest path from @xmath2 to @xmath59 follows the orientation .
the proof of theorem [ thm : converge ] consists of two main parts .
firstly , we show that an application of the hk - update rule always decreases the energy of a configuration , by an amount which is bounded below in terms of @xmath63 .
a similar result has been rediscovered several times over for euclidean space , the earliest reference we could find is theorem 2 of @xcite . our proof for the circle is almost identical to that for euclidean space given in theorem 4.3 of @xcite .
secondly , we show that if the digraphs @xmath64 and @xmath65 are different , that is , if at least one edge has been either added , deleted or changed orientation at time @xmath66 , then the sum of the decreases in energy at times @xmath66 and @xmath67 is bounded below by @xmath68 . since the initial energy is at most @xmath69 , this implies that the digraph @xmath70 can change at most @xmath71 times and , in particular , must be fixed from some @xmath66 onwards .
once we know this , convergence will follow from standard linear algebra arguments .
our reasoning in the second part of the proof also closely resembles known arguments for euclidean space , though the latter all seem to appeal at some point to the fact that the points of @xmath72 all lie inside the convex hull of those in @xmath49 .
there is no analogous fact for the circle , which gives the argument a new twist .
the final proof of convergence , although it only appeals to basic linear algebra , also has no direct counterpart in euclidean space since configurations always freeze eventually in the latter .
+ + in section 3 , we will prove another result .
given that freezing occurs in euclidean space after a time which is bounded by a universal polynomial function of the number @xmath0 of agents and given that , on the circle , the directed influence graph changes at most a polynomial - in-@xmath0 number of times , it is natural to ask whether freezing of the latter must also happen within a time which is bounded by a polynomial function of @xmath0 only .
this turns out not to be the case .
we shall prove that , for all sufficiently large @xmath0 and any @xmath73 , there is a choice of perimeter @xmath20 and a configuration of @xmath0 agents whose influence digraph remains the same up to time @xmath74 but eventually changes .
indeed , we can choose the configuration so that @xmath70 is strongly connected all the way around the circle until the first change occurs , at which point a pair of adjacent agents become disconnected in the underlying graph and the configuration subsequently collapses to a complete consensus after three additional time steps .
section 4 discusses some remaining open problems .
the first step in the proof is to show that the hk - dynamics decrease the energy of a configuration as defined in ( [ eq : energy ] ) .
the proof of the following result involves only minor modifications from that of theorem 4.3 in @xcite , but we will present it for the sake of completeness . [
prop : nrj ] if the sequence of configurations @xmath53 obeys the hk - update rule then , for any @xmath75 , @xmath76 fix an ( anti - clockwise ) orientation and an origin of @xmath19 , i.e. : a suitable bijection @xmath77 . in
what follows we shall often not distinguish in writing between a point @xmath78 and its image @xmath79 - hopefully , it will always be clear form the context which is being referred to . for each @xmath80 ,
let @xmath81^n$ ] be the vector such that @xmath82 note that ( [ eq : minsk ] ) can be written in the form @xmath83 where @xmath84 is the euclidean norm on @xmath85 . for every @xmath86 set @xmath87
we write @xmath88 if @xmath89 .
define functions @xmath90 as follows : @xmath91 where , if @xmath92 and @xmath93 , then @xmath94 clearly the functions @xmath95 are symmetric , i.e. : @xmath96 , and @xmath97 it is easy to see that the function @xmath98 is strictly convex on @xmath85 and attains its global minimum at @xmath99 .
hence if we define @xmath100 by @xmath101 then @xmath102 it is also clear that @xmath103 is a convex second - degree polynomial in @xmath104 and @xmath105 .
thus it must have the form @xmath106 in particular , @xmath107 which means that @xmath108 secondly , we claim that @xmath109 note that ( [ eq : ineq2 ] ) , ( [ eq : ineq1 ] ) and ( [ eq : ee ] ) would together imply ( [ eq : norminsk ] ) . to prove ( [ eq : ineq2 ] ) we simply note that , on the one hand , if @xmath30 then @xmath110 while , if @xmath111 , then it is easily verified that @xmath112 together ( [ eq : ineq4 ] ) and ( [ eq : ineq3 ] ) imply ( [ eq : ineq2 ] ) , so the proof of the proposition is complete .
the second step in the proof of theorem [ thm : converge ] is to establish a lower bound on the decrease in energy resulting from any change in the directed influence graph .
[ prop : graph ] let @xmath44 and suppose that @xmath113 . then , with notation as in ( [ eq : delta ] ) , @xmath114 then a priori at least one of the following four things must have happened at time @xmath66 : a. some agent , @xmath115 say , has a new neighbour @xmath116 to his right . in other words ,
the oriented edge @xmath117 is in @xmath70 but not in @xmath64 .
note that it could still be the case ( if @xmath118 ) that @xmath119 , it will not matter for the argument we present below .
b. some agent @xmath115 has a new neighbour @xmath116 to his left . c. some agent @xmath115 has a lost a neighbour @xmath116 on his right .
d. some agent @xmath115 has lost a neighbour @xmath116 on his left .
we suppose that ( i ) occurs - analogous arguments can be given in each of the other three cases .
pick an oriented edge @xmath117 which is added at time @xmath66 .
if either of these agents moved a distance of at least @xmath120 at step @xmath66 , then @xmath121 and we are done . otherwise , the distance between these two agents at time @xmath66 is still at least @xmath122 .
now agent @xmath115 will draw agent @xmath116 to the left ( i.e. : clockwise ) at step @xmath67 .
if agent @xmath116 were to move a distance of at least @xmath120 to the left at step @xmath67 then we would be done again .
but since @xmath116 gained a neighbour to his left at time @xmath66 and since the ordering of agents around the circle is always preserved by the hk - dynamics , he can not have lost any neighbours to his left at time @xmath66 .
hence , there are only two possible ways he can be prevented from moving at least @xmath120 to his left at time @xmath67 : a. there is some @xmath123 such that the oriented edge @xmath124 is also added to the digraph at time @xmath66 , or b. the neighbours of @xmath116 at time @xmath125 , including himself , move on average sufficiently far to the right at time @xmath66 to compensate for the appearance of @xmath115 .
let @xmath126 be the total net movement to the right of these neighbours at time @xmath66 . then
, since @xmath116 has certainly no more than @xmath0 neighbours at time @xmath66 and since @xmath115 is such a neighbour , his movement to the left at time @xmath67 is bounded below by @xmath127 .
thus @xmath128 .
since @xmath116 has at most @xmath129 neighbours at time @xmath125 , by the cauchy - schwarz inequality the sum of their contributions to @xmath130 is at least @xmath131 , which is greater than @xmath68 for all @xmath132 .
thus we are done again unless ( a ) occurs .
now we can apply the same analysis to the pair @xmath124 and so on . for the proposition to fail
, we would have to have an infinite , anti - clockwise sequence of agents @xmath133 such that each oriented edge @xmath134 appeared at time @xmath66 .
but this would mean that , for every @xmath12 , agent @xmath135 moved further to the left than @xmath136 at time @xmath66 .
since there are only finitely many agents , this is impossible .
_ of theorem [ thm : converge]_. the energy of a configuration of @xmath0 agents is certainly no more than @xmath69
. hence , if the sequence of configurations @xmath53 obeys the hk - update rule , it follows from propositions [ prop : nrj ] and [ prop : graph ] that there are at most @xmath71 values of @xmath66 for which @xmath137 .
thus it suffices to prove theorem [ thm : converge ] under the assumption that the directed influence graph @xmath138 is the same for all time .
furthermore , we may assume that the underlying graph @xmath139 is connected , as otherwise the dynamics are the same as for a configuration on the real line , in which case we know it will freeze in finite time .
once again , let us fix a choice of a suitable bijection @xmath140 .
the map @xmath141 induces an ordering of the components of any @xmath142 .
let us assume that , for our initial configuration @xmath143 , we have @xmath144 . then , since the hk - dynamics will never cause agents to cross , for all @xmath66 the agents @xmath145 will retain an anti - clockwise ordering , even though their ordering with respect to @xmath141 may be cyclically shifted .
let @xmath146 denote the set of probability vectors in @xmath85 , i.e. : @xmath147 .
then there is a natural map @xmath148 defined by @xmath149 since the directed influence graph is fixed , there is a fixed @xmath150 real matrix @xmath151 such that the dynamics satisfy @xmath152 by proposition [ prop : nrj ] , we know that the differences @xmath153 tend to @xmath154 as @xmath155 . hence , letting @xmath156 , we have @xmath157 hence , @xmath158 must be identically zero in those coordinates corresponding to each block in the jordan normal form of @xmath151 whose eigenvalue is at least one in absolute value .
it follows that there exists a @xmath159 , only depending on @xmath151 , such that @xmath160 .
thus @xmath161 must converge to a limit , call it @xmath162 , such that @xmath163 ) , it follows that there is some @xmath164 such that @xmath165 and hence in turn , by proposition [ prop : nrj ] , that @xmath166 .
this establishes convergence of the sequence @xmath53 , in the sense of definition 1.1 .
[ rem : irreda ] an alternative argument for obtaining ( [ eq : xinfty ] ) is as follows . it is easy to see that the matrix @xmath151 in ( [ eq : distdyn ] ) has non - negative entries .
we claim that @xmath151 is column - stochastic .
to prove this , it suffices to prove that @xmath167 by definition , ( [ eq : ones ] ) holds for @xmath168 and any @xmath66 .
now let @xmath158 be any probability vector .
the transformation @xmath169 describes an updating of a configuration of @xmath0 agents , possibly involving a cyclic shift of the agents indices .
this may not coincide with the hk - update rule for this configuration as there is no guarantee that agents interact precisely with those within unit distance of themselves .
however , exactly the same pairs of agents interact as would be the case if the configuration were some @xmath49 . in particular
, this means that the anti - clockwise ordering of agents will be preserved ( no agents will cross ) , which implies that @xmath170 will also be a probability vector .
thus ( [ eq : ones ] ) holds for any probability vector , and hence for any @xmath171 .
now @xmath151 may not be regular , since there may be pairs of agents that have exactly the same neighbours in @xmath138 at @xmath172 .
however , any such pair will immediately reach consensus at @xmath173 .
if we start at @xmath173 and replace the vector @xmath174 of ( [ eq : distform ] ) by the ( possibly shorter ) vector of non - zero distances , then it is easy to see that we have an analogue of ( [ eq : distdyn ] ) in which the transition matrix is now regular
. thus we will have exponential convergence to the steady - state , which yields ( [ eq : xinfty ] ) .
[ rem:4points ] with notation as above we have , for the configuration @xmath175 , @xmath176 , of figure 1 that @xmath177 , \;\;\;\ ; { \mbox{\boldmath $ { x } $ } } ^ { * } = a{\mbox{\boldmath $ { x } $ } } ^ { * } \;\ ; { \hbox{where } } \;\ ; a = \left [ \begin{array}{cccc } 1/6 & 1/4 & 0 & 1/12 \\ 1/2 & 5/12 & 1/3 & 1/4 \\ 1/4 & 1/3 & 5/12 & 1/2 \\ 1/12 & 0 & 1/4 & 1/6 \end{array } \right].\ ] ] the second eigenvalue of @xmath151 is @xmath178 with eigenvector @xmath179 .
if we choose @xmath51 such that @xmath180 then , for sufficiently small but non - zero @xmath181 ( how small depends on how close @xmath20 is to @xmath182 or to @xmath183 ) , the hk - updates will satisfy ( [ eq : distdyn ] ) and the configurations @xmath49 will converge to some translation of @xmath51 without ever actually reaching it .
in this section we will prove that the time taken for the ( directed or undirected ) influence graph to freeze is unbounded as a function of @xmath0 . to simplify some notation
, we will assume the perimeter @xmath20 to be fixed and instead allow @xmath8 to vary in the interval @xmath184 .
[ thm : longfreeze ] for any sufficiently large @xmath0 there is an initial configuration with @xmath0 agents such that the time it takes for the influence graph to freeze can be made arbitrarily large by picking @xmath8 appropriately .
hence , the time until the influence graph freezes is unbounded solely as a function of @xmath0 .
pick positive integers @xmath185 such that @xmath186 and @xmath187 .
this is possible provided @xmath0 is sufficiently large except if @xmath188 . ] . consider a state consisting of five clusters of sizes @xmath189 as illustrated in figure [ fig : slowconv ] .
if @xmath190 then such a configuration will be stable for any choice of @xmath191 .
let @xmath55 denote such a stable state and for @xmath192 sufficiently small positive numbers , let @xmath51 denote the perturbation of @xmath55 as indicated in the figure .
it is easy to see that , as long as every cluster in @xmath49 sees exactly one other cluster on either side , the state at time @xmath67 is the perturbation of @xmath55 by @xmath193 and @xmath194 , defined recursively by @xmath195 note that these sequences are positive , bounded by @xmath196 and tending to @xmath197 as @xmath155 . in particular
, this means that the distance between the adjacent clusters of equal size in @xmath49 is strictly less than , but tending to @xmath198 .
we assume @xmath199 and @xmath200 are sufficiently small such that the distance between any other pairs of clusters remains bounded away from @xmath198 . to complete the proof of the theorem , consider the hk - dynamics with initial state @xmath51 as described above for @xmath8 slightly smaller than @xmath198 . in this case , the influence graph will be constant until the distance between the clusters of equal size is greater than @xmath8 , at which time the model behaves as five clusters on @xmath201 .
the time before this happens can be made arbitrarily long by choosing @xmath8 sufficiently close to @xmath198 .
[ rem : consensus ] suppose we run the hk - dynamics on the real line , starting from five equally spaced clusters of sizes @xmath202 and @xmath203 , reading from left to right , with inter - cluster spacing @xmath204 and @xmath205 , where @xmath206 and @xmath207 .
it is easy to see that any such configuration will eventually reach a consensus .
hence , on the circle , the same will be true of the configuration in theorem [ thm : longfreeze ] for @xmath8 sufficiently close to @xmath198 .
we now show that the various parameters can be chosen so that , on the real line , consensus occurs after three time steps and hence , on the circle , three time steps after the graph disconnects .
first consider the real line again . at @xmath173
, the two extreme clusters will move inwards a distance @xmath208 while the other three clusters will remain fixed .
hence if @xmath209 , which can be satisfied provided @xmath210 , then the middle cluster of size @xmath203 will be visible to all the others at @xmath173 .
hence at @xmath211 we will have clusters of size @xmath212 on either side of the middle cluster , and each at a distance of @xmath213 from it . thus if @xmath214 then the configuration will collapse to a complete consensus at @xmath215 .
one may check that the open interval in ( [ eq : collapse ] ) is non - empty if @xmath216 , where @xmath217 is the unique real root of the equation @xmath218 .
it follows that also on a circle of fixed perimeter , for all @xmath0 sufficiently large and all @xmath73 , there is a choice of @xmath8 such that the configuration in theorem [ thm : longfreeze ] will remain connected up to time @xmath74 and , once it disconnects , will collapse to consensus after three further time steps .
it is natural to ask whether the methods of this paper can be adapted to prove convergence of the hk - dynamics in even more general spaces than the circle .
the case of higher - dimensional tori can be motivated in the usual manner , by imagining that there is more than one issue on which agents have opinions .
generalisation to even more abstract spaces would at present seem to be a purely mathematical curiosity .
one of the main challenges in such work would be to give a mathematically rigorous formulation of the problem .
on the circle , there remain some questions of possible interest .
one could investigate how the time to freezing of the influence digraph behaves for fixed @xmath20 and @xmath8 .
this would seem to depend very subtly on either parameter .
another question that occurred to us but which we can not presently resolve is whether the sequence of influence digraphs @xmath70 resulting from a given initial configuration must always be acyclic - in other words , if @xmath219 for some @xmath220 , must it be the case that @xmath221 for all @xmath222 ?
we thank bernadette charron - bost for pointing out a small error in an earlier version of the manuscript , in which we incorrectly stated that the matrix @xmath151 in ( [ eq : distdyn ] ) was regular .
remark [ rem : irreda ] presents the corrected assertions .
99 a. bhattacharya , m. braverman , b. chazelle and h. l. nguyen , _ on the convergence of the hegselmann - krause system _ , proceedings of the 4th innovations in theoretical computer science conference ( icts 2013 ) , berkeley ca , january 2013 . s. r. etesami and t. basar , _ game - theoretic analysis of the hegselmann - krause model for opinion dynamics _ , to appear in ieee trans .
available online at ` http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7024142 ` r. hegselmann and u. krause , _ opinion dynamics and bounded confidence : models , analysis and simulations _ , journal of artificial societies and social simulation 5 ( 2002 ) , no .
3 . fulltext at ` http://jasss.soc.surrey.ac.uk/5/3/2/2.pdf ` s. martinez , f. bullo , j. cortes and e. frazzoli , _ on synchronous robotic networks - part ii : time complexity of rendezvous and deployment algorithms _ , ieee trans .
control 52 ( 2007 ) , no .
12 , 22142226 .
m. roozbehani , a. megretski and e. frazzoli , _ lyapunov analysis of quadratically symmetric neighborhood consensus algorithms _ , proceedings of the 47th ieee conference on decision and control ( cdc 2008 ) , pp .
22522257 .
s. zhang and z. sun , _ consensus analysis of a @xmath182-dimensional kh model with various confidence level sets _ , proceedings of the second international conference on intelligent computation technology and automation ( icicta 2009 ) , ieee computer society , washington dc ( 2009 ) . | we consider the hegselmann - krause dynamics on a one - dimensional torus and provide the first proof of convergence of this system .
the proof requires only fairly minor modifications of existing methods for proving convergence in euclidean space . |
we describe the four - qubit dfs code , first proposed in the context of providing immunity against collective decoherence processes @xcite .
let @xmath93 and @xmath94 denote the quantum numbers associated with total spin and its projection , and let the singlet and triplet states of two electrons @xmath95 be denoted as @xmath96 here @xmath97 denotes a normalized basis state with the first ( second ) electron in the spin up ( down ) state , etc .
then a single encoded dfs qubit is formed by the two singlets of four spins , i.e. , the two states with zero total spin @xmath98 , where @xmath99 is the pauli spin vector - operator of electron @xmath38 .
these states are formed by combining two singlets of two pairs of spins ( @xmath100 ) , or triplets of two pairs of spins ( @xmath101 ) , with appropriate clebsch - gordan coefficients : @xmath102 the details of the sequences of exchange interactions needed to implement the hadamard , @xmath103 , and controlled - phase gates over this dfs code are too long to give here .
however , they have been well documented in the original works @xcite , and perhaps most concisely in the more recent ref . @xcite .
we performed numerically exact simulations since we required extremely high precision fidelity results , inaccessible via approximation techniques @xcite capable of handling much larger system and bath sizes .
such methods are required for long time scale simulations , which is not our case . moreover ,
large - scale simulations @xcite have confirmed earlier small - bath simulations of cdd pulse sequences @xcite .
our numerically exact simulations ran at 100 - 200 digits of numerical precision , for more than 90 hours on a computer with a dual core intel processor ( 2ghz + 2 g ram ) . in our simulations ,
each logical qubit was encoded using four physical qubits .
we tested for dependence on the initial encoded system state , and found a variation in output fidelity of less than an order of magnitude .
hence all our reported results are for the logical - one initial state . as a further test we considered several simple bath geometries : linear , circular , and polygonal .
our results did not depend appreciably on this geometry .
we took the initial bath state as the uniform superposition , zero temperature state @xmath104 , where @xmath105 is the number of available pure bath states when @xmath106 bath qubits are present .
we checked that finite temperature has only a small quantitative effect on our reported fidelities . in order to keep our numerically exact simulations feasible we had to restrict the total number of qubits to ten , and hence the size of the bath to only two physical qubits when considering two encoded 4-qubit dfs qubits . for consistency we also used two bath qubits in our single encoded qubit simulations .
however , we adjusted the strength of the coupling to the bath to account for this , using a scaling relation we found by testing bath sizes in the range @xmath107 . while we are fully aware of the potential problems associated with using a small bath at long time scales , we expect our small bath simulations to be reliable , as we are concerned only with short timescales , relative to the magnitudes of @xmath44 and @xmath108 , so that possible coherence recurrences due to non - markovian effects are well out of reach .
figures [ fullmemory]-[cphase ] present our complete gate fidelity results for the 4-qubit dfs code subject to free evolution or cdd , with the @xmath108 and @xmath44 parameters each varying over the range from 1hz to 1mhz .
the pulse interval @xmath46 is fixed at 1ns , and the pulse width assumes the values 0 , 1ps , and 1ns . in all cases
cdd leads to a fidelity improvement of many orders of magnitude relative to free evolution .
the plots surrounded by dashed line boxes are the ones shown in fig . 1 in the main text .
we picked the ranges of our @xmath108 and @xmath44 parameters shown in figs .
[ fullmemory]-[cphase ] to represent realistic quantum dot systems see table 1 .
the @xmath44 parameters in this table were estimated directly from the contact hyperfine interaction between a quantum dot electron and nuclear spins . for gaas
the hyperfine constant is @xmath109ev @xcite . for polarized nuclei ,
the interaction strength is reduced by the number of nuclei @xmath106 , typically in the range @xmath110 to @xmath111 per quantum dot , giving @xmath112 . for unpolarized nuclei ,
the interaction strength is reduced by @xmath113 @xcite . for si ( silicon ) , the contact hyperfine strength @xmath114 is computed from @xmath115 , where @xmath116 is the nuclear or electron ( bohr ) magnetic moment , @xmath117 is the fraction of non - zero spin @xmath118si nuclei , @xmath119 is the si nuclear number density , and @xmath120 is the concentration of the electron bloch wavefunction near the si nuclei @xcite . for natural abundance si we obtained @xmath121nev . for polarized nuclei ,
the interaction strength is reduced by the number @xmath122 of @xmath118si in a quantum dot . for unpolarized nuclei the interaction strength is again reduced by @xmath113 .
for both gaas and si the energy scale @xmath108 of the bath is taken to be @xmath123 , the inverse of the nuclear dephasing time . for gaas , the nuclear dephasing time has been estimated at @xmath124s @xcite , which is the precession time of a nucleus in the dipole magnetic field generated by neighboring nuclei . for si
, the nuclear dephasing time has been measured to be @xmath125s @xcite , again corresponding to the evolution time due to the dipole - dipole interaction between a @xmath118si nucleus and neighboring @xmath118si nuclei .
+ , i.e. , the maximum concatenation level before fidelity decreases , as a function of @xmath126 and @xmath127 , in hz .
note that slightly better performance is obtained for the @xmath103 gate .
this is due to the fact that it takes only a single elementary heisenberg exchange operation to implement , while the hadamard gate takes two such operations , i.e. , double the time.,width=307 ] | realizing the theoretical promise of quantum computers will require overcoming decoherence . here
we demonstrate numerically that high fidelity quantum gates are possible within a framework of quantum dynamical decoupling .
orders of magnitude improvement in the fidelities of a universal set of quantum gates , relative to unprotected evolution , is achieved over a broad range of system - environment coupling strengths , using recursively constructed ( concatenated ) dynamical decoupling pulse sequences .
= 1
_ _ introduction.__quantum systems are famously susceptible to interactions with their surrounding environments , a process which leads to a progressive loss of quantumness of these systems , via decoherence @xcite .
when a system performs a quantum information processing ( qip ) task this loss of quantumness is equivalent to the accumulation of computational errors , which leads to the eventual loss of any quantum advantage in information processing .
robust large - scale quantum information processing therefore requires that decoherence or any otherwise undesired evolution of a quantum state
be minimized to the largest extent possible by clever system choice and engineering .
one may then hope to apply the powerful techniques of fault tolerant quantum error correction ( ft - qec ) @xcite .
however , ft - qec imposes significant resource requirements , in particular rapidly growing spatial and temporal overhead , together with demanding gate and memory error rates which must remain below a certain threshold ( e.g. , refs .
@xcite ) .
this motivates the search for alternative strategies which can slow down decoherence and keep quantumness alive .
dynamical decoupling ( dd ) is a form of quantum error _ suppression _ that modifies the system - environment interaction so that its overall effects are very nearly self - canceling , thereby decoupling the system evolution from that of the noise - inducing environment @xcite .
dd has primarily been studied as a specialized control technique for quantum memory ( i.e. , arbitrary state preservation ) @xcite , as convincingly demonstrated by a number of recent experiments in qip platforms as diverse as electron - nuclear systems @xcite , photonics qubits @xcite , and trapped ions @xcite .
however , the holy grail of qip is not just to store states robustly , but rather to perform _ universal computation _
robustly @xcite .
fortunately , there are abstract results showing that dd is in principle compatible with computation , essentially by designing dd operations that commute with the computational operations @xcite .
additionally , recent theoretical results indicate that high fidelity dynamically error - corrected gates can be designed , using methods inspired by dd khodjasteh:080501klv:09 , and that dd can be merged with ft - qec to reduce resource overhead , or even improve gate error rates to below threshold nlp:09 .
dd can also be used to improve the fidelity of adiabatic quantum computation @xcite .
experimentally , dd has been successfully combined with qec in nuclear spin systems to demonstrate robust quantum memory @xcite . in principle , then , it appears that dd is a suitable control technique for overcoming decoherence and improving gate fidelity .
however , a very practical question still remains : what are the conditions under which dd can be used to perform universal quantum computation with a given fidelity ?
recent rigorous bounds devised for the popular `` periodic dd '' ( pdd ) protocol suggest that it is severely limited in this regard @xcite . here
we demonstrate , using numerical simulations of a logical qubit coupled to a small bath , that recursively constructed , concatenated dd ( cdd ) pulse sequences @xcite , can be used to endow a universal set of quantum logic gates with remarkably high fidelities . though our numerical results illustrate the effectiveness of cdd in a model inspired by quantum dot implementations of qip @xcite , the framework we describe is in principle more generally applicable , provided encoding and encoded gates can be implemented .
we therefore expect that our results will contribute to establishing dd as an indispensable tool in scalable qip .
_ _ dynamical decoupling.__the total hamiltonian without dd is @xmath0 , where @xmath1 includes all bath - only terms and @xmath2 includes all terms acting non - trivially on the system . to suppress error
, dd allows the joint evolution to proceed under @xmath3 for some time before applying a control pulse @xmath4 to the system alone [ generated by a time - dependent system - only hamiltonian @xmath5 which is added to @xmath3 ] , designed to refocus the evolution toward the error - free ideal , continually repeating this process until some total evolution has been completed : @xmath7=p_{n}u(\tau _ { 0})\cdots p_{2}u(\tau _ { 0})p_{1}u(\tau _ { 0})\equiv \widetilde{u}(n\tau _ { 0})$ ] , where @xmath8 represents the joint system - bath unitary evolution generated by @xmath3 , for a duration of length @xmath9 ( the pulse interval ) , decomposed so that @xmath10 determines the ideal , desired system - only error - free evolution , and @xmath11 is a unitary error operator acting jointly on the system and bath . here and
below we use a tilde to denote evolution in the presence of dd pulses .
dd schemes with non - uniform pulse intervals have also been considered and shown to be very powerful for quantum memory purposes @xcite , but since it is unclear how to use them for computation involving multiple qubits , we shall limit our discussion to uniform - interval schemes . for now , but not in our simulations presented below ,
we assume for simplicity of presentation that the pulses @xmath12 are sufficiently fast as to not contribute to the total time of the evolution .
the simplest example is quantum memory , where @xmath13 is the identity operation , and @xmath11 represents the deviation from the ideal dynamics caused by the presence of a bath . in this case ,
our goal is to choose pulses so that @xmath7=i_{s}\otimes \widetilde{b}$ ] , where @xmath14 is an arbitrary pure - bath operator .
uniform - interval dd schemes differ in precisely how the pulses @xmath12 are chosen , with the only common constraint that the following basic decoupling condition ( vanishing average hamiltonian , i.e. , vanishing first order term in the magnus series of the joint system - bath evolution ) is met @xcite : @xmath15 . to be concrete
, we will suppose that the pulses @xmath16 are pauli operators .
cdd generates pulse sequences by recursively building on a base sequence @xmath17x[\cdot ] z[\cdot ] x[\cdot ] $ ] ( motivated below ) , where @xmath18 $ ] denotes either free evolution or the insertion of gate operations between pulses .
the sequence is initialized as @xmath19=u(\tau _ { 0})=u_{0}(\tau _ { 0})b(\tau _ { 0})\equiv \widetilde{u}_{0}(\tau _ { 0})$ ] , and higher levels are generated via the rule @xmath20=z[\widetilde{u}_{n}(\tau _ { n})]x[\widetilde{u}_{n}(\tau _ { n})]z[\widetilde{u}_{n}(\tau _ { n})]x[\widetilde{u}_{n}(\tau _ { n})]\equiv \widetilde{u}_{n+1}(\tau _ { n+1})$ ] , where @xmath21 .
note that in contrast to previous work on cdd khodjastehlidar:04khodjastehlidar:07,witzel:07a , nlp:09 , we are allowing for the possibility of some non - trivial information processing operation @xmath10 , as this will be required in our discussion of universal computation below .
the choice of the base sequence is motivated by the observation that it satisfies the decoupling condition , in the quantum memory setting @xmath22 , under the dominant 1-local system - bath coupling term @xmath23 , where @xmath24 , @xmath25 , and @xmath26 denote the pauli matrices acting on system qubit @xmath27 , and @xmath28 are arbitrary bath operators .
( the next order 2-local coupling would have terms such as @xmath29 , etc . ) similarly , the most common pulse sequence used thus far in dd experiments ( e.g. , refs .
@xcite ) is pdd , which generates pulse sequences by _ periodically _ repeating the base sequence @xmath17x[\cdot ] z[\cdot ] x[\cdot ] $ ] : @xmath30=(\operatorname{pdd}_{1}[u(\tau _ { 0})])^{k}=\widetilde{u}_{k}(4k\tau _ { 0})$ ] , where @xmath31=\operatorname{cdd}_{1}[u(\tau _ { 0})]$ ] .
rigorous noise reduction bounds are known for both pdd and cdd in the quantum memory setting , and show that cdd is a much more effective strategy than pdd , provided @xmath32 is sufficiently small , where the norm is the largest eigenvalue @xcite .
it is convenient to characterize the leading - order dd behavior in terms of @xmath33 and @xmath34 , as these parameters capture the strength or overall rate of the internal bath and system - bath dynamics , respectively .
if @xmath35 , then the system - bath coupling is a dominant source of error . in this case , dd should produce significant fidelity gains as it removes the dominant error source . on the other hand , if @xmath36 , then the system - bath coupling induces relatively slow dynamics , while the environment itself has fast internal dynamics . in this case , suppressing the system - bath coupling will have less of an effect on the overall dynamics , so it may be considered a worst case scenario when assessing dd performance .
_ high fidelity universal quantum gates using cdd. _ our main goal in this work is to demonstrate that we can generate a universal set of logic gates which is highly robust in the presence of a decohering environment
. as a model system we consider electron spin qubits in semiconductor quantum dots @xcite , which we study numerically via full - quantum - state ( sometimes called numerically exact ) simulations over a wide range of system - bath coupling parameters . in such systems
the dominant bath is provided by the nuclear spins @xcite , and the interaction between system and environment is described by a heisenberg exchange hamiltonian with exponentially decreasing strength as a function of distance @xmath37 between system qubit @xmath27 and bath qubit @xmath38 .
thus , we let @xmath39 in the system - bath hamiltonian @xmath40 , so that @xmath41 .
we model the interaction between the bath nuclear spin qubits as dipole - dipole coupling , i.e. , @xmath42 so that @xmath43 , where now @xmath37 is the distance between bath qubits @xmath38 and @xmath27 . in our simulations
we pick the parameters @xmath44 , @xmath45 , and @xmath37 , as well as the pulse interval @xmath46 and the pulse width , to include a range of interest for gaas and si quantum dots @xcite .
the @xmath2 and @xmath1 hamiltonians are on during the entire pulse sequence execution , while @xmath5 pulses appropriately between dynamical decoupling ( @xmath47 ) and computational operations ( @xmath48 ) .
universal quantum computation requires that only a discrete set of universal gates be implemented ; a particularly simple choice are the hadamard , @xmath49 , and controlled - phase gates @xcite .
the first two are single - qubit gates , and the third is a two - qubit gate which can be used to generate entanglement .
a conundrum immediately presents itself when trying to combine computation with dd : how to make sure that the dd pulses do not cancel the ( system hamiltonian implementing the ) gates ?
one solution is to use an encoding so that the dd operations commute with the logical gate operations @xcite . to this end
we use logical qubits encoded into a four - qubit decoherence - free subspace ( dfs ) .
the logical basis states are the two orthonormal total spin - zero states of four spin-@xmath50 particles , first described as a dfs in ref .
@xcite .
we stress that our system - bath interaction does not exhibit any symmetries so that there is no naturally occuring dfs which can be used to store protected quantum information ; instead , our encoding choice is motivated by the fact that in this setting , a universal set of encoded computational operations can be generated by controllable heisenberg exchange hamiltonians between the system qubits ( @xmath51 ) , as first described in @xcite , and these commute with the global pauli operations @xmath52 used as decoupling pulses . to generate these pulses @xmath53 is modeled as a controllable uniform magnetic field
. however , we emphasize that these choices are by no means unique .
any choice of dd pulses @xmath12 such that @xmath54=0 $ ] @xmath55 will suffice @xcite , including , e.g. , the stabilizer quantum error correcting codes relevant in the theory of ft - qec used as dd pulses , and the normalizers of these codes used to generate computational gates @xcite . faced with several options for combining dd and computational operations @xcite , we chose the following decouple while computestrategy .
in this strategy we alternate between applying computational and dd operations , thus spreading a computational gate over the entire cdd pulse sequence .
we do this by applying the @xmath56 root of the gate @xmath57 times during a cdd pulse sequence involving @xmath58 pulses .
thus , if the ideal computational gate is @xmath59 , we implement it by applying @xmath60 between each of the @xmath57 pulses , where @xmath61 . as @xmath62 increases with concatenation level @xmath63 , the exchange couplings in @xmath64 are proportionately decreased so that the @xmath65 product remains constant ; in all simulations the pulse interval @xmath46 is held fixed at 1ns , the time - scale for exchange operations in semiconductor quantum dots @xcite .
this `` decouple while compute '' strategy is precisely the formulation presented in the expressions for @xmath20 $ ] and @xmath30 $ ] above , provided we identify @xmath66 there with @xmath67 here , and @xmath10 there with @xmath68 here .
other strategies are certainly also conceivable , e.g. , a decouple then compute strategy wherein @xmath10 is simply the identity operation , and the gate is implemented at the end of the pulse sequence . while the latter strategy was shown to be capable of reducing the resource requirements of ft - qec nlp:09 , we found in our simulations that we obtain a higher fidelity when we use decouple while compute , because then time is not wasted on free evolution during the intervals between pulses .
khz and @xmath69mhz , respectively ( we work in units of @xmath70 ) .
pulse intervals are fixed at @xmath71ns , while pulse widths are given by @xmath72 , @xmath73ps , and @xmath73ns , corresponding , from bottom to top , to the blue , green ( absent for controlled - phase ) , and magenta lines , respectively .
the red dashed line shows the unprotected evolution over a time period @xmath74 .
notice that the @xmath75 ranges change between plots . also , @xmath76 corresponds to free evolution for a duration @xmath71ns , whence the @xmath76 point starts at a relatively high fidelity .
results depend only slightly on the choice of initial system state.,width=336 ] we now present our simulation results ( details of the numerical procedure are given in @xcite ) . the worst - case scenario of @xmath77 is shown in fig
. [ ugatesfig ] , where we plot @xmath75 vs. concatenation level for each of the universal gates , with the fidelity defined as @xmath78 , where @xmath79 is the mixed output system state ( obtained from the joint system - bath evolution after partial trace over the bath ) and @xmath80 is the desired system state . in each of these plots , the red dashed line represents undecoupled free evolution for increasing total time , given by @xmath81 . as the evolution time increases , error accumulates and fidelity correspondingly worsens , while @xmath82 combats this effect with each successive level of concatenation .
to contrast , the blue line in these graphs shows @xmath82 with ideal , zero - width dd pulses , so that realistic , finite - width dd pulses lie somewhere between the blue and red lines , as shown . in each of the plots in fig .
[ ugatesfig ] cdd achieves impressive results , even when pulse widths are as long as the intervals , that is , when @xmath83ns as depicted with the magenta lines , cdd still manages more than five orders of magnitude improvement in fidelity over free evolution . as
the pulse width @xmath84 narrows relative to the pulse interval @xmath9 , that is , as the dd pulses becomes faster , fidelity improvement grows to between ten and twenty orders of magnitude over free evolution . the results for the encoded @xmath49 and hadamard gates are similar , which is not surprising given that they require , respectively , one and two elementary heisenberg exchange operations to be implemented @xcite .
the fidelity of the controlled - phase gate is several orders of magnitude lower , which is due to the fact that it involves a much longer sequence of @xmath85 elementary heisenberg operations bacon : thesiswoodworthmizellidar:05 . finally , while the quantum memory results are comparable to those of the hadamard and @xmath49-gates , we attribute the reduction in memory fidelity at the highest concatenation level to the absence of @xmath86 during the intervals between pulses
indeed , having the system hamiltonian on during the pulse intervals has a beneficial effect , as it effectively reduces the strength of the bath and system - bath hamiltonians . the overall conclusion from fig .
[ ugatesfig ] is rather encouraging : it appears to be possible to implement a universal set of quantum logic gates with a high fidelity in the presence of coupling to a spin bath . the results in fig .
[ ugatesfig ] are for specific coupling parameters chosen deliberately to represent a worst - case scenario for dd , in that @xmath77 .
as we next demonstrate , the conclusions are robust : cdd remains effective over a broad range of bath dynamics and system - bath coupling strengths .
figure [ contours ] shows the resilience of cdd to widely varying environments by displaying constant fidelity contours in @xmath87 space , at fixed concatenation level and pulse width , as indicated .
note that fixing @xmath63 and @xmath84 renders the total evolution time constant , so that fidelity becomes strictly a function of the dimensionless coupling parameters @xmath88 .
these plots show a strong fidelity dependence on @xmath89 , and a very weak dependence on @xmath90 , except in the quantum memory case .
more generally , our results show that cdd is effective over a broad range of coupling parameters , including the fundamentally different good ( @xmath91 ) and bad ( @xmath77 ) regimes .
this conclusion is further bolstered by our complete gate fidelity simulations @xcite , where the @xmath45 and @xmath44 parameters each vary over the range from @xmath92hz to @xmath92mhz . in these simulations
the non - memory gate fidelities improve monotonically as a function of concatenation level for all values of @xmath44 and @xmath45 . taken in their totality ,
our simulation results indicate that universal quantum computation can be combined with cdd to achieve very high fidelities . _
_ discussion.__the high gate fidelities we have reported here suggest that it is advantageous to incorporate cdd as a first layer of defense against decoherence , in a more complete ft - qec scheme . in this
regard our decouple while compute study complements the decouple then computestrategy for which it has already been shown that incorporating dd , and in particular cdd , into ft - qec can lead to substantial improvements nlp:09 .
it will be interesting to determine which strategy leads to better performance overall .
undoubtedly , optimal control methods will offer important additional performance improvements in the implementation of encoded logic gates @xcite .
we look forward to experimental tests of cdd - protected quantum memory and logic gates .
the views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies , either expressly or implied , of the united states department of defense or the u.s .
government . approved for public release , distribution unlimited . _
acknowledgments_. all authors were sponsored by the united states department of defense .
d.a.l .
was also sponsored by nsf under grants no .
chm-924318 , ccf-726439 and phy-803304 . 34 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . ,
_ _ ( , , ) . , , , * * , ( ) ; , * * , ( ) .
, , , * * , ( ) ; , * * , ( ) . , * * , ( ) . ,
* * , ( ) ; , * * , ( ) . , * * , ( ) . ,
* * , ( ) . ,
* * , ( ) . ,
* * , ( ) .
, , , * * , ( ) . , phys .
rev .
a * 82 * , 012301 ( 2010 ) . , * * , ( ) ; , * * , ( ) . , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , _ _ ( , , ) . , , , * * , ( ) . ,
* * , ( ) ; , , , * * , ( ) . , , , . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) ; , * * , ( ) .
see appendix . , * * , ( ) . , * * , ( ) . , * * , ( ) ; , * * , ( ) . , ph.d .
thesis , ( ) , ; , , , * * ( ) .
, * * , ( ) ; m. grace _ et al .
_ , j. phys .
b. * 40 * , s103 ( 2007 ) ; p. rebentrost _
et al_. , phys .
rev .
lett . * 102 * , 090401 ( 2009 ) . , , , ,
* * , ( ) . , , , , *
* , ( ) . ,
ph.d .
thesis , ( ) , .
, , , * * ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , , ,
* * , ( ) . |
we are aiming to describe the response of thermo - visco - elastic material to applied external forces and the heat flux through the boundary .
the system of equations capturing the displacement , temperature and visco - elastic strain of the body is a consequence of physical principles such as balance of momentum and balance of energy , cf .
@xcite , see also @xcite .
the equations are complemented by the constitutive relation for the cauchy stress tensor and the constitutive equation for the evolution of the visco - elastic strain tensor .
although we treat the case where the thermal expansion is negligible , but the changes of temperature affect the visco - elastic properties of the considered material . we shall observe it in the appearance of the temperature - dependent constitutive relation in the evolution equation for the visco - elastic strain tensor .
we assume that the body @xmath0 is an open bounded set with a @xmath1 boundary and moreover , the body is homogeneous in space .
the material undergoes two kinds of deformations : elastic and visco - elastic . by the first type we understand the deformations which are reversible and the second ones are irreversible .
the problem is captured by the following system @xmath2 which describes the quasi - static evolution of the displacement of the material @xmath3 , the temperature of the material @xmath4 and the visco - elastic strain tensor @xmath5 .
we denote by @xmath6 the set of symmetric @xmath7-matrices with real entries and by @xmath8 a subset of @xmath6 which contains traceless matrices .
by @xmath9 we mean the deviatoric part ( traceless ) of the tensor @xmath10 , i.e. @xmath11 , where @xmath12 is the identity matrix from @xmath6 .
additionally , @xmath13 denotes the symmetric part of the gradient of the displacement @xmath14 , i.e. @xmath15 . the volume force is denoted by @xmath16 .
the visco - elastic strain tensor is described by the evolutionary equation with prescribed constitutive function @xmath17 .
the function @xmath17 is assumed to be monotone and to satisfy polynomial growth and coercivity conditions .
the function @xmath18 is continuous with respect to @xmath19 and @xmath9 and satisfies for @xmath20 the following conditions : * @xmath21 , for all @xmath22 and @xmath23 ; *
@xmath24 , where @xmath25 , @xmath23 ; * @xmath26 , where @xmath25 , @xmath23 , [ ass_g ] where @xmath27 and @xmath28 are positive constants , independent of the temperature @xmath19 .
we complete the considered problem by formulating the initial conditions @xmath29 in @xmath30 and boundary conditions @xmath31 on @xmath32 .
the properties of the material under consideration determine the choice of the function @xmath33 .
such a framework includes the classical norton - hoff model , cf .
@xcite , which we shall briefly discuss in a sequel .
there are various different relations considered , e.g. * bodner - partom model @xcite : @xmath34 where @xmath35 describes the isotropic hardening of the metal , @xmath36 stands for the positive part of @xmath37 , @xmath38 and @xmath39 are given functions and @xmath40 is a positive constant .
moreover , functions @xmath41 , @xmath42 , @xmath43 and @xmath44 fulfill some specific properties .
* mrz model @xcite : @xmath45 where @xmath46 is a continuous function . * prandtl - reuss model with linear kinematic hardening @xcite @xmath47 where @xmath48 is the indicator function of the closed and convex subset @xmath49 and @xmath50 are material parameters .
furthermore , @xmath51 is a subdifferential of the function @xmath48 . for further examples of constitutive relations ( e.g. classical maxwell model , models proposed by chaboche , hart , miler , bruhns and many others ) we refer to (
* chapter 2.2 ) .
our motivation for current considerations were the results of alber and chemiski @xcite and of hmberg @xcite . in @xcite
the authors considered the quasi - static visco - elasticity models with norton - hoff constitutive function , namely of the power - law type @xmath52 with @xmath53 .
the parameter @xmath54 was either assumed to be a positive constant or dependent on an additional relaxation parameter described by a separate equation .
the scheme of the proof in @xcite was to formulate the problem in a way that it fits to the abstract theory of maximal monotone operators , cf .
@xcite . in the current paper
we include the thermal effects of the process through the dependence of the constitutive function @xmath33 on the temperature .
this dependence obstructs following the same scheme and requires different approach .
furthermore , we assume that @xmath55 depends only on the deviatoric part of the cauchy stress tensor and its range is the set of traceless matrices .
the last assumption , together with the fact that @xmath56 is traceless , provides that also @xmath57 is traceless .
vanishing of the trace of the deformation tensor corresponds to preserving the volume of the material .
indeed , the volume change is associated only with the elastic response of the material , and the plastic response is essentially incompressible , cf .
the dependence of @xmath55 only on @xmath9 is essential to maintain the coercivity of the model .
once we know that the range of @xmath33 is @xmath58 , then even for the isothermal process , namely the case of @xmath59 we observe that @xmath60 .
then e.g. taking as @xmath10 the identity matrix we immediately see that @xmath61 .
let us now comment on the technical consequences of this assumption .
contrary to the proof of alber and chemiski , where they showed that @xmath10 belongs to @xmath62 for @xmath63 , the estimates conducted in the current situation provide only that @xmath10 belongs to @xmath64 .
hmberg in @xcite considered more general physical phenomena including the electro - magnetic effects .
the changes of temperature influenced the concentration of different phases of materials and this dependence was prescribed by some general operator @xmath65 $ ] having _ good _ properties . then the constitutive function describing the evolution of visco - elastic strain depends no more on the temperature , but on these concentrations .
moreover it is linear with respect to the deviatoric part of the cauchy stress tensor , namely corresponds to the mrz model .
the similarities with our approach are related with the construction of the approximated problem , namely by the truncation of the terms which appear on the right - hand side of the heat equation and are only integrable .
the method also follows the framework of boccardo and gallout . nevertheless , because of the different structure of the problem
, hmberg can show the strong convergence of the approximated sequence of the cauchy stress tensor .
for the concept of showing this strong convergence observe that in the case of linear mrz relation , and in fact also in the case of norton - hoff relation , the stronger condition than monotonicity holds , namely the uniform monotonicity condition @xmath66 for the proof see e.g. @xcite .
the studies on the mrz model presented in @xcite essentially used the strong monotonicity of the function @xmath33 in the second variable . the existence proof used the methods developed in @xcite arising from the tools of young measures . in the present setting none of the assumptions of strong nor uniform monotonicity are needed .
we only assume monotonicity of @xmath33 .
following bartczak @xcite , chemiski @xcite , chemiski and racke @xcite , duvaut and j.l .
lions @xcite , johnson @xcite , neas and hlavek @xcite , suquet @xcite , temam @xcite and many others , we study the quasi - static evolution , i.e. the evolution , which is slow and we neglect the acceleration term in the equation for balance of momentum . moreover , we consider the model with infinitesimal displacement . in a consequence ,
the dependence between the cauchy stress tensor and the symmetric gradient of displacement is linear ( generalized hooke s law , for more details see @xcite or @xcite ) .
much of the approaches involve the models that are purely mechanical , namely concern the theory of inelastic and infinitesimal deformations with the nonlinear inelastic constitutive relation of monotone type , however neglect all thermal influences , see @xcite and also @xcite . on the other hand ,
the mathematical analysis of linear thermo - elasticity is also a classical , well understood topic , cf .
@xcite , contrary to an analysis of thermo - inelastic models . by the thermo - inelastic models we mean the systems consisting of balance of momentum for kind of inelastic deformation and the equation for an evolution of the temperature . in the equation for balance of momentum for inelastic deformation
the stress is not proportional to the strain , i.e. there appear term which absorbs the mechanical energy .
there are only some results for special models or for simplified models in the literature @xcite .
if we introduce thermal effects into various purely mechanical models , then the right hand side of the heat equation ( the product @xmath67 ) turns out to be only an integrable function .
in such a case the standard energy methods fail and one needs to search for more delicate tools . using the boccardo and gallout @xcite approach to prove the existence of solutions to the heat equation the essential point is to use the truncation of the solution as a test function .
this is however difficult to combine with a classical galerkin method as the truncation of a function may no longer be a linear combination of the functions from the galerkin basis .
therefore we appeal to non - standard energy methods , such as two - level galerkin approximation , see also @xcite . the new difficulty which arises here is the construction of the appropriate basis for approximation of the strain tensor @xmath57 , for details see appendix [ b ] .
all functions appearing in this paper are the functions of position @xmath68 and time @xmath69 .
we often omit the variables of the function and write @xmath14 instead of @xmath70 .
all of the computation are conducted in lagrangian coordinates . in view of the fact that the displacement is small , the stress tensor in the lagrangian coordinates
is approximated by the stress tensor in eulerian coordinates .
this is a standard way of considering the inelastic models , for more details see ( * ? ? ?
* chapter 13.2 ) .
before we formulate the definition of weak solutions and state the main result of the paper let us introduce the notation @xmath71 .
let @xmath63 , @xmath72 and @xmath73 be large enough .
the triple of functions @xmath74 and @xmath75,w^{-s,2}(\omega ) ) \nonumber\ ] ] is a weak solution to the system if @xmath76 where @xmath77 and @xmath78 holds for every test function @xmath79,c^{\infty}_c(\omega,\mathbb{r}^3))$ ] and @xmath80 .
furthermore , the visco - elastic strain tensor can be recovered from the equation on its evolution , i.e. @xmath81 for a.e . @xmath82 and @xmath83 .
moreover , @xmath84 .
let @xmath63 and let initial conditions satisfy @xmath85 , @xmath86 , boundary conditions satisfy @xmath87 , @xmath88 and volume force @xmath89 and function @xmath17 satisfy the assumption [ ass_g ] .
then there exists a weak solution to system .
[ thm : main2 ] there is nothing about the uniqueness of solutions in theorem [ thm : main2 ] . using boccardo and gallout approach to the heat equation
we obtain the existence of @xmath19 only in the space @xmath90 for all @xmath72 , see the appendix .
the lack of uniqueness of the temperature implies the lack of uniqueness of the solution to the whole system . in order to find the class of solutions providing both existence and uniqueness one should consider the renormalized solutions to the heat equation ,
see @xcite .
however , the existing theory concerns mostly the homogeneous dirichlet boundary - value problems . the rest of the paper is organized as follows : section [ druga ] is mostly dedicated to physical aspects of the problem .
therefore in section [ sec : model ] we introduce the complete model and present the assumptions which brought us to the simplified setting . then in section [ comp ] we concentrate on physical justification of the model after simplifications .
section [ sec:3 ] is only a technical part that prepares us to the proof of the main theorem , namely we transform the problem to a homogeneous boundary - value problem .
the whole section [ sec : proof ] is devoted to the proof of theorem [ thm : main2 ] .
the subsequent subsections correspond to the steps of the proof such as existence of the approximate solutions , boundedness of the approximate solution and the behaviour of the energy of the system .
finally we pass to the limit in the galerkin approximations .
we complete the introduction by introducing the notation . as a result of integration @xmath91
we write @xmath92 which is equal to @xmath93 . furthermore ,
we denote by @xmath94 standard lebesgue spaces , for @xmath95 and @xmath96 , by @xmath97 the sobolev spaces , by @xmath98 the fractional order sobolev space and by @xmath99 bochner spaces , by @xmath100 continuous functions on @xmath101 , by @xmath102 compactly supported smooth functions on @xmath101 .
we will start the current section with formulating the full system describing the evolution of visco - elastic body including thermal effects .
subsequently we describe the assumptions that were made due to simplify the system and motivate considering equations .
the second part concerns the issue of thermodynamical completeness of the considered system .
this part essentially follows @xcite .
however , since this is an important argument for choosing this model , we include the main steps for completeness . in the last subsection
we include the technical step which allows to reduce the problem to homogeneous boundary - value problem .
let us consider the system of equations in the bounded domain @xmath103 with a @xmath1 boundary @xmath104 @xmath105 derivation of this system was presented in @xcite , @xcite and @xcite .
the equation describes the balance of momentum .
equations and prescribe the constitutive relation for the cauchy stress tensor and presents the constitutive relation for the evolution of the visco - elastic strain tensor .
finally , stands for the balance of energy .
the function @xmath106 is the cauchy stress tensor .
the cauchy stress tensor can be divided into two parts : mechanical and thermal .
the mechanical part is @xmath107 , where the operator @xmath108 is linear , positively definite and bounded . assuming that @xmath30 is a homogeneous material , the operator @xmath109 is a four - index matrix , i.e. @xmath110 and the following equalities hold @xmath111 the evolution of the visco - elastic strain tensor @xmath57 is governed by the constitutive relation @xmath112 .
the visco - elastic strain tensor @xmath113 is traceless if @xmath114 is traceless .
the temperature @xmath115 is the reference temperature .
the function @xmath116 describes a given density of heat sources , @xmath117 is the material s conductivity , which in the case of homogeneous materials is a positive constant , @xmath118 is the constant density of the body .
moreover , @xmath119 describes the thermal expansion of the body .
we will study the simplified situation , namely under the following assumptions we consider only the problem with small inertial force , i.e. @xmath120 .
we assume that @xmath121 , i.e. the considered material is not subject to the thermal expansion .
the fact of neglecting the acceleration term implies that the system of equations may be supplemented only by the initial conditions .
moreover , we complete the system with boundary conditions . using
the dirichlet condition for the displacement means that we control the shape of the body , and by using the neumann condition for the temperature we control the flow of the energy through the boundary .
there are various simplifications that are proposed due to provide the mathematical analysis of the system . in the linear thermoelasticity ,
the term connected with thermal expansion in the heat equation is approximated by a linear one , i.e. @xmath122 with the argumentation that the temperature @xmath19 in the considered process is close to the reference temperature , cf .
bartczak @xcite , chemiski and racke @xcite . from the point of view of mathematical techniques used in the linear theory , such as e.g. linear semigroup theory
, this approximation seems to be accurate .
unfortunately , in a consequence of this procedure one obtains the model which is not consistent with the physical principles .
our simplification follows different way , we consider the case where no thermal expansion appears , hence @xmath121 . in the proceeding section
we discuss in detail the issue of thermodynamical completeness of the system after the simplifications .
finally , we also formulate the last assumption .
we assume that there are no heat sources in the system , hence @xmath123 .
the material s conductivity @xmath124 is for simplicity equal to @xmath125 .
taking into account the above conditions we obtain the considered system .
the purpose of the current section is to underline the physical advantages of the considered system .
the assumptions used in the construction of the simplified model do not effect the loss of physical properties , i.e. the system still conserves the energy , the temperature is positive and there exists a function of state , namely the entropy , which has a positive rate of production .
we shall say that the system is thermodynamically complete if these properties are satisfied . in @xcite
we showed the thermodynamical completeness of the system in the case it is isolated , i.e. @xmath126 , we assume homogeneous boundary values and there are no heat sources ( @xmath127 ) .
all of the calculation in this section are formal .
_ conservation of total energy _ in the first step
we intend to show that the global energy is preserved . multiplying the first equation of system by @xmath128 and integrating over an arbitrary set @xmath129
, we obtain @xmath130 and hence @xmath131 we multiply the evolutionary equation for the visco - elastic strain by @xmath10 and integrate over @xmath132 . subtracting
this equation from implies that @xmath133 finally , using the symmetry of @xmath10 we obtain @xmath134 since the global energy of the set @xmath132 is equal to @xmath135 and the density of the total energy is defined by @xmath136 , we obtain @xmath137 consequently equation may be written in the following from @xmath138 using @xmath139 , we obtain @xmath140 zero external forces , homogeneous boundary conditions and no heat sources implies that @xmath141 and @xmath142 on the boundary @xmath143 .
therefore , the global energy @xmath144 is constant in time .
_ positivity of the temperature _
let us assume that the initial temperature @xmath145 is positive .
the heat equation after simplifications has a form @xmath146 hence , the assumptions on the function @xmath17 imply that the right hand side of is positive , namely @xmath147 when the initial and boundary conditions for the temperature are positive , then the temperature @xmath19 is positive . _
entropy inequality _
multiplying by @xmath148 and integrating over an arbitrary set @xmath129 , we obtain @xmath149 thus @xmath150 by the properties of the function @xmath17 and positivity of @xmath19 , the right hand side of is positive .
therefore , an arbitrary choice of the domain @xmath132 implies that the inequality holds @xmath151 the above relation is the so - called clausius - duhem inequality and it is one of the equivalent formulations of the second principle of thermodynamics .
hence , the homogeneous boundary conditions and the definition of the heat flux ( @xmath152 ) implies that @xmath153 note that @xmath154 is one of the admissible entropies for system what furnishes a formal justification for the thermodynamical completeness of the model .
for the situation with linearization of the term @xmath155 one can show that none of the thermodynamical principles is fulfilled .
our aim is to reduce the problem to a homogeneous one . for this purpose
we are interested in a decoupled elastic systems and a heat equation .
the first system is subject to the same external forces as problem and both of the problems are complemented with the same boundary conditions as .
hence , given @xmath156 we study @xmath157 and @xmath158 let @xmath159 , @xmath160 , @xmath161 and moreover @xmath89 .
then there exists a solution to systems and .
additionally , the following estimates hold : @xmath162 [ wyrzucenie_war_brzeg ] moreover , @xmath19 belongs to @xmath163,l^2(\omega))$ ] . from the trace theorem ( * ? ? ?
* chapter ii ) there exist @xmath164 such that @xmath165
. then , finding the solution @xmath166 to is equivalent to finding the solution @xmath167 to the following problem @xmath168 and @xmath169 . using ( * ? ? ?
* corollary 4.4 ) , we obtain the estimates presented in lemma [ wyrzucenie_war_brzeg ] . instead of finding @xmath170 the solution to problem
-- we shall search for @xmath171 , where @xmath172 and @xmath173 and @xmath174 solve with @xmath175 and .
furthermore , we get @xmath176 hence , we consider the problem @xmath177 with the initial and boundary conditions @xmath178 where @xmath179 is the given initial condition for the temperature and @xmath180 is the initial condition for the system . from the proof provided in section [ sec : proof ] it follows that the displacement @xmath181 , which is a solution to the homogeneous problem belongs to the space @xmath163,l^{p'}(\omega))$ ] .
however , in theorem [ thm : main2 ] the information on the continuity of the solution to the nonhomogeneous problem does not appear .
this is the consequence of the the fact that @xmath182 may fail to be continuous under the assumptions that we have for the volume force @xmath183 and boundary data .
we are facing the problem of low regularity of the right hand side of the heat equation and the initial condition .
both functions are only integrable what enforces using some delicate methods , such as the approach of boccardo and gallout , cf .
@xcite , for showing the existence of solutions .
an essential step is testing the equation with the truncation of solution .
however , this truncation need not to be a linear combination of basis functions .
this is the reason why we use two level approximation , i.e. independent parameters of approximation in the displacement and temperature .
we pass to the limit , firstly with parameter @xmath187 corresponding to the dimension of the galerkin basis for the temperature to get the sequence of infinite dimensional approximate solutions .
passing to the limit with parameter @xmath188 corresponding to the dimension of the galerkin basis for the displacement requires closer attention .
we construct the approximated system using the galerkin method .
consider the space @xmath189 with a scalar product defined @xmath190 where @xmath191 .
let @xmath192 be the set of eigenfunctions of the operator @xmath193 with the domain @xmath194 and @xmath195 be the corresponding eigenvalues such that @xmath196 is orthogonal in @xmath197 with the inner product @xmath198 and orthonormal in @xmath199 .
hence @xmath200 using the eigenvalue problem for the operator @xmath193 we obtain @xmath201 moreover , let @xmath202 be the set of eigenfunctions of the laplace operator with the domain @xmath203 , let @xmath204 be the set of corresponding eigenvalues , let @xmath205 be orthogonal in @xmath206 and orthonormal in @xmath207 .
these two families of vectors shall be used to construct the finite dimensional approximations of the displacement and the temperature . to construct the basis for approximating the visco - elastic strain tensor we will proceed as follows .
let us consider the symmetric gradients of first @xmath188 functions from the basis @xmath208 .
due to the regularity of the eigenfunctions we observe that @xmath209 are elements of @xmath210 , namely the fractional sobolev space with a scalar product denoted by @xmath211 and @xmath212 .
define now @xmath213 which is the orthogonal complement in @xmath189 taken with respect to the scalar product @xmath214 and also @xmath215 let @xmath216 denote the orthonormal basis of @xmath217 , which is also an orthogonal basis of @xmath218 , for more details see appendix [ b ] . for @xmath219 , we are ready to define @xmath220 such that @xmath221 , @xmath222 and @xmath223 solve the system of equations @xmath224 \ten{t}_{k , l } & = \ten{d}(\ten{\varepsilon}(\vc{u}_{k , l } ) - \ten{\varepsilon}^{\bf p}_{k , l } ) , \\[1ex ] \int_{\omega}(\ten{\varepsilon}^{\bf p}_{k , l})_t : \ten{d}\ten{\varepsilon}(\vc{w}_n ) { \,{\rm{d}}x}&= \int_{\omega}\ten{g}(\theta_{k , l } + \tilde{\theta } , \ten{t}^d_{k , l } + \tilde{\ten{t}}^d ) : \ten{d}\ten{\varepsilon}(\vc{w}_n ) { \,{\rm{d}}x } & n=1, ... ,k , \\[1ex ] \int_{\omega}(\ten{\varepsilon}^{\bf p}_{k , l})_t
: \ten{d}\ten{\zeta}^k_m { \,{\rm{d}}x}&= \int_{\omega}\ten{g}(\theta_{k , l } + \tilde{\theta } , \ten{t}^d_{k , l } + \tilde{\ten{t}}^d ) : \ten{d}\ten{\zeta}^k_m { \,{\rm{d}}x } & m=1, ... ,l , \\[1ex ] \int_{\omega}(\theta_{k , l})_t v_m{\,{\rm{d}}x}+ \int_{\omega}\nabla\theta_{k , l}\cdot\nabla v_m { \,{\rm{d}}x } & \\[1ex ] = \int_{\omega } \mathcal{t}_k ( ( \ten{t}_{k , l}^d + \tilde{\ten{t}}^d ): & \ten{g}(\theta_{k , l } + \tilde{\theta } , \ten{t}^d_{k , l } + \tilde{\ten{t}}^d ) ) v_m { \,{\rm{d}}x } & m=1, ... ,l .
\end{array } \label{app_system}\ ] ] for a.a .
@xmath225 . for each approximate equation
we have the initial conditions in the following form @xmath226 \left ( \ten{\varepsilon}^{\bf p}_{k , l}(x,0 ) , \ten{\varepsilon}(\vc{w}_n ) \right)_{\ten{d } } & = & \left(\ten{\varepsilon}^{\bf p}_0 , \ten{\varepsilon}(\vc{w}_n ) \right)_{\ten{d } } & n=1, .. ,k , \\[1ex ] \left ( \ten{\varepsilon}^{\bf p}_{k , l}(x,0 ) , \ten{\zeta}_m^k ) \right)_{\ten{d } } & = & \left(\ten{\varepsilon}^{\bf p}_0 , \ten{\zeta}^k_m
\right)_{\ten{d } } & m=1, .. ,l , \end{array } \right . \label{eq : warunki_pocz_app}\ ] ] where @xmath227 denotes the inner product in @xmath207 and @xmath228 the inner product in @xmath189 .
let us define @xmath229 the selection of the galerkin bases and representation of the approximate solution allows to notice that @xmath230 and hence we obtain @xmath231 for @xmath232 and @xmath233 , where @xmath234 hence @xmath235 system with initial conditions can be equivalently written as the initial value problem @xmath236 where @xmath237 is a vector of initial conditions obtained from . for @xmath238 , we get @xmath239 , hence @xmath240 can be treated as a function only of @xmath241 , i.e. @xmath242 . according to carathodory theorem , see ( * ? ? ?
* theorem 3.4 ) or ( * ? ? ?
* appendix @xmath244 ) , there exist unique absolutely continuous functions @xmath245 , @xmath246 and @xmath247 for every @xmath248 and @xmath249 on some time interval @xmath250 $ ] .
moreover for every @xmath248 there exists a unique absolutely continuous function @xmath251 . in this section
we show the uniform boundedness of approximate solutions .
as the considered model describes the physical phenomena , then it is obvious that the total energy should be finite .
the total energy of the system consists of potential energy and thermal energy .
there exists a constant @xmath27 which is uniform with respect to @xmath188 and @xmath187 such that @xmath254 } \mathcal{e}(\ten{\varepsilon}(\vc{u}_{k , l } ) , \ten{\varepsilon}^{\bf p}_{k , l } ) ( t ) + c \|\ten{t}_{k , l}^d + \tilde{\ten{t}}^d\|^p_{l^p(0,t , l^p(\omega ) ) } \leq c.\ ] ] [ pom_2 ] the potential energy is an absolutely continuous function and calculating the time derivative of @xmath255 we get for a.a .
@xmath256 $ ] @xmath257 in the first step we multiply @xmath258 by @xmath259 for each @xmath238 . summing over @xmath232 we obtain @xmath260 in the second step we multiply @xmath261 by @xmath262 and summing over @xmath233
, we obtain the identity , which is equivalent to @xmath263 thus @xmath264 using assumption 1c and the young inequality we get @xmath265 where @xmath266 , with a constant @xmath27 coming from assumption 1b .
hence we estimate the last term as follows @xmath267 finally , integrating over @xmath268 , with @xmath269 we obtain @xmath270 from we immediately observe that the sequence @xmath271 is uniformly bounded in the space @xmath62 with respect to @xmath188 and @xmath187 . additionally , combining and we conclude the uniform boundedness of the sequence @xmath272 in the space @xmath273 and hence the uniform boundedness of the sequence @xmath274 in @xmath275 .
[ wsp_ogr_t ] let @xmath280 be a projection on @xmath281 , @xmath282 , then @xmath283 .
let @xmath284 be a projection on @xmath285 , @xmath286 .
since @xmath284 is the projection of a finite dimensional space , and the dimension of the space is independent of @xmath187 , there exists a constant , also independent of @xmath187 such that @xmath287 let @xmath288 and we may estimate as follows @xmath289 where the equality results from orthogonality of subspaces @xmath290 and @xmath291 .
then @xmath292 where @xmath293 is an optimal embedding constant of @xmath294 .
consequently , there exists @xmath295 such that latexmath:[\[\sup_{\varphi\in l^p(0,t , h^{s}(\omega))\atop \|\varphi\|_{l^p(0,t , h^{s}(\omega))}\le 1 } \int_0^t since it can be immediately observed that @xmath300 and lemma [ pom_2 ] holds , we omit the details of the proof .
the lemma provides that the internal energy of @xmath30 is finite at any time @xmath301 $ ] .
it is possible to prove better estimates for the temperature , however they are uniform only with respect to @xmath187 and not with respect to @xmath188 .
we provide the details in the proceeding lemma .
the uniform boundedness of solutions ( lemma [ pom_2 ] and lemma [ lm:7 ] ) implies the global existence of approximate solutions , i.e. existence of solutions @xmath305 on the whole time interval @xmath306 $ ] for each @xmath232 and @xmath233 . moreover , there exist global solutions @xmath307 for all @xmath232 .
before we pass to the limit let us multiply the system by smooth time - dependent functions , integrate over @xmath306 $ ] and then rewrite the system as follows @xmath309 @xmath310 and for @xmath311 @xmath312 holds for every test functions @xmath313)$ ] and @xmath314 .
firstly , we pass to the limit with @xmath315 - the galerkin approximation of temperature . from the previous section we get uniform boundedness with respect to @xmath187 for appropriate sequences . then at least for a subsequence , but still denoted by the index @xmath187 , we get the following convergences passing now to the limit in - yields @xmath317 @xmath318 holds for every test functions @xmath313)$ ] . by the density of @xmath319 in @xmath320
we conclude that @xmath321 holds for all @xmath322,l^{p}(\omega,\mathcal{s}^3))$ ] and then also for all @xmath323 . in the rest of this section
we identify the weak limit of the nonlinear term @xmath324 and then show the convergence of @xmath325 what shall allow to pass to the limit in . in view of lemma [ pom_2 ] the sequence @xmath333
is uniformly bounded in @xmath64 .
using the triangle inequality and boundedness of the operator @xmath109 we obtain @xmath334 integrating over @xmath335 and using that @xmath336 we get @xmath337 the tensor @xmath338 is the symmetric gradient of the displacement , thus using the korn inequality ( cf .
* theorem 1.10 ) ) we conclude that the sequence @xmath331 is uniformly bounded in @xmath332 .
next we shall use and multiply it by @xmath343 and integrate over @xmath302 @xmath344 let us now integrate by parts the left hand side of @xmath345 passing to the limit in with @xmath308 we obtain @xmath346 note that the last inequality holds due to the weak lower semicontinuity in @xmath347 . to complete the proof we choose in the test functions @xmath348 , and in @xmath349 , where @xmath350 is a standard mollifier and we mollify with respect to time .
thus we obtain @xmath351 for @xmath352 .
summing @xmath258 over @xmath232 we obtain @xmath353 and @xmath354 products in are well defined , since for the matrices @xmath355 and @xmath356 the equivalence @xmath357 holds and the sequence @xmath358 is uniformly bounded in @xmath359 . passing with @xmath360 we obtain the equality @xmath361 since @xmath362,l^2(\omega,\mathcal{s}^3))$ ] , then we may pass with @xmath363 and conclude @xmath364 multiplying by @xmath365 and integrating over the interval @xmath366 we get @xmath367 for brevity we denote @xmath368 which is obviously in @xmath369 .
then we may apply the fubini theorem @xmath370 the crucial observation is that @xmath371 hence using and we conclude @xmath372 which is nothing else than @xmath373 observe now that @xmath374 passing with @xmath375 yields .
the proof is complete . to identify the weak limit @xmath324 we use the minty - browder trick . from the monotonicity of the function
@xmath17 we obtain @xmath376 hence @xmath377 the pointwise convergence of @xmath298 implies the pointwise convergence of @xmath378 .
the function @xmath379 belongs to @xmath380 , hence the sequence @xmath381 is uniformly bounded in @xmath273 .
then , using the lebesgue dominated convergence theorem we obtain that @xmath382 in @xmath383 for every @xmath384 . letting @xmath308 in , we get @xmath385 and taking @xmath386 , where @xmath387 and @xmath388 , then @xmath389 hence @xmath390 letting @xmath391 we obtain @xmath392 choosing now @xmath393 we
obtain the opposite inequality and hence @xmath394 thus @xmath395 consequently for every @xmath303 @xmath396 as @xmath397 .
using monotonicity of the function @xmath17 @xmath399 passing with @xmath187 to @xmath400 we get that the second term from converges to zero . furthermore , using lemma [ lm:8 ] @xmath401 hence @xmath402 and @xmath403 which completes the proof .
we start this section with considerations on the sequence of temperatures .
we are using the result of boccardo and galllout @xcite for parabolic equation with only integrable data and dirichlet boundary conditions .
since our studies concern the problem with neumann boundary conditions , we include the modification of their result in the appendix a. consequently , we conclude for each @xmath407 @xmath408 moreover , the uniform estimates from the previous sections allow to conclude that at least for a subsequence the following holds @xmath409 consequently , passing to the limit in , we obtain @xmath410 @xmath411 for all @xmath322,l^2(\omega,\mathcal{s}^3))$ ] and then also for all @xmath412 and for all @xmath413 to characterize the limit @xmath414 and pass to the limit in the heat equation we follow the similar lines as in the limit passage with @xmath308 . due to
we can rewrite as follows @xmath416 we multiply the above identity by @xmath417 given by formula and integrate over @xmath302 . passing to the limit @xmath406 we proceed in the same manner as in the proof of lemma [ lm:8 ] and obtain @xmath418 for the final step of the proof of the lemma we need to show that the energy equality holds .
contrary to the case of previous section , we can not use the time derivative of the limit , namely @xmath419 as the test function . although we shall mollifty with respect to time , but the regularity with respect to space is not sufficient since possibly @xmath420 .
therefore we proceed differently .
we use an approximate sequence as a test function in the limit identity .
indeed , we take in the test function @xmath421 , where again @xmath350 is a standard mollifier and we mollify with respect to time @xmath422 then we use the approximate equation with a test function @xmath423 . in a consequence
we obtain , which together with yields @xmath424 products in are well defined , since for the matrices @xmath355 and @xmath356 the equivalence @xmath357 holds and tensor @xmath9 belongs to @xmath273 . subtracting from we get @xmath425 for every @xmath426 the sequence @xmath427 belongs to @xmath64 and is uniformly bounded in @xmath64 , hence we pass to the limit with @xmath428 and we obtain using the properties of convolution we get @xmath430 and finally passing to the limit with @xmath431 and then with @xmath432 @xmath433 we multiply by @xmath365 and integrate over @xmath434 and proceed now in the same manner as in the proof of lemma [ lm:8 ] to complete the proof . using the minty - browder trick to identify the weak limit @xmath414 and the same argumentation as in the previous section , we obtain that @xmath435 furthermore @xmath436 in @xmath275 . using convergences presented above we pass to the limit with @xmath437 in the equations and , include the previously removed boundary and volume force term and obtain @xmath438 where @xmath439 and @xmath440 and @xmath441 what completes the proof of theorem [ thm : main2 ] .
let @xmath185 be a standard truncation operator defined in . in @xcite
, the authors showed the existence of solutions for the heat equation with dirichlet boundary conditions .
the current section is devoted to the existence proof to the problem with neumann boundary conditions .
two dimensional case was considered in @xcite .
we consider the sequence of the heat equations with boundary and initial conditions and with the right hand side of equation in the form @xmath442 which for every @xmath303 belongs to @xmath443 and moreover is uniformly bounded @xmath444 and @xmath445 in @xmath275 as @xmath406 .
additionally , we have @xmath446 , @xmath447 and @xmath448 in @xmath449 . to simplify the notation in the remaining part of the appendix we denote @xmath450 by @xmath451 .
let us consider the following problem @xmath452 and its weak formulation @xmath453 holding for all @xmath454 . where @xmath462 .
thus @xmath463 the terms on the right side of the above equation can be estimated as follows @xmath464 for every @xmath95 .
additionally , @xmath465 is nonnegative .
hence , @xmath466 where the set @xmath467 . now let @xmath468 and @xmath469 ( in our case @xmath470 and @xmath471 ) . using the hlder inequality we obtain @xmath472 then @xmath473 where @xmath474 . using the hlder inequality
we observe that @xmath475 is bounded by the terms @xmath476 , @xmath477 and the measure of the set @xmath451 .
furthermore , @xmath478 and @xmath479 is summable . using the interpolation inequality for @xmath480
we obtain @xmath481 where @xmath482 ( @xmath483 ) and @xmath484 .
after simple calculations we get that @xmath485 ( and @xmath486 ) .
in lemma [ linftyl1 ] we showed that @xmath487 is uniformly bounded , hence @xmath488 using the hlder inequality we obtain @xmath489 let us notice that the exponent @xmath490 . using the interpolation inequality for @xmath491 we get @xmath492 where @xmath493 .
the parameters @xmath494 are different in each of the interpolation inequalities and .
simple calculations yield that @xmath495 . by lemma [ linftyl1 ]
we conclude that @xmath496 the sobolev embedding theorem implies that @xmath497 using the previous inequalities we obtain @xmath498 and @xmath499 and @xmath500 , so we have the uniform boundedness @xmath501 and from the previous inequalities we get the uniform boundedness of the sequence @xmath502 in the space @xmath503 . using this uniform boundedness and inequalities and we get the uniform boundedness of the sequence @xmath502 in the spaces @xmath90 , which completes the proof .
let @xmath506 be such that , for @xmath507 fixed .
let us define a test function @xmath508 subtracting equation with function on right side @xmath509 and @xmath510 , and using the test function @xmath511 we obtain @xmath512 where @xmath513 and @xmath514 .
the sequence @xmath515 is convergent to @xmath145 in @xmath449 , hence , we can find @xmath516 such that for every @xmath517 , @xmath518 greater than @xmath516 we have @xmath519 .
the function @xmath520 is nonnegative and the right hand side of the equation above is bounded ( @xmath521 ) , hence @xmath522 the hlder inequality yields @xmath523 using the decomposition of @xmath524 we have to consider the integral over the second set .
@xmath525 the first term on the right hand side is bounded , since the sequence @xmath526 is uniformly bounded in @xmath90 .
the sequence @xmath526 is a cauchy sequence in @xmath275 , so there exists @xmath516 such that for all @xmath527 occurs @xmath528 .
then from the previous inequalities we obtain @xmath529 which implies that @xmath530 is a cauchy sequence in @xmath275 .
let @xmath531 , @xmath532 be banach spaces , and @xmath533 be a metrizable hausdorff locally convex space , @xmath531 be separable and reflexive , @xmath534 ( a compact embedding ) , @xmath535 ( a continuous embedding ) , @xmath536 , @xmath537 .
then @xmath538 ( a compact embedding ) . from the uniform boundedness of the sequence @xmath539 in @xmath275 and from the uniform boundedness of the sequence @xmath502 in @xmath90
we obtain that @xmath540 is a sequence bounded in the space @xmath541 .
consequently the sequence @xmath526 is relatively compact in @xmath275 . due to lemma [ unif_boun_lqlq ] and lemma [ zbieznosc_nabla_teta ] we know that the sequence @xmath526 converges strongly to @xmath19 in @xmath90 .
moreover , for @xmath494 large enough @xmath542 converges strongly to @xmath543 in @xmath544 .
thus , @xmath545 converges strongly to @xmath19 in @xmath546 , w^{-1,s}(\omega))$ ] and @xmath547 converges to @xmath548 in @xmath549 . in the current section we present the construction of the basis used for approximation of the strain tensor .
we adapt it for our particular case , however the idea follows the lines of ( * ? ? ?
* theorem 4.11 ) .
the definitions of spaces @xmath217 and @xmath218 were introduced in section [ sec:4 ] by and .
let us consider the following problem : find @xmath561 and @xmath562 such that @xmath563 where by @xmath211 we denote the scalar product in @xmath564 and @xmath214 is the previously defined scalar product in @xmath189 .
* @xmath567 for all @xmath568 , * @xmath569 and @xmath570 as @xmath571 tends to @xmath400 , * @xmath572 for all @xmath568 , * the set @xmath566 is a basis of @xmath573 . * the set @xmath566 is a basis of @xmath217 .
let us define @xmath579 consequently , there exists a sequence @xmath580 such that @xmath581 as @xmath571 tends to @xmath400 and @xmath582 .
then , there exist a subsequence @xmath580 ( still denoted by i ) and @xmath583 such that @xmath584 if @xmath585 , then let us define @xmath586 and then @xmath587 which is contrary with and it implies that @xmath588 .
to finish the first step we show that @xmath578 is an eigenfunction .
let us take @xmath589 and define the function @xmath590 calculating the derivative of function @xmath591 , we obtain @xmath592 and then @xmath593 assume that for @xmath594 there exists the set of eigenvalues @xmath595 and the set of corresponding eigenfunctions @xmath596 .
let us define the space @xmath597 using the similar construction as in the previous step , we find the next eigenvalue and eigenfunction @xmath598 let us assume that the set of eigenvalues has a finite limit , i.e. @xmath600 . since @xmath601 , using subsequence if it is necessary , we get @xmath602 in @xmath189 as @xmath603 .
hence @xmath604 passing with @xmath605 to @xmath400 we obtain @xmath606 comparing and we get the contradiction . _ the set @xmath565 contains all eigenvalues _ + let us assume that there exists an eigenvalue @xmath607 such that @xmath608 .
let @xmath609 be the corresponding eigenfunction to the eigenvalue @xmath607 and @xmath610 without loss of generality , @xmath611 .
moreover , there exists @xmath612 such that @xmath613 .
then , for all @xmath614 @xmath615 hence , @xmath616 and therefore @xmath617 and @xmath618 which is a contradiction .
let us define @xmath619 and let us assume that @xmath620 .
then , there exists @xmath621 such that @xmath622 and @xmath623 for all @xmath612 .
moreover , for all @xmath612 @xmath624 which implies that @xmath625 . observe that the space @xmath218 is dense in @xmath217 in @xmath627 norm .
for this purpose consider an element @xmath628 of @xmath629 . to show there exists a sequence @xmath630 bounded in @xmath631 that converges to @xmath628 recall that if @xmath628 is in @xmath189 , then there exists an approximating sequence @xmath632 in @xmath210 .
then the sequence @xmath630 we construct as follows @xmath633 where the projection @xmath284 was defined in the proof of lemma [ wsp_org_epa ] . then using the continuity of @xmath284 in @xmath564
we immediately obtain that @xmath630 is bounded in @xmath564 and converges to @xmath634 .
consequently , @xmath566 is also a basis in @xmath217 .
* acknowledgement * p.g . is a coordinator and f.k . is a phd student in the international phd projects programme of foundation for polish science operated within the innovative economy operational programme 2007 - 2013 funded by european regional development fund ( phd programme : mathematical methods in natural sciences ) .
pg was supported by the national science center , project no .
6085/b / h03/2011/40 .
fk was partially supported by grant ncn opus 2012/07/b / st1/03306 , a. .- g .
was supported by the grant idp2011/000661 . | our research is directed to a quasi - static evolution of the thermo - visco - elastic model .
we assume that the material is subject to two kinds of mechanical deformations : elastic and inelastic .
moreover , our analysis captures the influence of the temperature on the visco - elastic properties of the body .
the novelty of the paper is the consideration of the thermodynamically complete model to describe this kind of phenomena related with a hardening rule of norton - hoff type .
we provide the proof of existence of solutions to thermo - visco - elastic model in a simplified setting , namely the thermal expansion effects are neglected .
consequently , the coupling between the temperature and the displacement occurs only in the constitutive function for the evolution of the visco - elastic strain . |
the physics of @xmath0 mesons ( those containing the @xmath4 [ bottom or beauty ] quark ) has greatly illuminated the study of the electroweak and strong interactions .
this brief review is devoted to some theoretical aspects of @xmath0 physics , with emphasis on current questions for @xmath5 and hadron collider experiments .
section 2 reviews weak quark transitions .
we note in section 3 progress and puzzles in the study of @xmath6 decays to cp eigenstates , turning in section 4 to direct cp asymmetries which require strong final - state phases for their observation .
some aspects of these phases are described in in section 5 .
we devote section 6 to the strange @xmath0 mesons , with section 7 treating the possibility of narrow @xmath7 states suggested by the recent observation of narrow @xmath2 mesons .
section 8 discusses the prospects for seeing heavier @xmath3 quarks .
we summarize in section 9 .
this review updates and supplements refs .
the relative strengths of charge - changing weak quark transitions are shown in fig .
[ fig : trans ] .
it is crucial to describe this pattern precisely in order to distinguish among theories which might predict it , and to see whether it can reproduce all weak phenomena including cp violation or whether some new ingredient is needed . .
[ fig : trans],height=288 ] the interactions in fig .
[ fig : trans ] may be parametrized by a unitary cabibbo - kobayashi - maskawa ( ckm ) matrix which can be written approximately @xcite in terms of a small expansion parameter @xmath8 as [ eqn : wp ] v_ckm = , where @xmath9 and @xmath10 .
the columns refer to @xmath11 and the rows to @xmath12 .
the parameter @xmath13 @xcite is @xmath14 , where @xmath15 is the cabibbo angle .
the value @xmath16 , obtained from @xmath17 decays , indicates @xmath18 , while @xmath19 , obtained from @xmath20 decays , implies @xmath21 .
we shall generally use the ckm parameters quoted in ref .
@xcite .
the unitarity of the ckm matrix can be expressed in terms of a triangle in the complex @xmath22 plane , with vertices at ( 0,0 ) ( angle @xmath23 ) , ( 1,0 ) ( angle @xmath24 ) , and @xmath25 ( angle @xmath26 ) .
the triangle has unit base and its other two sides are @xmath27 @xmath24 ) and @xmath28 .
the result is shown in fig .
[ fig : ut ] . , @xmath29 , and @xmath30 .
[ fig : ut],height=144 ] flavor - changing loop diagrams provide further constraints .
cp - violating @xmath31@xmath32 mixing is dominated by @xmath33 with virtual @xmath34 and @xmath35 intermediate states .
it constrains im@xmath36 , giving a hyperbolic band in the @xmath37 plane .
@xmath6@xmath38 mixing is dominated by @xmath34 and @xmath35 in the loop diagram for @xmath39 , and thus constrains @xmath40 and hence @xmath41 . by comparing @xmath1@xmath42 and @xmath6@xmath38 mixing ,
one reduces dependence on matrix elements and learns @xmath43 or @xmath44 .
the resulting constraints are shown in fig .
[ fig : ckmf ] @xcite .
plane as of july 2002 ( from the web page of ref .
[ fig : ckmf],height=566 ]
the decays of neutral @xmath0 mesons to cp eigenstates @xmath45 , where @xmath46 , @xmath47 , provide direct information on ckm phases without the need to understand complications of strong interactions . as a result of @xmath6@xmath48 mixing
, a state @xmath6 at proper time @xmath49 evolves into a mixture of @xmath6 and @xmath48 denoted @xmath50 .
thus there will be one pathway to the final state @xmath45 from @xmath6 through the amplitude @xmath51 and another from @xmath48 through the amplitude @xmath52 , which acquires an additional phase @xmath53 through the mixing .
the interference of these two amplitudes can differ in the decays @xmath54 and @xmath55 , leading to a time - integrated rate asymmetry a_cp as well as to time - dependent rates \
{ c + } ~e^- t [ 1 a_f m t s_f m t ] , where a_f , s_f , e^-2 i , where @xmath56 @xcite . for this decay one has @xmath58 .
the time - integrated asymmetry @xmath59 is proportional to @xmath60 . using this and related decays involving the same quark subprocess , babar @xcite finds @xmath61 while belle @xcite finds @xmath62 .
the world average @xcite is @xmath63 , consistent with other determinations @xcite .
two amplitudes contribute to the decay : a `` tree '' @xmath65 and a `` penguin '' @xmath66 : a = - ( |t|e^i + |p| e^i ) , |a = - ( |t|e^-i + |p| e^i ) , where @xmath67 is the relative @xmath68 strong phase .
the asymmetry @xmath59 would be proportional to @xmath69 if the penguin amplitude could be neglected .
one way to account for its contribution is via an isospin analysis @xcite of @xmath0 decays to @xmath70 , @xmath71 , and @xmath72 , separating the amplitudes for decays involving @xmath73 and @xmath74 final states .
information can then be obtained on both strong and weak phases . since the branching ratio of @xmath6 to @xmath72 may be very small , of order @xmath75 , alternative methods @xcite may be useful in which flavor su(3 ) symmetry is used to estimate the penguin contribution @xcite .
the tree amplitude for @xmath76 involves @xmath77 , with the spectator @xmath78 quark combining with @xmath79 to form a @xmath80 .
its magnitude is @xmath81 ; its weak phase is arg(@xmath82 ; by convention its strong phase is 0 .
the penguin amplitude involves the flavor structure @xmath83 , with the final @xmath84 pair fragmenting into @xmath70 .
its magnitude is @xmath85 .
the dominant @xmath86 contribution in the loop diagram for @xmath87 can be integrated out and the unitarity relation @xmath88 used .
the @xmath89 contribution can be absorbed into a redefinition of the tree amplitude , after which the weak phase of the penguin amplitude is 0 ( mod @xmath90 ) . by definition ,
its strong phase is @xmath67 .
the time - dependent asymmetries @xmath91 and @xmath92 specify both @xmath93 ( or @xmath94 ) and @xmath67 , if one has an independent estimate of @xmath95 .
one may obtain @xmath85 from @xmath96 using flavor su(3 ) @xcite and @xmath81 from @xmath97 using factorization @xcite .
( an alternative method discussed in refs .
@xcite uses the measured ratio of the @xmath98 and @xmath64 branching ratios to constrain @xmath95 . )
in addition to @xmath91 and @xmath92 , a useful quantity is the ratio of the @xmath64 branching ratio @xmath99 ( unless otherwise specified , branching ratios refer to cp averages ) to that due to the tree amplitude alone : r _ = 1 + 2 | | + | |^2 .
one also has r _ s _ = 2 + 2 | | ( - ) - | |^2 2 , r _ a _ = - 2 |p / t| .
we take @xmath100 .
the value of @xmath95 ( updating @xcite ) is @xmath101 . taking the central value , we plot in fig .
[ fig : sa ] trajectories in the ( @xmath102 ) plane for @xmath103 . and @xmath92 on @xmath67 @xcite
. from right to left the curves correspond to @xmath104 .
plotted point : average of babar and belle values ( see text ) .
as @xmath105 increases from 0 to @xmath90 , the values of @xmath91 become more positive , while the magnitudes @xmath106 increase from zero and then return to zero .
positive values of @xmath92 correspond to negative values of @xmath67 .
[ fig : sa],height=432 ] .values of @xmath91 and @xmath92 quoted by babar and belle and their averages . here
we have applied scale factors of @xmath107 to the errors for @xmath91 and @xmath92 , respectively . [
tab : sa ] [ cols="^,^,^,^",options="header " , ] if an amplitude @xmath51 for a process receives two contributions with differing strong and weak phases , one can write a = a_1 + a_2 e^i e^i , |a = a_1 + a_2 e^-i e^i .
the cp - averaged decay rate is proportional to @xmath108 , while the cp asymmetry is a_cp = - . in the case of @xmath109 the rates and cp asymmetry suggest that @xmath110 . by combining the branching ratio and @xmath111 rate asymmetry information of the @xmath112 modes
, one can extract the values of the relative strong phase @xmath67 and the weak phase @xmath113 , assuming maximal constructive interference between ordinary and singlet penguin amplitudes . on the basis of the range of amplitudes extracted from other processes , we find that the rates and cp asymmetries for @xmath114 and @xmath115 are correlated with one another @xcite : [ eq : eq1 ] a_cp(^+ ) = -(0.91 ) / ( 1 - 0.91 ) , [ eq : eq2 ] a_cp(^+ ) = -()/ ( 1 - ) , [ eq : eq3 ] ( ^+ ) = 5 10 ^ -6(1 - 0.91 ) , [ eq : eq4 ] ( ^+ ) = 3.4 10 ^ -6(1 - ) , where @xmath116 refers to a cp - averaged branching ratio .
one finds that @xmath117 ( below current upper bounds ) and that @xmath118 .
the amplitudes for @xmath119 may be written in the form a(^ ) ~e^i , + where @xmath120 ( estimated in ref . @xcite to be @xmath121 ) is the ratio of penguin to tree contributions to the @xmath119 decay amplitudes .
we define @xmath122 as the ratio of the observed @xmath111-averaged @xmath119 decay rate to that which would be expected in the limit of no penguin contributions and find [ eqn : reta ] r_= 1 + r_^2 - 2 r_= 1.18 0.30 .
one can then use the information on the observed @xmath111 asymmetry in this mode to eliminate @xmath67 and constrain @xmath113 .
( for a related treatment with a different convention for penguin amplitudes see ref .
the asymmetry is [ eqn : aeta ] a_= -2 r_/r_= -0.51 0.19 , so one can either use the result [ eqn : raeta ] r_= 1 + r_^2 with experimental ranges of @xmath122 and @xmath123 or solve ( [ eqn : raeta ] ) for @xmath122 in terms of @xmath113 and @xmath123 .
the result of this latter method is illustrated in fig .
[ fig : reta ] @xcite .
( ratio of observed @xmath111-averaged @xmath124 decay rate to that predicted for tree amplitude alone ) as a function of @xmath113 for various values of @xmath111 asymmetry @xmath125 @xcite .
( the values 0.70 and 0.32 correspond to @xmath126 errors on this asymmetry . ) [ fig : reta],height=336 ] the range of @xmath113 allowed at 95% c.l . in standard - model fits to ckm parameters
is @xmath127 @xcite . for comparison ,
[ fig : reta ] permits values of @xmath113 in the three ranges 14 ^ 53^ , 60 ^ 120^ , 127 ^ 166^ if @xmath122 and @xmath125 are constrained to lie within their @xmath128 limits .
the middle range overlaps the standard - model parameters , restricting them slightly .
better constraints on @xmath113 in this region would require reduction of errors on @xmath122 .
we have seen that final - state phases are needed in order to observe direct cp asymmetries .
it is interesting to obtain information on such phases in those @xmath129 decays in which weak phases are expected to play little role , so that magnitures of amplitudes directly reflect relative strong phases . as one example we illustrate such phases in the decays of @xmath130 and related processes in fig .
[ fig : dpi ] @xcite . the color - suppressed amplitude @xmath131 is found to have a non - trivial strong phase with respect to the color - favored tree amplitude @xmath65 , with a small exchange amplitude @xmath132 ( governing @xmath133 ) at an even larger phase with respect to @xmath65 .
such large phases can signal strong rescattering effects . and related decays @xcite .
the amplitude @xmath132 points from either @xmath134 or @xmath135 to the center of the small circle .
the amplitudes @xmath65 and @xmath131 are shown only for the first of these two solutions . here
@xmath136 , @xmath137 , @xmath138 , @xmath139 , and @xmath140 .
[ fig : dpi],height=201 ] in one method for measuring the weak phase @xmath93 in @xmath141 decays , the relative strong phase @xmath142 in @xmath143 and @xmath144 decays ( equivalently , in @xmath143 and @xmath145 ) plays a role @xcite .
a study of the dalitz plot in @xmath146 can yield information on this phase @xcite . by comparing such dalitz plots for constructive and destructive interference between the two
@xmath147 bands one finds that a clear - cut distinction is possible between @xmath148 and @xmath149 with a couple of thousand decays .
the ratio of the @xmath1@xmath150 mixing amplitude @xmath151 to the @xmath6@xmath38 mixing amplitude @xmath152 ( @xmath153 ) is given by = | while @xmath154 express the degree to which the mixing amplitude is due to vacuum intermediate states .
a lattice estimate of the ratio @xmath155 is @xmath156 @xcite .
we have taken @xmath157 with @xmath158 and @xmath159 . with @xcite @xmath160 ps@xmath161 one
then predicts @xmath162
. the lower portion of this range is already excluded by the bound @xcite @xmath163 .
when @xmath151 is measured it will constrain @xmath164 significantly . since the weak phase in @xmath166 is expected to be zero while that of @xmath1@xmath42 mixing is expected to be very small , one expects cp asymmetries to be only a few percent in the standard model for those @xmath1 decays dominated by this quark subprocess . the @xmath167 final state
is not a cp eigenstate but the even and odd cp components can be separated using angular analyses .
the final states of @xmath168 and @xmath169 are cp - even so no such analysis is needed .
a comparison of time - dependent asymmetries in @xmath170 and @xmath171 @xcite allows one to separate out strong and weak phases and relative tree and penguin contributions . in @xmath170
the @xmath172 penguin amplitude is dominant , while the strangeness - changing tree amplitude @xmath173 is smaller . in @xmath174
it is the other way around : the @xmath175 tree amplitude dominates , while the @xmath83 penguin is cabibbo - suppressed .
the u - spin subgroup of su(3 ) , which interchanges @xmath176 and @xmath78 quarks , relates each amplitude in one process to that in the other apart from the ckm factors . in comparing @xmath170 with @xmath64
, the mass peaks will overlap with one another if analyzed in terms of the same final state ( e.g. , @xmath70 ) @xcite .
thus , in the absence of good particle identification , a variant on this scheme employing the decays @xmath178 and @xmath179 ( also related to one another by u - spin ) may be useful @xcite .
for these final states , kinematic separation may be easier .
one can also study the time - dependence of @xmath170 while normalizing the penguin amplitude using @xmath180 @xcite .
the u - spin subgroup of su(3 ) allows one to relate many other @xmath1 decays besides those mentioned above to corresponding @xmath181 decays @xcite .
particularly useful are relations between cp - violating rate _
differences_. one thus will have the opportunity to perform many tests of flavor su(3 ) and to learn a great deal more about final - state phase patterns when a variety of @xmath1 decays can be studied .
a promising method for tagging the flavor of a neutral @xmath0 meson is to study the charge of the leading light hadron accompanying the fragmentation of the heavy quark @xcite .
for example , an initial @xmath4 will fragment into a @xmath38 by `` dressing '' itself with a @xmath182 .
the accompanying @xmath78 , if incorporated into a charged pion , will end up in a @xmath80 . thus a @xmath80 is more likely to be `` near '' a @xmath38 than to a @xmath6 in phase space .
this correlation between @xmath80 and @xmath38 ( and the corresponding correlation between @xmath183 and @xmath6 ) is also what one would expect on the basis of non - exotic resonance formation .
thus the study of the resonance spectrum of the excited @xmath0 mesons which can decay to @xmath184 or @xmath185 is of special interest @xcite .
the lowest such mesons are the p - wave levels of a @xmath186 antiquark and a light ( @xmath187 or @xmath78 ) quark . in april of this year
the babar collaboration @xcite reported a narrow resonance at 2317 mev decaying to @xmath190 .
this state was quickly confirmed by cleo @xcite , who also presented evidence for a narrow state at 2463 mev decaying to @xmath191 .
both states have been confirmed by belle @xcite .
we mention briefly why these states came as surprises .
the previously known p - wave levels of a charmed quark @xmath192 and an antistrange @xmath193 were candidates for @xmath194 and @xmath195 states at 2535 and 2572 mev @xcite .
these levels have narrow widths behave as expected if the spin of the @xmath193 and the orbital angular momentum were coupled up to @xmath196 .
( one expects @xmath197-@xmath197 rather than @xmath198-@xmath199 coupling in a light - heavy system @xcite . )
if the @xmath200 states were fairly close to these in mass one would then expect another @xmath194 state and a @xmath201 state somewhere above 2500 mev .
instead , the candidate for the @xmath201 @xmath2 state is the one at 2317 mev , with the state at 2463 mev the candidate for the second @xmath194 level .
belle s observation of the decay @xmath202 reinforces this interpretation @xcite .
both states are narrow since they are too light to decay respectively to @xmath203 or @xmath204 .
they decay instead via isospin - violating transitions .
they are either candidates for @xmath205 molecules @xcite , or indications of a broken chiral symmetry which places them as positive - parity partners of the @xmath188 and @xmath206 negative - parity @xmath2 ground states @xcite .
indeed , the mass splittings between the parity partners appear to be exactly as predicted ten years ago @xcite .
potential - based quarkonium models have a hard time accommodating such low masses @xcite , there should exist _ non - strange _
@xmath200 @xmath207 and @xmath208 states , lower in mass than the @xmath209 states at 2422 and 2459 mev @xcite but quite broad since their respective @xmath210 and @xmath211 channels will be open . the study of such states will be of great interest since the properties of the corresponding @xmath0-flavored states will be useful in tagging the flavor of neutral @xmath0 mesons .
if a strange antiquark can bind to a charmed quark in both negative- and positive - parity states , the same must be true for a strange antiquark and a @xmath4 quark .
one should then expect to see narrow @xmath213 and @xmath208 states with the quantum numbers of @xmath214 and @xmath215 but below those respective thresholds .
they should decay to @xmath216 and @xmath217 , respectively . to see such decays
one will need a multi - purpose detector with good charged particle and @xmath218 identification !
might there be heavier quarks visible at hadron colliders ? at present we have evidence for three families of quarks and leptons belonging to 16-dimensional multiplets of the grand unified group so(10 ) ( counting right - handed neutrinos as a reasonable explanation of the observed oscillations between different flavors of neutrinos ) .
just as so(10 ) was pieced together from multiplets of su(5 ) with dimensions 1 , 5 , and 10 , the smallest representation of a still larger grand unified group could contain the 16-dimensional so(10 ) spinor .
such a group is e@xmath220 @xcite .
its smallest representation , of dimension 27 , contains a 16-dimensional spinor , a 10-dimensional vector , and a singlet of so(10 ) .
the 10-dimensional vector contains vector - like isosinglet quarks `` @xmath221 '' and antiquarks @xmath222 of charge @xmath223 and isodoublet leptons .
the so(10 ) singlets are candidates for sterile neutrinos , one for each family .
the new exotic @xmath221 quarks can mix with the @xmath4 quark and push its mass down with respect to the top quark @xcite .
troy andre and i are looking at signatures of @xmath224 production in hadron colliders , to either set lower mass limits or see such quarks through their decays to @xmath225 , @xmath226 , and possibly @xmath227 .
the @xmath228 , for example , would be identified by its decays to @xmath229 , @xmath230 , or jet @xmath231 jet , while the higgs boson would show up through its @xmath232 decay if it were far enough below @xmath35 threshold .
the process @xmath57 has provided spectacular confirmation of the kobayashi - maskawa theory of cp violation , measuring @xmath233 to a few degrees .
now one is entering the territory of more difficult measurements .
the decay @xmath64 can give useful information on @xmath113 .
one needs either a measurement of @xmath234 @xcite , probably at the @xmath75 level ( present limits @xcite are several times that ) , or a better estimate of the tree amplitude from @xmath235 @xcite .
indeed , such an estimate has been presented recently @xcite .
the babar and belle experimental cp asymmetries @xcite will eventually converge to one another , as did the initial measurements of @xmath236 using @xmath57 .
the @xmath237 decay can display new physics via special @xmath238 operators or effects on the @xmath239 penguin .
some features of any new amplitude can be extracted from the data in a model - independent way if one uses both rate and asymmetry information @xcite . while the effective value of @xmath236 in @xmath240 seems to differ from its expected value by more than @xmath241
, cp asymmetries in @xmath242 do not seem anomalous .
the rate for @xmath243 is not a problem for the standard model if one allows for a modest flavor - singlet penguin contribution in addition to the standard penguin amplitude .
the cp asymmetries for this process are in accord with the expectations of the standard model at the @xmath128 level or better .
effects of the singlet penguin amplitude may also be visible elsewhere , for example in @xmath244 .
various ratios of @xmath245 rates , when combined with information on cp asymmetries , show promise for constraining phases in the ckm matrix .
these tests have steadily improved in accuracy in the past couple of years .
one expects further progress as @xmath5 luminosities increase , and as hadron colliders begin to provide important contributions .
the decays @xmath246 and @xmath247 show promise for displaying large cp asymmetries @xcite since they involve contributions of different amplitudes with comparable magnitudes .
strong final - state phases , important for the observation of direct cp violation , are beginning to be mapped out in @xmath0 decays . in the near term
the prospects for learning about the @xmath1@xmath42 mixing amplitude are good .
the potentialities of hadron colliders for the study study of cp violation and branching ratios in @xmath1 decays will be limited only by the versatility of detectors .
surprises in spectroscopy , as illustrated by the low - lying positive - parity @xmath2 candidiates , still can occur , and one is sure to find more of them .
finally , one can search for objects related to the properties of @xmath4 quarks , such as the exotic isosinglet quarks @xmath221 , with improved sensitivity in run ii of the tevatron and with greatly expanded reach at the lhc .
i thank my collaborators on topics mentioned here : troy andre , cheng - wei chiang , michael gronau , zumin luo , and denis suprun .
michael gronau and hassan jawahery also made helpful comments on the manuscript .
this work was supported in part by the united states department of energy under grant no .
de fg02 90er40560 .
# 1#2#3am .
j. phys . * # 1 * , # 2 ( # 3 ) # 1#2#3ann .
( n.y . ) * # 1 * , # 2 ( # 3 ) # 1#2#3acta phys .
polonica * # 1 * , # 2 ( # 3 ) # 1#2#3ann .
nucl . part .
sci . * # 1 * , # 2 ( # 3 ) # 1#2#3comments on nucl . part .
phys . * # 1 * ( # 3 ) # 2 89_cp violation , _ edited by c. jarlskog ( world scientific , singapore , 1989 ) # 1#2#3electronic conference proceedings * # 1 * , # 2 ( # 3 ) # 1#2#3eur .
j. c * # 1 * ( # 3 ) # 2 79_proceedings of the 1979 international symposium on lepton and photon interactions at high energies , _ fermilab , august 23 - 29 , 1979 , ed . by t. b. w. kirk and h. d. i. abarbanel ( fermi national accelerator laboratory , batavia , il , 1979 87_proceeding of the 1987 international symposium on lepton and photon interactions at high energies , _ hamburg , 1987 , ed . by w. bartel and r. rckl ( nucl .
b , proc .
3 ) ( north - holland , amsterdam , 1988 ) # 1#2#3 * # 1 * ( # 3 ) # 2 72_proceedings of the xvi international conference on high energy physics _ ,
chicago and batavia , illinois , sept .
6 13 , 1972 , edited by j. d. jackson , a. roberts , and r. donaldson ( fermilab , batavia , il , 1972 ) # 1#2#3int .
j. mod .
phys . a * # 1 * ( # 3 ) # 2 # 1#2#3jhep * # 1 * ( # 3 ) # 2 # 1#2#3j .
b * # 1 * , # 2 ( # 3 ) 87_selected topics in electroweak interactions _ ( proceedings of the second lake louise institute on new frontiers in particle physics , 15 21 february , 1987 ) , edited by j. m. cameron ( world scientific , singapore , 1987 ) # 1#2#3kong .
danske vid .
selsk . , matt - fys
* # 1 * , no.#2 ( # 3 ) # 1#2#3mod . phys .
a * # 1 * ( # 3 ) # 2 # 1#2#3nature * # 1 * , # 2 ( # 3 ) # 1#2#3nuovo cim
. * # 1 * ( # 3 ) # 2 # 1#2#3nucl . instr .
meth . a * # 1 * , # 2 ( # 3 ) # 1#2#3nucl .
# 1 * ( # 3 ) # 2 # 1#2#3nucl .
suppl . * # 1 * ( # 3 ) # 2 # 1#2#3nucl .
suppl . *
# 1 * ( # 3 ) # 2 # 1#2#3#4pisma zh . eksp .
teor . fiz .
* # 1 * , # 2 ( # 3 ) [ jetp lett .
* # 1 * , # 4 ( # 3 ) ] # 1#2#3phys .
# 1 * ( # 3 ) # 2 # 1#2#3phys .
lett . a * # 1 * , # 2 ( # 3 ) # 1#2#3phys . lett .
b * # 1 * ( # 3 ) # 2 # 1#2#3phys .
* # 1 * ( # 3 ) # 2 # 1#2#3phys . rev .
c * # 1 * ( # 3 ) # 2 # 1#2#3phys .
d * # 1 * ( # 3 ) # 2 # 1#2#3phys .
* # 1 * ( # 3 ) # 2 # 1#2#3phys . rep .
* # 1 * ( # 3 ) # 2 # 1#2#3prog .
# 1 * ( # 3 ) # 2 # 1#2#3rev .
* # 1 * ( # 3 ) # 2 # 1 rp # 1 9025th international conference on high energy physics , singapore , aug . 2 - 8 , 1990 87_proceedings of the salt lake city meeting _ ( division of particles and fields , american physical society , salt lake city , utah , 1987 ) , ed .
by c. detar and j. s. ball ( world scientific , singapore , 1987 ) 89_proceedings of the xivth international symposium on lepton and photon interactions , _ stanford , california , 1989 , edited by m. riordan ( world scientific , singapore , 1990 ) 82_proceedings of the 1982 dpf summer study on elementary particle physics and future facilities _ , snowmass , colorado , edited by r. donaldson , r. gustafson , and f. paige ( world scientific , singapore , 1982 ) 90_research directions for the decade _ ( proceedings of the 1990 summer study on high energy physics , june 25july 13 , snowmass , colorado ) , edited by e. l. berger ( world scientific , singapore , 1992 ) # 1#2#3#4yad .
fiz . * # 1 * , # 2 ( # 3 ) [ sov .
j. nucl .
# 4 ( # 3 ) ] # 1#2#3#4#5#6zh .
fiz . * # 1 * , # 2 ( # 3 ) [ sov.phys
. - jetp * # 4 * , # 5 ( # 6 ) ] # 1#2#3zeit .
c * # 1 * , # 2 ( # 3 ) # 1#2#3zeit .
d * # 1 * , # 2 ( # 3 ) j. l. rosner , enrico fermi institute report no . efi-03 - 26 ,
hep - ph/0305315 , invited talk at 4th international symposium on lhc physics and detectors ( lhc 2003 ) , batavia , illinois , 13 may 2003 , to be published in eur .
j. g. hiller , ; m. ciuchini and l. silvestrini , ; a. datta , ; m. raidal , ; b. dutta , c. s. kim , and s. oh , ; s. khalil and e. kou , and hep - ph/0303214 ; g. l. kane , p. ko , h. wang , c. kolda , j. h. park and l. t. wang , ; s. baek , ; a. kundu and t. mitra , ; k. agashe and c. d. carone , hep - ph/0304229 .
babar , b. aubert , quoted by s. playfer at lhcb workshop , cern , february 2003 ; updated results on @xmath248 presented by j. ocariz at eps 2003 ( international europhysics conference on high energy physics , aachen , 1723 july 2003 .
w. a. bardeen and c. t. hill , .
chiral partners of the ground states of heavy mesons were independently predicted by m. a. nowak , m. rho , and i. zahed , .
see also d. ebert , t. feldmann , r. friedrich , and h. reinhardt , ; d. ebert , t. feldmann , and h. reinhardt , . | some theoretical aspects of @xmath0 physics are reviewed .
these include a brief recapitulation of information on weak quark transitions as described by the cabibbo - kobayashi - maskawa ( ckm ) matrix , descriptions of cp asymmetries in @xmath0 decays to cp eigenstates and to self - tagging modes , a discussion of final - state phases in @xmath0 and charm decays , some topics on @xmath1 properties and decays , prospects for unusual excited @xmath0 states opened by discovery of some narrow @xmath2 resonances , and the search for heavier @xmath3 quarks predicted in some extended grand unified theories .
address = enrico fermi institute and department of physics + university of chicago , 5640 s. ellis avenue , chicago il 60637 |
the analysis of the unitarity triangle ( ut ) and cp violation represents one of the most stringent tests of the standard model ( sm ) and , for this reason , also an interesting window on new physics ( np ) . the input of this analysis is a large number of both experimental and theoretical parameters , the most relevant of which are listed in table [ tab : inputs ] . [
cols="^,^,^,^,^ " , ] to the angle @xmath0 can be also extracted , in principle , from the measurements of two - body charmless hadronic b decays .
a lot of theoretical investigations have been recently made for these decay channels .
in particular , important progress has been made with the calculation of the amplitudes in the heavy quark limit using the factorization approach @xcite , though there is still some controversy on the importance of non - leading corrections @xcite . from the experimental point of view
an impressive effort has been made to measure as many branching fractions and cp asymmetries as possible ( for a collection of results see @xcite ) .
it would be very interesting in the near future to compare the determination of the angle @xmath0 from the ut analysis and from two - body decays measurements . in the future , at the lhc / btev experiments , @xmath0 will be cleanly measured in the tree - level @xmath1 decays .
the plot in figure [ fig : pull_gamma ] shows the compatibility of the indirect determination of @xmath0 with a future determination of the same angle obtained from b decays .
it can be noted that even in case the angle @xmath0 can be measured with a precision of 10@xmath2 from b decays , the predicted 3@xmath3 region is still rather large , corresponding to the interval [ 25 - 100]@xmath2 . if all the theoretical errors are divided by a factor of two , as indicated by the darker curve , the predicted 3@xmath3 region will be consistently reduced .
although the ckm mechanism is extremely successful , leading in particular to the precise prediction of the value of sin2@xmath4 , it is nevertheless worth to investigate whether the analysis of the ut still allows some room for np effects .
this is the issue we would like to address in this section .
the physical processes entering the analysis and the related physical observables determined from the experiments , are listed in table [ tab : schema ] . barring the possibility of significant np effects in the determination of the ratio @xmath5 from tree - level processes
, we have explored the possible contributions to @xmath6 mixing ( @xmath7 ) and @xmath8 mixing .
np contributions introduce in general a large number of new parameters : flavour changing couplings , short distance coefficients and matrix elements of new local operators . the specific list and
the actual values of these parameters depend on the details of the np model . nevertheless ,
each of the mixing process listed in table [ tab : schema ] , being described by a single amplitude , can be effectively parameterized in a completely general way in terms of only two new parameters , which we choose to quantify the difference of the amplitude in absolute value and phase with respect to the sm one @xcite .
thus , for instance , in the case of @xmath6 mixings we define @xmath9 where @xmath10 includes only the sm box diagrams , while @xmath11 includes also the np contributions . by definition , in the absence of np effects , @xmath12 and @xmath13 .
the experimental quantities determined from the @xmath6 mixings and listed in table [ tab : schema ] are related to their sm counterparts and the np parameters by the following relations : @xmath14 and @xmath15 as far as the @xmath8 mixing is concerned , we find it convenient to introduce a single parameter which relates the imaginary part of the amplitude to the sm one @xmath16 } { { \rm im}[\langle k^0|h_{eff}^{sm}|\bar{k}^0\rangle]}\ , .
\label{eq : ceps}\ ] ] this definition implies in fact a simple relation for @xmath17 , @xmath18 thus , all np effects which may enter the present analysis of the ut are parameterized in terms of four real coefficients , @xmath19 , @xmath20 , @xmath21 and @xmath22 . due to
the limited available number of constraints , we now make the hypothesis that np effects can appear ( at least to a large extent ) only in one of the three mixing amplitudes .
this is the only restrictive assumption we make in this analysis . clearly , this hypothesis is not valid in several specific models , as for instance in the minimal flavour violating susy models @xcite in which np effects simultaneously appear in all the mixing amplitudes but with coefficients which are related one to the other .
the result from the fit is : @xmath23~ \rm{at}~95\% ~c.l .
\label{eq : cepsk}\end{aligned}\ ] ] the value is compatible with unity but the distribution is rather broad .
thus , in this respect , large np contributions to @xmath8 mixing are still allowed .
we note , however , that the experimental constraint coming from @xmath17 can only determine the product @xmath24 .
therefore , the large width of the distribution of @xmath22 simply reflects the uncertainty existing on the hadronic parameter @xmath25 .
we also find that , in this scenario , the distributions of the other ut parameters ( @xmath26 , @xmath27 , sin2@xmath4 , @xmath28 ) are not really different from those obtained using the sm parameterization ( @xmath29 ) . the value of @xmath30 peaks at 1 .
however , it is not limited from above due to the fact the @xmath32 is not yet determined . also in this case ,
the distributions of the other ut parameters are not really different from those obtained using the sm parameterization ( @xmath33 ) .
finally we explore the possibility of np contributions in @xmath34 mixing .
figure [ fig : cd ] shows the distributions of @xmath35 and @xmath20 . is determined from the measurement of @xmath36 up to a discrete ambiguity , @xmath37 . in figure
[ fig : cd ] only one of the two determinations ( the one containing @xmath38 ) is shown for clarity . ]
we see that two solutions are possible in this case .
the first one peaks around the sm solution , @xmath39 and @xmath40 .
the second one represents instead the possibility of a really distinct np contribution .
the presence of two solutions corresponds to the fact that , as shown in figure [ fig : cd2 ] , the constraint from @xmath17 intercepts the circle defined by @xmath41 in two regions .
the region on the quadrant with positive value of @xmath42 corresponds to the sm solution .
the presence of these two solutions is also visible in several distributions of the ut parameters shown in figure [ fig:1d_angle_cd ] .
as can be seen from these plots , in order to discriminate between the two solutions , an independent determination of either @xmath26 , as obtained for instance from the study of the ratio @xmath43 of radiative @xmath44 decays , or @xmath0 would be necessary . the np solution has been also recently discussed in the literature @xcite .
we want to conclude this discussion with a comment concerning the naturalness " of this np solution " .
first , we observe that np , though possible , is certainly not required in order to explain the results of the ut analysis . in addition , suppose that the np solution in @xmath34 mixing is indeed the correct one .
then it is somewhat surprising to find that in the sm , i.e. in the
wrong " theory of @xmath34 mixing , the predicted values of @xmath45 and sin2@xmath4 actually select in the ( @xmath26 , @xmath27 ) plane just one of the two regions which are also allowed by the other constraints , namely @xmath46 and @xmath17 . in this sense
, the consistency of this np solution would be accidental .
y. grossman and m. p. worah , phys .
b * 395 * ( 1997 ) 241 [ arxiv : hep - ph/9612269 ] .
+ m. ciuchini _ et al .
_ , phys .
* 79 * ( 1997 ) 978 [ arxiv : hep - ph/9704274 ] .
d. london and a. soni , phys .
b * 407 * ( 1997 ) 61 [ arxiv : hep - ph/9704277 ] .
+ y. grossman , g. isidori and m. p. worah , phys .
d * 58 * ( 1998 ) 057504 [ arxiv : hep - ph/9708305 ] .
b. aubert _ et al .
_ [ babar collaboration ] , arxiv : hep - ex/0207070
. + k. abe _ et al . _ [ belle collaboration ] , arxiv : hep - ex/0207098 .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , phys .
* 83 * ( 1999 ) 1914 [ arxiv : hep - ph/9905312 ] ; nucl .
b * 591 * ( 2000 ) 313 [ arxiv : hep - ph/0006124 ] ; nucl .
b * 606 * ( 2001 ) 245 [ arxiv : hep - ph/0104110 ] .
m. ciuchini _ et al .
_ , aip conf .
* 602 * ( 2001 ) 180 [ arxiv : hep - ph/0110022 ] ; phys . lett .
b * 515 * ( 2001 ) 33 [ arxiv : hep - ph/0104126 ] .
a. stocchi , nucl .
* 117 * ( 2003 ) 145 [ arxiv : hep - ph/0211245 ] .
j. m. soares and l. wolfenstein , phys .
d * 47 * ( 1993 ) 1021 .
+ n. g. deshpande , b. dutta and s. oh , phys . rev .
* 77 * ( 1996 ) 4499 [ arxiv : hep - ph/9608231 ] .
+ j. p. silva and l. wolfenstein , phys .
d * 55 * ( 1997 ) 5331 [ arxiv : hep - ph/9610208 ] .
+ a. g. cohen _ et al .
_ , phys .
* 78 * ( 1997 ) 2300 [ arxiv : hep - ph/9610252 ] .
+ y. grossman , y. nir and m. p. worah , phys .
b * 407 * ( 1997 ) 307 [ arxiv : hep - ph/9704287 ] .
m. ciuchini _ et al .
_ , nucl . phys .
b * 534 * ( 1998 ) 3 [ arxiv : hep - ph/9806308 ] .
+ a. j. buras _ et al .
_ , phys .
b * 500 * ( 2001 ) 161 [ arxiv : hep - ph/0007085 ] .
r. fleischer , g. isidori and j. matias , jhep * 0305 * ( 2003 ) 053 [ arxiv : hep - ph/0302229 ] .
+ g. isidori , arxiv : hep - ph/0307014 . | by using the most recent determinations of the several theoretical and experimental input parameters , we update the unitarity triangle analysis in the standard model and discuss the sensitivity to new physics effects .
we investigate the interest of measuring with a better precision the various physical quantities entering the unitarity triangle analysis and study in a model independent way whether , despite the undoubted success of the ckm mechanism in the standard model , the unitarity triangle analysis still allows the presence of new physics . |
the correlation energy is a central quantity in quantum chemistry .
it is usually defined as the error in the electronic energy calculated within the independent - particle model of hartree fock ( hf ) theory with respect to the exact solution of the electronic schrdinger equation @xcite , @xmath4 the exact electronic energy @xmath5 can be obtained from the full configuration interaction ( fci ) approach . although there exists no rigorous distinction between different types of electron correlation effects , the correlation energy is typically divided into three categories : _ dynamic _ , _ static _ and _ nondynamic _ @xcite . the dynamic correlation energy is considered to be responsible for keeping electrons apart and is attributed to a large number of configurations ( determinants ) with small absolute weights in the wave function expansion , while the nondynamic and static contributions involve only some determinants with large absolute weights which are necessary for an appropriate treatment of the quasi - degeneracy of orbitals @xcite . in particular , static electron correlation embraces a suitable combination of determinants to account for proper spin symmetries and their interactions , whereas nondynamic correlation is required to allow a molecule to separate correctly into its fragments @xcite . over the past decades
, a number of quantum chemical methods has been developed to accurately describe either dynamic or nondynamic / static correlation effects .
for instance , mller
plesset perturbation theory and single - reference coupled cluster ( cc ) theory @xcite are successful in capturing dynamic correlation effects in single - reference cases , while the complete active space self consistent field ( casscf ) approach @xcite is suitable to describe static correlation in multi - reference problems .
still , the neglect of one kind of correlation effects may lead to non - negligible errors , which led to the development of `` hybrid '' approaches , among which are the complete active space second - order perturbation theory ( caspt2 ) @xcite and the multi - reference cc ansatz @xcite , both with their own intrinsic limitations . _ a priori _ knowledge about the interplay of dynamic , nondynamic and static electron correlation effects is required to select an appropriate electron correlation method in order to obtain reliable results .
this issue becomes most severe when spectroscopic accuracy of , say , 0.01 ev for relative energies is desired .
the consideration of dynamic , static and nondynamic correlation effects on an equal footing still remains a challenge for quantum chemistry .
the quality of single- and multi - reference quantum chemical methods can be estimated by a number of diagnostic tools @xcite .
examples are the absolute or squared weights of the reference or principal configuration ( the @xmath6 coefficient ) obtained from a ci calculation @xcite and the euclidean norm of the @xmath7 amplitudes , which are denoted as @xmath8 @xcite and @xmath9 diagnostics @xcite in cc and perturbation theory , respectively . related
are the @xmath10 and @xmath11 measures based on single and double excitations in single - reference cc theory @xcite .
a conceptually different group of diagnostic measures is based on concepts from quantum information theory and exploits knowledge about the one - particle reduced density matrix in terms of natural occupation numbers @xcite , the two - particle reduced density matrix or its cumulant in terms of the frobenius norm @xcite , the weights from excited configurations of some wave function expansion , @xcite and the distribution of effectively unpaired electrons @xcite .
a complementary classification of electron correlation effects that exploits entanglement measures among molecular orbitals was recently proposed by us @xcite .
our analysis is based on the assessment of the entanglement among any pair of orbitals and the entanglement of one orbital with all other orbitals , respectively , as encoded in a fci - type wave function .
an in - depth study of iron nitrosyl complexes @xcite , featuring complicated electronic structures , showed that the static , nondynamic , and dynamic contributions to the correlation energy and , as a consequence , the single- and multi - reference nature of a quantum system , can be distinguished by examining the entanglement patterns of the orbitals .
our entanglement analysis comprises two entropic measures : ( i ) the single - orbital entropy@xcite , @xmath12 ( @xmath13 denotes the four different occupations of a spatial orbital ) which quantifies the entanglement between one particular orbital and the remaining set of orbitals contained in the active orbital space from the eigenvalues @xmath14 of the one - orbital reduced density matrix @xmath15 of a given orbital . and , ( ii ) the mutual information@xcite , @xmath16 which measures the entanglement of two orbitals @xmath17 and @xmath18 embedded in the environment of all other active - space orbitals .
@xmath19 is the two - orbital entropy between a pair @xmath20 of orbitals , which is calculated from the eigenvalue of the two - orbital reduced density matrix @xmath21 ( analogously to eq .
( [ eq : s1 ] ) , but with @xmath13 now enumerating the 16 possible no- , one- , and two - electron states defined on the pair of orbitals ) , and @xmath22 is the kronecker delta . the one- and two - orbital density matrices can be determined from a many - particle density matrix by tracing out all many - electron states defined on the active - space orbitals that complement those orbital @xmath17 ( and @xmath18 , respectively ) whose entanglement with the others shall be studied @xcite . the general occupation - number - vector expansion of the electronic wave function constructed from @xmath23 active - space orbitals @xmath24 denoting the occupation of orbital @xmath17 )
may be decomposed into states @xmath25 and @xmath26 defined on a system comprising the single or pair of orbitals , respectively , and into those of its environment defined on the remaining orbitals . if , for the system consisting of one orbital @xmath17 , we split the environment states @xmath27
( @xmath27 then being a composite index ) as @xmath28 and @xmath29 , we can write the total electronic state as @xmath30 for the case of the system consisting of two orbitals @xmath17 and @xmath18 , we may split the environment states @xmath27 as @xmath31 , @xmath32 , and @xmath33
. then , the total electronic state reads in this basis @xmath34 the one- and two - orbital density matrix operator can now be expressed as @xmath35 and @xmath36 respectively . from these operators
one may derive the one- and two - orbital density matrices , which are calculated from the expansion coefficients @xmath37 and @xmath38 , respectively .
the total quantum information embedded in a wave function can be calculated from the set of single - orbital entropies @xcite @xmath39 as found in ref . ,
large single - orbital entropies ( @xmath40 ) and large values for the mutual information @xmath41 indicate orbitals which are important for nondynamic correlation effects .
medium - sized single - orbital entropies ( @xmath42 ) together with moderately entangled orbitals ( medium values of @xmath41 ) and small single - orbital entropies ( @xmath43 ) accompanied by small values of @xmath41 can be attributed to static and dynamic correlation effects , respectively .
an advantage of our approach is that it reduces electron correlation effects encoded in a wave function ( optimized by any quantum chemical method ) to quantities defined for the orbital basis , which can then be easily compared to each other .
the aim of this paper is twofold .
first , it provides a computational prescription for the calculation of the entanglement measures introduced in refs . and exploited in ref . , which is given in section [ sec :
second , these entanglement measures are discussed in the context of bond - forming ( or equivalently , bond - breaking ) processes in section [ sec : bond ] ( section [ comput_det ] contains the computational details ) , where we demonstrate how the single - orbital entropy can be employed to monitor the cleavage of chemical bonds .
conceptual understanding of electronic structures in terms of entangled orbitals results and a pictorial representation of how many bonds ( single , double , triple , etc . ) are formed between two atoms emerges from the single - orbital entropy diagrams .
we focus on a benchmark set of small diatomic molecules , which constitute paradigms of both single- and multi - reference problems : the n@xmath44 , f@xmath44 and csh molecules .
the matrix representation of the one and two - orbital reduced density matrices @xmath15 and @xmath21 introduced above can be constructed from fermionic correlation functions @xcite or from generalized correlation functions@xcite . in the following , we present the formalism relying on the latter ones . for a spin-1/2 fermionic model , like the one under consideration here , single - electron basis states ( orbitals )
can be empty , occupied with an @xmath13-(spin - up ) or a @xmath45-(spin - down ) electron , or doubly occupied with two electrons of paired spin .
these states we denote as @xmath46* * -0.31 cm @xmath47 , @xmath46* * -0.31 cm @xmath48 @xmath47 , @xmath46* * -0.31 cm @xmath49 @xmath47 , and @xmath46* * -0.364 cm @xmath50 @xmath47 , respectively .
since the local basis is four dimensional , 16 possible operators @xmath51 arise , @xmath52 which operate on the basis states of a single orbital @xmath17 ( with @xmath53 being the four - dimensional unit matrix and the action of @xmath54 as summarized in table [ tab : fermion - ops ] ( for @xmath55 ) ) .
note that the dimension of @xmath54 is four , while the ( full ) operator @xmath51 acting on the total state is 4@xmath56 .
c|cccc @xmath54 & * * -0.31 cm & * * -0.31 cm @xmath48 & * * -0.31 cm @xmath49 & * * -0.364 cm @xmath50 + * * -0.31 cm & @xmath57 & @xmath58 & @xmath59 & @xmath60 + * * -0.31 cm @xmath48 & @xmath61 & @xmath62 & @xmath63 & @xmath64 + * * -0.31 cm @xmath49 & @xmath65 & @xmath66 & @xmath67 & @xmath68 + * * -0.364 cm @xmath50 & @xmath69 & @xmath70 & @xmath71 & @xmath72 + the structure of the @xmath73 one - orbital operators @xmath54 is rather trivial : each operator contains a single element being equal to one at matrix position @xmath74 , @xmath75 where @xmath76 is the initial state and @xmath77 is the final state .
it thus acts like a transition matrix from state @xmath76 to @xmath77 .
explicitly , we may write the one - orbital operator @xmath54 as @xmath78 such that the matrix elements of @xmath54 can be expressed as a kronecker delta , @xmath79),m}$ ] for @xmath55 , and the one - orbital states are labeled by @xmath80 and @xmath81 . for the many - electron wave function in occupation - number - vector representation
the multi - site reduced density matrices can be expressed using products of @xmath51 operators acting on specific sites .
the one - orbital reduced density matrix @xmath15 can then be calculated by taking those operators which do not change the single - orbital basis state @xcite , @xmath82 where the expectation value is calculated from the total electronic state .
we may abbreviate this structure in tabular form as summarized in table [ tab : rho1-fermion ] .
once the one - orbital reduced density matrix @xmath15 is constructed , @xmath83 can be determined from its eigenvalues @xmath84 according to eq .
( [ eq : s1 ] ) .
.expressing the single - orbital reduced density matrix , @xmath85 , in terms of single - orbital operators , @xmath86 , defined in table [ tab : fermion - ops ] . for better readability only the operator number indices , @xmath87 are shown corresponding to @xmath88 . [ cols="^,^,^,^,^",options="header " , ] table [ tab : energiescsh ] summarizes all electronic energies obtained from dmrg calculations for the dissociation of the csh molecule . as aforementioned , the maximum number of renormalized active system states was set to 1024 with a quantum information loss of @xmath89 , while the minimum number of renormalized active system states was increased as mentioned in the table until convergence with respect to @xmath87 was reached .
again , the maximum number of renormalized active - system states was kept for at most one microiteration step , while the quantum information loss was considerably smaller than the chosen threshold of 10@xmath90 .
hence , increasing @xmath91 would result in an energy decrease of considerably less than 1 mhartree as observed for the n@xmath44 molecules .
lccccccc & + method & @xmath92 & @xmath93 & @xmath94 & @xmath95 + + @xmath96 & @xmath977783.946 277 & @xmath97 & @xmath977783.884 358 & @xmath977784.900 746 + @xmath98 & @xmath977783.970 497 & @xmath977783.949 555 & @xmath977783.907 986 & @xmath977784.905 747 + @xmath99 & @xmath977783.971 143 & @xmath977783.950 064 & @xmath977783.908 452 & @xmath977784.906 309 + for dmrg(10,46)[512,1024,@xmath89 ] calculations for the n@xmath44 molecule at different internuclear distances .
the orbitals are numbered and sorted according to their ( casscf ) natural occupation numbers .
each orbital index in the @xmath83 diagram indicates one molecular orbital and corresponds to the same natural orbital as numbered in the mutual information plot .
those orbitals which are included in the cas(10,8)scf calculations are marked in red in the single - orbital entropy diagram .
the total quantum information sums up to 7.63 , and 8.26 for increasing distances . ] | the accurate calculation of the ( differential ) correlation energy is central to the quantum chemical description of bond - formation and bond - dissociation processes . in order to estimate the quality of single- and multi - reference approaches for this purpose , various diagnostic tools have been developed . in this work
, we elaborate on our previous observation [ _ j .
phys .
chem .
lett . _ * 3 * , 3129 ( 2012 ) ] that one- and two - orbital - based entanglement measures provide quantitative means for the assessment and classification of electron correlation effects among molecular orbitals .
the dissociation behavior of some prototypical diatomic molecules features all types of correlation effects relevant for chemical bonding .
we demonstrate that our entanglement analysis is convenient to dissect these electron correlation effects and to provide a conceptual understanding of bond - forming and bond - breaking processes from the point of view of quantum information theory .
* orbital entanglement + in bond - formation processes * katharina boguslawski@xmath0 , pawel tecmer@xmath0 , gergely barcza@xmath1 , rs legeza@xmath2 , and markus reiher@xmath3 + @xmath0eth zurich , laboratorium fr physikalische chemie , + wolfgang - pauli - strasse 10 , 8093 zurich , switzerland + @xmath1wigner research center for physics , strongly correlated systems + `` lendlet '' research group , h-1525 budapest , hungary + |
one important clue about the nature of quantum gravity is that the einstein field equation and quantum field theory in curved spacetime imply that black holes behave as thermodynamical objects consistent with the four laws of thermodynamics .
they are endowed with an entropy proportional to the cross - sectional area of the event horizon @xcite and a temperature due to quantum hawking radiation @xcite .
although this interplay of gravitation , quantum field theory , and thermodynamics has been a focus of research for over 30 years , it has yet to be completely understood .
one fascinating idea was the proposal by jacobson @xcite to reverse the logic of black hole thermodynamics and derive the einstein equations as a consequence of spacetime thermodynamics and quantum field theory .
the idea is that the local acceleration horizons that exist through every point in spacetime are analogous to tiny pieces of a black hole event horizon and have an entropy proportional to their area .
the equivalence principle is then invoked to view the local neighborhood around any point in a general curved spacetime as a piece of flat spacetime . even in flat spacetime accelerated observers can never receive information from certain regions . for these observers , quantum fields
are localized to the rindler wedge of flat spacetime and they view the usual minkowski vacuum as precisely a thermal state @xcite .
this supplies the notion of a local temperature . by demanding the clausius relation @xmath7 at every point , where @xmath2 is the flow of energy across the horizon and
@xmath3 is the unruh temperature , jacobson was able to show the einstein equation appears as an equation of state .
this derivation supports the idea that macroscopic spacetime dynamics is just the thermodynamics of the quantum vacuum . in the past several years other related
work has shown that in certain spacetimes ( spherically symmetric , axisymmetric , and cosmological ) the gravitational field equations near a horizon can also be re - expressed as the thermodynamic identity @xmath8 @xcite .
these results have deep implications , but their true significance ( if any ) is not yet clear .
recently , in order to probe jacobson s derivation further , a horizon entropy proportional to a function of the ricci scalar was considered @xcite .
the idea was to determine whether higher curvature correction terms that one expects from effective field theory to appear in the gravitational field equation ( or action ) can be derived from the thermodynamical prescription .
it was found that the field equations of @xmath9 gravity can be derived only if the setting is shifted from equilibrium to non - equilibrium thermodynamics .
an extra internal entropy production term proportional to the squared expansion of the horizon is required for the clausius relation ( or now more properly the _ entropy balance law _ ) to hold .
perhaps more importantly , it was realized that the original derivation of the einstein equation via an area entropy can be easily generalized to a non - equilibrium setting . in this case
an entropy production term proportional to the squared shear of the horizon appears when the entropy balance law holds .
consistency with the einstein equation requires that the ratio of the shear term coefficient to area entropy density be @xmath10 . ] .
the purpose of this paper is to explore the meaning of these additional terms .
the existence of entropy production terms proportional to squared shear and expansion is reminiscent of the shear and bulk viscosity terms that appear in viscous fluid hydrodynamics .
a connection between horizon dynamics and fluid dynamics was first noticed by damour @xcite . in the 1980 s price , thorne , and
collaborators further developed this picture into the black hole
membrane paradigm " @xcite , using a timelike _ stretched horizon _ to approximate the null horizon .
the shear and expansion of the horizon also appear in the membrane paradigm , where they are interpreted to be viscous terms in the law describing how the horizon area / entropy changes .
however , this interpretation is just an analogy and it is not clear if there is a deep relationship between the dynamics of causal horizons and hydrodynamics .
if there is a relationship we must not only understand horizons as a fluid system , but also the physics of this system must be consistent with hydrodynamics as an effective theory . in hydrodynamics , viscosities are phenomenological coefficients in the linear constitutive relations between fluxes of momentum in the fluid and the thermodynamic forces " given by gradients of the fluid velocity
. these velocity gradients must be small ( both in space and time ) compared to some microscopic scale for hydrodynamics to be an accurate description of the near - equilibrium physics .
after reviewing the thermodynamic derivation in detail we will show that it can be consistently reformulated in the language of hydrodynamics .
we argue the minkowski vacuum , which is a thermal state when localized in the rindler wedge of spacetime , is holographic @xcite ; its properties are encoded into the 2 + 1 dimensional system near the rindler horizon .
this is because the degrees of freedom in this thermal atmosphere are essentially piled up near the horizon boundary .
the velocity and temperature gradients associated with these degrees of freedom can be made small enough so that hydrodynamics is the appropriate effective theory .
therefore we think of these thermal atmosphere degrees of freedom as a fluid living on a stretched horizon approximating the local acceleration horizon . using hydrodynamics and the properties of the stretched horizon
, we will derive the entropy balance law postulated in @xcite , showing that it is appropriate to interpret the coefficients of the shear and expansion terms in this law as shear and bulk viscosities respectively .
just as in the thermodynamic derivation , the entropy balance law requires that spacetime dynamics is governed by the einstein equation and that the shear viscosity to entropy density ratio is @xmath10 . therefore in this more general formulation einstein s equation
is a consequence of the hydrodynamics of spacetime vacuum .
the value @xmath10 has attracted considerable attention over the past few years from work on the celebrated anti - de sitter ( ads ) /conformal field theory ( cft ) correspondence @xcite .
this correspondence states that certain gauge theories in flat spacetime are equivalent to a quantum gravity theory in a higher dimensional ads spacetime .
hydrodynamics arises as the description for the dynamics of long wavelength perturbations about equilibrium in high temperature gauge theory plasmas .
working on the gravity side of the duality it has been found that @xmath10 is the value of the shear viscosity to entropy density ratio for all gauge theories with an einstein gravity dual @xcite .
kovtun , son , and starinets conjectured that these theories may saturate a universal lower bound on the ratio @xcite , but the true significance of the ratio is unclear and the existence of a viscosity bound is controversial @xcite .
for instance , it is not clear why the ratio , derived using relativistic field theories , is independent of the speed of light @xmath11 . similarly ,
if gravity is somehow involved in saturation of the bound , why does @xmath12 not appear ? here , since the ratio holds universally for any local acceleration horizon , it appears more fundamental than previous results involving ads spacetimes .
it seems that @xmath10 is the shear viscosity to entropy density ratio of the local vacuum as a thermal state .
we discuss the significance of this result and its connection to the gauge / gravity literature .
we conclude by examining open questions and possible extensions of our work .
we now review the derivation @xcite of the einstein equation as the equation of state arising from the thermodynamics of local horizons .
the motivating idea is that the origin of the thermodynamic behavior of black holes is rooted in the thermodynamic behavior of the local vacuum . in minkowski spacetime
when quantum fields are restricted to a rindler wedge @xmath13 , the vacuum density matrix takes the form of a thermal state @xmath14 @xcite .
notice that neither the boost hamiltonian " @xmath15 nor boost temperature " @xmath16 have dimensions of energy .
this is because @xmath15 generates translations of dimensionless hyperbolic boost angle .
when @xmath15 is rescaled to generate proper time along a worldline with proper acceleration @xmath17 , @xmath18 is rescaled to the usual unruh temperature @xmath19 @xcite .
the null surface @xmath20 forming the edge of the rindler wedge is one part of the boundary of the past for the bifurcation plane @xmath21 , and therefore acts as a causal horizon .
accelerated observers in the rindler wedge can only access information on spacelike slices bounded by the bifurcation plane . since vacuum fluctuations are correlated between the inside and outside of the wedge , these observers will see an entanglement entropy . in a continuum field theory
this entropy scales with the area of the boundary , but is divergent because of the ultraviolet ( uv ) divergence in the density of states .
when a uv regulator is introduced the entropy is proportional to the area of a horizon cross - section , with a proportionality constant that could depend on the nature and number of the quantum fields @xcite .
this motivates jacobson s assumption of a universal entropy density @xmath22 per unit horizon area , with @xmath22 possibly dependent on the field content .
a horizon entropy @xmath23 will be contributed by a little horizon patch of area @xmath24 . when the thermal density matrix @xmath25 at temperature @xmath18 is perturbed , the change in entanglement entropy is related to the change in mean energy via ds = /t_b .
because the change in the mean energy is due to the flux into the unobservable region of spacetime , which is perfectly thermalized by the horizon system , it is assumed to consist entirely of heat .
thus , we have the thermodynamic clausius relation @xmath26 .
jacobson s second assumption was this relation should hold for all causal horizons , with @xmath2 as the flow of boost energy across the horizon and @xmath1 the change in area entropy .
since the area of the horizon is no longer fixed , the spacetime must become dynamical .
now that the spacetime is no longer flat everywhere a _ local _
horizon is defined in analogy with a black hole horizon .
a global definition of the latter is the boundary of the past of future null infinity .
the segment of a black hole horizon to the past of a spatial cross - section is the boundary of the past of that cross section .
a local horizon at a point @xmath27 is defined in a similar way : choose a spacelike 2-surface patch @xmath28 including @xmath27 , and choose one side of the boundary of the past of @xmath28 . near @xmath27
this boundary is a congruence of null geodesics orthogonal to @xmath28 .
these comprise the horizon . at @xmath27
one can invoke the equivalence principle to view the spacetime in the neighborhood of @xmath27 as approximately flat . in this small patch
the idea is to construct a future pointing approximate boost killing vector @xmath29 that vanishes at @xmath27 and whose flow leaves the tangent plane to @xmath28 at @xmath27 invariant .
the normalization of @xmath29 is chosen so that @xmath30 .
the construction can be done explicitly by solving killing s equation @xmath31 order by order in riemann normal coordinates @xmath32 .
eventually , at @xmath33 , no solution exists because a general curved spacetime has no killing vectors . up to this ambiguity
, our notion of time along the horizon is given by the parameter @xmath34 such that @xmath35 .
this killing time is related to the affine parameter along the horizon generators by @xmath36 , so the point @xmath27 is located at infinite killing time and @xmath37 .
the expansion @xmath38 and shear @xmath39 of the horizon in terms of killing time are related to the expansion @xmath40 and shear @xmath41 in affine time as follows = e^-v = - , = e^-v = - .
[ falloff]thus , the killing expansion and shear vanish at @xmath27 ( as long as the affine quantities are not diverging ) and the horizon area is instantaneously stationary at this point .
this defines our notion of equilibrium .
the system is defined as the degrees of freedom just behind the horizon and we will consider transitions that terminate in the equilibrium state at @xmath27 .
we define the heat as the flux of the boost energy current of matter across the horizon , q = t^m_^d^ , where @xmath42 is the matter stress tensor .
this and all subsequent integrals are taken over a thin pencil of horizon generators centered on the one that terminates at @xmath27
. it will be convenient to work in terms of affine parameter and the affinely parameterized horizon tangent vector @xmath43 . using the relation @xmath44 and the definition of @xmath18 we thus have = t^m _ k^k^(- ) dd^2a .
[ dq / t ] to compute the entropy change @xmath23 we must follow the area change of the horizon , a = d d^2a , [ da ] where @xmath45 is the expansion of the congruence of null geodesics generating the horizon . using the raychaudhuri equation = -^2-_^-r_k^k^[ray]the entropy change
is thus given up to @xmath46 by s = s dd^2a , [ ds]where all quantities in the integrand are evaluated at @xmath27 .
in @xcite jacobson chose @xmath47 at @xmath27 , which is required for equilibrium if the affine parameter @xmath48 is assumed to be the natural time " of the system . here and in @xcite we consider the killing parameter @xmath34 to be the natural time , so vanishing affine expansion and shear at @xmath27 are not necessarily needed a priori for our notion of equilibrium .
on the other hand , if it is required that @xmath49 at all points @xmath27 and for all null vectors @xmath43 , we first find that the affine expansion at @xmath27 must vanish since the heat flux ( [ dq / t ] ) vanishes at @xmath27 . at @xmath50
the integrands of ( [ dq / t ] ) and ( [ ds ] ) then imply ( 2/ ) t^m_k^k^= s ( _ ^+ r_k^k^ ) .
[ clausiusfinal]note that the shear squared term can be written in terms of derivatives of @xmath43 , which can be independently chosen at @xmath27 .
therefore the @xmath43 derivative part of ( [ clausiusfinal ] ) implies the shear must also vanish at @xmath27 if the clausius relation is to hold .
however , ( [ falloff ] ) tells us the shear and expansion with respect to killing time fall off to zero at @xmath27 as @xmath51 when @xmath40 and @xmath41 vanish , while only as @xmath52 when @xmath40 and @xmath41 are non vanishing . in @xcite
it was hypothesized that for a slower approach to equilibrium the clausius relation may not apply and thus @xmath53 . in this case
there is an entropy balance law ds = q / t_b+d_i s [ balance]where @xmath4 represents internal entropy production for a system out of equilibrium .
as is standard in non - equilibrium thermodynamics we consider the system to be still near enough to equilibrium so that the entropy and temperature take their local equilibrium values .
the production term @xmath4 should be of @xmath50 to be consistent with the notion of equilibrium at @xmath27 .
we also assume that @xmath4 depends only on squared gradients of @xmath43 .
entropy production from the squared gradients of state variables is a universal property of non - equilibrium thermodynamics @xcite .
given these assumptions , if the conjectured balance law ( [ balance ] ) is to hold at all @xmath27 and for all @xmath43 , @xmath40 is still required to vanish and the entropy production term is d_i s = -s _ ^
dd^2 a. in terms of killing time this new term has the form d_i s = s _ ^ dv d^2 a , [ shearprod ] which looks like the standard @xcite entropy production term for a fluid with shear viscosity @xmath54 , d_i s = _ ^ dv d^2 a , if we identify @xmath55 .
the remaining @xmath43 part of the entropy balance law yields r _
= ( 2/s ) t^m _ [ eos ] where @xmath56 is a so far undetermined function .
this corresponds to the tracefree part of the einstein equation , with newton s constant determined by the universal entropy density @xmath22 , g_n=. conversely , @xmath57 , so the entropy is identified as one quarter the area in planck units , like the bekenstein - hawking black hole entropy .
the free function @xmath56 can be fixed if it is assumed that the matter stress tensor is divergence free , corresponding to the usual local conservation of matter energy . taking the divergence of both sides of ( [ eos ] ) and using the contracted bianchi identity @xmath58
we then find that @xmath59 , corresponding to the einstein equation with ( undetermined ) cosmological constant @xmath60 .
as noted above , the shear squared entropy production term ( [ shearprod ] ) is very similar to the entropy production term due to the shear of the fluid flow in viscous hydrodynamics .
however , at this stage it is not at all clear whether it makes sense to identify @xmath54 as shear viscosity .
the shear in ( [ shearprod ] ) is a gradient of a null horizon tangent vector .
can we think of this horizon tangent as a flow velocity and a cross - section of the horizon as a fluid " system ? also , are we in a near - equilibrium regime where the horizon shear is small ( both in space and time ) compared to some microscopic length scale so that the constitutive relations are valid and hydrodynamics is an accurate description ?
it turns out the answers to these questions are in the affirmative and it is possible to _ derive _ the entropy balance law and entropy production term we postulated in ( [ balance ] ) and ( [ shearprod ] ) from hydrodynamics . before we start , it will be necessary to briefly review ( relativistic ) hydrodynamics . in modern terminology , hydrodynamics is an effective theory describing the dynamics of perturbations about an equilibrium state on long length scales .
although the fluid as a whole is not in equilibrium , we assume it is close enough to it such that equilibrium states with a local temperature @xmath3 exist at every point .
long " wavelengths mean long compared to a relevant microscopic scale for the fluid .
this is normally taken as the mean free path @xmath61 , which sets the characteristic length scale over which a system equilibrates locally . in this
regime quantum fluctuations are suppressed and the theory is classical . since this is an effective theory our knowledge of the fluid is restricted to a finite set of variables : typically the equilibrium proper energy density @xmath62 and pressure @xmath63 , temperature @xmath64 and local fluid velocity @xmath65 , where @xmath66 . the hydrodynamic equations are simply the conservation of energy and momentum @xmath67 in the simplest case where there are no other conserved currents @xmath68 and the relativistic chemical potential is zero .
the stress tensor @xmath69 is a function of the fluid variables and has the form of an equilibrium perfect fluid plus a dissipative part @xmath70 t^ = ( ( t)+p(t))g^+p(t ) u^u^+ ^(t , u , t ) .
this form of the stress tensor alone is not sufficient to determine the dynamics .
however , if @xmath71 and @xmath72 are @xmath73 , the dissipative part of the stress tensor @xmath70 is smaller than the zeroth order perfect fluid part and can be expanded in terms of the derivatives of the fluid velocity . at each point
@xmath74 there is the freedom to boost @xmath65 such that @xmath75 . with this standard choice of gauge , at first ( linear ) order
the corrections depend explicitly on the velocity derivatives and have the form @xcite ^ = p^ p^ [ ( _ u_+ _ u_- g _
_ u^ ) + g _ _ u^ ] where @xmath76 and we have considered a fluid of 3 spatial dimensions . in this landau gauge " the spatial parts of the stress tensor are related to momentum fluxes .
the shear viscosity @xmath54 and bulk viscosity @xmath77 in this constitutive relation are phenomenological coefficients at this level and are determined by experiment or matching to the complete microscopic theory .
the derivative expansion can in principle be continued on to higher orders , with each extra term suppressed by powers of @xmath78 , where @xmath79 is the characteristic length scale of variations in @xmath80 and @xmath3 .
returning to the problem at hand , what can we take to be the fluid system ? as stated above , an obvious candidate for the fluid " here is the local acceleration horizon itself , with @xmath43 as the fluid velocity .
but the technical drawback is the horizon is a null surface and @xmath43 is a null vector instead of being unit timelike .
however , we can employ the notion of a stretched horizon pioneered by price and thorne to describe the physics of globally defined black hole event horizons @xcite .
the stretched horizon is a timelike surface that lives close enough to the event horizon that it can capture its essential physics .
more specifically , the idea is to perform a 2 + 1 + 1 split of spacetime . the foliating spacelike surfaces are surfaces of constant time according to a family of accelerated observers with 4-velocity @xmath81 defined such that @xmath82 for lapse function @xmath83 . in the familiar special case of a schwarzschild geometry
@xmath84 and @xmath85 are surfaces of constant schwarzschild time .
the event horizon itself is characterized by the null generator @xmath86 and can be foliated into spacelike 2-surfaces by surfaces of constant horizon time . in the schwarzschild example @xmath87 and @xmath88 , where @xmath34 is the killing time on the horizon .
the distance from the horizon is naturally parameterized by the affine parameter along the ingoing null rays ( for example just the radial coordinate @xmath89 in schwarzschild ) , or @xmath83 equivalently by a change of variables .
the stretched horizon is defined as a surface of fixed constant lapse ( or radial coordinate ) such that @xmath90 .
the stretched horizon itself has unit spacelike normal @xmath91 .
since this vector field can be extended throughout the spacetime as the normal to all surfaces of constant @xmath83 , we have a 2 + 1 + 1 split defined by @xmath92 and @xmath93 .
we will always work in the limit of the true horizon @xmath94 , where u^ l^ + n^ l^ [ vectorlimit ] a local stretched horizon in the limit where @xmath94 will be used to approximate the acceleration horizon , which is the boundary of the past of @xmath28 .
as shown in figure 1 , the stretched horizon lives just inside " the true causal / acceleration horizon .
representation of how the stretched horizon @xmath95 approximates the past horizon @xmath96 .
note that 2 spatial dimensions are suppressed .
each point in the stretched horizon represents the fluid at constant time @xmath97 . as @xmath94 , points in the stretched horizon are mapped to points in the past horizon along the ingoing null rays as indicated . a point at infinite @xmath97
is mapped to the bifurcation point @xmath27.,title="fig:",width=166 ] + the fluid itself lives on the spacelike cross section of the stretched horizon defined with the fluid velocity @xmath92 .
although we have formally identified the stretched horizon system as a fluid " , one may wonder if this choice has any physical interpretation .
since the stretched horizon is a 2 + 1 dimensional surface , the fluid must be 2 + 1 dimensional . in the literature , @xcite considered the null vector @xmath98 associated with null hypersurfaces as an elastic displacement vector of a spacetime solid " in a long wavelength limit .
the entropy of the solid is assumed to be a quadratic functional of derivatives of this vector . when this entropy is maximized
gravitational field equations can be obtained .
the idea of a fluid living on a lower dimensional surface is also consistent with the hydrodynamic limit of ads / cft correspondence .
however , there the fluid is taken to be a real gauge theory fluid living on the timelike boundary of ads spacetime .
is there a real fluid on the stretched horizon here ?
we argue that the minkowski vacuum , which looks like a thermal state when localized into the rindler wedge of spacetime , obeys the holographic principle @xcite .
its properties are encoded into the 2 + 1 dimensional stretched horizon boundary of the wedge .
some similar ideas can be found in @xcite .
this appears counterintuitive at first because the vacuum in the rindler wedge looks like a 3 + 1 dimensional bath of thermal radiation .
however , the entanglement entropy associated with the vacuum is ( in the absence of a uv cutoff ) a formally divergent quantity that scales like the cross - sectional area of the horizon boundary , not the volume of the wedge .
heuristically , the entropy density of the radiation bath goes like @xmath99 .
the key point is that @xmath3 is a local unruh temperature that is a function of the proper length to the horizon : @xmath100 .
the total entropy is s = ^-3 dd^2 a ~a _ c^-2 , where @xmath24 is the cross - sectional area of a horizon patch and @xmath101 is the uv cutoff length .
a similar calculation for total energy using density @xmath102 also yields an area scaling and stretched horizon energy density @xmath103 . since these quantities scale like area densities instead of the usual volume densities they will not be extensive unless we identify them with the stretched horizon boundary surface .
the degrees of freedom in the vacuum thermal state are effectively packed into the stretched horizon .
further evidence for this picture is brustein and yarom s work @xcite showing that vacuum fluctuations in any sub - volume of minkowski space scale as the area of the boundary and diverge unless there is a uv cutoff .
they used this result to argue that these fluctuations have a representation in terms of a high temperature theory on the boundary , which in the case of the rindler wedge , is reminiscent of the near - horizon thermal atmosphere with its diverging local temperature . in light of the heuristic arguments above and these results in the literature
, we postulate that if hydrodynamical perturbations of the thermal atmosphere exist they should be manifested in the dynamics of a stretched horizon .
in particular , in the hydrodynamic limit the degrees of the freedom in the thermal atmosphere can be represented as a real fluid living on the boundary .
we now examine the dynamics of the horizon fluid in detail .
first we discuss the fluid in equilibrium . as in section [ thermo ] around the arbitrary point @xmath27 in @xmath28
we can invoke the local flat spacetime approximation to define the local vacuum state . around this point
we have the approximate set of poincare symmetries , including boosts generated by killing vector @xmath104 , which is defined to vanish at @xmath27 . using rindler coordinates adapted to this boost killing field @xmath105
, the metric in the neighborhood of @xmath27 has the approximate form ds^2 - ^2 ^2 d ^2 + d ^2 + dx^2 + dy^2 .
[ rindler]the lapse function @xmath106 , where @xmath107 is an arbitrary constant associated with the normalization of boost time @xmath97 . in section [ thermo ] , @xmath107 was scaled to be unity and @xmath97 was a dimensionless boost angle .
the local rindler ( killing ) horizon at @xmath108 can be approximated by its own stretched horizon for constant @xmath94 .
the killing vector @xmath104 describes the horizon fluid rest frame .
however there is the freedom to boost to a moving frame in the @xmath109 directions & = & ( -^i x_i ) + x^i & = & ( x^i - ^i ) , [ boost ] to characterize a moving horizon fluid . in this state the flat spacetime ( boosted ) rindler horizon has fixed area and as expected the entropy is unchanging .
just as in section [ thermo ] we identify the unruh temperature @xmath110 with the local equilibrium temperature .
notice this has the tolman law form @xmath111 , where @xmath112 is analogous to a position independent hawking temperature . in this equilibrium state
we expect the fluid is described by a surface stress tensor in the perfect fluid form t^s _ = ( + p)u_u_+ p _ [ stress ] where @xmath113 and the superscript @xmath95 indicates this is a surface tensor . just like entropy density @xmath22 ,
the surface energy density @xmath114 and pressure @xmath63 are formally divergent quantities that may depend on the number and nature of fields in the thermal atmosphere .
we will allow for a uv cutoff length @xmath101 , whose value is initially unknown , which will render all quantities finite .
the stretched horizon boundary metric ( ( [ rindler ] ) with @xmath115 fixed ) is flat and invariant under translations in time and space .
these local translational symmetries in the boundary imply the surface stress tensor is conserved . using the thermodynamic relations @xmath116 , @xmath117 , and @xmath118 we find the entropy density current _
( s u^ ) = 0 [ conserveds]is conserved , as expected .
the stretched horizon system of section [ prelim ] and the equilibrium fluid do not agree in general : the fluid velocity @xmath119 is not proportional to @xmath104 except as @xmath120 at @xmath121 .
we have chosen this point because in the limit @xmath122 , it approaches the bifurcation point at @xmath27 along the null ray shown in figure [ horizons ] .
this supplies the notion of local equilibrium in the general fluid .
for this non - equilibrium horizon fluid entropy is created externally via heat flux from the outside of the system , and internally from the friction of expansions and shears .
this implies the horizon area is not fixed , the entropy current in ( [ conserveds ] ) is not conserved , and the spacetime can no longer be exactly flat . to parameterize the near - horizon curved metric
we follow the construction used by @xcite to study perturbations of black brane metrics and assume the previously constant @xmath107 in ( [ rindler ] ) and boost parameter @xmath123 in ( [ boost ] ) are functions of stretched horizon coordinates @xmath124 : & & ( x^ ) + u^&= & ^-1 ( x^ ) ( + ^i(x^ ) ) . @xmath125 and
the boost parameter @xmath126 will approach constant values at @xmath127 , where there is no entropy production and the expansion and shear must vanish . for hydrodynamics to be
an applicable description , the horizon gradients @xmath128 ( or equivalently of @xmath129 ) and @xmath130 ( or of @xmath131 ) in the local rindler coordinates need to be @xmath132 at @xmath127 . by dimensional analysis
the inverse mean free path of this thermal state is position dependent and @xmath133 , where @xmath134 is an unknown dimensionless parameter the diverging temperature will be much larger than any other scale . ] .
the gradients and the inverse mean free path are divergent as we approach the true causal horizon @xmath94 , but their ratios are finite .
the horizon gradient of the local temperature is @xmath135 , while eqn .
( [ vectorlimit ] ) implies that the gradient has the form _
l^. thus , we need @xmath136 where now @xmath137 for horizon killing time @xmath34 .
this criterion is clearly satisfied for derivatives in @xmath34 .
this can be seen because the local equilibration time for the system is @xmath138 , while the process is assumed to occur for an _ infinite _ amount of killing time before terminating in the equilibrium state .
furthermore in section [ thermo ] there was no requirement on the size of the changes in @xmath139 directions of the horizon fluid .
the stretched horizon cross - section at @xmath140 ( or equivalently the 2-surface @xmath28 as @xmath94 ) can be tuned so that the changes in @xmath141 and @xmath142 are @xmath143 near @xmath27 .
thus , there is no obstruction to working in the hydrodynamic regime and therefore an order by order expansion in derivatives is justified . in the next subsection
we will use the equations of hydrodynamics and the properties of stretched horizons to derive the near - equilibrium entropy balance law ( [ balance ] ) postulated in section [ thermo ] . following our above review of hydrodynamics
, we can proceed to add a dissipative part to the perfect fluid stress tensor ( [ stress ] ) and expand it in derivatives of the flow velocity . using conservation of the stress tensor in the stretched horizon , the thermodynamic relations @xmath144 , @xmath145 , @xmath118 , and making the gauge choice @xmath146 , it follows @xcite that entropy balance law for the horizon fluid is _ ( s u^ ) = + _ ^ + ^2 , where @xmath147 and @xmath148 .
the clausius term is the flux of bulk matter energy into the fluid as heat .
we will see below that the entropy change on the left hand side of this equation is a finite quantity ; the ratios of the divergent quantities on the right hand side will be finite . integrating over a volume in the horizon fluid
we find _ ( s u^ ) dd^2a = dd^2a[entropybal1]using stokes theorem on the left hand side and then taking the limit @xmath94 along with ( [ vectorlimit ] ) yields s(v ) = dv d^2a , [ entropybal2]where the @xmath39 and @xmath38 are now the expansion and shear of the null @xmath98 .
notice how the @xmath83 s have also canceled out of the right hand side and the relativistic entropy balance law ( [ entropybal1 ] ) has been reduced to a non - relativistic form in the true horizon limit , with the left hand side just a change in total entropy in killing time .
this result agrees with the equation for the long - time " evolution of black hole entropy in the membrane paradigm @xcite , if we identify @xmath149 as a hawking temperature . what is new here is the conceptual picture of ( [ entropybal2 ] ) as a consequence of relativistic hydrodynamics
this is not present in the damour - price - thorne membrane paradigm because no hydrodynamic limit was identified . , where @xmath150 the hawking temperature .
the schwarzschild radius @xmath151 is the characteristic size of the system .
spatial gradients of velocity necessarily scale as @xmath152 so there can be no hydrodynamic limit .
] thus , @xmath54 and @xmath77 are not just analogous to viscosities ; in our framework it is consistent to identify them as the shear and bulk viscosity of the horizon fluid .
working with the bifurcation point parameterized as @xmath153 is not convenient ; therefore we change to the affine parameter @xmath154 so that @xmath27 is at the origin : @xmath155 and @xmath121 . using the relations @xmath156 , @xmath157 , @xmath158 ,
yields s( ) = - dd^2a , [ entropybal2]which is consistent with the form of the entropy balance law ( [ balance ] ) written in terms of horizon quantities .
the matter stress tensor term is the expected flux of boost energy , while the viscous terms form the internal production piece @xmath4 .
notice that in general the horizon bulk viscosity @xmath77 also appears in addition to the shear viscosity @xmath54 .
we can now proceed to expand the left hand side of ( [ entropybal2 ] ) order by order in @xmath48 just as in the original thermodynamic derivation discussed at the end of in section [ thermo ] , assuming that the entropy density @xmath159 is an undetermined quantity such that @xmath160 . at zeroth order the affine expansion at @xmath27 is again required to be zero . demanding the linear order equation hold for all null @xmath161 and at any arbitrary point @xmath27 in spacetime , along with local conservation of the bulk matter tensor @xmath162 yields the einstein equation and @xmath22 equal to @xmath163 .
thus the uv cutoff length is fixed to be @xmath164 .
the extra shear term in the area change that comes from the raychaudhuri equation ( [ ray ] ) is consistent with the balance law only if the shear viscosity to entropy density is @xmath165 , just like in the thermodynamical argument .
the bulk viscosity @xmath77 is not determined by the balance law at this order because the affine expansion must be zero at @xmath27 .
we now conclude with some remarks on the meaning and possible implications of these results .
first , to summarize , we argued that the thermodynamic derivation of the einstein field equations can consistently be reformulated using hydrodynamics .
the hydrodynamic degrees of freedom for the local vacuum state are associated with local acceleration horizons through any point .
these are closely approximated by surrogate timelike stretched horizons and can be thought of as 2 + 1 dimensional fluids .
the equivalence principle is invoked to view the neighborhood of each point as a piece of flat spacetime with a local rindler horizon .
the entropy of these horizons is fixed and they describe local equilibrium for the horizon fluid .
the unruh effect is then used to effectively assign these equilibrium states a local temperature in the limit where the stretched horizon approaches the true one . on length scales much larger than the mean free path
, hydrodynamics must be an accurate description of the physics .
when the horizon fluid is out of the equilibrium state , entropy is produced , the horizon area is no longer fixed , and the spacetime can no longer be flat .
the entropy balance law is then re - derived using the equations of hydrodynamics .
together with the local conservation of bulk energy - momentum , the balance law implies entropy changes must be governed by the einstein equation .
the einstein equation thus arises from the hydrodynamics of the local vacuum .
remarkably , this argument also fixes the entropy density and shear viscosity of the vacuum such that their ratio is @xmath10 .
our picture seems to imply that microscopic dynamics ( which could include quantum gravity below the cutoff ) leads to ( semi-)classical einstein gravity as collective hydrodynamic behavior at low energies .
some ideas in the same spirit can be found in @xcite .
what is interesting here is that some hydrodynamic properties turn out to be universal although we initially allowed for the properties of the horizon fluid to depend on the number and nature of the quantum fields and treated the viscosities as being purely phenomenological .
once the value of the the uv cutoff scale @xmath101 was fixed to be roughly a planck length , the entropy density associated with all local rindler horizons is the bekenstein - hawking entropy density and @xmath166 is universally @xmath5 .
all the dependence on the number and nature of the quantum fields is apparently absorbed into the low energy newton constant @xmath12 .
this in accord with arguments that the bekenstein - hawking entropy is dependent implicitly on the nature of quantum fields through the renormalization of the gravitational constant and is either partly or wholly the entanglement entropy of the thermal atmosphere @xcite .
these results are puzzling here since no knowledge of microscopic physics was needed to obtain them , only the balance law .
we did , on the other hand , fix the value of cutoff scale to be the planck length experimentally " by requiring the einstein equation inferred from the entropy balance law to agree with the observed einstein equation . low energy physics ( the balance law ) and this one observation turn out to be enough to determine the entropy density and the shear viscosity of the fluid .
the bulk viscosity though is one fluid property not fixed by the balance law in this case and therefore it seems one would have to know about the details of the microscopic physics in order to determine it . as we noted in the introduction @xmath10
also appears in the ads / cft literature as the universal value of the shear viscosity to entropy density ratio of gauge theories with an einstein gravity dual .
could this be merely a coincidence or is there a connection between this gauge / gravity duality result and our hydrodynamic derivation ?
first , in both cases holography is crucial : we postulated the thermal vacuum state is holographic , while ads / cft is a precise realization of the equivalence of a higher dimensional gravity theory to a lower dimensional non - gravitational theory on a boundary .
furthermore , in the duality , @xmath167 dimensional gauge theories in high temperature deconfining phases are dual to large black hole or black brane spacetimes in @xmath168 ads @xcite
. therefore one can use classical perturbations of the large black hole or black brane spacetimes ( see @xcite for a review ) to perform analytical computations of the hydrodynamic transport coefficients .
according to the ads / cft dictionary the notion of viscosity is meaningful in the infrared regime of the gauge theory , which corresponds to the near horizon limit of the translationally invariant black object . in this sense
these black objects have viscosities , just like the viscosity we found for local stretched horizons .
in both cases the hydrodynamics of a flat spacetime system is manifested in the dynamics of a horizon boundary . however , since @xmath10 holds for all local acceleration horizons it seems more fundamental than the ads / cft results for large black holes and black branes in ads spacetimes .
the dynamics of the local vacuum is governed by gravity itself in the form of einstein s equations at each point in an arbitrary spacetime , while the gauge theory dynamics is encoded in the perturbation theory about an ads gravity background . in the duality the @xmath10 result holds for both conformal and non - conformal gauge theories .
the common feature is strong coupling , in particular very large t hooft coupling . in general the shear viscosity to entropy density ratio of a gauge theory
depends on the value of the t hooft coupling @xcite . for weakly coupled theories
there is a large separation between the mean free path and any other microscopic scale , for example a thermal de broglie wavelength . in this intermediate region
we can use a kinetic theory description where viscosity is due to momentum transfer by quasiparticle motion .
larger mean free paths correspond to an easier momentum transfer and higher viscosity . as the coupling is tuned up the viscosity decreases and the kinetic theory description begins to break down .
nevertheless , extrapolating all the way to strong coupling correctly indicates @xmath169 @xcite .
this viewpoint suggests the dynamics of the local vacuum thermal state should also be strongly coupled in some sense .
in fact , using the results of section [ balancelaw ] we can argue this is the case . from @xmath165 , we can roughly determine the other undetermined parameter in our analysis : the dimensionless parameter @xmath170 in the mean free path .
first , from kinetic theory we can estimate @xmath171 . using @xmath172 and @xmath159 for the thermal state
, this implies that ~. [ coupling ] thus , consistency with the entropy balance law implies the mean free path @xmath173 is of order the uv cutoff scale @xmath101 . since @xmath174 , where @xmath175 is the local temperature at the cutoff
, we find that @xmath170 must be roughly of order unity .
we can think of the dimensionless parameter @xmath134 in @xmath176 as a coupling which controls the size of the mean free path compared to the microscopic scale ( here the uv cutoff scale ) . if @xmath177 the mean free path would be much larger than the uv cutoff length and the ratio much larger than @xmath178 .
however the hydrodynamic derivation requires @xmath179 and @xmath180 , which is indicative of strong coupling .
typically one would not consider the local vacuum thermal state a strongly coupled system .
for example , the vacuum fluctuations of a free field do not appear to be strongly coupled .
on the other hand , a free ( scalar , for example ) field theory in flat spacetime is a continuum field theory and has infinite entanglement entropy .
this conflicts with the requirement of a finite entropy and a non - zero cutoff that allowed us to derive the entropy balance law .
the balance law implies we must have backreaction effects that distort the flat background spacetime .
these gravitational dynamics are imposed up to the uv cutoff , where the physics is strongly coupled . in this regime
the entropy density and shear viscosity are universal constants proportional to one another .
since the bulk viscosity is not determined , it may depend on the field content . if this is the case it would be similar to a non - conformal gauge theory , where the bulk viscosity is determined by the mass scales associated with the particular fields that break the conformal symmetry .
since our picture of the local vacuum thermal state and its dynamics is consistent with key aspects of the gauge / gravity dualities , perhaps it can provide a new perspective on the puzzling aspects of the @xmath10 ratio .
for example , although the ratio is derived for relativistic field theories , it is independent of the speed of light @xmath11 ( when we return to cgs dimensions ) .
the hydrodynamic derivation indicates the ratio is tied crucially to the physics of null horizons .
as we noticed in ( [ entropybal2 ] ) the entropy balance law for the local acceleration horizon reduces to a non - relativistic form .
the same type of behavior was first noticed in the membrane paradigm @xcite , where the behavior of a black hole event horizon is analogous to non - relativistic fluid dynamics .
intuitively , the value of the speed of light should not affect the behavior of the intrinsically ultra - relativistic degrees of freedom living on the stretched horizon boundary surface .
an important open question is whether @xmath10 is a universal bound on the shear viscosity to entropy density ratio for all systems , even those that are non - relativistic .
experimental data and the fact that viscosity is larger than @xmath181 for weakly coupled systems indicate the bound is plausible @xcite . it is curious that the conjectured bound is independent of @xmath12 even though it is saturated in the special class of theories with a gravity dual .
however , in the hydrodynamic derivation we relied only the general thermal properties of the minkowski vacuum .
@xmath12 only appears ( in the planck length ) when we require agreement with the experiment and fix @xmath164 .
this indicates that the bound , if it exists , may be a consequence of the behavior of quantum fields when they are localized into regions of flat spacetime .
the conjectured bound may also be related to the bekenstein entropy bound @xcite , a result also first derived in a gravitational setting , yet which ultimately does not depend on @xmath12 . in the future
it would be interesting to consider higher curvature corrections to the assumed entropy density and compare the results for the viscosities to the gauge / gravity literature .
for example , in the case where the acceleration horizon entropy density is assumed to be a ( non - constant , polynomial ) function of the ricci scalar @xmath9 it was found previously that @xmath165 , while the bulk viscosity ratio is @xmath182 @xcite .
corresponding results in the duality would require study of corrections to non - conformal gauge theories , although an exact comparison is complicated by our inability to determine the bulk viscosity even when an area entropy is assumed . in ads
/ cft , general corrections to einstein gravity involving contractions of the ricci and riemann tensors were found to modify the shear viscosity to entropy density ratio at strong t hooft coupling . with a choice of one parameter
the ratio can now be less than @xmath10 @xcite .
a comparison here would require assuming the local horizon has an entropy density proportional to these contractions and checking the effects in the entropy balance law ( [ entropybal2 ] ) .
finally , we have only considered linearized hydrodynamics , which was sufficient to derive the einstein equations and fix the shear viscosity of the local rindler stretched horizon .
however , @xcite recently showed the form of the hydrodynamic stress tensor on the ads boundary is determined up to 2nd order in derivatives by demanding the einstein equations ( with negative cosmological constant ) hold for perturbations about black brane spacetimes .
the procedure is roughly the inverse of our hydrodynamic derivation : instead of starting with the hydrodynamics of local rindler stretched horizon and deriving einstein s equations , they impose einstein s equation order by order in a derivative expansion to derive the stress - tensor at the ads boundary .
the resulting perturbative metrics are dual to solutions of the navier - stokes equations .
the set of hydrodynamic coefficients at the next order in the stress tensor characterize relaxation times @xcite .
in our case it would be interesting to see whether higher derivative terms in the stress tensor of the local rindler stretched horizon can be meaningfully defined and fixed by the entropy balance law .
i am grateful to jacob bekenstein and ted jacobson for helpful discussions on drafts of this paper .
i also thank itzhak fouxon for a suggestion .
this research was supported by the lady davis foundation at hebrew university , and by grant 694/04 of the israel science foundation , established by the israel academy of sciences and humanities .
t. padmanabhan , class .
* 19 * , 5387 ( 2002 ) [ arxiv : gr - qc/0204019 ] ; a. paranjape , s. sarkar and t. padmanabhan , phys
. rev .
d * 74 * , 104015 ( 2006 ) [ arxiv : hep - th/0607240 ] ; d. kothawala , s. sarkar and t. padmanabhan , phys .
b * 652 * , 338 ( 2007 ) [ arxiv : gr - qc/0701002 ] .
t. damour , these de doctorat detat , university of paris vi , ( 1979 ) ( unpublished ) ; t. damour , in _ proceedings of the second marcel grossmann meeting on general relativity _ , edited by r. ruffini ( north - holland , amsterdam , 1982 ) , p. 587 .
r. h. price and k. s thorne , phys .
d * 33 * , 915 ( 1986 ) ; _ black holes : the membrane paradigm _ , edited by k. s. thorne , r. h. price , and d. a. macdonald ( yale university press , london , 1986 ) ; m. parikh and f. wilczek , phys .
d * 58 * , 064011 ( 1998 ) [ arxiv : gr - qc/9712077 ] .
j. m. maldacena , adv .
* 2 * , 231 ( 1998 ) [ int .
j. theor .
* 38 * , 1113 ( 1999 ) ] [ arxiv : hep - th/9711200 ] ; o. aharony , s. s. gubser , j. m. maldacena , h. ooguri and y. oz , phys . rept . *
323 * , 183 ( 2000 ) [ arxiv : hep - th/9905111 ] .
t. d. cohen , phys .
lett . * 99 * , 021602 ( 2007 ) [ arxiv : hep - th/0702136 ] ; a. cherman , t. d. cohen and p. m. hohler , jhep * 0802 * , 026 ( 2008 ) [ arxiv:0708.4201 [ hep - th ] ] ; d. t. son , phys .
* 100 * , 029101 ( 2008 ) [ arxiv:0709.4651 [ hep - th ] ] ; t. d. cohen , arxiv:0711.2664 [ hep - th ] . for ideas on rindler horizon holography using action functionals ,
see a. mukhopadhyay and t. padmanabhan , phys .
d * 74 * , 124023 ( 2006 ) [ arxiv : hep - th/0608120 ] ; t. padmanabhan , aip conf .
proc . * 939 * , 114 ( 2007 ) [ arxiv:0706.1654 [ gr - qc ] ] and references therein .
l. susskind and j. uglum , phys .
d * 50 * , 2700 ( 1994 ) [ arxiv : hep - th/9401070 ] ; t. jacobson , arxiv : gr - qc/9404039 ; s. n. solodukhin , [ arxiv : hep - th/9504022 ] ; f. larsen and f. wilczek , nucl .
b * 458 * , 249 ( 1996 ) [ arxiv : hep - th/9506066 ] . | it has recently been shown that the einstein equation can be derived by demanding a non - equilibrium entropy balance law @xmath0 hold for all local acceleration horizons through each point in spacetime .
the entropy change @xmath1 is proportional to the change in horizon area while @xmath2 and @xmath3 are the energy flux across the horizon and unruh temperature seen by an accelerating observer just inside the horizon .
the internal entropy production term @xmath4 is proportional to the squared shear of the horizon and the ratio of the proportionality constant to the area entropy density is @xmath5 . here
we will show that this derivation can be reformulated in the language of hydrodynamics .
we postulate that the vacuum thermal state in the rindler wedge of spacetime obeys the holographic principle .
hydrodynamic perturbations of this state exist and are manifested in the dynamics of a stretched horizon fluid at the horizon boundary . using the equations of hydrodynamics we derive the entropy balance law and show the einstein equation is a consequence of vacuum hydrodynamics .
this result implies that @xmath5 is the shear viscosity to entropy density ratio of the local vacuum thermal state .
the value @xmath5 has attracted much attention as the shear viscosity to entropy density ratio for all gauge theories with an einstein gravity dual .
it has also been conjectured as the universal lower bound on the ratio .
we argue that our picture of the vacuum thermal state is consistent with the physics of the gauge / gravity dualities and then consider possible applications to open questions .
l * hydrodynamics of spacetime + and vacuum viscosity * christopher eling@xmath6 0.5 cm @xmath6_racah institute of physics _ _ the hebrew university of jerusalem _
_ jerusalem 91904 , israel _ 1 cm |
ngc 4472 ( m49 ) is a giant elliptical galaxy in the virgo cluster , located at 4@xmath10 south of ngc 4486 ( m87 ) at the center of the cluster .
ngc 4472 is the brightest member of the virgo cluster , some 0.2 mag brighter than the cd galaxy m87 .
ngc 4472 is an outstanding example of giant elliptical galaxies showing a bimodality in the color distribution of globular clusters ( geisler , lee & kim 1996 ; lee , kim & geisler 1998 ) .
the bimodal color distribution of the globular clusters in ngc 4472 has shown that there are two kinds of cluster populations in this galaxy : a metal - poor population with a mean metallicity of [ fe / h ] @xmath11 dex and a more spatially concentrated metal - rich population with a mean metallicity of [ fe / h ] @xmath12 dex .
interestingly it is found that the metal - rich globular clusters show some properties in common with the galaxy halo stars in their spatial distribution and color profiles , while the metal - poor globular clusters do not show such behavior .
this result indicates that there may exist some connection between the metal - rich globular clusters and halo stars in ngc 4472 ( lee et al .
1998 ) .
[ cols="<,^,^,^ , < " , ] it is found that there is clearly a negative color gradient for @xmath13 (= 15 kpc ) in ngc 4472 in the sense that the color gets bluer with increasing radius : @xmath14 mag arcsec@xmath4 for @xmath15 . in general ,
color gradients in giant elliptical galaxies are interpreted as evidence for a metallicity gradient , while color gradients in dwarf galaxies may be due to age effects ( vader et al .
1988 ; peletier et al .
the color gradient in the inner region of ngc 4472 corresponds to a metallicity gradient of @xmath6[fe / h ] @xmath7 dex .
this result is similar to that derived from the line - strength gradient for @xmath16 of ngc 4472 by davies et al .
this value for ngc 4472 is similar to the mean value known for giant elliptical galaxies , @xmath6[fe / h]@xmath17 dex ( kormendy & djorgovski 1989 ; peletier et al .
1990 ; davies , sadler & peletier 1993 ) .
the metallicity gradient in ngc 4472 may result from dissipational collapse .
however , the metallicity gradients predicted by conventional models of galaxy formation based on dissipational collapse ( larson 1975 ; carlberg 1984 ) are much steeper than the value observed in ngc 4472 . considering that the metallicity gradients can be diluted by a factor of 2 over three merger events @xcite , davies et al .
( 1993 ) pointed out that the shallow metallicity gradients support the hypothesis that giant elliptical galaxies form by stellar mergers , and that the line - strength ( metallicity ) gradient may be due to their progenitors which formed predominantly by dissipational collapse .
some of the observed properties of the globular clusters in ngc 4472 also support the merger hypothesis for the formation of ngc 4472 , while some do not .
this point was discussed in detail by lee et al .
( 1998 ) .
on the other hand , it is difficult to understand in terms of galaxy formation that the color profile of ngc 4472 shows a positive gradient in the outer region at @xmath18 .
this trend was also shown by the @xmath19 color profile given by cohen ( 1986 ) , but this @xmath19 color profile showed a positive color gradient even for the inner region which is contrary to others , as seen in fig .
other deep photometry of the outer region of ngc 4472 and other giant elliptical galaxies are needed to investigate this feature further .
we have presented surface photometry in washington @xmath8 and @xmath9 filters for a wide field centered on the brightest elliptical galaxy in virgo , ngc 4472 .
our data cover a wider field than those of any previous ccd surface photometry , and our photometry goes deeper than any previous photometry .
primary results obtained in this study are summarized as follows : \(1 ) surface brightness profiles for the outer region of ngc 4472 obtained in this study lie between those found in the wide field photographic studies of king ( 1978 ) and caon et al .
the surface brightness profiles of ngc 4472 are not fit well by a single king model , but they can be fit approximately by two king models : one for the inner region and the other for the outer region .
surface brightness profiles for the outer region can be fit approximately also by a devaucouleurs law .
\(2 ) there is obviously a negative color gradient for the region at @xmath13 of ngc 4472 in the sense that the colors get bluer with increasing radius .. the slope of the color gradient for this region is derived to be @xmath14 mag arcsec@xmath4 for @xmath5 , which corresponds to a metallicity gradient of @xmath6[fe / h ] @xmath7 dex
. however , the surface color appears to get redder slowly with increasing radius for the region with @xmath18 .
\(3 ) a comparison of the structural parameters of ngc 4472 in @xmath8 and @xmath9 images has shown that there is little difference in the ellipse shapes between isochromes and isophotes .
\(4 ) photometric and structural parameters of ngc 4472 have been determined , which are listed in table 3 .
this research is supported by the ministry of education , basic science research institute grant no.bsri-98-5411 ( to m.g.l . ) .
barsony , m. 1989 , phd thesis , caltech bender , r. , & mllenhoff , c. 1987 , a&a , 177 , 71 boroson , t. a. & thompson , i. b. 1987 , aj , 93 , 33 caon , n. , capaccioli , m. & donofrio , m. 1994 , a&as , 126 , 199 carlberg , r. g. 1984 , apj , 286 , 403 canterna , r. 1976 , aj , 81 , 228 cohen , j. g. 1986 , aj , 92 , 1039 davies , r. l. , sadler , e. m. , & peletier , r. f. 1993 , mnras , 262 , 650 davis , r. , & birkinshaw , m. 1988 , apjs , 68 , 409 devaucouleurs , g. , devaucouleurs , a , corwin , jr . , h. g. , buta , r. j. , paturel , g. & fouque , p. 1991
, _ third reference catalogue of bright galaxies _ , springer - verlag elias , j. 1994 , noao newsletter , no .
37 , 1 ferrarese , l. , van den bosch , f. c. , ford , h. c. , jaffe , w. & oconnell , r. w. 1994 , aj , 108 , 1598 forbes , d. a. , brodie , j. p. & grillmair , c. j. 1997 , aj , 113 , 1652 franx , m. , illingworth , g. , & heckman , t. 1989 , apj , 344 , 613 geisler , d. & forte , j. c. 1990 , apj , 350 , l5 geisler , d. 1996 , aj , 111 , 480 geisler , d. , lee , m. g. & kim , e. 1996 , aj , 111 , 1529 jaffe , w. , ford , h. c. , oconnell , r. w. , van den bosch , f. c. & ferraresse , l. 1994 , aj , 108 , 1567 kent , s. m. 1985 , pasp , 97 , 165 king , i. r. 1966 , aj , 71 , 64 king , i. r. 1978 , aj , 221 , 1 kormendy , j. , & djordovski , s. 1989 , ara&a , 27 , 235 larson , r. b. 1975 , mnras , 173 , 671 lauer , t. r. 1985 , apjs , 57 , 473 lee , m. g. , kim , e. & geisler , d. 1998 , aj , 115 , 947 mclaughlin , d. e. 1999 , aj , 117 , 2398 michard , r. 1985 , a&as , 59 , 205 mihalas , d. , & binney , j. 1981 , galactic astronomy ( w.h freeman and company , san francisco ) , 309 peletier , r. f. , davis , r. l. , illingworth , g. d. , davis , l. e. & cawson , m. 1990 , aj , 100 , 1091 pilachowski , c. , goodrich , b. , binkert , b. & africano , j. 1989 , noao newsletter , no .
15 , 14 taylor , b. j. 1985 , apjs , 60 , 577 tifft , w. g. 1969 , aj , 74 , 354 vader , j. p. , vigroux , l. , lachize - rey , m. , & souviron , j. 1988 , a&a , 203 , 217 van den bosch , f. c. , ferraresse , l. , jaffe , w. , ford , h. c. & oconnell , r. w. 1994 , aj , 108 , 1579 white , s. d. m. 1980 , mnras , 191 , 1p | we present deep wide field ( @xmath0 ) washington @xmath1 ccd surface photometry of the giant elliptical galaxy ngc 4472 , the brightest member in the virgo cluster .
our data cover a wider field than any previous ccd photometry as well as going deeper .
surface brightness profiles of ngc 4472 are not well fit by a single king model , but they can be fit approximately by two king models : with separate models for the inner and outer regions .
surface brightness profiles for the outer region can also be fit approximately by a devaucouleurs law .
there is clearly a negative color gradient within @xmath2 of ngc 4472 , in the sense that the color gets bluer with increasing radius .
the slope of the color gradient for this region is derived to be @xmath3 = 0.08 mag arcsec@xmath4 for @xmath5 , which corresponds to a metallicity gradient of @xmath6 [ fe / h ] @xmath7 dex .
however , the surface color gets redder slowly with increasing radius beyond @xmath2 .
a comparison of the structural parameters of ngc 4472 in @xmath8 and @xmath9 images has shown that there is little difference in the ellipse shapes between isochromes and isophotes .
in addition , photometric and structural parameters of ngc 4472 have been determined
. = = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] galaxies : ngc 4472 galaxies : elliptical galaxies : abundances surface photometry color gradient |
in this paper , we introduce the notion of a subanalytic bundle @xmath4 ( generated by a finite dimensional space @xmath0 of global sections ) on a ( not necessarily closed ) subanalytic set @xmath1 inside a real analytic manifold @xmath2 , as a natural generalisation of real analytic bundles on real analytic spaces to the subanalytic setting .
we prove ( in theorem [ strattubular2 ] below ) that for @xmath2 compact , there exists a baire subset @xmath3 of sections in @xmath0 , such that for @xmath5 , there exist tubular neighbourhoods of the zero - locus @xmath6 of @xmath7 in @xmath1 , i.e. which are homeomorphic to the restriction of the given bundle to @xmath8 . to keep the account self - contained we recall basic facts about subanalytic sets in 2 and strong whitney ( sw ) stratifications ( defined by verdier ) in 4 .
1 we remark here that the main theorem [ strattubular2 ] would follow from theorem 1.11 on p. 48 of [ g - m ] . however , the proof ( ` deformation to the normal bundle ' ) sketched in [ g - m ] is incomplete , at least in the generality that it is stated . in this generality ,
the stratified submersion they construct is not proper ( as was pointed out by v srinivas ) , and hence thom s first isotopy lemma is inapplicable . to circumvent this , we have imposed the hypothesis of compactness on the ambient real analytic manifold @xmath2 containing the subanalytic set @xmath1 , but no compactness assumption on @xmath1 .
our hypotheses are general enough to cover most situations arising in real or complex algebraic geometry ( see example [ subanalyticexample ] and remark [ strattubular3 ] ) .
let @xmath2 be a real - analytic manifold .
we will always assume @xmath2 to be connected , hausdorff , second countable and paracompact .
@xmath9 we say @xmath10 is a _
subanalytic set _ of @xmath2 if there exists an open covering @xmath11 of @xmath2 ( _ not _ just of @xmath1 ) such that for each @xmath12 , @xmath13 where @xmath14 for @xmath15 and @xmath16 , are real analytic maps of real analytic manifolds @xmath17 , @xmath18 are closed analytic subsets of @xmath17 and @xmath19 are proper maps ( see proposition 3.13 in [ b - m ] and definition 3.1 in [ hi]).[subanaldef ] all real ( resp .
complex ) analytic subsets of a real ( resp .
complex ) analytic manifold are subanalytic sets . in particular , ( real or complex ) algebraic subsets of a ( real or complex ) algebraic manifold ( such as projective space , or grassmannians ) are subanalytic sets . also since subanalytic subsets of a real analytic manifold form a boolean algebra ( see ( i ) of proposition [ subanalfacts ] below ) all real ( resp .
complex ) analytically ( or algebraically ) constructible sets in a real ( resp .
complex ) analytic ( or algebraic ) manifold are subanalytic sets .
in particular , all ( real or complex ) affine algebraic varieties are subanalytic in both affine space , and projective space . real or complex quasiprojective varieties
are subanalytic sets in the corresponding projective spaces.[subanalyticexample ] a real analytic subset @xmath1 ( subspace ) of a real analytic manifold @xmath2 is a closed subset of @xmath2 by definition . in particular if @xmath2 is compact , so is @xmath1 . by contrast , a subanalytic set @xmath1 of a real analytic manifold @xmath2 need not be closed , and need not be compact even if @xmath2 is compact .
[ analsubanal ] @xmath9 let @xmath10 and @xmath20 be subanalytic sets in the real analytic manifolds @xmath21 respectively .
we say that a map @xmath22 is a _ subanalytic map _ if @xmath23 is a continuous map , and the graph @xmath24 is a subanalytic set in @xmath25 ( see [ ha ] , 4.1 , or definition 3.2 in [ b - m ] ) .
note that although the map @xmath26 is defined only on @xmath1 , its subanalyticity depends on the ambient @xmath2 , @xmath27 , as we shall see in remark [ subanalrmk ] below .
[ subanalmapdef ] if @xmath28 are as above , and @xmath29 is a subanalytic map , we shall write @xmath30 is a subanalytic map , for notational convenience .
the subanalyticity ( or analyticity ) of a set , or of a map _ depends on the ambient spaces
_ @xmath21 .
for example , @xmath31 is a subanalytic set in @xmath32 , but not in @xmath33 . in the former ,
it is the zero set of the analytic function @xmath34 , so analytic and hence subanalytic in @xmath32 .
it is not subanalytic in @xmath33 because the connected components of its germ at @xmath35 in @xmath33 do not form a locally finite collection ( see ( viii ) of proposition [ subanalfacts ] below ) . similarly , the map @xmath36 defined by @xmath37 is clearly subanalytic , because its graph @xmath38 is an analytic ( hence subanalytic ) subset of @xmath39 . on the other hand , the same mapping regarded as a map @xmath40 is _ not _ subanalytic , since @xmath41 is not a subanalytic subset in @xmath42 .
( by ( i ) of proposition [ subanalfacts ] , if @xmath41 were subanalytic in @xmath43 , its intersection with the @xmath44-axis would have to be subanalytic in @xmath45 .
but this intersection is the set @xmath46 , which is not subanalytic because the connected components of its germ at @xmath47 in @xmath45 is not a locally finite collection ( see ( viii ) of proposition [ subanalfacts ] below.).[subanalrmk ] ( facts on subanalytic sets and maps)@xmath9 we collect some well - known facts on subanalytic sets and maps : 1 . the collection of subanalytic sets of a real analytic manifold @xmath2 forms a boolean algebra .
2 . if @xmath48 is a proper real analytic map of real analytic manifolds , and @xmath10 a subanalytic set , then @xmath49 is subanalytic in @xmath27 . in particular ,
if @xmath2 is compact , the image @xmath49 is subanalytic .
if @xmath50 is a subanalytic map and @xmath1 is relatively compact in @xmath2 , then @xmath51 is subanalytic .
3 . if @xmath10 and @xmath20 are subanalytic , then @xmath52 is subanalytic in @xmath25 .
4 . if @xmath10 is subanalytic , then the diagonal @xmath53 is subanalytic in @xmath54 .
thus the inclusion map @xmath55 is always a subanalytic map .
let @xmath48 be a real analytic map of real analytic manifolds .
if @xmath10 is a subanalytic set , then the restricted map @xmath56 is a subanalytic map .
let @xmath48 be a real analytic map of real analytic manifolds .
if @xmath20 is subanalytic , then @xmath57 is subanalytic . 7
the closure @xmath58 of a subanalytic set @xmath10 is also subanalytic .
let @xmath10 be a subanalytic set in a real analytic manifold @xmath2 .
then each connected component of @xmath1 is also subanalytic .
the collection of connected components of @xmath1 is a locally finite collection in @xmath2 .
[ subanalfacts ] for a proof of ( i ) , see proposition 3.2 of [ hi ] or 3 of [ b - m ] . for a proof of the first statement of
( ii ) , see [ hi ] , proposition 3.8 . for a proof of the second statement of ( ii ) , see the remark after definition 3.2 in [ b - m ] .
easy examples can be constructed to show that the properness condition can not be dropped from thehypothesis . to see ( iii ) , first note that @xmath10 subanalytic implies @xmath59 is also subanalytic .
for , in the definition [ subanaldef ] above , one merely takes the open covering @xmath60 , the closed sets @xmath61 and the maps @xmath62 , which are proper on @xmath63 since @xmath64 are proper on @xmath18 .
similarly @xmath65 is also subanalytic . by ( i ) above , the intersection @xmath66 is also subanalytic .
this proves ( iii ) .
to see ( iv ) , note that by ( iii ) @xmath67 is subanalytic .
the diagonal @xmath68 is subanalytic since it is analytic in @xmath54 .
the intersection @xmath69 is therefore subanalytic by ( i ) .
since @xmath70 is the graph of the inclusion @xmath55 inside @xmath54 , it follows that @xmath71 is a subanalytic map . to see ( v )
, we note that the graph of @xmath72 in @xmath25 is just the intersection of the graph @xmath73 of @xmath26 and @xmath74 inside @xmath25 .
since @xmath73 is an analytic set in @xmath25 , it is subanalytic , and since by ( iii ) @xmath74 is subanalytic , their intersection is subanalytic by ( i ) . for ( vi )
, let @xmath20 be subanalytic and let @xmath11 be an open covering of @xmath27 such that for each @xmath75 we have @xmath76 where @xmath14 are real analytic maps of real analytic manifolds @xmath17 , @xmath18 are closed analytic subsets of @xmath17 and @xmath19 are proper maps . now take the open covering @xmath77 of @xmath2 , and set @xmath78 , with @xmath79 being the second projection .
let @xmath80 ( the fibre product , a closed analytic subset of @xmath81 ) .
observe that the restriction to @xmath82 of the natural real analytic projection @xmath79 , is the ` base change ' to @xmath83 of the restriction @xmath84 , and this last map is given to be proper .
hence this restriction @xmath85 is proper .
it is easily verified that @xmath86 which shows that @xmath87 is subanalytic . for ( vii ) , see the immediate consequences following definition 3.1 in [ b - m ] , and also corollary 3.2.9 in [ hi ] . for ( viii ) , see the immediate consequences following definition 3.1 in [ b - m ] .
( also see proposition 3.6 and corollary 3.7.10 in [ hi ] . ) for an analytic subset @xmath1 inside a real analytic manifold @xmath2 , there is a structure sheaf , making it a locally ringed space .
thus mappings ( @xmath88 morphisms ) of real analytic spaces are easy to define , and obey the usual functorial properties . for subanalytic sets inside a real analytic manifold , we note that there is no such structure sheaf , and the definition of a subanalytic map is dependent on the ambient manifold @xmath2 .
thus the notion of ` subanalytic equivalence ' of subanalytic sets @xmath10 and @xmath20 requires some care .
we propose one such below , which may not be the most general , but is good enough for our purposes .
we are unaware if this notion exists in the literature .
let @xmath2 , @xmath89 be real analytic manifolds .
let @xmath10 be a subanalytic set , and suppose @xmath90 is a subanalytic map .
suppose there exists a _
proper _ real analytic map @xmath91 such that @xmath92
. then 1 .
@xmath93 is a homeomorphism , and its image @xmath94 is subanalytic in @xmath89 .
further @xmath10 is relatively compact if and only if @xmath95 is relatively compact .
2 . for each subanalytic map
@xmath96 a real analytic manifold , the composite map : @xmath97 is a subanalytic map .
3 . for each subanalytic map @xmath98 a real analytic manifold ,
the composite : @xmath99 is a subanalytic map .
[ properlemma ] it is clear that @xmath100 is a homeomorphism , with inverse @xmath101 . consider the real analytic map @xmath102 z & \mapsto ( p(z),z).\end{aligned}\ ] ] also let @xmath103 be the graph of @xmath100 . since @xmath100 is a subanalytic map , @xmath104 is subanalytic .
we claim that the inverse image @xmath105 is precisely @xmath106 . for @xmath107 and @xmath108 .
conversely , @xmath109 for @xmath110 which is clearly in @xmath111 .
hence the claim .
now , since @xmath112 is real analytic , and @xmath111 is subanalytic , we have by ( vi ) of proposition [ subanalfacts ] that @xmath113 is subanalytic .
this proves the first assertion of ( i ) .
for the second assertion , note that the continuity of @xmath114 implies @xmath115 .
since @xmath114 is proper , it is a closed map , and so @xmath116 is a closed set containing @xmath117 .
hence @xmath118 .
thus if @xmath106 is relatively compact in @xmath89 , @xmath1 is relatively compact in @xmath2 .
conversely if @xmath1 is relatively compact in @xmath2 , @xmath119 and @xmath114 is proper implies that @xmath120 is a closed subset of the compact set @xmath121 , and hence also compact .
that is , @xmath106 is relatively compact in @xmath89 .
this proves ( i ) . to see ( ii ) , let @xmath122 be a subanalytic map .
thus the graph @xmath123 is a subanalytic set .
the real analytic map : @xmath124 is proper , since @xmath114 is proper . but the image @xmath125 & = \{(pj(y),fj(y))\in m\times n:\;y\in x\}\\[.2pc ] & = \{(y , fj(y))\in m\times n:\;y\in x\}\\[.2pc ] & = \gamma_{f\circ j},\end{aligned}\ ] ] the graph of the composite @xmath126 .
since @xmath127 is a proper real analytic map , and @xmath73 is subanalytic , it follows by ( ii ) of proposition [ subanalfacts ] that this image , the graph @xmath128 is subanalytic in @xmath25 , so that @xmath129 is a subanalytic map . thisproves ( ii ) . to see ( iii ) ,
let @xmath50 be a subanalytic map .
this means that the graph @xmath130 is a subanalytic set .
consider the set @xmath131
\hskip -1pc & = \{(m , n)\in m_{1}\times n:\ ; n = gp(m ) , \;m\in x_{1}\}\\[.2pc ] \hskip -1pc & = \gamma_{g\circ p_{|x_{1}}}.\end{aligned}\ ] ] since @xmath132 is subanalytic in @xmath133 by ( iii ) of proposition [ subanalfacts ] , @xmath134 is subanalytic in @xmath25 by definition , and @xmath135 is real analytic , ( iii ) follows from ( i ) and ( vi ) of proposition [ subanalfacts ] .
the above lemma [ properlemma ] shows that under the hypotheses stated there , the subanalytic sets @xmath136 and @xmath137 are ` equivalent ' in some sense .
more precisely , we make the following definition : ( pseudoequivalence of subanalytic sets)@xmath9 let @xmath138 be real analytic manifolds , with @xmath10 a subanalytic set and @xmath90 a subanalytic map . if there exists a
_ proper real analytic map _
@xmath91 such that @xmath92 , then we say that the subanalytic sets @xmath136 and @xmath139 are _ subanalytically pseudoequivalent_. the map @xmath100 is called a _
subanalytic pseudoequivalence_. we note that @xmath1 and @xmath140 are therefore _ a fortiori _ homeomorphic , and also ( i ) of the lemma [ properlemma ] implies that @xmath1 is relatively compact in @xmath2 iff @xmath106 is relatively compact in @xmath89 .
[ properequivdef ] the prototypical example of such a subanalytic pseudoequivalence of interest to us in the sequel is the following .
( graph embeddings ) .
let @xmath10 be a subanalytic set , and @xmath30 a subanalytic map .
_ assume that @xmath27 is compact_. set @xmath141 , and let @xmath90 be the graph embedding defined by @xmath142 for @xmath143 .
@xmath100 is a subanalytic map because its graph in @xmath144 is the set @xmath145 , which is precisely the intersection of the two subanalytic sets @xmath146 and @xmath147 in @xmath144 , and therefore subanalytic ( by proposition [ subanalfacts](i ) , ( iii ) , and ( iv ) ) .
the projection @xmath148 is proper since @xmath27 is compact , and thus we have the requirements of definition [ properequivdef ] .
that is , @xmath149 is a subanalytic pseudoequivalence.[graphexample ]
we review some basic notions of bundles from the real - analytic set - up , with a view to generalising them to the subanalytic set - up .
suppose that @xmath10 is a real analytic subset ( @xmath88 subspace ) in a real analytic manifold @xmath2 . by definition , its germ at each point of @xmath2 ( not just @xmath1 ) is given by the vanishing of some ideal ,
so by definition @xmath1 is closed in @xmath2 .
then @xmath1 comes equipped with a structure sheaf @xmath150 , whose stalk @xmath151 at @xmath143 consists of germs of real analytic functions on @xmath1 at @xmath44 .
it is , by definition , the local ring @xmath152 , where @xmath153 is the local ring of @xmath2 at @xmath44 consisting of convergent power series at @xmath154 , and @xmath155 is the ideal of functions in @xmath156 vanishing on the germ of @xmath1 at @xmath44 .
let @xmath157 be a _
real analytic vector bundle of rank _
@xmath158 on @xmath1 .
that is , its transition cocycles @xmath159 are real - analytic functions for all @xmath160 .
the sheaf ( of germs of analytic sections ) of an analytic vector bundle @xmath4 on @xmath1 is a locally free sheaf @xmath161 of modules over the structure sheaf @xmath162 .
global sections of this sheaf are called global sections of @xmath4 .
@xmath4 is said to be _ generated by a vector space _ @xmath163 _ of global sections _ if the natural sheaf map : @xmath164 is a surjective map of @xmath150-modules .
this is equivalent to demanding that the evaluation map : @xmath165 is surjective for all @xmath143 , where @xmath166 is the fibre of @xmath4 at @xmath44 .
it is clear from the cocycle formulation of analytic vector bundles that the natural bundle operations , such as direct sums , tensor products , homs , duals and pullbacks under analytic maps of analytic vector bundles are again analytic .
real analytic sections of the real analytic bundle @xmath167 are defined to be real analytic bundle morphisms .
( universal exact sequence)@xmath9 let @xmath168 denote the grassmannian of @xmath169-dimensional subspaces of @xmath0 , where @xmath0 is a finite dimensional real vector space of dimension @xmath170 .
on @xmath168 , there is the short exact sequence of real analytic vector bundles and real analytic morphisms : @xmath171 where @xmath172 is the trivial rank-@xmath170 bundle on @xmath168 , @xmath173 is the tautological rank-@xmath169 real analytic bundle on @xmath168 ( having fibre @xmath174 over the point @xmath175 ) .
the bundle @xmath176 is the universal quotient bundle of rank @xmath158 on @xmath168 .
@xmath0 gets identified with the constant sections of @xmath172 , and the bundle @xmath176 is generated by the global sections @xmath177 , @xmath178 .
[ univdef ] indeed , the bundles and morphisms defined above are all real algebraic , and hence real analytic .
( classifying maps ) .
let @xmath2 be a real analytic manifold , and @xmath10 be a real analytic subset ( @xmath88 subspace ) .
let @xmath157 be a real analytic vector bundle of rank @xmath158 with corresponding sheaf @xmath161 .
let @xmath0 be an @xmath170-dimensional real vector subspace of @xmath179
. then the following are equivalent : 1 .
@xmath180 is generated by the global sections @xmath0 .
2 . there exists a real analytic map @xmath181 called the _ classifying map _ such that the pullback of the universal short exact sequence @xmath182 under @xmath26 yields the short exact sequence @xmath183 on @xmath1 . here
@xmath184 is the evaluation map @xmath185 .
3 . there is a real analytic manifold @xmath89 , a real analytic map @xmath186 , and a proper real analytic map @xmath91 such that : + .2pc ( a ) @xmath187 and @xmath140 is isomorphic to @xmath1 as a real analytic space via @xmath100 .
+ \(b ) there is an exact sequence of real analytic bundles on @xmath89 : @xmath188 with @xmath189 generated by a space of global sections equal to @xmath0 , and such that the pullback of the last two terms to @xmath1 via the analytic isomorphism @xmath100 is the morphism of bundles @xmath190 , defined by evaluation ( i.e. @xmath191 ) .
+ ( in keeping with definition one may want to call @xmath100 an _ analytic pseudoequivalence_. ) [ analclass ] @xmath9
\(i ) @xmath192 ( ii ) .
first we note that by choosing a basis of @xmath0 of real analytic sections @xmath193 , that the maps : @xmath194 are all real analytic , so that the map : @xmath195 is also real analytic . in particular , the map @xmath196 defined by @xmath197 is also real analytic . the classifying map @xmath26 is now defined as the map @xmath198 , where @xmath199 is the evaluation map . to show that this map @xmath26 is real analytic , it is enough to do it locally .
since @xmath4 is analytically locally trivial , we may assume without loss of generality that @xmath4 is trivial .
in this case , the map @xmath26 is just the composite : @xmath200 where @xmath201 is the open subset of @xmath202 consisting of surjective maps , the first arrow is @xmath203 , and the second arrow is the map @xmath204 . since both these maps are real analytic , the composite is real analytic .
thus the classifying map @xmath205 is real analytic .
so the bundle @xmath206 is a real analytic bundle , and it is clear that its fibre is @xmath207 finally , since @xmath208 , and @xmath209 , we have @xmath210 , and ( ii ) follows .
\(ii ) @xmath192 ( iii ) .
this follows by using the graph of the classifying map @xmath205 .
more precisely , let us define @xmath211 , a real analytic manifold , and set @xmath212 where @xmath26 is the classifying map of ( ii ) .
define @xmath213 to be the graph embedding @xmath142 .
let @xmath91 be the first projection .
clearly @xmath92 . that @xmath106 is a real analytic space in @xmath89 is clear from the corresponding local fact , i.e. that the germ of the graph @xmath106 , at any point @xmath214 is defined by the real analytic ideal @xmath215 generated by @xmath216 and ( the component functions of ) @xmath217 in the completed tensor product @xmath218 where @xmath219 and @xmath220 are local coordinates around @xmath154 and @xmath221 .
the graph embedding : @xmath222 is a real analytic map , with image @xmath106 . since the first projection @xmath91 provides the analytic inverse to @xmath213 , the map @xmath100 is an analytic isomorphism of the real analytic spaces @xmath1 and @xmath106 .
this proves ( a ) .
if we let @xmath223 denote the real analytic map defined by the second projection , and set @xmath224 , and @xmath225 , we have the @xmath226-pullback of the universal exact sequence of ( [ univseq ] ) : @xmath227 where @xmath228 .
this is a short exact sequence of real analytic vector bundles on @xmath89 .
if we pullback this sequence via the real analytic isomorphism @xmath100 , the fact that @xmath229 implies the short exact sequence ( [ exactseq ] ) on @xmath1 .
thus we have the assertion ( b ) .
\(iii ) @xmath192 ( i ) .
we pull back the given exact sequence of real analytic bundles on @xmath89 of ( iii ) to @xmath1 via @xmath100 .
it continues to be an exact sequence of real analytic bundles on @xmath1 .
if we let @xmath184 denote the pullback morphism @xmath230 , and set @xmath231 , the last two terms of this pulled back exact sequence read : @xmath232 which shows that the @xmath184 images of the constant sections @xmath233 generate @xmath4 , and hence ( i ) follows .
the equivalence of ( i ) and ( iii ) of the above lemma shows that in considering analytic bundles generated by global sections on an analytic subset @xmath10 , we lose no generality in assuming ( up to analytic pseudoequivalence ) that such bundles are _ restrictions _ of similar ( viz . generated by global sections ) real analytic bundles on the _ ambient smooth _
@xmath2 to @xmath1 .
we will generalise all this to a subanalytic setting in the next section.[restrictrmk ]
let @xmath10 a subanalytic set in @xmath2 , with @xmath2 a real analytic manifold . note that unlike analytic subsets , @xmath1 need not be closed anymore .
we need to define ` subanalytic bundles ' in some reasonable fashion , but are hindered by the fact that there is no structure sheaf for a subanalytic space . motivated by the equivalence of ( i ) and ( ii ) of lemma [ analclass ] of the last section , we _ propose _ the following : @xmath9 a _ subanalytic real vector bundle _ @xmath4 _ of rank _
@xmath158 _ generated by an @xmath170-dimensional vector space _ @xmath0 _ of global sections _ on a subanalytic set @xmath10 is the pullback @xmath234 , where @xmath176 is the universal rank @xmath158 analytic quotient bundle on @xmath168 defined in definition [ univdef ] , and @xmath235 is a _
subanalytic _ map ( see definition [ subanalmapdef ] ) .
then , by pulling back the last two terms of the exact sequence ( [ univseq ] ) , there is , by definition , an exact sequence : @xmath236 of continuous vector bundles on @xmath1.[subanalbundledef ] if @xmath10 is an analytic subset , and @xmath4 is a real analytic vector bundle of rank @xmath158 , generated by an @xmath170-dimensional vector space of real analytic global sections @xmath0 , we have by ( i)@xmath192 ( ii ) of lemma [ analclass ] that there is a real analytic classifying map @xmath205 . since @xmath10 is an analytic subset , it is subanalytic in @xmath2 , and since the graph @xmath73 is an analytic space in @xmath237 , it is subanalytic , so that @xmath26 is subanalytic .
thus ( ii ) of lemma [ analclass ] shows that the definition above is a generalisation of the real analytic setup to the subanalytic setting .
( a typical subanalytic bundle ) .
let @xmath10 be a subanalytic set , @xmath2 a real - analytic manifold , and let @xmath238 be a _ real analytic vector bundle _ generated by a vector space @xmath0 of global real analytic sections .
thus by definition , there is an exact sequence of analytic bundles and morphisms : @xmath239 by the remark above , there is a real analytic classifying map @xmath240 such that the exact sequence above is the pullback via @xmath26 of the universal sequence @xmath241 if one restricts @xmath26 to @xmath1 , the restricted map @xmath242 is a subanalytic map , by ( v ) proposition [ subanalfacts ] . thus the restricted bundle @xmath243 ( which is the pullback via @xmath72 of the above sequence on @xmath168 ) is by definition a subanalytic bundle generated by global sections @xmath0.[typicalexample ] again , motivated by the equivalence of ( ii ) and ( iii ) of lemma [ analclass ] , we would like to assert that _ up to a subanalytic pseudoequivalence _ of @xmath1 ( see definition [ properequivdef ] ) , the above example [ typicalexample ] of a subanalytic bundle generated by global sections @xmath0 is the _ only _ one . more precisely , in complete analogy with lemma [ analclass ] of the analytic setting in 3 , we have the following : the following are equivalent : 1 .
@xmath4 is a rank @xmath158 subanalytic vector bundle on @xmath1 generated by an @xmath170-dimensional vector space @xmath0 of global sections in the sense of definition [ subanalbundledef ] above . 2 .
there exists a real analytic manifold @xmath89 , a subanalytic map @xmath90 and a proper map @xmath91 such that : + \(a ) @xmath187 , and hence @xmath244 is a _
subanalytic pseudoequivalence _ in the sense of definition [ properequivdef ] .
+ \(b ) there is an exact sequence of _ real analytic bundles _ on the real analytic manifold @xmath89 : @xmath245 with @xmath189 generated by a space of global sections equal to @xmath0 , and such that the pullback of the last two terms via @xmath100 to @xmath1 , is a surjective morphism of bundles @xmath246 .
[ subanallemma ] @xmath9 \(i ) @xmath192 ( ii ) . as in lemma
[ analclass ] , set @xmath211 . by definition of a subanalytic map ,
the graph of the subanalytic classifying map @xmath26 : @xmath247 is a subanalytic space in @xmath89 . by example
[ graphexample ] , the graph embedding : @xmath248 is a subanalytic map .
@xmath168 is compact , so by example [ graphexample ] , the map @xmath249 is a subanalytic pseudoequivalence in the sense of definition [ properequivdef ] .
the proof of ( b ) of ( ii ) is exactly as in the proof of part ( b ) of ( iii ) in lemma [ analclass ] , and therefore omitted .
\(ii ) @xmath192 ( i ) . from the given exact sequence of real analytic bundles on @xmath89 : @xmath250 it follows from ( i ) and ( ii ) of lemma [ analclass ] that there is a real analytic classifying map @xmath251 ( defined by @xmath252 ) such that the above exact sequence is the @xmath253-pullback of the universal exact sequence ( [ univseq ] ) on @xmath168 . if one composes this classifying map with the the subanalytic map @xmath90 , which is given to be a subanalytic pseudoequivalence of @xmath136 with @xmath137 , we have : 1 .
@xmath254 is a subanalytic set in @xmath89 by ( i ) of lemma [ properlemma ] , and 2 .
the composite @xmath255 is a subanalytic map , by ( ii ) of lemma [ properlemma ] ( note that @xmath251 a real analytic map implies that @xmath256 is subanalytic by ( v ) of proposition [ subanalfacts ] ) .
clearly this subanalytic composite map : @xmath257 is the classifying map for the bundle @xmath258 .
thus , letting @xmath259 , and taking the @xmath260 of the last two terms of the given exact sequence ( [ exactm1 ] ) of real analytic bundles on @xmath89 , we have the exact sequence @xmath261 of bundles on @xmath1 , where @xmath262 .
since the exact sequence ( [ exactm1 ] ) is the @xmath253-pullback of the universal exact sequence ( [ univseq ] ) , the above exact sequence ( [ exactx ] ) is the pullback of the last two terms of universal sequence ( [ univseq ] ) via the above subanalytic map @xmath263 . in particular
, @xmath4 is a subanalytic bundle of rank @xmath158 generated by the global sections @xmath0 , by definition [ subanalbundledef ] .
this implies ( i ) . in complete analogy with remark [ restrictrmk ]
, we observe that ( ii ) of the above lemma says that in considering subanalytic bundles on a subanalytic set @xmath10 , generated by a vector space @xmath0 global sections , we lose no generality ( up to subanalytic pseudoequivalence ) in assuming that they are _ restrictions of real analytic bundles _ ( generated by the vector space of analytic sections @xmath264 ) on the _ ambient smooth _ @xmath2 .
[ restrictrmk2 ]
in this section , we recall some known definitions and results from the theory of stratifications of subanalytic sets .
the general references for this section are the papers by verdier [ ve ] and hironaka [ hi ] .
( the verdier condition)@xmath9 let @xmath2 and @xmath265 be two locally closed @xmath266 submanifolds of some real finite dimensional inner product space @xmath4 , and such that @xmath267 , with @xmath268 .
the _ property ( w ) _ or _ verdier condition _ ( see 1.4 in [ ve ] ) for the pair @xmath269 is the following : for each @xmath270 , there exists a neighbourhood @xmath3 of @xmath271 in @xmath4 and a constant @xmath272 such that , for all @xmath273 and @xmath274 , we have @xmath275 where @xmath276 is the distance between two vector subspaces @xmath277 and @xmath278 of @xmath4 ( see [ ve ] , 1.1 ) defined by @xmath279 @xmath280 being the euclidean distance ( in the given norm @xmath281 on @xmath4 ) between @xmath44 and @xmath278 in @xmath4 ( viz . , @xmath282 ) .
this property is invariant under smooth local diffeomorphisms of @xmath4 , and hence makes sense for @xmath283 contained in a smooth manifold @xmath4.[verdiercond ] now we can define stratifications and _ verdier _ ( or _ strong whitney _
stratifications ) .
( stratification and strong whitney stratification ) ( see [ ve ] ,
2.1)@xmath9 let @xmath2 be a real analytic manifold , countable at @xmath284 ( i.e. @xmath2 is the countable union of compact sets ) .
a _ stratification _ @xmath285 of @xmath2 is a partition of @xmath2 as @xmath286 satisfying : 1 .
@xmath287 for @xmath288 .
each @xmath289 is a locally closed real analytic submanifold of @xmath2 , smooth , connected , subanalytic in @xmath2 .
the family @xmath289 is locally finite .
the family @xmath289 has the boundary property : i.e. @xmath290 implies @xmath291 .
the @xmath289 are called the _ strata _ of the stratification @xmath285 . a _ strong whitney stratification _ ( or _ sw - stratification _ ) @xmath285 of @xmath2 is a stratification @xmath286 with the following additional property : 1 . if @xmath292 and if @xmath288 , then the pair @xmath293 has the property ( w ) of [ verdiercond ] above .
more generally , if @xmath1 is a subset of @xmath2 , we may define a _
@xmath285 of @xmath1 to be a partition of @xmath1 into @xmath294 where @xmath295 satisfy ( sw1 ) through ( sw3 ) above .
similarly , a strong whitney or sw - stratification of @xmath1 to be a stratification of @xmath1 having the additional property ( sw4 ) .
note that all the conditions ( such as subanalyticity , real analyticity , or the verdier condition ) _ are required to hold inside the ambient real analyticmanifold _
@xmath2 .
finally , we say a stratification @xmath296 of @xmath1 is _ finer _ or a _ refinement _ of the stratification @xmath285 if each stratum @xmath295 of the stratification @xmath285 is a union of strata @xmath297 of the stratification @xmath298.[stratdef ] @xmath9 1 .
the condition ( sw4 ) above is stronger than whitney s condition ( b ) .
see theorem 1.5 ( due to kuo ) in [ ve ] .
2 . since strata are disjoint sets , and constitute a locally finite collection , a compact set @xmath299 can intersect only finitely many strata . otherwise , choose a point in @xmath300 for each @xmath301 in some countably infinite index set of strata @xmath302 , and observe that it will have a limit point @xmath303 since @xmath304 is compact .
every neighbourhood of this @xmath44 will meet infinitely many @xmath302 , contradicting local finiteness .
in particular , if @xmath2 itself is compact , @xmath2 is automatically countable at infinity , and the number of strata in _ any _ stratification of @xmath2 is finite .
3 . if @xmath2 is sw - stratified as above by strata @xmath305 , and @xmath10 is any subset which is a union of strata , then @xmath1 is also sw - stratified ( by the strata of which it is a union ) .
[ stratrmk ] the following is a key proposition due to verdier ( [ ve ] , theorem 2.2 ) .
let @xmath2 be a real analytic manifold , and @xmath306 a locally finite family of subanalytic subsets of @xmath2 .
then there exists a strong whitney ( or sw)-stratification of @xmath2 such that each @xmath306 is a union of strata . if @xmath2 is compact , then the stratification is _ finite _ , i.e. the number of strata is finite .
[ refine1 ] in particular , for any subanalytic set @xmath10 , there is a sw - stratification of @xmath2 such that @xmath1 becomes a union of strata , and hence sw - stratified with the induced stratification .
note that the assertion for @xmath2 compact follows from ( ii ) of remark [ stratrmk ] above .
the analogous statement proving the existence of a whitney-(b ) stratification for the above setting is theorem 4.8 of [ hi ] .
( stratified transversality ) ( see 1.3.1 on p. 38 in [ g - m])@xmath9 let @xmath48 be a smooth map of real analytic manifolds .
let @xmath10 and @xmath20 be sw - stratified subsets , with strata @xmath307 and @xmath308 respectively .
we say that @xmath309 is _ transverse to _ @xmath310 ( denoted by @xmath311 ) if : 1 . for each stratum @xmath295 of @xmath1 , @xmath312 is a smooth map , and 2 . for each stratum @xmath295 of @xmath1 , and each stratum @xmath313 of @xmath310 , the map @xmath312 is transverse to @xmath313 .
[ transversedef ] note that the above notion does not depend on the analytic , but just the underlying smooth structures .
let @xmath2 be a real analytic manifold with a sw - stratification @xmath285 , and let @xmath10 be a subanalytic set which is a union of strata .
let @xmath295 be the strata of @xmath1 , and let @xmath314 denote this induced stratification of @xmath1 .
let @xmath315 be a smooth real analytic submanifold , subanalytic in @xmath2
. then if the inclusion map @xmath316 is transverse to the stratification @xmath314 , the intersected stratification @xmath317 of @xmath318 is defined by @xmath319 this stratification @xmath317 defines a sw - stratification of @xmath318 .
in particular , each @xmath320 is a union of strata @xmath321 of @xmath317 of the _ same dimension_.[interstrat ] since each @xmath295 is subanalytic and locally closed , and @xmath27 being a real analytic submanifold , is also locally closed and given to be subanalytic , the sets @xmath320 are subanalytic and locally closed , by ( i ) of proposition [ subanalfacts ] . thus the connected components @xmath321 of @xmath320 are locally closed , and subanalytic by ( viii ) of proposition [ subanalfacts ] .
since @xmath295 meets @xmath27 transversely for each @xmath301 , the @xmath322 are real analytic submanifolds of @xmath2 .
hence ( sw1 ) follows .
( sw2 ) ( local finiteness ) also follows from the second statement of ( viii ) in [ subanalfacts ] . to see ( sw3 ) , we need to show that if @xmath323 , then @xmath324 .
clearly , @xmath325 is closed in @xmath321 .
we claim that it is also open in @xmath321 . for , let @xmath326 .
this last intersection being non - empty implies by ( sw3 ) applied to @xmath295 and @xmath327 , that @xmath328 . by the topological local triviality of whitney ( b ) stratifications ( see [ g - m ] , 1.4 or [ ma ] , corollary 8.4 ) , and the remark in definition [ stratdef ] that sw - stratifications are whitney ( b ) stratifications , there exists a connected neighbourhood @xmath3 of @xmath44 in @xmath295 such that a neighbourhood @xmath329 of @xmath44 in @xmath327 is of the form @xmath330 , where @xmath331 is the _ normal slice to @xmath44 _ inside @xmath327 . note that @xmath331 , being the cone on the link @xmath332 , is connected . since the real analytic submanifold @xmath27 intersects each stratum
transversely , it follows by intersecting everything with @xmath27 , that there is a neighbourhood @xmath333 of @xmath44 in @xmath320 such that the neighbourhood @xmath334 of @xmath44 in @xmath335 is of the form @xmath336 .
if we choose @xmath337 to be the connected component of @xmath333 containing @xmath44 , and @xmath338 to be the connected component of @xmath339 containing @xmath44 , it follows that @xmath338 is of the form @xmath340 , where @xmath341 is the intersection of @xmath331 and @xmath342 .
thus @xmath343 , and so this last intersection is open in @xmath321 , and our claim is proven .
the connectedness of @xmath321 implies that the open and closed subset @xmath344 is equal to @xmath321 , and ( sw3 ) follows for our stratification @xmath345 . to prove the verdier condition ( sw4 ) for @xmath346 , we first need the following linear algebraic claim .
let @xmath347 be a finite dimensional inner product space , with a fixed linear subspace @xmath348 .
for every subspace @xmath278 of @xmath4 intersecting @xmath348 transversely ( viz .
@xmath349 ) there exists a positive constant @xmath350 , depending continuously on @xmath278 , such that for all linear subspaces @xmath351 , we have @xmath352 where @xmath276 is the verdier distance with respect to @xmath353 introduced in definition [ verdiercond ] . first note that @xmath354 implies @xmath355 , where @xmath356 denotes orthogonal complement with respect to @xmath353 .
thus the orthogonal projection @xmath357 onto @xmath278 is an isomorphism when restricted to @xmath358 .
hence the subspace @xmath359 has dimension @xmath360 which , by transversality of @xmath278 and @xmath348 , is precisely the codimension of @xmath361 in @xmath278 .
also , for @xmath362 , we have @xmath363 , thus implying that @xmath359 is orthogonal to @xmath361 .
hence @xmath364 is the orthogonal complement of @xmath361 in @xmath278 .
define a new inner product @xmath365 on @xmath4 by setting @xmath366 for @xmath367 , @xmath366 for @xmath368 , and @xmath369 .
the orthogonal complement of a subspace @xmath329 with respect to @xmath365 will be denoted @xmath370 .
then , with respect to this new inner product @xmath365 , we have an orthogonal direct sum decomposition @xmath371 where ` @xmath372 ' signifies that the decomposition is orthogonal with respect to @xmath365 .
also , @xmath373 .
thus @xmath374 .
hence , by counting dimensions , and noting that @xmath375 , it follows that @xmath376 hence for any @xmath377 , we will have @xmath378 .
thus , denoting the verdier distance with respect to @xmath365 by @xmath379 , we have @xmath380 ( all orthogonal projections in the last six lines are with respect to @xmath381 . )
now , by the above definition of @xmath365 , it follows that there is a positive constant @xmath382 depending continuously on @xmath278 such that for every @xmath383 : @xmath384 from which it is easy to deduce that there is an inequality @xmath385 where @xmath350 is a positive constant depending continuously on @xmath278 . also since @xmath353 and @xmath365 agree on @xmath348
, it follows that @xmath386 .
hence the inequality ( [ verdierineq ] ) follows by applying ( [ deltaprime ] ) , and hence the claim . if @xmath387 , then @xmath388 and we have by ( sw4 ) applied to @xmath389 , that there is some neighbourhood @xmath3 of @xmath271 such that for all @xmath390 and @xmath391 we have @xmath392 let @xmath337 be the connected component of @xmath393 containing @xmath271 , and simultaneously locally trivialise on @xmath3 the manifold pair @xmath394 so that the bundle pair @xmath395 is isomorphic to @xmath396
. set @xmath397 and @xmath398 , so that @xmath399 and @xmath400 , by the fact that @xmath27 meets @xmath295 and @xmath401 transversely . for @xmath402
, one can find a bound @xmath189 for the constant @xmath350 in ( [ verdierineq ] ) by the continuity of @xmath350 in @xmath278 .
now we apply the inequality ( [ verdierineq ] ) above to get the verdier condition ( w ) on @xmath337 .
thus ( sw4 ) is verified for the stratification @xmath346 .
this proves the lemma .
we also need a straighforward extension of proposition [ interstrat ] .
namely , in the above situation let @xmath27 be a closed connected subset , which is a real analytic submanifold with real analytic boundary @xmath403 .
this can be regarded as a sw - stratified subset of @xmath2 with two strata , viz . @xmath404 and @xmath403 . for this stratification , ( sw1 ) , ( sw2 ) and ( sw3 )
are straightforward , and ( sw4 ) holds because at any point @xmath405 , the germ of the triple @xmath406 is ( by definition ) real analytically isomorphic to @xmath407 , which obviously satisfies ( sw4 ) .
let @xmath10 be a sw - stratified subset , such that @xmath408 and @xmath409 for each stratum @xmath295 of @xmath1 .
then the intersection stratification on @xmath318 defined by taking the connected components of all the intersections @xmath410 and @xmath411 is an sw - stratification of @xmath318 .
the proof is exactly the same as in the case of [ interstrat ] above .
[ interstrat2 ] ( stratified submersion ) ( see [ g - m ] , 1.5 on p. 41 and [ ve ] , 3.2)@xmath9
let @xmath48 be a smooth map , and let @xmath10 be a sw - stratified subset , with strata @xmath307 .
we say @xmath26 is a _ stratified submersion _ if for each stratum @xmath295 of @xmath1 , the restriction @xmath312 is smooth and a submersion .
[ stratsubdef ] again , the above notion depends only on the underlying smooth structures . in [ ve ] , definition 3.2 , such a map is referred to as a map _ ` transverse to the stratification ' _
( on @xmath1 ) .
the following is a key proposition due to verdier .
( see @xmath412}$ ] , theorem 4.14 ) .
suppose that @xmath1 is a _ closed _ sw - stratified subset in @xmath2 , with stratification @xmath285 .
let @xmath27 be a smooth real analytic manifold , @xmath48 be a real analytic map , _ proper on _ @xmath1 , and a stratified submersion .
let @xmath413 , and @xmath414 and @xmath415 be the fibres over @xmath416 of @xmath2 and @xmath1 respectively .
( note that by the hypothesis , @xmath414 meets all the strata of @xmath1 transversely , so that @xmath417 acquires an induced stratification @xmath418 via the connected components of @xmath419 , as in [ interstrat ] above ) .
then there exists an open neighbourhood @xmath174 of @xmath416 in @xmath27 , and a homeomorphism of the stratified spaces @xmath420 onto @xmath421 preserving the stratifications and compatible with the projections to @xmath174 .
( again note that @xmath422 , being an open subset of @xmath1 , has a natural induced stratification , also denoted by @xmath285 , from @xmath1 , and @xmath423 has a natural product stratification @xmath424.)[firstiso ] in fact , in theorem 4.14 of [ ve ] , verdier proves that the homeomorphism above is a ` rugeux ' ( coarse ) homeomorphism , which is slightly stronger than saying that it is a homeomorphism .
we do not need this stronger statement . the corresponding statement for whitney
( b ) stratifications is thom s first isotopy lemma , and due to thom and mather ( see [ 7 ] in [ ve ] ) .
let @xmath1 be a topological space , and @xmath4 be a continuous real vector bundle of rank @xmath158 on @xmath1 .
let @xmath425 be some continuous bundle metric , and for @xmath426 let @xmath427 denote the @xmath184-disc bundle of @xmath4 with respect to this bundle metric .
denote the zero - section of this bundle by @xmath428 .
( tubular neighbourhoods)@xmath9 let @xmath157 be as above and let @xmath429 be a continuous section .
we will say that @xmath7 _ has a tubular neighbourhood in _
@xmath1 if there exists a neighbourhood @xmath174 of @xmath430 in @xmath1 , and @xmath426 , and a homeomorphism @xmath431 such that the composite @xmath432 is the map @xmath433 for all @xmath434 .
[ tubnbddef ] @xmath9 1 .
if @xmath180 and @xmath435 are isomorphic real vector bundles , via a continuous bundle isomorphism @xmath436 , then clearly a section @xmath7 of @xmath4 will have a tubular neighbourhood in @xmath1 iff the section @xmath437 of @xmath277 has a tubular neighbourhood in @xmath1 .
we just need to observe that @xmath438 , and that @xmath439 will induce a homeomorphism of the disc bundle @xmath440 with the disc bundle @xmath427 where @xmath4 is given the @xmath439-pullback bundle metric from @xmath277 .
2 . more generally , if there is a bundle diagram : @xmath441 with @xmath26 a _ homeomorphism _ , and @xmath439 a continuous vector bundle equivalence , then the section @xmath442 of @xmath443 has a tubular neighbourhood in @xmath444 iff the section @xmath445 of @xmath446 has a tubular neighbourhood in @xmath106 .
this is clear because we will have a continuous bundle equivalence @xmath447 on @xmath1 , the statement ( i ) above , and the fact that the homeomorphism @xmath26 induces a homeomorphism of any neighbourhood @xmath174 of @xmath448 with the neighbourhood @xmath449 of @xmath450 .
[ tubnbdrmk1 ] before proving the main result , we need some lemmas .
( transversality to the @xmath35-section in smooth case ) ( see theorem 1.3.6 on p. 39 in [ g - m ] ) .
let @xmath2 be a real analytic manifold , and @xmath238 be a real analytic vector bundle of rank @xmath158 on @xmath2 , generated by an @xmath170-dimensional vector space @xmath0 of real analytic global sections .
let @xmath451 be a real analytic submanifold ( in particular , locally closed in @xmath2 ) .
then there exists a baire subset ( i.e. whose complement is of measure @xmath35 , in particular , dense ) @xmath452 such that @xmath5 implies that @xmath453 , is transverse to the zero - section @xmath428 of @xmath4.[transmooth ] by definition , and ( ii ) of lemma [ analclass ] , we have the exact sequence of real analytic bundles on @xmath2 : @xmath454 which is the pullback via the analytic classifying map @xmath240 of the last two terms of the universal exact sequence ( [ univseq ] ) on @xmath168 .
in particular @xmath455 above is real analytic .
the section @xmath178 is viewed as an analytic section of @xmath4 by taking the map @xmath456 of @xmath4 . since @xmath184 is a smooth epimorphism of bundles
, it is a smooth submersion . restricting the above real analytic map @xmath184 to @xmath310
, we have a real analytic map : @xmath457 we claim that this map @xmath112 is transverse to the zero - section @xmath428 of @xmath4 . for , if @xmath458 such that @xmath459 , the fact that @xmath460 is a linear surjection shows that the partial derivative @xmath461 takes the linear subspace @xmath462 of @xmath463 surjectively onto the tangent space @xmath464 of the fibre @xmath465 .
hence the image of the total derivative @xmath466 contains @xmath464 . since @xmath467
, it follows that @xmath468 and @xmath469 together span @xmath470 .
hence @xmath112 is transverse to the zero - section @xmath428 of @xmath4 .
since @xmath428 is a codimension @xmath158 smooth submanifold of @xmath4 , the inverse image @xmath471 is a smooth real - analytic submanifold of @xmath472 , of dimension @xmath473 .
let @xmath474 denote the second projection , and consider the followingdiagram : @xmath475 note that the fibre @xmath476 is precisely the set of zeroes of the section @xmath477 of the restricted bundle @xmath478 .
we make the following : the section @xmath479 is transverse to the zero - section @xmath428 of @xmath4 iff @xmath178 is a regular value of @xmath480 .
let @xmath481 denote the natural quotient map .
since @xmath482 is transverse to @xmath428 , for each @xmath483 we have the diagram : @xmath484 in which the columns and last two rows are exact . by the snake lemma , and the fact that @xmath485
, we have an exact sequence : @xmath486 consequently , the map @xmath487 is surjective iff @xmath488 is surjective .
since @xmath489 , @xmath7 is a regular value of @xmath490 iff @xmath487 is surjective for all @xmath491 , that is , iff @xmath492 is transverse to @xmath428 . hence the claim . by sard
s theorem , there is a baire subset @xmath452 of regular values for @xmath480 .
hence the lemma follows , for this choice of @xmath3 .
we have the following immediate corollary for subanalytic sets .
( stratified transversality to the @xmath35-section)@xmath9 let @xmath2 be a real analytic manifold which is countable at @xmath284 .
let @xmath285 be a sw - stratification of @xmath2 , and let @xmath493 denote the strata ( see definition [ stratdef ] ) .
let @xmath4 be a real analytic vector bundle of rank @xmath158 on @xmath2 , generated by an @xmath170-dimensional real vector space @xmath0 of real - analytic global sections .
then there exists a baire ( in particular dense ) subset @xmath452 such that for @xmath5 , we have 1 .
@xmath494 is transverse to @xmath428 , and 2 .
@xmath495 ( see definition [ transversedef ] ) for each @xmath496 .
3 . for any @xmath10 which is a union of strata ( and hence itself sw - stratified ) , @xmath497 .
[ transubanal ] since the manifold @xmath2 is a countable union of compact sets , and a compact set can intersect only finitely many members of locally finite collection , and sw - stratifications are locally finite , it follows that the sw - stratification above consists of only countably many strata .
we thus assume that @xmath498 .
in particular , @xmath2 is the union : @xmath499 of countably many strata .
set @xmath500 , as a notational convention . by ( i ) of definition [ stratdef ] , each @xmath302 is a locally closed real analytic submanifold of @xmath2 . applying the previous lemma [ transmooth ] to @xmath501 for @xmath502
, we find a baire subset @xmath503 such that for @xmath504 , @xmath505 .
the set @xmath506 , being the countable intersection of baire sets , is a baire set ( the countable union of measure zero sets is measure zero ) . clearly for @xmath5 , ( i ) and
( ii ) follow .
( iii ) is obvious from ( ii ) bydefinition [ transversedef ] .
let @xmath2 be a _ compact _ real analytic manifold , and @xmath10 a subanalytic set ( not necessarily closed ) in @xmath2 . in accordance with theorem [ refine1 ]
above , equip @xmath2 with a sw - stratification @xmath507 by strata @xmath508 such that the subanalytic set @xmath1 is a union of strata .
( note that the strata form a finite collection by ( ii ) of remark [ stratrmk ] ) .
let @xmath238 be a real analytic vector bundle of rank @xmath158 , generated by an @xmath170-dimensional vector space @xmath0 of real analytic global sections .
then , there exists a baire subset @xmath452 such that for @xmath178 , the section @xmath509 of the restricted subanalytic bundle @xmath510 has a tubular neighbourhood in the sense ofdefinition [ tubnbddef ] .
[ strattubular ] the proof is rather involved , though the essential idea is contained in the sketch of the proof of theorem 1.11 , p. 47 in [ g - m ] .
the main steps are as follows : equip the compact manifold @xmath2 with a finite ( sw)-stratification by @xmath302 so that the subanalytic set @xmath1 is a union of strata , by theorem [ refine1 ] . by corollary [ transubanal ] , for a baire subset @xmath452 and @xmath5 , the restriction of @xmath7 to _ each stratum
_ is transverse to @xmath428 . let @xmath8 denote the zero locus of @xmath7 in @xmath2 , and @xmath511 .
next we choose a thin enough @xmath184-disc bundle in @xmath4 which intersects only those strata @xmath512 of @xmath513 that meet @xmath428 ( i.e. for which @xmath514 ) .
now , shrink down this @xmath184-disc bundle to the zero - section @xmath428 via the scaling map @xmath515 of @xmath4 , @xmath516 .
the final step is to prove , using the first isotopy lemma [ firstiso ] that the @xmath517-family of intersections of these shrinking neighbourhoods with @xmath513 is a stratified product @xmath518 , where @xmath519 is some interval containing @xmath35 . for this family ,
the fibre over @xmath520 is the bundle restricted to @xmath8 , and the fibre over @xmath521 , is a homeomorphic image of the above neighbourhood .
thus these fibres are both stratified homeomorphic to @xmath310 .
because it is a stratified product , and @xmath1 is a union of strata , one can restrict the ( stratification preserving ) homeomorphisms above to @xmath1 .
we carry out the details below .
define @xmath500 for notational convenience .
we know by corollary [ transubanal ] that for the given stratification of @xmath522 above , there exists a baire subset @xmath452 such that @xmath5 implies that @xmath523 is transverse to the zero section @xmath428 of @xmath4 for each @xmath524 .
( note that @xmath525 is transverse to the zero section of @xmath526 iff @xmath527 is transverse to the zero section @xmath428 of @xmath4 . ) in particular @xmath509 is transverse to the zero section @xmath428 of @xmath4 ( or , in the notation of definition [ transversedef ] , , for the real analytic map @xmath494 ) .
thus the zero - locus @xmath528 is a real analytic submanifold of @xmath2 , and the restriction of @xmath8 to @xmath302 , viz .
the zero - locus @xmath529 is a real analytic submanifold of @xmath302 , and hence @xmath2 , for each @xmath530 .
denote the restricted bundle @xmath526 by @xmath531 .
we shall prove that for such an @xmath5 , the section @xmath532 has a tubular neighbourhood in @xmath1 , in the sense of definition [ tubnbddef ] .
since @xmath238 is isomorphic as a real analytic bundle to @xmath234 , and since @xmath533 is a real analytic subbundle of @xmath172 via the real analytic splitting coming from a constant ( hence real analytic ) bundle metric on @xmath534 , it follows that @xmath238 has a real analytic pullback bundle metric @xmath281 .
let @xmath535 denote the first projection .
then the bundle @xmath536 , which is precisely @xmath537 , has the pulled back metric , also denoted @xmath281 , also real analytic .
note that @xmath538 has the product stratification denoted @xmath539 , and @xmath540 is a unionof strata .
similarly @xmath4 has the sw - stratification @xmath541 , with strata @xmath542 .
( it is easy to verify , using the local analytic triviality of @xmath4 , that @xmath543 satisfies ( sw1 ) through ( sw4 ) .
also @xmath531 are subanalytic inside @xmath4 because @xmath302 are subanalytic in @xmath2 and ( vi ) of proposition [ subanalfacts ] . )
similarly , @xmath544 is sw - stratified by @xmath545 , which is the same as the sw - stratification @xmath546 . consider the real analytic map : @xmath547 and denote @xmath548 by @xmath549 .
denote the restricted bundle map @xmath550 by @xmath551 . the map @xmath552 is transverse to the smooth real analytic submanifold @xmath553 . in fact we shall show that for each @xmath554
, the map @xmath555 is transverse to @xmath513 . since for each @xmath556 , the image @xmath557 contains @xmath558
, this will prove the assertion .
if @xmath559 , we have @xmath555 is a real analytic diffeomorphism , and hence transverse to @xmath513 .
if @xmath560 , we have @xmath561 equal to projection onto the zero - section @xmath428 , viz .
the map @xmath562 .
thus for @xmath563 , if we denote @xmath564 , we have @xmath565 and @xmath566 .
if @xmath567 , it follows that @xmath568 , so that @xmath434 . by the choice of @xmath7 , @xmath7 is transverse to @xmath428 , so that we have @xmath569 .
that is , @xmath570 .
thus @xmath571 is also transverse to @xmath513 , and our claim follows . by the transversality above , it follows that @xmath572 is a real analytic submani- fold of @xmath573 , whose fibre over @xmath556 is the real analytic manifold @xmath574 .
since @xmath494 is a real analytic embedding , and sections of bundles are proper maps , it follows that @xmath513 is a smooth real analytic subspace of @xmath4 .
in particular it is closed .
since @xmath575 is real analytic , it follows that @xmath576 is a real analytic subspace in @xmath544 which is closed . by the transversality of @xmath575 to @xmath513
, it follows that @xmath27 is a smooth real analytic subspace of @xmath544 .
next we have the real analytic submanifold @xmath27 is transverse to the zero section @xmath577 , and the intersection @xmath578 is @xmath579 , where @xmath580 is the zero locus of @xmath7 .
since @xmath494 is transverse to @xmath428 by the choice of @xmath7 above , it follows that for each @xmath581 ( @xmath582 ) we have @xmath583 now for @xmath584 , we have the identifications @xmath585 and @xmath586 . with this identification , clearly @xmath587 for @xmath588 ,
since @xmath575 restricted to @xmath589 is the first projection to @xmath428 .
let us denote @xmath590 for notational convenience . then by the foregoing remark and ( [ tangeqn1 ] ) above
, we have : @xmath591 for any linear map @xmath348 , any two linear subspaces @xmath592 , we have the identity @xmath593 . substituting @xmath594 and @xmath595 , and using ( [ tangeqn2 ] ) above we find @xmath596 since by claim @xmath597
, @xmath575 is transverse to @xmath513 , we have @xmath598 .
thus we have @xmath599 that is , @xmath27 meets @xmath600 transversely .
the first part of claim @xmath601 follows . to see the second part of claim @xmath601 , note that @xmath602 iff @xmath603 is in @xmath513 . but
@xmath581 iff @xmath604 iff @xmath605 .
thus @xmath606 , and claim @xmath601 is proved . in the proofs of the above two claims ,
the transversality of @xmath7 to the zero section was the only fact used . by the choice of @xmath5 we have @xmath607 is transverse to the zero section @xmath608 for all @xmath71 .
thus , in the proofs of the above two claims , we replace @xmath2 by the stratum @xmath302 , @xmath4 by @xmath531 , @xmath7 by @xmath523 , @xmath575 by @xmath609 , and repeat everything ( noting that @xmath607 is transverse to @xmath608 iff @xmath527 is transverse to @xmath428 ) . by so doing , for @xmath524 , we obtain : @xmath610 is transverse to @xmath512 , and thus @xmath611 is a smooth real analytic subspace of @xmath531 , and is equal to the intersection @xmath612 .
( for notational consistency , we set @xmath613 . )
@xmath614 intersects @xmath577 transversely , and this intersection is @xmath615 . from the given sw - stratification @xmath285 in the hypothesis of the theorem
, the smooth analytic subspace @xmath513 gets an induced sw - stratification @xmath616 , defined by the strata @xmath617 .
( note @xmath512 are subanalytic in @xmath4 by ( ii ) of proposition [ subanalfacts ] because @xmath7 , being a continuous section of a bundle , is proper . )
the manifold @xmath513 meets the strata @xmath531 of @xmath618 transversely ( inside @xmath4 ) , and @xmath27 meets the strata @xmath619 of @xmath620 transversely ( inside @xmath573 ) . to prove the first assertion ,
let @xmath621 .
since @xmath238 satisfies @xmath622 , the image @xmath623 is a vector space complement to @xmath624 inside @xmath625 , and hence @xmath626 since @xmath627 for all @xmath628 , it follows that @xmath629 and hence @xmath513 is transverse to @xmath531 for all @xmath71 .
the first assertion of claim ( 3 ) follows .
now we prove the second assertion of claim ( 3 ) , i.e. @xmath27 meets @xmath619 transversally for all @xmath630 inside @xmath544 .
we have already observed that @xmath631 is surjective for all @xmath559 ( because @xmath555 is a diffeomorphism ) .
also for such a @xmath559 , @xmath632 maps @xmath633 onto @xmath166 .
hence , from ( [ smtrans ] ) we conclude that @xmath634 for @xmath635 with @xmath559 , and @xmath636 . applying the identity @xmath637 to the linear map @xmath638 , and noting that @xmath639 by definition and claim @xmath597
, we obtain @xmath640 since @xmath641 for all @xmath524 , we obtain that the intersection of @xmath27 with @xmath642 is transverse at all points @xmath643 for @xmath559
. now we look at the situation where @xmath560 .
if a point @xmath644 , we have that @xmath645 and @xmath646 . from the paragraph
preceding the proof of claim @xmath647 , we have that @xmath648 is transverse to the zero - section @xmath608 in @xmath531 , so that @xmath649 since @xmath650 is a complement to @xmath166 in @xmath651 , for each @xmath628 , and @xmath630 , it follows that @xmath652 let @xmath653 denote the normal bundle to @xmath302 in @xmath2 , and @xmath654 its fibre at @xmath628 . since @xmath655 and @xmath656 , we have @xmath657 @xmath561 is just the projection of @xmath4 to its zero section @xmath428 , and hence @xmath658 @xmath659 , so that @xmath660 contains @xmath661 . combining with eq .
( [ smtrans4 ] ) , we have @xmath662 now we apply our lemma @xmath663 to the linear map @xmath664 , @xmath665 , @xmath666 , and noting that by definition and claim @xmath597 , @xmath667 , we obtain @xmath668 which shows that @xmath27 meets @xmath619 transversely at @xmath669 and our claim ( 3 ) follows . since @xmath670 , the first assertion of claim ( 3 ) shows that the stratification @xmath616 is just the intersection stratification @xmath671 as defined in lemma [ interstrat ] .
in particular , by that same lemma , @xmath616 is a ( sw)-stratification of @xmath513 . by the second assertion of claim ( 3 ) and lemma [ interstrat ]
, we can give the sw - stratification @xmath672 to the smooth real analytic subspace @xmath576 .
the strata are precisely the connected components of @xmath673 .
now we need another assertion : for each @xmath524 , define @xmath674 to be the restriction of the second projection @xmath675 to @xmath676 .
the derivative @xmath677 is surjective at all points @xmath678 .
we now prove claim ( 4 ) . for the curve @xmath679 in @xmath573 ,
starting at @xmath680 with initial velocity @xmath681 , we have @xmath682 for all @xmath7 so that @xmath683 for all @xmath554 .
thus the subspace @xmath684 lies in @xmath685 . since ( by claim @xmath647 above ) , @xmath686 , it contains this kernel @xmath685 .
thus @xmath687 for all @xmath688 .
since @xmath689 maps the subspace @xmath690 isomorphically onto @xmath691 , it follows that @xmath677 is surjective at all points @xmath692 and all @xmath524 . hence claim ( 4 )
is proved .
recall that @xmath693 , and by the second part of claim @xmath694 , we have @xmath695 . from claim @xmath696 , and by continuity of @xmath689 , we have the following claim . for each @xmath71 such that @xmath514 , and for @xmath678 , there exists a neighbourhood @xmath697 of @xmath680 in @xmath573 such that the derivative @xmath698 is surjective for all @xmath699 .
renumber the strata so that @xmath700 for @xmath701 and @xmath702 for @xmath703 .
again , for notational convenience , set @xmath704 . for @xmath705 , let @xmath706 and @xmath707 denote the open @xmath276-disc bundle and the @xmath276-sphere bundles of @xmath4 respectively ( with respect to the above real analytic bundle metric @xmath281 ) . for each @xmath708 , and for @xmath709
, there exists a neighbourhood @xmath697 of @xmath680 in @xmath573 such that @xmath710 meets @xmath711 transversely for all @xmath712 .
( note that for @xmath276 very large , the intersection @xmath711 might be empty , in which case this assertion is vacuously true . ) to see this , we first prove that for @xmath708 and @xmath713 , there is a neighbourhood @xmath714 of @xmath715 in @xmath4 such that @xmath716 meets @xmath717 transversely for all @xmath718 .
since this assertion is local around @xmath715 , and @xmath4 is locally trivial , we may assume without loss of generality that @xmath4 is trivial . by this triviality
, there is a natural linear surjection : @xmath719 for each @xmath563 .
the fact that @xmath512 meets @xmath428 transversely at @xmath715 implies that the composite @xmath720 is surjective . by continuity
, there exists a neighbourhood @xmath714 of @xmath715 in @xmath4 such that the composite @xmath721 is surjective for all @xmath722 . at a point @xmath723 ,
it is obvious that the one - dimensional quotient space @xmath724 is a quotient of @xmath725 ( by the tangent space to the sphere fibre @xmath726 ) .
thus the composite map @xmath727 is also surjective for all @xmath728 .
but this is precisely the statement that @xmath716 meets @xmath717 transversely inside @xmath4 . for the sake of convention
, we define @xmath729 for @xmath730 .
then , by the fact that @xmath512 is transverse to @xmath428 , we have that @xmath716 meets @xmath717 transversely in @xmath4 for all @xmath731 .
now let @xmath556 , and let @xmath697 be a neighbourhood of @xmath680 such that @xmath732 , @xmath714 as above .
hence @xmath733 implies that @xmath734 . by the above choice of the neighbourhood @xmath714 , at the point @xmath735
, we have @xmath736 if @xmath737 with @xmath738 , then @xmath739 is a diffeomorphism . in this event ,
the derivative @xmath740 is surjective . taking the inverse image under this surjective map @xmath741 of both sides of the equality ( [ tangineq3 ] ) , noting that @xmath575 is transverse to @xmath512 , and that @xmath742 is a diffeomorphism for @xmath738 , we have @xmath743
since @xmath744 contains the subspace @xmath745 , we see that at all points @xmath746 with @xmath738 , which lie in @xmath697 and in @xmath747 , the intersection of @xmath748 and @xmath676 is transverse . to check the case @xmath749 , let @xmath750 .
note @xmath751 where @xmath752 and @xmath238 is the bundle projection . since @xmath753
, we have @xmath754 .
thus , setting @xmath755 , we have @xmath756 , since the projection @xmath757 is a submersion .
since @xmath758 , and @xmath512 intersects @xmath428 transversely , we have @xmath759 from which it follows that @xmath760 which implies @xmath761 again we use the identity @xmath637 for any linear map @xmath348 , and any subspaces @xmath762 and obtain , by setting @xmath763 and @xmath764 , that @xmath765 which is to say , @xmath766 thus the intersection of @xmath710 and @xmath767 is transverse .
this proves claim @xmath768 .
since @xmath2 is compact , @xmath428 is a compact subset of @xmath4 .
since , by claim @xmath694 above , @xmath676 intersects @xmath577 transversely for each @xmath524 , each connected component of each @xmath676 intersects @xmath577 transversely inside @xmath573 .
thus no such connected component can be contained in @xmath577 .
that is , no stratum of the stratification @xmath769 of the smooth manifold @xmath576 is contained in @xmath577 .
we now note that since every stratum @xmath512 of @xmath513 meets @xmath428 transversely , @xmath770 for @xmath771 will imply that the codimension of @xmath772 in @xmath773 is the same as the codimension of @xmath512 in @xmath774 , which is non - zero . thus if @xmath775 , the union @xmath776 has non - zero codimension in @xmath773 .
thus @xmath777@xmath9 that is , @xmath775 if and only if @xmath778 , if and only if @xmath779 . by noting that @xmath780 , and @xmath781 , the same fact obtains for the @xmath782 , viz .
@xmath783 if and only if @xmath784 which , in turn , happens if and only if @xmath779 . consider the closed subset @xmath189 of @xmath573 defined by @xmath785 which is disjoint from @xmath577 , by the preceding paragraph . since @xmath189 is closed and @xmath786 $ ] is compact , there will exist an @xmath787 such that the restricted @xmath788-disc bundle @xmath789 $ ] does not intersect @xmath189 . thus @xmath790 $ ]
is disjoint from @xmath676 for @xmath703 .
we saw above in claim @xmath694 that @xmath695 , and similarly @xmath791)=s(z_{i})\times [ -2,2]$ ] .
thus @xmath676 intersects @xmath789 $ ] if and only if @xmath676 intersects @xmath786 $ ] , and this happens if and only if @xmath792 , i.e. if and only if @xmath708 .
we record this fact in claim ( 7 ) @xmath793)\cap n_{i}=\phi$ ] iff @xmath794 . by reducing @xmath184 if necessary , and from claims ( 5 ) and ( 6 )
, using the compactness of @xmath795 $ ] , we can further assert the following claim ( 8) . for @xmath708 ,
the derivative @xmath796 is surjective for all @xmath797\right)\cap n_{i}$ ] . for @xmath708 ,
the intersection of @xmath798 $ ] with @xmath676 is transverse for all @xmath799 .
we need an analogue of claim ( 8) above for @xmath800 .
that is for @xmath184 small enough , and @xmath801 such that the intersection the derivative @xmath796 is surjective for all @xmath802)\cap n_{i}$ ] and all @xmath803 .
we note that the tangent subspace @xmath804 has the line complement @xmath805 in the tangent space @xmath806 .
since @xmath807 annihilates all vectors in @xmath808 , it annihilates the subspace @xmath809 , and hence @xmath810 .
thus the @xmath689 image of @xmath811)\cap n_{i})$ ] is the same as the @xmath689 image of @xmath812)\cap n_{i})$ ] , which is all of @xmath813)$ ] by claim ( 5 ) above .
thus claim ( 10 ) is proved .
consider @xmath814 as a sw - stratified subspace of the analytic manifold @xmath815 with just the two strata @xmath816 and @xmath817 , as in remark [ interstrat2 ] . by the fact that @xmath816 is open ( and subanalytic in @xmath818 ,
since it is the inverse image of @xmath819 under the analytic map @xmath820 ) , and the fact that @xmath27 is sw - stratified by the connected components of @xmath676 , it follows that @xmath821 is sw - stratified by the connected components of the analytic submanifolds @xmath822 . by claim ( 7 )
above , we have @xmath823 similarly by claim ( 9 ) above , and remark [ interstrat2 ] , the subset is sw - stratified by the connected components of the analytic submanifolds ( note that @xmath800 is a smooth real analytic subspace of @xmath4 since @xmath820 is a real analytic function , and @xmath824 is a real analytic manifold with boundary @xmath800 ) .
in fact , by remark [ interstrat2 ] , @xmath825 is sw - stratified by the connected components of the real analytic submanifolds @xmath826 and @xmath827 for @xmath828 .
let us call these connected components @xmath829 , where @xmath830 for some finite set @xmath277 .
thus the subset @xmath831 is a sw - stratified space in @xmath815 with stratification @xmath832 by @xmath829 . by claims ( 8) and ( 10 ) above , for @xmath708 the maps @xmath833 and @xmath834 are submersions .
thus they are submersions when restricted to each connected component .
in particular , @xmath835 is a submersion for each @xmath830 .
that is @xmath836 is a stratified submersion in the sense of definition [ stratsubdef ] . for any compact subset @xmath837
, @xmath838 is a closed set , and contained in the compact set @xmath839 .
( @xmath824 is compact since @xmath2 is compact ! ) .
thus @xmath840 is compact , implying that @xmath841 is proper . by the first isotopy lemma [ firstiso ] ,
applied to the analytic map @xmath842 and the closed sw - stratified subset @xmath843 , it follows that for each point @xmath844 , there is a neighbourhood @xmath845 of @xmath517 , and a stratum preserving rugeux homeomorphism @xmath846 where @xmath847 , such that @xmath848 ( here @xmath849 is the second projection on the right hand side ) .
that is , @xmath850 is a topologically locally - trivial stratified fibre bundle . by the compactness and connectedness of @xmath851\subset ( -2,2)$ ]
, there is therefore a stratification preserving rugeux homeomorphism @xmath852 between the fibres @xmath853 and @xmath854 .
but @xmath855 by definition .
since @xmath856 is the identity map , this last set is stratified homeomorphic to @xmath857 . on the other hand , the map @xmath561 is the bundle projection map @xmath858 .
thus @xmath859 where @xmath6 is the zero - locus of @xmath7 .
thus @xmath854 is stratified homeomorphic to @xmath860 .
note that @xmath861 is an open neighbourhood of @xmath862 . by the last para ,
this neighbourhood is stratified homeomorphic to @xmath863 .
the fact that this homeomorphism preserves strata shows that it maps @xmath864 homeomorphically to @xmath865 , where @xmath511 for @xmath701 ( recall for @xmath866 , the above sets are empty ) . since @xmath1 is a union of strata from among the @xmath302 , and
this last homeomorphism is stratum preserving , it follows that this homeomorphism when restricted to the open neighbourhood @xmath867 of @xmath868 , is again a stratum preserving homeomorphism @xmath869 .
since the analytic embedding @xmath870 defines a homeomorphism between a neighbourhood of @xmath871 and @xmath872 , theorem [ strattubular ] follows .
let @xmath2 be a compact real analytic manifold , and @xmath10 be a subanalytic set .
let @xmath4 be a subanalytic bundle of rank @xmath158 over @xmath1 generated by an @xmath170-dimensional vector space of global sections @xmath0 in the sense of definition [ subanalbundledef ] .
then , for a baire subset @xmath452 , and @xmath873 there exists a neighbourhood of @xmath874 in @xmath1 which is homeomorphic to an @xmath184-disc bundle @xmath863 for some @xmath426 ( i.e. @xmath875 has a tubular neighbourhood in the sense of definition [ tubnbddef ] ) . [ strattubular2 ] by lemma [ subanallemma ] , there is a subanalytic pseudoequivalence @xmath876 , with @xmath106 a subanalytic set in @xmath89 , such that our given bundle @xmath4 is the pullback @xmath877 of a real analytic rank-@xmath158 vector bundle @xmath189 on @xmath89 , generated by a space of global analytic sections @xmath0 .
further , since @xmath2 is compact , and @xmath211 , @xmath89 is also compact . by ( ii ) of remark [ tubnbdrmk1 ]
, the tubular neighbourhood theorem for @xmath4 on @xmath1 follows from the stratified tubular neighbourhood theorem 1 ( i.e. theorem [ strattubular ] ) applied to @xmath189 , @xmath106 and @xmath89 above .
theorem [ strattubular2 ] covers the case of @xmath1 being any real projective or affine variety .
in fact , any real projective or affine algebraic constructible set can be regarded as a subanalytic subset in projective space . in the analytic situation too , any real analytically constructible subset of a compact real analytic manifold automatically becomes a subanalytic set .
the main theorem [ strattubular2 ] applies in all of the above situations , provided the bundle @xmath4 is a subanalytic bundle generated by global sections in the sense of definition [ subanalbundledef ] .
[ strattubular3 ]
i would like to thank v srinivas for asking me the question addressed in this paper , and many fruitful discussions .
i also thank jishnu biswas for carefully reading this paper and pointing out errors , and for many fruitful conversations on subanalytic sets .
finally , i thank the referee for valuable comments and corrections . | we introduce the natural and fairly general notion of a subanalytic bundle ( with a finite dimensional vector space @xmath0 of sections ) on a subanalytic subset @xmath1 of a real analytic manifold @xmath2 , and prove that when @xmath2 is compact , there is a baire subset @xmath3 of sections in @xmath0 whose zero - loci in @xmath1 have tubular neighbourhoods , homeomorphic to the restriction of the given bundle to these zero - loci .
= msam5 at 10pt [ theo]*theorem * [ theo]lemma [ theo]proposition [ theo]remark [ theo]definition [ theo]example [ theo]corollary [ theo]_notation _ |
in this paper we study three kinds of functionals for a one - parameter family of infinitely long double helices with straight axes that have radius @xmath2 and _ slope _ @xmath3 , i.e. _ pitch _ @xmath4 :
@xmath5 \g_1(a)=\{(\cos\theta , \sin\theta , a\theta)\ } , \>\g_2(a)=\{(-\cos\theta , -\sin\theta , a\theta)\ } , \end{array}\ ] ] and compute the optimal slopes with respect to them numerically .
the first functional is the `` _ average ropelength _ '' which is the smallest length of a rope with unit thickness which is needed to make one twist of the double helix .
it is a dual notion of _ thickness_@xcite which is the radius of the `` fattest tube '' around a curve , or equivalently , the minimum _ global radius of curvature_@xcite . to be precise
, the thickness of a curve @xmath6 , which we denote by @xmath7 , is given by the supremum of the radii @xmath8 so that a tubular @xmath9-neighbourhood of @xmath6 ( denoted by @xmath10 ) can be embedded , i.e. the supremum of @xmath8 so that the normal discs to @xmath6 of radius @xmath9 at distinct points never intersect @xcite .
the _ ropelength _ is the ratio of the length @xmath11 of @xmath6 and the thickness @xmath7@xcite .
a knot that minimizes the ropelength in its isotopy class is called an _ ideal knot _ after stasiak . in other words ,
an ideal knot is a knot whose fattest tubular neighbourhood maximizes a scale invariant functional on the space of knots defined by @xmath12 in its isotopy class .
the ideal double helix was computed by sylwester przybyl and piotr pieraski@xcite .
the second is the `` _ packing proportion _ '' , in other words , the ratio of the volume of the fattest tubes and that of the circumscribed cylinder of the tubes ( figure [ 0635805_cylinder ] ) .
the best volume packing double helix was computed by kasper olsen and jakob bohr@xcite .
the last is an `` _ average mutual energy _ '' , which is a central topic of the paper . an _ energy of knots _ is a functional on the space of knots which was introduced to produce a representative embedding for each isotopy class as an energy minimizer in the isotopy class .
the first example was obtained as the renormalization of @xmath13-modified potential energy of knots@xcite .
this energy is sometimes called the _ mbius energy _ because it is invariant under mbius transformations of the ambient space , but in fact , there are many other energies which are also invariant under mbius transformations@xcite .
the energy blows up if a knot degenerates to a singular knot with double points .
therefore a knot can be deformed to decrease its energy without having self - crossing , and hence one can expect that the isotopy class would be kept the same during the deformation .
many kinds of energies have been studied intensively for about twenty years , forming a branch of mathematics called `` physical knot theory '' . in this paper
we use two one parameter families of energies for @xmath14 , @xmath15 where @xmath9 denotes the distance @xmath16 between a pair of points .
namely , we consider the integration of modified coulomb s potential and the screened coulomb s potential ( yukawa potential ) . as both diverges because of non - compactness of the domain of integration , we take the `` average per one twist '' , in other words , we consider the integration on @xmath17 , where @xmath18 is a subarc of @xmath19 that winds up just once around the axis . as we only consider the interaction between two strands @xmath19 and @xmath20 , the energies are well - defined without renormalization .
we compute the slopes that minimize the above energies and see how they depend on the parameters @xmath21 and @xmath22 .
we will compare our numerical results with experimental data of dna reported by stasiak and maddocks .
we remark that our strategy of looking for optima double helices with respect to one of the three kinds of functionals above mentioned does not work for single helices .
in fact , our functionals are all optimized when the slope @xmath23 goes to @xmath24 , i.e. when the curve approaches a straight line .
let @xmath6 be a curve .
let @xmath25 be the distance function : @xmath26 .
the _ doubly critical self distance _ of @xmath6@xcite , denoted by @xmath27 , is the smallest positive critical value of @xmath28 , i.e. @xmath29 where @xmath30 denotes the tangent vector to @xmath6 at @xmath31 .
then the thickness of a knot @xmath6 is given by @xcite @xmath32 remark that , for a fixed point @xmath33 , the radius of curvature at @xmath33 is a local quantity whereas @xmath34 is a global quantity .
we remark that , in the case of a _ single _ helix with slope @xmath23 , maritan , micheletti , trovato , and banavar@xcite showed that there is a critical slope @xmath35 such that the thickness is given by the half of the doubly critical self distance if @xmath36 and by the minimum radius of curvature if @xmath37 .
let @xmath38 be the length of @xmath39 per one twist .
it is equal to @xmath40 .
define the _ average ropelength _ of a double helix @xmath39 with slope @xmath23 , denoted by @xmath41 , by @xmath42 where @xmath43 denotes the thickness of @xmath39 .
let us compute the slope @xmath23 of an _ ideal double helix _
, i.e. the slope @xmath23 that attains the minimum average ropelength .
the thickness @xmath43 of @xmath39 is given as follows .
the radius of curvature is constantly equal to @xmath44 . on the other hand ,
the doubly critical self distance is equal to the distance between a point @xmath45 and @xmath20 .
the latter is given by @xmath46 if @xmath47 and by @xmath48 where
@xmath49 @xmath50 is given by @xmath51 if @xmath52 .
therefore , by , the thickness is given by the half of the doubly critical self distance for any @xmath23 : @xmath53 \frac12\sqrt{2 + 2\cos\theta_0+a^2{\theta_0}^2}\>,&\ > ( 0<\theta_0<\pi , \sin\theta_0=a^2\theta_0 ) & \>\>(0<a<1 ) . \end{array } \right.\ ] ] when @xmath47 we have @xmath54 which takes the minimum value @xmath55 at @xmath56 .
the numerical computation using maple implies that the average ropelength takes the minimum which is approximately equal to @xmath57 at @xmath58 ( figure [ av_t_g_c2 ] ) , when the thickness @xmath59 is @xmath60 and the ratio of the pitch and the thickness is @xmath61 .
0.2 cm the above result is not new : this double helix with a straight axis that is optimal with respect to the ropelength ( figure [ 082074_wogrid ] ) had already been reported by pieranski@xcite , fig .
12 ( b ) .
the equation implies that the closest point in @xmath20 from a point in @xmath19 can be found in the opposite side when @xmath47 and in an upward ( or downward ) direction when @xmath52 .
it means that the curve of the contact of two fattest tubes around @xmath19 and @xmath20 is the straight axis when @xmath47 i.e. when the ratio of pitch and thickness satisfies @xmath62 , and a helix between @xmath19 and @xmath20 when @xmath63 i.e. when @xmath64 , as was explained by stasiak and maddocks@xcite .
next we compute the slope of `` best packed '' tubular double helix , namely , a double helix that maximizes the ratio of the volume of the fattest tubular neighbourhood and the volume of the circumscribed cylinder of the tubes ( figure [ 0635805_cylinder ] ) .
this `` packing proportion function '' is given by @xmath65 where @xmath43 is the thickness of @xmath39 .
when @xmath47 we have @xmath66 which takes the maximum value at @xmath67 at @xmath56 . the numerical computation using maple implies that the packing proportion takes the maximum which is approximately equal to @xmath68 at @xmath69 ( figure [ pr ] ) , when the thickness @xmath59 is @xmath70 and the ratio of the pitch and the thickness is @xmath71 . ] 0.4 cm ] 0.2 cm the best volume packing double helix had already been studied by kasper olsen and jakob bohr@xcite .
define the _ @xmath0-modified coulomb s potential _ of @xmath20 at a point @xmath33 in @xmath19 by @xmath72
we assume @xmath73 so that @xmath74 is finite . as @xmath21 increases @xmath74
is dominated by the contribution of near - by points .
remark that it does not depend on the point @xmath75 because of the symmetry .
define the _ average cross - term @xmath0-modified coulomb s potential energy _ of @xmath39 by @xmath76
the asymptotic behavior of @xmath77 as @xmath23 goes to @xmath78 or @xmath79 is given as follows : @xmath80 as @xmath81 and @xmath82 as @xmath83 , where @xmath84 and @xmath85 .
numerical experiments using maple imply that , for each @xmath73 , @xmath77 is a convex function of @xmath23 . the graphs of @xmath77 when @xmath86 and @xmath87 are illustrated in figures [ m2 - 1 ] and [ m1000 - 1 ] . ] 0.4 cm ] [ cols="^,^,^ " , ] + table ii .
@xmath88-minimizing slopes @xmath89 .
the thickness is always @xmath2 as @xmath47 .
0.1 cm ( diamond ) and a fitting curve @xmath90 ( red ) ] we like to end the section with conjectures implied by numerical experiments .
* conjecture*. let @xmath73 and @xmath91 . 1 . for fixed @xmath21 ,
@xmath77 is a convex function of @xmath23 .
+ for fixed @xmath22 , @xmath92 is a convex function of @xmath23 .
the slope @xmath93 that minimizes @xmath77 is an increasing function of @xmath21 . + the slope @xmath89 that minimizes @xmath88 is an increasing function of @xmath22 .
both @xmath94 and @xmath95 are bounded subsets of @xmath96 , and we have @xmath97 and @xmath98 .
we have studied three kinds of functionals for double helices with straight axes and compute the optimal slopes with respect to them numerically ( i.e. without mathematical proofs ) . if we consider the average ropelength , or equivalently , the ratio of the volume of the fattest tubular neighbourhood and the cube of the length , the maximum is attained by an _ ideal _
double helix , whose slope is @xmath99 ( figure [ graph ] , * _ c _ * ) . if we consider the ratio of the volume of the fattest tubular neighbourhood and that of a circumscribed cylinder of the tubes , the maximum is attained by a best volume packing double helix , whose slope is @xmath100 ( figure [ graph ] , * _ b _ * ) .
if we consider two kinds of one parameter families of interaction energies between two strands of types @xmath0 ( modified coulomb potential ) and @xmath1 ( screened coulomb s potential or yukawa potential ) , the optimal slopes seem to be between @xmath2 and @xmath101 in both cases , approaching @xmath2 ( or @xmath102 ) ( figure [ graph ] , * _ e , f _ * ) as the energy is dominated by the global ( or respectively local ) contribution , i.e. the contribution of near - by points gets weaker ( or respectively stronger ) .
stasiak and maddocks@xcite reported that , in the case of dna , the ratio @xmath103 is @xmath104 ( figure [ graph ] , * _ d _ * ) , which is within 4@xmath105 error of that of optimal double helix if we use `` @xmath106-potential energy '' , i.e. the limit of either interaction energy as the parameter approaches to the critical value when the energy fails to be well - defined , when the contribution of near - by points gets weakest . and
the ratio of the pitch and the thickness @xmath103 , and the optimal ones with respect to various functionals : ( * _ b _ * ) volume packing ratio ( * _ c _ * ) ropelength ( ideal double helix ) ( * _ d _ * ) dna ( * _ e _ * ) the limit of the interaction energy @xmath107 as @xmath108 or @xmath88 as @xmath109 ( conjecture ) ( * _ f _ * ) the limit of the interaction energy @xmath107 as @xmath110 or @xmath88 as @xmath111 ( conjecture ) .
any point between @xmath112 and @xmath113 can be realized as an optimal slope with respect to some interaction energy . ]
the author would like to thank the organizers and participants of the conference `` statistical physics and topology of polymers with ramifications to structure and function of dna and proteins '' for stimulating conversations .
in fact , he started to compute the energy @xmath88 after a suggestion by a participant after the talk .
the author is sorry not to have recognized the name .
the author would also like to thank andrzej stasiak and the referee for helpful suggestions and the references , and kasper olsen and jakob bohr for new information and references . | we study optimal double helices with straight axes ( or the fattest tubes around them ) computationally using three kinds of functionals ; ideal ones using ropelength , best volume packing ones , and energy minimizers using two one - parameter families of interaction energies between two strands of types @xmath0 and @xmath1 .
we compare the numerical results with experimental data of dna .
2010 _ mathematics subject classification .
_ 53c65 , 53a04 , 51p05 , 31c99 , 57m25 |
nowadays various estimates of lyapunov dimension of attractors of generalized lorenz systems @xcite @xmath0 are actively developed . in the work @xcite the following result on the estimate of lyapunov dimension of the generalized lorenz system
is formulated .
[ errortheorem ] let @xmath1 be invariant compact set of the generalized lorenz system .
suppose that @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 . then if it is satisfied one of the following conditions : 1 .
@xmath7 , @xmath8 2 .
@xmath9 , @xmath10 3 .
d = 0 ; and also one of the conditions : 1 .
@xmath11 , @xmath12 2 .
@xmath13 , @xmath14 then @xmath15 is valid .
however in the proof of this result in @xcite there is an inaccuracy , the consideration of which the present work is devoted .
introduce a notion of lyapunov dimension .
consider , for this purpose , the linearization of dynamical system along a solution @xmath16 @xmath17 where @xmath18 is the @xmath19 jacobian matrix evaluated along the trajectory @xmath20 of system .
the fundamental matrix @xmath21 of linearized system is defined by the variational equation @xmath22 let @xmath23 be singular values of a fundamental matrix @xmath24 ( the square roots of the eigenvalues of the matrix @xmath25 are ordered for each @xmath26 ) .
[ def : le ] the _ lyapunov exponents ( les)_[multiblock footnote omitted ] in the point @xmath27 are the numbers ( or symbols @xmath28 ) @xmath29 following @xcite , introduce a notion of lyapunov dimension . a local lyapunov dimension of a point @xmath27 in the phase space of dynamical system , generated by equation , is a number @xmath30 where @xmath31 are lyapunov exponents ; @xmath32 $ ] is the smallest natural number @xmath33 such that @xmath34 the lyapunov dimension of invariant set @xmath1 of dynamical system is defined by @xmath35 note that les and lyapunov dimension are invariant under the linear change of variable ( see , e.g. , @xcite ) .
alongside with widely used numerical methods for estimating and computing the lyapunov dimension there exists analytical approach suggested by g.leonov @xcite that is based on the direct lyapunov method .
let us outline the main theorems describing this approach .
let @xmath36 be the eigenvalues of the following matrix @xmath37 [ theorem : th1 ] given an integer @xmath38 $ ] and @xmath39 $ ] , suppose that there is a continuously differentiable scalar function @xmath40 and a nonsingular matrix @xmath41 such that @xmath42 then @xmath43 . here
@xmath44 is derivative of the function @xmath45 with respect to the system .
[ theorem : th2 ] assume that there is a continuously differentiable scalar function @xmath45 and a nonsingular matrix @xmath41 such that @xmath46 then any solution of system , bounded on @xmath47 , tends to an equilibrium as @xmath48 .
thus , if condition holds , then the global attractor of system coincides with its stationary set .
for proving theorem [ errortheorem ] in @xcite it is used above described approach with @xmath49 where @xmath50 and the function @xmath51 where @xmath52 the parameters @xmath53 are chosen in dependence of system parameters in such a way that the conditions of theorem [ theorem : th1 ] or [ theorem : th2 ] are satisfied .
in proving theorem [ errortheorem ] in the work @xcite there is committed an inaccuracy , which does not permit one to obtain a condition for estimating the attractor dimension in the case @xmath54 .
namely , in the proof there is an inaccurate assertion that in the case @xmath55 the inequality @xmath56 yields the inequality @xmath57 where @xmath58 .
let us show that this assertion may be not true .
consider specific values of system parameters and running parameters for which the assertion is not satisfied .
take @xmath59 , @xmath60 , @xmath61 , @xmath62 for which @xmath63 , and @xmath64 , @xmath65 .
then one has @xmath66 .
condition is satisfied since @xmath67 .
however @xmath68 and consequently for @xmath69 the relation @xmath70 is impossible , i.e. condition under such parameters is not satisfied . correcting the above - mentioned inaccuracy ,
we obtain the following result . [ our ] denote by @xmath1 the bounded invariant set of system , involving the point @xmath71 .
suppose that @xmath72 , @xmath73 , @xmath74 , and there exists @xmath75 such that the inequality @xmath76 is satisfied .
let @xmath77 @xmath78 where @xmath79 is defined as in . in this case 1 .
if @xmath80 then any bounded on @xmath81 solution of system tends to a certain equilibrium as @xmath82 ; 2 .
if @xmath83 then @xmath84 for proving theorem [ our ] we uses the matrix @xmath41 , defined according to , and the function @xmath85^{\frac{1}{2}}},\ ] ] where @xmath86 exact value of dimension is obtained here from the comparison of dimension estimate and the value of lyapunov dimension in zero point .
the obtained value corresponds to the dimension of global attractor or a ( local ) b - attractor , involving zero equilibrium .
in 2012 g.a . leonov suggested to consider for the chen and lu systems the following change of variables @xcite @xmath87 with @xmath88 . under this transformation for @xmath89 the chen and lu systems are transformed to the generalized lorenz system with the parameters @xmath90 note that for @xmath91 the chen and the lu systems become linear and their dynamics have minor interest .
thus , without loss of generality , one can assume that @xmath92 .
the transformation with @xmath88 does not change the direction of time for the positive chaotic parameters considered in the works @xcite .
remark that the applying of transformations with the time inversion ( see corresponding discussion of the lorenz , chen and lu systems @xcite ) is not suitable for the study of lyapunov dimension of invariant sets since the absolute values of lyapunov exponents and corresponding local lyapunov dimension in direct and backward time may be different ( see , e.g. , @xcite ) .
below there are represented domains in the obtained two - dimensional parameter space for which the condition of theorem [ our ] is satisfied . on a plane there
are also shown the points , corresponding to standard parameters of chaotic attractors .
for the lu system such parameters are @xmath93 , and after the change @xmath94 ; for the chen system @xmath95 and @xmath96 . 2 as shown in figs [ figleft]-[figright ] , nor the lu system nor the chen system standard parameters are situated in the obtained domain .
while lyapunov dimension of the classical lorenz attractor has been obtained analytically , for the chen and lu attractors it is still an open problem . also by numerical simulation
we do not find any chaotic attractor of the chen and lu systems with the parameters from the obtained domains . applying the same approach ,
we extend the results , obtained in the case @xmath97 in theorem [ errortheorem ] . [ yangtheorem ] suppose that @xmath98 , and the inequality @xmath99 is satisfied .
let @xmath100 be valid , where @xmath101 is the greater root of the equation @xmath102 where @xmath79 is given according to for @xmath103 . in this case 1 .
if @xmath104 then any bounded on @xmath81 solution of system tends to a certain equilibrium as @xmath82 ; 2 . if @xmath105 then @xmath106 consider , for example , the parameters @xmath107 for which there exists a chaotic attractor . for them
@xmath108 , i.e. @xmath11 is satisfied .
however @xmath109 .
thus , for the above - mentioned parameters theorem [ errortheorem ] can not be applied .
but for them theorem [ yangtheorem ] can be applied since for them a higher root of equation is equal to @xmath110 . ]
similarly it is possible to consider the generalized lorenz system not only with positive @xmath111 , but with the negative ones . [
main ] suppose that @xmath1 is a bounded invariant set of system @xmath112 involving the point @xmath71 .
suppose that @xmath113 , the inequality @xmath114 is valid , and there exist numbers @xmath115 such that @xmath116 @xmath117 @xmath118 @xmath119 where @xmath120 . in this case 1 .
if @xmath121 then any bounded on @xmath81 solution of system tends to a certain equilibrium as @xmath82 ; 2 . if @xmath122 then @xmath123
10 url # 1`#1`urlprefixhref # 1#2#2 # 1#1 g. a. leonov , n. v. kuznetsov , on differences and similarities in the analysis of lorenz , chen , and lu systems , applied mathematics and computation 256 ( 2015 ) 334343 .
http://dx.doi.org/10.1016/j.amc.2014.12.132 [ ] .
y. chen , q. yang , the nonequivalence and dimension formula for attractors of lorenz - type systems , international journal of bifurcation and chaos 23 ( 12 ) ( 2013 ) 1350200 . http://dx.doi.org/10.1142/s0218127413502003 [ ] .
n. v. kuznetsov , t. a. alexeeva , g. a. leonov , invariance of lyapunov characteristic exponents , lyapunov exponents , and lyapunov dimension for regular and non - regular linearizations , arxiv e - printshttp://arxiv.org / abs/1410.2016 [ ] . | nowadays various estimates of lyapunov dimension of lorenz - like systems attractors @xcite are actively developed . within the frame of this study
the question arises whether it is possible to obtain the corresponding estimates of dimension for the chen and lu systems using the reduction of them to the generalized lorenz system @xcite . in the work
@xcite leonov s method was applied for the estimation of lyapunov dimension , and as a consequence the lyapunov dimension of attractors of the chen and lu systems with the classical parameters was estimated . in the present work
an inaccuracy in @xcite is corrected and it is shown that the revised domain of parameters , where the estimate of lyapunov dimension is valid , does not involve the classical parameters of the chen and lu systems . lorenz - like system , lorenz system , chen system , lu system , kaplan - yorke dimension , lyapunov dimension , lyapunov exponents . |
it is important to understand the magnetic structure of two - dimensional antiferromagnetic in the light - doping region in the study of cuprate high - temperature superconductors@xcite .
the physics of underdoped region , which is expected to be closely related to the pseudogap state , is still not clear .
the influence of doping holes on the antiferromagnetic state in the parent materials of cuprate superconductors is one of the most interesting problems in strongly correlated electron systems .
it has been experimentally found that the stripe order is stabilized in the underdoped region of la@xmath0nd@xmath1sr@xmath2cuo@xmath3,@xcite la@xmath4sr@xmath2cuo@xmath3 ( lsco)@xcite , and la@xmath4ba@xmath2cuo@xmath3@xcite .
a checkerboard - like charge density modulation with a roughly @xmath5 period ( @xmath6 is a lattice spacing ) has also been observed by scanning tunneling microscopy experiments in bi@xmath7sr@xmath7cacu@xmath7o@xmath8 ( bi-2212)@xcite , bi@xmath7sr@xmath4la@xmath2cuo@xmath9@xcite , and ca@xmath4na@xmath2cuo@xmath7cl@xmath10 ( na - ccoc)@xcite .
it has been pointed out that these types of modulated structures can be understood within the framework of correlated electrons by using the variational monte carlo method for the two - dimensional hubbard model@xcite .
it is expected that there appears a new state in an extremely light - doping region where the charge modulated structures such as stripes and checkerboards are instable and only spin - modulations , for example , a spin - vortex state , will be formulated .
the purpose of this paper is to investigate spin - vortex excitation on the basis of the two - dimensional quantum antiferromagnet . in this paper
, we use the mapping of a two - dimensional magnet onto a nonlinear sigma model .
the nonlinear sigma model is a nonlinear field theory and its renormalization properties were applied to ferromagnets in two dimensions@xcite .
the o(3 ) nonlinear sigma model can be interpreted as the continuum limit of an isotropic ferromagnet , and the action is @xmath11 where the three - component scalar field @xmath12 is under the condition @xmath13 .
the two - dimensional quantum heisenberg antiferromagnet is mapped to a ( 2 + 1 ) dimensional nonlinear sigma model .
the ( 2 + 1)d nonlinear sigma model is@xcite @xmath14,\ ] ] where @xmath15 is a three - component scalar field on the sphere @xmath16 . the parameters @xmath17 and @xmath18 are @xmath19 and @xmath20 where where @xmath21 is the coupling constant of the antiferromagnetic nearest - neighbor interaction , @xmath22 is the magnitude of the spin and @xmath6 is the lattice spacing . in this paper
we consider the two - dimensional case @xmath23 .
we change the scale of @xmath24 by @xmath25 , and then the action is @xmath26.\ ] ] here , we defined in two - space dimensions @xmath23@xcite , @xmath27 for @xmath28 . to be
more precisely , @xmath22 should be replaced by @xmath29 .
although @xmath30 has dimension in this expression , @xmath30 can be dimensionless by scaling the field variables @xmath31 and @xmath24 .
this model describes the nambu - goldstone mode , that is , the spin - wave mode of the quantum antiferromagnetic model .
the renormalization of the spin - wave mode was investigated using the wilson - kogut renormalization method@xcite . in the case of hole doping ,
when the doping rate is small , the nonlinear sigma model is expected to be still relevant if we use the coupling constant @xmath30 renormalized by the doping effect .
it is plausible that we can use the nonlinear sigma model for the extremely light - doping case .
the 2d nonlinear sigma model has an instanton solution @xcite ( being interpreted as skyrmion or meron@xcite ) which describes an excitation of spin - vortex type .
quantum fluctuations of instantons were computed and found to be reduced to the study of the coulomb gas@xcite . according to this study ,
the gas of instantons of the 2d nonlinear sigma model is in the plasma phase .
this means that spin - vortex excitations are independent each other and never form dimers .
this is the case for the two - dimensional classical heisenberg model .
then , we necessarily have a question such as : is this picture still correct for the ( 2 + 1)d nonlinear sigma model ? to investigate this , we generalize the instanton gas approximation to the ( 2 + 1)d nonlinear sigma model .
we use the parametrization of the field @xmath15 in terms of a single complex field @xmath32 : @xmath33 where @xmath34 is the complex conjugate of @xmath32 .
then , using the complex variable @xmath35 for @xmath36 , the action is written as @xmath37 here , @xmath38 is @xmath39 a solution that satisfies @xmath40 is called instanton solution .
a typical instanton solution is , for an integer @xmath41 , @xmath42 @xmath43 and @xmath44 ( @xmath45 ) indicate positions of vortex - like excitations in the antiferromagnetic spin order .
this each local structure can be interpreted as a skyrmion . in general ,
@xmath43 and @xmath44 have @xmath24-dependence : @xmath46 , @xmath47 ( @xmath48 . when @xmath43 and @xmath44 are constants , @xmath38 is the integer quantum number equal to @xmath41@xcite .
we assume that @xmath49 and @xmath50 are continuous function of @xmath24 , then @xmath51 is also a continuous function of @xmath24 and @xmath52 since @xmath38 is an integer . for this instanton solution ,
the action is given as @xmath53
fluctuation effect to the partition function was evaluated for the 2d nonlinear sigma model as@xcite @xmath54 where @xmath55 here @xmath56 is the overall constant factor of @xmath32 .
this is the model of classical coulomb gas interacting with the logarithmic interaction .
this contribution comes from the measure of the functional integral by changing the variable @xmath32 to @xmath43 , @xmath44 and @xmath56 .
this is straightforwardly generalized to the present model ; the measure is given as @xmath57 this formula will be used in the following .
let us consider one - instanton state @xmath58 .
if we set @xmath59 , for simplicity , then we obtain @xmath60 at @xmath61 , @xmath62 at @xmath63 and @xmath64 at @xmath65 . in the limit @xmath66 , @xmath67 .
this indicates a pair of spin vortices that disturbs the antiferromagnetic spin order being located at @xmath68 . there is a spin disorder in the region that includes @xmath68 . for @xmath69 ,
the action in eq.(9 ) is given as @xmath70 where @xmath71 is a constant that depends on @xmath72 , and @xmath73 and @xmath74 are complex conjugates of @xmath6 and @xmath75 .
we replace @xmath30 by @xmath76 so that @xmath30 is dimensionless and add the fluctuation effect to this action to obtain @xmath77.\ ] ] this is the model of single particle in the logarithmic potential if we set @xmath78 .
we define @xmath79 , then the energy is determined by the angular momentum @xmath80 and the potential @xmath81 , which is shown in fig.1 .
the particle dynamics corresponds to the oscillation of the size of spin - vortex pair ( fig.2 ) .
we generalize the one - pair skyrmion to many - skyrmion gas .
the action is written as @xmath82,\end{aligned}\ ] ] where @xmath83 are constants and @xmath84 indicate both @xmath43 and @xmath44 : @xmath85 .
@xmath83 are given by integrals in terms of @xmath32 and show logarithmic divergence @xmath86 for the cutoff @xmath87 of the order of the system size .
when we take into account only the divergent term @xmath86 , @xmath88 are constants . in this case , the kinetic term is simply @xmath89 this means that the kinetic term is only concerned with the movement of centers of @xmath6-particles and @xmath75-particles .
this term is expected to be unimportant and we neglect it .
hence , the model is reduced to the coulomb gas model without the kinetic term
. then we use the static approximation as a first step . under this approximation ,
the action is @xmath90.\ ] ] this yields that the transition from the plasma phase to the molecular phase occurs at @xmath91@xcite : @xmath92 therefore , there is a kosterlitz - thouless transition in the ( 2 + 1)d nonlinear sigma model .
we have formulated skyrmion ( instanton ) gas picture for the two - dimensional quantum antiferromagnet on the basis of the ( 2 + 1)d nonlinear sigma model .
we have shown that there is a kosterlitz - thouless transition from the plasma phase to the molecular phase as the temperature is lowered by using a static approximation to the instanton gas . in the molecular phase two instantons form a bound state , namely , a dipole of two spin vortices .
we expect that this is related to the pairing mechanism of superconductivity in doped antiferromagnets .
9 k. h. bennemann and j. b. ketterson eds . : _ the physics of superconductors _ ( springer - verlag , berlin , 2004 ) . t. suzuki , t. goto , k. chiba , t. shinoda , t. fukase , h. kimura , k. yamada , m. ohashi , and y. yamaguchi : phys .
rev . b*57 * ( 1998 ) 3229 .
m. fujita , h. goka , t. adachi , y. koike , and k. yamada : physica c*426 * -*431 * ( 2005 ) 257 .
j. m. tranquada , j. d. axe , n. ichikawa , y. nakamura , s. uchida , and b. nachumi : phys .
rev . b*54 * ( 1996 ) 7489 .
j. e. hoffman , e. w. hudson , k. m. lang , v. madhavan , h. eisaki , s. uchida , and j. c. davis : science * 295 * ( 2002 ) 466 .
w. d. wise , m. c. boyer , k. chatterjee , t. kondo , t. takeuchi , h. ikuta , y. wang , and e. w. hudson : nature phys .
* 4 * ( 2008 ) 696 .
t. hanaguri , c. lupien , y. kohsaka , d. h. lee , m. azuma , m. takano , h. takagi , and j. c. davis : nature * 430 * ( 2004 ) 1001 .
t. yanagisawa , s. koike , m. miyazaki , and k. yamaji : j. phys .
matter * 14 * ( 2002 ) 21 .
m. miyazaki , t. yanagisawa , and k. yamaji : j. phys .
* 73 * ( 2004 ) 1643 ; t. yanagisawa , m. miyazaki and k. yamaji : j. phys .
* 78 * ( 2009 ) 013706 .
m. miyazaki , k. yamaji , t. yanagisawa , and r. kadono : j. phys .
* 78 * ( 2009 ) 043706 .
a. m. polyakov : phys .
lett . b*59 * ( 1975 ) 79 .
e. brezin and j. zinn - justin : phys .
rev . b*14 * ( 1976 ) 3110 .
w. a. bardeen , b. w. lee , and e. r. shrock : phys .
d*14 * ( 1976 ) 985 .
s. chakravarty , b. i. halperin and d. r. nelson : phys .
rev . b*39 * ( 1989 ) 2344 .
t. yanagisawa : phys .
rev . b*46 * ( 1992 ) 13896 .
a. a. belavin and a. m. polyakov : jetp lett .
* 22 * ( 1975 ) 245 .
a. jevicki : nucl .
phys . b*127 * ( 1977 ) 125 .
a. dadda , m. lscher , and p. di vecchia : nucl .
phys . b*146 * ( 1978 ) 63 .
a. m. polyakov : _ gauge fields and strings _ ( harwood academic publisher , switerland , 1987 ) . h. koizumi : cond - mat/0506747 .
t. morinari : phys .
rev . b*73 * ( 2006 ) 064504 .
i. v. frolov and a. s. schwarz : jetp lett .
* 28 * ( 1978 ) 249 .
v. a. fateev , i. v. frolov and a. s. schwarz : nucl .
phys . b*154 * ( 1979 ) 1 .
b.berg and m. lscher : commun .
* 69 * ( 1979 ) 57 . | we investigate spin - vortex excitations in the two - dimensional antiferromagnet on the basis of the nonlinear sigma model .
the model of two - dimensional heisenberg quantum antiferromagnet is mapped onto the ( 2 + 1)d nonlinear sigma model .
the 2d nonlinear sigma model has an instanton ( or skyrmion ) solution which describes an excitation of spin - vortex type .
quantum fluctuations of instantons are reduced to the study of the coulomb gas , and the gas of instantons of the 2d nonlinear sigma model is in the plasma phase .
we generalize this picture of instanton gas to the ( 2 + 1)d nonlinear sigma model .
we show , using some approximation , that there is a kosterlitz - thouless transition from the plasma phase to the molecular phase as the temperature is lowered . |
physics of quark - gluon plasma is one of the most important problems in high energy physics .
quark - gluon plasma state is expected to be produced in extremely high temperature , which will be realized in ultra - relativistic heavy ion collisions . to obtain the higher temperature ,
experimental setting has been growing in three directions : the higher incident energy par nucleon , the heavier incident nucleus , and the heavier target nucleus . in any case , colliding energy is getting larger , but the larger energy does not directly mean the higher temperature .
it is not obvious , whether large fraction of colliding energy will be transferred into thermal energy ( temperature ) or kinetic energy ( systematic flow ) .
especially , the heavier target does not correspond to the larger colliding velocity in the center of mass system .
hence , if we stand on the landau - type simple stopping " picture , achieved energy density must become lower for larger target mass . in this paper
, we will analyze the recent experimental data for various target nuclei and discuss systematic change of initial parameters in hydrodynamical model . in a previous paper@xcite we formulated a semi - phenomenological quantum transport theory for a quark - gluon plasma fluid based on an operator valued langevin equation .
putting a mode spectrum and a damping into this formula , we can easily evaluate thermodynamical quantities and transport coefficients . introducing a simple model spectrum to this formula
, we have already discussed the space - time evolution of ( 1 + 1)-dimensional viscous quark - gluon plasma fluid @xcite and ( 3 + 1)-dimensional perfect fluid quark - gluon plasma with phase - transition @xcite .
we have also discussed the case of baryon - rich quark - gluon plasma with phase - transition @xcite , in which one of the conclusions was that the baryon - number effect was not so much large .
hence , although the existing experiments do not seem to be baryon - free , we apply a simple baryon - free quark - gluon plasma model to analyses of available experiments .
following @xcite , we here introduce a simple model mode spectrum , @xmath1 where @xmath2 is a parameter that adjusts the degrees of modes and @xmath3 is the critical temperature responsible for the reliant phase transition .
supposing that the high temperature phase is dominated by massless u- , d- , s - quarks and gluons and the hadronic phase is dominated by pions and kaons , we can put @xmath4 , @xmath5 mev , @xmath6 mev , and @xmath7 mev . with these parameters
, we can easily obtain an equation of state with phase transition - like behavior ( fig.1 ) , which seems to reproduce the lattice qcd result @xcite .
the hydrodynamical equation is given by , @xmath8 for perfect fluid . here
@xmath9,@xmath10 and @xmath11 are , respectively , energy density , pressure and local four velocity . for cylindrically symmetric expansion along the collision axis ,
it is convenient to introduce new variables , @xmath12 , @xmath13 , @xmath14 and @xmath15 defined by @xmath16 instead of ordinary coordinate @xmath17 , @xmath18 , @xmath19 and @xmath20 . as for the local four velocity
, we represent it by four components , @xmath21 , @xmath22 , @xmath23 and @xmath24 associated with the new variables , @xmath25 because of @xmath26 and the cylindrical symmetry , the four components can be reduced to two variables @xmath27 and @xmath28@xcite , @xmath29 taking account of the formation time , let us put the temperature distribution on @xmath13 and @xmath14 at @xmath30 , @xmath31 assuming that the hydrodynamical expansion starts at @xmath32 fm later than the collision instance at @xmath33 fm .
@xmath34 represents the initial temperature , @xmath35 and @xmath36 are measures of the longitudinal and transverse spreads in @xmath13 and @xmath14 , respectively .
@xmath34 , @xmath35 , @xmath36,@xmath37 and @xmath38 are input parameters to characterize our model on which we should impose a constraint given by a fixed value of the initial fluid energy @xmath39 ( @xmath40 standing for the total collision energy and @xmath41 for the inelasticity ) . as for the initial condition of the local velocity
, we use bjorken s scaling solution , @xmath42 , in the longitudinal direction , by which we can take the initial longitudinal flow into account .
initial values of these parameters should be determined on the basis of an appropriate physical discussion , but details of these parameters are deeply connected to the interaction mechanism and the thermalization process . in this paper , standing on _ phenomenological _ point of view , we try to choose these parameters so as to reproduce experimental results of both momentum spectrums .
the numerical solution of the hydrodynamical equation gives us the momentum distribution of hadrons , coming out from a local system with volume @xmath43 in local equilibrium with freeze - out temperature @xmath44 , through the formula @xmath45 where @xmath46 with hadron mass @xmath47 ( see ref.@xcite ) . integrating eq.([5.3 ] ) on the hypersurface with @xmath44
, we obtain @xmath48 for the y - distribution , @xmath49 for @xmath50-distribution , where @xmath51 stands for the pseudorapidity distribution , and @xmath52 for the @xmath53-distribution . our numerical results given by eq.([5.5 ] )
is shown in fig.2 .
as is well known , the freeze - out temperature @xmath54 is directly connected with the @xmath53 slope and we should determine it from the experimental @xmath53-spectrum .
however , throughout this paper , we fix it equal to @xmath55mev because of the lack of sufficient experimental data .
our numerical solution of hydrodynamical model is so designed as to describe the head - on collision of the same kind of nuclei in the center of mass system . in order to apply the solution to the asymmetric collision , such as wa80 and na35
, we estimated effective participants in large target nucleus as , @xmath56,\ ] ] @xmath57 and @xmath58 being target mass and projectile mass , respectively .
our hydrodynamical model contains many parameters to be adjusted .
we have already fixed freeze - out temperature as @xmath55mev from fig.2 in the previous section . we may put formation time equal to @xmath59 fm and initial transverse size equal to the size of smaller nucleus , @xmath60{a_{proj } } - \sigma_{r } \mbox{\quad ( fm)},\ ] ] where @xmath61 . as for the local velocity ,
we use bjorken s scaling solution , @xmath42 , neglecting initial transverse flow .
therefore , the residual parameters to be adjusted are only initial temperature @xmath62 at central point , and longitudinal size at initial time , @xmath63 and @xmath37 .
we try to adjust these parameters so as to reproduce existing experiments .
our results for sulphur beam experiments are shown in fig.3 , and the values of parameters for sulphur and for oxygen beam are summarized in table 1 . in any case , our model seems to reproduce the experimental results well with plausible values of the parameters .
table 2 stands for the results given by a equation of state in the limit @xmath64 , which corresponds to a hot hadron - gas model . in this case
, @xmath41 became an unphysical value , larger than unity .
it means that _ we failed to reproduce experimental results with the hot hadron - gas model_. these results tell us that larger nuclei are more effective to make a high temperature fluid than smaller nuclei can do even in target experiment . for larger targets ,
the initial temperature must become higher , the longitudinal size smaller , and the inelasticity larger .
this kind of tendency is just desired for the experimental quark - gluon plasma production .
sometimes , it has been pointed out that the present cern energy is not high enough to make the bjorken s scaling quark - gluon plasma , and the stopping picture of landau type is then realized . however , the simple stopping picture is not in agreement with our results . in the simple stopping picture ,
the only collision parameter is the colliding relative velocity @xmath65 , and the longitudinal size contracts with @xmath66 and the energy density increases with @xmath67 . in target experiments ,
the larger target mass means the higher collision energy , but does not means the higher collision velocity .
table 3 shows the our results and relative velocities .
the tendency is opposite to the simple landau picture .
of course , it only means the failure of the _ simple _ stopping picture and we may describe the phenomena which require to introduce additional phenomenological parameters , such as energy dependent inelasticity @xcite . in this paper , we have fitted hydrodynamical parameters in a purely phenomenological manner without resorting to any dynamical model for initial collision process .
although we need a dynamical model for the quark - gluon plasma formation to understand the results in the table 1 , we may extrapolate our phenomenological results in order to obtain the rough sketch of coming experiments .
figure 4 indicates us that parameters in pb + pb 150 gev / a collision are @xmath68mev and @xmath69 .
the possible psuedorapidity - distribution of hadrons is shown in fig .
5 and the total charged hadronic multiplicity ( pions and kaons ) reaches 2095 .
we have applied our ( 3 + 1)-dimensional hydrodynamical model with phase - transition to the recent heavy - ion experiments .
our results show that the inelasticity increases with target mass , the longitudinal size is getting smaller with target mass , and the initial temperature becomes higher .
such a tendency is very promising for our aim of producing quark - gluon plasma experimentally and also for understanding the future experimental results at cern and rhic .
99 m. mizutani , s. muroya and m. namiki , phys .
* d37 * ( 1988 ) , 3033 . y. akase , s. dat@xmath70 , m. mizutani , s. muroya m. namiki and m. yasuda , prog .
* 82 * ( 1989 ) , 591 .
y. akase , m. mizutani , s. muroya and m. yasuda , prog.theor .
phys . * 85 * ( 1991 ) , 305 .
1990 - 16 .
t. ishii and s. muroya , phys .
rev . * d46 * ( 1992 ) , 5156 .
r. albrecht _ et al . _
: wa80 , z. phys .
* c 55 * ( 1992 ) , 539 .
a. bamberger _
: na35 , nucl . phys . * a498 * ( 1989 ) , 133c
. t. kkeson _ et al .
_ : helios , nucl . phys .
* b342 * ( 1990 ) , 279 .
r. albrecht _ et al . _
: wa80 , phys .
* b202 * ( 1988 ) , 596 u. ornic , f. pottag and r. m. weiner phys .
( 1989),2641 .
c|c|c|c|c|c + collision & @xmath71 & @xmath72 & @xmath73 & @xmath62(mev ) & @xmath74 + s + al @xmath75 & 27@xmath76 & 1.5 & 1.0 & 169 & 0.52 + s + s @xmath77 & 32 & 1.5 & 1.0 & 167 & 0.46 + s + cu @xmath75 & 64 & 1.5 & 0.8 & 180 & 0.69 + s + ag @xmath75 & 107 & 1.5 & 0.7 & 188 & 0.75 + s + w @xmath78 & 184 & 1.2 & 0.5 & 202 & 0.68 + s + au @xmath75 & 197 & 1.5 & 0.7 & 195 & 0.88 + o + cu @xmath79 & 64 & 167 & 3.0 & 0.3 & 0.38 + o + ag @xmath79 & 107 & 172 & 3.0 & 0.2 & 0.54 + o + au @xmath79 & 197 & 176 & 3.0 & 0.1 & 0.62 + + c|c|c|c|c|c + target & @xmath71 & @xmath72 & @xmath73 & @xmath62(mev ) & @xmath74 + al@xmath75 & 27@xmath76 & 0.8 & 0.5 & 335 & 0.82 + s@xmath77 & 32 & 0.6 & 1.0 & 312 & 0.93 + cu@xmath75 & 64 & 1.0 & 0.5 & 363 & 1.37 + ag@xmath75 & 107 & 1.0 & 0.5 & 380 & 1.49 + w@xmath80 & 184 & 0.8 & 0.7 & 405 & 1.65 + au@xmath75 & 197 & 0.9 & 0.7 & 400 & 1.77 + + c|c|c|c|c|c + collision & effective & @xmath81 & @xmath82 & @xmath83 & @xmath84 + & participants & ( gev ) & ( gev ) & & + s + al & 30.9 + 27 & 6175.72 & 561.218 & 11 & 1.5 + 1.0 + s + cu & 32 + 49.6 & 6400 & 774.461 & 8.3 & 1.5 + 0.8 + s + ag & 32 + 63.0 & 6400 & 873.332 & 7.4 & 1.5 + 0.7 + s + au & 32 + 81.1 & 6400 & 990.954 & 6.5 & 1.5 + 0.7 + o + cu & 16 + 34.0 & 3200 & 453.783 & 7.1 & 3.0 + 0.3 + o + ag & 16 + 41.9 & 3200 & 503.616 & 6.4 & 3.0 + 0.2 + o + au & 16 + 52.7 & 3200 & 565.62 & 5.7 & 3.0 + 0.1 + the energy density as a function of temperature ; the solid line stands for our phase - transition model , the dotted line for the simple hadron - gas of pions and kaons , and the dot - dashed line for the quark - gluon plasma of massless u- , d- , s - quarks and gluons . in the phase - transition model
, phase - transition takes place at @xmath85mev in the width of @xmath86mev .
the @xmath87 spectrum of pions in s+s 200 gev / a collision together with the data of na35 .
the solid line stands for @xmath88 mev , the dashed line for @xmath89mev , the dotted line for @xmath90mev , and dot - dashed line for @xmath91mev , respectively .
the pseudorapidity distributions of charged hadron in 200gev / a sulphur beam collision .
the solid line stands for s+al , the dashed line for s+cu , the dotted line for s+ag , and the dot - dashed line stands for s+au , respectively . in our calculation , we take account of pions and kaons only .
plots are the data from @xcite . in the case of s+al
, @xmath92 is determined from the size of al .
the values of initial parameters as a function of total collision energy .
the solid line stands for initial temperature,@xmath62 , and the dashed line for longitudinal extent , @xmath93 .
the plots from the left to the right correspond to o+cu , o+ag , o+au , s+al , s+cu , s+ag , and s+au , respectively .
plots at the right end are the extrapolation of parameters to pb+pb 150 gev / a collision . | we numerically analyze recent high energy heavy - ion collision experiments based on a hydrodynamical model with phase transition and discuss a systematic change of initial state of qgp - fluid depending on colliding - nuclei s mass . in a previous paper
, we formulated a ( 3 + 1)-dimensional hydrodynamical model for quark - gluon plasma with phase transition and discussed numerically the space - time evolution in detail .
we here compare the numerical solution with the hadronic distributions given by cern wa80 and na35 .
systematic analyses of the experiments with various colliding nuclei enable us to discuss the dependences of the initial parameters of the hydrodynamical model on colliding nuclei s mass .
furthermore , extrapolating the present experiments , we derive the possible hadronic distributions for lead - lead 150gev / a collision .
wu - hep-95 - 1 + twc-95 - 1 + january 1995 numerical analyses of cern 200gev / a heavy - ion collisions based on a hydrodynamical model with phase transition + + + _ @xmath0department of physics , waseda university + okubo 3 - 4 - 1 , shinjuku - ku , tokyo 169 - 50 , japan _ + |
for @xmath7 , the _ grid _
@xmath8 is the ( infinite ) graph with vertex set @xmath9 , and with edges between vertices with ( euclidean ) distance @xmath10 . for @xmath11 , the _ strip _
@xmath0 is the subgraph of @xmath12 induced by the vertex subset @xmath13 , where @xmath14 . for @xmath15 , any subgraph of @xmath0 induced by @xmath16 with @xmath17
is called a _ bar _ with length @xmath18 , or an _
@xmath18-bar_. for @xmath19 and @xmath20 , the subgraphs of @xmath0 induced by @xmath21 and @xmath22 , respectively , are called a _ row _ and a _ column _ of @xmath0 ; similarly we also talk about rows and columns of a bar .
let @xmath23 be a ( finite or infinite ) graph .
for any vertex @xmath24 , the _ open neighborhood _
@xmath25 is the subset of vertices adjacent to @xmath26 in @xmath23 , and the _ close neighborhood _ @xmath27 $ ] is @xmath28 . for @xmath29 and @xmath24 , the _ ball _ of radius @xmath30 centered at @xmath26 , denoted by @xmath31 , is the set of vertices with distance at most @xmath30 from @xmath26 in @xmath23 . in particular , @xmath32 and @xmath33 $ ] .
a vertex subset @xmath34 is an _ identifying code _ in @xmath23 if 1 . for each vertex @xmath24 , @xmath27 \cap c \neq \emptyset$ ] , 2 . for each pair of distinct vertices
@xmath35 , @xmath36 \cap c \neq n[v ] \cap c$ ] .
let @xmath37 be an arbitrary vertex in @xmath23 .
for any @xmath34 , the _ upper density _ and _ lower density _ of @xmath38 in @xmath23 are , respectively , @xmath39 if these two numbers are equal , then their common value is simply called the _ density _ of @xmath38 in @xmath23 , @xmath40 in particular , if @xmath23 is a finite graph , then @xmath41 always exists , and @xmath42 the infimum density of an identifying code in @xmath23 is customarily defined as @xmath43 where @xmath38 ranges over all identifying codes in @xmath23 . here
the upper density @xmath44 is used because the density @xmath45 does not always exist when @xmath23 is an infinite graph .
we show in this paper that if @xmath23 is @xmath0 , for any @xmath11 , then the infimum density can always be achieved by a _ periodic _ identifying code repeating a bar pattern whose length @xmath18 is bounded by a function of @xmath1 , and hence @xmath46 where @xmath38 ranges over all periodic identifying codes @xmath38 , whose densities exist . identifying codes in the square grid @xmath12 and its subgraphs
have been extensively studied .
the concept was introduced by karpovsky , chakrabarty , and levitin @xcite for its application in fault diagnosis of multiprocessor systems .
cohen et al .
@xcite proved an upper bound of @xmath47 by giving two periodic identifying codes achieving this density .
subsequently , ben - haim and litsyn @xcite proved the matching lower bound of @xmath48 .
and @xmath49 , respectively , in @xmath50 and @xmath51.,title="fig : " ] and @xmath49 , respectively , in @xmath50 and @xmath51.,title="fig : " ] daniel , gravier , and moncel @xcite initiated the study of identifying codes in strips , and proved that @xmath52 and @xmath53 .
the infimum density of @xmath54 for @xmath55 is achieved by a periodic identifying code with pattern length @xmath56 , consisting of every other vertex in @xmath55 .
the infimum density of @xmath57 for @xmath50 is achieved by a periodic identifying code with pattern length @xmath58 , as illustrated in figure [ fig : s2s3 ] left .
bouznif , havet , and preissmann @xcite recently proved that @xmath59 by a sophisticated charging argument ; see figure [ fig : s2s3 ] right for the periodic identifying code with pattern length @xmath60 that achieves this density . extending the previous result of @xmath61 for the square grid @xmath12 , bouznif , havet , and preissmann @xcite also proved that , for any @xmath11 , @xmath62 and @xmath6 , respectively , in @xmath3 and @xmath5.,title="fig : " ] and @xmath6 , respectively , in @xmath3 and @xmath5.,title="fig : " ] in this paper , we prove that @xmath63 and @xmath64 , with the assistance of a computer program justified by .
refer to figure [ fig : s4s5 ] : the infimum density of @xmath4 for @xmath3 is achieved by a periodic identifying code with pattern length @xmath65 ; the infimum density of @xmath6 for @xmath5 is achieved by a periodic identifying code with pattern length @xmath66 .
it is interesting to note that @xmath67 ( @xmath68 ) , in contrast to the strict monotonicity of @xmath69 , and to the asymptotic bound of @xmath70 .
consider any subset @xmath38 of vertices in a graph @xmath23 . for any pair of distinct vertices
@xmath71 , if @xmath36 \cap c \ne \emptyset$ ] , @xmath27 \cap c \ne \emptyset$ ] , and @xmath36 \cap n[v ] = \emptyset$ ] , then @xmath36 \cap c \ne n[v ] \cap c$ ] .
thus to verify whether @xmath38 is an identifying code in @xmath23 , it suffices to verify that 1 . for each vertex @xmath24 , @xmath27 \cap c \neq \emptyset$ ] , 2 . for each pair of distinct vertices @xmath35 with @xmath36 \cap n[v ]
\ne \emptyset$ ] , @xmath36 \cap c \neq n[v ] \cap c$ ] .
any two vertices with intersecting closed neighborhoods are within distance @xmath56 from each other . in particular , in the strip @xmath0 , such a pair of vertices must be located in some @xmath72-bar , and the union of their closed neighborhoods is contained in some @xmath73-bar .
let @xmath74 be any bar with length @xmath75 in the strip @xmath0 , and let @xmath76 be the sub - bar with length @xmath77 consisting of the middle columns ( except the first column and the last column ) of @xmath74 .
then the closed neighborhood ( in @xmath0 ) of each vertex in @xmath76 is contained in @xmath74 .
we say that a subset @xmath78 of vertices in @xmath74 is a _
barcode _ of @xmath74 , if @xmath78 satisfies the two conditions of an identifying code _ locally _ for the vertices in @xmath76 , that is , for each vertex @xmath79 , @xmath27 \cap p \ne \emptyset$ ] , and for each pair of distinct vertices @xmath80 , @xmath36 \cap p \ne
n[v ] \cap p$ ] . construct an edge - weighted directed graph @xmath81 as follows .
for each barcode @xmath78 of any @xmath82-bar , let @xmath81 have a corresponding vertex .
for each barcode @xmath83 of any @xmath73-bar , which induces two barcodes @xmath84 and @xmath85 of @xmath82-bars consisting of the first @xmath82 columns and the last @xmath82 columns , respectively , of the @xmath73-bar , let @xmath81 have an edge , directed from the vertex corresponding to @xmath84 , to the vertex corresponding to @xmath85 , then set the weight of this edge to the number of vertices of @xmath83 in the last column of the @xmath73-bar .
then @xmath81 has at most @xmath2 vertices and at most @xmath86 edges , where each vertex is incident to at most @xmath87 incoming edges and at most @xmath87 outgoing edges , and each edge has an integer weight between @xmath88 and @xmath1 .
note that @xmath81 is strongly connected . given any barcode @xmath78 of a @xmath82-bar , we can always extend it to a barcode @xmath83 of a @xmath73-bar , which consists of the vertices of @xmath78 in the first @xmath82 columns and all @xmath1 vertices in the last column .
this implies that @xmath81 contains an edge progression ( a directed path ) of at most @xmath82 edges from any vertex @xmath26 , to the vertex @xmath89 corresponding to the barcode containing all @xmath90 vertices of a @xmath82-bar .
symmetrically , there is also an edge progression of at most @xmath82 edges from @xmath89 to any vertex @xmath26 in @xmath81 .
note also that @xmath81 contains self - loops , for example , from @xmath89 to itself .
indeed if a vertex is connected to itself by an edge , then the barcode corresponding to this vertex must include either all or none of the @xmath82 vertices in each row of the @xmath82-bar .
define the _ mean weight _ of any finite edge progression in an edge - weighted directed graph as the total weight of the edges divided by the number of edges ( allowing duplicates ) .
define the _ minimum cycle mean _ of any edge - weighted directed graph as the minimum mean weight of any cycle in it .
the following proposition is easy to prove : [ prp ] each bar pattern of a periodic identifying code in @xmath0 corresponds to a cycle in @xmath81 , and vice versa .
the minimum density a periodic identifying code in @xmath0 is exactly the minimum cycle mean of @xmath81 divided by @xmath1 .
we next show that the infimum density of an identifying code in @xmath0 can always be achieved by a periodic identifying code with pattern length bounded by a function of @xmath1 : [ thm : periodic ] for any @xmath11 , there is a periodic identifying code @xmath91 with pattern length @xmath92 in @xmath0 satisfying @xmath93 .
the proof follows the line of reasoning of ( * ? ? ?
* hint for exercise 3.2 ) , where a similar question on the ( lattice ) packing density of a convex body in the plane is considered .
the following lemma will be used : [ lem : super ] for any graph @xmath23 , if @xmath34 is an identifying code in @xmath23 , then any subset @xmath94 with @xmath95 is also an identifying code in @xmath23 . since @xmath96
, we have for each vertex @xmath24 , @xmath97 \cap c \ne \emptyset \implies n[v ] \cap d \ne \emptyset,\ ] ] and for each pair of distinct vertices @xmath35 , @xmath98 \cap c \ne n[v ] \cap c \implies n[u ] \cap d \ne n[v ] \cap d.\ ] ] the lemma then follows by the definition of identifying codes .
now we proceed with the proof of theorem [ thm : periodic ] .
fix @xmath11 .
for each @xmath99 , we will construct a periodic identifying code @xmath100 in @xmath0 with pattern length at most @xmath2 such that @xmath101 fix @xmath99 . by the definition of @xmath102 in , there exists a sequence of ( not necessarily periodic ) identifying codes in @xmath0 , whose upper densities tend arbitrarily close to @xmath103
. in particular , we can find @xmath104 such that @xmath105 in the following we will derive @xmath100 from @xmath104 . partition @xmath0 into an infinite sequence of disjoint @xmath106-bars @xmath107 $ ] , induced by disjoint vertex subsets @xmath108 : = \ { in,\ldots,(i+1)n-1 \}\times{\mathbb{z}}_k$ ] , @xmath109 .
let @xmath110 : = a_n \cap v[i]$ ] .
then for all @xmath111 , @xmath112 , r[i ] ) = \frac{|a[i]|}{nk},\ ] ] and hence there exists @xmath113 such that @xmath114 , r[j ] ) \le \overline{d}(a_n , s_k).\ ] ] let @xmath115 and @xmath116 be the vertices in the first @xmath82 columns of @xmath117 $ ] and of @xmath118 $ ] , respectively . by lemma [ lem : super ] , @xmath119 remains an identifying code in @xmath0 .
let @xmath120 \cup q$ ] .
then simply repeating the pattern @xmath78 of @xmath117 $ ] results in an identifying code too .
let @xmath100 denote this periodic identifying code in @xmath0 , with pattern length @xmath106 .
note that @xmath121| + 4k$ ] .
thus @xmath122| + 4k}{nk } = d(a[j ] , r[j ] ) + 4n^{-1 } \le \overline{d}(a_n , s_k ) + 4n^{-1 } \le d^*(s_k ) + 5 n^{-1},\ ] ] as desired in .
recall proposition [ prp ] on the equivalence between patterns of periodic identifying codes in @xmath0 and cycles in @xmath81 .
since @xmath81 has at most @xmath2 vertices , any cycle in @xmath81 with more than @xmath2 edges must repeat some vertex .
then the edges of the cycle can be partitioned to form two shorter cycles , and the mean weights of the two parts can not be both greater than that of the whole .
correspondingly , if the length @xmath106 of the pattern @xmath78 exceeds @xmath2 , then we can replace @xmath78 by a shorter pattern with equal or smaller density .
thus we can assume , without loss of generality , that @xmath100 is a periodic identifying code with pattern length at most @xmath2 .
we have obtained an infinite sequence of periodic identifying codes @xmath100 , for @xmath99 , with pattern length at most @xmath2 .
since @xmath1 is finite , the number of distinct periodic identifying codes with pattern length at most @xmath2 is finite .
so there must exist a periodic identifying code @xmath91 with pattern length at most @xmath2 , such that @xmath123 is identical to @xmath91 for an infinite sequence of increasing positive integers @xmath124 , @xmath125 .
then by we have @xmath126 for @xmath127 , and consequently @xmath128 .
this completes the proof of theorem [ thm : periodic ] .
we briefly review karp s algorithm @xcite for computing the minimum cycle mean @xmath129 of a strongly connected directed graph @xmath23 with @xmath106 vertices and @xmath130 edges , where each edge @xmath131 has a weight @xmath132 .
let @xmath89 be an arbitrary vertex in @xmath23 . for each vertex @xmath24
, and each integer @xmath133 , define @xmath134 as the minimum weight of any edge progression of length @xmath1 from @xmath89 to @xmath26 ; if no such edge progression exists , then @xmath135 .
karp proved ( * ? ? ?
* theorem 1 ) ( see also ( * ? ? ?
* problem 24.5 ) for a step - by - step sketch of the main argument in this elegant proof ) that @xmath136 all values of @xmath134 can be computed in @xmath137 time by dynamic programming , with the recurrence @xmath138 \quad\textrm{for } 1 \le k \le n-1,\ ] ] and the initial conditions @xmath139 finally , by , the value of @xmath129 can be computed in @xmath140 time .
the overall running time is @xmath137 .
if the actual cycle yielding the minimum cycle mean is desired , it may be extracted from a minimum - weight edge progression of length @xmath106 between @xmath89 and the minimizing vertex @xmath26 in .
such edge progressions for all vertices @xmath26 can be recorded , in parallel to the computation of @xmath134 by , by recording back references to the vertices @xmath141 of the minimizing edges @xmath142 .
it is easy to verify that karp s algorithm remains valid for directed graphs with self - loops .
now recall proposition [ prp ] . by applying karp s algorithm to the directed graph @xmath81
corresponding to the strip @xmath0 , we immediately have the following theorem : for any @xmath11 , a minimum - density periodic identifying code in @xmath0 , with pattern length at most @xmath2 , can be found by karp s algorithm in @xmath143 time . a straightforward implementation of karp s algorithm uses @xmath144 space for all values of @xmath134 and the corresponding back references to @xmath141 .
the space for @xmath134 can be easily reduced to @xmath145 by running the dynamic programming algorithm in two passes . in each pass , by , the values of @xmath146 for each @xmath11 depend only on @xmath147 .
thus the space for @xmath134 in each round @xmath1 can be kept for the next round , while the space for @xmath148 can be recycled for @xmath149 . at the end of the first pass ,
the values of @xmath150 are recorded . then in the second pass , with ready access to @xmath150 , we can compute @xmath151 for each @xmath26 on the fly , without keeping the values of @xmath134 for all @xmath1 . finally , after the second pass , we can compute @xmath152 in @xmath145 time .
the overall running time remains @xmath137 .
the @xmath144 space for the back references is harder to reduce .
however , it turns out that in our application , a rather short cycle yielding the minimum cycle mean can be found near the end of the minimum - weight edge progression of length @xmath106 from @xmath89 to @xmath26 .
thus we only need to record back references when @xmath1 gets close to @xmath106 in , and hence can reduce the total space usage to @xmath153 .
refer to appendix [ sec : code ] for the source code of a computer program written in the c programming language .
the subscript @xmath1 of @xmath134 in karp s algorithm and the subscript @xmath1 of @xmath0 have different meanings , and are denoted by variable ` k ` and symbolic constant ` k ` , respectively , in the program .
the values of @xmath146 and @xmath147 are recycled in the two arrays ` d0 ` and ` d1 ` , while the values of @xmath154 are recorded in the array ` dn ` . to avoid numerical inaccuracy of floating point arithmetics , the numerator and denominator of each @xmath155
are recorded separately , in the two ` int ` arrays ` dd ` and ` nk ` .
the back references for each vertex @xmath26 are recorded in the array ` pp ` in ` struct vertex ` .
the program has been tested on a laptop computer with modest computing power : macbook air ( 13-inch , mid 2011 ) ; 1.8 ghz intel core i7 processor ; 4 gb 1333 mhz ddr3 memory . with ` k ` set to @xmath56 and @xmath72 , respectively
, the program runs for less than one second , confirms the known results of @xmath53 and @xmath59 , and rediscovers the minimum - density periodic identifying codes in figure [ fig : s2s3 ] . with ` k ` set to @xmath82 ,
the program runs for about three minutes , determines @xmath63 , and finds the minimum - density periodic identifying code in figure [ fig : s4s5 ] left . with ` k ` set to @xmath73 ,
the program runs for about @xmath156 hours , determines @xmath64 , and finds the minimum - density periodic identifying code in figure [ fig : s4s5 ] right . 9 y. ben - haim and s. litsyn .
exact minimum density of codes identifying vertices in the square grid .
_ siam journal on discrete mathematics _
, 19:6982 , 2005 .
m. bouznif , f. havet , and m. preissmann .
minimum - density identifying codes in square grids . ` https://hal.inria.fr/hal-01259550 ` , research report rr-8845 , inria sophia antipolis
i3s , 2016 . also in _ proceedings of the 11th international conference on algorithmic aspects in information and management ( aaim16 )
_ , pages 7788 , 2016 .
g. cohen , s. gravier , i. honkala , a. lobstein , m. mollard , c. payan , and g. zmor . improved identifying codes for the grid . _ the electronic journal of combinatorics _
, comment to 6:#r19 , 1999 .
g. cohen , i. honkala , a. lobstein , and g. zmor .
new bounds for codes identifying vertices in graphs . _
the electronic journal of combinatorics _
, 6:#r19 , 1999 .
t. h. cormen , c. e. leiserson , r. l. rivest , and c. stein .
_ introduction to algorithms_. third edition . the mit press , 2009 .
m. daniel , s. gravier , and j. moncel . identifying codes in some subgraphs of the square lattice .
_ theoretical computer science _ , 319:411421 , 2004 .
r. m. karp .
a characterization of the minimum cycle mean in a digraph .
_ discrete mathematics _ , 23:309311 , 1978 .
m. g. karpovsky , k. chakrabarty , and l. b. levitin . on a new class of codes for identifying vertices in graphs .
_ ieee transactions on information theory _ , 44:599611 , 1998 .
j. pach and p. k. agarwal .
_ combinatorial geometry_. wiley , 1995 . | an identifying code in a graph is a subset of vertices having a nonempty and distinct intersection with the closed neighborhood of every vertex .
we prove that the infimum density of any identifying code in @xmath0 ( an infinite strip of @xmath1 rows in the square grid ) can always be achieved by a periodic identifying code with pattern length at most @xmath2 .
assisted by a compute program implementing karp s algorithm for minimum cycle mean , we find a periodic identifying code in @xmath3 with the minimum density @xmath4 , and a periodic identifying code in @xmath5 with the minimum density @xmath6 .
* keywords : * identifying code , minimum cycle mean . |
magnetic field lines that pass through the light cylinder of a rotating neutron star are twisted and carry electric currents @xmath7 .
these currents are sustained by electric field @xmath8 induced along the magnetic field @xmath9 , and ohmic dissipation @xmath10 feeds the observed pulsar activity . voltage associated with @xmath8 controls the energies of accelerated particles , creation of secondary electron - positron pairs , and emission of radio waves .
the accelerating voltage has been discussed in a number of works on pulsars beginning from early papers in the 1970s ( sturrock 1971 ; ruderman & sutherland 1975 ; arons & scharlemann 1979 ) .
the key dimensionless parameter of the polar - cap accelerator is [ eq : alpha ] = , where @xmath11 is the local corotation charge density of the magnetosphere ( goldreich & julian 1969 ) .
for a special value of @xmath12 ( close to unity ) a steady state was found for the polar - cap flow with significant particle acceleration ( e.g. arons & scharlemann 1979 ; muslimov & tsygan 1992 ) .
however , @xmath13 is not , in general , expected to take this special value ( e.g. kennel et al .
global solutions for approximately force - free pulsar magnetospheres give @xmath13 that significantly varies across the polar cap ( timokhin 2006 ) . in general , @xmath13 can take any value from @xmath14 to @xmath15 , depending on the polar cap distance from the rotation axis and the location inside the polar - cap region .
the character of the polar - cap accelerator strongly depends on @xmath13 ( mestel et al .
1985 ; beloborodov 2008 , hereafter b08 ) .
the steady solution with @xmath16 is a separatrix between two opposite regimes of efficient and inefficient acceleration .
, neglecting the deviation of @xmath17 from unity .
the precise @xmath17 is controlled by the curvature of magnetic field lines and the general relativistic effects ( muslimov & tsygan 1992 ) ; its exact value is close to unity and is not essential for the rest of the paper . ] in particular , if @xmath3 , @xmath8 is quickly screened in the charge - separated plasma flowing from the polar - cap surface .
the electric field satisfies maxwell equations that read ( in the co - rotating frame of the star , see e.g. fawley et al . 1977 ; levinson et al . 2005 ) , [ eq : gauss ] = 4(- ) , [ eq : maxw ] = 4(_b- ) .
if @xmath3 , there exists a velocity @xmath18 that allows the charge - separated flow @xmath19 to simultaneously satisfy @xmath20 and @xmath21 .
if the flow started from the conducting boundary ( which has @xmath22 ) with @xmath18 , no electric field would be generated ( then @xmath23 and @xmath24 ) .
the actual boundary has @xmath25 , as charges are lifted from the polar - cap surface with a small initial @xmath26 , comparable to the thermal velocity in the surface material .
the deviation of @xmath26 from @xmath27 implies @xmath28 or @xmath29 , which generates electric field .
b08 argued that ( [ eq : gauss ] ) and ( [ eq : maxw ] ) with @xmath3 always drive the flow toward @xmath18 , like a pendulum is driven by gravity toward its equilibrium position .
the resulting oscillations occur in space or time , according to ( [ eq : gauss ] ) or ( [ eq : maxw ] ) , respectively .
for example , the steady - state solution for a cold flow exhibits oscillations in space ( mestel et al . 1985 ; b08 ) .
the oscillatory behavior of the flow with @xmath3 is , in essence , langmuir oscillations ; they are generated near the boundary where the flow is injected with @xmath30 and accelerated toward @xmath18 . in this paper , we investigate the accelerator with @xmath3 in more detail . in 2 ,
we write down the steady - state solution for the charge - separated flow , generalized to non - zero temperature of the polar - cap .
we argue that the flow is unstable to small perturbations and can develop into a complicated time - dependent state with a broad momentum distribution . to explore the behavior of the flow , we perform fully kinetic time - dependent simulations .
the method of simulations is described in 3 , and the results are presented in sections 4 and 5 .
our simulations confirm the predicted turbulent langmuir oscillations with a small voltage .
particle acceleration in the flow with @xmath3 is insufficient to ignite pair creation .
implications of this `` dead zone '' for radio emission and outer gaps in pulsars are discussed in 6 .
it is natural first to attempt to construct a simple model assuming that the polar - cap flow is steady in the ( rotating ) frame of the neutron star . given the steady magnetic field in this frame , and the steady boundary conditions at the stellar surface excellent static conductor that can supply charges with a given temperature , one could expect a steady state to be established unless the flow is prone to an instability .
consider a charge - separated flow from the polar cap that carries electric current @xmath31 along magnetic field @xmath9 . in a steady state @xmath21 ( [ eq : maxw ] ) .
for simplicity , let us assume that @xmath9 is approximately perpendicular to the polar cap and let @xmath32 measure the altitude above the stellar surface .
a particle of mass @xmath33 and charge @xmath34 that starts with a lorentz factor @xmath35 at @xmath36 will accelerate as it moves along the magnetic field line , [ eq : a ] ( z)=_0+a(z ) , a=- , where @xmath37 is the electric potential and @xmath38 . gravitational acceleration ( and centrifugal acceleration in the rotating frame ) is neglected compared to the electric acceleration .
the electric potential satisfies poisson equation , [ eq : phi ] = -4(- ) , where we assumed that the potential varies along @xmath32 much faster than it does in the transverse directions , i.e. the acceleration length @xmath39 is much smaller than the characteristic transverse scale of the problem @xmath40 , which may be associated with the size of the polar cap .
this condition is satisfied for the flows considered below .
could be moved to the right - hand side of ( [ eq : phi ] ) and included in the effective @xmath2 . ]
the term @xmath41 may be viewed as a fixed background charge density .
the charge density of the flow itself is given by [ eq : rho ] ( z)=j_b _ 1^ . here
@xmath42 is the velocity of particles that start at @xmath36 with initial lorentz factor @xmath43 ; note that @xmath44 where @xmath45 is given by ( [ eq : a ] ) .
function @xmath46 describes the probability distribution of @xmath43 .
the width of this distribution is controlled by the temperature of polar cap @xmath47 .
for example , @xmath48 describes a cold polar cap ( @xmath49 ) where all particles have @xmath50 .
we multiply both sides of ( [ eq : phi ] ) by @xmath51 , substitute ( [ eq : rho ] ) , and find [ eq : diff ] ( ) ^2 = 4 . on the right - hand side , we used @xmath52 ( [ eq : a ] ) and @xmath53 .
integration of ( [ eq : diff ] ) in @xmath32 gives [ eq : steady ] ( ) ^2 = _
1^ w(_0)d_0 - , where p^2(_0,z)=^2 - 1=^2 - 1 . in ( [ eq :
steady ] ) we used @xmath54 and the boundary condition @xmath55 ( the stellar surface is modeled as a perfect conductor that can freely emit charges with @xmath56 )
. we also used @xmath57 and @xmath58 , as @xmath31 and @xmath2 do not significantly vary on the characteristic acceleration length @xmath59 , which is defined by [ eq : lambda ] _
p^2=. this length may be thought of as the plasma skin depth ; it is related to the plasma frequency @xmath60 , _
p^2= , where @xmath61 is the characteristic plasma density .
a quick estimate for @xmath31 and @xmath59 in pulsars may be obtained from the following consideration . the magnetic flux through the polar cap @xmath62 equals the flux through the light cylinder @xmath63 .
the bundle of open field lines is strongly twisted at the light cylinder ( toroidal component comparable to poloidal ) , and hence it carries electric current @xmath64 , according to stokes theorem . the current density near the star satisfies @xmath65 ( which follows from the fact that @xmath66 flows along @xmath9 ) ; this yields [ eq : j_b ] j_b~. then the plasma skin depth in the polar - cap accelerator may be expressed as @xmath67 the scale @xmath59 is much smaller than the typical size of the polar cap @xmath68 , where @xmath69 cm is the radius of the neutron star . once the injection distribution @xmath46 is specified , it is straightforward to numerically integrate ( [ eq : steady ] ) and find @xmath70 . in our sample models we chose
@xmath71 $ ] with @xmath72 ( cold ) and 0.03 ( warm ) ; the average injection momentum @xmath73 in the warm model equals @xmath74 .
figure [ fig : steady ] shows @xmath75 for the cold and warm solutions .
figure [ fig : steady ] also shows a third model where all particles injected at the polar cap have @xmath76 , i.e. @xmath46 is a delta - function . in this model , ( [ eq : steady ] ) simplifies to [ eq : cold ] ( ) ^2 = p(_0,z)-p_0 + , .
this flow is everywhere cold , i.e. its momentum distribution is described by @xmath77 $ ] . as one can see in figure [ fig : steady ] , the cold model with @xmath78 provides an excellent approximation to the exact warm model that has the same _ average _ value of @xmath73 .
the cold flow solution was discussed in earlier works ( mestel et al .
1985 ; b08 ) . for @xmath79
, the oscillation period is approximately given by ( b08 ) z_0 2 ^ 3/2 , 1 - 1 .
the precise period is obtained by numerical integration ; e.g. @xmath80 for @xmath81 .
the momentum of the steady cold flow @xmath82 oscillates between the injection momentum @xmath83 and a maximum value @xmath84 .
the minima and maxima are where @xmath85 , and from ( [ eq : cold ] ) one finds [ eq : pmax ] p_=. the above equations assumed @xmath86 . in real pulsars ,
@xmath13 varies due the field - line curvature and general relativistic effects ( muslimov & tsygan 1992 ) . the length - scale of this variation @xmath87 is typically larger or comparable to the stellar radius , which exceeds @xmath59 by several orders of magnitude .
when @xmath13 varies with @xmath32 , the analytical integration of the dynamic equation is not possible and one has to solve the two coupled differential equations ( b08 ) , @xmath88 the solution is similar to the case where @xmath13 is constant , as long as @xmath3 .
the momentum @xmath82 quasi - periodically passes through maxima and minima .
the only difference is that the period @xmath89 and @xmath90 now gradually change with @xmath32 ( see figure 1 in b08 ) .
although the cold flow solution with @xmath76 reproduces very well the electric potential @xmath75 of the exact warm solution with the same average @xmath73 , the warm and cold flows are qualitatively different . their different momentum distribution functions @xmath91 leads to a qualitatively different response to small perturbations
consider first the cold - flow solution shown by the blue dotted curve in figure [ fig : steady ] .
since all particles are injected with the same momentum @xmath76 , all of them follow a single trajectory in the phase space @xmath92 .
they periodically reach the minimum momentum @xmath73 at @xmath93 ( @xmath94 ) where potential @xmath37 reaches maximum .
there are no particles with momenta @xmath95 , so a small perturbation can not force any particles to reverse their direction of motion , and hence the perturbation will be advected along the flow .
this flow is expected to be stable .
in contrast , the warm flow ( dashed curve in figure [ fig : steady ] ) has a broad distribution of @xmath73 that extends from @xmath96 . at each peak of the electric potential ( at @xmath97 ) there is a population of particles with nearly zero velocities .
consider a perturbation at @xmath98 .
for example , suppose a small bunch @xmath99 of particles with momenta in a range @xmath100 are slightly pushed forward while the rest of particles are unperturbed .
this perturbation implies a local increase in electric current @xmath101 and hence @xmath102 ( [ eq : maxw ] ) , generating negative electric field @xmath103 at @xmath98 that tends to restore the condition @xmath21 .
in contrast to the initial perturbation , the induced @xmath103 affects _ all _ local particles , regardless of their momenta , not just bunch @xmath99 .
this has two implications : ( 1 ) the induced @xmath104 will easily and quickly reduce @xmath1 back to @xmath31 but will be unable to decelerate bunch @xmath99 to the momentum it would have in the steady state flow bunch @xmath99 will continue to move to @xmath105 with a larger momentum .
( 2 ) @xmath106 will give very slow particles @xmath95 negative velocities , creating a new bunch @xmath107 that slides backward down the potential hill .
bunch @xmath107 creates @xmath108 at @xmath109 , and the system reacts there by inducing a small @xmath110 , which accelerates all local particles , regardless their momenta , not just bunch @xmath107 . as a result , @xmath1 quickly recovers to @xmath31 , however , bunch @xmath107 is not stopped from moving backward and away from @xmath97 .
one concludes that the perturbation creates a permanent damage to the steady state that broadens the momentum distribution by creating backflowing particles .
this perturbation is _ not _ advected away along the flow , and can develop further .
the backflowing particles turn out to be trapped between two peaks of the electrostatic potential .
further development can be studied with kinetic time - dependent simulations ; it eventually completely destroys the steady state solution .
our numerical method is similar to that used by beloborodov & thompson ( 2007 , hereafter bt07 ) .
the plasma is modeled as a large number @xmath111 of individual particles that flow along the magnetic field lines .
we assume that the magnetic field is fixed in the co - rotating frame of the star ; thus @xmath31 and @xmath2 are fixed .
then the problem becomes essentially one - dimensional , as discussed in detail in bt07 . in the present paper , we consider only charge - separated flows , with no pair creation .
three other differences from the magnetar simulation in bt07 are as follows : ( 1 ) the magnetar problem had @xmath112 ( @xmath2 was negligible compared with @xmath113 ) ; in contrast , @xmath2 is crucial for polar - cap flows considered here .
( 2 ) the presence of gravity was essential for the closed - field circuit considered in bt07 , where the global plasma flow was studied on a scale comparable to the radius of the star ; in the problem considered here the electric fields are screened on a much smaller scale @xmath114 and the gravitational acceleration plays no role .
( 3 ) the flow behavior on the small scales @xmath115 may be studied using a smal computational box @xmath116 with an open outer boundary ( see below ) . in the absence of pair creation , the flow is composed of particles lifted from the surface . in most simulations presented below
we assume that all particles have the same mass @xmath33 and charge @xmath34 .
the particle motion is described by the equation , @xmath117 where @xmath118 is the momentum of the @xmath119-th particle in units of @xmath120 , and @xmath121 is the self - consistent electric field at the particle location @xmath122 .
the field is found by integrating gauss law ( [ eq : gauss ] ) along the magnetic field line , @xmath123.\ ] ] here @xmath124 is the column density of particles between @xmath36 and @xmath125 , and we used the boundary condition @xmath56 , as the material below the stellar surface is assumed to be a very good conductor that can emit free charges .
divergence of the perpendicular component of electric field @xmath126 is neglected in ( [ eq : e ] ) ( see bt07 for discussion of this approximation ) .
the approximation @xmath127 is valid if the characteristic scale of the flow acceleration @xmath89 is smaller than the transverse scale @xmath40 , which is limited by the polar - cap size @xmath128 ; the condition @xmath129 is satisfied in the dead - zone models presented below .
we also assume that @xmath2 is approximately constant on scale @xmath89 .
( [ eq : p ] ) and ( [ eq : e ] ) in essence describe a relativistic , time - dependent diode problem with an additional fixed background charge density @xmath41 .
as we track the motion of all particles individually , the continuity equation is automatically satisfied ; for a charge - separated flow it is equivalent to charge conservation , [ eq : charge ] + = 0 .
( [ eq : maxw ] ) follows from ( [ eq : gauss ] ) and ( [ eq : charge ] ) , so we will not need ( [ eq : maxw ] ) . instead , the parameter @xmath31 enters the problem as a boundary condition .
the magnetic field lines are frozen in the excellent conductor below the stellar surface , which sustains @xmath130 .
this condition is enforced in the simulation by injecting the charges in the computational box at @xmath36 with the fixed rate @xmath31 ( bt07 ) .
the electric current @xmath31 is enforced at one boundary @xmath36 .
since the computational box has a finite size @xmath131 , we also have to choose a boundary condition at @xmath132 and the value of @xmath131 . in all sample models shown in this paper
we use the simplest boundary condition : particles moving out of the box are lost and no particles enter the box at @xmath132 .
this condition may be refined by allowing a small inflow of returning particles at the outer boundary .
we ran test simulations that show that the refinements are not important as long as the boundary is sufficiently far , so that @xmath131 is much larger than the characteristic scale of the flow acceleration . in the one - dimensional model ,
the transverse gradients are neglected and the flow effectively has a slab geometry . then it is sufficient to follow particles flowing through a small area @xmath133 of the slab .
this allows one to chose a reasonable number of particles in the computational box , @xmath134 , e.g. @xmath111 , so that their dynamics can be followed in a reasonable computational time . on the other hand
, @xmath135 should be large enough so that the plasma scale @xmath59 contains many particles @xmath136 . in summary , we choose @xmath135 and @xmath131 so that 1 , n_p = n 1 . in this limit
, the results are expected to be independent of the choice of @xmath135 and @xmath131 ( we verified this by varying the two parameters ) . for most of our simulations @xmath137 and @xmath111 .
another requirement is a small time step of the simulation , @xmath138 , so that plasma oscillations are well resolved .
in our simulations and in reality the plasma above pulsar polar caps is collisionless . in the absence of pair creation it must satisfy the vlasov equation , @xmath139 where @xmath140 is the particle distribution function in phase space .
the electric current is @xmath141 where @xmath142 is the average velocity of the particles . as a first simple test ,
consider a uniform flow with @xmath143 , @xmath144 , and @xmath145 .
it is easy to see from ( [ eq : p ] ) , ( [ eq : e ] ) and ( [ eq : vlasov ] ) that the flow must remain in this state .
this behavior is reproduced by our simulations .
the steady uniform flow can have any momentum distribution @xmath146 as long as @xmath147 .
note that it requires a continual injection of particles at @xmath36 with the average velocity @xmath147 ( which also requires @xmath3 ) . as a second test ,
consider a `` cold '' flow where all particles move with momentum @xmath82 , with zero momentum dispersion .
suppose the flow is injected at @xmath36 with velocity @xmath148
. then @xmath8 must be generated , accelerating the flow . in a steady state , the solution for the cold flow must have the form , @xmath149 $ ] , where @xmath82 and @xmath150 can be described analytically .
we first test the special case @xmath151 ( michel 1974 ) .
the flow is accelerated by the self - consistent @xmath152 , and @xmath153 exceeds unity at @xmath154 . at heights
@xmath155 , velocity approaches @xmath156 , charge density of the flow @xmath157 approaches @xmath2 , and electric field @xmath8 asymptotes to a constant value , e_=^1/2 , where @xmath158 and @xmath159 .
then the flow momentum keeps growing linearly with @xmath32 , p(z)=^1/2 , z_p .
this analytical solution is reproduced by our simulations .
after an initial relaxation period ( comparable to the light crossing time of the computational box ) the system forgot initial conditions and relaxed to the steady state shown in figure [ fig : alpha1 ] ( in this example , @xmath160 ) .
the charge density of the flow is large near the polar cap surface and asymptotes to @xmath2 at @xmath161 , as expected .
+ then we studied cold flows with @xmath3 with a fixed injection velocity @xmath162 .
we chose in our sample numerical model @xmath81 and @xmath163 .
the computational box was initially empty ; the plasma injected at @xmath36 filled the box on the dynamical timescale @xmath164 and established a steady state shown in figure [ fig : alpha0.8cold ] .
the steady state is in perfect agreement with the analytical model of 2 .
the charge density @xmath165 has spikes at @xmath166 ( @xmath94 ) where the flow has the minimum velocity @xmath162 ; the height of each spike is @xmath167 .
the charge spikes are associated with maxima of the electric potential ( figure [ fig : alpha0.8cold]c ) .
the oscillating momentum has maxima @xmath168 , in excellent agreement with ( [ eq : pmax ] ) .
the period of oscillation is @xmath169 , same as found using the method of 2 .
as anticipated in 2.3 , we find that the steady state becomes unstable if we reduce @xmath162 to zero .
then any small perturbation ( e.g. due to numerical error ) completely destroys the steady state ; instead , a time - dependent state forms , with a broadened momentum distribution function . a steady flow with a finite @xmath170 can also be destroyed , although in this case a finite , sufficiently large perturbation is required .
in fact , this case provides a better setup for a numerical analysis of the instability , as we can control the form of the initial perturbation and then observe how it destroys the flow that was stable before the perturbation was applied .
we made such an experiment with the flow with @xmath81 and @xmath163 .
we applied a perturbation that was localized in space and time a small `` kick '' @xmath171 was given to all particles located in a small region @xmath172 ; in this experiment @xmath171 had a gaussian distribution with the mean value and dispersion equal to 0.02 .
we observed the following evolution .
as the localized perturbation moved along with the background flow , it was greatly amplified when it reached the potential maximum ( which corresponds to the minimum @xmath173 of the steady - state solution , see figure [ fig : alpha0.8cold ] ) , and some particles acquired a negative momentum , i.e. reversed their direction of motion .
most of the reversed particles became trapped between two potential maxima , and some of them were able to penetrate even further back , beyond the preceding potential peak .
the perturbation further spread in the phase space and the damage to the initial steady - state solution was further amplified with time , in particular near the potential maxima .
eventually , the entire flow became strongly time - dependent and the regular periodic structure of potential peaks disappeared .
the amplification of small ( linear ) perturbations at the potential maximum can be understood as follows .
consider a particle whose lorentz factor differs from that of the background cold flow by a small @xmath174 . as the particle moves along with the flow , its deviation @xmath174 remains constant , because it travels in the same electrostatic potential of the background flow ( cf .
[ eq : a ] ) . using the relation @xmath175
, we find the perturbation of momentum @xmath171 that corresponds to @xmath174 , p = ^-1 .
it grows as the particle ( and the background flow ) decelerates near the potential maximum ; the corresponding amplification factor @xmath176 is particularly large if @xmath162 is small .
the generation of backflowing particles at the potential peaks @xmath177 plays a key role in disrupting the steady state .
we also observe that , in a flow with a finite minimum velocity @xmath178 , only a finite , sufficiently large perturbation can destroy the steady state .
the perturbation would need to steal from particles energy @xmath179 so that they can be reflected by the potential hill .
the energy gap @xmath180 stabilizes the flow against infinitesimal perturbations , and only a sufficiently strong kick disrupts the flow .
the trapped / backflowing particles have a deteriorating effect on the steady state because they are not advected away with the flow and instead repeatedly approach the same potential peaks , amplifying the perturbations .
in addition , one can view the trapped particles as extra charge that distorts the electric field .
let @xmath181 be the number of particles trapped between two potential peaks @xmath182 and @xmath177 ; they create electric field @xmath183 at @xmath105 .
the corresponding distortion of the electrostatic potential @xmath184 grows linearly with @xmath32 and becomes significant at sufficiently large @xmath32 even if @xmath181 is small .
the distance @xmath32 required to produce @xmath185 is @xmath186 .
this behavior is qualitatively confirmed by our numerical experiments with larger simulation boxes @xmath131 the flow was found to become more unstable with increasing @xmath131 . in a more realistic model ,
particles are lifted from the polar cap with a thermal velocity dispersion @xmath187 .
the flow still starts with a small velocity @xmath188 and hence with a large charge density @xmath189 , which self - consistently generates the accelerating electric field .
the basic acceleration mechanism is the same as for the cold flow shown in figures [ fig : alpha1 ] and [ fig : alpha0.8cold ] .
however , there is a new feature : particles with different initial velocities behave differently in the collective electric potential , and the charge density @xmath165 is changed from the cold - flow solution , even though @xmath190 .
some particles have @xmath191 and can reverse their motion in the regions of growing potential ( @xmath104 ) , which greatly complicates the behavior of the distribution function @xmath192 . in our simulations , we modeled the warm injection by a one - dimensional maxwell distribution , which is a simple gaussian with dispersion @xmath193 equal to the mean value @xmath194 ; we chose @xmath195 .
as initial conditions we took the steady - state solution ( 2 ) .
the main parameter of the flow is @xmath13 , and we performed simulations for several values of @xmath13 in the range @xmath3 .
as expected , the steady state was quickly destroyed and the flow kept oscillating in space and time .
the basic parameters of the flow remained , however , similar to the steady cold model .
the average charge density ( averaged over oscillations ) is nearly equal to @xmath2 and the average velocity @xmath196 is nearly equal to @xmath27 , so that the condition @xmath197 is satisfied .
figure [ fig : meanvelocity ] shows the evolution of the hydrodynamic velocity @xmath198 measured at a fixed location @xmath199 ( we chose @xmath200 , in the middle of the computational box ; @xmath196 was calculated by averaging over particles inside a small bin around @xmath199 , of width @xmath201 ) .
the hydrodynamic velocity @xmath198 oscillates around @xmath27 ; these oscillations have a relatively small amplitude @xmath202 .
+ + the moderate value of the hydrodynamic velocity does not , in principle , exclude acceleration of a fraction of particles to much higher energies .
we therefore also studied the momentum distribution of particles in the flow .
figure [ fig : phasetraj]a shows a random snapshot of the particle distribution in the phase space for the flow with @xmath81 .
we randomly chose 1000 particles between @xmath36 and @xmath203 and the figure shows their locations in the two - dimensional phase space @xmath92 .
the simulation demonstrates the following : \(1 ) there is no high - energy tail in the momentum distribution .
\(2 ) at each @xmath32 , the momentum distribution has a pronounced narrow peak at @xmath204 .
thus , a large fraction of particles form a cold stream ; this fraction is approximately equal to @xmath13 ( see below ) .
the momentum of the cold stream @xmath204 is above ( but comparable to ) @xmath84 predicted by the steady - state model .
\(3 ) there is a low - energy wing in the momentum distribution which extends to negative momenta .
this broad component of the particle distribution has a hydrodynamic velocity close to zero ; these particles are `` trapped '' and do not contribute much to the current density ; however they make a significant contribution to charge density . in our sample model ,
about 20% of particles reside in the broad trapped component , and this fact has a simple explanation .
from the point of view of the cold stream dynamics , the broad component provides a background that offsets the effect of vacuum charge density @xmath2 by the fraction of 20% .
this fraction approximately equals to @xmath205 , so that the other particles ( fraction @xmath206 ) may move in the cold stream with @xmath207 and carry @xmath31 without the mismatch in charge density that would generate strong @xmath8 .
in essence , the broad component with backflowing particles allows the plasma to self - organize so that the cold stream can keep @xmath207 .
this is in contrast to the steady - state solution in 2 where all particles formed a stream with a positive velocity @xmath208 , which must be periodically decelerated and accelerated .
\(4 ) the cold stream momentum @xmath204 fluctuates in time ( the corresponding curve in figure [ fig : phasetraj ] moves in time ) .
however , the qualitative form of the phase - space distribution remains similar to that in figure [ fig : phasetraj ] .
as seen in figure [ fig : phasetraj]a , the flow momentum @xmath204 decreases near the outer boundary of the computational box @xmath132 .
this is an artifact of the boundary condition ( free escape with no backflow ) , which suppresses the density of backflowing particles near the boundary . as a result
, a modest negative electric field is induced near the boundary , decreasing @xmath204 so that the flow carries the required electric current @xmath31 . for comparison ,
figure [ fig : phasetraj]b shows a random snapshot of a similar model ( in the same interval @xmath209 ) that has twice as large computational box , @xmath210 .
as we increase @xmath131 , the boundary effect moves away to larger @xmath32 , affecting the flow properties only at @xmath211 .
the flow structure inside the box ( away from the boundary ) does not depend on @xmath131 . to check
whether the flow momentum depends on the size of the computational box , we ran several simulations with the same @xmath212 and different box sizes @xmath131 .
in each simulation , we measured the fluid momentum @xmath213 in the center of the box ( using a bin @xmath214 ) at time @xmath215 .
the results are shown in figure [ fig : zetacompare ] .
there is no systematic variation in @xmath213 with the box size ; the small variations ( @xmath216% ) are consistent with the fluctuations of @xmath213 in time for each model
. does not correspond to the average velocity @xmath196 shown in figure [ fig : meanvelocity ] , in the sense that @xmath217 , because of the broad low - energy tail of the distribution function .
compared with @xmath196 , the calculation of @xmath213 gives a higher weight to fast particles due to the additional factor @xmath218 in @xmath219 .
the average velocity remains close to @xmath27 , and the averaged momentum is larger than @xmath220 . ]
the polar - cap flows in pulsars extend through altitudes @xmath32 much larger than our box size @xmath131 . the fact that our results are independent of @xmath131 confirm the expected behavior
the plasma keeps oscillating and particle acceleration is quenched everywhere as long as the flow satisfies the condition @xmath3 .
we observe a quasi - uniform and quasi - steady behavior in the computational box ( apart from the initial acceleration region of length @xmath221 ) . in a realistic polar - cap flow ,
each segment of length @xmath222 should behave like our computational box .
we also ran simulations with varying @xmath223 .
we ran models with @xmath224 ( realistically , @xmath225 should be even smaller , @xmath226 , where @xmath227 is the star radius ) .
the results are similar to the case of @xmath228 .
the maximum momentum remains comparable to that given by ( [ eq : pmax ] ) as long as @xmath3 .
if @xmath229 ( which is equivalent to @xmath230 for @xmath231 ) , the charge - separated flow puled out from the polar cap is made of ions .
different ion species may end up in such a flow , and they will be accelerated to different velocities .
the mixed ion flow shares many features with the identical - particle model studied in the previous sections .
steady state solutions can be obtained using the method described in 2 .
ions with different masses and charges move with different hydrodynamical momenta and co - exist in a common , periodic electrostatic potential .
this steady solution is prone to kinetic instability similar to that described in sections 2 and 4 .
there is , however , an important new feature : the ion streams with different hydrodynamical momenta are prone to two - stream instability . to study the behavior of the mixed ion flow we slightly change the setup of our numerical simulation .
consider , e.g. , a mixture of protons and helium nuclei ( alpha - particles ) .
the particle injection at @xmath36 now consists of two ions species ; they have charges @xmath232 and @xmath233 , and masses @xmath234 and @xmath235 . the two species are injected with equal rates @xmath236 .
then alpha - particles carry electric current @xmath237 that is two times larger than the proton current @xmath238 .
thus , @xmath239 and @xmath240 are maintained at the boundary . to define a characteristic plasma skin depth @xmath59 we use ( [ eq : lambda ] ) where we replace @xmath241 by @xmath242 or , equivalently , by @xmath243 (
note that @xmath244 ) .
the characteristic plasma frequency is defined by @xmath245 .
figure 7 shows a snapshot of the phase - space distribution of ions long after the beginning of the simulation . in this sample model @xmath246 .
the modest value of @xmath13 ( not close to unity ) implies modest lorentz factors of particles and the fast development of instabilities .
the flow exhibits the following features : \(1 ) one period of the steady state solution is reproduced near the injection boundary @xmath36 .
the period @xmath247 agrees with the result from numerical integration of the corresponding steady state model .
( this feature is stable in our sample model because we chose a high injection velocity @xmath248 . )
\(2 ) at larger @xmath32 the periodic flow becomes unstable and develops into a configuration similar to that in figure 5 , except that now we have _ two _ cold variable streams .
besides the cold streams , there is a broad distribution of ions with smaller momenta and a negligible hydrodynamic velocity .
the origin of this broad component of trapped particles was discussed in 4.2 and plays here a similar role it is self - organized so that the streams may move with a relativistic speed without a mismatch in charge density .
\(3 ) further from the boundary ( at @xmath249 ) , the two streams develop a two - stream instability .
the growth rate of the instability may be estimated using an idealized model of two cold fluids with densities @xmath250 , @xmath251 and velocities @xmath252 .
it is straightforward to derive the dispersion relation for langmuir modes with frequency @xmath253 and wave - vector @xmath254 ( e.g. melrose 1986 ) ; it gives , @xmath255 where @xmath256 and @xmath257 .
using @xmath258 and the characteristic values of velocities @xmath252 from our simulation , we find from ( [ eq : disp ] ) that the most unstable modes have @xmath253 comparable to @xmath60 and their growth rate is @xmath259 . in the simulation
we observe a slightly smaller @xmath260 .
the distance over which the langmuir waves are amplified is roughly @xmath261 . as a result of the instability ,
the two streams are smeared out at large @xmath32 , in particular the stream of lighter ions .
no significant particle acceleration is seen in the simulation .
we have presented detailed one - dimensional time - dependent simulations of the plasma flow extracted from the polar caps of neutron stars .
the simulations provide a fully kinetic description of the flow , with self - consistent electric field and particle distribution function . in this paper
, we focused on the regime @xmath3 , where @xmath13 is the main parameter of the flow defined by ( [ eq : alpha ] ) . in agreement with the estimates of b08
, we find that the particles are accelerated to lorentz factors , [ eq : gam2 ] , and are not capable of igniting pair creation . in this sense , flows with @xmath3 are `` dead . ''
they are sustained by a modest voltage , oscillating in space and time .
although the simulation is limited to regions close to the pulsar surface , the result does not depend on the simulation box size , and hence should describe the entire polar cap flow , as long as @xmath13 remains between @xmath262 and @xmath263 .
the parameter @xmath13 is expected to vary along the magnetic field lines , on a scale comparable to the stellar radius ; we have verified that this variation does not change the oscillating behavior of the flow ( see also b08 ) .
the simulations show how a kinetic instability develops and disrupts the ideal periodic structure found in the analytical models of the dead zone ; the instability mechanism is described in 2 and 4 .
we find that the momentum distribution function has two distinct parts a variable `` cold stream '' and a broad wing at low momenta , which includes particles flowing backward to the polar cap .
the fraction of particles in the cold stream is approximately equal to @xmath13 ; the remaining fraction @xmath205 forms the broad component . even though the flow is turbulent , it shows no signs of particle acceleration to energies higher than that of the cold stream .
the value of parameter @xmath13 depends on the location and geometry of the polar cap .
a simplest magnetospheric configuration is that of a centered dipole .
then the parameter @xmath13 depends on the angle between the magnetic and spin axes , @xmath264 ; besides , it varies across the polar cap . for nearly aligned rotators ( @xmath265 ) , @xmath3 in the central part of the polar cap and @xmath266 in a ring - shaped zone near the edge of the polar cap ( timokhin 2006 ; parfrey et al .
2012 ) . in this case
, the dead zone occupies the central part of the polar cap , and @xmath267 discharge must be confined to the ring , matching the phenomenological `` hollow cone '' model of pulsar emission .
in contrast , the polar cap of an orthogonal rotator ( @xmath268 ) has @xmath269 , which enables @xmath267 discharge for the entire polar cap . at arbitrary misalignment @xmath270 ,
the values of @xmath13 are provided by global three - dimensional simulations of the magnetospheric structure ( e.g. spitkovsky 2006 ) and should play a key role for the geometry of the radio beam .
we presented our results using plasma skin depth @xmath59 as a unit of length and particle rest - mass @xmath271 as a unit of energy . in this form ,
the results do not depend on the charge or mass of the particles extracted from the polar cap , as long as the flow is made of identical particles .
in particular , ( [ eq : gam2 ] ) is valid for both electron flow ( @xmath272 ) and ion flow ( @xmath230 ) , and the phase - space distribution shown in figure 5 describes both cases .
note that the accelerating voltage is proportional to the particle mass ; voltage implied by ( [ eq : gam2 ] ) is different for ions and electrons by the factor of @xmath273 . the relatively high voltage in the ion flow
, @xmath274 is still hardly sufficient to ignite @xmath267 pair discharge by a seed electron or positron .
the identical - particle model may not hold for an ion flow ; in this case , new effects may enter the problem .
firstly , heavy ions pulled out from the polar cap may not be completely ionized and begin to lose electrons as they are accelerated and interact with the x - rays above the stellar surface ; this process effectively creates new charges , reminiscent of pair creation ( e.g. jones 2012 ) .
secondly , the ion flow may be a mixture of different nuclei which will be accelerated to different lorentz factors .
the mixed ion flow is prone to two - stream instability , possibly leading to formation of plasma clumps and generation of coherent radio emission . in our simulations
, we observe the expected two - stream instability , however do not observe significant structure ( clumps ) in the turbulent flow .
this may change in three - dimensional simulations .
the frequency of excited waves ( comparable to the ion plasma frequency ) is in the radio band , and coherent emission from clumps could create bright coherent emission . it remains to be seen whether this mechanism can contribute to the pulsar emission .
if it does , it would create an additional component of the radio pulse . in the case of approximately aligned rotator ,
the additional component would be generated in the central region of the polar cap , leading to a `` hollow cone + core '' structure of the radio pulse .
the charge - separated model of the dead zone can be modified to include possible backflowing particles from distant parts of the open field - line bundle ( e.g. from a pair - producing outer gap ) .
these particles can contribute to the current density and also serve as an additional background charge density , which may be modeled as a contribution to the effective `` vacuum '' charge density @xmath41 .
this would change the effective @xmath13 ( lyubarsky 1992 ; b08 ) , most likely reducing it .
an outer gap is expected to form in a charge - separated flow near the null surface @xmath275 ( cheng et al .
1986 ) . on a given field line
, the outer gap will be screened if it is loaded by multiple @xmath267 pairs produced by the discharge at the polar cap .
thus , the suppression of @xmath267 discharge near the field - line footpoint is an essential condition for the existence of an outer - gap accelerator .
therefore , one can expect an outer gap to form on field lines with footpoints in the dead zone .
we did not simulate in this paper flows with @xmath276 or @xmath266 ; in these cases particles must be strongly accelerated .
this regime leads to @xmath267 discharge that must be unsteady , with a significant intermittent backflow ( b08 ) . a model for oscillating discharge
may be constructed in hydrodynamical approximation ( levinson et al . 2005 ) , however a fully kinetic description is essential , as demonstrated by our results for the dead zone , where a significant fraction of particles are trapped and form a broad wing at low momenta in the distribution function .
the discharge simulation can be done using our setup of a fixed current @xmath21 at the stellar surface ( bt07 ) and incorporating pair creation .
we defer the simulations with @xmath276 and @xmath266 to a future work .
when this work was completed , the preprint by timokhin & arons ( 2012 ) came out .
they present simulations of charge separated flows , using a similar method , and the results agree with our results for @xmath3 .
they also consider flows with @xmath266 and @xmath276 , and find a strong unsteady @xmath267 discharge confirming the analysis in b08 . | we study plasma flows above pulsar polar caps using time - dependent simulations of plasma particles in the self - consistent electric field .
the flow behavior is controlled by the dimensionless parameter @xmath0 where @xmath1 is the electric current density and @xmath2 is the goldreich - julian charge density .
the region of the polar cap where @xmath3 is a `` dead zone '' in this zone particle acceleration is inefficient and pair creation is not expected even for young , rapidly rotating pulsars .
pulsars with polar caps near the rotation axis are predicted to have a hollow - cone structure of radio emission , as the dead zone occupies the central part of the polar cap .
our results apply to charge - separated flows of electrons ( @xmath4 ) or ions ( @xmath5 ) . in the latter case
, we consider the possibility of a mixed flow consisting of different ion species , and observe the development of two - stream instability .
the dead zone at the polar cap is essential for the development of an outer gap near the null surface @xmath6 . |
in their papers , kreussler - kurke @xcite and poon @xcite investigated algebraic structure of generic twistor spaces on @xmath2 , the connected sum of 3 copies of complex projective planes .
they showed that if the half - anticanonical system of a twistor space of @xmath0 is base point free , then the morphism associated to the system becomes a generically 2 to 1 covering map whose branch divisor is a quartic surface .
further , they determined defining equation of the quartic surface ; for the most generic twistor spaces , with respect to homogeneous coordinates on @xmath3 , the equation is of the form @xmath4 where @xmath5 is a ( homogeneous ) quadratic polynomial with real coefficients . from the equation , the intersection of the quadratic surface @xmath6 and any of the 4 plane @xmath7 is a double conic , and the intersection points of these 4 conics ( consisting of 12 points ) are ordinary double points of the quartic surface . for a generic quadratic polynomial @xmath5 , these are all singularities of the surface , but they showed that when is actually the branch divisor of the twistor spaces , the surface has one more node , which is necessarily real , so that the branch surface has 13 ordinary nodes in total . in this paper , we shall find moishezon twistor spaces on @xmath1 which can be regarded as a direct generalization of these twistor spaces on @xmath0 . more concretely , we show the following .
there exist twistor spaces on @xmath1 such that ( i ) the anticanonical system is 4-dimensional as a linear system , and the image of the associated rational map is a scroll @xmath8 of 2-planes in @xmath9 over a conic , ( ii ) there is an explicit and simple elimination of the indeterminacy locus of the anticanonical map , whose resulting morphisms is a generically 2 to 1 covering map onto the scroll @xmath8 , ( iii ) the branch divisor of the last covering , which will be denoted by @xmath10 throughout this paper , is an intersection of @xmath8 with a quartic hypersurface in @xmath9 , ( iv ) if we take homogeneous coordinates on @xmath9 such that the scroll @xmath8 is defined by @xmath11 , then the quartic hypersurface is defined by the equation @xmath12 where @xmath13 and @xmath5 are linear and quadratic polynomials with real coefficients respectively .
the double covering structure and similarity of the equations with would justify to call these twistor spaces a direct generalization of those by kreussler - kurke - poon .
the main tool of the present investigation is the anticanonical system of the twistor spaces . in section [ s:1 ]
we start by constructing a rational surface @xmath14 which will be contained in the twistor space @xmath15 on @xmath1 as a real half - anticanonical divisor , and then clarify the structure of bi - anticanonical system on @xmath14 .
next in section [ s : anticanonical ] we study the structure of the anticanonical map of the twistor spaces in detail . in section [ ss : acs ]
we show that the anticanonical map induces a rational map to @xmath9 , whose image is a scroll @xmath8 of planes over a conic , and give an explicit elimination of the indeterminacy locus , obtaining a degree 2 morphism @xmath16 .
next in section [ ss : modify ] we analyze the structure of the anticanonical map more in detail , and by applying some explicit blowups and blowdowns we modify the degree 2 morphism @xmath16 to another morphism @xmath17 so as to have no divisor to be contracted .
up to contraction of curves , this gives the stein factorization of the morphism @xmath16 .
although the modifications therein are a little bit complicated , they are rather natural in light of the structure of the anticanonical system , and indispensable for obtaining explicit construction of the twistor spaces . we also obtain a key technical result that the branch divisor of @xmath16 is a cut of @xmath8 by a quartic hypersurface . in section [ s : defeq ] ,
we explicitly determine a defining equation of the branch divisor of the double covering .
for this , in section [ ss : dc ] we find 5 hyperplanes in @xmath9 such that the intersection of the branch divisor with the hyperplanes becomes double curves , i.e.a curve of multiplicity 2 .
these double curves are analogous to the above 4 conics appeared in the case of @xmath0 , but to understand how they intersect each other requires some effort . in addition , in the present case , finding all the double curves is not so easy , since not all of the double curves are obtained as an image of twistor lines as in the case of @xmath0 . in section
[ ss : quadric ] we show that the 5 double curves are contained in a quadratic hypersurface in @xmath9 , and that such a hyperquadric is unique up to the defining equation of the scroll @xmath8 . in section [ ss : defeq ]
we prove the main result which determines the defining equation of the quartic hypersurface ( theorem [ thm : b ] ) .
the equation includes not only the quadratic polynomial obtained in section [ ss : quadric ] but also a linear polynomial , which might look strange at first sight .
we give an account for a geometric meaning of it . in section [ ss :
sing ] we investigate singularities of the branch divisor @xmath10 of the double covering . as above @xmath10 has 5 double curves , and at most of the intersection points of them , @xmath10 has ( non - real ) ordinary double points .
this is totally parallel to the case of @xmath0 explained at the beginning .
but in the present case there are exactly 2 special intersection points , at which @xmath10 has @xmath18-singularities . besides these ordinary double points and @xmath18-singular points ,
we show that @xmath10 has other isolated singularities and determine their basic invariants ( theorem [ thm:6sing ] ) .
the result means that in general @xmath10 has extra 6 ordinary double points . for obtaining this result , as in kreussler @xcite and kreussler - kurke @xcite in the case of @xmath0 , we compute the euler numbers of the relevant spaces , especially the branch divisor @xmath10 .
we also note that the concrete modification of the anticanonical map obtained in section [ ss : modify ] is crucial for determining the invariants of the singularities . in section [ ss : moduli ]
we compute dimension of the moduli space of the present twistor spaces .
we first compute the dimension by counting the number of effective parameters ( coefficients ) involved in the defining quartic polynomials .
next we show that some cohomology group of the twistor space an be regarded as a tangent space of the moduli space , and see that it coincides with the dimension obtained by counting the number of effective parameters .
this implies a completeness of our description obtained in section [ s : defeq ] . in section [ ss : gen ]
we discuss genericity of our twistor spaces among all moishezon twistor spaces on @xmath1 , indicated in the title of this paper .
the genericity implies a kind of density in the moduli space of moishezon twistor spaces on @xmath1 . in the course we also give a rough classification of moishezon twistor spaces on @xmath1 under some genericity assumption . in appendix
we show that an inverse of a non - standard contraction map employed in section [ ss : modify ] can be realized by an embedded blowup with a non - singular center in the ambient space , and point out in a concrete form that the modification can be regarded as a singular version of the well - known operation of hironaka @xcite for constructing non - projective moishezon threefolds .
we should also mention a relationship between this work and our previous paper @xcite . as we explained in the beginning , in the case of @xmath0 , the branch quartic divisor of the double
covering becomes of the form under a genericity assumption . in non - generic cases ,
as showed by kreussler - kurke @xcite , the branch divisor becomes similar but more degenerate form , and in the most degenerate situation , the branch divisor has a @xmath19-action . as a consequence , in that case the twistor spaces have a @xmath19-action .
the twistor spaces studied in @xcite is a generalization of these twistor spaces ( with @xmath19-action ) to the case of @xmath20 , @xmath21 . in this respect
we remark that the twistor spaces in this paper is obtained as a deformation of the twistor spaces studied in @xcite in the case of @xmath1 .
it is very natural to expect that we can obtain a further generalization to the case of @xmath20 , @xmath22 .
we hope to discuss this attractive topic in a future paper
. * acknowledgement . *
i would like to thank masaharu ishikawa for kindly answering a question about deformation of singularities , which helps me to be confident about computations of the euler number of the singular branch surface .
* notations and conventions . * the natural square root of the anticanonical bundle of a twistor space is denoted by @xmath23 .
if two varieties @xmath24 and @xmath25 are birational under some blowup or blowdown , and if @xmath8 is a subvariety of @xmath24 , then we often use the same symbol @xmath8 to mean the strict transform or birational image of @xmath8 under the blowup or blowdown , as far as it makes sense . for a line bundle @xmath26 and a non - zero section @xmath27 of @xmath26
, @xmath28 means the zero divisor of @xmath27 . for a linear subspace @xmath29 ,
we denote by @xmath30 to mean the linear system @xmath31 .
@xmath32 means the base locus of @xmath30 .
we mean @xmath33 and @xmath34 .
we are going to investigate twistor spaces on @xmath1 which contain a particular type of non - singular rational surface @xmath14 as a real member of @xmath35 . in this section
we first construct the surface @xmath14 as a blowup of @xmath36 , and then study the bi - anticanonical system on it . for this
, we define the line bundles @xmath37 and @xmath38 on @xmath36 as the pullback of @xmath39 by the projection to the first and second factors respectively .
we simply call members of the linear system @xmath40 as @xmath41-curves .
we define a real structure on @xmath36 as a product of the complex conjugation and the antipodal map .
next take any non - real @xmath42-curves @xmath43 , any @xmath44-curve @xmath45 , any distinct 3 points on @xmath46 , and 1 point on @xmath47 . by taking the images under the real structure , we obtain distinct 8 points in total ( see figure [ fig : square ] ) .
let @xmath48 be the blowup at these 8 points .
this surface @xmath14 has a natural real structure induced from that on @xmath36 , and also has an anticanonical curve @xmath49 where this time @xmath50 means the strict transform of the original @xmath50 , so that @xmath51 and @xmath52 .
the identity component of the holomorphic automorphism group of @xmath14 is trivial . in the sequel
we investigate the anticanonical and bi - anticanonical systems on @xmath14 . .
] [ prop : s1 ] ( i ) @xmath53 , so that @xmath54 is the unique anticanonical curve on @xmath14 , ( ii ) @xmath55 , @xmath56 , and @xmath57 .
\(i ) is immediate . for
( ii ) , @xmath58 means the inclusion @xmath59 .
riemann - roch formula and the rationality of @xmath14 imply @xmath60 .
these mean @xmath61 . therefore restricting @xmath62 to @xmath63
, we obtain the exact sequence @xmath64 this implies @xmath65 , and also @xmath66 . on the other hand by restricting the same system to @xmath67 , we obtain the exact sequence @xmath68 as @xmath69 clearly , the image of the restriction map @xmath70 in is 2-dimensional . projecting this to @xmath71 gives either a 1-dimensional subspace or @xmath71 itself . in order to prove @xmath72 ,
it is enough to exclude the former possibility .
suppose it is the case .
let @xmath73 be the zero point of a generator of the 1-dimensional subspace .
then we have @xmath74 and @xmath75 as we have already seen .
let @xmath76 be the blowup at @xmath77 and @xmath78 , and @xmath79 and @xmath80 the strict transforms of @xmath43 and @xmath81 respectively .
as @xmath82 , for any member @xmath83 with @xmath84 , we have @xmath85 , and the intersections are transversal .
this means that the system @xmath86 on @xmath87 is base point free .
then as @xmath88 , the morphism associated to the system contracts @xmath79 and @xmath80 to points .
on the other hand , by , @xmath45 and @xmath89 are mapped to mutually different points by the same morphism .
this is a contradiction because @xmath90 , @xmath91 , and @xmath43 is connected .
@xmath92 let @xmath93 be the morphism associated to the system @xmath94 . since @xmath95 and @xmath96 , @xmath97 factors as @xmath98 , where @xmath99 denotes the blowdown of @xmath45 and @xmath89 .
[ prop:2to1 ] the morphism @xmath97 is generically 2 to 1 , and the branch divisor is a quartic curve .
further , the images @xmath100 and @xmath101 are the same line , and @xmath102 and @xmath103 are different 2 points on the line .
the morphism @xmath97 is surjective since @xmath104 .
this also means it is generically 2 to 1 .
further , a general member @xmath105 of @xmath106 , which is an irreducible non - singular curve by bertini s theorem , is an elliptic curve because @xmath107 .
hence the branch curve of @xmath97 must be of degree 4 . for the images @xmath100 and @xmath101 , as in the proof of proposition [ prop : s1 ]
, we have the exact sequence , and the image of @xmath70 projects isomorphically to @xmath71 and @xmath108 .
hence @xmath100 and @xmath101 are lines .
further these lines are identical , since they are precisely the 2-dimensional images into the dual space @xmath109 by the dual map of @xmath70 .
also the last claim for @xmath102 and @xmath110 is clear from the exact sequence , @xmath90 and @xmath111 . @xmath92 the following property of the branch quartic
will also be needed later .
[ prop : s3 ] if we denote the line and the pair of points on it by @xmath112 and @xmath113 and @xmath114 respectively , then @xmath115 and @xmath116 are smooth points of the branch quartic curve , and the curve is tangent to @xmath117 at these 2 points .
( hence @xmath117 is a bitangent of the branch curve . )
let @xmath118 be the branch curve of @xmath97 .
as @xmath119 , there is a line @xmath120 such that @xmath121 .
but as @xmath122 , we obtain @xmath123 .
since @xmath124 and both @xmath125 and @xmath126 are non - empty , it follows that @xmath127 , so that @xmath128 by the real structure .
further , after the blowdown @xmath99 , the curve @xmath67 is of course locally reducible at the images of @xmath45 and @xmath89 .
hence we have @xmath129 as divisors .
this means that either @xmath118 has double points at @xmath115 and @xmath116 , or otherwise @xmath118 is smooth at @xmath115 and @xmath116 and is tangent to @xmath117 at these 2 points .
but in the former case by smoothness of @xmath14 there have to be extra exceptional curves over @xmath115 and @xmath116 , which contradicts @xmath130 .
hence the claim follows .
let @xmath14 be the rational surface equipped with the real structure constructed in the previous section , and @xmath131 the unique anticanonical curve on @xmath14 .
let @xmath15 be a twistor space on @xmath132 and suppose that @xmath15 contains @xmath14 as a real member of @xmath35 .
the following property of @xmath35 is immediate to see and we omit a proof . [
prop : f ] the system @xmath35 satisfies the following : ( i ) @xmath133 , ( ii ) @xmath134 , ( iii ) the number of reducible members of @xmath35 is two , and both of the members are real . and
a cube in @xmath135 we note that it readily follows from ( i ) and ( ii ) that a general member @xmath87 of the pencil @xmath35 is also obtained from @xmath36 by blowing up 8 points arranged as in figure [ fig : square ] , where the positions of the 8 points are not identical to the original ones .
we denote the 2 reducible members of @xmath35 by @xmath136 where we make distinction between @xmath137 and @xmath138 by declaring that @xmath139 and @xmath140 contains the component @xmath81 .
we denote @xmath141 and @xmath142 , both of which are twistor lines by @xcite .
then these divisors and curves form a tetrahedron as illustrated in figure [ fig : trans0 ] , ( a )
. these will be significant for our analysis of the anticanonical system on the twistor spaces .
we show the the following basic properties of the anticanonical system .
note that ( iv ) means that @xmath15 is moishezon .
[ prop:2f ] the anticanonical system @xmath143 of the twistor space @xmath15 satisfies the following : ( i ) @xmath144 , ( ii ) @xmath145 , ( iii ) if @xmath146 denotes the blowup at @xmath147 , @xmath148 the exceptional divisor , and @xmath149 , then @xmath150 , ( iv ) if @xmath151 denotes the morphism associated to @xmath152 , then the image @xmath153 is a scroll of 2-planes over a conic , and the morphism @xmath154 is generically 2 to 1 over the scroll . by the blowup @xmath155 , the tetrahedron in @xmath15
is transformed to be a cubic in @xmath156 as in figure [ fig : trans0 ] , ( b ) .
_ proof of proposition [ prop:2f ] . _
( i ) is immediate from proposition [ prop : s1 ] ( ii ) , proposition [ prop : f ] ( i ) , and the exact sequence @xmath157 where the last zero is a consequence of @xmath158 and the riemann - roch formula applied to @xmath23 . the claim ( ii ) also follows from this exact sequence and proposition [ prop : s1 ] ( ii ) . for ( iii )
, let @xmath159 be the strict transform of @xmath14 .
then @xmath159 is biholomorphic to @xmath14 , and @xmath160 .
hence we have an exact sequence @xmath161 since @xmath162 , we obtain that the restriction map @xmath163 is surjective .
further , as @xmath164 under the biholomorphism @xmath165 , we have @xmath166 .
further , @xmath167 . hence we obtain an isomorphism @xmath168 .
therefore by the third claim of proposition [ prop : s1 ] ( ii ) , we obtain @xmath169 .
let @xmath170 be the rational map associated to the anticanonical system @xmath171 , so that @xmath172 .
for ( iv ) it is enough to show that @xmath173 is the 3-dimensional scroll as in the statement , and the rational map @xmath174 is generically 2 to 1 .
let @xmath175 be the subspace of @xmath176 generated by all sections of the form @xmath177 where @xmath178 .
this is a 3-dimensional subspace .
then we have the following left commutative diagram of rational maps : @xmath179^{{\phi } } \ar@{->}[rd]_{{f } } & \mathbb{cp}^{4 } \ar@{->}[d]^{\pi } \\ & \mathbb{cp}^{2}\\ } \qquad\qquad \xymatrix { z \ar@{->}[r]^{{\phi } } \ar@{->}[rd]_{{f } } & y \ar@{->}[d]^{\pi } \\ & \lambda } \ ] ] where @xmath180 is the linear projection induced by the inclusion @xmath181 and @xmath13 is the rational map associated to the subsystem @xmath182 .
clearly the image @xmath183 is a conic , for which we denote by @xmath184 . hence writing @xmath185 ,
@xmath8 is exactly the scroll as in the statement of ( iv ) , and we obtain the right commutative diagram in .
we have to show that @xmath186 is surjective and generically 2 to 1 . for these , we note that by the definition of @xmath13 , for any @xmath187 , @xmath188 belongs to the pencil @xmath35 . then by for _
any _ non - singular member @xmath189 , the restriction @xmath190 is exactly the rational map associated to the system @xmath191 , where the target space is the fiber plane @xmath192 . by proposition [ prop:2to1 ] , this means that @xmath193 is surjective as far as @xmath188 is non - singular .
therefore @xmath194 itself is also surjective to @xmath8 .
now the final claim ( 2 to 1 over @xmath8 ) is immediate from these considerations and proposition [ prop:2to1 ] .
@xmath92 thus the anticanonical map is 2 to 1 over the scroll @xmath8 .
further , by the above argument and proposition [ prop:2to1 ] , the branch locus of the 2 to 1 map has degree 4 on the planes @xmath188 , from which one might find similarity with those in the case of @xmath0 @xcite .
but we are yet far from the goal . in the next subsection
we shall investigate structure of the anticanonical map more closely .
we use the notations @xmath194 , @xmath184 , @xmath8 , @xmath13 and @xmath180 given in the proof of the proposition .
further define @xmath117 to be the singular locus of the scroll @xmath8 .
@xmath117 is a line , and is exactly the indeterminacy locus of the projection @xmath180 .
this line plays an important role throughout this paper .
note that for a hyperplane @xmath195 , the intersection @xmath196 splits to planes iff @xmath197 projects to a line in @xmath198 ( in which the conic @xmath184 is contained ) , and otherwise @xmath196 is a cone over @xmath184 whose vertex is the point @xmath199 .
further , in the former situation , @xmath200 is a double plane ( i.e.a non - reduced plane of multiplicity 2 ) iff the line is tangent to @xmath184 .
let @xmath201 be the blowup at @xmath117 , and @xmath202 the exceptional divisor .
@xmath203 is biholomorphic to the total space of the @xmath198-bundle @xmath204 , and @xmath202 is identified with the subbundle @xmath205 , so that it is biholomorphic to @xmath36 .
more invariantly , we have a natural isomorphism @xmath206 .
the composition @xmath207 is a morphism which is identified with the bundle projection , for which we denote by @xmath208 . in order to treat the branch locus of the degree 2 morphism @xmath151 ( see proposition [ prop:2f ] ) properly , we consider a lifting problem of the morphism @xmath151 to @xmath203 . by definition of @xmath13 ,
the composition @xmath209 is the rational map associated to the 2-dimensional linear system @xmath210 on @xmath156 ( whose members are the total transforms of members of the linear system @xmath182 ) .
this linear system on @xmath156 is a subsystem of @xmath211 .
but since the pencil @xmath35 has @xmath43 and @xmath81 as components of the base locus , all the above total transforms contain the divisor @xmath212 .
hence subtracting @xmath213 and recalling @xmath214 , we can regard linear system @xmath210 as a subsystem of @xmath152 .
this 2-dimensional subsystem of @xmath152 still has @xmath213 as the fixed components , so by subtracting it we obtain a 2-dimensional subsystem of @xmath215 , which is readily seen to be coincide with @xmath216 itself .
hence the composition @xmath217 can be regarded as the rational map associated to @xmath216 .
however , the curves @xmath45 and @xmath89 are contained in bs@xmath35 ( proposition [ prop : f ] ( ii ) ) , and the strict transforms of these curves to @xmath156 are exactly the base locus of @xmath216 . thus the composition @xmath217 has the strict transforms of @xmath45 and @xmath89 as its indeterminacy locus .
therefore , the morphism @xmath218 can not be lifted to @xmath219 as a morphism , because the composition with @xmath220 is not a morphism .
so let @xmath221 be the blowup at @xmath63 , and @xmath222 and @xmath223 the exceptional divisors .
here we are regarding @xmath45 and @xmath89 as curves in @xmath156 .
let @xmath224 , and @xmath225 .
( this time we do not subtract @xmath226 because @xmath45 and @xmath89 are not base curves of @xmath152 . ) obviously @xmath227 is the rational map associated to @xmath228 , and it is clearly a morphism . then we have the following : the morphism @xmath229 can be lifted to a morphism @xmath230 . namely there is a morphism @xmath231 such that @xmath227 factors as @xmath232 . as in the above explanation
, the composition @xmath233 is the rational map associated to the system @xmath234 . in the same way for the identification between a conic and a line on a plane by means of the projection from a point ,
once we fix any non - zero element of @xmath235 , by taking a product with it , members of the system @xmath216 can be identified with those of the pencil @xmath236 , and the rational map associated to @xmath216 is identified with the rational map associated to @xmath236 . therefore the composition @xmath233 can be identified with the rational map associated to the pencil @xmath237 .
this pencil has @xmath238 as the fixed component , and if we subtract this , the pencil becomes free , since @xmath239 and @xmath240 consists of a single member @xmath241 ( a reduced curve ) by proposition [ prop : s1 ] ( i ) .
hence the composition @xmath242 has no point of indeterminacy .
we write @xmath243 for this morphism .
thus we are in the following left situation : @xmath244_{f_2 } \ar@{->}[rd ] & \tilde y \ar@{->}[d]^{\nu } \ar@{->}[ld ] \\
\lambda & y \ar@{->}[l]^{\pi } } \qquad\qquad \xymatrix { z_2 \ar@{->}[d]_{f_2 } \ar@{->}[r]^{\tilde{\phi}_2 } \ar@{->}[rd]^{\phi_2 } & \tilde y \ar@{->}[d]^{\nu } \\
\lambda & y \ar@{->}[l]^{\pi } } \ ] ] where all maps except @xmath180 are morphisms , and the 2 triangles are commutative .
( the map @xmath245 is @xmath227 and the map @xmath246 is @xmath208 . ) for lifting @xmath227 to @xmath203 , we need to assign a point of @xmath203 for each point of @xmath247 . for this ,
recall that @xmath248 is isomorphic outside @xmath249 .
for @xmath250 , of course , we assign the point @xmath251 . for @xmath252
define @xmath253 and @xmath254 .
the inverse image @xmath255 is a fiber of the projection @xmath256 . in accordance with the natural isomorphism @xmath257 , this fiber and the fiber @xmath258 intersect at a unique point .
let @xmath259 be this point , and we assign @xmath260 to @xmath261 .
define @xmath231 to be the map thus obtained .
then @xmath262 is clearly continuous and is a lift of @xmath227 .
as @xmath262 is holomorphic on the complement of the analytic subset @xmath263 , riemann s extension theorem means that it is automatically holomorphic on the whole of @xmath247 .
thus we get the situation right in and obtained the desired lift @xmath262 . @xmath92 since the lift @xmath264 is a degree 2 morphism between non - singular spaces , we can speak about its branch divisor .
namely we first define the ramification divisor @xmath265 on @xmath247 as a zero divisor of a natural section ( defined by the jacobian ) of the line bundle @xmath266 , and then let the branch divisor @xmath267 to be the image @xmath268 , which is necessarily a divisor .
we can determine the cohomology class of this divisor as follows : [ branchdivisor1 ] let @xmath267 be the branch divisor of the lift @xmath269 as above .
then @xmath270 , where @xmath271 .
as before let @xmath202 be the exceptional divisor of the blowup @xmath272 , and let @xmath273 be the cohomology class of the fiber class @xmath274 .
the cohomology group @xmath275 is a free @xmath276-module generated by @xmath202 and @xmath273 .
@xmath202 is isomorphic to @xmath277 , and the restriction @xmath278 can be identified with the projection to @xmath117 .
define @xmath44 to be the bidegree of a fiber of this projection .
then the normal bundle is @xmath279 , while @xmath273 is restricted to the class @xmath42 . from these
we can readily deduce that the restriction map @xmath280 is isomorphic .
let @xmath281 be any hyperplane containing the line @xmath117 .
then since @xmath184 is a conic , we have @xmath282 , which means @xmath283 then by using the above explicit form of the restriction map , we obtain @xmath284 hence in order to prove @xmath270 , it suffices to show @xmath285 . in order to obtain the restriction @xmath286 , for each @xmath187 we write @xmath287 , which is the strict transform of a member of the pencil @xmath35 .
then it is not difficult to see that the restriction @xmath288 is naturally identified with the restriction of the original restriction @xmath289 .
if @xmath290 is non - singular ( which is the case for almost all @xmath291 ) , the last restriction @xmath292 is exactly the bi - anticanonical map of @xmath290 .
hence by propositions [ prop:2to1 ] and [ prop : s3 ] , the branch curve is a quartic curve which is tangent to the line @xmath117 at 2 points .
the last 2 points are independent of the choice of @xmath291 , since they are exactly @xmath293 and @xmath294 .
therefore @xmath286 contains the fibers of @xmath256 ( the restriction of @xmath272 to @xmath202 ) over the points @xmath115 and @xmath116 by multiplicity 2 respectively .
moreover since the intersection of the line @xmath117 with the branch quartic curve of @xmath295 consists of the 2 points @xmath115 and @xmath116 , if @xmath286 contains an irreducible curve of bidegree @xmath296 with @xmath297 , then we have @xmath298 . but this can not occur since the original @xmath194 does not contain @xmath117 as a branch locus .
thus we have obtained @xmath299 .
@xmath92 define a divisor @xmath10 on @xmath8 by @xmath300 . by proposition [ branchdivisor1 ] , @xmath301 , and since @xmath302 ,
the morphism @xmath229 is a generically 2 to 1 covering with branch @xmath10 .
( the former implies that @xmath10 is a cut of @xmath8 by a quartic hypersurface .
we will explicitly obtain a defining equation of this hypersurface in the next section . )
however @xmath227 is _ not _ a finite map but contracts the divisors @xmath303 and @xmath223 as we see next until completing as proposition [ prop : factor1 ] . for this
, we first notice that the exceptional divisors @xmath222 and @xmath223 of @xmath304 are isomorphic to @xmath36 and the normal bundles satisfy @xmath305 ( see figure [ fig : trans ] ( c ) . ) therefore @xmath222 and @xmath223 can also be blowdown along the projection different from the original @xmath306 and @xmath307 .
let @xmath308 be this blowdown .
( see ( c ) @xmath309 ( d ) in figure [ fig : trans ] . ) @xmath310 is still non - singular .
the birational transformation from @xmath156 to @xmath310 is exactly atiyah s flop at @xmath45 and @xmath89 .
the divisors @xmath311 and @xmath312 in @xmath156 are also isomorphic to @xmath36 , and they are respectively blown up at 2 points through @xmath313 .
we use the same letters @xmath311 and @xmath312 to mean these divisors in @xmath247 .
these divisors are not affected by the blowdown @xmath314 , and we still denote by @xmath311 and @xmath312 for their images in @xmath310 , as displayed in figure [ fig : trans ] ( d ) . next we show that these 2 divisors @xmath311 and @xmath312 in @xmath310 can be contracted to non - singular rational curves simultaneously . for this
we first consider the divisor @xmath311 in @xmath156 , so that @xmath315 , and take the cohomology class of a fiber of the projection to @xmath316 which is different from the projection to @xmath43 .
next pullback the class by the blowup @xmath313 and push it to @xmath317 by @xmath314 .
( in figure [ fig : trans ] these cohomology classes are represented by non - dotted lines on @xmath311 . )
thus we obtain a cohomology class on @xmath317 whose self - intersection number is zero . the linear system on this @xmath311 having this cohomology class is clearly a free pencil , and induces a morphism to @xmath316 .
let @xmath318 be this morphism .
general fibers of @xmath319 are non - singular rational curves , and there exist precisely 2 singular fibers , both of which consist of 2 non - singular rational curves intersecting at a point .
the same is true for @xmath312 , and let @xmath320 the morphism corresponding to @xmath319 .
now @xmath319 and @xmath321 naturally fit on the intersection @xmath322 and form a morphism @xmath323 . here
note that these two @xmath316-s are identified at 2 points , and @xmath324 has reducible fibers exactly over these 2 points , both of which consist of _ three _ rational curves . in figure
[ fig : trans ] ( d ) , these 2 reducible fibers are written by 3 bold lines respectively . to @xmath310 .
the picture ( b ) is identical to ( b ) in figure [ fig : square ] ; the present ( b ) is obtained from the original ( b ) by just cutting out ( just for presentation ) along the 2 twistor lines @xmath325 and @xmath326 .
so in each of ( b ) , ( c ) and ( d ) , the two @xmath325-s are identified in the direction indicated by the arrows , and the same for @xmath326-s .
( e ) is obtained from ( d ) by contracting four @xmath327-curves , and will be used later . ]
we are going to show that the reducible connected divisor @xmath148 on @xmath310 can be contracted along @xmath324 .
for this we need to examine the normal bundle , @xmath328|_{e_1\cup{\overline}e_1}$ ] .
the restriction of the normal bundle @xmath329|_{e_1}$ ] is degree @xmath330 on irreducible fibers of @xmath319 , and degree @xmath331 on the 2 irreducible components of the ( two ) singular fibers .
( see figure [ fig : trans ] ( d ) . ) from this we deduce that the restriction of the line bundle @xmath328|_{e_1\cup{\overline}e_1}$ ] is @xmath330 on irreducible fibers of @xmath324 , and @xmath331 on the end components of the reducible fibers , while @xmath332 on the middle component of the reducible fibers .
now by the relative version of nakai - moishezon criterion for ampleness , these numerical data imply that the dual line bundle @xmath328^*|_{e_1\cup{\overline}e_1}$ ] is @xmath333-ample . moreover again from the numerical data , for a direct image of the dual bundle
, we have @xmath334 ^ * | _ { e_1\cup{\overline}e_1 } ) ^ { \otimes m}=0 { \hspace{2mm}}{\text{for any}}{\hspace{2mm}}m>0.\ ] ] therefore by a theorem of fujiki ( * ? ? ?
* theorem 2 ) , the divisor @xmath335 can be contracted to @xmath336 along the morphism @xmath324 .
let @xmath337 be the birational morphism obtained this way , and put @xmath338 and @xmath339 , so that @xmath340 can be naturally identified with the target space of @xmath324 .
as the degree of the restriction of @xmath341 to irreducible fibers of @xmath319 is @xmath330 as above , we have @xmath342 and possibly outside the 2 points @xmath343 , @xmath344 has ordinary double points along @xmath340 .
( in section [ ss : blowup ] we will obtain an explicit defining equation of @xmath344 around the 2 points . ) this way by contracting @xmath148 in @xmath310 we have obtained a singular variety @xmath344 .
then the morphism @xmath227 descends to @xmath344 : [ prop : factor1 ] let @xmath229 be the generically 2 to 1 covering as before .
then @xmath227 descends to a morphism @xmath17 .
namely there is a morphism @xmath345 such that @xmath346 .
we first show that @xmath227 descends to a morphism @xmath347 .
recall that @xmath227 is induced by the system @xmath228 , where @xmath348 and @xmath214 .
in accordance with those on @xmath349 in the proof of proposition [ branchdivisor1 ] , let @xmath44 be the fiber class of the projection @xmath350 .
then we obtain @xmath351 ( see figure [ fig : trans ] ( b ) for @xmath352 . )
further , as the curve @xmath353 intersects @xmath311 transversally at exactly 1 point and the same for @xmath312 ( again see figure [ fig : trans ] ( b ) ) , we have @xmath354 therefore pulling back to @xmath247 we obtain @xmath355 and analogous result for the restrictions to @xmath312 and @xmath223 .
these imply that the direct image sheaf @xmath356 is still invertible and @xmath357 .
if we use the projection @xmath358 , the last isomorphism can be rewritten as @xmath359 then since @xmath360 , the morphism associated to @xmath228 factors through the morphism associated to @xmath361 .
letting @xmath362 be the last morphism , this means @xmath363 as claimed . in a similar way
we next show that @xmath362 descends to a morphism @xmath345 . from the direct image @xmath364
is still an invertible sheaf ( on @xmath344 ) , whose restriction to @xmath365 is of degree 1 . then by the natural isomorphisms @xmath366 the map associated to @xmath361 factors through the map induced by @xmath367 .
therefore if we define @xmath368 to be the map associated to @xmath367 , we have @xmath369 , as claimed .
hence we have obtained @xmath370 .
@xmath92 thus we arrived at the following situation : @xmath371_{\mu_2 } \ar@{->}[r]^{\mu_4\circ\mu_3 } \ar@{->}[rd]^{\phi_2 } & z_4 \ar@{->}[d]^{\phi_4 } \\
z_1 \ar@{->}[r]_{\phi_1 } & y } \ ] ] where all maps are morphisms and the 2 triangles are commutative . since @xmath372 is birational
, @xmath368 is still a degree 2 morphism branching at the divisor @xmath10 .
but in contrast with @xmath227 , it does not contract divisors anymore : [ prop : phi4 ] the morphism @xmath368 does not contract any divisor to a point or a curve . it is enough to show that the morphism @xmath218 does not contract any irreducible divisor other than @xmath311 and @xmath312 .
let @xmath105 be such a divisor .
if @xmath105 is real , then @xmath373 for some @xmath374 and @xmath375 , and @xmath376 by the contractedness property . for computing this intersection number ,
we notice , as @xmath377 , that we have @xmath378 .
hence we have @xmath379 and the same for @xmath312 . from these , recalling @xmath380 ( as we are over @xmath1 ) , we obtain @xmath381 therefore we have @xmath382 .
hence @xmath105 is not contracted to a curve or a point by @xmath151 .
when @xmath105 is not real , by applying the above computations for @xmath383 instead of @xmath105 , we obtain @xmath384 . hence , since @xmath385 , we again conclude @xmath382 .
hence in the non - real case too , @xmath105 can not be contracted to a point or a curve by @xmath151 , as claimed .
@xmath92 as a consequence , we obtain the following [ prop : isolated ] the branch divisor @xmath10 has only isolated singularities .
let @xmath386 be the stein factorization of the morphism @xmath368 .
@xmath387 is necessarily birational .
then @xmath388 is just a double covering with branch @xmath10 . hence if @xmath10 has singularities along a curve , so is @xmath389 .
since @xmath10 does not contain @xmath117 and @xmath390 , this means that the birational morphism @xmath387 resolves the singularities along the curve .
hence @xmath387 contracts a divisor .
this contradicts proposition [ prop : phi4 ] .
@xmath92 we note that the proof of proposition [ prop : phi4 ] means that the original anticanonical map @xmath186 does not contract any divisor .
we also note that the morphism @xmath387 in the proof of proposition [ prop : isolated ] contracts ( at worst ) finitely many curves , and all these curves are over singular points of @xmath10 .
we will investigate these singularities in detail in section [ ss : sing ] .
in the last section we analyzed the anticanonical system on the twistor space in detail and obtained the space @xmath344 and a degree 2 morphism @xmath345 which is explicitly birational to the original anticanonical map , and which does not contract any divisor .
we further showed that the branch divisor @xmath10 of @xmath368 is a cut of @xmath8 by a quartic hypersurface in @xmath9 . in this section
we shall determine defining equation of this quartic hypersurface .
we also determine the number of singularities of the branch divisor . our way for obtaining the equation includes finding hyperplanes @xmath391 such that , regarding the intersection @xmath392 ( which is either a plane or a cone as in the beginning of section [ ss : modify ] ) as a reduced divisor on @xmath8 ,
the restriction @xmath393 is a double curve ( i.e.a non - reduced curve of multiplicity 2 ) .
so it is similar to the method of poon @xcite ( for the case of @xmath0 ) , but the origin of some of the double curves is different from the case of @xmath0 . we keep the notations from the last section .
we are going to show the existence of _ five _ double curves , two of which are easy to find as we see now .
first we recall there are diagrams in ; @xmath184 is a conic in @xmath198 and for each @xmath187 , @xmath394 is a member of the pencil @xmath35 , i.e.@xmath184 is a parameter space of @xmath35 . for any @xmath395 , @xmath192 is a plane containing the line @xmath117 .
let @xmath396 and @xmath397 be the reducible members as in , and let @xmath398 and @xmath399 be the points of @xmath184 such that @xmath400 and @xmath401 hold .
let @xmath402 and @xmath403 be the hyperplanes in @xmath9 which are the inverse images of the tangent line of @xmath184 at the points @xmath398 and @xmath399 respectively under the projection @xmath180 . then the restrictions @xmath404 and @xmath405 are double planes , and we have @xmath406 and @xmath407 . letting @xmath408 and @xmath409 be the twistor lines as before , we define @xmath410 then since @xmath194 is a real map and @xmath398 and @xmath399 are real points , and since @xmath194 does not contract any divisor by proposition [ prop : phi4 ] , @xmath139 and @xmath411 are mapped birationally to the plane @xmath412 and @xmath140 and @xmath413 are mapped birationally to the plane @xmath414 . hence @xmath415 and @xmath416 are contained in the branch divisor @xmath10 in such a way that the restrictions @xmath417 and @xmath418 respectively contain @xmath415 and @xmath416 by multiplicity 2 . but as we know that @xmath10 is a cut of @xmath8 by a hyperquartic surface by proposition [ branchdivisor1 ] , @xmath415 and @xmath416 must be conics .
so we call these two _ double conics_. these double conics are analogous to the 4 conics contained in the coordinates tetrahedron in @xmath3 used in @xcite in the case of @xmath0 , but in the present case there are only 2 since there are only 2 reducible members of @xmath35 .
next we find other 3 double curves .
for this recall that our twistor space @xmath15 contains the surface @xmath14 constructed in section [ s:1 ] as a real member of @xmath35 .
take up the blowup @xmath48 which was concretely given when constructing @xmath14 , and let @xmath419 and @xmath420 , @xmath421 , be the exceptional curves of the blowup . here
we take indices such that @xmath422 for @xmath423 and @xmath424 .
next let @xmath425 be an orthonormal basis of @xmath426 determined by @xmath427 ; namely letting @xmath428 be the twistor fibration , @xmath429 in @xmath430
. then we have the following proposition which is technically significant for our purpose : [ prop : redivisor ] for each @xmath431 with @xmath432 ( not 4 ) , the linear systems @xmath433 and @xmath434 consist of a single element
. moreover , all these 6 divisors are irreducible . in this proof for simplicity
we write @xmath435 for @xmath436 .
it is enough to prove the claim for the system @xmath437 .
fix any @xmath431 with @xmath432 .
we have @xmath438 , which can be rewritten as @xmath439 the intersection number between @xmath440 and the above @xmath441 is easily computed to be @xmath330 . hence @xmath81 is a fixed component of @xmath442 .
further counting dimension , the remaining system @xmath443 consists of a single member , which is the strict transform of a @xmath444-curve passing through the 5 points @xmath445 and @xmath446 with @xmath447 and @xmath448 , where @xmath449 and @xmath450 .
thus we have @xmath451 .
on the other hand from riemann - roch formula and hitchin s vanishing theorem @xcite we deduce @xmath452 .
then by the standard exact sequence @xmath453 , we obtain the exact sequence @xmath454 .
as @xmath455 , we obtain @xmath456 .
hence we get @xmath457 . for the irreducibility ,
all the divisors @xmath458 and @xmath459 ( the irreducible components of reducible members of @xmath35 ) are degree 1 on @xmath15 .
moreover , it is not difficult to show that the chern classes of these divisors are given by _
the half of _ the following classes : @xmath460 then it is a easy to check that a sum of any two of these 4 classes ( allowing to choose the same one ) are not equal to @xmath461 , for any @xmath432 .
this implies the desired irreducibility .
( on the other hand , by , the systems @xmath462 are also non - empty , but both of them consist of a single _ reducible _ member . )
@xmath92 in the following for @xmath432 we denote by @xmath463 for the unique member of @xmath464 . then @xmath465 , and @xmath466 .
thus we obtained 3 reducible real members of the anticanonical system on @xmath15 .
we remark that from the proof of proposition [ prop : redivisor ] these 3 members originally come from the choice of 3 points on @xmath43 in the construction of @xmath14 at the beginning of section [ s:1 ] . by using these
we obtain a special basis of @xmath467 as follows : [ prop : basis4 ] for any @xmath431 with @xmath432 , let @xmath468 be an element such that @xmath469 .
then @xmath470 @xmath471 and any two among @xmath472 generate @xmath473 .
let @xmath189 be any real irreducible member and take @xmath474 with @xmath475 .
let @xmath476 be any element satisfying @xmath477 .
then @xmath478 is a basis of @xmath479 and @xmath480 is a basis of @xmath481 .
we consider the exact sequence @xmath482 appeared in the proof of proposition [ prop:2f ] . for proving the claim of the proposition
, it suffices to show that for any subset @xmath483 , the images of @xmath484 by the restriction map to @xmath14 generate @xmath485 . for this , the divisor @xmath486 is exactly @xmath487 , where @xmath54 is the unique anticanonical curve ( i.e.the cycle of 4 rational curves ) . on the other hand we have @xmath488 , and from the proof of proposition [ prop : redivisor ] we know the curves @xmath489 and @xmath490 in concrete forms , and it is not difficult to verify that the 3 bi - anticanonical curves @xmath491 are linearly independent . hence the 3 images generate @xmath485 .
@xmath92 in the sequel for obtaining nice coordinates , we choose a slightly different basis @xmath492 of @xmath493 as follows .
namely respecting the reducible members , we choose those satisfying @xmath494 and @xmath495 . by proposition [ prop : basis4 ] the collection @xmath496 is a basis of @xmath497 .
the target space of the anticanonical map @xmath498 is nothing but the dual projective space @xmath499 , and if we put @xmath500 then @xmath501 can be used as homogeneous coordinates on it .
( here we remark that there is no special reason to choose @xmath502 and @xmath503 .
any choice of two among @xmath504 leads to the same description below . )
as @xmath505 as above , @xmath9 is equipped with a real structure and by it is just the complex conjugation with respect to the above coordinates . in these
coordinates the scroll @xmath506 is explicitly defined by the equation @xmath507 and the ridge @xmath117 of @xmath8 is given by @xmath508 for each @xmath509 we define a hyperplane by @xmath510 .
obviously @xmath511 for @xmath512 , and @xmath513 for @xmath514 .
in particular @xmath515 and @xmath516 are double planes , @xmath517 is the sum of these 2 planes , and @xmath518 and @xmath519 are cones whose vertices are the points @xmath520 and @xmath521 respectively .
let @xmath522 be an element such that @xmath523 .
then by proposition [ prop : basis4 ] we can write @xmath524 for some @xmath525
. let @xmath526 . since @xmath527 clearly , @xmath528 is also a cone .
now the following proposition provides the promised 3 double curves on @xmath10 : [ prop : dq1 ] for @xmath529 , the intersection of the branch divisor @xmath10 with the cone @xmath530 is a double curve of @xmath10 .
let @xmath431 be any one of @xmath531 . by definition of the hyperplane @xmath532
, we have @xmath533 . since @xmath194 does not contract any divisor by proposition [ prop : phi4 ] , this implies @xmath534 .
as @xmath194 is degree 2 , this means that the restrictions @xmath535 and @xmath536 are birational over the cone @xmath537 .
( in particular , the non - real degree 2 divisors @xmath538 and @xmath539 are birational to a cone . )
therefore the curve @xmath540 is the ramification divisor of the restriction @xmath541 , and hence @xmath542 are branch divisors when restricted to @xmath543 .
this implies the claim of the proposition .
@xmath92 as in the proof , we use the letters @xmath544 and @xmath545 to mean the 3 double curves in the proposition .
then because we know that @xmath10 is a cut of @xmath8 by a quartic hypersurface , we have @xmath546 , where @xmath547 .
namely , @xmath544 and @xmath545 are intersection of the cone @xmath548 with a quadratic in @xmath549 . from this
it follows that these 3 curves are of degree 4 in @xmath9 .
so in the following we call these double curves _ double quartic curves_. from our choice of the coordinates , any double curves can be written as , as sets , @xmath550 since @xmath10 and @xmath532 are real , all @xmath551-s are real curves . in this section we show that there exists a quadratic hypersurface in @xmath9 which contains the double conics @xmath415 , @xmath416 and the double quartic curves @xmath544 and @xmath545 , and also show that such hyperquadric is unique up to the defining equation of the scroll @xmath8 .
first we make it clear how the 5 double curves of @xmath10 intersect each other . for this for each @xmath509
we denote by @xmath552 for the point whose coordinates are zero except @xmath553-component , and define some lines as follows : for each pair @xmath554 with @xmath555 and @xmath556 , define @xmath557 . since @xmath530 is a ( double ) plane for @xmath555 , this is a line .
thus we get 6 lines . if @xmath558 , these are coordinate lines and in figure [ fig : dc ] , @xmath559 .
( we do not write pictures of @xmath560 and @xmath561 because these are not coordinate lines .
but this is just a matter of a choice of coordinates and these two play the same role as other 4 lines . ) also , for the ridge @xmath117 we have @xmath562 ( the bold line on the left picture in figure [ fig : dc ] ) . , except @xmath545 .
both pictures lie on the same @xmath9 , and the common numbers represent the same point , all of which are ordinary double points of @xmath10 .
the left picture indicates all coordinate hyperplanes , 2-planes and lines , as well as double conics . in the right picture
the upper and lower halves are the cones @xmath563 and @xmath564 respectively , on which the double quartic curves @xmath565 and @xmath566 lie . ] then since the intersection of the branch divisor @xmath10 with the plane @xmath567 @xmath568 is the double conics @xmath551 and since the line @xmath569 ( @xmath570 ) is contained in this plane , the intersection @xmath571 consists of ( not 4 but ) 2 points , and @xmath572 . moreover , as @xmath573 , these 2 points can not be identical .
thus for each of the 6 lines @xmath569 , @xmath571 consists of 2 points .
( in figure [ fig : dc ] these points are represented by numbered points @xmath574 . ) on the other hand , we have @xmath575 .
hence as @xmath576 , we obtain @xmath577 for @xmath555 and @xmath556 .
but since @xmath578 is an intersection of @xmath579 with a quadric surface in @xmath580 , @xmath581 consists at most 2 points .
hence we have the coincidence @xmath582 for these @xmath583 and @xmath584 .
therefore we have @xmath585 for @xmath555 and @xmath556 . by a similar reason , the intersection @xmath586 also consists of 2 points , which are exactly @xmath587 . in figure
[ fig : dc ] these points are displayed as @xmath115 and @xmath116 .
on the other hand , for each pair @xmath588 with @xmath589 we define a plane @xmath590 by @xmath591 .
then since @xmath592 is contained in a quadric surface , and @xmath593 is a conic , @xmath594 ( @xmath589 ) consists of 4 points , and it coincides with @xmath595 . in figure
[ fig : dc ] , for the case @xmath596 , these are represented by numbered points @xmath597 .
( for avoiding confusion we do not write a picture for @xmath598 and @xmath599 .
the way how these curves intersect is completely analogous to that of @xmath566 and @xmath565 . )
we list all these intersections : * 2 points @xmath587 , which are exactly @xmath115 and @xmath116 , * @xmath600 points @xmath601 with @xmath555 and @xmath556 , * @xmath600 points @xmath602 , @xmath598 and @xmath599 . collecting these
, we obtain 26 points in total . since all the double curves are the image of curves in @xmath15 by a map which is degree 1 on these curves , these 26 points form 13 conjugate pairs .
among these 26 points , the 2 points @xmath603 are on the singular locus @xmath117 of @xmath8 , and other 24 points are ordinary double points of @xmath10 .
( in some sense these 24 points are analogous to the 12 ordinary double points of the branch quartic surface appeared in @xcite and @xcite . ) in section [ ss : sing ] , we will show that the 2 points @xmath115 and @xmath116 are @xmath18-singular points of @xmath10 . with these situation in hand ,
we next show the existence of a hyperquadric which contains all the double conics : [ prop : q ] there exists a real quadratic hypersurface in @xmath9 which contains all the 5 double curves @xmath551 , @xmath604 , and which is different from the scroll @xmath8 .
moreover , such a hyperquadric is unique in the following sense : if @xmath5 and @xmath605 are defining quadratic polynomials of two such hyperquadrics , then there exists @xmath606 with @xmath607 such that @xmath608 .
( note that since the scroll @xmath8 contains all the double curves , presence of this ambiguity is obvious from the beginning . ) as we have already seen , for @xmath514 the intersection @xmath530 is a quadratic cone in @xmath549 , and the double quartic curve @xmath551 belongs to @xmath609 .
in the above homogeneous coordinates the intersection @xmath610 is a plane defined by @xmath611 , and @xmath612 is a conic defined by @xmath613 , on which the 4 points @xmath614 lie .
conics on the plane passing through these 4 points form a pencil , which is invariant under the real structure .
choose any real one of such conics , and let @xmath615 be its defining equation with real coefficients , which is of course uniquely determined up to rescaling . among the above 26 points
there are exactly 8 points lying on @xmath616 ( which are the points @xmath617 and @xmath618 to @xmath600 in figure [ fig : dc ] ) , four of which are the above 4 points on @xmath610 ( the points @xmath597 in figure [ fig : dc ] ) .
any quadratic polynomial on @xmath616 whose restriction to @xmath610 equals @xmath619 is of the form @xmath620 imposing that the quadric @xmath621 passes the remaining 4 points ( @xmath622 ) , the coefficients @xmath623 and @xmath624 are uniquely determined ( without an ambiguity of rescaling ) , and they are real since the set of 4 points ( 3,4,7,8 ) is real . then since elements of @xmath625 which go through the 8 points is unique by dimension counting
, it follows that the quadratic surface @xmath621 automatically contains @xmath566 .
the situation is the same for @xmath626 , and let @xmath627 be the quadratic polynomial on @xmath626 with real coefficients , which is uniquely determined by @xmath619 and the 8 points ( @xmath628 and @xmath629 to @xmath600 ) on @xmath626
. then @xmath630 .
since @xmath631 ( @xmath632 ) , the pair @xmath633 is naturally regarded as a section of a line bundle @xmath634 . by the exact sequence @xmath635 we obtain that the section @xmath633 can be extended to a quadratic polynomial on @xmath9 , and that such polynomial is unique up to adding a constant multiple of @xmath636 .
explicitly such an extension has to be of the form @xmath637 where @xmath638 is an arbitrary constant .
of course , the hyperquadric satisfies @xmath639 .
then if we further impose that @xmath640 contains the point @xmath641 , then from a linear equation for @xmath642 is obtained , from which @xmath638 is uniquely determined .
@xmath638 is real since @xmath624 and @xmath643 are real . summarizing up
, we have obtained that once we fix a real quadratic polynomial @xmath615 , then there exists a unique real quadratic polynomial @xmath644 whose restriction to @xmath610 equals @xmath619 and which goes through the 9 points ( 1 to 8 and @xmath115 ) . in particular , if @xmath645 , then @xmath646 .
next we show that this polynomial @xmath5 ( which is uniquely determined from @xmath619 ) always contains the double curves @xmath551 for any @xmath604 .
it remains to show @xmath647 for @xmath648 . for @xmath555
this is immediate since @xmath640 already goes through 5 points on the conic @xmath551 , which means that it goes through the remaining 1 point @xmath116 .
@xmath649 is also immediate if we notice that as @xmath650 for @xmath421 , @xmath640 already goes through all the 12 points @xmath651 for @xmath421 on @xmath652 among the 26 points obtained in section [ ss : quadric ] and that since @xmath653 , eight points already and uniquely determine the quadric .
thus we have proved the existence of a quadratic polynomial @xmath5 satisfying @xmath647 for any @xmath604 . for the uniqueness in the sense of the proposition , let @xmath5 and @xmath605 be as stated in the proposition .
then since both @xmath654 and @xmath655 are real and belong to the pencil ( determined from the 4 points @xmath602 ) , there exists @xmath606 with @xmath656 such that @xmath657 .
further , the hyperquadric @xmath658 goes through the points @xmath659 to @xmath660 and @xmath115 at least , and therefore must belong to the ideal @xmath661 by the uniqueness which was already proved .
thus we get the required uniqueness . @xmath92 with the results in the previous 2 subsections , we are ready to provide the main result in this paper : [ thm : b ] let @xmath15 be any twistor space on @xmath1 containing the surface @xmath14 ( constructed in section 2 ) as a real member of @xmath35 .
let @xmath345 be the generically 2 to 1 morphism canonically obtained from the explicit birational transformations in section 3 , and @xmath10 the branch divisor of @xmath368 .
let @xmath644 be a defining equation of the hyperquadric containing all the 5 double curves , obtained in proposition [ prop : q ] .
then @xmath10 is an intersection of the scroll @xmath662 with the quartic hypersurface defined by the equation of the form @xmath663 where @xmath664 is a linear polynomial with real coefficients
. by proposition [ branchdivisor1 ] there exists a real hyperquartic such that the intersection with @xmath8 is the branch divisor @xmath10 .
let @xmath665 be any one of such hyperquartics and @xmath666 a defining equation of @xmath667 .
we note that @xmath667 is not unique in the sense that @xmath23 is determined only up to quartic polynomials in the ideal @xmath668 .
then for @xmath555 , the restriction of @xmath667 to a plane @xmath567 is the twice of the double conic @xmath551 ( see ) .
also , by the choice of @xmath5 , the restriction of the hyperquadric @xmath640 to the same plane is @xmath551 .
these two mean that there exists a constant @xmath669 such that @xmath670 belongs to the ideal @xmath671 .
namely there exist cubic polynomials @xmath672 and @xmath673 ( in @xmath674 satisfying @xmath675 taking the difference , we obtain @xmath676 if @xmath677 , substituting @xmath678 , the hyperquadric @xmath640 restricted to the plane @xmath679 is defined by @xmath680 .
this means that @xmath415 is reducible , which can not happen since it is the image of the twistor line @xmath325 ( see ) .
hence we obtain @xmath681 .
similarly , for @xmath514 , considering the restrictions of @xmath23 and @xmath682 to the cone @xmath683 , again by coincidence , there exist a constant @xmath669 , a cubic polynomial @xmath672 and a quadratic polynomial @xmath673 satisfying @xmath684 by with @xmath685 and we obtain @xmath686 from this we again obtain that the hyperquadric @xmath687 restricted to the plane @xmath679 is given by @xmath688 , contradicting the irreducibility of @xmath415 .
hence we obtain @xmath689 for @xmath514 .
thus we get @xmath690 .
if @xmath691 , by , we have @xmath692 . but this can not happen since this means @xmath693 , contradicting @xmath694 . hence @xmath695 .
so replacing @xmath5 with @xmath696 , we may assume that all the four @xmath697-s in and are one .
next in the expression we take @xmath672 and @xmath673 in such a way that @xmath673 does not contain @xmath698 . then since the right hand side of is zero
, @xmath699 follows .
hence @xmath700 , and we can write @xmath701 by a quadratic polynomial @xmath702 which does not contain @xmath698 . similarly , in the expression we take @xmath672 and @xmath673 in such a way that @xmath703 and @xmath704 do not belong to the ideal @xmath668 . then this time from for the case @xmath705 and @xmath706
, we obtain @xmath707 from the choice of @xmath703 and @xmath704 , it follows @xmath708 and @xmath709 .
hence we can put @xmath710 for some quadratic polynomial @xmath711 . from these ,
we obtain @xmath712 then since @xmath702 does not contain @xmath698 , from the latter equality we can readily deduce that if we write @xmath713 in a way that @xmath714 does not contain @xmath698 , then @xmath714 is a multiple of @xmath715 , so that @xmath716 for some @xmath717 . hence by we
obtain @xmath718 defining a linear polynomial @xmath719 by @xmath720 ( so that @xmath719 may contain @xmath698 ) and substituting into , we finally get @xmath721 thus we obtain @xmath722 . hence modulo quartic polynomials in the ideal @xmath668 , @xmath667 is defined by the equation of the form .
this completes a proof of the theorem .
@xmath92 from the quartic equation it is immediate to see that the intersection of the hyperquartic @xmath640 and the hyperplanes @xmath616 and @xmath626 are double quadric surfaces , and this is of course consistent with the fact that the restrictions @xmath723 and @xmath724 are double curves . on the other hand , for @xmath555 , in order to see that @xmath725 are also double curves from the equation , we just need to notice that , on the scroll @xmath8 , @xmath7 means @xmath726 . in comparison with the case of @xmath0 , appearance of the linear polynomial @xmath13 in our defining equation
might look strange at first sight . as the following proposition shows
, @xmath13 comes from the _ fifth _ double curve @xmath545 , which does not exist in the case of @xmath0 : up to non - zero constants , the linear polynomial @xmath13 in is exactly @xmath727 we have defined in . in other words , for the third double quartic curve , we have @xmath728 .
we will find all hyperplanes in @xmath9 ( which is the target space of the anticanonical map @xmath194 ) , which correspond to reducible members of the anticanonical system @xmath729 .
first as above for any hyperplane @xmath197 defined by the equation of the form @xmath730 , the corresponding member @xmath731 is clearly reducible .
( we are including a non - reduced case . ) also , @xmath732 and @xmath733 are reducible since @xmath734 an @xmath735 are double curves . by the same reason ,
if @xmath736 , the divisor @xmath737 is reducible .
then recalling that the double quartic curve @xmath545 is obtained as an image of the third reducible member of @xmath729 obtained in proposition [ prop : redivisor ] , in order to prove the claim of the proposition , it suffices to show that there exists no other hyperplane @xmath197 such that @xmath738 is reducible .
if @xmath197 is such a hyperplane , then either @xmath196 is reducible , or @xmath196 is irreducible ( i.e.a cone ) and @xmath739 is a double curve .
the former occurs exactly when @xmath197 is defined by the equation of the form @xmath730 .
so suppose the latter happens .
then recalling @xmath740 and @xmath741 is reduced by the assumption , @xmath739 can be a double curve only when @xmath742 is a double surface .
we show by algebraic mean that this happens only when @xmath197 is defined by one of the 4 factors of the left - hand side of .
if @xmath743 is a double surface , there exists a quadratic polynomial @xmath619 on @xmath197 such that @xmath744 ; namely @xmath745 if @xmath197 is defined by the equation of the form @xmath746 , then even after substitution the left - hand side of does not have monomial of the form @xmath747 for any @xmath748 .
therefore the coefficient of @xmath747 of the right - hand side of must be zero for any @xmath748 . by an elementary argument , it is possible to show this can happen only when @xmath749 . hence by @xmath197 is equal to one of @xmath750 and @xmath751 . by symmetry of the equation ,
if @xmath197 is defined by a equation of the form @xmath752 , then is possible only when @xmath197 is one of @xmath750 and @xmath751 .
if @xmath197 is of the form @xmath753 , then the left - hand side of can not contain monomials of the form @xmath754 , @xmath755 , @xmath756 ( for any @xmath431 ) , and @xmath757 , @xmath758 , @xmath759 .
therefore the coefficients of these monomials of the right - hand side of must vanish . from these , again by an elementary argument it is possible to show that @xmath749 .
hence again @xmath197 has to be one of @xmath750 and @xmath751 .
the remaining 2 cases immediately follow from symmetry of the equation .
thus we have shown that @xmath742 is a double surface only when @xmath197 is one of @xmath750 and @xmath751 .
@xmath92 we again emphasize that the role of the 3 double quartic curves is symmetric , and any choice of two leads to the equation of the form .
_ one may wonder whether the linear polynomial @xmath13 can be taken as one of the homogenous coordinates on @xmath9 .
at least in generic situation this is possible , but if we do so , we lose simplicity of the defining equation of the scroll @xmath8 , and it makes more difficult the counting the number of effective parameters in defining equations of @xmath760 which will be done in section [ ss : moduli ] . _ in this subsection by using the quartic equation obtained in the previous subsection we determine the number of singularities of the branch divisor of the double covering . similarly to the method by kreussler @xcite and kreussler - kurke @xcite , we resort to _ topology _ ; more precisely we compute the euler number of the relevant spaces to determine the number of singularities . though we require much more complicated computation than the case of @xmath0 , we do it since this result is crucial for determining the dimension of the moduli space of the present twistor spaces . as in the proof of proposition
[ prop : isolated ] let @xmath761 be the stein factorization of the degree 2 morphism @xmath345 .
we already know that @xmath387 contracts finitely many curves , whose images are singular points of the branch divisor @xmath10 . if we put @xmath762 and @xmath763 , @xmath389 also has ordinary double points along @xmath764 .
all other singularities of @xmath389 are lying on singularities of the branch divisor @xmath10 . among these singularities
we already know that there are 26 singularities listed in section [ ss : quadric ] , and 24 points among them are ordinary double points .
the 2 points excluded here are exactly the points @xmath587 , for which we still denote by @xmath115 and @xmath116 ( see figure [ fig : dc ] again ) .
we begin with determining the type of singularities of these 2 points : [ prop : a3 ] at the 2 points @xmath115 and @xmath116 , the branch divisor @xmath10 has @xmath18-singularities .
recall that @xmath765 , and @xmath766 .
as above let @xmath115 be any one of the 2 points and we work in a neighborhood of @xmath115 .
we put @xmath767 and @xmath768 .
then by transversality for the intersection of @xmath5 and @xmath117 , we can use @xmath769 as coordinates in a neighborhood of @xmath115 in @xmath770 , and noticing @xmath771 at @xmath115 , we may suppose that the hyperquartic is defined by a very simple equation , @xmath772 .
since @xmath8 is defined by @xmath773 , we deduce that @xmath115 is an a@xmath774-singular point of the surface @xmath10 . by reality , @xmath116 is also an a@xmath774-singular point .
@xmath92 next , as the transformation from @xmath15 to @xmath344 is explicit , it is easy to show the following : [ euler4 ] for the variety @xmath344 we have @xmath775 .
since @xmath15 is a twistor space on @xmath1 , we have @xmath776 . because the blowup @xmath777 replaces two disjoint @xmath316-s by two @xmath36-s
, we have @xmath778 .
then since a flop does not change the euler number we obtain @xmath779 . finally looking figure [ fig : trans ] ( d )
, the exceptional divisor @xmath148 of the contraction @xmath337 has euler number 8 , and the image @xmath780 of the exceptional divisor has euler number 2 .
hence we obtain @xmath781 .
@xmath92 the next result means that the 26 points that we have already found are _ not _ all singularities of the branch divisor @xmath10 , but in the generic situation @xmath10 has extra 6 ordinary double points : [ thm:6sing ] let @xmath782 be the set of all singular points of @xmath10 which are different from the 26 singular points listed in section [ ss : quadric ] .
let @xmath783 be the milnor number of the singular point @xmath784 , and put @xmath785 for the exceptional curve of @xmath387 over the point @xmath784 .
( of course we do not assume irreducibility of @xmath786 . )
then we have the relation @xmath787 in particular , if all the singularities are ordinary double points , we have @xmath788 .
first since @xmath789 replaces each of the 24 ordinary double points by smooth @xmath316 and also replaces the singular point @xmath784 by the curve @xmath786 for @xmath790 , we obtain @xmath791 on the other hand by the double covering @xmath792 we have @xmath793 further as @xmath8 is obtained from the @xmath198-bundle @xmath203 over @xmath316 , and it replaces @xmath36 by @xmath794 , we have @xmath795 .
hence we have @xmath796 , giving @xmath797 next for computing @xmath798 let @xmath105 be a general member of the system @xmath799 . then since the scroll @xmath8 has ordinary double points along the line @xmath117 , the divisor @xmath105 has ordinary double points at the 4 points @xmath800 . as before
let @xmath201 be the blowup at @xmath117 , and let @xmath801 be the strict transform of @xmath105 . by bertini s theorem we may suppose that @xmath801 is non - singular .
we shall compute @xmath802 .
as before write @xmath803 . from the standard relationship of the total chern class @xmath804 and adjunction formula ,
we readily obtain @xmath805 where the dot means the product in @xmath806 . as in the proof of proposition [ branchdivisor1 ]
let @xmath807 , so that @xmath808
. then by using @xmath809 and the adjunction formula applied to a fiber of @xmath208 , we readily obtain @xmath810 .
further , in the cohomology ring of @xmath203 , we have @xmath811 , @xmath812 , @xmath813 and @xmath814 . furthermore by recalling @xmath815 ( see ) , we obtain @xmath816 . from these
we readily get @xmath817 . for computing @xmath818 , as generators of @xmath819
we take any element @xmath820 and @xmath821 , viewed as submanifolds in @xmath203 , and put @xmath822 . from the exact sequence associated to the inclusion @xmath823 , we immediately obtain @xmath824 .
then since @xmath825 and @xmath826 , we obtain @xmath827 . on the other hand , from the inclusion @xmath828
we readily obtain @xmath829 .
further , in the cohomology ring of @xmath203 we have @xmath830 and @xmath831 .
therefore by restricting to @xmath202 and @xmath273 respectively , we get @xmath832 and @xmath833 .
. therefore we obtain @xmath835 which is readily computed to be @xmath836 . hence from we obtain @xmath837 . as @xmath838 contracts four @xmath316-s to 4 points
, we have @xmath839
. then @xmath105 is obtained from the actual branch divisor @xmath10 by ( a ) smoothing the @xmath840 nodes , ( b ) smoothing @xmath841 singular points @xmath842 , and ( c ) deforming each of the two @xmath18-singularities ( which is exactly @xmath843 ) to two @xmath844 singularities . adding the milnor number of the singularities for the cases ( a ) and ( b ) , and also taking the difference of the milnor number of @xmath18-singularity and two @xmath844-singularities into account
, we obtain @xmath845 and hence @xmath846 . substituting this into and using proposition [ euler4 ] , we obtain .
@xmath92 _
if the blown - up 8 points on @xmath36 are arranged as in figure [ fig : square ] is in a general position ( in certain precise sense ) , then @xmath45 and @xmath89 are all curves that are contracted to points by the bi - anticanonical map .
but if the 8 points are in a special position ( in certain precise sense ) , then the map contracts extra curves .
it is not difficult to classify all positions which yield this situation .
the appearance of this kind of curves is exactly the reason why the anticanonical map of the twistor spaces contracts some rational curves which can not be found from the equation of the branch divisor . _
in this subsection we compute the dimension of the moduli space of our twistor spaces by counting the number of effective parameters , and verify that it agrees with the dimension of the cohomology group which is relevant to the present case . for the former purpose , we recall from section [ s : anticanonical ] that @xmath15 canonically determines a birational model @xmath344 and the degree 2 morphism @xmath345 , and from section [ ss : defeq ] the branch divisor of @xmath368 is an intersection of @xmath8 with the quartic surface defined by @xmath847 where @xmath13 and @xmath5 are linear and quadratic polynomials with real coefficients respectively . in this subsection we denote this quartic hypersurface by @xmath848 .
since the quartic hypersurface @xmath848 uniquely determines the double cover via the natural quadratic map @xmath849 ( which takes squares ) , up to small resolutions , @xmath15 is uniquely determined by the quartic hypersurface .
of course , @xmath13 has 5 coefficients and @xmath5 has 15 coefficients , so the equation contains @xmath850 parameters .
further it is elementary to see that two pairs @xmath851 and @xmath852 of linear and quadratic polynomials ( over @xmath853 ) determine the same hyperquartic surface if and only if @xmath854 for some @xmath855 .
this decreases the number of parameters by one . on the other hand , projective transformations which preserve @xmath8 and which preserves
_ the form of _ the equation have to be @xmath856 for some @xmath857 .
( here we are only considering transformations which are homotopic to the identity . )
if a pair @xmath852 is obtained from a pair @xmath851 by one of these transformations , then the intersections @xmath858 and @xmath859 are mutually biholomorphic , so that they define mutually isomorphic double cover . but taking an effect of the above equivalence @xmath860 into account , we can suppose @xmath861 and therefore these projective transformations decrease the number of parameters by 3 .
thus up to now the number of parameters is @xmath862 .
however what we have to consider is not the hyperquartics themselves but the intersection with @xmath8 ; namely if @xmath863 for some @xmath855 , then we have the coincidence @xmath864 .
clearly these transformations are not included in the above projective transformations , so they drop the dimension by one .
thus we have obtained that the number of effective parameters in the quadratic hypersurface is 15 .
finally , by theorem [ thm:6sing ] , the pair @xmath851 must satisfy the constraint that @xmath865 has extra 6 ordinary double points in general , which decreases the number of parameters by 6 . therefore we conclude that _ the space of isomorphic classes of the divisors of the form @xmath865 , which can be the branch divisor for the twistor spaces under consideration , is @xmath866-dimensional_. next we compute the dimension of the moduli space of our twistor spaces by determining the dimension of the first cohomology group of an appropriate subsheaf of the tangent sheaf .
we begin with a computation for the full moduli space .
[ prop:13dim ] let @xmath15 be a twistor space on @xmath1 which contains the surface @xmath14 constructed in section [ s:1 ] as a real member of @xmath35 .
then we have @xmath867 for @xmath868 and @xmath869 .
as computed in @xcite , for any twistor space on @xmath20 , by the riemann - roch formula , we have @xmath870 . also , since @xmath15 is moishezon and @xmath35 has an irreducible member , we have @xmath871 by @xcite .
further we always have @xmath872 .
hence it suffices to show @xmath873 .
let @xmath874 be the identity component of the holomorphic automorphism group of @xmath15 .
then the real part @xmath875 is naturally identified with the identity component of conformal automorphism group of the self - dual structure . also , since @xmath35 has just 2 irreducible components , @xmath875 acts on @xmath876 .
hence as the twistor projections @xmath877 and @xmath878 are of degree 1 , @xmath875 acts effectively on @xmath876 .
furthermore , the degree 1 divisor @xmath139 and @xmath411 are obtained from @xmath198 by blowing - up 4 points , exactly 3 of which are collinear . from this
it readily follows that the subgroup of @xmath879 which consists of automorphism preserving the twistor line @xmath325 is 0-dimensional .
hence so is @xmath880 .
thus @xmath874 can not be of positive dimension .
therefore @xmath873 .
@xmath92 from the proposition , the real part of the kuranishi family of our twistor space @xmath15 is 13-dimensional . of course , generic twistor spaces on @xmath1 is algebraic dimension 1 and generic members of the kuranishi family have the same property . in order to restrict to the moishezon twistor spaces under consideration , we show the following . [
prop : dfmpair ] let @xmath15 and @xmath14 be as in proposition [ prop:13dim ] and @xmath43 and @xmath81 the @xmath881-curves on @xmath14 .
then deformation theory of the pair @xmath882 is unobstructed and its kuranishi family is 9-dimensional .
further , for all sufficiently small deformations preserving the real structure in the kuranishi family , the twistor spaces contain a non - singular surface constructed in section [ s:1 ] as a real member of @xmath35 .
of course , the last property means that the deformed spaces are still the moishezon twistor spaces under consideration .
_ proof of proposition [ prop : dfmpair ] .
_ let @xmath883 be the subsheaf of @xmath884 whose germs are vector fields that are tangents to @xmath43 and @xmath81 .
for the former claim on the kuranishi family , it suffices to show that @xmath885 and @xmath886 .
recalling @xmath887 , we obtain the standard exact sequence @xmath888 which induces an exact sequence @xmath889 hence with the aid of proposition [ prop:13dim ] we have only to show @xmath890 . for this
we first deduce from duality and rationality that @xmath891 , which implies , from the exact sequence @xmath892 , that @xmath893 moreover , noting @xmath894 , we have an exact sequence @xmath895 for the last term we have @xmath896 , and it is easy to see @xmath897 .
hence for the middle term we obtain @xmath898 .
then by the exact sequence @xmath899 and @xmath900 @xcite , we finally obtain @xmath901 , as claimed .
for the latter claim about the existence of the surface in the deformed space , let @xmath902 be any one of the deformed twistor space which is sufficiently close to the original @xmath15 , and @xmath903 the curves corresponding to the original curves @xmath43 and @xmath81 .
let @xmath904 be the fundamental line bundle on @xmath902 .
then as @xmath133 , we may suppose @xmath905 by upper - semicontinuity of dimensions of cohomology groups under deformations and the riemann - roch formula @xmath906 we also have an invariance @xmath907 , and the latter is equal to @xmath908 , and therefore @xmath909 .
this means that @xmath910 and @xmath911 are base curves of the pencil @xmath912 .
let @xmath913 be any real irreducible member . through the kuranishi family
, this surface can be regarded as a small deformation of some real irreducible @xmath189 , which means that @xmath914 is obtained from @xmath36 by moving the blownup 8 points from the original positions ( indicated as in figure [ fig : square ] ) .
but since @xmath914 contains the curves @xmath910 and @xmath911 as @xmath881-curves , the property that 3 points belong to a @xmath42-curve must be preserved .
this means that the structure of @xmath914 is the same as that of the original @xmath14 , and we are done .
@xmath92 in this subsection , by using a theorem of pedersen - poon about structure of real irreducible members of @xmath35 , we show that the present twistor spaces are in a sense generic among all moishezon twistor spaces on @xmath1 .
we first recall the theorem of pedersen - poon @xcite in a precise form : [ prop : pp94 ] let @xmath15 be a twistor space on @xmath20 and @xmath189 a real irreducible member .
then @xmath14 is non - singular with @xmath915 , and the set of twistor lines lying on @xmath14 is exactly the real part of a real pencil whose self - intersection number is zero .
moreover there is a birational morphism @xmath916 preserving the real structure , such that the twistor lines are mapped to @xmath42-curves .
thus @xmath14 is always obtained from @xmath36 by blowing up @xmath917 points , where some of the points might be infinitely near in general . as is well - known the position of the blowing up points has a strong effect on algebraic structure of twistor spaces .
for example , if a twistor space @xmath15 contains @xmath14 that is obtained from the @xmath917 points lying on an irreducible @xmath444-curve , then it follows @xmath918 , and detailed structure of such twistor spaces is investigated by campana - kreussler @xcite .
then in terms of the configuration of the blowing up points , the genericity of our twistor spaces refers the following property : [ prop : density ] let @xmath15 be a moishezon twistor space on @xmath1 which is not of campana - kreussler type .
suppose that there exists a real irreducible member @xmath189 such that the images of the 8 exceptional curves of the blowing - down @xmath916 in proposition [ prop : pp94 ] can be taken as distinct points .
then the configuration of the 8 points falls into exactly one of figure [ fig : square2 ] . for moishezon twistor spaces , except for the campana - kreussler s case .
] for the proof , we first show the following .
[ prop : gen1 ] if @xmath15 is a twistor space on @xmath1 which satisfies @xmath918 , then @xmath15 is either a campana - kreussler twistor space , or otherwise non - moishezon .
let @xmath189 be a real irreducible member , which is necessarily non - singular as above . by the assumption
, @xmath14 satisfies @xmath919 .
let @xmath48 be the birational morphism fulfilling the properties of proposition [ prop : pp94 ] .
( the images of the exceptional curves of @xmath920 can be infinitely near . ) since @xmath920 is a composition of blowdowns , by the canonical bundle formula for blowups , the image of the pencil @xmath240 by @xmath920 necessarily has to be a pencil of anticanonical curves on @xmath36 ; namely a pencil of @xmath921-curves .
let @xmath922 be this pencil on @xmath36 .
then again by the canonical bundle formula all the images of the exceptional curves of @xmath920 must be contained in the base locus of @xmath923 .
suppose that general members of the pencil @xmath923 are irreducible .
then the pencil @xmath240 does not have a fixed component .
if this pencil has a base point , by taking a sequence of blowups @xmath924 , we obtain a morphism @xmath925 , which is , again by the canonical bundle formula , necessarily the anticanonical map on @xmath159 .
therefore , the morphism @xmath925 must be an elliptic fibration .
but then by the canonical bundle formula for elliptic surfaces , we obtain @xmath926 .
since @xmath927 , this means that @xmath159 and @xmath14 are biholomorphic .
hence the pencil @xmath928 is base point free , and the anticanonical map induces an elliptic fibration @xmath929 .
this implies the anti - kodaira dimension of @xmath14 is one , which means that @xmath15 is non - moishezon .
so in the sequel we suppose that general members of the pencil @xmath923 on @xmath36 are reducible .
then if @xmath923 does not have a fixed component , we have @xmath930 , which contradicts our assumption .
hence @xmath923 has a fixed component .
let @xmath931 be any one of its irreducible components .
then among the image points of the exceptional curves of @xmath920 , there exists at least 1 point on @xmath931 because otherwise @xmath931 is not a fixed component .
suppose that @xmath932 .
then @xmath933 by the induced real structure on @xmath36 , and @xmath934 is also a fixed component of @xmath923 .
hence the movable part of @xmath923 must be a free 1-dimensional subsystem of @xmath935 , or the system @xmath936 itself with another fixed @xmath42-curve @xmath937 .
but the former can not occur because general members of the movable part of the 1-dimensional subsystem would be reducible by freeness and both components actually move , so that all the images of the exceptional curves of @xmath920 have to be contained in @xmath938 , which means @xmath939 .
so suppose the latter is the case .
then the fixed @xmath42-curve @xmath937 can not be real , since if so , we would have @xmath940 for some twistor line @xmath941 by the property of @xmath920 , whereas on @xmath937 there is at least one point among the images of the exceptional curves of @xmath920 , which means @xmath942 on @xmath14 .
hence @xmath943 .
but this is impossible since @xmath944 , which is clearly @xmath921-curves , would be fixed components of the pencil @xmath923 of @xmath921-curves .
thus we obtained @xmath945 ; namely @xmath923 does not have a @xmath44-curve as a fixed component .
also if a fixed component @xmath931 is a @xmath42-curve , then it can not be real by the same reason .
hence the movable part of @xmath923 is a free 1-dimensional subsystem of @xmath946 , or the system @xmath947 with another fixed component @xmath948 .
but the former implies @xmath949 by the same argument as above , and the latter can not occur since this time there is no real @xmath44-curve .
thus the fixed component @xmath931 of @xmath923 can not be a @xmath42-curve .
further we have @xmath950 , since any @xmath951-curve is not real and hence @xmath933 has to be also a base curve , which contradicts that @xmath923 is a pencil .
similarly we have @xmath952 by the real structure .
hence @xmath931 must be in the remaining possibility , @xmath953 . in this case
@xmath954 means that all the images of the exceptional curves of @xmath920 belong to @xmath931 .
this implies that the structure of @xmath14 is exactly as in the case of campana - kreussler , and we are done .
@xmath92 _
proof of proposition [ prop : density]_. first by a result by kreussler ( * ? ? ?
* theorem 6.5 ) , on @xmath20 with @xmath955 we always have @xmath956 and the equality holds iff @xmath15 is a lebrun twistor space @xcite .
suppose that @xmath15 is a lebrun twistor space .
then it is well - known that a configuration of @xmath917 points for generic real irreducible member @xmath189 is as in ( i ) of figure [ fig : square2 ] .
so suppose that @xmath15 is a moishezon twistor space on @xmath1 which is different from lebrun s nor campana - kreussler s , and let @xmath14 be a real irreducible member of @xmath35 such that the @xmath917 points on @xmath36 are distinct . by proposition [ prop : gen1 ]
we have @xmath133 .
this means @xmath957 .
let @xmath54 be the unique anticanonical curve , @xmath958 the birational morphism as in proposition [ prop : pp94 ] whose images of the exceptional curves are distinct .
we put @xmath959 , which is necessarily a real @xmath921-curve .
all the 8 points are on @xmath931 .
if @xmath931 is irreducible , @xmath931 must be a non - singular elliptic curve by using the real structure , and from this we readily see that @xmath960 for all @xmath961 , which means that @xmath15 is non - moishezon .
hence @xmath931 is reducible . taking the form of the induced real structure on @xmath962 into account
, we can easily show that the decomposition of @xmath931 into irreducible components is one of the following 3 types : ( a ) @xmath963 , ( b ) @xmath964 , both components being irreducible , or ( c ) @xmath965 , both components being irreducible .
we note that since @xmath966 , on any of these components , there exists at least 1 point among the 8 points . repeating an argument in the last part of the proof of proposition [ prop : gen1 ]
, we deduce that ( c ) can not happen under our assumption . if @xmath931 has a multiple component , since there exists no real @xmath44-curve , it must be a real @xmath42-curve . but this can not happen since as remarked above among the 8 points there is at least one point on any irreducible component of @xmath931 , contradicting the family of twistor lines on @xmath14 . hence in both cases
( a ) and ( b ) @xmath931 has no multiple component .
next we show that in the case ( a ) there is an irreducible component of @xmath931 on which precisely @xmath967 points among the 8 points lie . if not , then because we are excluding lebrun twistor spaces , on each of the 4 irreducible components exactly 2 points are lying among the 8 points . in this case , the restriction @xmath968|_{c}$ ] belongs to pic@xmath969 , which again implies @xmath970 for any @xmath961 as in the above case .
this implies that @xmath15 is not moishezon .
hence the component actually exists .
next we prove that if @xmath931 is in the case ( b ) , there exists another birational morphism @xmath971 preserving the real structure such that @xmath972 falls into the case ( a ) , and such that the images of twistor lines in @xmath14 are @xmath42-curves . for this
we write @xmath973 with @xmath43 and @xmath81 being irreducible @xmath951-curves .
if exactly 4 points belong to @xmath43 , then the remaining 4 points belong to @xmath81 , and also no point coincides with the 2 points @xmath974 . then by a similar reason for the case ( a ) , this implies that @xmath15 is not moishezon .
so the 2 points @xmath974 are included in the 8 points .
therefore @xmath920 factors as @xmath975 where the latter arrow is the blowup at @xmath974 .
we obtain 6 points on @xmath976 as the images of the exceptional curves of @xmath977 .
these 6 points are not on the exceptional curves of @xmath978 by the assumption that the 8 points are distinct .
this implies that the unique anticanonical curve @xmath54 on @xmath14 is a cycle of 4 rational curves , whose self - intersection numbers are @xmath979 .
then for another blowdown @xmath980 in the proposition , it is enough to choose a blowingdown @xmath981 which does not contract the 2 exceptional curves of the original @xmath982 , and composite it with the morphism @xmath977 .
thus we obtained another @xmath980 as claimed , and we can neglect the case ( b ) . hence @xmath931 can be supposed to be in the case ( a ) and that there is at least one component on which exactly 3 points among 8 points lie .
this directly means the 8 points have to be put on @xmath931 arranged as in ( ii ) , ( iii ) or ( iv ) , as claimed .
@xmath92 needless to say , the case ( ii ) of proposition [ prop : density ] is exactly the situation we have investigated in this paper . since it is clear that all other 3 cases ( ( i ) , ( iii ) and ( iv ) ) can be obtained as small deformations of the case ( ii ) , it would be reasonable to say that among the surface @xmath14 obtained from the 8 points arranged as in ( i)(iv ) , the case ( ii ) is most generic . by deformation theory including a co - stability theorem of horikawa @xcite , the same is true for the twistor spaces containing these surfaces .
namely any twistor spaces on @xmath1 which has @xmath14 obtained from ( i ) , ( iii ) and ( iv ) as a real member of @xmath35 can be obtained as a limit of the twistor spaces investigated in this paper . in particular , the present twistor spaces can be obtained as a small deformation of a lebrun twistor space , and this proves the existence of our twistor spaces .
we also remark that by using the horikawa s theorem , it is possible to show that the present twistor spaces can also be obtained as a small deformation of the twistor spaces studied in @xcite ( on @xmath1 , of course ) .
these are the reason why we call the present twistor spaces to be generic .
finally we remark that a converse of proposition [ prop : density ] also holds .
namely if a twistor space @xmath15 on @xmath1 has real irreducible @xmath189 which is obtained from the 8 points in the case ( ii ) , ( iii ) , or ( iv ) , then @xmath133 and @xmath15 is moishezon .
concerning structure of these twistor spaces , the case ( iii ) can be regarded as a mild degeneration of the present twistor spaces , in the sense that the twistor space still has a double covering structure over the scroll @xmath8 by the anticanonical system .
these twistor spaces ( having @xmath14 obtained from the configuration ( iii ) ) are analogous to a double solid twistor spaces on @xmath0 of a degenerate form found by kreussler - kurke ( * ? ? ?
* , case ( b ) ) .
here we mention that there is _ one more _ another type of twistor spaces on @xmath1 of a degenerate form which do not admit @xmath19-action .
we will study these 2 kinds of double solid twistor spaces on @xmath1 in a separate paper . on the other hand , although moishezon , it turns out that the remaining case ( iv ) does not have structure of double solids , because the anticanonical map of the twistor spaces becomes birational .
so they are rather similar to the twistor spaces of joyce metrics on @xmath1 of non - lebrun type @xcite .
but contrary to the joyce s case , explicit realization of the anticanonical model seems difficult .
we recall from section [ ss : modify ] that the singular variety @xmath344 is obtained from the twistor space @xmath15 via non - singular spaces @xmath984 and @xmath310 , and the transformations therein are standard until getting @xmath310 . on the other hand , the map @xmath337 contracts the reducible connected divisor @xmath985 to a reducible connected curve @xmath780 , and there we used fujiki s contraction theorem .
one would wish to find such a birational morphism through the usual procedure of blowups with non - singular center . in this subsection
, we explicitly see that an embedded blowup at and a small resolution provide the desired situation and point out that the process is in a sense a singular version of the hironaka s construction of non - projective moishezon 3-folds .
for this we consider the double covering of @xmath8 with branch @xmath10 which is the intersection with the quartic hypersurface .
recall that @xmath765 , and @xmath586 consists of two points @xmath986 . as before let @xmath115 be any one of the two points and we work in a neighborhood of @xmath115 as the situation around @xmath116 can see by just taking the image under the real structure . as in the proof of proposition
[ prop : a3 ] , putting @xmath987 and @xmath988 , we can use @xmath769 as coordinates in a neighborhood of @xmath115 in @xmath770 , and we may suppose that the hyperquartic is defined by @xmath772 , while @xmath8 is defined by @xmath773 .
next for studying the structure of the double covering , we introduce another coordinate @xmath989 over the neighborhood of @xmath115 , so that the double cover is defined by @xmath990 ( this is the equation of @xmath344 around the points @xmath343 we promised in section [ ss : modify ] . )
let @xmath991 be this double covering and @xmath992 the projection .
( of course this is valid only in a neighborhood of @xmath115 . )
the singular locus of @xmath991 is @xmath993 which is a union of 2 lines , and @xmath991 has a@xmath994-singularities along these lines minus the origin .
we note that substituting @xmath995 to @xmath773 , we obtain that @xmath991 contains the following distinguished 4 surfaces @xmath996 then obviously we have @xmath997 .
so if we let @xmath998 to be the blowup at the plane @xmath999 and @xmath1000 to mean the strict transform of @xmath991 , then @xmath1001 is an embedded blowup at sing@xmath991 .
then by concrete computations using coordinates it is not difficult to see that the exceptional locus of @xmath1001 consists of 2 irreducible divisors which are over the 2 lines respectively , that the inverse image of the origin is a non - singular rational curve , and that the singularities of @xmath1000 consists of 2 points lying on this rational curve , both of which are ordinary double points .
further , by the effect of the blowup , the pair of planes , both of which contain the same line @xmath1002 are separated by one of the exceptional divisors , and the same for another pair .
this way we get the situation of figure [ fig : trans ] , ( e ) .
then an appropriate small resolution ( which is obvious from the figure ) gives the desired space @xmath310 .
as above the center of the blowup is a reducible curve whose fundamental group is @xmath276 .
thus , together with an inspection of the choice of the small resolution displayed in figure , it would be possible to say that the transformation from @xmath344 to @xmath310 is a singular version of hironaka s well - known example of non - projective moishezon 3-folds @xcite in the sense that the center of the blowup in the present situation is a singular locus of the 3-fold .
m. atiyah , n. hitchin , i. singer , _ self - duality in four - dimensional riemannian geometry _ , proc .
london , ser . a * 362 * ( 1978 ) 425461 .
f. campana , _ the class @xmath1003 is not stable by small deformations _ , math
. ann . * 229 * ( 1991 ) 1930 .
f. campana , b. kreuler , _ a conic bundle description of moishezon twistor spaces without effective divisor of degree one _ , math . z.
* 229 * ( 1998 ) 137162 .
a. fujiki , _ on the blowing down of analytic spaces _ , publ .
rims , kyoto university * 19 * ( 1975 ) 473507 . h. hironaka , _ on the theory of birational blowing - up _ , thesis , 1960 .
n. hitchin , _ linear field equations on self - dual spaces _ , proc .
london ser .
a * 370 * ( 1980 ) 173 - 191 .
n. honda , _ on a construction of the twistor spaces of joyce metrics _ , j. algebraic geom .
* 17 * ( 2008 ) , 709 - 750 .
( available as math.dg/0603242 and math.dg/0604306 . )
b. kreuler , _ on the algebraic dimension of twistor spaces over the connected sum of four complex projective planes _ , geom .
dedicata * 71 * ( 1998 ) 263 - 285 . b. kreussler and h. kurke , _ twistor spaces over the connected sum of 3 projective planes
. _ compositio math . * 82*:2555 , 1992 . c. lebrun , _ explicit self - dual metrics on @xmath1004 _ , j. differential geom .
* 34 * ( 1991 ) 223253 .
c. lebrun , _ twistors , khler manifolds , and bimeromorphic geometry i _ , j. amer .
soc . * 5 * ( 1992 ) 289316 .
h. pedersen , y. s. poon , _ self - duality and differentiable structures on the connected sum of complex projective planes _
* 121 * ( 1994 ) 859 - 864 .
y. s. poon , _ on the algebraic structure of twistor spaces _
, j. differential geom .
* 36 * ( 1992 ) , 451491 . | in this paper we investigate a family of moishezon twistor spaces on the connected sum of 4 complex projective planes , which can be regarded as a direct generalization of the twistor spaces on @xmath0 of double solid type studied by poon and kreussler - kurke .
these twistor spaces have a natural structure of double covering over a scroll of 2-planes over a conic .
we determine the defining equations of the branch divisors in an explicit form , which are very similar to the case of @xmath0 . using these explicit description
we compute the dimension of the moduli spaces of these twistor spaces .
also we observe that similarly to the case of @xmath0 , these twistor spaces can also be considered as generic moishezon twistor spaces on @xmath1 .
we obtain these results by analyzing the anticanonical map of the twistor spaces in detail , which enables us to give an explicit construction of the twistor spaces , up to small resolutions . |
the standard model @xcite of the electroweak interactions predicts a mixing of the quark mass eigenstates with the weak interaction ones .
this mixing is described by the cabibbo - kobayashi - maskawa@xcite ( ckm ) matrix .
four real parameters describe this @xmath7 unitary matrix @xcite : @xmath8 as @xmath9 , @xmath10 and @xmath11 are of order unity , and @xmath12 is chosen as the sine of the cabibbo angle , this parametrisation shows immediately the hierarchy of the couplings of the quarks in the charged current part of the standard model lagrangian .
moreover in this parametrisation the parameter @xmath11 is the complex phase of the matrix and is thus directly related to the known violation of the cp symmetry produced by the weak interactions .
the measurement of the parameters of the ckm matrix is thus of fundamental importance for both the precision description of the weak interaction of quarks and the investigation of the mechanism of cp violation .
the parameters @xmath9 and @xmath12 are known with an accuracy of a few percent and the determination of @xmath10 and @xmath11 is the subject of this letter .
a large number of physical processes can be parametrised in terms of the values of the elements of the ckm matrix , together with other parameters of theoretical and experimental origin .
four of them show good sensitivity for the indirect determination of @xmath10 and @xmath11 and are discussed in what follows .
a fit based on this information has been performed , as suggested in @xcite , and its results are presented below . as it is well known
the measurement of @xmath10 and @xmath11 is equivalent to the determination of the only unknown vertex and the angles of a triangle in the @xmath13 plane whose other two vertices are in ( 0,0 ) and ( 1,0 ) .
figure 1 shows this triangle , called the unitarity triangle .
the value of the sine of the cabibbo angle is known with a good accuracy @xcite as : @xmath14 the parameter @xmath9 depends on @xmath12 and on the ckm matrix element @xmath15 . using the value @xcite : @xmath16
it can be extracted : @xmath17 the mass eigenstates of the neutral kaons can be written as @xmath18 and @xmath19 .
the relation @xmath20 implies the violation of cp that , in the wu - yang phase convention @xcite , is described by the parameter @xmath21 defined as : @xmath22 the precise measurements of the @xmath23 and @xmath24 decay rates imply @xcite : @xmath25 the relation of @xmath26 to the ckm matrix parameters is @xcite : @xmath27 .
\label{equation : ek}\ ] ] the functions @xmath28 and @xmath29 of the variables @xmath30 and @xmath31 are given by @xcite : @xmath32 from the value of the mass of the top quark reported by the cdf and d0 collaborations @xcite , @xmath33 , and the scaling proposed in @xcite one obtains : while the mass of the charm quark is @xcite : @xmath35 the current values of the parameters that include the calculated qcd corrections are @xcite : @xmath36 the largest theoretical uncertainty which affects this constraint is that on the `` bag '' parameter @xmath37 , that reflects non - perturbative qcd contributions to the process . using the value of the jlqcd collaboration @xcite , @xmath38 , with a calculation similar to that reported in @xcite the value used in the following can be derived as : @xmath39 neutral mesons containing a @xmath41 quark show a behaviour similar to neutral kaons .
the heavy and light mass eigenstates , @xmath42 and @xmath43 respectively , are different from the cp eigenstates @xmath0 and @xmath44 : @xmath45 is the key feature of the physics while the lifetime difference dominates the effects in the neutral kaon system .
this mass difference can be measured by means of the study of the oscillations of one cp eigenstate into the other .
the high precision world average is @xcite : @xmath46 the relation of @xmath47 with the ckm parameters , making use of the standard model description of the box diagrams that give rise to the mixing and the parametrisation ( [ equation : ckm ] ) of the ckm matrix , reads : @xmath48 .
\label{equation : dmd}\ ] ] the function @xmath29 is given by ( [ equation : f2 ] ) , the value of the calculated qcd correction @xmath49 is @xcite : @xmath50 and the equivalent of the @xmath37 parameter for the kaon system , @xmath51 , is taken as @xcite : @xmath52 @xmath1 mesons are believed to undergo a mixing analogous to the @xmath0 ones .
their larger mass difference @xmath55 is responsible for oscillations that are faster than the @xmath0 ones , and have thus still eluded direct observation .
a lower limit has been set by the lep b oscillation working group combining the results of the searches performed by the lep experiments with a contribution from the sld and cdf collaborations , as @xcite : @xmath56 the expression for @xmath55 in the standard model is similar to that for @xmath47 . from the ratio of these two expressions the value of @xmath55 can be written as : @xmath57 where all the theoretical uncertainties are included in the quantity @xmath58 , known as @xcite : @xmath59 the three constraints described above are all affected by a large theoretical uncertainty on some of the parameters that enter their expression , namely @xmath37 , @xmath51 and @xmath58 . a determination of either @xmath2 or the ratio @xmath60 allows a more sensitive constraint not relaying on any non - perturbative qcd calculation .
it follows from the ckm matrix parametrisation of ( [ equation : ckm ] ) that : @xmath61 the cleo collaboration has measured this ratio by means of the endpoint of inclusive @xcite charmless semileptonic b decays as : @xmath62 .
the aleph and l3 collaborations have recently measured at lep the inclusive charmless semileptonic branching fraction of beauty hadrons , @xmath63 , from which the value of @xmath2 can be extracted @xcite as : @xmath64 the experimental results are : | the precise measurements of the @xmath0 oscillation frequency and the limit on the @xmath1 one as well as the determination of the cabibbo - kobayashi - maskawa matrix element @xmath2 improve the constraints on the other elements of this matrix .
a fit to the experimental data and the theory calculations leads to the determination of the vertex of the unitarity triangle as : @xmath3 the values of its angles , in their customary definition in terms of sines for @xmath4 and @xmath5 , are found to be : @xmath6 indirect information on non - perturbative qcd parameters , on the presence of a cp violating complex phase in the ckm matrix and on the @xmath1 oscillation frequency are also extracted .
european organization for nuclear research cern - ep/98 - 133 + august , 21 1998 + * indirect measurement of the vertex and angles of the unitarity triangle * * salvatore mele * + cern , ch1211 , genve 23 , switzerland _ to be submitted to physics letters b _ |
in @xcite , a two - variable polynomial invariant of finite quandles was introduced .
this polynomial quantifies the way in which the trivial action of one quandle element on another is distributed throughout the quandle as opposed to concentrated in a single identity element as in a group . in @xcite
the quandle polynomial was generalized to a family of @xmath1 polynomials where @xmath2 is the least common multiple of the exponents of the columns of the quandle matrix considered as elements of of the symmetric group @xmath3 on the elements of the quandle .
in both cases , the quandle polynomials were used to enhance the quandle counting invariants to obtain new invariants which specialize to the original quandle counting invariants but contain more information . in this paper
we study the natural generalization of quandle polynomials to finite racks .
we are able to show that for at least one class of finite racks , the generalized rack polynomials determine the rack structure up to isomorphism , and we identify another class of quandles for which the generalized rack polynomials do not determine the isomorphism class .
we then use these rack polynomials to enhance the rack counting invariants from @xcite .
the paper is organized as follows . in section [ rrp ]
we review the definitions of racks and rack polynomials and give some examples . in section [ grpcar ] we show that constant action racks are classified by their generalized rack polynomials . in section [ nsata ] we show that alexander quandles have quandle polynomial of the form @xmath0 and that unlike constant action racks , alexander quandles and other quandles with quandle polynomial @xmath0 are not classified by their generalized quandle polynomials . in section [ rpinv ]
we define rack polynomial enhanced counting invariants . in section [ quest ] we collect some questions for future investigation .
in @xcite , joyce defined a kind of self - distributive algebraic structure which he dubbed a `` quandle . '' in @xcite , quandles were generalized to a larger class of self - distributive algebraic systems known as `` racks . ''
both concepts appear under other names in the literature such as `` distributive groupoids , '' `` automorphic sets '' and `` kei . ''
see @xcite . axiom ( i ) requires that each element @xmath4 of a rack @xmath5 acts on @xmath5 bijectively , while axiom ( ii ) requires these bijections to be automorphisms of the rack structure .
the bijectivity of the action of @xmath4 gives us a right inverse action @xmath6 defined by @xmath7 where @xmath8 .
the reader can check that @xmath9 is also a rack , called the _ dual _ of @xmath10 standard examples of rack and quandle structures include : @xmath11 any union of conjugacy classes in a group @xmath12 with operation @xmath13 , @xmath14 the set of right cosets in a group @xmath12 of a subgroup @xmath15 ( not necessarily normal ) fixed by an automorphism @xmath16 with rack operation @xmath17 any set @xmath5 with a permutation @xmath18 with @xmath19 ( these are _ constant action racks _ or _ permutation racks _ ) the subset of a vector space @xmath20 on which a bilinear form @xmath21 with @xmath22 where @xmath23 is a non - zero scalar ( these are called _ coxeter racks _ ; see @xcite ) any module over @xmath24/s(1-t - s)$ ] with @xmath25 .
racks of the last type in which @xmath26 are known as _
alexander quandles_. a rack is _ abelian _ if for all @xmath27 we have @xmath28 in addition to being right - distributive , abelian quandles are also left - distributive , since we have @xmath29 a quandle is a _ crossed set _ ( see @xcite ) if we have @xmath30 the reader can check that alexander quandles are abelian and coxeter quandles ( set @xmath31 ) are crossed sets .
we can express rack structures on a finite set @xmath32 in an algebra - agnostic way , i.e. without needing a formula for @xmath33 , by giving the rack operation table as a matrix @xmath34 whose @xmath35 entry is @xmath36 where @xmath37 .
we call this the _ rack matrix _ of @xmath5 . @xmath38.\ ] ] for a rack @xmath5 , say that an equivalence relation @xmath39 on @xmath5 is a _ congruence _ if @xmath40 and @xmath41 imply @xmath42 .
the set @xmath43 of equivalence classes then forms a _ quotient rack _ under the operation @xmath44=[x]\tr[y]$ ] .
see @xcite for more .
next , we have a definition from @xcite : @xmath45 @xmath46 @xmath47
in this section we show that constant action racks are classified by their generalized rack polynomials .
[ genrpofconstact ] let @xmath5 be the constant action rack of a given permutation @xmath48 on @xmath49 .
then the generalized rack polynomial of @xmath5 is @xmath50 where @xmath51 is the number of @xmath52 such that @xmath53 and @xmath54 is the number of @xmath52 such that @xmath55 . for any @xmath56
, @xmath57 so each term of @xmath58 contains a factor of @xmath59 .
furthermore , @xmath60 therefore , each @xmath4 such that @xmath61 corresponds to a term of @xmath62 , and each @xmath4 such that @xmath63 corresponds to a term of @xmath59 .
there are @xmath54 distinct @xmath4 such that @xmath64 , and thus @xmath65 distinct @xmath4 such that @xmath66 .
therefore , @xmath67 note that the original `` non - generalized '' rack polynomial corresponds to @xmath68 . for this specific case , we have @xmath51 and @xmath54 both equal to the number of fixed points of @xmath48 , and
so we have [ rpofconstact ] let @xmath5 be the constant action rack of a given permutation @xmath48 on @xmath49 , and suppose @xmath48 has @xmath54 fixed points .
then the rack polynomial of @xmath5 is @xmath69 [ rpcompleteinvariant ] the set of generalized rack polynomials is a complete invariant of constant action racks . to prove this result
we will need a pair of lemmas .
[ isom = cycles ] suppose @xmath5 and @xmath70 are constant action racks given by permutations @xmath48 and @xmath71 , respectively
. then @xmath5 and @xmath70 are isomorphic if and only if @xmath48 and @xmath71 have the same cycle structure .
suppose that @xmath5 and @xmath70 are isomorphic .
then for any positive integer @xmath72 , @xmath73 and thus @xmath5 and @xmath70 must have the same number of cycles whose length divides @xmath72 . for @xmath74 ,
this means that @xmath5 and @xmath70 have the same number of cycles of length @xmath75 .
proceeding inductively , @xmath5 and @xmath70 have the same number of cycles of all positive integer lengths @xmath72 , and so have the same cycle structure .
conversely , suppose that @xmath48 and @xmath71 have the same cycle structure , that is , we can write them as @xmath76 and @xmath77 let @xmath78 be the bijection given by @xmath79 .
then @xmath80 and so @xmath23 is a rack isomorphism from @xmath5 to @xmath70 .
[ rp = cycles ] suppose @xmath5 and @xmath70 are constant action racks given by the permutations @xmath48 and @xmath71 , respectively
. then @xmath81 for all @xmath82 if and only if @xmath48 and @xmath71 have the same cycle structure . by proposition [ genrpofconstact ] , if @xmath5 and @xmath70 are constant action racks of the same cardinality , then @xmath81 if and only if @xmath83 and @xmath84 if @xmath48 and @xmath71 have the same cycle structure , then they have the same number of elements in cycles whose length divides @xmath85 and the same number of elements in cycles whose length divides @xmath86 .
thus , both and hold and so @xmath81 .
if @xmath48 and @xmath71 do not have the same cycle structure , then there exists some minimal @xmath72 such that @xmath48 and @xmath71 do not have the same number of cycles of length @xmath72 .
thus , for @xmath87 can not hold , and so @xmath88 .
combining lemmas [ isom = cycles ] and [ rp = cycles ] immediately gives us proposition [ rpcompleteinvariant ] .
the following example shows that the generalized rack polynomials must be used to get a complete invariant on constant action racks ; the @xmath89rack polynomial is not sufficient .
in this section we study the quandle polynomials of alexander quandles .
[ alex=>nsata ] let @xmath91 be an alexander quandle such that @xmath92 .
then @xmath93 for some positive integer @xmath94 .
let @xmath95 be the function given by @xmath96 since @xmath97 @xmath98 is just left multiplication by @xmath99 . since alexander quandles
are abelian and thus left distributive , @xmath98 is automatically a homomorphism . by the first isomorphism theorem @xmath98 partitions @xmath91 into cosets by the congruence
@xmath100 these cosets are all of size @xmath101 ; clearly @xmath94 .
we will show that @xmath102 if and only if @xmath103 .
suppose @xmath102 .
then @xmath104 so @xmath105 and so @xmath103 .
suppose @xmath103 .
then @xmath106 implies @xmath107 and hence @xmath108 therefore , for any @xmath109 there are exactly @xmath51 choices of @xmath110 such that @xmath102 and also exactly @xmath51 choices of @xmath110 such that @xmath111 .
thus , @xmath112 for all @xmath109 , and so @xmath113 note that proposition [ alex=>nsata ] provides a second , equivalent definition for @xmath114-equivalence
. [ cor7 ] an equivalent definition of @xmath39 is @xmath115 if and only if @xmath116 .
this also gives us another proof of the fact noted in @xcite that all alexander quandles are crossed sets .
this follows immediately from corollary [ cor7 ] and the reflexivity of equivalence relations .
these results suggest the possibility of defining @xmath39 for all crossed sets , or proving the converse of [ alex=>nsata ] .
however , there is a counterexample to both of these natural conjectures .
[ ce ] @xmath117 this also provides a counterexample to the conjecture in @xcite that distinct non - latin quandles are distinguished by at least one of their generalized rack polynomials .
@xmath118 @xmath119 @xmath120
in @xcite , the quandle counting invariant @xmath121 was extended to the case of finite non - quandle racks . in this section
we will enhance this invariant with rack polynomials .
we begin by recalling how this was done in the quandle case .
@xmath122 the subquandle polynomials of the image subquandles in @xmath123 are used to enhance the quandle counting invariants in @xcite .
specifically , instead of counting @xmath75 for each element of @xmath123 to obtain the quandle counting invariant @xmath121 , we count @xmath124 to obtain a multiset of subquandle polynomials .
we can express these multisets in a polynomial - style form by writing the elements of the multiset as powers of a variable @xmath125 and the multiplicities as coefficients .
let @xmath126 be a link and @xmath127 a finite quandle .
the @xmath128_subquandle polynomial invariant _ of @xmath126 with respect to @xmath127 is then @xmath129 now , let @xmath126 be an oriented link with @xmath130 ordered components . for any diagram @xmath131 of @xmath126 , we can regard @xmath131 as a framed link using the blackboard framing , i.e. giving each component of @xmath126 a framing number @xmath132 equal to its self - writhe .
thus , such a diagram has a framing vector @xmath133 for any finite rack @xmath127 , let @xmath134 be the _ rack rank _ of @xmath127 , i.e. the exponent of the permutation in @xmath135 along the diagonal of the rack matrix of @xmath127 .
if two ambient isotopic diagrams of @xmath131 have writhe vectors which are componentwise congruent modulo @xmath134 , then there is a bijection @xmath136 between the sets of rack homomorphisms from the fundamental racks of @xmath137 and @xmath138 into @xmath127 defined by sending a coloring of one diagram to a coloring of the same diagram with @xmath139 kinks added . indeed ,
since any subrack containing an element @xmath140 must also contain the rack powers @xmath141 for all @xmath14 ( see @xcite ) , @xmath142 preserves image subracks . hence , as far as @xmath127 is concerned , the framing vectors of @xmath131 live in @xmath143 , and we have an invariant of unframed links given by @xmath144 called the _ simple rack counting invariant_. a refinement obtained by keeping track of which colorings belong to which framings is the _ rack counting polynomial _ invariant latexmath:[\[pr(l , t)=\sum_{\mathbf{w}\in w } \left ( @xmath127 is a quandle , @xmath146 and we have @xmath147 .
see @xcite for more .
we would like to jazz up these rack counting invariants with the generalized rack polynomials . to this end , we propose the following @xmath148 @xmath149 @xmath150 @xmath151 specializing @xmath152 ( or , indeed , @xmath153 ) in the subrack polynomial @xmath154 yields the rack counting polynomial . since every finite quandle @xmath127 is a rack , the fact that subquandle polynomial invariants are stronger than unenhanced quandle counting invariants means _ a fortiori _ that subrack polynomial invariants are stronger than unenhanced rack counting invariants .
the next example shows how subrack enhancement gives more information about a knot or link than the unadorned rack counting invariant .
@xmath155\ ] ] indeed , the example suggests the following if @xmath156 is a quandle and @xmath157 a knot , then the contributions to @xmath158 from @xmath159 are equal for all powers of @xmath160 .
that is , @xmath154 includes the term @xmath161 quandle colorings do not depend on framing , so we get the same contribution , namely @xmath162 from each framing .
we have shown that some classes of racks are classified by their generalized rack polynomials ( the constant action racks ) while others are not ( quandles with polynomial @xmath0 ) .
what conditions are sufficient for a type of rack to be determined by its generalized rack polynomials ? since @xmath0 quandles are not determined by their generalized quandle polynomials , what extra information is necessary to determine these quandles up to isomorphism ?
how can such extra information be incorporated into the enhanced rack counting invariants ? for every rack @xmath163 , the quotient rack under operator equivalence ( @xmath164 ) is a quandle .
what is the relationship between the subrack polynomial invariant with respect to @xmath163 and the subquandle polynomial invariant with respect to @xmath165 ? | we study rack polynomials and the link invariants they define .
we show that constant action racks are classified by their generalized rack polynomials and show that @xmath0-quandles are not classified by their generalized quandle polynomials .
we use subrack polynomials to define enhanced rack counting invariants , generalizing the quandle polynomial invariants .
keywords : finite racks , rack polynomials , knot and link invariants 2000 msc : 57m25 , 57m27 , 17d99 |
scattering theory for manifolds with cylindrical ends deals with the following geometric situation .
let @xmath1 be an oriented , connected , compact riemannian manifold with boundary @xmath2 such that the metric is a product near the boundary , i.e. , there is a tubular neighborhood of @xmath3 which is isometric to @xmath4 \times y,$ ] equipped with the product metric @xmath5 , where @xmath6 is a riemannian metric on @xmath3 .
the non - compact elongation @xmath7 of @xmath1 is then obtained from @xmath1 by attaching the half - cylinder @xmath8 over the boundary ( see figure [ fiig01 ] ) : @xmath9 of @xmath1.,width=340 ] the riemannian metric on @xmath1 is extended to one on @xmath7 in the obvious way so that @xmath10 such a manifold is called a _
manifold with cylindrical ends_. scattering theory on @xmath7 investigates how wave packets coming in from infinity are scattered in @xmath1 .
the scattering of @xmath0-forms on @xmath7 is described by the scattering matrix @xmath11 where @xmath12 is the laplace - beltrami operator acting on @xmath0-forms on @xmath3 and @xmath13 the eigenspace of @xmath12 with eigenvalue @xmath14 .
in particular for small values of the spectral parameter @xmath15 we have @xmath16 where @xmath17 is the space of harmonic @xmath0-forms on @xmath3 .
for the purposes of this article one can think of the scattering matrix for small values of the spectral parameter as being defined by the statement of theorem [ main1 ] . in this case
it can be shown that @xmath18 leaves the direct summands invariant and is of the form @xmath19 where @xmath20 is the scattering matrix describing the scattering of coclosed forms . again
@xmath21 can be defined by the statement of theorem [ main2 ] as the matrix relating the incoming and outgoing waves .
the first observation is that the total scattering matrix for coclosed @xmath0-forms at energy @xmath22 , @xmath23 is a self - adjoint involution which anti - commutes with the hodge star operator .
the @xmath24 eigenspace of @xmath25 coincides with the space of harmonic forms that represent cohomology classes in @xmath26 .
the eisenbud - wigner time - delay operator @xmath27 describes the time - delay a @xmath0-form - wave undergoes when being scattered in @xmath1 ( see appendix [ app2 ] ) .
it can be calculated from the eisenbud - wigner formula ( see appendix [ app2 ] ) , and for small @xmath15 it is given by @xmath28 of course @xmath29 where @xmath30 is the time delay operator for coclosed forms defined by @xmath31 its value @xmath32 at zero energy is of particular interest and we call it the _ scattering length_. the physical interpretation of the scattering length is as follows .
if a coclosed wave packet has very low energy then , by the uncertainty relation , it will be far spread out .
in particular it will not be able to `` feel '' details of the geometry of @xmath1 .
the effect of the manifold @xmath1 in the scattering process for a wave is then close to that of a cylindrical obstacle of length given by the scattering length .
it is therefore an interesting question to determine what geometric properties of @xmath1 have an effect on the scattering length . since @xmath33 commutes with the hodge star operator it is enough to know its restriction to the @xmath34 eigenspace of @xmath25 .
denote by @xmath35 the stable norm of a homology class , and by @xmath36 the comass norm on the cohomology groups ( see @xcite ) .
let @xmath37 be the smallest positive eigenvalue of @xmath38 .
put @xmath39 furthermore , in section [ lzweiestimates ] we introduce for each @xmath40 and @xmath41 constants @xmath42 , which are related to the estimation of the comass norm on @xmath43 .
they are equal to 1 for @xmath44 and @xmath45 .
one of our main results relates the scattering length to certains norms in homology ( theorem [ main3 ] ) and gives rise to the following estimation of the scattering length in terms of geometric data .
[ main00 ] let @xmath41 . for every @xmath46 in the @xmath34-eigenspace of @xmath47 we have @xmath48 \cap \partial \phi\|_{st}^2 \leq
\langle \phi , t(0)^{-1 } \phi \rangle \leq \frac{1}{2 } c(n , p+1 ) \mathrm{vol}(m ) \|\partial \phi \|_\infty^2 . \ ] ] as an example we treat the case when @xmath3 has two connected components @xmath49 and @xmath50 and @xmath44 . in this case
there is a canonical basis in @xmath51 with respect to which @xmath52 has the form @xmath53 so that @xmath54 and the constants @xmath55 and @xmath56 are given by @xmath57)\|_{st}^2 ( \mathrm{vol}(y_1)+ \mathrm{vol}(y_2))},\\ c_2= 2 \mathrm{vol}(m)^{-1 } \frac{\mathrm{dist}(y_1,y_2)^2 \mathrm{vol}(y_1)\mathrm{vol}(y_2 ) } { \mathrm{vol}(y_1)+ \mathrm{vol}(y_2)}.\end{gathered}\ ] ] the map @xmath58 is the inclusion of @xmath3 into @xmath1 .
so we get an estimate for the scattering length by purely geometric quantities .
the physical interpretation of this is as follows . for a wave of low energy the reflection coefficient @xmath59 and the transmission coefficient @xmath60 for a wave coming in at the boundary component @xmath49
are approximated by their values at zero , namely ( see section [ scatterex1 ] ) @xmath61 the time - delay is then determined by @xmath62 and @xmath63 .
for example in the case where @xmath64 the reflection coefficient at zero energy is zero and the time delay of the transmitted wave is equal to @xmath65 ( see section [ scatterex1 ] ) .
another example , the full - torus , is treated in section [ scatterex2 ] .
given @xmath66 , let @xmath67 be the manifold that is obtained from @xmath1 by attaching the cylinder @xmath68 \times y$ ] to @xmath1 .
we also investigate how the @xmath69-norm of a class in @xmath70 and the @xmath69-norm of its image in @xmath71 under the connecting homomorphism are related for the manifold with boundary @xmath67 .
there is an operator @xmath72 that relates the @xmath69-norm of classes in the complement of the kernel of the connecting homomorphism to the @xmath69-norm of the image of this class under the connecting homomorphism .
so @xmath72 measures to what extent the connecting homomorphism is a partial isometry .
theorem [ theoremquadratic ] shows that the operator @xmath72 has an expansion of the form @xmath73 as @xmath74 .
this shows that one can calculate the scattering length by approximating @xmath7 by the compact manifolds @xmath67 and consider the constant term in the above expansion .
the exponential decay of the remainder term may also be useful for numerical computations .
the paper is organized as follows . in sections [ one ]
we review stationary scattering theory for @xmath0-forms on manifolds with cylindrical ends .
section [ three ] and section [ four ] deal with cohomology of @xmath1 and @xmath7 , and their relation to scattering theory and the continuous spectrum of the laplacian of @xmath7 .
we also derive a cohomological formula for the scattering length . in sections
[ lzweiestimates ] and [ systoles ] it is shown that the @xmath69-scalar products on the cohomology groups of @xmath7 can be estimated in terms of geometric quantities and that these estimates imply estimates on the scattering length .
section [ examples ] treats the case of functions in the case of two boundary components and the case of a full - torus . in this section
we demonstrate how our main result can be used to obtain estimates of the scattering length in terms of geometric data . in appendix [ app2 ]
we discuss the relation between the stationary and the dynamical approach to scattering theory and we establish the eisenbud - wigner formula for manifolds with cylindrical ends .
whereas for the sake of notational simplicity we restricted ourselves in this paper to manifolds with cylindrical ends most of our analysis carries over in a straightforward manner to waveguides if neumann boundary conditions are imposed .
* acknowledgements . *
much of the work on this paper has been done during the visit of the second author to the mpi in bonn and he would like to thank the mpi for the kind support .
both authors would also like to thank the msri in berkeley for hospitality during the program `` analysis on singular spaces '' .
we are grateful to werner ballmann , alexej bolsinov , peter perry , and sasha pushnitski for useful discussions and comments .
as before let @xmath1 be an oriented , connected , compact riemannian manifold with boundary @xmath2 such that the metric is a product near the boundary .
let @xmath7 be the elongation of @xmath1 .
for any riemannian manifold @xmath75 we denote by @xmath76 ( resp .
@xmath77 , @xmath78 ) the space of smooth @xmath0-forms ( resp .
smooth @xmath0-forms with compact supports , @xmath69-forms ) on @xmath75 .
let @xmath79 be the laplace operator on @xmath76 . throughout this paper a harmonic @xmath0-form will mean a @xmath0-form @xmath80 with @xmath81 .
since @xmath7 is complete , the laplace - beltrami operator @xmath79 on @xmath0-forms is essentially self - adjoint when regarded as operator in @xmath82 with domain @xmath83 ( @xcite ) .
we continue to denote its self - adjoint extension by @xmath79 . in this section
we recall some facts concerning the generalized eigenforms of @xmath79 and derive some properties of the scattering matrix for low energy .
let @xmath12 denote the laplacian on @xmath0-forms of @xmath3 .
let @xmath84 be the distinct eigenvalues of of @xmath85 .
let @xmath86 the minimal riemann surface on which @xmath87 is a single - valued function for all @xmath88 .
as proved by melrose@xcite the resolvent @xmath89 , regarded as operator @xmath90 , admits a meromorphic extension from the half - plane @xmath91 to @xmath92 .
let @xmath93 be the first non - zero eigenvalue of @xmath85 .
then it follows in particular that @xmath89 extends to a meromorphic function on the disc @xmath94 . as a consequence
one can define analytic families of generalized eigenforms . for @xmath95
we denote by @xmath96 the hypersurface @xmath97 .
note that the restriction of the bundle @xmath98 to @xmath96 is canonically isomorphic to the direct sum @xmath99 since each vector @xmath100 at some point @xmath101 can be uniquely decomposed as @xmath102 with @xmath103 and @xmath104 .
accordingly , the restriction of any @xmath0-form @xmath105 to the cylinder @xmath106 is of the form @xmath107 where @xmath108 and @xmath109 are sections of the pulled back bundles @xmath110 and @xmath111 , respectively , and @xmath112 is the canonical projection .
we think of @xmath113 and @xmath114 as forms on @xmath3 that depend smoothly on the additional parameter @xmath115 .
the map @xmath116 is an isomorphism of vector bundles .
the exterior differential of such a form @xmath117 is then given by @xmath118 where @xmath119 denotes the exterior differential on @xmath3 . in matrix notation
this means @xmath120 where we denote @xmath121 . since the metric has product structure the decomposition is orthogonal and the formal adjoint @xmath122 of @xmath123 is therefore @xmath124 here again we use the notation @xmath125 for the codifferential on @xmath3 . we therefore have @xmath126 and @xmath127 for all @xmath0 .
this form of an operator allows for a separation of variables on the cylinder @xmath128 .
suppose that @xmath129 is an orthonormal sequence of eigenforms of @xmath130 with eigenvalues @xmath131 .
then for @xmath132 , any solution @xmath133 of the equation @xmath134 has an expansion of the form @xmath135 the series converges in the @xmath136-topology .
the square roots are chosen throughout the article as @xmath137 if @xmath138 and @xmath139 .
from the analytic continuation of the resolvent one gets the following result .
[ main1 ] for each @xmath140 in @xmath141 there exists a @xmath0-form @xmath142 which is meromorphic in @xmath143 such that the following conditions hold 1 .
@xmath142 is holomorphic in @xmath15 for @xmath91 .
2 . @xmath144 .
there exists @xmath145 such that on @xmath146 we have @xmath147 4 .
@xmath148 is a linear operator , and @xmath149 and @xmath150 are meromorphic functions of @xmath15 .
moreover , @xmath149 , @xmath150 and @xmath142 are uniquely determined by these properties .
this follows from@xcite , @xcite . for the convenience of the reader
we include the details .
let @xmath151 be the minimal riemann surface to which all the functions @xmath152 extend to be holomorphic . by ( * ?
* thorm 0.2 ) , ( * ? ? ? * theorem7.1 ) the resolvent @xmath153 , regarded as operator @xmath154 , extends to a meromorphic function of @xmath155 .
especially @xmath153 extends to a meromorphic function of @xmath156 as an operator @xmath154 .
let @xmath157 be a smooth function with support in @xmath146 which is equal to @xmath158 outside a compact set .
put @xmath159 then @xmath142 is a smooth @xmath0-form on @xmath7 which depends meromorphically on @xmath155 .
moreover , it satisfies @xmath160 since @xmath161 is square integrable for @xmath162 , the expansion ( [ expansion1 ] ) has the form @xmath163 where @xmath164 and @xmath165 moreover , @xmath150 is square integrable for @xmath91 .
this implies that @xmath166 for @xmath91 .
since @xmath167 is meromorphic , it follows that @xmath166 for @xmath132 .
thus we conclude that @xmath168 is a meromorphic function in the disc @xmath132 with values in @xmath82 .
the uniqueness is an immediate consequence of the self - adjointness of @xmath79 .
namely , if @xmath169 and @xmath170 both have the above properties then their difference @xmath171 is square integrable for @xmath91 and is contained in the kernel of @xmath172 .
using that @xmath79 is self - adjoint , we get @xmath173 for @xmath91 .
since it is meromorphic in @xmath15 we conclude that @xmath173 .
the requirement that @xmath150 is in @xmath174 in fact implies a much faster decay at infinity .
[ rapiddec ] if @xmath175 is regular at @xmath15 , then all derivatives of @xmath175 in @xmath115 and @xmath176 are exponentially decaying as @xmath177 .
more precisely @xmath178 for all @xmath179 and some positive constants @xmath180 and @xmath181 .
the proof is already implicitly contained in the proof of the previous theorem .
namely , @xmath175 is smooth by elliptic regularity and thus the expansion ( [ expansion ] ) converges in @xmath182 .
since @xmath183 , we get the decay for @xmath184 with @xmath185 .
let @xmath186 .
put @xmath187 then the expansion ( iii ) of theorem [ main1 ] takes the form @xmath188 where @xmath189 as endomorphism of @xmath190 .
a priory there is no reason why @xmath149 should leave the summands invariant . nevertheless this is guaranteed by a continuous version of the hodge decomposition as the proof of the following proposition shows .
[ vanishing ] @xmath149 leaves the spaces @xmath191 and @xmath192 invariant , i.e. , @xmath193 and @xmath194 .
let @xmath195 . using the expansion of [ expansion11 ]
, it follows that on @xmath106 we have @xmath196 comparing the leading terms , it follows that @xmath197 since all derivatives of @xmath168 and @xmath198 are exponentially decaying , the uniqueness statement in theorem [ main1 ] implies immediately @xmath193 and @xmath194 .
note that the splitting @xmath199 indeed corresponds to the hodge decomposition .
let @xmath200 and @xmath201 the projection on the first and second summand , respectively .
then we have we have @xmath202 , i.e. @xmath203 if and only if @xmath204
. similarly @xmath205 , i.e. @xmath206 if and only if @xmath207 .
let @xmath208 .
suppose that @xmath204 .
then , in particular , we have @xmath209 . applying @xmath122 to the second equation of ( [ expansion11 ] ) and using proposition [ vanishing ]
, it follows that @xmath210 . for the other direction ,
observe that by ( [ expansion11 ] ) , @xmath211 on @xmath146 .
hence , @xmath212 is exponentially decaying .
in particular it is square integrable for @xmath213 . since @xmath79 is self - adjoint and @xmath214
we get @xmath215 .
thus , if @xmath210 , it follows that @xmath216 .
the proof of the other case is analogous .
the following relation holds for @xmath217 .
@xmath218 let @xmath203 . using ( [ expansion11 ] ) , we get @xmath219 comparing the leading terms , it follows from theorem [ main1 ] that @xmath220 using ( [ expansion11 ] ) , we get @xmath221 , which is equivalent to the statement of the proposition .
let us use the notation @xmath222 for the restriction of @xmath149 to @xmath223 .
then the above proposition shows that @xmath224 in the following we will suppress the index @xmath0 and write @xmath225 , meaning that @xmath225 is acting on the space of harmonic forms , leaving the space of @xmath0-forms invariant .
it is the scattering matrix at low energy for the scattering problem for coclosed forms .
summarizing we have established the following theorem .
[ main2 ] for each harmonic @xmath0-form @xmath226 there exists a @xmath0-form @xmath227 on @xmath7 , which is meromorphic in @xmath15 in the disc @xmath184 such that 1 .
@xmath228 2 .
@xmath227 is holomorphic in @xmath15 for @xmath91 .
3 . @xmath229 .
4 . @xmath230 on @xmath106 , 5 .
@xmath231 and @xmath232 are meromorphic functions of @xmath15 .
moreover , @xmath225 , @xmath233 and @xmath227 are uniquely determined by these properties .
the scattering matrix has the following properties [ main22 ] the function @xmath225 satisfies the following equations 1 .
2 . @xmath235 .
3 . @xmath236 , where @xmath237 is the hodge star operator on @xmath3 . for @xmath66 let @xmath238 be the manifold @xmath239 with boundary @xmath240 .
let @xmath241 be the volume form of @xmath7 . by theorem [ main2 ] , ( iii )
, we have @xmath242 using green s formula , we obtain @xmath243 in the limit @xmath74 this expression can be evaluated using theorem [ main2 ] .
we obtain @xmath244 which proves the first statement .
the second statement follows from the functional equation @xmath245 which is a simple consequence of the uniqueness statement in theorem [ main1 ] . to show that @xmath18 anticommutes with the hodge star operator on @xmath3 we note that that hodge star operator @xmath246 on @xmath7 commutes with the laplace operator @xmath247 , i.e. @xmath248 .
applying this to @xmath227 and using the uniqueness statement we obtain immediately @xmath249 for @xmath203 we get @xmath250 and consequently @xmath251 , where we used that @xmath252 anticommutes with @xmath18 . as an application we obtain the following well known result about the signature @xmath253 of a closed manifold @xmath3 .
let @xmath3 be a closed oriented manifold .
assume that @xmath3 is the boundary of a compact manifold
. then @xmath254 .
we may assume that @xmath255 .
otherwise the signature is zero .
pick a riemannian metric on @xmath3 .
let @xmath256 be the @xmath257 eigenspaces of @xmath237 acting in @xmath258 .
then the signature @xmath253 of @xmath3 is given by @xmath259 if @xmath3 is the boundary of a compact riemannian manifold @xmath1 , it follows from theorem [ main22 ] , that @xmath225 is regular at @xmath260 , @xmath261 and @xmath25 intertwines @xmath262 and @xmath263 .
hence we get @xmath254 .
the same proof works equally well for dirac type operators .
it implies the cobordism invariance of the index of dirac operators .
[ reg ] @xmath225 , @xmath227 , and @xmath233 are regular for real @xmath15 . if @xmath264 then @xmath265 . for real @xmath15
it follows from theorem [ main22 ] that @xmath266 and therefore @xmath267 .
in particular , @xmath268 is bounded on the real line and can not have a pole there .
it remains to show that @xmath133 is regular for real @xmath15 .
suppose that @xmath46 is a square integrable eigensection of @xmath247 with real eigenvalue @xmath269 .
then the expansion ( [ expansion1 ] ) of @xmath46 on @xmath270 takes the form @xmath271 this implies that @xmath46 is exponentially decaying . since for real @xmath15 @xmath272 and @xmath273 is a meromorphic function we get @xmath274 .
suppose now that @xmath227 has a pole of order @xmath275 at @xmath269 .
then @xmath276 is an eigenform with eigenvalue @xmath269 and it also is square integrable since @xmath268 is regular at @xmath269 . by the above @xmath277 .
it follows that @xmath278 and therefore @xmath227 is regular at @xmath269 . if @xmath264 then @xmath279 is square integrable and harmonic and by the same argument @xmath265 .
in particular , it follows that @xmath227 is regular at @xmath260 .
therfore , @xmath279 and @xmath280 are well defined . from the proof of proposition [ reg ]
we obtain the following corollary .
[ orthogonal ] @xmath279 and @xmath281 are orthogonal to the space @xmath282 of square integrable harmonic forms .
the scattering matrix @xmath225 is also regular at @xmath22 and it follows from theorem [ main22 ] that @xmath25 is a self - adjoint involution .
hence , @xmath283 decomposes into @xmath24 and @xmath34 eigenspaces . for @xmath284 with @xmath285
we get that @xmath279 is a smooth coclosed harmonic @xmath0-form whose restriction to @xmath146 equals @xmath286 where @xmath287 and its derivatives are exponentially decaying . that is , @xmath140 is a limiting value of @xmath288 in the sense of atiyah , patodi , and singer ( @xcite ) .
it turns out that the converse is also true .
[ limiting ] the @xmath24 eigenspace for @xmath25 is the set of limiting values of coclosed harmonic forms on @xmath7 , i.e , it equals @xmath289 furthermore for each @xmath290 , @xmath279 satisfies @xmath291 and @xmath292 .
suppose that @xmath293 and @xmath294 are both coclosed harmonic forms on @xmath7 with limiting values @xmath140 and @xmath46 , respectively .
since @xmath295 and @xmath296 are both exponentially decaying and @xmath140 and @xmath46 are closed and coclosed on @xmath3 the forms @xmath297 and @xmath298 are exponentially decaying . using green s formula ,
we get @xmath299 thus @xmath300 .
similarly we get @xmath301 . using stokes formula
, it follows that @xmath302 thus , @xmath303 .
now suppose that @xmath46 is a limiting value which is in the @xmath34 eigenspace to @xmath25 .
since @xmath237 anticommutes with @xmath25 , it follows that @xmath304 is in the @xmath24 eigenspace .
it is therefore a limiting value . since @xmath304 and @xmath46 are both limiting values it follows from the above that @xmath305 and therefore , @xmath306 . since the set of limiting vectors contains the @xmath24 eigenspace it has to coincide with the @xmath24 eigenspace .
finally we derive some formulas concerning @xmath281 which we are going to use in the next section .
note that the restriction of @xmath307 to the cylinder @xmath146 has the form @xmath308 and for @xmath260 : @xmath309 differentiating the equation @xmath310 it follows that @xmath311 hence , @xmath312 is in the kernel of @xmath247 and its restriction to the cylinder has the form @xmath313 by theorem [ main22 ] , ( iii ) , we get @xmath314 thus , @xmath315 is an extended harmonic form with limiting value @xmath316 . since @xmath315 is coexact and bounded it is orthogonal to the space @xmath317 of square integrable harmonic forms . the proof of prop .
[ reg ] shows that @xmath318 is harmonic and orthogonal to @xmath317 .
their difference is therefore square integrable , harmonic and orthogonal to @xmath317 .
thus , it vanishes and we have the following nice formula @xmath319 in particular for @xmath320 we have @xmath321 or equivalently @xmath322
as before let @xmath1 be a compact manifold with boundary @xmath3 and @xmath323 the associated manifold with a cylindrical end .
we consider the long exact sequence @xmath324^(0.3)\partial & h^k(m , y,{{\mathbb{r } } } ) \ar[r]^(0.5)e & h^k(m,{{\mathbb{r } } } ) \ar[r]^{r } & h^k(y,{{\mathbb{r } } } ) \ar[r]^(0.4){\partial } & h^{k+1}(m , y,{{\mathbb{r } } } ) \ar[r]^(0.6)e & \ldots , } \end{xy}\ ] ] in de rham cohomology . here
@xmath325 is the canonical embedding and @xmath326 is the restriction homomorphism .
there are three cochain complexes which compute the relative de rham cohomology .
let @xmath327 where @xmath328 is the inclusion . since @xmath123 commutes with @xmath329 , we get a complex @xmath330 .
its cohomology is denoted by @xmath331 .
there is an exact sequence of complexes @xmath332 & \lambda^*(m , y)\ar[r]^j & \lambda^*(m)\ar[r]^{i^ * } & \lambda^*(y)\ar[r ] & 0 , } \end{xy}\ ] ] where @xmath333 is the inclusion map .
it gives rise to the long exact sequence ( [ exactsequ ] ) .
the connecting homomorphism @xmath334 is defined as follows .
let @xmath335\in h^k(y,{{\mathbb{r}}})$ ] .
extend @xmath46 to a @xmath275-form @xmath117 on @xmath1 such that @xmath336 in a neighborhood of the boundary
. then @xmath337=[d \omega].\end{gathered}\ ] ] for the second description consider the cochain complex @xmath338 of the mapping cone of @xmath329 which is defined by @xmath339 with differential @xmath123 given by @xmath340 let @xmath341 be defined by @xmath342 and @xmath343 , respectively .
then @xmath344 and @xmath345 are cochain maps and we get a second exact sequence of cochain complexes @xmath332 & \lambda^{*-1}(y)\ar[r]^\alpha & \lambda^*_{rel}(m , y)\ar[r]^{\beta } & \lambda^*(m)\ar[r ] & 0 } \end{xy}\ ] ] there is a natural inclusion of cochain complexes @xmath346 it follows from the corresponding commutative diagram of long exact sequences that @xmath347 induces an isomorphism @xmath348 finally @xmath349 is also naturally isomorphic to the cohomology with compact supports @xmath350 . the isomorphism can be described as follows .
let @xmath351 be the canonical projection .
integration over the fibre @xmath352 of @xmath0 induces a mapping @xmath353 define a map @xmath354 this is a chain map . if the support of @xmath117 is contained in @xmath355 , then @xmath356 .
since every cohomology class in @xmath357 has a representative of this form , it follows that @xmath358 induces an isomorphism @xmath359 if we fix a metric on @xmath3 we may identify @xmath360 with the space of harmonic forms @xmath361 .
in fact , the image of @xmath325 , i.e. the kernel of @xmath326 can be read off from the scattering matrix at @xmath22 .
[ imageofs ] the @xmath24 eigenspace of the scattering matrix @xmath25 on @xmath362 coincides with @xmath363 .
let @xmath364 with @xmath365 . by proposition
[ limiting ] , @xmath279 is closed and coclosed .
therefore the restriction of @xmath279 to @xmath1 defines a cohomology class in @xmath366 .
expand @xmath279 on @xmath146 in terms of an orthonormal basis of @xmath367 .
using that @xmath279 is closed and coclosed , it follows that its expansion on @xmath146 has the form @xmath368 where @xmath369 is an orthonormal basis of @xmath370 consisting of eigenforms of @xmath371 with eigenvalues @xmath372 .
in particular @xmath373 and therefore the image of the cohomology class @xmath374 $ ] under @xmath326 is precisely @xmath375 $ ] .
therefore we have shown that the @xmath24 eigenspace of @xmath25 is contained in the image of @xmath376 in @xmath362 .
now let @xmath46 be an element in the image of @xmath326 , i.e. @xmath46 is a harmonic form that is in the same cohomology class as the restriction of a closed form @xmath377 on @xmath1 .
if @xmath140 is in the @xmath34 eigenspace of @xmath25 then by theorem [ main22 ] , @xmath378 is in the @xmath24 eigenspace and we have @xmath379 hence , any element in @xmath363 is in the orthogonal complement to the @xmath34 eigenspace of @xmath25 which is exactly the @xmath24 eigenspace .
this shows that @xmath380 is contained in the @xmath24 eigenspace and this concludes the proof .
hence , the scattering matrix at @xmath22 is determined completely by the metric on the boundary .
namely , it is equal to @xmath158 on the kernel of @xmath334 and equal to @xmath34 on the orthogonal complement of the kernel of @xmath334 .
recall that by proposition [ limiting ] , @xmath279 is a closed and coclosed @xmath0-form .
let @xmath381 be the map defined by @xmath382.\end{gathered}\ ] ] then , by construction , @xmath383 is the orthogonal projection onto the kernel of @xmath334 .
hodge theory for manifolds with boundary shows that absolute and relative cohomology classes have unique harmonic representatives that satisfy certain boundary conditions .
we recall the definition of the relative and absolute boundary conditions for the laplace operator .
the operator @xmath384 is the closure of the laplace operator with respect to the relative boundary conditions @xmath385 the operator @xmath386 is the closure of the laplace operator with respect to the absolute boundary conditions @xmath387 both operators are self - adjoint and have compact resolvents .
their kernels are the space of harmonic forms satisfying relative and absolute boundary conditions .
equivalently , they are given by @xmath388 hodge theory for manifolds with boundary shows that the canonical maps @xmath389 are isomorphisms , that is , every absolute / relative cohomology class has a unique harmonic representative satisfying absolute / relative boundary conditions ( see e.g. @xcite ) .
the harmonic representative @xmath46 of the cohomology class @xmath335 \in h^p(m , y,{{\mathbb{r}}})$ ] is the unique minimizer of the functional @xmath390 in @xmath335 $ ] .
similarly , any harmonic form satisfying absolute boundary conditions minimizes the @xmath69-norm in its absolute cohomology class .
apart from these minimax principles there is another interesting minimizing problem which is described in the following proposition . [ mini ]
let @xmath391 . consider the functional @xmath133 on @xmath392 which is defined by @xmath393 then there exists a unique coclosed harmonic form @xmath394 with @xmath395 such that @xmath394 is orthogonal to @xmath396 .
the minimum of @xmath133 is attained at @xmath394 and @xmath394 is the unique coclosed minimizer that is orthogonal to @xmath396 . if @xmath46 is closed @xmath397 is the harmonic representative in @xmath398 $ ] .
we divide the proof into several steps .
+ * uniqueness : * if two forms @xmath394 and @xmath399 are harmonic , coclosed , and their restrictions to @xmath3 coincide , it follows that @xmath400 is harmonic , coclosed and satisfies relative boundary conditions .
therefore , @xmath401 .
if both @xmath394 and @xmath399 are orthogonal to @xmath396 it follows that @xmath402 . +
* existence : * choose any extension @xmath403 of @xmath46 to @xmath1 which in a neighborhood of @xmath3 is of the form @xmath404 then @xmath405 vanishes near @xmath3 .
next we claim that the form @xmath406 is in the orthogonal complement of the kernel of @xmath384 . indeed , if @xmath407 , then @xmath408 where the first integral vanishes because @xmath403 is coclosed near @xmath3 and the second integral vanishes because @xmath358 satisfies relative boundary conditions .
let us denote by @xmath409 the orthogonal projection onto the kernel of @xmath384 .
since @xmath406 is in the orthogonal complement of the kernel of @xmath384 the following form is well defined and harmonic @xmath410 by construction it is also in the orthogonal complement of @xmath396 and satisfies @xmath395 .
moreover , @xmath394 is harmonic and since @xmath384 commutes with @xmath122 , we have @xmath411 since @xmath412 vansihes near @xmath3 , it is in the domain of @xmath384 and therefore , the right hand side vanishes .
hence @xmath413 . +
* minimizing property : * suppose that @xmath140 is a @xmath0-form with @xmath414 , then @xmath415 because @xmath416 .
thus , @xmath394 is a minimizer .
one can see immediately that the euler - lagrange equations for the minimizer are the equations @xmath417 .
thus , any coclosed minimizer has to be harmonic .
the uniqueness statement for the minimizer thus follows from the above uniqueness statement . since restriction to the boundary commutes with the differential the map which sends @xmath46 to @xmath394 commutes with the differential .
therefore , if @xmath46 is exact or closed , so is @xmath394 . now consider the manifold @xmath67 which is obtained from @xmath1 by attaching the cylinder @xmath68\times y$ ] to @xmath1 .
then @xmath67 is a manifold with boundary @xmath96 .
let @xmath418 .
we regard it as a harmonic form on @xmath96 . by the hodge theorem
there is a unique harmonic form in @xmath419 which represents @xmath398\in h^p(m_a,\partial m_a)$ ] .
let us denote this form by @xmath420 , where the notation @xmath421 indicates that @xmath421 maps @xmath422 to different spaces depending on @xmath423 .
for each @xmath423 we have the @xmath69-inner product on @xmath419 .
it is a natural question to ask whether @xmath334 as a map from one hilbert space to another one is a partial isometry . the scalar product on @xmath419 , however , depends on @xmath423 .
[ theoremquadratic ] let @xmath424 be the sesquilinear form on @xmath425 which is defined by @xmath426 and let @xmath427 be the unique linear operator @xmath428 such that @xmath429 then , as @xmath74 : @xmath430 recall that @xmath431 .
let @xmath432 .
then by ( [ expansion2 ] ) the restriction of @xmath433 to the cylinder @xmath146 has the following form @xmath434 it follows that @xmath433 is a coclosed harmonic form .
thus , by theorem [ mini ] @xmath433 is a coclosed minimizer of the functional @xmath435 with boundary condition @xmath436 .
let @xmath437 be the minimizer with boundary conditions @xmath438 so that @xmath439 is the unique harmonic representative of @xmath440 .
again by theorem [ mini ] @xmath441 minimizes the functional @xmath435 with boundary conditions @xmath442 then for every @xmath443 we have @xmath444 the second term on the right hand side can be estimated using the cauchy - schwarz inequality @xmath445 the first term in ( [ eqncxcx ] ) can be explicitly calculated . using ( [ fprime ] )
we get @xmath446 where we used that @xmath447 is orthogonal to @xmath140 .
thus , @xmath448 the terms on the right hand side can be estimated as follows .
first note that @xmath449 minimizes @xmath450 over all forms @xmath451 which restrict to @xmath452 on @xmath96 .
moreover , @xmath453 is exponentially decaying in @xmath423 .
define the form @xmath454 by @xmath455 on the cylinder and @xmath22 elsewhere .
then , @xmath456 by lemma [ rapiddec ] we have @xmath457 which implies @xmath458 to estimate @xmath459 , recall that by theorem [ mini ] , @xmath460 , where @xmath394 minimizes the functional @xmath461 with boundary conditions @xmath462 .
let @xmath463 such that @xmath464 for @xmath465 and @xmath466 for @xmath467 . for @xmath468 define @xmath469 by @xmath470\times y;\\ 0 , & \mathrm{otherwise}. \end{cases}\ ]
] then it follows that there exists @xmath471 such that @xmath472 thus we get @xmath473 by compactness of the unit sphere in @xmath425 the constant can be chosen independent of @xmath46 if we use vectors of norm @xmath158 only .
hence @xmath474 which implies @xmath475 from this inequality we deduce that @xmath476 . combined with ( [ eqnfin ] ) the statement of the theorem follows .
in this section we give a description of the long exact cohomology sequence of @xmath7 in terms of harmonic forms and we derive a cohomological formula for the scattering length .
let @xmath350 denote the de rham cohomology groups with compact supports .
it is well known ( see @xcite ) that @xmath477 and @xmath478 are canonically isomorphic to certain spaces of extended harmonic forms on @xmath7 .
we recall some details .
the space of extended harmonic forms @xmath479 is defined to be the subspace of all ( real valued ) @xmath480 satifying 1 ) @xmath481 and 2 ) there exist @xmath482 and @xmath483 such that @xmath484 note that for a given @xmath485 the sections @xmath486 and @xmath487 are uniquely determined .
we regard @xmath486 ( resp .
@xmath487 ) as the tangential ( resp .
normal ) boundary value of @xmath140 at infinity and we denote them by @xmath488 and @xmath489 , respectively . the spaces satisfying absolute and relative boundary conditions at infinity are then defined as @xmath490 since @xmath491 is harmonic and the form @xmath492 is square integrable , it follows from ( [ expansion1 ] ) that there exists @xmath493 such that @xmath494 moreover @xmath495 and @xmath496 are also exponentially decaying . applying greens formula to @xmath67 , we get @xmath497 which implies that @xmath498 the intersection @xmath499 is the space @xmath500 of square integrable harmonic forms .
on @xmath479 we introduce an inner product as follows . for @xmath501 let @xmath502 to verify that this is an inner product , we only need to show that @xmath503 implies @xmath306 .
so suppose that @xmath504 .
then , in particular , we have @xmath505 and the unique continuation property for harmonic forms ( see ( * ? ? ?
* corollary 3 ) ) implies @xmath306 .
we note that the inner product can be also defined by the following formula : @xmath506 this inner product coincides on the subspace @xmath282 with the usual inner product on @xmath282 .
the orthogonal projections define canonical maps @xmath507 moreover , we have the maps @xmath508 next we define maps into the de rham cohomology .
. by ( [ closed ] ) , @xmath46 is closed and we get a canonical map @xmath510 now consider @xmath511 . by
( [ closed ] ) @xmath140 is closed and on the cylinder @xmath140 is of the form @xmath512 where @xmath513 is exponentially decaying .
let @xmath157 be a function with support on the cylinder @xmath146 which is equal to 1 outside a compact set .
following @xcite we can then define a map @xmath514.\end{gathered}\ ] ] this map is well defined and independent of the choice of @xmath157 . indeed changing @xmath157 on a compact subset
changes @xmath515 by the differential of a compactly supported form .
let @xmath516 be the canonical pairing defined by @xmath517,[\psi])=\int_x\phi\wedge \psi,\quad [ \phi]\in h^p_c(x),\ ; [ \psi]\in h^{n - p}(x).\ ] ] define a pairing @xmath518 by taking the constant term in the asymptotic expansion of @xmath519 as @xmath74 .
applying green s formula to @xmath67 , it follows that there exists @xmath493 such that @xmath520,[\phi ] ) + o(e^{-c a } ) \end{gathered}\ ] ] as @xmath74 .
this implies that the following diagram commutes @xmath521 \obj(20,40)[a]{$\mathcal{h}_{ext , rel}^p(x)$ } \obj(25,40)[b]{$\times$ } \obj(30,40)[d]{$\mathcal{h}^{n - p}_{ext , abs}(x)$ } \obj(40,35)[e]{${{\mathbb{r}}}$ } \obj(20,30)[f]{$h^p_c(x,{{\mathbb{r}}})$ } \obj(25,30)[g]{$\times$ } \obj(30,30)[h]{$h^{n - p}(x,{{\mathbb{r}}})$ } \mor{d}{e } { } \mor{h}{e } { } \mor{a}{f}{$r_c$ } \mor{d}{h}{r } \enddc,\ ] ] where the horizontal maps are given by the corresponding pairing .
let @xmath522 and @xmath523 .
since @xmath46 is co - closed ( [ closed ] ) , it follows that @xmath524 .
therefore , for @xmath522 , @xmath375\in h^p_c(x)$ ] , and @xmath525 $ ] , the inner product @xmath526 is independent of the representative of the cohomology class @xmath375 $ ] and will be denoted by @xmath527,\phi\rangle$ ] .
this leads to the following alternative description of the inner product in @xmath528 .
[ innpro3 ] for all @xmath529 we have @xmath530 applying stoke s theorem and using that @xmath513 is rapidly decreasing , we get @xmath531 by ( [ innprod2 ] ) we get @xmath532 our next goal is to describe the connecting homomorphism @xmath533 on the level of harmonic forms . to this end we need some preapration .
let @xmath140 be in the @xmath34 eigenspace of @xmath25 . then by ( [ expansion2 ] ) , we have on @xmath146 @xmath534 where @xmath513 is exponentially decaying .
let @xmath157 be a smooth function with support in @xmath146 which is a equal to 1 outside a compact set .
then @xmath535 is equal to @xmath536 outside a compact set and we conclude that @xmath537 represents @xmath538 $ ] in @xmath539 .
let @xmath540 be the canonical isomorphism .
then we have shown that for each @xmath541 we have @xmath542,\ ] ] where @xmath543 is defined by ( [ maps1 ] ) .
[ invert ] the operator @xmath544 in @xmath545 is invertible . differentiating equations ( ii ) and ( iii ) of theorem [ main22 ] , it follows that @xmath544 commutes with @xmath25 and anti - commutes with @xmath237 .
therefore , it suffices to show that the restriction of @xmath544 to the @xmath34-eigenspace @xmath546 of @xmath25 is invertible .
. then @xmath548 . by theorem [ imageofs ]
we have @xmath549 . using ( [ equ5 ] )
, it follows that it suffices to show that @xmath550 whenever @xmath551 . by lemma [ innpro3 ]
we have @xmath552 for all @xmath553 .
recall that @xmath543 is a harmonic form , which is non - zero , if @xmath551 .
therefore , the left hand side of the above equality is non - zero , if @xmath551 .
now we define maps @xmath554 as follows .
let @xmath555 be the composition of the orthogonal projection @xmath556 and the inclusion @xmath557 .
@xmath558 assigns to @xmath559 its limiting value @xmath560 . to define @xmath561
, we note that by lemma [ invert ] , @xmath562 is an invertible operator .
put @xmath563 the sequence @xmath564
\obj ( 0,20)[d]{$\ldots$ } \obj(10,20)[a]{${\mathcal{h}}^p_{ext , rel}(x)$ } \obj(20,20)[b]{${\mathcal{h}}^p_{ext , abs}(x)$ } \obj(30,20)[c]{${\mathcal{h}}^p(y)$ } \obj(40,20)[e]{${\mathcal{h}}^{p+1}_{ext , rel}(x)$ } \obj(50,20)[f]{$\ldots$ } \mor{d}{a}{$\tilde\partial$ } \mor{a}{b}{$\tilde e$ } \mor{b}{c}{$\tilde r$ } \mor{c}{e}{$\tilde\partial$ } \mor{e}{f}{$\tilde e$ } \enddc\ ] ] is exact .
let @xmath565 .
by proposition [ limiting ] and ( [ niceform ] ) it follows that @xmath566 since @xmath562 preserves @xmath567 , we get @xmath568 . by definition
we have @xmath569 and this is also equal to @xmath570 .
finally by corollary [ orthogonal ] it follows that @xmath571 . on the other hand , by definition we have @xmath572 .
thus @xmath573 . using the definition of @xmath561 and ( [ equ5 ] )
, it follows that @xmath574 by @xcite every element in the image of @xmath575 in @xmath576 can be represented by a unique square integrable harmonic form . using these facts we obtain the following commutative diagram .
@xmath564 \obj ( 0,30)[a]{$\ldots$ } \obj(10,30)[b]{$\mathcal{h}^p_{ext , rel}(x)$ } \obj(10,40)[cc]{$\mathcal{h}^p(y)$ } \obj(20,40)[c]{$\mathcal{h}^p_{(2)}(x)$ } \obj(20,30)[d]{$\mathcal{h}^p_{ext , abs}(x)$ } \obj(30,30)[e]{$\mathcal{h}^p(y)$ } \obj(30,40)[ee]{$\mathrm{ker}(s(0)-{1 \hspace{-1 mm } \mbox{\sf i}})$ } \obj(40,40)[ccc]{$\mathcal{h}^{p}(y)$ } \obj(40,30)[f]{$\mathcal{h}^{p+1}_{ext , rel}(x)$ } \obj(50,30)[g]{$\ldots$ } \obj ( 0,20)[h]{$\ldots$ } \obj(10,20)[i]{$h^p_c(x,{{\mathbb{r}}})$ } \obj(20,20)[j]{$h^p(x,{{\mathbb{r}}})$ } \obj(30,20)[k]{$h^p(y,{{\mathbb{r}}})$ } \obj(40,20)[l]{$h^{p+1}_c(x,{{\mathbb{r}}})$ } \obj(50,20)[m]{$\ldots$ } \mor{h}{i}{$\partial$ } \mor{i}{j}{$e$ } \mor{j}{k}{$r$ } \mor{k}{l}{$\partial$ } \mor{l}{m}{$e$ } \mor{a}{b}{$\tilde\partial$ } \mor{c}{b } { } \mor{c}{d } { } \mor{d}{e}{$\tilde r$ } \mor{e}{f}{$\tilde\partial$ } \mor{f}{g}{$\tilde e$ } \mor{b}{i}{$r_c$ } \mor{d}{j}{$r$ } \mor{e}{k}{$\kappa$ } \mor{f}{l}{$r_c$ } \mor{ee}{e } { } \mor{ee}{d}{$\hat f$ } \mor{b}{d}{$\tilde e$ } \mor{cc}{a}{$\frac{{\mathrm{i}}}{2 } s'(0)$ } \mor{cc}{b}{$\hat g$ } \mor{ccc}{e}{$\frac{{\mathrm{i}}}{2 } s'(0)$ } \mor{ccc}{f}{$\hat g$ } \enddc\ ] ] the maps @xmath577 are isomorphisms .
we first consider @xmath578 .
let @xmath579 .
by @xcite , @xmath578 induces an isomorphism of @xmath282 onto @xmath580 .
let @xmath559 and suppose that @xmath581 .
then it follows that @xmath582 .
hence @xmath583 . since @xmath578 is an isomorphism on @xmath584 , we get @xmath306 .
this proves injectivity .
let @xmath585 .
using @xmath586 and theorem [ imageofs ] , it follows that @xmath587 .
thus by ( [ imageofr ] ) there exists @xmath509 such that @xmath588 . then @xmath589 . hence @xmath590 . by the above remark there is @xmath591 such that @xmath592 .
thus @xmath578 is surjectiv and hence an isomorphism . applying the commutativity of the diagramm and the 5-lemma
, it follows that @xmath593 is an isomorphism too .
this can also be proved by slightly different methods ( e.g. @xcite ) . in every class in @xmath594
there is a unique representative in @xmath595 .
[ extrel ] for every class @xmath375 $ ] in @xmath596 there is a unique element @xmath597 in @xmath598 such that for any @xmath599 : @xmath600,[\phi ] \rangle = \langle \hat\psi , \phi \rangle .
\end{gathered}\ ] ] moreover , the map @xmath601 \mapsto \hat \psi \end{gathered}\ ] ] is an isomorphism .
we can now consider the scattering length @xmath602 let @xmath603 be the connecting homomorphism .
we identify @xmath604 with @xmath605 and @xmath539 with @xmath606 via corollary [ extrel ] .
thus we may regard the connecting homomorphism as a map @xmath607 let @xmath608 be the orthogonal complement of @xmath609 .
[ main3 ] the scattering length @xmath33 is a positive , invertible operator in @xmath605 .
it is uniquely determined by the following conditions .
@xmath610 by theorem [ imageofs ] , @xmath611 equals the @xmath24 eigenspace of @xmath25 .
therefore @xmath612 equals the @xmath34 eigenspace of @xmath25 . by theorem [ main22 ] , @xmath237 anti - commutes with @xmath25 .
hence it interchanges the @xmath257-eigenspaces .
@xmath237 also anti - commutes with @xmath562 and consequently the scattering length @xmath33 commutes with the hodge star operator .
it is therefore completely determined by its restriction to @xmath612 .
let @xmath613 . using ( [ equ5 ] ) and lemma [ innpro3 ] , we have @xmath614 let @xmath615 be the pull - back of @xmath46 to @xmath146 and let @xmath616 such that @xmath617 on @xmath618\times y$ ] and @xmath619 outside a compact set .
then @xmath620 represents @xmath621\in h^{p+1}_c(x,{{\mathbb r}})$ ] . by the same argument as in the proof of lemma [ innpro3 ] we get @xmath622 by ( [ limitvalue ] ) , @xmath623 is an extended harmonic form with limiting value @xmath378 . using stokes theorem ,
applied to @xmath67 , it follows that @xmath624 this concludes the proof of the theorem . for the scattering length @xmath625
we have the following formula @xmath626 this implies that @xmath627 is equal to the orthogonal projection onto the orthogonal complement of @xmath628 .
by the results of the previous sections we have canonical isomorphisms @xmath629 in this section we establish relations between some norms on these spaces . on the cohomology groups @xmath376 and @xmath630 there is the so - called comass norm ( see ( * ? ? ?
4c ) ) which is defined as follows .
if @xmath631 is a finite dimensional inner product space , then @xmath632 has a natural inner product as well and we denote the norm that is induced by this inner product by @xmath633 .
the comass norm @xmath634 on @xmath632 is defined by @xmath635 since the norms are equivalent there is a constant @xmath636 such that @xmath637 and we denote by @xmath638 the optimal such constant .
since all @xmath639-dimensional inner product spaces are unitarily equivalent the constant depends only on @xmath639 and @xmath0 .
of course ( see also @xcite ) , @xmath640 moreover , since the hodge star operator leaves the space of primitive forms invariant , we have @xmath641 it is also known that @xmath642\end{gathered}\ ] ] ( see @xcite ) .
now let @xmath643 be a differentiable manifold .
let @xmath644 .
the comass @xmath645 of @xmath117 is defined by @xmath646 for a compact manifold @xmath643 with smooth boundary @xmath647 this induces a norm on @xmath648 by @xmath649,\ ; \omega\in\lambda^p(b,\partial b),\,d\omega=0\},\end{gathered}\ ] ] which we also refer to as the comass norm . to compare the norms on the various cohomology groups , we need some preparation .
let @xmath650 .
we define an extension @xmath651 of @xmath140 in the following way .
the restriction @xmath652 can be expanded into eigensections of the laplace - beltrami operator on @xmath3 : @xmath653 where @xmath46 is harmonic and @xmath654 is an orthonormal basis in the orthogonal complement of the space of harmonic forms such that @xmath655 now define @xmath656 as usually @xmath657 .
the map @xmath658 is of course linear and maps into the space of bounded sections .
note that in general @xmath597 is not continuous .
however , it satisfies @xmath659 therefore , using green s formula , it follows that the distributional derivative @xmath660 satisfies @xmath661 in particular @xmath660 is again bounded . if @xmath140 is closed , then @xmath597 is also closed . note that the extension map @xmath662 is chosen so that it inverts the restriction operator on the space @xmath595 .
namely , for @xmath663 it follows from ( [ expans6 ] ) that @xmath664 we can now establish the comparison results for the norms . [ hilfreichlemma ] let @xmath650 be closed and @xmath665 a harmonic form such that @xmath666 and @xmath140 represent the same element in @xmath376 .
let @xmath667 be the unique harmonic representative of the cohomology class of @xmath652 .
then @xmath668 where @xmath669 is the smallest positive eigenvalue of @xmath670 .
in particular @xmath671 \|_\infty^2 , \end{gathered}\ ] ] where we define the effective volume @xmath672 by @xmath673 let @xmath665 .
let @xmath674 be a closed form such that @xmath675 for some @xmath676 .
denote by @xmath46 the unique harmonic representative of the class of @xmath652 .
if @xmath677 is the characteristic function of @xmath678 , then the norm of @xmath133 is by definition the @xmath69-norm of @xmath679 .
therefore , @xmath680 now observe that by ( [ closed2 ] ) and the definition of @xmath597 , the expansion of @xmath681 on @xmath146 contains no harmonic form .
hence by ( [ estimation8 ] ) and ( [ extension1 ] ) the restriction of @xmath682 to @xmath146 is exponentially decreasing .
this implies @xmath683 by ( [ closed2 ] ) and ( [ derivative ] ) we have @xmath684 , where the latter means the distributional derivative .
since @xmath133 is closed and coclosed , it follows that @xmath685 therefore we get @xmath686 on the other hand @xmath687 [ hilfreichlemma2 ] let @xmath688 .
let @xmath689 be a representative of the class @xmath375 \in h^p(m , y,{{\mathbb{r}}})$ ] which corresponds to @xmath46 with resepect to the isomorphism ( [ caniso ] ) .
then @xmath690 and in particular @xmath691 \|_\infty^2 .
\end{gathered}\ ] ] let @xmath688 .
recall the definition of @xmath593 by ( [ derham1 ] ) .
choose @xmath157 such that the support of @xmath692 is contained in @xmath1 .
then @xmath693 $ ] is the image of @xmath46 w.r.t .
the isomorphism ( [ caniso ] ) .
since @xmath694 represents this cohomology class , there is @xmath695 such that @xmath696 let @xmath403 ( resp .
@xmath697 ) be the differential form on @xmath7 which is equal to @xmath140 ( resp .
@xmath117 ) on @xmath1 and @xmath22 on @xmath146 .
then @xmath698 by the definition ( [ innprod1 ] ) of the norm in @xmath528 , we have @xmath699 by ( [ equ6 ] ) we have @xmath700 since @xmath695 , it follows from green s formula that @xmath701 similarly , by green s formula and ( [ restrict ] ) we have @xmath702 thus @xmath703 and by ( [ ungleichungsstifter ] ) we get @xmath704
we recall some notions from geometric measure theory .
suppose @xmath643 is a compact oriented riemannian manifold and let @xmath705 be a closed submanifold
. in our case @xmath705 will be either the boundary of @xmath643 or the empty set . for @xmath706 let the minimal volume be defined as the infimum of the volumes of all its representatives , i.e. @xmath707,\;\alpha_i\in{{\mathbb z}}\}.\end{gathered}\ ] ] where the infimum is over all lipschitz continuous simplices @xmath708 .
the stable norm @xmath709 of an element @xmath710 is defined similarly by @xmath711,\;\alpha_i\in{{\mathbb r}}\}.\end{gathered}\ ] ] this defines indeed a norm ( see @xcite , and @xcite , also ( * ? ? ?
4c ) , and @xcite ) .
the stable norm of an element in @xmath706 is by definition the stable norm of its image in @xmath712 .
clearly , @xmath713 and equality does not hold in general .
however , as shown by federer ( see @xcite ) , @xmath714 moreover , for @xmath715 we have equality , i.e. @xmath716 . by a general result in geometric measure theory the stable norm and the comass are dual to each other , i.e. , @xmath717 ( see @xcite , and also @xcite , and @xcite for a sketch of the proof in the case without boundary ) .
now we will apply this result to our problem in the case where @xmath719 and @xmath720 .
we equip @xmath721 with the norm induced from @xmath605 by the de rham isomorphism .
suppose that @xmath117 is a @xmath0-form on @xmath3 . then using ( [ estconst ] )
one obtains @xmath722 now let @xmath46 be a harmonic @xmath0-form representing an element @xmath335 \in h^p(y,{{\mathbb{r}}})$ ] . using ( [ inform ] ) , we get @xmath723 \cup \alpha , [ y ] \rangle \mid \alpha \in h^{n - p-1}(y,{{\mathbb{r } } } ) , \| \alpha \|_\infty \leq 1 \right\ } \end{split}\end{gathered}\ ] ] since the stable norm is dual to the comass norm , we finally get @xmath724 \cap \phi \|_{st},\end{gathered}\ ] ] for any @xmath725 . on the other hand the inequality ( [ inform ] )
may also be used directly . since @xmath726 is the infimum of the @xmath69-norms of all representatives of the cohomology class @xmath335 $ ] , we have [ randabsch ] the hilbert space norm on @xmath362 , induced from the harmonic forms , satisfies @xmath727 \cap \phi \|_{st } \leq \|\phi\|_2 \leq c(n-1,p)^{1/2 } \mathrm{vol}(y)^{1/2 } \|\phi \|_\infty.\end{gathered}\ ] ] we can use lemmas [ hilfreichlemma ] and [ hilfreichlemma2 ] in the same way as above to get similar estimates of the norms of elements in @xmath729 . first note that @xmath237 induces an isomorphism @xmath730 let @xmath731 be the pairing ( [ pairing ] )
. then we have @xmath732 let @xmath733 .
then we get @xmath734 choose a representative @xmath46 of the cohomology class @xmath735 with @xmath736
. then @xmath737 $ ] .
since the diagramm ( [ diagram ] ) commutes , we get @xmath738,[*\psi|_m ] ) , \quad \phi,\psi\in h^p_{ext , rel}(x).\ ] ] using lemma [ hilfreichlemma ] , it follows from ( [ norm10 ] ) that @xmath739,[\psi|_m ] ) \mid \psi\in \mathcal{h}^{n - p}_{ext , abs}(x),\ , \|\psi\|\le 1\}\\ & \ge c(n , p)^{-1/2 } \mathrm{vol}_*(m)^{-1/2 } \sup \ { \langle [ \phi ] \cup \alpha , [ m ] \rangle \mid \alpha \in h^{n - p}(m,{{\mathbb{r } } } ) , \| \alpha \|_\infty \leq 1 \}. \end{split}\ ] ]
since the stable norm is dual to the comass norm , we finally get @xmath740 \cap [ \phi ] \|_{st},\end{gathered}\ ] ] note that the poincare - lefschetz dual @xmath741 \cap [ \phi]$ ] of @xmath335 $ ] is in @xmath742 and its stable norm equals its stable norm as an element of @xmath742 .
lemma [ hilfreichlemma2 ] gives an upper bound for @xmath743 and we get let @xmath744 and @xmath745 .
then we have @xmath746 \cap \phi \|_{st } \leq \|f\|
\leq c(n , p)^{1/2 } \mathrm{vol}(m)^{1/2 } \|\phi \|_\infty .
\end{gathered}\ ] ] using theorem [ main3 ] , we obtain theorem [ main00 ] .
if we denote by @xmath58 the inclusion map @xmath747 and by @xmath748 the induced map in homology , then @xmath749 \cap \partial \phi = \iota_*\left([y ] \cap \phi \right).\end{gathered}\ ] ] that is @xmath741 \cap \partial \phi$ ] coincides with the image of the poincare dual of the class @xmath46 in @xmath750 . the stable norm of @xmath751 can be calculated in many cases explicitly in terms of geometric data using the fact that the stable mass is dual to the comass norm . in order to make statements about the spectrum of the map @xmath33 one can combine these estimates with the estimates of the @xmath69-norms of the cohomology classes on @xmath3 ( see proposition [ randabsch ] ) .
note that the extended harmonic functions are exactly the constant functions .
therefore , @xmath752 is spanned by the function equal to @xmath158 on @xmath3 .
thus , the @xmath24 eigenspace to @xmath753 is spanned by @xmath158 . moreover , since @xmath268 anticommutes with the hodge @xmath237 operator we have @xmath754 . by stokes formula @xmath755)[m]= [ * 1][y]\end{gathered}\ ] ] and
therefore @xmath756 = \frac{\mathrm{vol}(y)}{\mathrm{vol}(m ) } [ * _ m 1]\end{gathered}\ ] ] and by the above equation we immediately obtain @xmath757 this in turn implies that @xmath758 in this case @xmath759 consists of the constant functions only . by the above we have @xmath760 this in turn implies that @xmath761 note that if @xmath762 the only non - vanishing cohomology groups are @xmath763 and @xmath764 .
thus , in this case the above formulas determine @xmath33 completely .
now we have @xmath766 and therefore on @xmath767 is a direct sum of the one dimensional spaces @xmath768 and @xmath769 . under the splitting @xmath770
the operator @xmath52 is of the form @xmath53 where we have already seen , that @xmath771 our formula now allows us to give an estimate of @xmath63 in purely geometric terms .
the harmonic functions in @xmath768 are multiples of the function @xmath158 on @xmath3 .
the complement @xmath772 has dimension @xmath158 and is spanned by the function @xmath773 clearly , @xmath774 it remains to estimate the @xmath69-norm of the class @xmath775 in @xmath776 .
the comass norm of @xmath775 may be calculated using the duality between the stable norm and the comass norm .
let @xmath777 be a lipschitz continuous chain whose boundary is homologuous to @xmath778 , where @xmath779 and @xmath780 .
then @xmath777 defines a relative cycle @xmath781 $ ] in @xmath782 and we have @xmath783 ) = \int_c \partial \chi = \mathrm{vol}(y_1 ) + \mathrm{vol}(y_2)=\mathrm{vol}(y ) .
\end{gathered}\ ] ] we observe that the cycle @xmath777 can be written as a linear combination of curves which are either closed or whose end points are in the boundary .
this implies @xmath784 \|_{st } \geq \mathrm{dist}(y_1,y_2),\end{gathered}\ ] ] with equality if for example @xmath781 $ ] is in the same homology class as a shortest curve connecting @xmath49 with @xmath50 .
duality implies @xmath785 and therefore @xmath786 to get the estimate from above we need to look at @xmath787 \cap \partial \chi \|_{st}$ ] , that is the stable norm of the poincare dual of the class @xmath775 . by equation ( [ anders ] ) we have @xmath741 \cap \partial \chi = \iota_*([y ] \cap \chi)$ ] . of course ,
@xmath788 \cap \chi ) = \iota_*(\mathrm{vol}(y_2)[y_1 ] - \mathrm{vol}(y_1)[y_2 ] ) = \mathrm{vol}(y ) \iota_*([y_1]),\end{gathered}\ ] ] where we have used that @xmath789)+\iota_*([y_2])=0 $ ] . therefore , @xmath790)\|_{st}.\end{gathered}\ ] ] combining these two estimates we obtain @xmath791 with @xmath57)\|_{st}^2 ( \mathrm{vol}(y_1)+ \mathrm{vol}(y_2))},\\ c_2= 2 \mathrm{vol}(m)^{-1 } \frac{\mathrm{dist}(y_1,y_2)^2 \mathrm{vol}(y_1)\mathrm{vol}(y_2 ) } { \mathrm{vol}(y_1)+ \mathrm{vol}(y_2)}.\end{gathered}\ ] ] note that with respect to this basis , the scattering matrix at zero has the form @xmath792 in order to interpret this in terms of reflection and transmission coefficients we choose another basis @xmath793 of @xmath767 , where @xmath794 is constant equal to @xmath158 on @xmath795 and equal to zero on the other boundary component . in this basis
we get @xmath796 the reflection coefficient @xmath59 and the transmission coefficient @xmath60 of a wave of low energy that comes in at the end @xmath49 and is scattered in @xmath1 therefore are @xmath797 and @xmath798 . transforming @xmath52 into this basis
gives @xmath799 as a remark we would like to add here that in physics the scattering matrix for one dimensional scattering problems has the transmission coefficients on the diagonal and the reflection coefficients off the diagonal .
our notation differs here as we consider the operator with neumann boundary conditions at each end as the unperturbed operator .
let @xmath1 be the full torus @xmath800 with boundary @xmath801 .
we view both @xmath802 and @xmath803 as subsets of the complex planes and use coordinates @xmath804 on @xmath802 and @xmath805 on @xmath803 .
we assume that we are given a metric on @xmath1 which has product structure in a small neighborhood of the boundary and that the metric on the boundary is equal to the product metric @xmath806 with positive real numbers @xmath807 .
then , the volume is @xmath808 .
moreover , @xmath809 and it is gerated by the class of the one form @xmath810 .
the group @xmath811 is isomorphic to @xmath812 and is generated by the classes of the two harmonic @xmath158-forms @xmath810 and @xmath813 .
the @xmath69-norms of these forms is easily calculated .
@xmath814 and , moreover , @xmath810 and @xmath813 are orthogonal to each other .
the restriction of the form @xmath810 on @xmath1 to @xmath3 is the form @xmath810 regarded as a form on @xmath3 .
therefore , the kernel of the connecting homomorphism @xmath334 is spanned by @xmath815 $ ] .
hence , @xmath816 is spanned by @xmath817 $ ] .
since the hodge star operator commutes with @xmath818 the map @xmath818 has the form @xmath819 with respect to the decomposition @xmath820 . in order to get
an estimate for @xmath821 we need to calculate @xmath822\|_\infty$ ] and @xmath823 \cap \partial [ dy]|_{st}$ ] .
the stable norm of @xmath822\|_\infty$ ] can again be calculated by duality . for this note that @xmath824 by alexander duality and it is generated by the relative cycle @xmath825 .
now , of course @xmath826,[\alpha ] \rangle= \int_{\partial d \times \{1\ } } dy = 2 \pi.\end{gathered}\ ] ] duality of comass norm and stable norm now implies that @xmath827\|_{\infty}=2 \pi \|[d \times \{1\}]\|_{st}^{-1}.\end{gathered}\ ] ] the class @xmath741 \cap \partial [ dy]$ ] is given by @xmath828 .
so @xmath829 \cap \partial [ dy]\|_{st}$ ] is equal to the infimum of the lengths of all representatives of the cycle @xmath830 in @xmath831 .
the geometric picture is described in fig .
[ fiig02 ] . and
@xmath832 on @xmath1.,width=264 ] theorem [ main00 ] now gives an estimate of @xmath821 : @xmath833 since @xmath834 we have @xmath835 .
in this appendix we discuss the relation between the stationary and the dynamical approach to scattering theory and we establish the eisenbud - wigner formula for manifolds with cylindrical ends . for details concerning scattering theory
we refer to @xcite . for original papers on the eisenbud - wigner formula see @xcite and @xcite . in order to simplify the relation of the scattering length to the time - delay operator we consider scattering theory for the square root of the laplace beltrami operator , i.e. we consider scattering theory and the time - delay in relativistic quantum mechanics .
let @xmath836 be the closure in @xmath69 of the operator @xmath79 with domain @xmath837 . denote by @xmath838 the self - adjoint extension of @xmath79 with respect to neumann boundary conditions along @xmath3 .
note that @xmath838 is the self - adjoint operator associated to the quadratic form @xmath839 on @xmath840 .
with respect to the decomposition @xmath841 we have @xmath842 where @xmath843 and @xmath844 are the corresponding self - adjoint extensions of @xmath845 and @xmath846 , respectively , with neumann boundary conditions imposed .
let @xmath437 be the square root of @xmath836 defined by spectral calculus and similarly define @xmath847 , @xmath848 , and @xmath849 . since @xmath1 is compact , @xmath850 has purely discrete spectrum .
the spectral resolution of @xmath851 can be determined , using separation of variables .
the spectrum is absolutely continuous .
a complete set of generalized eigensections is given by @xmath852 where @xmath46 runs through an orthonormal basis of eigensection of the operator @xmath853 with eigenvalue @xmath854 .
denote by @xmath855 ( resp .
@xmath856 ) the orthogonal projection onto the absolutely continuous subspace of @xmath437 ( resp .
@xmath857 ) . denote by @xmath858 and @xmath859 the restrictions of @xmath437 and @xmath857 , respectively , to the absolutely continuous subspaces .
furthermore , for an open interval @xmath860 we denote by @xmath861 ( resp .
@xmath862 ) the projection onto the continuous part of the spectral subspace of the interval . as explained above , the absolutely continuous subspace for @xmath857 is @xmath174 .
thus @xmath863 is the orthogonal projection of @xmath82 onto the subspace @xmath174 . by ( * ?
* thorm 3.6 ) and the birman - kato invariance principle the wave operators @xmath864 exist and are complete .
this means that the strong limit exists and and the operators @xmath865 define isometries of @xmath866 onto @xmath867 , intertwining @xmath859 and @xmath858 . in this context
the scattering operator @xmath268 is defined as @xmath868 this is a unitary operator in @xmath869 which commutes with @xmath859 .
let @xmath870 be the absolutely continuous spectrum of @xmath857 .
it equals @xmath871 with @xmath872 .
let @xmath873 be the spectral family of @xmath859 . since @xmath268 commutes with @xmath859 , we have @xmath874 where @xmath875 acts in the finite - dimensional hilbert space @xmath876 for the purposes of this paper we need only to investigate the structure of the continuous spectrum of @xmath437 in the interval @xmath877 in the situations when @xmath878 .
the generalized eigensections @xmath879 constructed in theorem [ main1 ] are well defined as distributions on @xmath880 with values in @xmath82 .
this follows easily from the properties of the fourier transform and the expansion of @xmath879 on @xmath146 .
of course , in the weak topology on the space of distributions we have @xmath881 the smooth function @xmath882 can be determined as the continuous extension of @xmath883 this integral can be simplified using green s formula and the limit @xmath74 can be explicitly evaluated ( see equ .
( [ parequa ] ) ) .
a simple exercise in distribution theory shows that indeed we have the orthogonality relations @xmath884 as distributions on @xmath885 .
moreover , in the same way as in @xcite , theorem 6.2 , one shows that @xmath886 therefore , in the distributional sense for @xmath887 @xmath888 which shows that the dynamical and the stationary scattering matrices coincide @xmath889 the time delay operator @xmath890 is defined in the following way .
if @xmath891 then , according to the laws of quantum mechanics , the probability of finding the particle with wave function @xmath46 in @xmath67 at time @xmath821 is given by @xmath892 the total time spent in @xmath67 is then given by @xmath893 this expression is not necessarily finite for all @xmath46 .
now , according to scattering theory , for @xmath894 the states @xmath895 and @xmath896 are asymptotically the same for @xmath897 . thus
, the time excess due to the interaction ( the presence of m ) is @xmath898 the eisenbud - wigner time - delay operator @xmath890 is formally defined by @xmath899 in many situations in potential scattering it can be shown that the above defines a closable quadratic form and @xmath890 is a self - adjoint operator that commutes with @xmath857 and the eisenbud - wigner formula @xmath900 holds .
since we are only interested in the spectrum near @xmath22 , we prove this formula only for elements in @xmath901 .
suppose that @xmath902 , @xmath903 .
let @xmath904 .
then , @xmath905 moreover , @xmath906 where @xmath907 hence , @xmath890 is a self - adjoint operator on @xmath908 which has the form @xmath909 with respect to the spectral family @xmath873 of @xmath859 .
we can do this calculation in the explicit spectral decomposition .
@xmath910 and similarly , @xmath911 in the limit @xmath74 both integrals can be computed up to exponentially small errors since we have the expansions on @xmath146 : @xmath912 therefore , integration by parts yields @xmath913 similarly , we have @xmath914 therefore , @xmath915 and one obtains @xmath916 where the second limit is in the distributional sense and @xmath917 by the properties of the scattering matrix @xmath918 commutes with the hodge star operator and leaves the summands in @xmath919 invariant ( see ( [ invsummands ] ) ) .
we therefore have @xmath29 where @xmath30 is the time delay operator for coclosed forms defined by @xmath920 this operator describes the time - delay of coclosed forms of energy @xmath921 scattered in @xmath1 . in the physics literature the time delay of the @xmath922 partial wave for potential scattering in @xmath923 is called the scattering length ( e.g. @xcite ) . as the @xmath922 partial wave corresponds to the constant function on the sphere at infinity we call @xmath32 in analogy with this the scattering length .
note however , that in the physics literature the time - delay is usually considered for non - relativistic schrdinger mechanics so that it differs from the relativistic time - delay by an energy dependent factor .
a simple relation that equates time - delay and scattering length can for dimensional reasons only hold in relativistic theories .
misha gromov , _ metric structures for riemannian and non - riemannian spaces _ , progress in mathematics , vol .
152 , birkhuser boston inc . ,
boston , ma , 1999 , based on the 1981 french original [ mr0682063 ( 85e:53051 ) ] , with appendices by m. katz , p. pansu and s. semmes , translated from the french by sean michael bates . | scattering theory for @xmath0-forms on manifolds with cylindrical ends has a direct interpretation in terms of cohomology . using the hodge isomorphism ,
the scattering matrix at low energy may be regarded as operator on the cohomology of the boundary .
its value at zero describes the image of the absolute cohomology in the cohomology of the boundary .
we show that the so - called scattering length , the eisenbud - wigner time delay at zero energy , has a cohomological interpretation as well .
namely , it relates the norm of a cohomology class on the boundary to the norm of its image under the connecting homomorphism in the long exact sequence in cohomology .
an interesting consequence of this is that one can estimate the scattering lengths in terms of geometric data like the volumes of certain homological systoles . |
in the study of the evolution of the universe , it is important to understand the origin of the cosmic density fluctuations .
the most conventional scenario is inflation @xcite where quantum fluctuation of the inflaton field during the inflation becomes the origin of the cosmic density fluctuations . from the particle - physics point of view , it is desirable to construct a model of inflation which is testable by collider ( or other laboratory ) experiments ; if it is possible , we can study the mechanism of the cosmological density fluctuations with collider experiments .
it is , however , difficult to construct such a testable model of inflation since the requirements on the inflaton potential is very stringent.within the minimal supersymmetric standard model , however , it may be possible to use some of the scalar quarks and higgs bosons as the inflaton . for detail , see @xcite . recently , a new mechanism of generating the cosmic density fluctuations has been attracting many attentions , where a late - decaying scalar condensation provides the dominant source of the cosmic density fluctuations . in this scenario , dominant part of the cosmic density fluctuations
originate from the primordial fluctuation of a new scalar field , called `` curvaton , '' which is different from the inflaton field @xcite .
even though , in a large class of the curvaton scenario , inflation is assumed as a solution to the horizon and flatness problems as well as to generate the primordial fluctuation of the curvaton field , constraints on the inflaton potential can be relaxed in the curvaton scenario . in addition , since the requirements on the curvaton is not so stringent , it is possible to use ( some of the ) well - motivated particles as the curvaton ; for example , scalar fields in the minimal supersymmetric standard model ( mssm ) may play the role of the curvaton .
in such a case , it may be possible to study the properties of the fields responsible for the structure formation by collider experiments .
( for detailed studies of the curvaton scenario , see @xcite . )
here , i will review the curvaton scenario and discuss some of its implications to particle cosmology .
organization of the rest of this article is as follows . in section [ sec : history ] , i first discuss the thermal history we consider .
then , in section [ sec : mechanism ] , i explain the curvaton scenario .
the cosmic microwave background ( cmb ) angular power spectrum generated in the curvaton scenario is studied in section [ sec : cl ] . in section [ sec : implications ] , implications to the sneutrino cosmology is discussed .
section [ sec : conclusion ] is devoted for conclusions and discussion .
i first discuss the basic scenario . in the curvaton scenario , there are two scalar fields which play important roles ; one is the inflation @xmath0 and the other is the curvaton field @xmath1.i assume the inflation to solve the horizon and flatness problems .
the curvaton mechanism may be implemented in the pre - big - bang @xcite and the ekpyrotic @xcite scenarios . in this article , however , i will not discuss those cases .
the most important features of the curvaton scenario do not depend on the details of the inflation .
thus , i do not assume any specific form of the inflaton potential .
on the contrary , resultant density fluctuations depend on the potential of the curvaton field . here , for simplicity , i adopt the simplest form of the curvaton potential , i.e. , the parabolic one : @xmath2 where @xmath3 is smaller than the expansion rate of the universe during the inflation @xmath4 . in addition , the curvaton field is assumed to have non - vanishing initial amplitude @xmath5 . in the scenario ,
the universe starts with the inflationary epoch and , after the inflation , the universe is reheated by the decay of the inflaton .
then , the universe is dominated by the radiation ( which i call @xmath6 ) .
i call the radiation - dominated epoch filled with radiation generated from the decay products of the inflaton as the first radiation dominated epoch , or rd1 epoch , since , in the curvaton scenario , there are two radiation - dominated epochs . denoting the decay rate of the inflaton as @xmath7 , reheating temperature after the inflation is given byin the following discussions , i neglect @xmath8 coefficients which are not important .
@xmath9 where @xmath10 is the reduced planck scale . in the early stage of the rd1 epoch ,
slow - roll condition is satisfied for the curvaton field .
as the universe expands , however , the curvaton field starts to oscillate .
when @xmath11 ( with @xmath12 being the decay rate of @xmath1 ) , energy density of @xmath1 behaves as that of non - relativistic matter .
then , the energy density of the curvaton @xmath13 is proportional to @xmath14 while the energy density of @xmath6 is proportional to @xmath15 , where @xmath16 is the scale factor .
thus , as the universe expands , curvaton dominates the universe ( if the lifetime of the curvaton is long enough ) .
i call this epoch as curvaton - dominated epoch , or @xmath1d epoch .
the curvaton field decays when the expansion rate becomes comparable to the decay rate of the curvaton .
the reheating temperature due to the curvaton decay is given by @xmath17
in the curvaton scenario , @xmath1 also acquires the primordial fluctuation during the inflation . denoting the curvaton field ( with comoving momentum coordinate @xmath18 ) ashere
, i use the same notation for the zero - mode and for the total amplitude since there should be no confusion .
@xmath19 let us consider the two - point correlation function of the fluctuation : @xmath20 then , the fourier amplitude generated during the inflation is given by @xmath21_{k = ah_{\rm inf } } , \nonumber \\ \label{dphi_init}\end{aligned}\ ] ] where the subscript `` @xmath22 '' implies that the quantity is evaluated at the time of the horizon - exit during the inflation . as i discuss below , the primordial fluctuation of @xmath1 given in eq.([dphi_init ] ) becomes the origin of the cosmic density fluctuations in the curvaton scenario .
now , we are at the position to discuss the evolutions of the cosmological density fluctuations generated from @xmath23 . for this purpose ,
i use the newtonian gauge where the line element is described with the metric perturbations @xmath24 and @xmath25 ashere , i use the notation and convention of @xcite unless otherwise mentioned .
@xmath26 , \label{ds}\end{aligned}\ ] ] where @xmath27 is the conformal time .
in addition , the variable @xmath28 is defined as @xmath29 where the subscript @xmath30 denotes the individual components ( like radiation , cold dark matter ( cdm ) , baryon , and so on ) and @xmath31 is the fluctuation of the energy density of @xmath30 .
substituting eq .
( [ ds ] ) into the einstein equation , we obtain the generalized poisson equation for @xmath25 : @xmath32 .
\nonumber \\ \label{poisson}\end{aligned}\ ] ] here , `` tot '' denotes the total matter and the variable @xmath33 denotes the velocity perturbation of the component @xmath30 .
furthermore , @xmath34 is the equation - of - state parameter for the total matter , and @xmath35 i consider the situation where the temperature of the universe is so high that the momentum - exchange of relativistic particles are efficient enough . in this case ,
perturbation of the radiation becomes locally isotropic and the anisotropic stress perturbation vanishes , resulting in @xmath36 .
when the perturbation of the radiation becomes locally isotropic , the equations for the density and velocity perturbations of the radiation are given by @xmath37 where the `` prime '' denotes the derivative with respect to the conformal time @xmath27 .
in addition , if a very weakly interacting non - relativistic component exists , its perturbations obey the following equations : @xmath38 where the subscript `` @xmath39 '' is for non - relativistic matters . notice that , when a scalar field is oscillating , the equation - of - state parameter of the scalar condensation vanishes and hence the density and velocity perturbations of the scalar field also obey eqs .
( [ eq : cdm ] ) and ( [ eq : cdm_velocity ] ) . in the curvaton scenario
, it is assumed that the primordial fluctuation @xmath40 becomes the dominant source of the cosmic density fluctuations .
thus , hereafter , i will discuss the density fluctuations generated from @xmath40 .
when @xmath11 , the situation is like the case with isocurvature fluctuation in non - relativistic matter component @xcite . in the curvaton case , as in the conventional isocurvature case , it is convenient to define the ( primordial ) entropy fluctuation between @xmath1 ( and its decay products ) and the photon from the inflaton @xmath6 : @xmath41_{\rm rd1}.\end{aligned}\ ] ] then , the cosmological density fluctuations are parameterized by using @xmath42 . in order to discuss the evolutions of the fluctuations , it is important to know the equation - of - state parameters of individual components in the universe . if all the components behave as the relativistic or the non - relativistic matter , evolutions of the perturbations are described by eqs .
( [ dr ] ) @xmath43 ( [ eq : cdm_velocity ] ) . in this case , it is convenient to distinguish the photon ( or any other components ) from the decay product of @xmath1 from that from the inflaton field , which i call @xmath44 and @xmath6 , respectively.in fact , these photons are mixed each other and they can not be defined separately . in other words , their velocity perturbations should be the same since they form a single fluid .
even so , the following arguments are unchanged as far as we consider the leading terms in the density perturbations since the velocity perturbation is at most @xmath45 . in the following discussion , @xmath44 and @xmath6
should be understood as representatives of the components which are and are not generated from the decay product of @xmath1 , respectively . in order to consider @xmath46 in the rd2 epoch , we can neglect @xmath6 since the cmb radiation at this epoch is dominantly from the @xmath1 field .
then , we find that , in the rd2 epoch , @xmath24 and @xmath46 become constant while @xmath47 up to higher order corrections . indeed , combining eq .
( [ poisson ] ) with eqs .
( [ dr ] ) and ( [ vr ] ) , and using @xmath48 and @xmath49 , we obtain @xmath50 and @xmath51 where @xmath52 is the metric perturbation induced by the primordial fluctuation of the amplitude of @xmath1 .
( in the following , the superscript `` @xmath53 '' is for perturbations generated from the primordial fluctuation of @xmath1 . ) as i mentioned , @xmath52 is constant up to a correction of @xmath54 .
behavior of @xmath55 is also easily understood . in discussing the effects of the primordial fluctuation of @xmath1
, we neglect the initial fluctuation of the inflaton field and hence @xmath56 in the deep rd1 epoch .
in addition , from eqs .
( [ dr ] ) and ( [ vr ] ) , @xmath57 becomes higher order in @xmath58 than @xmath55 and @xmath59 .
thus , we obtain @xmath60 the above relation holds in the rd1 , @xmath1d , and rd2 epochs up to corrections of @xmath54 . @xmath61_{\rm rd2}$ ] can be related to @xmath42 .
using the fact that the entropy fluctuation is a conserved quantity for superhorizon mode , the following relation holds : @xmath62_{\rm rd2 } \nonumber \\ & ~=~ & \left [ \delta_{\phi}^{(\delta\phi ) } - \frac{3}{4 } \delta_{\gamma_\chi}^{(\delta\phi ) } \right]_{\rm \phi d , rd1}.\end{aligned}\ ] ] with this relation , in particular , we obtain @xmath63 density fluctuations of various components are also parameterized by @xmath42 .
detailed properties of the density fluctuations , however , depend on how the various components in the universe are produced .
if a component @xmath30 is generated from the decay product of @xmath1 , then there is no entropy between the photon ( i.e. , @xmath44 ) and @xmath30 . on the contrary , if some other scalar field @xmath64 generates @xmath30 , the entropy between the photon and @xmath30 is the same as @xmath42 .
thus , if all the components in the universe are generated from @xmath1 , the density fluctuations become purely adiabatic and @xmath65_{\rm rd2 } & ~=~ & \frac{4}{3}\left [ \delta_b^{(\delta\phi ) } \right]_{\rm rd2 } \nonumber \\ & ~=~ & \frac{4}{3}\left [ \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } \nonumber \\ & ~=~ & -2 \psi_{\rm rd2}^{(\delta\phi)},\end{aligned}\ ] ] where the subscripts @xmath66 , @xmath67 , and @xmath68 are for the photon , baryon , and cdm , respectively . in this case , the isocurvature perturbation in the @xmath1 field is converted to the purely adiabatic density perturbation after the decay of @xmath1 . on the contrary , if the baryon asymmetry is generated by the scalar field @xmath64 , the entropy between the radiation and the baryon becomes @xmath69 and hence @xcite @xmath70_{\rm rd2 } = \frac{4}{3}\left [ \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } = -2 \psi_{\rm rd2}^{(\delta\phi ) } , \\ & & \left [ \delta_b^{(\delta\phi ) } \right]_{\rm rd2 } = \frac{3}{4 } \left [ \delta_\gamma^{(\delta\phi ) } \right]_{\rm rd2 } + \frac{9}{2 } \psi_{\rm rd2}^{(\delta\phi ) } , \label{sb}\end{aligned}\ ] ] and in the case where @xmath64 is responsible for the cdm while the baryon number is somehow generated from the decay product of @xmath1 , @xmath70_{\rm rd2 } = \frac{4}{3}\left [ \delta_b^{(\delta\phi ) } \right]_{\rm rd2 } = -2 \psi_{\rm rd2}^{(\delta\phi ) } , \\ & & \left [ \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } = \frac{3}{4 } \left [ \delta_\gamma^{(\delta\phi ) } \right]_{\rm rd2 } + \frac{9}{2 } \psi_{\rm rd2}^{(\delta\phi)}. \label{sc}\end{aligned}\ ] ] in addition , if the baryon and the cdm are both generated from sources other than @xmath1 , we obtain @xmath71_{\rm rd2 } & = -2 \psi_{\rm rd2}^{(\delta\phi ) } , \\ & \left [ \delta_b^{(\delta\phi ) } \right]_{\rm rd2 } & = \left [ \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } \nonumber \\ & & = \frac{3}{4 } \left [ \delta_\gamma^{(\delta\phi ) } \right]_{\rm rd2 } + \frac{9}{2 } \psi_{\rm rd2}^{(\delta\phi)}. \label{sbc}\end{aligned}\ ] ] it is important to notice that , for the cases given in eqs .
( [ sb ] ) @xmath43 ( [ sbc ] ) , the isocurvature perturbation is correlated with the adiabatic perturbation .
so far , we have seen that the primordial fluctuation of the curvaton field may generate the metric and density fluctuations . before closing this section
, we should compare the size of the curvaton contribution with the inflaton contribution .
even with the curvaton , there is also a contribution from the inflaton fluctuation since i assume inflation as a solution to the horizon and flatness problems .
if we consider the situation where there is no entropy fluctuation , inflaton contribution to the metric perturbation ( in the rd2 epoch ) is given by @xcite @xmath72_{k = ah } , \label{psi(inf)}\end{aligned}\ ] ] with @xmath73 , while the curvaton contribution is @xmath74_{k = ah}. \label{psi(curv)}\end{aligned}\ ] ] ( here and hereafter , the superscript `` ( inf ) '' is for quantities generated from the inflaton fluctuation . ) as one can see , the curvaton contribution to @xmath24 is inversely proportional to @xmath5 and hence , if @xmath5 is small enough , curvaton contribution dominates over the inflaton contribution .
thus , in the curvaton scenario , there are two conflicting requirements on the initial amplitude of the curvaton field .
one is the upper bound on @xmath5 ; upper bound on @xmath5 for @xmath75 depends on the model of the inflation .
for the case of the chaotic inflation , for example , the curvaton contribution becomes larger than the inflaton contribution if @xmath76 . the other is the lower bound on the @xmath5 , since the @xmath1d epoch can not be realized if @xmath5 is too small .
lower bound depends on the reheating temperatures @xmath77 and @xmath78 given in eqs .
( [ t_r1 ] ) and ( [ t_r2 ] ) , respectively . in fig .
[ fig : phimin ] , i plot the lower bound on @xmath5 to realize the @xmath1d epoch .
now , i discuss the cmb angular power spectrum in the curvaton scenario .
one of the most important consequences of the curvaton scenario is that , if all the components in the universe are generated only from the decay products of @xmath1 , no entropy fluctuation is generated and the primordial density fluctuations ( after the rd2 epoch ) becomes purely adiabatic .
let us first consider such a case .
one important check point is that the scale dependence of the primordial density fluctuations . in the case where the cosmic density fluctuations are generated from the primordial fluctuation of the inflaton
, scale - dependence originates from the change of the slope of the inflaton potential as well as the expansion rate during the inflation . as can be seen from eq .
( [ psi(curv ) ] ) , on the contrary , scale - dependence of the curvaton contribution is only from the change of the expansion rate ( as far as @xmath79 ) .
thus , even though both the curvaton and inflation contributions are from primordial fluctuations of some scalar fields , their scale - dependences are different . defining the spectral index @xmath80 as @xmath81 ^ 2 \simeq n^{({\rm inf } , \delta\phi)}_{\rm s}-1,\end{aligned}\ ] ] the spectral indices for the inflaton and curvaton contributions are calculated as @xmath82 where @xmath83 and @xmath84 are slow - roll parameters which are given by @xmath85 due to the change of the scale - dependence
, we can see that the observational constraints on inflation models are relaxed in the curvaton scenario .
for example , for the case of the chaotic inflation with the inflaton potential @xmath86 , spectral indices for the inflaton and curvaton contributions are estimated as @xmath87 and @xmath88 for @xmath89 , @xmath90 and @xmath91 for @xmath92 , and @xmath93 and @xmath94 for @xmath95 .
thus , using currently available constraint on the spectral index @xmath96 @xcite from the wilkinson microwave anisotropy probe ( wmap ) , simple chaotic inflation model with @xmath97 is excluded by the observations while , with the curvaton mechanism , @xmath95 becomes consistent with the observational constraints .
for other class of inflation models , change of the constraint may be more drastic .
next , let us discuss the effects of the possible entropy fluctuation in the curvaton scenario .
as mentioned in the previous section , in the curvaton scenario , correlated mixture of the adiabatic and isocurvature fluctuations may be generated . in order to discuss effects of the entropy fluctuation to the cmb angular power spectrum , it is convenient to parameterize the density fluctuation of the non - relativistic matter as @xmath98_{\rm rd2 } & ~\equiv~ & \left [ \frac{\omega_b}{\omega_m } \delta_b^{(\delta\phi ) } + \frac{\omega_c}{\omega_m } \delta_c^{(\delta\phi ) } \right]_{\rm rd2 } \nonumber \\ & ~=~ & \frac{3}{4 } \left [ \delta_\gamma^{(\delta\phi ) } \right]_{\rm rd2 } + \kappa_m \psi_{\rm rd2}^{(\delta\phi)},\end{aligned}\ ] ] where @xmath99 , @xmath100 and @xmath101 are the ( present ) density parameters of the baryon , the cdm , and the non - relativistic component ( and hence @xmath102 ) .
although the density fluctuations for the baryon and the cdm are independent , shape of the cmb angular power spectrum is determined once @xmath103 is fixed .
for the purely adiabatic case , @xmath104 . with the relations ( [ sb ] ) , ( [ sc ] ) , and ( [ sbc ] ) , @xmath103 becomes @xmath105 , @xmath106 , and @xmath107 , respectively .
= 0.5 in fig .
[ fig : clkdep ] , i show how the cmb angular power spectrum depends on the @xmath103 parameter .
as one can see , @xmath108 strongly depends on @xmath103 and , if @xmath109 , deviation of the cmb angular power spectrum from the adiabatic result ( i.e. , @xmath108 with @xmath104 ) becomes sizable .
importantly , the wmap results strongly suggest that the primordial density fluctuations be ( almost ) purely adiabatic ; with the wmap data , @xmath110 is obtained @xcite .
thus , the wmap results impose stringent constraint on the curvaton scenario .
a possible small contamination of the entropy fluctuation in some case will be discussed in the next section .
now , i would like to discuss the case where the curvaton field is also responsible for the scenario of baryogenesis . among various scenarios , there are some cases where the baryon asymmetry of the universe is generated from scalar - field condensations . in those scenarios ,
scalar fields often dominate the universe at some epoch and hence , if they have primordial fluctuations , they may play the role of the curvaton .
probably , two of the most famous examples of the possible scalar fields responsible for the baryon asymmetry are right - handed sneutrino @xcite and the affleck - dine field @xcite.in some case , inflaton can be responsible for the scenario of the baryogenesis .
for example , right - handed neutrinos can be produced by the decay of the inflaton , which may become the origin of the baryon asymmetry of the universe @xcite . here
, i consider the case where the right - handed sneutrino , which becomes the origin of the baryon asymmetry of the universe @xcite , plays the role of the curvaton @xcite .
first , i briefly summarize the model and the thermal history . here , the relevant part of the superpotential is given by @xmath111 where @xmath112 is the yukawa matrix for the neutrino while @xmath113 is the majorana mass matrix for the right - handed ( s)neutrinos . here ,
@xmath114 and @xmath115 are generation indices of the right - handed neutrino @xmath116 while @xmath117 is that of the left - handed lepton doublet @xmath118 . in addition
, @xmath119 is the up - type higgs field .
i work in the basis where the matrix @xmath113 is diagonalized . for simplicity ,
let us consider the case where the lightest right - handed sneutrino @xmath120 has non - vanishing initial amplitude .
( hereafter , mass of @xmath121 is denoted as @xmath122 . ) with a non - vanishing primordial amplitude , @xmath121 starts to oscillate when @xmath123 and decays when @xmath124 , where @xmath125 is the decay rate of @xmath121 .
since the majorana mass term breaks the lepton - number symmetry , lepton - number asymmetry may be generated at the time of the sneutrino decay if non - vanishing cp violation exists .
such a lepton - number asymmetry becomes the source of the baryon - number asymmetry of the universe with the sphaleron process .
the mechanism of generating the baryon - number asymmetry is basically the supersymmetric version of the fukugita - yanagida mechanism @xcite .
expression for the resultant amount of the baryon asymmetry is , however , rather complicated since , in this case , the primordial abundance of the right - handed sneutrino has non - thermally determined .
assuming @xmath126 , we obtain @xcite @xmath127 where @xmath128 is the temperature at the epoch of the decay of @xmath121 , @xmath129 the mass of the heaviest ( left - handed ) neutrino mass .
( so , if @xmath121 decays after dominating the universe , @xmath128 becomes the reheat temperature due to the decay of @xmath121 . )
in addition , in the basis where the majorana mass matrix for the right - handed neutrinos @xmath130 is real and diagonalized , the effective cp violating phase is given by @xmath131_{11 } } { [ \hat{h } \hat{h}^\dagger]_{11 } } .\end{aligned}\ ] ] notice that , with maximum cp violation , @xmath132 .
in addition , @xmath133 is the energy fraction of the radiation generated from the decay product of @xmath121 . with the relation @xmath126
, @xmath121 decays after the inflaton decay . in this case , it is convenient to define the following quantity : @xmath134 if @xmath135 , @xmath136 is realized when @xmath137 .
thus , when @xmath138 , @xmath121 decays in the @xmath6-dominated universe and hence @xmath139 on the contrary , if @xmath140 , the right - handed sneutrino decays after it dominates the universe and we obtain @xmath141 thus , as is easily seen , energy fraction of @xmath142 becomes close to 1 when @xmath143 while , in the opposite limit , most of the radiation are generated from the decay product of the inflaton .
now , i consider the cosmological density fluctuations in this scenario
. in particular , i concentrate on the case where the condensation of the right - handed sneutrino plays the role of the curvaton .
( thus , in this case , the epoch after the decay of @xmath121 is identified as the rd2 epoch . )
the scenario is basically the same as the curvaton scenario which i have explained before . in the previous discussion ,
i have discussed the cases where all the components in the universe are generated from the decay product of the curvaton , which corresponds to the case of @xmath144 .
as is explained above , however , ( small ) contamination of the radiation from the decay product of the inflaton may always exist .
if the baryon asymmetry is generated from thermally produced particles after the rd2 epoch is realized , then entropy fluctuations vanish and the primordial density fluctuations become purely adiabatic . in the scenario we consider , however , the situation is quite different since the baryon asymmetry is generated from the decay product of @xmath121 . in particular ,
if the energy fraction of the radiation from the decay product of the inflaton becomes sizable , non - vanishing baryonic entropy fluctuation may be generated .
evolution of the density fluctuations in this case is also understood by solving the einstein and boltzmann equations for the fluctuations given in the previous section .
we can calculate the entropy in the radiation - dominated universe after the decay of both @xmath0 and @xmath121 . in the radiation - dominated universe @xcite
@xmath145 then , with the relation @xmath146 with @xmath147 , we obtain @xmath148 using the fact that the entropy between any component produced from @xmath121 and that from the inflaton field is conserved , we can relate @xmath149 with @xmath150 ; with the relation @xmath151 , we obtain @xmath152 thus , if @xmath121 decays much after it dominates the universe , @xmath144 and hence the metric perturbation becomes comparable to the primordial entropy perturbation . on the other hand , if @xmath153 , the metric perturbation becomes negligibly small .
since the baryon asymmetry is generated from @xmath121 , the density fluctuation in the baryonic component is given by @xmath154 thus , the entropy between the baryon and the radiation is given by @xmath155 and defining @xmath156 we obtain @xmath157 if the right - handed sneutrino once dominates the universe , @xmath158 and hence the perturbation becomes adiabatic .
( thus , the situation is like the case discussed in the previous section . )
it is interesting if a sizable amount of the energy density of the radiation is from the inflaton . in this case , @xmath159 becomes smaller than @xmath160 and correlated mixture of the adiabatic and isocurvature fluctuations is generated .
as mentioned in the previous section , wmap data suggests that the baryonic and cdm entropy fluctuations should be very small ; @xmath161 @xcite . if @xmath162 , however , @xmath163 can be smaller than the present upper bound but the deviation of the cmb angular power spectrum from the adiabatic result may still be within the reach of the future precise observations .
thus , it is desirable to find the signal from the ( correlated ) entropy fluctuation in the future precise observations of the cmb anisotropy for the test of the curvaton scenario .
finally , let me comment on the case where the curvaton also generates the baryon asymmetry of the universe via affleck - dine mechanism .
if the primordial fluctuation of the affleck - dine field becomes the dominant source of the cosmological density fluctuations , large baryonic entropy fluctuation is generated even if the universe is once completely dominated by the affleck - dine field .
this is due to the fact that the affleck - dine field is a complex scalar field and also that the resultant baryon asymmetry of the universe is sensitive to the initial value of the affleck - dine field .
if the affleck - dine field acquires primordial fluctuation , its initial phase as well as the initial amplitude is expected to have fluctuations .
here , i have discussed the basic features of the curvaton scenario and its implications to particle cosmology . in particular , i have discussed that the cosmic density fluctuations can dominantly originate from the primordial fluctuation of the `` curvaton '' field , which is a late - decaying scalar condensation other than the inflaton . in the curvaton scenario ,
scale - dependence of the cosmic density fluctuation becomes different from the case of the simple inflation , and in many cases , the spectral index @xmath164 becomes close to @xmath160 compared to the simple inflation case .
thus , given the fact that the cmb anisotropy observed by the wmap suggests ( almost ) scale - invariant primordial density fluctuations , observational constraints on the inflaton potential can be relaxed by adopting the curvaton scenario .
one of the important feature of the curvaton scenario is the possible contamination of the ( correlated ) entropy fluctuation in the non - relativistic matter .
the cmb angular power spectrum is sensitive to the entropy fluctuation .
thus , if such correlated entropy fluctuation is sizable , it affects the shape of the cmb angular power spectrum and hence it will be interesting and important to try to find its consequence in particular in the cmb anisotropy at future precision observations of the universe .
the author is grateful to the organizers of the fujihara seminar `` seesaw 1979 2004 : neutrino mass and seesaw mechanism '' for their hospitality during the stay .
the work of t.m . is supported by the grant - in - aid for scientific research from the ministry of education , science , sports , and culture of japan , no .
15540247 .
s. kasuya , t. moroi and f. takahashi , arxiv : hep - ph/0312094 . k. enqvist and m. s. sloth , nucl .
b * 626 * ( 2002 ) 395 .
d. h. lyth and d. wands , phys .
b * 524 * ( 2002 ) 5 .
t. moroi and t. takahashi , phys .
b * 522 * ( 2001 ) 215 [ erratum - ibid .
b * 539 * ( 2002 ) 303 ] .
n. bartolo and a. r. liddle , phys .
d * 65 * ( 2002 ) 121301 ; t. moroi and t. takahashi , phys .
d * 66 * ( 2002 ) 063501 ; d. h. lyth , c. ungarelli and d. wands , phys .
d * 67 * ( 2003 ) 023503 ; m. s. sloth , nucl .
b * 656 * ( 2003 ) 239 ; r. hofmann , arxiv : hep - ph/0208267 ; k. dimopoulos and d. h. lyth , arxiv : hep - ph/0209180 ; t. moroi and h. murayama , phys .
b * 553 * ( 2003 ) 126 ; k. enqvist , s. kasuya and a. mazumdar , phys .
* 90 * ( 2003 ) 091302 ; f. di marco , f. finelli and r. brandenberger , phys .
d * 67 * ( 2003 ) 063512 ; k. a. malik , d. wands and c. ungarelli , phys .
d * 67 * ( 2003 ) 063516 ; m. postma , phys .
d * 67 * ( 2003 ) 063518 ; c. gordon and a. lewis , arxiv : astro - ph/0212248 ; k. dimopoulos , phys .
d * 68 * ( 2003 ) 123506 ; a. r. liddle and l. a. urena - lopez , phys .
d * 68 * ( 2003 ) 043517 ; j. mcdonald , phys .
d * 68 * ( 2003 ) 043505 ; k. dimopoulos , g. lazarides , d. lyth and r. ruiz de austri , jhep * 0305 * ( 2003 ) 057 ; k. enqvist , a. jokinen , s. kasuya and a. mazumdar , phys .
d * 68 * ( 2003 ) 103507 ; k. dimopoulos , d. h. lyth , a. notari and a. riotto , jhep * 0307 * ( 2003 ) 053 m. endo , m. kawasaki and t. moroi , phys .
b * 569 * ( 2003 ) 73 ; m. postma and a. mazumdar , arxiv : hep - ph/0304246 ; m. postma , arxiv : astro - ph/0305101 ; s. kasuya , m. kawasaki and f. takahashi , phys .
b * 578 * ( 2004 ) 259 ; d. h. lyth and d. wands , phys .
d * 68 * ( 2003 ) 103515 ; d. h. lyth and d. wands , phys . rev .
d * 68 * ( 2003 ) 103516 ; k. dimopoulos , g. lazarides , d. lyth and r. r. de austri , phys .
d * 68 * ( 2003 ) 123515 ; k. hamaguchi , m. kawasaki , t. moroi and f. takahashi , arxiv : hep - ph/0308174 ; d. h. lyth , phys .
b * 579 * ( 2004 ) 239 ; n. bartolo , s. matarrese and a. riotto , phys .
d * 69 * ( 2004 ) 043503 ; j.
mcdonald , jcap * 0312 * ( 2003 ) 005 ; j. mcdonald , arxiv : hep - ph/0310126 ; c. gordon and k. a. malik , arxiv : astro - ph/0311102 ; m. bastero - gil , v. di clemente and s. f. king , arxiv : hep - ph/0311237 ; s. gupta , k. a. malik and d. wands , arxiv : astro - ph/0311562 ; r. allahverdi and m. drees , arxiv : hep - ph/0401054 ; l. pilo , a. riotto and a. zaffaroni , arxiv : astro - ph/0401302 ; e. j. chun , k. dimopoulos and d. lyth , arxiv : hep - ph/0402059 ; m. postma , arxiv : astro - ph/0403213 ; d. langlois and f. vernizzi , arxiv : astro - ph/0403258 ; k. enqvist , arxiv : hep - ph/0403273 ; j. mcdonald , arxiv : hep - ph/0404154 ; m. ikegami and t. moroi , arxiv : hep - ph/0404253 .
g. veneziano , phys . lett .
b * 265 * ( 1991 ) 287 ; m. gasperini and g. veneziano , astropart .
* 1 * ( 1993 ) 317 ; m. gasperini and g. veneziano , phys .
d * 50 * ( 1994 ) 2519 ; m. gasperini and g. veneziano , phys . rep .
* 373 * ( 2003 ) 1 .
j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , phys .
d * 64 * ( 2001 ) 123522 ; j. khoury , b. a. ovrut , n. seiberg , p. j. steinhardt and n. turok , phys .
d * 65 * ( 2002 ) 086007 ; j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , phys .
d * 66 * ( 2002 ) 046005 .
w. t. hu , ph .
d thesis ( arxiv : astro - ph/9508126 ) .
s. mollerach , phys .
d * 42 * ( 1990 ) 313 .
t. moroi and t. takahashi , in @xcite .
j. m. bardeen , p. j. steinhardt and m. s. turner , phys .
d * 28 * ( 1983 ) 679 .
d. n. spergel _ et al .
_ , astrophys .
j. suppl .
* 148 * ( 2003 ) 175 .
k. hamaguchi , m. kawasaki , t. moroi and f. takahashi , in ref .
@xcite .
h. murayama , h. suzuki , t. yanagida and j. yokoyama , phys .
* 70 * ( 1993 ) 1912 .
k. hamaguchi , h. murayama and t. yanagida , phys .
d * 65 * ( 2002 ) 043512 .
i. affleck and m. dine , nucl .
b * 249 * ( 1985 ) 361 .
t. asaka , k. hamaguchi , m. kawasaki and t. yanagida , phys .
d * 61 * ( 2000 ) 083512 .
t. moroi and h. murayama , in @xcite . | i describe basic features of the curvaton scenario where the primordial fluctuation of a late - decaying scalar field , called `` curvaton , '' becomes the dominant source of the cosmic density fluctuations .
i also discuss its implications to sneutrino cosmology . |
we use a simple model of a cubic sqd with @xmath23 ( we will use @xmath11 nm for estimates ) , covered with an oxide layer with thickness @xmath24 nm , and contacted with current terminals from below and from above .
the @xmath25-axis is oriented from top to bottom along the current flow , as shown in fig . 1 .
a tunneling current is injected into the dot via the oxide barrier from the top ( source at @xmath26 ) and leaves the dot at the bottom ( drain at @xmath27 ) . due to the presence of the oxide barriers
this current is non - ballistic and non - thermal .
we assume that the high potential barriers associated with the oxide layers are not much affected by the voltage and the tunneling charges .
we concentrate on what happens between these effective source and drain ( fig .
2 ) . in the case
when the dot can be regarded as an insulating system it is reasonable to assume that the applied voltage equally drops over the potential barriers and the dots . for simplicity
we neglect the difference of the dielectric constants of the oxide barriers and the dot . in this approximation
we can introduce an effective voltage @xmath28 describing the part of the total transport voltage @xmath29 applied between effective source and drain which drops across the dot itself . in this rude approximation
we neglect the effect of spatial quantization upon values on the ionization energy , the conductivity gap and material parameters of silicon .
at @xmath30 the fermi level inside the dot is situated in the middle of the gap , i.e. @xmath31 bellow the conduction band edge ( @xmath32 ev at 300 k ) .
as @xmath33 grows , the bottom of the ( still empty ) conduction band bends down accordingly .
when the conduction band in the dot close to the drain aligns with the fermi level of the emitter we expect a drastic increase in the tunneling current . this threshold @xmath34 voltage ( fig .
2 ) for @xmath33 is given by @xmath35 , regardless of the number @xmath36 of dopants in the dot ( as long as the dot is not yet metallic , of course ) .
in the following we therefore limit our studies to voltages @xmath37 in this voltage range we have a @xmath1-thick barrier ( formed by the dot ) with always finite hight between effective source and drain . the intrinsic concentration of electrons and holes at 300 k is @xmath38 @xmath6 .
even at this high temperature the probability to have at least one intrinsic electron in a dot with size @xmath11 nm is only @xmath39 .
so we would expect virtually no current in this mode .
this is confirmed by direct electrical tests @xcite of sqd with the required properties .
let us now consider one single single donor in the dot located at @xmath25 with ionization energy @xmath40 ev ( for p as a donor ) .
the evident channel for current flow is single - electron tunneling from the source to the empty impurity , and then from populated impurity to the drain .
this channel opens as soon as @xmath33 reaches a threshold @xmath41 leading to a step - like increase in the total conductance of the dot .
if the impurity is located near the drain , i.e. @xmath42 ( as donor 1 in fig .
2 ) , then @xmath41 is given by @xmath43 in contrast , for an impurity located at distances @xmath44 from the drain ( i.e. further away than the threshold case of donor 2 in fig .
2 ) , no additional current channel via a single impurity can be opened at low enough voltages defined in ( [ threshold ] ) where virtually no background current is present . in the present case
this value @xmath45 nm , which returns us to the above criterion : only impurities located in the immediate vicinity ( defined within the accuracy @xmath9 ) of the drain contribute to the single - impurity channel .
the probability to populate an impurity from the source , and then to depopulate it towards the drain is directly related to the overlap of the atom - like impurity wave - functions with the corresponding contacts leading to a conductance @xmath46 of this current channel @xmath47 this shows that in first approximation the conductance of this channel does not depend on @xmath25 . as shown above , a single - impurity channel already only selects impurities located within a very narrow range of @xmath25 close to the drain , eq .
( [ g1 ] ) gives an additional argument for the independence of this channel conductance @xmath46 on the actual location of the impurity inside this thin layer near the drain .
the above consideration shows , that due to the bend of the bottom of conduction band following the transport voltage , there is no chance to notice current flowing through a sequential chain of impurities ( such as donors 1 and 3 in fig .
2 ) , connecting source and drain .
the contribution of such a chain will be totally masked by the current flowing directly via the conduction band .
the only way for multiple impurities to manifest themselves in quantized conductance is to form multiple _ parallel _ singe - impurity channels situated close enough to the drain as considered above .
therefore , if @xmath48 impurities fall into the thin layer near the drain to approximately the same @xmath25 coordinate as that of donor 1 in fig .
2 ( within the bohr radius ) , we will see a switching - on of an @xmath17-fold channel with conductance @xmath49 at the same threshold voltage @xmath50 v as for a single - donor channels .
all the above considerations are only valid as long as the dot itself can be regarded as an insulating system .
as the number of donors in a sqd grows , the dot becomes a metallic particle , and the conduction band edge in the dot aligns with the fermi level of the electrodes . in a very simple estimate
we define this transition to a metal when the total volume of @xmath17 donors with an individual volume of @xmath51 exceeds the volume of the dot .
this is an exaggerated version of the mott criterion ( [ mott ] ) which holds not only in bulk , but in a small structure , too . for the analyzed example from above this
gives @xmath52 as a limiting value .
the practically interesting set @xmath53 for both @xmath36 and @xmath17 considered above is still far bellow this limit .
quite a number of other mechanisms of electron transport might take place in this system , the main one being resonant tunneling .
surprisingly , even taking into account such other mechanisms @xcite does not change much the main idea of the present paper .
useful discussions with m. kupriyanov and s. oda are gratefully acknowledged .
this work was supported in part by the russian foundation for basic research and by the russian program for future nanonelectronic devices .
a. dutta , m. kimura , y. honda , m. otobe , a. itoh , and s. oda , _ jpn .
_ , * 35 * , p. 4038
+ s. vyshenski , a. ohata , y. kinugasa , y. hayafune , h. hara , s .-
lee , k. nishiguchi , a. dutta , s. oda .
active probing of single - electron effects in a silicon quantum dot .
_ 4-th int . symp . on quantum effect electronics _ ,
p. 45 - 50 , tokyo ( 1997 ) . | considering a double - barrier structure formed by a silicon quantum dot covered by natural oxide , we derive simple conditions for the conductance of the dot to become a step - like function of the number of doping atoms inside the dot , with negligible dependence on the actual position of the dopants .
the found conditions are feasible in experimentally available structures .
* quantum mott transition in a silicon quantum dot * s. v. vyshenski , u. zeitler , and r. j. haug + institute of nuclear physics , moscow state university , moscow 119899 , russia + institut fr festkrperphysik , universitt hannover , appelstr .
2 , d-30167 hannover , germany february 21 , 1998 the fabrication of @xmath0 nanostructures became possible through very recently developed new technologies @xcite .
one unique preparation technology for individual silicon quantum dots ( sqd ) has been reported in @xcite .
they are spherical @xmath0 particles with diameters @xmath1 in the range 512 nm covered by a 12 nm - thick natural @xmath2 film .
metallic current terminals made from degenerately doped @xmath0 are defined lithographically to touch each individual dot from above and from below . to ensure metallic electrodes
the donor concentration @xmath3 should be @xmath4 , where @xmath5 @xmath6 .
the critical concentration @xmath7 is defined by the _
mott criterion _
@xcite , introducing the transition to a metallic type of conductivity in a semiconductor at : @xmath8 where @xmath9 nm is the bohr radius of an electron bound to a donor inside the @xmath0 crystal , in the case of phosphorus - donors @xmath10 nm @xcite . as for the doping of the dot ,
the situation concerning a mott transition in that small dots is much less trivial than the one described by eq .
( [ mott ] ) .
let us consider dots with diameters @xmath11 nm formed from n - doped @xmath0 with @xmath12 as an illustrative example .
then each dot contains in average one donor .
note that we will consider degenerately @xmath13-doped electrodes with @xmath14 which ensures metallic conduction up to the borders of the dot .
real fabrication technology @xcite provides a wafer with hundreds of sqds on it with current leads towards each individual sqd .
dots in average have the same value of mean dopant concentration @xmath3 , which is determined by the parent material of bulk silicon the dots are formed from .
however , on the level of each individual sqd we will always have exactly _ integer _ number of doping atoms .
if , as in the example above , the average number of dopants @xmath15 the actual number of donors in the dot can have values @xmath16 , with values larger than these very unlikely .
our objective is to illustrate , that sqds from the same wafer fall into several distinct sets of approximately the same conductance .
the typical value of conductance for each set is nearly completely determined by the number @xmath17 of donors present in a certain part of a sqd so that @xmath17 labels each set of sqds . summarizing the above , we need for a quantization of the conduction through a dot with @xmath17 donors the following conditions : * size @xmath1 of the dot comparable with bohr radius : @xmath18 . *
average doping @xmath3 of the dot @xmath19 , leading to a mean number of dopants @xmath20 , so that @xmath21 are the most probable configurations of an individual sqd . *
doping of the electrodes @xmath22 , so that current leads are perfectly metallic . * dot covered by an oxide layer thick enough to suppress ballistic transport through the dot .
in fact all these condition can be simultaneously satisfied for sqd fabricated with the method mentioned above @xcite . |
together with the active developments of the heavy - ion collision ( hic ) experiments , one can now explore the hot and dense qcd matter within experimental facilities .
theoretically , the hot qcd matter can be studied appropriately from the first principle via the lattice qcd ( lqcd ) simulations , unlike the dense - matter lqcd , which suffers from the notorious sign problem . beside the lqcd methods , various effective approaches
have been also frequently applied to investigate the hot and/or dense qcd matter . in our previous works using the liquid instanton model at finite @xmath1 ( lim-@xmath1 ) ,
we have studied various nonperturbative quantities at finite @xmath1 and zero quark chemical potential @xmath11 , i.e. electric conductivity , shear viscosity , quark condensate , magnetic susceptibility , and so on @xcite . by comparing our numerical results with other theoretical results , including the lqcd data
, we verified that lim-@xmath1 provides relatively good agreement with them .
hence , in the present work , we want to provide theoretical results for the dimension - five quark - gluon mixed condensate ( mc ) at finite @xmath1 using lim-@xmath1 .
mc is defined by the vacuum expectation value ( vev ) for the quark - gluon dimension - five operator in minkowski space as follows : @xmath12 where @xmath13 and @xmath14 stands for the quark and gluon fields .
the antisymmetric tensor is given as @xmath15 .
as shown in eq .
( [ eq : qdef ] ) , mc is usually factorized into a mass constant @xmath16 and the quark condensate @xmath17 ( qc ) for a practical usage in the operator - product expansion ( ope ) of the qcd sum rule ( qcdsr ) calculations . in other words , @xmath6 in the right - hand - side of eq .
( [ eq : qdef ] ) plays an important input value in applying the qcdsr technique . for vacuum @xmath18
, many theoretical works have been done so far for mc using lqcd @xcite , holographic qcd ( hqcd ) @xcite , global color - symmetry model @xcite , dyson - schwinger method @xcite , lim @xcite , hybrid - current method @xcite , truncated quark - quark interaction model @xcite , and so on .
note that , however , there have been only a few theoretical studies for mc at finite @xmath1 so far : lqcd @xcite and effective approach , such as the global color - symmetry model @xcite .
moreover , the lqcd work was done via the kogut - susskind ( ks ) fermion at the quenched level , giving @xmath19 mev , which is much higher than those from the recent full lqcd simulations @xcite .
hence , considering the recent progress in the realistic lqcd simulations at finite @xmath1 , it is worth to providing sophisticated theoretical estimations for mc , as done in the previous works , and the numerical results will be a useful guide for future lqcd simulations . here
, we want to make a brief description for lim-@xmath1 as a theoretical framework : lim-@xmath1 is basically based on lim , which manifests the nonlocal quark - quark interactions via the fermionic zero mode . to extend lim to a finite-@xmath1 system , the model parameters of lim , i.e.
the average ( anti)instanton size @xmath20 and inter-(anti)instanton distance @xmath21 , are modified as functions of @xmath1 using the trivial - holonomy caloron solution @xcite .
note that these modifications result in the partial chiral restoration of the parameters .
constructing an effective thermodynamic potential within the same model , we compute the constituent - quark mass at zero virtuality as a function of @xmath1 , as an chiral order parameter , by solving the saddle - point equation with respect to the variation parameter @xcite . as a result ,
one observe the second - order and crossover chiral restoration patterns for the zero and finite current - quark masses ( @xmath2 ) , satisfying the universal restoration patterns .
we note that the chiral transition @xmath1 is determined as @xmath22 mev for @xmath23 mev . along with these @xmath1-modified parameters and constituent - quark mass ,
we employ the fermionic matsubara formula for the relevant matrix elements .
we also perform the two - loop renormalization - group ( rg ) evolution for the both condensates , mc and qc , for @xmath3 gev , to compare the numerical results with the lqcd data . note
that @xmath10 gev is a typical scale of lim , i.e. @xmath24 . from the numerical results , it turns out that mc is insensitive to the evolution , whereas qc gets larger by the evolution .
numerically , we obtain @xmath4 gev at @xmath5 within the present theoretical framework .
moreover , mc as well as qc play the role of the order parameters of the chiral phase transition . the ratio of the two condensates @xmath6 turns out to be almost flat below the chiral transition @xmath1 ( @xmath7 ) , and increases stiffly beyond it . from the numerical results , we parametrize the ratio approximately with a simple linear functions at @xmath10 gev : @xmath25 we organize the present work as follows : in section ii
, we briefly introduce the liquid instanton model ( lim ) and how to compute the mixed condensate in terms of the field theoretical manner . in section iii
, the temperature modifications of the relevant model parameters are performed using the trivial caloron solution .
we also show the correct universal chiral restoration patterns , computed within the present model .
the numerical results for mc , qc , and their ratio @xmath6 as functions of temperature are presented with relevant discussions in section iv .
final section is devoted to summary , conclusion , and future perspectives .
in eq . ( [ eq : y ] ) . ( c ) the quark - gluon mixed condensate in eq .
( [ eq : mc0 ] ) , constructed by attaching the external quark lines in the effective quark - gluon vertex ( b).,width=529 ] in this section , we elucidate the theoretical framework and how to compute the relevant condensates .
all the calculations will be performed in the leading large-@xmath26 limit .
we make use of an effective action , derived from the liquid - instanton model ( lim ) in euclidean space for the su(2 ) light - flavor sector , and it reads @xmath27,\ ] ] where the current - quark mass matrix for su(@xmath28 ) is designated as @xmath29 . by assuming the isospin symmetry , we set @xmath30 mev throughout the present work , when the current - quark mass is finite
. @xmath31 denotes the effective quark mass at zero virtuality , i.e. constituent - quark mass . in the euclidean momentum space , the quark momentum distribution @xmath32 is given by @xmath33 , \,\,\,\,\tau=\frac{|k|\bar{\rho}}{2}.\ ] ] here , @xmath34 and @xmath35 indicate the modified bessel functions , whereas @xmath36 stands for the average ( anti)instanton size , which corresponds to the inverse of the model scale , @xmath24 .
first , we want to calculate @xmath17 ( qc ) by functional differentiating @xmath37 with respect to @xmath2 , resulting in @xmath38,\ ] ] where we have used a simplified notation @xmath39 , and @xmath26 indicates the number of color : @xmath40 .
note that @xmath36 and @xmath21 at @xmath5 were estimated by the variational method @xmath41 fm @xcite , phenomenological way @xmath42 fm @xcite , and the lqcd simulation @xmath43 fm @xcite . among them , we employ the phenomenological values , considering about @xmath44 uncertainties within those estimations .
the value for @xmath31 for vacuum can be fixed by the self - consistent equation : @xmath45 where @xmath46 indicates the instanton number density or packing fraction of the instanton ensemble . by solving eq .
( [ eq : sce ] ) with the phenomenological instanton parameters given above , one acquires @xmath47 mev .
now , we are in a position to discuss the quark - gluon operator in terms of the quark fields . the local operator inside mc in the left - hand - side of eq .
( [ eq : qdef ] ) corresponds to the quark - gluon interaction of yukawa type , as shown in ( a ) of figure [ fig1 ] .
however , in this model calculation , the gluon field strength ( @xmath48 ) can be expressed in terms of the nontrivial quark - instanton interaction @xcite .
first , the one flavor quark and one instanton interaction can be written as a function of @xmath49 and color orientation matrix @xmath50 .
@xmath51\left[iq^{\dagger}(k)_{\alpha i}\frac{1\pm\gamma_5}{2}q(p)^{\beta j}\right].\ ] ] here , @xmath52 and @xmath53 for @xmath54denote the color and spinor indices , respectively . in deriving eq .
( [ eq : y ] ) , we assumed the @xmath55-function type instanton distribution @xmath56 as usual . a schematic diagram for eq .
( [ eq : y ] ) is given in ( b ) of figure [ fig1 ] .
then , we write the field strength tensor @xmath57 in terms of instanton configuration : @xmath58f^b_{\pm\mu\nu}(x'-x).\ ] ] @xmath59 stand for the field strength consisted of a certain instanton configuration . using eqs .
( [ eq : y ] ) and ( [ eq : fa ] ) , we can define the field strength tensor in momentum space in terms of the quark - instanton interactions : @xmath60 following the course of ref .
@xcite , finally , we obtain the mixed - condensate as follows : @xmath61[p^2+\bar{m}^2_p]},\ ] ] where @xmath62 , which relates to the fourier transform of the instanton field strength , and @xmath63 are defined by the followings : @xmath64 , \,\,\,\,t=|k - p|\bar{\rho } \cr n_{k , p}&=&\frac{1}{4}\mathrm{tr}_\gamma \left[\sigma_{\mu\nu}(\rlap{/}{k}+i\bar{m}_k)\gamma_{\mu\nu}(\rlap{/}{p}+i\bar{m}_p ) \right],\,\,\,\ , \gamma_{\mu\nu}=\sigma_{\alpha\nu}\frac{q_\rho q_\mu}{q^2 } + \sigma_{\mu\alpha}\frac{q_\rho q_\nu}{q^2}-\frac{1}{2}\sigma_{\mu\nu},\,\,\,\ , q = k - p.\end{aligned}\ ] ] performing the trace of the lorentz index and simplifying the expression , we arrive at @xmath65 } { [ k^2+\bar{m}^2_k][p^2+\bar{m}^2_p]},\ ] ] which is schematically represented by ( c ) in figure [ fig1 ] .
mc in eq .
( [ eq : mcb ] ) can be written as a function of @xmath1 , employing the fermionic matsubara formula .
note that one of the technical difficulties here is the double summation over the matsubara frequencies @xmath66 and @xmath67 , corresponding to the integral over @xmath68 and @xmath69 .
however , this difficulty can be easily removed by assuming that @xmath70 in the numerator in eq .
( [ eq : mcb ] ) , whereas we rewrite the denominator by @xmath71 .
according to this simplification , one can separate the integrand in eq . ( [ eq : mcb ] ) into independent summations over @xmath72 and @xmath73 , resulting in @xmath74 } { [ \omega^2_n+e^2_{\bm{k}}][\omega^2_{n'}+e^2_{\bm{p } } ] } \cr & = & n_c\int\frac{d^3\bm{k}}{(2\pi)^3}\frac{d^3\bm{p}}{(2\pi)^3 } \frac{\sqrt{m_{\bm{k}}m_{\bm{p}}}g_{\bm{k},\bm{p}}}{e_{\bm{k}}e_{\bm{p}}}\left[(\bm{k}\cdot \bm{p})-\frac{4(\bm{k}\cdot \bm{q})(\bm{p}\cdot \bm{q})}{\bm{q}^2 } \right ] \mathrm{tanh}\left[\frac{e_{\bm{k}}}{2 t } \right ] \mathrm{tanh}\left[\frac{e_{\bm{p}}}{2 t } \right],\end{aligned}\ ] ] where we have used @xmath75 for brevity .
note that @xmath76 and @xmath77 in the above equation represent those in eqs .
( [ eq : fk ] ) and ( [ eq : gn ] ) by replacing @xmath78 as @xmath79 ^ 3,\,\,\,\ , g_{k , p}(\bar{\rho})\to g_{\bm{k},\bm{p}}(\bar{\rho})\ ] ] to reproduce the correct low - energy constants @xcite .
note that @xmath76 in eq .
( [ eq : massss ] ) show stronger decrease with respect to @xmath80 , in comparison to those used in the previous works @xcite .
the reason to use this stronger one is to reproduce the chiral transition @xmath1 , @xmath81 mev for @xmath82 , which was estimated by the recent lqcd simulations , using the @xmath83 flavor qcd thermodynamics with improved staggered fermions @xcite .
detailed discussions on @xmath7 will be given in section iv .
now , we want to address how to determine the @xmath1-dependent effective quark mass @xmath31 in eq .
( [ eq : massss ] ) . in refs .
@xcite , we derived it by using the caloron distribution with the trivial holonomy , i.e. harrington - shepard caloron @xcite .
firstly , we want to explain briefly how to modify @xmath36 and @xmath21 as functions of @xmath1 , using the caloron solution .
details can be found in ref .
an instanton distribution function for arbitrary @xmath26 and @xmath84 can be written with a gaussian suppression factor as a function of @xmath1 and an arbitrary instanton size @xmath85 for pure - glue qcd @xcite : @xmath86.\ ] ] we note that the cp - invariant vacuum was taken into account in eq .
( [ eq : ind ] ) , and we assumed the same analytical form of the distribution function for the ( anti)instanton .
note that the instanton number density ( packing fraction ) @xmath87 and @xmath36 have been taken into account as functions of @xmath1 implicitly . for simplicity ,
we take the numbers of the anti - instanton and instanton are the same , i.e. @xmath88 .
we also assigned the constant factor in the right - hand - side of the above equation as @xmath89 for simplicity .
the abbreviated notations read : @xmath90,\,\,\,\ , \bar{\beta}=-b\ln[\lambda_\mathrm{rs}\langle r\rangle],\,\,\ , c_{n_c}=\frac{4.60\,e^{-1.68\alpha_{\mathrm{rs } } nc}}{\pi^2(n_c-2)!(n_c-1 ) ! } , \cr a_{n_c}&=&\frac{1}{3}\left[\frac{11}{6}n_c-1\right]\pi^2,\,\,\,\ , \gamma=\frac{27}{4}\left[\frac{n_c}{n^2_c-1}\right]\pi^2,\,\,\,\ , b=\frac{11n_c-2n_f}{3}.\end{aligned}\ ] ] note that we defined the one - loop inverse charges @xmath91 and @xmath92 at certain phenomenological cutoffs @xmath93 and @xmath94 .
@xmath95 denotes a scale , depending on a renormalization scheme , whereas @xmath96 for the three - dimensional volume . using the instanton distribution function in eq .
( [ eq : ind ] ) , we can compute the average value of the instanton size @xmath97 straightforwardly as follows @xcite : @xmath98^{\frac{1}{2 } } -a_{n_c}t^2}{2\bar{\beta}\gamma n},\ ] ] where @xmath99 .
it can be easily shown that eq .
( [ eq : rho ] ) satisfies the following asymptotic behaviors @xcite : @xmath100 here , the second relation of eq .
( [ eq : asym ] ) indicates a correct scale - temperature behavior at high @xmath1 , i.e. , @xmath101 . substituting eq .
( [ eq : rho ] ) into eq .
( [ eq : ind ] ) , the caloron distribution function can be evaluated further : @xmath102,\,\,\,\ , \mathcal{f}(t)=\frac{1}{2}a_{n_c}t^2+\left[\frac{1}{4}a^2_{n_c}t^4 + \nu\bar{\beta}\gamma n \right]^{\frac{1}{2}}.\ ] ] the instanton packing fraction @xmath103 can be computed self - consistently , using the following equation : @xmath104^\frac{1}{\nu},\ ] ] where we replaced @xmath105 , and @xmath106 stands for the @xmath107-function with an argument @xmath108 .
note that @xmath89 and @xmath92 can be determined easily using eqs .
( [ eq : rho ] ) and ( [ eq : novv ] ) , incorporating the vacuum values for @xmath109 and @xmath110 : @xmath111 and @xmath112 .
finally , in order for estimating the @xmath1-dependence of @xmath31 , it is necessary to consider the normalized distribution function , defined as follows , @xmath113}{\gamma(\nu)}.\ ] ] here , the subscript @xmath114 denotes the normalized distribution . for brevity , we want to employ the large-@xmath26 limit to simplify the expression for @xmath115 . in this limit , as understood from eq .
( [ eq : nid ] ) , @xmath115 can be approximated as a @xmath55-function : @xmath116.\ ] ] the numerical result for @xmath117 is given in the left panel of figure [ fig23 ] .
the curve for @xmath117 shows that the average ( anti)instanton size smoothly decreases with respect to temperature .
this behavior indicates that the instanton ensemble gets diluted and the nonperturbative effects via the quark - instanton interactions are diminished . at @xmath118 mev , which is close to the chiral phase transition temperature , the instanton size decreases by about @xmath119 in comparison to its value at @xmath1=0 .
considering that the instanton size corresponds to the scale parameter of the model , i.e. uv cutoff mass , @xmath120 , the temperature - dependent cutoff mass is a clearly distinctive feature in comparison to other low - energy effective models , such as the njl model .
in addition , we also show the temperature dependence of the average ( anti)instanton number density or ( anti)instanton packing fraction , @xmath46 , in the left panel of figure [ fig23 ] .
again , the instanton number density decreases as temperature increases : the instanton ensemble becomes diluted with respect to @xmath1 .
as in ref .
@xcite , the lim-@xmath1 thermodynamic potential per volume in the leading @xmath121 contributions at zero quark chemical potential can be written as follows : @xmath122 + 2\sigma^2 - 2n_cn_f\int^\infty_0\frac{d^3\bm{k}}{(2\pi)^3 } \left[e_{\bm{k}}+2t\ln\left[1+e^{-\frac{e_{\bm{k}}}{t } } \right ] \right],\end{aligned}\ ] ] where @xmath123 and @xmath124 represent a lagrange multiplier to exponentiate the effective quark - instanton action and an arbitrary mass parameter to make the argument for the logarithm dimensionless .
@xmath125 stands for the isosinglet scalar meson field corresponding to the effective quark mass . in the leading large-@xmath26 contributions
, we have the relation @xmath126 @xcite .
then , the saddle - point equation can be derived from eq .
( [ eq : tp ] ) by differentiating @xmath127 by the lagrange multiplier @xmath123 : @xmath128=0.\ ] ] here , @xmath129 and @xmath130 implicitly .
note that one can write the instanton number density in terms of the effective quark mass @xmath31 and @xmath36 @xcite : @xmath131 the value of the real - positive parameter @xmath132 is determined to reproduce @xmath133 mev for @xmath23 at @xmath5 , resulting in @xmath134 , respectively , by solving the saddle - point equation in eq .
( [ eq : limgap ] ) . by solving eq .
( [ eq : limgap ] ) with respect to @xmath31 numerically , the numerical results for @xmath31 as a function of @xmath1 are given in the right panel of figure [ fig23 ] for the zero and finite current quark mass : @xmath135 ( solid ) and @xmath136 mev ( dotted ) .
these results show proper universal patterns for the chiral phase transition , i.e. the second - order chiral phase transition for the massless quark and the crossover for the finite mass . from those numerical results ,
the phase transition temperatures are given as @xmath22 mev for @xmath23 mev .
@xmath7 is marked by the vertical line in the right panel of figure [ fig23 ] .
note that @xmath7 for the second - order and crossover chiral phase transition are almost consistent to each other .
the obtained @xmath7 is compatible with the lqcd estimation as mentioned above @xcite . [ cols="^,^ " , ] in eq .
( [ eq : qdef ] ) as functions of @xmath1 for different @xmath2 and @xmath137 values given in ( @xmath138 ) : ( 0 mev , 0.6 gev ) ( solid ) , ( 0 mev , 2.0 gev ) ( dot ) , ( 5 mev , 0.6 gev ) ( dash ) , and ( 5 mev , 2.0 gev ) ( long - dash ) .
the vertical line indicates the chiral transition temperature @xmath22 mev.,width=321 ]
we have investigated the quark - gluon mixed condensate ( mc ) at finite temperature ( @xmath1 ) , employing the @xmath1-modified liquid - instanton model ( lim-@xmath1 ) . in the present theoretical framework , one can write the quark - gluon yukawa interaction in terms of the effective nonlocal four - quark vertex . using it , we calculated mc at a low renormalization @xmath10 gev , then performed the rg evolution up to @xmath139 gev .
since we were interested in the @xmath1 dependence of mc , we obtained the @xmath1-dependent instanton parameters , @xmath117 and @xmath140 from the trivial - holonomy caloron solution .
we also made use of the effective thermodynamic potential and fermionic matsubara formula to determine @xmath141 and to compute the relevant condensates .
numerical results for mc , quark condensate ( qc ) , and their ratio @xmath142 were presented with discussions .
important observations in the present work are listed below : * the @xmath1-modified instanton parameters show partial partial chiral restoration behaviors as @xmath1 increases : the average instanton size @xmath143 and instanton number density @xmath144 slowly decrease with respect to @xmath1 , indicating the reduction of the nonperturbative effect , which is given by the nontrivial interaction of the quarks and ( anti)instantons via the quark zero mode .
note that this @xmath1 dependence of the relevant model parameters is a peculiar feature of the present model .
* by solving the saddle - point equation for the effective thermodynamic potential , we obtain the chiral order parameter @xmath31 , i.e. constituent - quark mass , and the chiral phase transition @xmath1 is given by @xmath22 mev for @xmath23 .
the chiral phase transitions are second - order and crossover for @xmath135 and @xmath145 mev , respectively , satisfying the universal chiral restoration patterns . *
numerical results for qc are well compatible with the phenomenological values and the lqcd simulation data below @xmath7 .
however , we observe sizable deviation , in comparison to the lqcd data for @xmath82 , beyond @xmath7 .
as @xmath1 grows beyond @xmath7 , qc keeps decreasing but finite up to @xmath146 mev . by running up the scale @xmath3 gev
, we had a multiplicable factor @xmath147 from the two - loop rg evolution . * those for mc exhibit similar curve shapes to that of qc as functions of @xmath1 .
it turned out that the mc curves increase slightly up to @xmath148 mev , then start to decrease .
this tendency is found to be similar to that from the global color - symmetry model . as @xmath136 mev ,
mc curves are almost saturated beyond a certain value in the vicinity of @xmath149 .
we had a rg - evolution factor @xmath150 for @xmath3 gev . * the ratios @xmath6
are given numerically by @xmath151 for @xmath23 mev for vacuum .
these values are well compatible with those from other effective models and qcdsr estimation .
it turned out that @xmath6 is almost flat with the slope @xmath152 for @xmath153 as a function of @xmath154 , being consistent with the ( quenched ) lqcd simulation . above @xmath7 , the curve of @xmath6 increased rapidly , depending on the different @xmath1 dependence between qc and mc .
we suggest an approximated linear parametrization for @xmath6 as : @xmath155 the present results for mc as a function of @xmath1 will be a useful guide for future lqcd simulations and various model calculations .
note that the effects of the external strong electromagnetic field , which is expected to be created from the peripheral ( non - central ) hic , has been widely discussed @xcite .
in addition , the distribution of the gluons in the quark - gluon plasma ( qgp ) is an important input for the hydrodynamic studies for hic @xcite .
it is worth mentioning that , using the effective quark - gluon vertex as in the present work , one can explore the effects of the external em field , which interacts with the gluon fields in terms of the quark ones .
related works are under progress and appear elsewhere .
s.i.n . is grateful to g. endrdi ( regensburg ) for fruitful comments .
the numerical calculations were partially performed via the computing server abacus2 at kias .
99 s. i. nam and c. -w .
kao , arxiv:1304.0287 [ hep - ph ] .
s. i. nam , arxiv:1304.1265 [ hep - ph ] .
t. doi , n. ishii , m. oka and h. suganuma , nucl .
a * 721 * , 934 ( 2003 ) .
h. -ch .
kim and y. kim , jhep * 0810 * , 011 ( 2008 ) .
h. -s .
zong , f. wang , j. -l .
ping , x. -f .
lu and e. -g .
zhao , commun .
* 38 * , 205 ( 2002 ) . c. -f .
lu and x. -f .
lu , commun .
phys . * 53 * , 893 ( 2010 ) .
m. v. polyakov and c. weiss , phys .
b * 387 * , 841 ( 1996 ) s. i. nam and h. -ch .
kim , phys .
b * 647 * , 145 ( 2007 ) .
k. aladashvili and m. margvelashvili , phys .
b * 372 * , 299 ( 1996 ) .
t. meissner , phys .
b * 405 * , 8 ( 1997 ) .
t. doi , n. ishii , m. oka and h. suganuma , hep - lat/0409077 .
z. zhang and w. -q .
zhao , phys .
b * 610 * , 235 ( 2005 ) .
b. j. harrington and h. k. shepard , nucl .
b * 124 * , 409 ( 1977 ) .
d. diakonov and a. d. mirlin , phys .
b * 203 * , 299 ( 1988 ) .
s. i. nam , j. phys .
g * 37 * , 075002 ( 2010 ) .
d. diakonov , prog .
phys . * 51 * , 173 ( 2003 ) .
e. v. shuryak , nucl .
b * 203 * , 93 ( 1982 ) .
m. c. chu , j. m. grandy , s. huang and j. w. negele , phys .
d * 49 * , 6039 ( 1994 ) .
s. i. nam , in preparation .
a. bazavov _ et al .
_ , phys .
d * 85 * , 054503 ( 2012 ) .
a. bazavov [ hotqcd collaboration ] , pos lattice * 2011 * , 182 ( 2011 ) .
t. schafer and e. v. shuryak , rev .
* 70 * , 323 ( 1998 ) .
t. muta , _ foundations of quantum chromodynamics _ , ( world - scientific , 1998 ) .
g. s. bali _ et al .
_ , phys . rev .
d * 86 * , 094512 ( 2012 ) .
g. s. bali , f. bruckmann , g. endrodi , z. fodor , s. d. katz and a. schafer , phys .
d * 86 * , 071502 ( 2012 ) s. i. nam , h. -ch .
kim , a. hosaka and m. m. musakhanov , phys .
d * 74 * , 014019 ( 2006 ) .
b. i. abelev _ et al . _ [ star collaboration ] , phys .
c * 79 * , 034909 ( 2009 ) .
k. tuchin , arxiv:1301.0099 [ hep - ph ] .
k. fukushima , d. e. kharzeev and h. j. warringa , phys .
d * 78 * , 074033 ( 2008 ) .
s. ozonder , phys .
d * 87 * , 045013 ( 2013 ) | we investigate the quark - gluon mixed condensate @xmath0 for the su(2 ) light - flavor sector at finite temperature ( @xmath1 ) .
relevant model parameters , such as the average ( anti)instanton size , inter-(anti)instanton distance , and constituent - quark mass at zero virtuality , are modified as functions of @xmath1 , employing the trivial - holonomy caloron solution . by doing that
, we observe correct chiral restoration patterns depending on the current - quark mass @xmath2 .
we also perform the two - loop renormalization - group ( rg ) evolution for the both condensates by increasing the renormalization scale @xmath3 gev .
it turns out that the mixed condensate is insensitive to the rg evolution , whereas the quark condensate become larger considerably by the evolution .
numerically , we obtain @xmath4 gev at @xmath5 within the present theoretical framework , and the mixed condensate plays the role of the chiral order parameter for finite @xmath1 .
the ratio of the two condensates @xmath6 is almost flat below the chiral transition @xmath1 ( @xmath7 ) , and increases rapidly beyond it . from a simple linear parametrization
, we obtain @xmath8 for @xmath9 at @xmath10 gev .
the present results are compared with other theoretical ones including the lattice qcd simulations , and show qualitatively good agreement with them . |
inflation @xcite ( for review , @xcite ) is an vacuum - dominated epoch in the early universe when the scale factor grew exponentially .
this scenario is used to set the initial condition for the hot big bang model and to provide the primordial density perturbation as the seed of structure formation . in the framework of slow - roll inflation ,
the slow - roll parameters are defined by @xmath8 @xmath9 where @xmath10 is the reduced planck mass .
the spectral index can be expressed in terms of the slow - roll parameters as @xmath11 the latest wmap 5-year result prefers the spectral index around @xmath12 @xcite .
the spectrum is given by @xmath13 with the slow - roll approximation the value of the inflaton field @xmath14 , in order to achieve @xmath15 e - folds inflation , is @xmath16 from observation @xcite @xmath17 at @xmath18 ( we call this cmb ( cosmic microwave background ) normalization ) .
the scalar potential for the inflaton field @xmath14 of supernatural inflation @xcite ( during inflation ) is @xmath19 where @xmath20 is the susy breaking scale with @xmath21 gev and the second term is the soft mass term with @xmath22 tev in order to address the hierarchy problem of the standard model ( sm ) .
supernatural inflation is very attractive but predicts the spectral index @xmath23 . in @xcite ,
a model is proposed with the potential includes the a - term and non - renormalizable terms @xmath24 where @xmath25 for the minimal supersymmetric standard model ( mssm ) @xcite . when @xmath26 is the gravity mediated susy breaking scale it is shown that the spectral index can naturally be @xmath12 , because the potential is naturally of a hilltop form ( so called hilltop inflation @xcite ) .
the soft mass @xmath27 is related to @xmath26 via @xmath28 where @xcite @xmath29 as we can see from eq .
( [ eq3 ] ) , for gravity - mediated susy breaking , @xmath30 gev . for gauge
mediated susy breaking , @xmath26 can be as low as @xmath31 tev . from eq .
( [ eq1 ] ) , it is clear that the original model of supernatural inflation can not work for lower @xmath26 .
however , for the model eq .
( [ eq2 ] ) , it certainly can .
the reason is even if we have @xmath32 the potential still can accommodate for a successful inflation model if inflation occurs near a saddle point where @xmath33 , and it is called mssm inflation or a - term inflation in the literature @xcite . in this paper , we will explore the parameter space for a successful inflation model from eq .
( [ eq2 ] ) near a saddle point with a wide range of @xmath0 from @xmath34 tev to @xmath35 gev .
the paper is organized as follows .
we present the calculation and result in section [ 2 ] .
our conclusions are summarized in section [ 3 ] .
as in the case of mssm inflation , the potential ( eq .
( [ eq2 ] ) ) has a saddle point where @xmath33 if @xmath27 and @xmath36 are related via @xmath37 the saddle point is at @xmath38 where @xmath39 the potential at the saddle point is @xmath40 near the saddle point , the potential can be described as @xmath41 where @xmath42 from eq .
( [ efolds ] ) , the number of e - folds is @xmath43 the slow roll parameters @xmath44 and @xmath45 are @xmath46 @xmath47 inflation ends when @xmath48 .
the spectral index is ( for @xmath49 ) @xmath50 which is the same as a - term inflation .
the spectrum is give by @xmath51 after impose cmb normalization ( @xmath52 ) for @xmath53 tev , we obtain the relation between @xmath0 and @xmath54 .
we show the result on fig .
( [ result ] ) . and @xmath54 .
here we assume @xmath55 tev.,scaledwidth=45.0% ] as we can see from fig .
( [ result ] ) , for @xmath56 , the predictions is just the same as mssm ( or a - term ) inflation .
actually the lower bound can go down to zero , and recovers mssm ( or a - term ) inflation . for @xmath57 ,
the existence of @xmath0 allows also @xmath3 and @xmath4 to fit cmb normalization with @xmath58 . for @xmath3 , from eq .
( [ eq4 ] ) , we have @xmath59 , corresponds to @xmath60 .
similarly , @xmath4 corresponds to @xmath61 .
in this paper , we generalize the model of @xcite by considering smaller @xmath0 .
our model is similar to hybrid inflation because the additional @xmath0 which eventually should gracefully exist by a waterfall field .
however , it is different from the usual hybrid inflation because inflation occurs near a saddle point .
the result is for @xmath62 and @xmath1 tev , the @xmath6 flat direction can works with @xmath63 .
the @xmath3 and @xmath4 flat directions can also play the role of an inflaton field with appropriate @xmath0 .
this work was supported in part by the nsc under grant no .
nsc 96 - 2628-m-007 - 002-my3 , by the ncts , and by the boost program of nthu .
this work is also supported in parts by the wcu program through the kosef funded by the mest ( r31 - 2008 - 000 - 10057 - 0 ) .
we are grateful to david lyth and kazunori kohri for discussions .
a. h. guth , phys .
d * 23 * , 347 ( 1981 ) .
d. h. lyth and a. riotto , phys .
rept . * 314 * , 1 ( 1999 ) [ arxiv : hep - ph/9807278 ] .
d. h. lyth , lect .
notes phys .
* 738 * , 81 ( 2008 ) [ arxiv : hep - th/0702128 ] .
a. linde , lect .
notes phys .
* 738 * , 1 ( 2008 ) [ arxiv:0705.0164 [ hep - th ] ] .
l. randall , m. soljacic and a. h. guth , arxiv : hep - ph/9601296 . c. m. lin and k. cheung , arxiv:0901.3280 [ hep - ph ] .
t. gherghetta , c. f. kolda and s. p. martin , nucl .
b * 468 * , 37 ( 1996 ) [ arxiv : hep - ph/9510370 ] .
l. boubekeur and d. h. lyth , jcap * 0507 * , 010 ( 2005 ) [ arxiv : hep - ph/0502047 ] .
k. kohri , c. m. lin and d. h. lyth , jcap * 0712 * , 004 ( 2007 ) [ arxiv:0707.3826 [ hep - ph ] ] .
k. dimopoulos and d. h. lyth , phys .
d * 69 * , 123509 ( 2004 ) [ arxiv : hep - ph/0209180 ] .
r. allahverdi , k. enqvist , j. garcia - bellido and a. mazumdar , phys .
lett . * 97 * , 191304 ( 2006 ) [ arxiv : hep - ph/0605035 ] .
d. h. lyth , jcap * 0704 * , 006 ( 2007 ) [ arxiv : hep - ph/0605283 ] . j. c. bueno sanchez , k. dimopoulos and d. h. lyth , jcap * 0701 * , 015 ( 2007 ) [ arxiv : hep - ph/0608299 ] .
r. allahverdi , mod .
a * 23 * , 2799 ( 2008 ) [ arxiv:0812.3628 [ hep - ph ] ] . | following @xcite , we explore the parameter space of the case when the supersymmetry ( susy ) breaking scale is lower , for example , in gauge mediated susy breaking model . during inflation
, the form of the potential is @xmath0 plus mssm ( or a - term ) inflation .
we show that the model works for a wide range of the potential @xmath0 with the soft susy breaking mass @xmath1 tev .
the implication to mssm ( or a - term ) inflation is that the flat directions which is lifted by the non - renormalizable terms described by the superpotential @xmath2 with @xmath3 and @xmath4 are also suitable to be an inflaton field for @xmath5 provided there is an additional false vacuum term @xmath0 with appropriate magnitude .
the flat directions corresponds to @xmath6 also works for @xmath7 . |
the rutherford centennial is celebrating rutherford s discovery of the atomic nucleus in 1911 by scattering alpha - particles on a gold foil . since then scattering experiments have come a long way , improving the resolution from about @xmath2 times the proton radius ( @xmath3 fm ) in 1911 to about @xmath4 as achieved by the hera experiments @xcite at the so far only @xmath5 collider . here , i want to describe the current achievements in the extraction of the parton density functions ( pdfs ) of the proton and of the strong qcd coupling from inclusive and jet measurements in dis at hera .
inclusive neutral current ( nc ) and charged current ( cc ) measurements , in leading order ( lo ) represented by the quark - parton model , are sensitive to the valence and sea quarks of the proton . only at next - to - leading order ( nlo ) and beyond
do they become sensitive to the gluon density ( scaling violations ) and the strong coupling , which are however strongly correlated .
inclusive jets in dis , when measured for example in the breit frame , are sensitive to the gluon density and the strong coupling already in lo , via the boson - gluon fusion process .
this process dominates for the negative photon momentum transfer squared , or photon virtuality , @xmath6 , less than about @xmath7 gev@xmath8 . above that value
the qcd - compton process becomes dominant , providing sensitivity to the strong coupling without being correlated to the gluon .
therefore , one can expect that the addition of jet measurements to the hitherto mainly used inclusive measurements alone will allow a reduction of the correlation between the gluon density and the strong coupling , thereby making it possible to extract simultaneously both pdfs and the strong coupling from dis data .
the idea behind the herapdfs is to use only hera nc and cc data in the fits .
combined inclusive cross sections from the h1 and zeus experiments are used , which have increased precision due to statistics but also due to intercalibration effects made possible by the two complementary detectors .
thus , precise combined data sets with total uncertainties between @xmath9 ( for nc ) over most of the phase space are used .
the systematic correlated and uncorrelated errors are well controlled , such that the standard @xmath10 uncertainty criterion can be used .
since hera data only include @xmath11 data , there are no uncertainties due to for example deuterium or heavy target corrections .
two parametrizations have been studied : a @xmath12 parameter fit which was used for the published herapdf1.0 @xcite and for the preliminary herapdf1.5 @xcite , and a @xmath13 parameter fit , for the preliminary herapdf1.5f and herapdf1.6 @xcite .
the latter parametrization provides more flexibility , in particular for the gluon density at low @xmath14 , which is preferred because of the additional jet data . at the @xmath15
starting scale the pdfs are parametrized as follows : @xmath16 in the @xmath13 parameter fit @xmath17 , @xmath18 and @xmath19 are constrained by the momentum and quark sum rules .
furthermore it is assumed that @xmath20 , @xmath21 and @xmath22 , where @xmath23 is the strange sea fraction at @xmath15 .
these choices define the central fit , as adding more parameters does not significantly reduce the @xmath24 .
the parametrization uncertainty is estimated by adding one additional parameter at a time , changing the parametrizations and varying the starting scale @xmath15 around the default value of @xmath25 gev@xmath8 .
a model uncertainty is determined by varying the heavy quark masses , @xmath23 and @xmath26 around their default values .
@xmath26 specifies the minimal @xmath6 for a data point to be included in the fit .
the fit of the pdfs is performed using the nlo dglap equations , employing the qcdnum @xcite evolution code .
charm and beauty quarks are treated as massive in the thorne - roberts variable - flavor - number scheme @xcite .
the jet production cross sections are calculated with the fastnlo code @xcite , which efficiently convolutes the matrix elements , calculated using the nlojet++ program @xcite , with the fitted pdfs . here
is not the place to discuss this , but it should be mentioned that herapdf1.0 and herapdf1.5 fits have been also performed using evolution equations in next - to - next - to - leading order ( nnlo ) . for the reader who wants to know more details and updates on the herapdfs
, they can be found here @xcite .
the data used in the extraction of the pdfs and the strong coupling @xmath0 are the combined nc and cc @xmath27 and @xmath28 inclusive cross sections from the hera - i @xcite and hera - ii running periods @xcite .
in addition , the following jet measurements from the h1 and zeus collaborations were used : normalized inclusive jet cross sections for @xmath29 gev@xmath8 @xcite , corresponding to an integrated luminosity of @xmath30 pb@xmath31 , and inclusive jet cross sections for @xmath32 gev@xmath8 @xcite , corresponding to an integrated luminosity of @xmath33 pb@xmath31 , and inclusive jet cross sections from two running periods in hera - i for @xmath34 gev@xmath8 @xcite , corresponding to an integrated luminosity of @xmath35 pb@xmath31 , and for @xmath36 gev@xmath8 @xcite , corresponding to an integrated luminosity of @xmath37 pb@xmath31 .
the pdfs as a function of @xmath14 at @xmath38 gev@xmath8 , when going from herapdf1.5 ( 10 parameters ) to herapdf1.5f ( 14 parameters ) with fixed @xmath39 , show an overall similar total uncertainty .
the latter yields a softer sea density at high @xmath14 and a suppressed d - valence density at low @xmath14 . keeping @xmath0 fixed and including the jet measurements , which leads to the herapdf1.6 fit , one observes little difference in the pdfs and their uncertainties , only the high @xmath14 gluon uncertainty is somewhat reduced . the respective figures can be found in @xcite .
more interesting results are obtained when the fits for herapdf1.5f ( no jets ) and herapdf1.6 ( with jets ) are performed with @xmath0 treated as a free fit parameter .
the corresponding pdfs and their uncertainty bands are shown in fig.[fig1 ] . for @xmath38 gev@xmath8 with @xmath0 as an additional free parameter in the fit , and ( c ) @xmath40 for both fits .
, title="fig : " ] ( -126.0,121.0)(a ) for @xmath38 gev@xmath8 with @xmath0 as an additional free parameter in the fit , and ( c ) @xmath40 for both fits .
, title="fig : " ] ( -126.0,121.0)(b ) ( -126.0,121.0)(c ) adding the jet data leads to a dramatic decrease of the low @xmath14 gluon uncertainty , not only of the experimental uncertainty , but also of the model and parametrization ones . in fig.[fig1]c
the @xmath40 distributions are shown as a function of @xmath0 .
the herapdf1.5f fit shows a shallow minimum , while the herapdf1.6 fit provides a strong constraint on the value of @xmath0 .
the addition of the jet data successfully reduces the correlation of the strong coupling and the gluon density , for reasons discussed in the introduction .
the increase in the uncertainty of the gluon density due to @xmath0 being a free parameter is not large when jet data are included .
the fit yields the following value for the strong coupling : 68 mm * figure 2 . * @xmath0 values from herapdf1.6 , from jet production in dis , by the h1 and zeus collaborations , and the world average .
[ fig3 ] @xmath41 the hadronization uncertainty is due to the corresponding corrections of the nlo jet cross sections .
the uncertainty due to the scale is estimated by a variation of the renormalization and factorization scale by the conventional factor of @xmath42 and @xmath43 when calculating the jet cross sections .
for the inclusive cross sections this uncertainty was assumed to be negligible . in fig.2
the results on @xmath0 are displayed from herapdf1.6 and as obtained from recent extractions by h1 and zeus using only jet production data in dis .
the experimental and theoretical uncertainties are indicated separately .
for herapdf1.6 the uncertainty due to the pdfs is part of the experimental uncertainty , while for the jets only results it is part of the theory uncertainty .
the values obtained using different methods and experiments are experimentally very precise .
they are consistent with each other and with the world average .
a nlo qcd fit with simultaneous determinations of the proton pdfs and @xmath0 was performed using hera dis data only .
combined h1 and zeus inclusive nc and cc cross sections together with inclusive jet cross sections from the h1 and zeus collaborations have been used . including jet data in the fit and treating @xmath0 as a free parameter , dramatically reduces the correlations between the gluon density and the strong coupling compared to the fit without jets .
the precision of the gluon density is improved , and an accurate and unbiased determination of @xmath0 is achieved , yielding a value consistent with the world average .
h1 and zeus collaboration , `` qcd analysis and determination of @xmath0 using the combined h1 and zeus inclusive cross sections together with the jet production cross section measured by the h1 and zeus experiments , '' _
h1prelim-11 - 034 _ , _ zeus - prel-11 - 001_. f. d. aaron _ et al .
_ [ h1 collaboration ] , _ eur .
j. _ c * 65 * , 363 ( 2010 ) .
[ arxiv:0904.3870 [ hep - ex ] ] .
f. d. aaron _ et al .
_ [ h1 collaboration ] , _ eur .
j. _ c * 67 * , 1 ( 2010 ) .
[ arxiv:0911.5678 [ hep - ex ] ] . | a simultaneous nlo qcd fit of the parton density functions ( pdfs ) of the proton and of the strong coupling @xmath0 is presented .
the analysis is based on combined h1 and zeus inclusive dis cross section measurements together with inclusive jet cross section measurements provided by both h1 and zeus collaborations .
the inclusion of the jet data significantly reduces the correlation between the gluon pdf and the strong coupling , improving the precision of the gluon pdf and providing an accurate determination of @xmath0 using dis data from hera only .
address= max - planck - institute fr physik , fhringer ring 6 , 80805 munich , germany@xmath1(mail:[email protected] ) |
many neural systems composed of a number of interacting neurons exhibit self - sustained oscillatory behavior leading to the formation of various space - time patterns .
such patterns are believed to play a key role in signal and information processing functions of the brain @xcite .
one of the fundamental problems is the understanding of possible dynamic mechanisms of such patterns to appear and to evolve in time and space @xcite .
there are two basic phenomena of emergence of oscillations ( regular or chaotic ) in neuron assemblies .
the first one is obvious and deals with the presence of local intra - cellular oscillations .
being coupled such units are capable of various oscillatory patterns , clustering and synchronization .
the second one , found in recent theoretical and experimental studies , concerns the possibility of oscillations in assemblies of non - oscillatory cells @xcite .
the oscillations may also appear in coupled non - identical cells for sufficiently strong coupling .
the assembly is characterized by an oscillatory `` average '' cell dynamics which makes the non - oscillatory cells oscillating .
another studies have reported that coupling even identical excitable cells can modify the dynamics to form oscillatory attractors co - existing with a stable fixed point @xcite .
such attractors are characterized by anti - phase spiking .
the effects of electrical coupling between neural cells also include the appearance of bursting in two coupled pacemaker cells , modification of burst period in coupled bursters , synchronization and chaos @xcite . a model approach , in order to display the dynamical origin of neural oscillations ,
is to use a simplified behavior - based description of the system . for
this purposes nonlinear maps could be helpful as they can provide an appropriate qualitative description of complex dynamic processes including chaotic behavior in lower dimensional systems @xcite .
in this paper we study the system of two coupled piece - wise linear one - dimensional maps .
the single map is derived from the dynamics of an isolated neural cell modelled by the fitzhugh - nagumo excitable system @xcite .
the fitzhugh - nagumo neuron model can be derived from hodgkin - huxley conductance - based model , for some parameter values , when we take into account the difference of kinetics between the _ potential dependent gating variables _ @xcite .
the first variable describes the evolution of neuron membrane potential , the second mimics the dynamics of outward ionic currents .
then , the model can describe the salient features of neuron dynamics including the action potential generation , excitability and excitation threshold.the map has one globally stable fixed point and a discontinuity corresponding to the excitation threshold of the cell .
we shall show how linear diffusive coupling between the two maps leads to the appearance of chaotic oscillations with anti - phase spiking .
a number of studies of coupled chaotic maps have shown that anti - phase chaotic oscillations may appear when the synchronization manifold looses transverse stability and the system evolves to the off - diagonal attractors @xcite .
the paper is organized as follows . in sect .
ii we show how the dynamics of excitable fitzhugh - nagumo model can be described by a piece - wise continuous point map and introduce two - dimensional map modeling a pair of coupled cells . in sect .
iii we analyze the dynamics of the map .
we numerically show the emergence of strange attractor in an invariant domain defined by invariant manifolds of saddle fixed point and saddle periodic orbit .
iv describes the statistical characteristics of the attractor set and the emergent chaotic oscillations with anti - phase spiking .
section v contains a brief discussion of the results .
to replicate the excitable dynamics of an isolated neural cell one can use the fitzhugh - nagumo - like model .
it can be taken in the following form @xmath0 the @xmath1-variable describes the evolution of the membrane potential of the neuron , @xmath2 describes the dynamics of the outward ionic currents ( the recovery variable ) @xcite.the function @xmath3 is taken piece - wise linear , @xmath4 , if @xmath5 , and @xmath6 , if @xmath7 .
the parameters @xmath8 and @xmath9 control the shape and the location of the @xmath2-nullcline , hence the dynamics of the recovery term .
the parameter @xmath10 defines the time scale of excitation pulse and the parameter @xmath11 is a constant stimulus .
the excitable behavior of eqs .
( [ fhn ] ) is illustrated in fig .
appropriate values of the parameters provide the existence of three fixed points @xmath12 , @xmath13 and @xmath14 .
the points @xmath15 and @xmath16 are stable and unstable foci , respectively , the point @xmath17 is a saddle with the incoming separatrix defining the excitation threshold .
then , if a perturbation of the rest state , @xmath15 , is large enough , i.e. lies beyond the separatrix , the system responds with an excitation pulse , otherwise it decays to the stable rest point @xmath15 ( fig .
1 ) . ) .
@xmath18 illustrates the transversal half - line . ] in order to describe the dynamics of the unit ( [ fhn ] ) by a point map , we introduce the transversal half - line @xmath19 ( fig .
it is found that the return map of the flow given by eqs .
( [ fhn ] ) at the section @xmath20 defines a map , @xmath21 , for all points excluding the one , @xmath22 , corresponding to the intersection of the incoming saddle separatrix with the half - line @xmath20 .
this point never returns to the cross - section .
consequently , we also must exclude all the pre - images of the point @xmath22 , @xmath23 , @xmath24 .
then , the poincar map @xmath25 can be written as @xmath26 @xmath27 with @xmath28 accounting for @xmath29-coordinates of the points at the poincare section .
the map is invertible and defined in the interval @xmath30 excluding the points @xmath31 .
the shape of the curve @xmath32 calculated numerically is shown in fig .
it is given by a piece - wise continuous curve with one stable fixed point .
then , the dynamics of the map is trivial .
all its trajectories represent sequences of points monotonically approaching to the fixed point .
the discontinuity point plays the role of the excitation threshold , i.e. the neuron exhibits spiking if the map evolves above this point .
note , that such map obtained from ( [ fhn ] ) gives an `` average '' description of the cell dynamics .
obtained from eqs .
( [ fhn ] ) at the section half - line @xmath20 .
parameter values : @xmath33 .
] let us approximate the piece - wise continuous curve @xmath32 by a piece - wise linear function and define the map for all points @xmath34 excluding the points @xmath31 .
then , for two cells we introduce difference coupling term ( electrical coupling ) and consider the following two - dimensional map @xmath35 : @xmath36 @xmath37 where the variables @xmath38 refer to the dynamics of the two cells , @xmath39 is the coupling coefficient , the function @xmath40 is taken in the form @xmath41 @xmath42 .
* a. general properties of map * since @xmath40 is given by the piece - wise linear function , then the differential @xmath43 of the map @xmath44 is a constant matrix , with eigenvalues @xmath45 .
hence , lyapunov exponents are defined by @xmath46 , for a arbitrary trajectory of the map , except zero lebesgue measure set of initial conditions .
let us consider the system ( [ 2d ] ) in the parameter region @xmath47 for convenience , we shall fixed @xmath48 .
this region @xmath49 is chosen in order to have lyapunov exponents @xmath50 , hence any trajectory , @xmath51 , where @xmath52 is a set of discontinuous points of the map @xmath35 , is heavily dependent on initial conditions , and to have the modulus of jacobian of the map @xmath35 less than one , in @xmath49 .
all these features allow us to speak of chaotic behavior .
we shall study the trajectories of the map @xmath35 missing the discontinuity set @xmath53 , where @xmath54 for convenience , let us change variables , according to the eigenvectors of @xmath43 @xmath55 then , the discontinuity lines become @xmath56 the lines @xmath57 and @xmath58 divide the phase plane @xmath59 of the map @xmath35 into four regions @xmath60 in each region @xmath61 the map @xmath35 is continuous and has the form @xmath62 with @xmath63 @xmath64 the map @xmath35 has one hyperbolic fixed point , @xmath65 . in region @xmath66
its invariant curves coincide with the coordinate axes . analyzing map @xmath35 in the regions @xmath67 , @xmath68 we find that it has a hyperbolic periodic orbit , @xmath69 , of period two with coordinates @xmath70 if the parameters satisfy the inequality @xmath71}.\ ] ] stable and unstable invariant manifolds of the orbit @xmath69 are @xmath72 @xmath73 it follows from ( [ 9 ] ) that @xmath74 then periodic orbit @xmath69 appears from infinity in the phase plane @xmath59 for the parameters belonging to the curve @xmath75 in the parameter plane @xmath76 .
note , that on this curve one of the multiplier hits the bifurcation value , @xmath77 .
then fixed point @xmath78 changes its stability becoming of saddle type in @xmath49 . *
b. invariant region and chaotic attractor of the map @xmath35 * let us consider the location of invariant curves of the points @xmath78 and periodic orbit @xmath69 in the phase plane with respect to the lines @xmath79 .
the unstable invariant curve ( the separatix ) @xmath80 intersects the line @xmath58 at point @xmath81 with coordinates @xmath82 and invariant curve @xmath83 at the point @xmath84 with coordinates @xmath85 figure 3 illustrates the location of the invariant curves of fixed point @xmath78 and orbit @xmath69 in the phase plane @xmath59 .
let us introduce region @xmath86 in the phase plane defined by @xmath87 let us find the conditions for region @xmath86 to be an invariant region of map @xmath35 .
since the system ( [ y1y2 ] ) is symmetric with respect to @xmath88-axis it is sufficient to consider only @xmath89 region . assuming @xmath90 , where @xmath91 , let us obtain the conditions on the parameter values for which @xmath92 . in region
@xmath93 we find that @xmath94 it follows from ( [ 13 ] ) that @xmath92 if the following inequalities @xmath95 are satisfied . using the symmetry of @xmath35 with respect to @xmath96 , we can restrict the above condition to the right part of @xmath86 , and since @xmath97 the inequalities ( [ 14 ] ) are satisfied if @xmath98 the first inequalities in ( [ 15])impose the following condition on @xmath39 : @xmath99 thus , under the condition ( [ 16 ] ) @xmath100 .
similarly we find that the image of @xmath101 by @xmath35 is also included in @xmath86 , for the parameter in @xmath49 . and periodic orbit @xmath69 , and discontinuity lines @xmath102 on the phase plane of map @xmath35 . ] finally , @xmath86 is invariant region of map @xmath35 in the parameter region @xmath103 on the parameter plane @xmath76 . ] figure @xmath104 shows region @xmath105 in the parameter plane @xmath106 . in this plane
the boundary of @xmath105 consists of three components : @xmath107 the line @xmath108 corresponds to the appearance of periodic orbit @xmath69 and changing stability of the fixed point @xmath78 .
the line @xmath109 is the boundary of the monostability of the uncoupled , @xmath110 , maps ( [ 2d ] ) .
the points of the curve @xmath111 corresponds to the `` tangency '' of the separatricies @xmath112 and @xmath113 , and for these parameter values @xmath114 .
therefore , if the parameter values of system ( [ y1y2 ] ) belongs to region @xmath105 , then invariant ( absorbing ) region @xmath86 and @xmath115 exists in the phase plane .
consequently , @xmath86 contains strange ( chaotic ) attractor @xmath116 of the map @xmath35 .
figure 5 illustrates possible structure of the attractor @xmath116 in the phase plane . on the phase plane @xmath117 .
parameter values : @xmath118 . ]
to characterize the complexity of the chaotic attractor @xmath116 we calculated numerically its fractal dimension @xmath119 .
it appears that , indeed , @xmath120 takes non - integer values greater than @xmath121 ( fig .
note that the box dimension of the set increases with the increase of coupling coefficient @xmath39 .
corresponding estimate of the dimension @xmath122 using lyapunov exponents @xmath123 is shown by dashed curve .
of the chaotic attractor @xmath116 calculated by box counting method versus coupling coefficient @xmath39 .
parameter values : @xmath124 . ] ) .
@xmath125 are shown by solid and dashed curves , respectively .
( b ) spike - burst behavior of binary variables @xmath126 .
parameter values : @xmath118 . ]
* c. chaotic oscillations and attractor . * figure 7 ( a ) illustrates time evolution of the variables @xmath127 corresponding to chaotic attractor @xmath116 .
let us characterize the oscillations in terms of original model ( [ fhn ] ) describing neuron excitability .
note that the neuron excitation threshold accounted by saddle separatrix ( fig .
1 ) corresponds to the discontinuity lines @xmath128 in the map description .
thus , if a map trajectory jumps above these lines then we can refer this event as a neuron spike , if it evolves below @xmath128 , the neuron is not excited .
in such a way the map oscillations can be described with binary variables @xmath129 : @xmath130 the evolution of @xmath129-variables is shown in fig . 7 ( b ) .
it appears that map oscillation represent sequence of bursts containing anti - phase spiking .
note that characteristic time scales of oscillations ( burst duration and inter - burst period ) depend coupling @xmath39 . for smaller
@xmath39 one can find longer lasting subthreshold period and shorter burst period .
then , the map describes a kind of chaotic spike - burst behavior typical for many neural systems .
the global behavior of the chaotic dynamics is described by the probability of finding the trajectory in any given region of the attractor
. it can be visualized by a distribution of a cloud of points , each moving under the deterministic mapping .
a stationary distribution may be reached by the system after long term evolution . to obtain this distribution ,
we start from a histogram of clouds of points in each small `` box '' of the phase space .
these points are mapped by @xmath35 and a new cloud is thus obtained .
thus to the initial probability density @xmath131 correspond a new density @xmath132 .
the operator which maps @xmath133 is called the perron - frobenius operator : @xmath134 defined by : @xmath135 for any region @xmath116 of the phase space .
the stationary density @xmath136 is an eigenfunction of @xmath137 corresponding to the eigenvalue @xmath121 .
the existence and uniqueness of the stationary distribution has been the object of many works @xcite .
it is nevertheless difficult to obtain analytical exact expression of @xmath138 , apart from the restricted case of markov maps .
we shall use an approximating algorithm inspired from the method of ding and zhou @xcite .
the phase space of the coupled system is divided into @xmath139 identical rectangles @xmath140 .
we consider an initial probability density @xmath141 that is constant on each @xmath140 : @xmath142 where @xmath143 , @xmath144 is the lebesgue measure of @xmath140 and @xmath145 is the probability of finding a phase point in @xmath146 is the characteristic function of @xmath140 : @xmath147 it is clear that @xmath132 has no reason to be of the `` coarse - grained '' form ( [ 18 ] ) .
smoothing this density by integrating it on each @xmath148 we obtain by using the definition ( [ 17 ] ) : @xmath149dy \\ & = & \sum_i \frac{a_i(0)}{m(\delta_i ) } \int_{\delta_j } \chi_{\delta_i}(y)]dy \\ & = & \sum_i a_i(0 ) \frac{m(f^{-1}(\delta_j ) \bigcap \delta_i)}{m(\delta_i ) } \\ \end{array}\ ] ] thus , the transition probability from @xmath140 to @xmath148 is given by the stochastic matrix : @xmath150 the matrix @xmath151 is an approximate value of the perron - frobenius operator @xmath137 .
it can be proved that @xmath152 converges to @xmath137 as @xmath153 @xcite .
( [ 19 ] ) can be written under the matrix form : @xmath154 where @xmath155 is the row vector @xmath156 with @xmath157 .
thus , an approximate stationary probability is the row eigenvector @xmath158 such that : @xmath159 we first calculate the matrix elements of @xmath152 and then we compute the eigenvector @xmath2 . to calculate @xmath160 we use ( [ 20 ] ) . in each rectangle
@xmath140 we put @xmath161 points uniformly distributes - then we compute the number of point @xmath162 such that @xmath163 , and we divide this number by the number of points in @xmath140 .
this provides @xmath160 .
we used this method to approximate the stationary distribution for @xmath164 and @xmath165 ( fig .
8) showing a density distributed over some region , which is to be compared with the fig .
5 where we have used only one trajectory .
it is to be noted that the last figure shows the attractor generated by one trajectory whereas fig.8 shows the approximate density of the attractor .
the above method allows to obtain the statistical distribution of synchronized spikes .
recall that the variable @xmath166 is spiking if and only if @xmath167 and @xmath168 is spiking if and only if @xmath169 , and no one is spiking if @xmath170 which we denote @xmath66 .
recall also that the region @xmath68 corresponds to @xmath166 and @xmath168 are simultaneously spiking , which lies outside the invariant region .
moreover , because @xmath171 are disjoint , @xmath166 and @xmath168 can not be simultaneously spiking .
thus we shall consider a partition of the invariant domain @xmath86 corresponding to one of the three possible states of spiking @xmath172 : @xmath173 where @xmath174 means that the only spiking variable is @xmath175 ; and @xmath176 means that no one is spiking .
this partition induces symbolic dynamics , that we shall study .
we calculate the stochastic matrix @xmath177 from any one of these states to any other , where @xmath178 are the states @xmath179 , for two values of the coupling parameter @xmath180 ( see fig .
stochastic matrices of transition probabilities for three values of the coupling d. ] this table shows that we have only the following probable transitions in one step : @xmath181 and @xmath182 .
less probable transitions are : @xmath183 @xmath184 but these transition probabilities depend monotonically on the coupling @xmath39 .
thus the transitions @xmath185 and @xmath186 ( i.e. successive spikes ) decreases slightly with coupling .
so , strengthening the coupling in this range decreases the bursting probability .
we can understand these occurrences in considering the intersections of the image of each region by @xmath35 with these same regions ( see fig .
. the most probable transition @xmath187 and @xmath188 correspond respectively to @xmath189 and @xmath190 .
the less probable transitions @xmath191 and @xmath192 correspond respectively to @xmath193 and @xmath194 .
the positive lebesgue measure of these intersected regions give the transition probabilities of each of these occurrences .
some of the transition probabilities given by the calculation of the lebesgue measures of the intersected regions are shown in fig .
11 , with their dependance over the coupling constant @xmath39 , which can be compared to those obtained numerically in the table fig .
all the calculations are made in appendix a , where we also show that , when the parameters are in @xmath105,a neuron can not fire twice , i.e. @xmath195 . of the partitions of the invariant region @xmath86 . ] ; @xmath124 . ]
we can now check the memory effects in studying the conditional probability of a state given the past .
if we denote by @xmath196 the values of the variable @xmath18 at time , @xmath197 etc , we like to know if conditional probability @xmath198 is equal to @xmath199 or not . in the first case
there is no memory of the process . in order to compute the above conditional probability we have to find @xmath200 which implies @xmath201 calculations .
but , neglecting rare events like @xmath202 having @xmath203 probability for @xmath165 , we have only @xmath104 significant transitions .
@xmath204 the first one is the probability of having @xmath205 if it is preceded by a sequence of @xmath205 and @xmath206 and for the others the definition is the same .
thus we obtain the following table ( fig.12 ) . the fact that the probability @xmath207 is not equal to @xmath208 means that the system has acquired a memory , the chain is not a markov chain .
the memory of a chain is given by the smallest positive integer @xmath209 such that : @xmath210 . it is also remarkable that the probability of having @xmath176 if it is preceded by the sequence @xmath211 is @xmath212 the double of the probability of having @xmath176 preceded by @xmath206 simply @xmath213 .
this explain the frequent appearance of sequence of spikes .
how could we estimate the memory of the process depends on the full calculation of all conditional probabilities for all past .
another very interesting aspect is the dependence of this memory on the coupling coefficient as shown in fig .
it is clear that the memory effects increase with coupling ( the distance between the two curves increases with coupling ) . .
we have investigated chaotic dynamics of two coupled maps each formally derived from fitzhugh - nagumo model of neuron excitability .
being uncoupled such maps are trivial with only one stable fixed point corresponding to neuron rest state .
the excitation threshold is given by the discontinuity point .
we have introduced the formal definition of spike if the map evolves above this threshold .
the linear `` diffusive '' coupling term introduced for the maps can be treated as a kind of electrical inter - neuron coupling that mimics an `` integral '' coupling current between the two cells .
we have shown that increasing the coupling coefficient above certain threshold leads to the appearance of chaotic attractor .
the attractor appears in the invariant region of the phase space confined by the invariant manifolds of saddle periodic orbit .
this region attracts trajectories from outside and does not contain any stable trajectory inside .
the oscillations emerging with such an attractor have a spike - burst shape with synchronous bursts with anti - phase spiking . using probabilistic description
we have found the probabilities to find a spike in different conditions .
it has been also found that the system acquires a memory ( in contrast with markov s processes ) .
this research has been supported by russian foundation for basic research ( grants 03 - 02 - 17135 ) and by grant of president of russian federation ( mk 4586.2004.2 ) .
v.b.k . acknowledges russian science support foundation for financial support .
v.i.n . acknowledges the financial support from university paris vii .
in order to understand the oscillatory behavior it is interesting to look at the image of the invariant region after one iteration ( see fig . 10 ) .
first of all , we have to determine the images of the regions of the invariant region , defined by the discontinuities , namely @xmath66 , @xmath214 and @xmath215(e.q . to @xmath216 ) .
@xmath66 is a polygon having for vertices @xmath217 , @xmath218 , @xmath219 and @xmath220 ; @xmath214 , @xmath221 , @xmath218 , @xmath222 and @xmath220 ; and @xmath215 , @xmath217 , @xmath223 , @xmath224 and @xmath220 ( when @xmath225 or equivalent , which is always satisfied for @xmath226 ) .
let us compute the images of all these vertex , according to the definition of @xmath35 in each region .
we have to keep in mind that some of these points belong to the lines of discontinuity ; in this case , the computation is taken in the sense of a limit starting from the interior of each region .
we first shall see that @xmath231 is completely included in @xmath232 and excluded of @xmath214 , which implies that one neuron can not spike twice . for the simplicity of proof
, we shall use indirect calculations .
we are going to show that : the first three inequalities allow us to say that @xmath234 is under and on the left of @xmath235 ; but , as @xmath217 is on the segment @xmath236 $ ] , @xmath237 is on the segment @xmath238 $ ] , and so it is under and on the left of @xmath235.the last inequality show that @xmath235 ( and by consequence@xmath237)is not in @xmath239 .
99 johnson s.w . , seutin v. and north r. a. , science , * 258 * , 665 ( 1992 ) e.r .
kandel , j.h .
schwartz , t.m .
jessell ( eds . ) _ principles of neural science .
third edition _
( prentice - hall intern .
a. scott , neuroscience : a mathematical primer ( oxford univ .
press , oxford , 1998 ) .
murray , mathematical biology ( springer - verlag , berlin , 1989 ) .
y. manor , j. rinzel , i. segev , and y. yarom , j. neurophysiol . * 77 * , 2736 ( 1997 ) .
y. loewenstein , y. yarom , and h. sompolinsky , proc .
98 * , 8095 ( 2001 ) .
a. sherman and j. rinzel , proc .
89 * , 2471 ( 1992 ) x .- j .
wang , d. golomb and j. rinzel , proc . natl .
92 * , 5577 ( 1995 ) n.f .
rulkov , phys .
86 * , 183 ( 2001 ) ; phys .
e * 65 * 041922 ( 2002 ) .
g. de vries , phys .
e * 64 * , 051914 ( 2001 ) .
b. cazelles , m. courbage and m. rabinovich , europhys .
lett . * 56*(4 ) 504 ( 2001 ) .
c.c . chow and n. kopell , neural computation * 12 * , 1643 ( 2000 ) r.d .
et al _ , phys .
e * 62 * , 2644 ( 2000 ) .
n.f . rulkov _
et al . _ ,
e * 64 * , 016217 ( 2001 ) .
izhikevich , int .
j. bifurc . chaos *
10*(6 ) , 1171 ( 2000 ) ; s.p .
dawson , m.v .
dangelo and j.e .
pearson , phys .
a * a265 * , 346 ( 2000 ) .
kazantsev , phys .
e * 64 * , 056210 ( 2001 ) . | the dynamics of two coupled piece - wise linear one - dimensional monostable maps is investigated .
the single map is associated with poincar section of the fitzhugh - nagumo neuron model .
it is found that a diffusive coupling leads to the appearance of chaotic attractor .
the attractor exists in an invariant region of phase space bounded by the manifolds of the saddle fixed point and the saddle periodic point .
the oscillations from the chaotic attractor have a spike - burst shape with anti - phase synchronized spiking .
* spiking - bursting activity is a common feature of the temporal organization in many neural firing patterns @xcite .
bursting activity means that clusters of spikes ( of action potentials ) occur more or less rhythmically and separated by phases of quiescence .
spiking - bursting dynamics can be regular or chaotic depending on the concentration of neuromodulators , currents and other control parameters .
for example , many of the talamocortical neurons from central patterns generators generate chaotic spiking - bursting dynamics . in order to understand the emergence of chaotic oscillations in a neurones network , we use a two variables fitzhugh - nagumo model of the membrane potential of an isolated neural cell . for appropriate values of the parameters ,
the system possesses a stable fixed point a focus surrounded by a separatrix such that for large enough perturbations , the system responds by an excitation pulse which also decays to the stable focus .
hence , introducing a section by a half - line , we obtain a one - dimensional piece - wise smooth map with one discontinuity point representing the excitation threshold .
then we consider a difference coupling between two such maps . in the linear approximation of the maps and for suitable values of the parameters ,
this two - dimensional system is locally hyperbolic having discontinuity lines and a strange attractor .
the interesting fact is that these lines define regions in the phase space corresponding to the thresholds of excitability of each one of these neurones , or none of them .
the symbolic dynamics represents then the bursting activity of the neurones .
it is then possible to apply the approximation tools of statistical analysis of time series generated by the system , namely through perron - frobenius operator , in order to study and interpret some important features of neural networks , like synchronization and time correlations . * |
in their pioneering work @xcite , gupta and kumar posed and studied the limits of communication in ad hoc wireless networks . assuming @xmath0 nodes are uniformly distributed in a plane and grouped into source - destination ( s - d ) pairs at random , they showed that one can achieve a sum throughput of @xmath8 .
this is achieved using a multihop transmission scheme in which nodes transmit to one of the nodes in their neighboring cells , requiring full connectivity with at least one node per cell . a trade - off between throughput and delay of fully - connected networks
was studied in @xcite and was extended in @xcite to trade - offs between throughput , delay as well as energy .
the work in @xcite has studied relay networks in which a single source transmits its data to the intended destination using the other nodes as relays . using percolation theory @xcite , they showed that a constant rate is achievable for a single s - d pair if we allow a small fraction of nodes to be disconnected .
this result can be applied to ad hoc networks having multiple s - d pairs and the work in @xcite proposed an indirect multihop routing protocol based on such partial connectivity , that is all s - d pairs perform multihop transmissions based on this partially - connected sub - network .
they showed that the indirect multihop routing improves the achievable sum throughput as @xmath9 .
information - theoretic outer bounds on throughput scaling laws of ad hoc wireless networks were derived in @xcite .
these bounds showed that the multihop routing using neighbor nodes is order - optimal in the power - limited and high attenuation regime .
recently , a hierarchical cooperation scheme was proposed in @xcite and was shown to achieve better throughput scaling than the multihop strategy in the interference - limited or low attenuation regime , achieving a scaling very close to their new outer bound .
a more general hierarchical cooperation was proposed in @xcite , which works for an arbitrary node distribution in which a minimum separation between nodes is guaranteed .
recently _ hybrid network _
models have been studied as well .
hybrid networks are ad hoc networks in which the nodes communication is aided by additional infrastructures such as base stations ( bss ) .
these are generally assumed to have high bandwidth connections to each other . in @xcite
the connectivity of hybrid networks has been analyzed . in @xcite
the throughput scaling of hybrid networks has been studied . in order for a hybrid network s throughput scaling to outperform that of a strictly ad hoc network , it was determined that the number of bss should be greater than a certain threshold @xcite .
the existing literatures have focused on the throughput scaling of a _ single _ network .
however , the necessity of extending and expanding results to capture _ multiple _ overlapping networks is becoming apparent .
recent measurements have shown that despite increasing demands for bandwidth , much of the currently licensed spectrum remains unused a surprisingly large portion of the time @xcite . in the us , this has led the federal communications commission ( fcc ) to consider easing the regulations towards _ secondary spectrum sharing _ through their _ secondary markets initiative _ @xcite . the essence of secondary spectrum sharing involves having primary license holders allow secondary license holders to access the spectrum .
different types of spectrum sharing exist but most agree that the primary users have a higher priority access to the spectrum , while secondary users _ opportunistically _ use it .
these secondary users often require greater sensing abilities and more flexible and diverse communication abilities than legacy primary users .
secondary users are often assumed to be _ cognitive radios _ , or wireless devices which are able to transmit and receive according to a variety of protocols and are also able to sense and independently adapt to their environment @xcite .
these features allow them to behave in a more `` intelligent '' manner than current wireless devices . in this paper , we consider _ cognitive networks _ , which consist of secondary , or cognitive , users who wish to transmit over the spectrum licensed to the primary users .
the single - user case in which a single primary and a single cognitive s - d pairs share the spectrum has been considered in the literature , see for example @xcite and the references therein . in @xcite the primary and cognitive s - d pairs are modeled as an interference channel with asymmetric side - information . in @xcite the communication opportunities are modeled as a two - switch channel . recently , a single - hop cognitive network was considered in @xcite , where multiple secondary s - d pairs transmit in the presence of a single primary s - d pair .
it was shown that a linear scaling law of the single - hop secondary network is obtained when its operation is constrained to guarantee a particular outage constraint for the primary s - d pair .
we study a more general environment in which a _ primary ad hoc network _ and a _ cognitive ad hoc network _ both share the same space , time and frequency dimensions . two types of primary networks are considered in this paper @xmath10 : an ad hoc primary network and an infrastructure - supported primary network . for the ad hoc primary model ,
the primary network consists of @xmath0 nodes randomly distributed and grouped into s - d pairs at random . for the infrastructure - supported primary model , additional @xmath1 bss are regularly deployed and used to support the primary transmissions . in both cases , the cognitive network consists of @xmath2 secondary nodes distributed randomly and s - d pairs are again chosen randomly .
our main assumptions are that ( 1 ) the primary network continues to operate as if no secondary network were present , ( 2 ) the secondary nodes know the locations of the primary nodes and ( 3 ) the secondary network is denser than the primary network . under these assumptions , we will illustrate routing protocols for the primary and secondary networks that result in the _ same throughput scaling _ as if each were a single network .
note that the constraint that the primary network does _ not _ alter its protocol because of the secondary network is what makes the problem non - trivial .
indeed , if the primary network were to change its protocol when the secondary network is present , a simple time - sharing scheme is able to achieve the throughput scaling of homogeneous networks for both primary and secondary networks . for the ad hoc primary model
, we use a routing protocol that is a simple modification of the nearest neighbor multihop schemes in @xcite . for the infrastructure - supported primary model
, we use a bs - based transmission similar to the scheme in @xcite .
we propose novel routing protocols for the secondary network under each primary network model .
our proposed protocols use multihop routing , in which the secondary routes _ avoid _ passing too close to the primary nodes , reducing the interference to them .
we show that the proposed protocols achieve the throughput scalings of homogeneous networks _
simultaneously_. this implies that when a denser `` intelligent '' network is layered on top of a sparser oblivious one , then both may achieve the same throughput scalings as if each were a single network .
this result may be extended to more than two networks , provided each layered network obeys the same three main assumptions as in the two network case .
this paper is structured as follows . in section [ sec : model ]
we outline the system model : we first look at the network geometry , co - existing primary and secondary ad hoc networks , then turn to the information theoretic achievable rates before stating our assumptions on the primary and secondary network behaviors . in section [ sec : adhoc ]
we outline the protocols used for the ad hoc primary model and prove that the claimed single network throughput scalings may be achieved .
we also prove the claimed single network throughput scalings for the infrastructure - supported primary model in section [ sec : infra ] .
we conclude in section [ sec : conclusion ] and refer the proofs of the lemmas to the appendix .
in order to study throughput scaling laws of ad hoc cognitive networks , we must define the underlying network models .
we first explain the two geometric models that will be considered in sections [ sec : adhoc ] and [ sec : infra ] .
we then look at the transmission schemes , resulting achievable rates , and assumptions made about the primary and secondary networks . throughout this paper
, we use @xmath11 to denote the probability of an event @xmath12 and we will be dealing with events which take place with high probability ( w.h.p . ) , or with probability 1 as the node density tends to infinity . ] .
we consider a planar area in which a network of primary nodes and a network of secondary nodes co - exist .
that is , the two networks share the same space , time , code , and frequency dimensions .
two types of networks are considered as the primary network : an _ ad hoc network _ and an _ infrastructure - supported network _ , while the secondary network is always ad hoc .
the two geometric models are illustrated in fig .
[ fig:2models ] . as shown in fig .
[ fig:2models ] .
( a ) , in the ad hoc primary model , nodes are distributed according to a poisson point process ( p.p.p . ) of density @xmath0 over a unit square , which are randomly grouped into primary s - d pairs . for the secondary network ,
nodes are distributed according to a p.p.p . of density
@xmath2 over the same unit square and are also randomly grouped into secondary s - d pairs .
our second model is the infrastructure - supported primary model , shown in fig .
[ fig:2models ] .
( b ) . there , primary nodes are still randomly distributed over the square according to a p.p.p . of density @xmath0 , but these nodes are supported by additional @xmath1 regularly spaced bss ( the number of bss is equal to @xmath1 , which is also the density of bss ) .
the bss sole purpose is to relay data for the primary network , they are neither sources nor destinations .
we assume that the bss are connected to each other through wired lines of capacity large enough such that the bs - bs communication is not the limiting factor in the throughput scaling laws .
secondary nodes again form an ad hoc network with random s - d pairs , distributed according to a p.p.p . of density @xmath2 .
the densities of the @xmath0 primary nodes , @xmath2 secondary nodes , and @xmath1 bss are related according to @xmath13 where @xmath14 and @xmath15 .
we focus on the case where the density of the secondary nodes is higher than that of the primary nodes .
we also assume that the densities of both the primary nodes and secondary nodes are higher than that of the bss , which is reasonable from a practical point of view .
the wireless propagation channel typically includes path loss with distance , shadowing and fading effects .
however , in this work we assume the channel gain depends only on the distance between a transmitter and its receiver , and ignore shadowing and fading .
thus , the channel power gain @xmath16 , normalized by a constant , is given by @xmath17 where @xmath18 denotes the distance between a transmitter ( tx ) and its receiver ( rx ) and @xmath19 denotes the path - loss exponent .
each network operates based on slotted transmissions .
we assume the duration of each slot , and the coding scheme employed are such that one can achieve the additive white gaussian noise ( awgn ) channel capacity . for a given signal to interference and noise ratio ( sinr ) , this capacity is given by the well known formula @xmath20 bps / hz assuming the additive interference is also white , gaussian , and independent from the noise and signal .
we assume that primary slots and secondary slots have the same duration and are synchronized with each other .
we further assume all the primary , secondary , and bs nodes are subject to a transmit power constraint @xmath21 .
we now characterize the rates achieved by the primary and secondary transmit pairs .
suppose that @xmath22 primary pairs and @xmath23 secondary pairs communicate simultaneously . before proceeding with a detailed description ,
let us define the notations used in the paper , given by table [ table : simbols ] .
then , the @xmath24-th primary pair can communicate at a rate of @xmath25 where @xmath26 denotes the euclidean norm of a vector .
@xmath27 and @xmath28 are given by @xmath29 and @xmath30 similarly , the @xmath31-th secondary pair can communicate at a rate of @xmath32 where @xmath33 and @xmath34 are given by @xmath35 and @xmath36 throughout the paper , the achievable per - node throughput of the primary and secondary networks are defined as follows .
a throughput of @xmath37 per primary node is said to be achievable w.h.p .
if all primary sources can transmit at a rate of @xmath37 ( bps / hz ) to their primary destinations w.h.p . in the presence of the secondary network .
let @xmath38 denote an outage probability of the secondary network , which may vary as a function of @xmath2 .
a throughput of @xmath39 per secondary node is said to be @xmath40-achievable w.h.p .
if at least @xmath41 fraction of secondary sources can transmit at a rate of @xmath39 ( bps / hz ) to their secondary destinations w.h.p . in the presence of the primary network . for both ad hoc and infrastructure - supported primary models
, we will propose secondary routing schemes that make @xmath42 as @xmath43 is equivalent to @xmath44 since @xmath45 . ] .
thus , although we allow a fraction of secondary s - d pairs to be in outage , for sufficiently large @xmath2 , almost all secondary s - d pairs will be served at a rate of @xmath39 .
let us define @xmath46 as the sum throughput of the primary network , or @xmath37 times the number of primary s - d pairs since the nodes are thrown at random according to a p.p.p . of density @xmath0 .
the actual number of nodes in the network will vary in a particular realization . ] .
similarly , we define @xmath47 as the sum throughput of the secondary network , or @xmath39 times the number of _ served _ secondary s - d pairs at a rate of @xmath39 . while @xmath37 and @xmath46 represent the per - node and sum throughputs of the primary network _ in the presence of the secondary network _ , we use the notations @xmath48 and @xmath49 to denote the per - node and sum throughputs of the primary network _ in the absence of the secondary network _ , respectively . as primary and secondary nodes must share the spectrum , the rules or assumptions made about this co - existence are of critical importance to the resulting achievable throughputs and scaling laws .
primary networks may be thought of as existing communication systems that operate in licensed bands .
these primary users are the license holders , and thus have higher priority access to the spectrum than secondary users .
thus , our first key assumption is that _ the primary network does not have to change its protocol due to the secondary network . _ in other words , all primary s - d pairs communicate with each other as intended , regardless of the secondary network .
the secondary network , which is opportunistic in nature , is responsible for reducing its interference to the primary network to an `` acceptable level '' , while maximizing its own throughput @xmath39 .
this acceptable level may be defined to be one that does not degrade the throughput scaling of the primary network .
more strictly , the secondary network should satisfy w.h.p .
@xmath50 during its transmission , where @xmath51 is the maximum allowable fraction of throughput loss for the primary network .
notice that the above condition guarantees @xmath52 .
the secondary network may ensure ( [ eq : fraction_loss ] ) by adjusting its protocol based on information about the primary network .
thus , our second key assumption is that _ the secondary network knows the locations of all primary nodes .
_ since the secondary network is denser than the primary network , each secondary node can measure the interference power from its adjacent primary node and send it to a coordinator node . based on these measured values ,
the secondary network can establish the locations of the primary nodes .
we first consider the throughput scaling laws when both the primary and secondary networks are ad hoc in nature . since the primary network needs not change its transmission scheme due to the presence of the secondary network , we assume it transmits according to the direct multihop routing similar to those in @xcite and @xcite .
we also consider the indirect multihop routing proposed in @xcite as a primary protocol . of greater interest
is how the secondary nodes will transmit such that the primary network remains unaffected in terms of throughput scaling .
the main results of this section describe achievable throughput scaling laws of the primary and secondary networks .
we simply state these results here and derive them in the remainder of this section .
_ suppose the ad hoc primary model .
for any @xmath51 , the primary network can achieve the following per - node and sum throughputs w.h.p .
: @xmath53 where @xmath54 and @xmath55 .
the following per - node and sum throughputs are @xmath40-achievable w.h.p . for the secondary network : @xmath56 where @xmath57 , which converges to zero as @xmath58 . _
this result is of particular interest as it shows that not only can the primary network operate at the same scaling law as when the secondary network does not exist , but the secondary network can also achieve , with an arbitrarily small fraction of outage , the exact same scaling law obtained by the direct multihop routing as when the primary network does not exist .
thus almost all secondary s - d pairs can communicate at a rate of @xmath39 in the limit of large @xmath2 .
in essence , whether the indirect multihop or the direct multihop is adopted as a primary protocol , the secondary network can achieve the sum throughput of @xmath59 w.h.p . while preserving @xmath60 fraction of the primary network s stand - alone throughput . in the remainder of this section , we first outline the operation of the primary network and then focus on the design of a secondary network protocol under the given primary protocol .
we analyze achievable throughputs of the primary and secondary networks , which will determine the throughput scaling of both co - existing networks . throughout this work ,
we place the proofs of more technical lemmas and theorems in the appendix and outline the main proofs in the text .
we assume the primary network communicates according to the direct multihop routing protocol .
the indirect multihop routing will be explained in section [ subsec : indirect_routing ] , which can be extended from the results of the direct routing .
the challenge is thus to prove that the secondary nodes can exchange information in such a way that satisfies @xmath61 w.h.p .. we first outline a primary network protocol , and then design a secondary network protocol which operates in the presence of the primary network .
we assume that the primary network delivers data using the direct multihop routing , in a manner similar to @xcite and @xcite .
the basic multihop protocol is as follows : * divide the unit area into square cells of area @xmath62 . *
a @xmath63- time division multiple access ( tdma ) scheme is used , in which each cell is activated during one out of @xmath63 slots .
* define the horizontal data path ( hdp ) and the vertical data path ( vdp ) of a s - d pair as the horizontal line and the vertical line connecting a source to its destination , respectively .
each source transmits data to its destination by first hopping to the adjacent cells on its hdp and then on its vdp .
* when a cell becomes active , it delivers its traffic .
specifically , a tx node in the active cell transmits a packet to a node in an adjacent cell ( or in the same cell ) .
a simple round - robin scheme is used for all tx nodes in the same cell .
* at each transmission , a tx node transmits with power @xmath64 , where @xmath18 denotes the distance between the tx and its rx .
this protocol requires full connectivity , meaning that each cell should have at least one node .
let @xmath65 denote the area of a primary cell .
the following lemma indicates how to determine @xmath65 satisfying this requirement . [ thm : node_number ] the following facts hold .
+ ( a ) the number of primary nodes in a unit area is within @xmath66 w.h.p . , where @xmath67 is an arbitrarily small constant .
+ ( b ) suppose @xmath68 .
then , each primary cell has at least one primary node w.h.p .. the proof is in the appendix .
based on lemma [ thm : node_number ] , we set @xmath68 . under the given primary protocol ,
@xmath69 and @xmath70 are achievable w.h.p . when the secondary network is absent or silent .
results similar to lemma [ thm : node_number ] can be found in @xcite and @xcite , where their proposed schemes also achieve the same @xmath48 and @xmath49 .
note that the gupta - kumar s model @xcite , @xcite assumes uniformly distributed nodes in the network and a constant rate between tx and rx if sinr is higher than a certain level .
although we assume that the network is constructed according to a p.p.p .
( rather than uniform ) and that the information - theoretic rate @xmath71 is achievable ( rather than a constant rate ) , the above primary network protocol provides the same throughput scaling as that under the gupta - kumar s model .
since the secondary nodes know the primary nodes locations , an intuitive idea is to have the secondary network operate in a multihop fashion in which they circumvent each primary node in order to reduce the effect of secondary transmissions to the primary nodes . in @xcite
a network with holes is considered and geographic forwarding algorithms that establish routing paths around holes are proposed . around
each primary node we define its _ preservation region : _ a square containing @xmath63 secondary cells , with the primary node at the center cell .
the secondary nodes , when determining their routing paths , need to avoid these preservation regions : our protocol for the secondary ad hoc network is the same as the basic multihop protocol except that * the secondary cell size is @xmath72 . * at each transmission a secondary node transmits its packet _ three _ times repeatedly ( rather than once ) using three slots .
* the secondary paths avoid the preservation regions ( see fig .
[ fig : secodary_routing_adhoc ] ) .
that is , if the hdp or vdp of a secondary s - d pair is blocked by a preservation region , this data path circumvents the preservation region by using its adjacent cells .
if a secondary source ( or its destination ) belongs to preservation regions or its data path is disconnected by preservation regions , the corresponding s - d pair is not served .
* at each transmission , a tx node transmits with power @xmath73 , where @xmath18 denotes the distance between the tx and its rx and @xmath74 .
since @xmath75 converges to zero as @xmath58 , there exists @xmath76 such that the power constraint is satisfied for any @xmath74 if @xmath77
. we will show in lemma [ thm : primary_rate_adhoc1 ] that adjusting @xmath78 induces a trade - off between the rates of the primary and secondary networks while the scaling laws of both networks are unchanged , which allows the condition ( [ eq : fraction_loss ] ) to be meet .
unlike the primary protocol , each secondary cell transmits a secondary packet three times repeatedly when it is activated . as we will show later
, the repeated secondary transmissions can guarantee the secondary receivers a certain minimum distance from all primary interferers for at least one packet , thus guaranteeing the secondary network a non - trivial rate .
therefore , the duration of the secondary @xmath63-tdma scheme is three times longer than that of the primary @xmath63-tdma .
the main difference between this scheme and previous multihop routing schemes is that the secondary multihop paths must circumvent the preservation regions and that a portion of secondary s - d pairs is not served .
but this portion will be negligible as @xmath43 . by re - routing the secondary nodes transmission around the primary nodes preservation regions
, we can guarantee the primary nodes a non - trivial rate .
similar to lemma [ thm : node_number ] , we can also prove that the total number of secondary nodes is within @xmath79 w.h.p . and that each secondary cell has at least one secondary node w.h.p .. in this subsection
, we analyze the per - node and sum throughputs of each network under the given protocols and derive throughput scaling laws with respect to the node densities .
let us consider the primary network in the presence of the secondary network .
we first show that each primary cell can sustain a constant aggregate rate ( lemma [ thm : primary_rate_adhoc1 ] ) , which may be used in conjunction with the number of data paths each primary cell must transmit ( lemma [ thm : primary_traffic_adhoc ] ) to obtain the per - node and sum throughputs in theorem [ thm : primary_throughput_adhoc ] .
let @xmath80 and @xmath81 denote the achievable aggregate rate of each primary cell in the presence and in the absence of the secondary network , respectively .
we define @xmath82 having a finite value for @xmath19 , which will be used to derive an upper bound on the interference power of the ad hoc primary and secondary networks . then the following lemma holds .
[ thm : primary_rate_adhoc1 ] suppose the ad hoc primary model .
if @xmath83 , then @xmath84 where @xmath85 and @xmath86 is given by ( [ eq : def_i ] ) . moreover , @xmath81 is lower bounded by @xmath87 , where @xmath88 is a constant independent of @xmath0 .
the proof is in the appendix .
the essence of the proof of lemma [ thm : primary_rate_adhoc1 ] lies in showing that the secondary nodes , even as @xmath43 , do not cause the aggregate rate of each primary cell to decay with @xmath2 .
this is done by introducing the preservation regions , which guarantee the minimum distance of @xmath89 from all secondary txs to the primary rxs .
this lemma will be used to show that ( [ eq : fraction_loss ] ) can be satisfied w.h.p . if @xmath83 in theorem [ thm : primary_throughput_adhoc ] .
the next lemma determines the number of data paths that each cell should carry . to obtain an upper bound
, we extend each hdp to the entire horizontal line and all cells through which this horizontal line passes should deliver the corresponding data of hdp ( see fig .
[ fig : routing_analysis_adhoc ] ) .
similarly , we extend each vdp to the entire vertical line .
we define this entire horizontal and vertical line as an _ extended _ hdp and an _ extended _ vdp , respectively . throughout the rest of the paper , our analysis deals with extended hdps and vdps instead of original hdps and vdps .
since we are adding hops to our routing scheme , the extended traffic gives us a lower bound on the achievable throughput .
[ thm : primary_traffic_adhoc ] under the ad hoc primary model , each primary cell needs to carry at most @xmath90 data paths w.h.p .. the proof is in the appendix .
lemma [ thm : primary_traffic_adhoc ] shows how the number of data paths varies with the node density @xmath0 .
lemmas [ thm : node_number]-[thm : primary_traffic_adhoc ] may be used to prove the main theorem , stated next . [ thm : primary_throughput_adhoc ] suppose the ad hoc primary model . for any @xmath51 , by setting @xmath91 , the primary network can achieve @xmath92 and @xmath93 w.h.p . , where @xmath94 and @xmath95 the definitions of @xmath96 and @xmath87 are given in lemma [ thm : primary_rate_adhoc1 ] .
first consider the stand - alone throughput of the primary network .
since each primary cell can sustain a rate of @xmath87 ( lemma [ thm : primary_rate_adhoc1 ] ) , each primary s - d pair can achieve a rate of at least @xmath87 divided by the maximum number of data paths per primary cell .
the number of data paths is upper bounded by @xmath90 w.h.p .
( lemma [ thm : primary_traffic_adhoc ] ) .
therefore , @xmath48 is lower bounded by @xmath97 w.h.p .. now the whole network contains at least @xmath98 primary s - d pairs w.h.p .
( lemma [ thm : node_number ] ) .
therefore , @xmath49 is lower bounded by @xmath99 w.h.p .. finally lemma [ thm : primary_rate_adhoc1 ] shows that , for any @xmath51 , if we set @xmath83 , then @xmath100 is achievable in the limit of large @xmath0 . since the number of primary data paths carried by each primary cell and the total number of primary s - d pairs in the network holds regardless of the existence of the secondary network , @xmath92 and @xmath93 are also achievable w.h.p . , which completes the proof .
let us now consider the per - node throughput of the secondary network in the presence of the primary network .
the main difference between the primary and secondary transmission schemes arises from the presence of the preservation regions .
recall that the secondary nodes wish to transmit according to a multihop protocol , but their path may be blocked by a preservation region . in this case , they must circumvent the preservation region , or possibly the _
cluster _ of primary preservation regions .
however , as we will see later circumventing these preservation regions ( clusters ) does not degrade the secondary network s throughput scaling due to the relative primary and secondary node densities : the secondary nodes increase at the rate @xmath101 and @xmath14 .
thus , intuitively , as the density @xmath0 of the primary nodes increases , the area of each preservation region ( which equals 9 secondary cells ) decreases faster than the increase rate of the primary node density ( and thus number of preservation regions ) .
these clusters of preservation regions remain bounded in size , although their number diverges as @xmath102 .
this can be obtained using percolation theory @xcite .
let us introduce a poisson boolean model @xmath103 on @xmath104 .
the points @xmath105 are distributed according to a p.p.p . of density @xmath106 and each point @xmath107
is the center of a closed ball with radius @xmath108 .
notice that @xmath108 s are random variables independent of each other and independent of @xmath109 , whose distributions are identical to that of @xmath110 .
the _ occupied _
region is the region that is covered by at least one ball and the _ vacant _ region is the complement of the occupied region .
note that the occupied ( or vacant ) region may consists of several occupied ( vacant ) components that are disjointed with each other
. then the following theorem holds .
[ thm : percolation ] for a poisson boolean model @xmath103 on @xmath104 , for @xmath111 , if @xmath112 , then there exists @xmath113 such that for all @xmath114 , @xmath115 we refer readers to the proof of theorem 3.3 in @xcite . by scaling the size of the above poisson boolean model and setting @xmath110 as a deterministic value , we apply theorem [ thm : percolation ] to our network model .
[ thm : clusters_of_preservation_regions ] any cluster of preservation regions has at most @xmath116 preservation regions w.h.p . , where @xmath117 is an integer independent of @xmath0 .
let us consider a poisson boolean model @xmath118 on @xmath119 .
all balls in this model have deterministic radii of @xmath120 and the density of the underlining p.p.p .
is a function of @xmath0 decreasing to zero as @xmath121 .
thus , @xmath122 and there exists @xmath123 such that @xmath124 for all @xmath125 . as a consequence , ( [ eq : percolation ] ) holds for all @xmath125 .
since this result holds on @xmath119 , the same result still holds if we focus on the area of @xmath126 ^ 2 $ ] instead of @xmath119 .
moreover , two poisson boolean models @xmath127 on @xmath126 ^ 2 $ ] and @xmath128 on @xmath129 ^ 2 $ ] show the same percolation result ( see proposition 2.6.2 in @xcite ) . therefore , under the poisson boolean model @xmath128 on @xmath129 ^ 2 $ ] , the number of balls in any occupied component is upper bounded by @xmath116 w.h.p . , where @xmath117 is an integer independent of @xmath0 .
in the case of @xmath128 on @xmath129 ^ 2 $ ] , the underlining p.p.p . is the same as that of the primary network and each ball contains the corresponding preservation region shown in fig .
[ fig : percolation ] .
thus preservation regions can not form a cluster if the corresponding balls do not form an occupied component , meaning the number of preservation regions in any cluster is also upper bounded by @xmath116 w.h.p .
, which completes the proof .
this corollary is needed to ensure that the secondary network remains connected , to bound the number of data paths that pass through secondary cells , and to prove the next lemma . as mentioned earlier , whenever a secondary source or destination lies within a primary preservation region or there is no possible data path , this pair is not served .
the next lemma shows that the fraction of these unserved secondary s - d pairs is arbitrarily small w.h.p .. [ thm : number_of_served_sd_adhoc ] under the ad hoc primary model , the fraction of unserved secondary s - d pairs is upper bounded by @xmath130 w.h.p . , which converges to zero as @xmath131 .
the proof is in the appendix .
next , lemma [ thm : secondary_rate_adhoc ] shows that , in the presence of the primary network , each secondary cell may sustain a constant aggregate rate .
[ thm : secondary_rate_adhoc ] under the ad hoc primary model , each secondary cell can sustain traffic at a rate of @xmath132 , where @xmath133 is a constant independent of @xmath2 and @xmath86 is given by ( [ eq : def_i ] ) .
the proof is in the appendix . the main challenge in proving lemma [ thm : secondary_rate_adhoc ] is the presence of the primary txs
. since the primary node density is smaller than the secondary node density , the primary cells are relatively further away from each other , thus requiring higher power to communicate .
although the relatively higher power could be a potential problem because the secondary nodes repeat their transmissions for three slots , the interfering primary transmission occurs at a certain minimum distance away from the secondary rx on one of these slots .
although the actual rate of the secondary network is reduced by a factor of three , this allows us to bound the interference of the more powerful primary nodes , without changing the scaling laws . from lemma [ thm : primary_rate_adhoc1 ]
, the value of @xmath78 , which is a normalized transmit power of the secondary txs , should be smaller than @xmath134 in order to satisfy ( [ eq : fraction_loss ] ) .
we also notice that the range of @xmath78 does not affect the throughput scalings of the secondary network .
let us define the secondary cells that border the preservation regions as _ loaded _ cells and the other cells as _ regular _ cells .
the loaded cells will be required to carry not only their own traffic , but also re - routed traffic around the preservation regions and , as a result , could deliver more data than the regular cells .
the next lemma bounds the number of data paths that each regular cell and each loaded cell must transport .
as the number of data paths each cell could carry was essentially the limiting factor in the sum throughput of the primary network , the following lemma is of crucial importance for the secondary sum throughput scaling law . [ thm : secondary_traffic_adhoc ] under the ad hoc primary model , each regular secondary cell needs to carry at most @xmath135 data paths and each loaded secondary cell carries at most @xmath136 data paths w.h.p . , where @xmath116 is given in corollary [ thm : clusters_of_preservation_regions ] .
the proof is in the appendix .
as it will be shown later , for @xmath137 the loaded cells are the bottleneck of the overall throughput . but even in this case , only a constant fraction of throughput degradation occurs , which does not affect the throughput scaling . for @xmath138 , since the secondary network is much denser than the primary network , the fraction of secondary data paths needing to be re - routed diminishes to zero as the node densities increase .
thus in the limit , almost all secondary cells behave as regular cells .
finally , we can use the previous corollary and lemmas to obtain the per - node and sum throughputs of the secondary network in the following theorem .
[ thm : secondary_throughput_adhoc ] suppose the ad hoc primary model . for any @xmath51 , by setting @xmath83 ,
the following per - node and sum throughputs are @xmath40-achievable w.h.p . for the secondary network : @xmath139 and @xmath140 where @xmath57 , which converges to zero as @xmath43 .
the definitions of @xmath96 , @xmath132 , and @xmath116 are given in lemma [ thm : primary_rate_adhoc1 ] , lemma [ thm : secondary_rate_adhoc ] , and corollary [ thm : clusters_of_preservation_regions ] , respectively .
note that by setting @xmath83 , the secondary network satisfies ( [ eq : fraction_loss ] ) during its transmission .
let us first consider @xmath141 .
let @xmath142 ( similarly , @xmath143 ) denote the number of secondary s - d pairs whose original or re - routed hdps ( vdps ) pass through loaded cells .
suppose the following two cases where the projections of two preservation regions on the @xmath144-axis are at a distance greater than @xmath145 ( fig .
[ fig : high_loaded ] . ( a ) ) and less than @xmath145 ( fig .
[ fig : high_loaded ] .
( b ) ) , respectively . for the first case ,
all extended hdps in the area of @xmath146 will pass through the loaded cells generated by two preservation regions .
but for the second case , the number of extended hdps passing through the loaded cells is less than the previous case w.h.p . because the corresponding area is smaller than @xmath146 .
thus , assuming that projections of all preservation regions on the @xmath144-axis are at a distance of at least @xmath145 from each other gives an upper bound on @xmath142 . in this worst - case scenario ,
all sources located in the area of @xmath147 generate extended hdps w.h.p .
, which must pass through the loaded cells , where we use the fact that the number of preservation regions is upper bounded by @xmath148 w.h.p .. by assuming that all nodes are sources , the resulting upper bound follows @xmath149 .
similarly , an upper bound on @xmath143 follows @xmath149 . if @xmath150 , we obtain @xmath151 if @xmath152 , from lemma [ thm : upper_bound_poission ] , we obtain @xmath153 then , @xmath154 hence , if @xmath141 , we obtain w.h.p .
@xmath155 where @xmath156 . in conclusion ,
the fraction of s - d pairs whose data paths pass through the loaded cells is upper bounded by @xmath157 w.h.p .
, which tends to zero as @xmath43 .
this indicates that almost all data paths will pass through regular cells rather than loaded cells .
if we treat the s - d pairs passing through the loaded cells and the s - d pairs not served as outages , @xmath40 is obviously upper bounded w.h.p . by @xmath158 where we use the fact that the fraction of s - d pairs not served is upper bounded by @xmath159 w.h.p .
( lemma [ thm : number_of_served_sd_adhoc ] ) .
then the achievable per - node throughput is determined by the rate of s - d pairs passing only the regular cells . since each secondary cell can sustain a constant rate of @xmath132 w.h.p .
( lemma [ thm : secondary_rate_adhoc ] ) , from the result of lemma [ thm : secondary_traffic_adhoc ] , each served secondary s - d pair that passes only through regular cells can achieve a rate of at least @xmath160 w.h.p .. therefore , @xmath39 is lower bounded by @xmath161 w.h.p .. let us now consider the case when @xmath137 . unlike the previous case ,
most served s - d pairs in this case pass through loaded cells , which will become bottlenecks . by assuming that all served s - d pairs
pass through loaded cells , we obtain a lower bound on @xmath39 with @xmath162 , which is an upper bound on the fraction of unserved s - d pairs . therefore , based on lemmas [ thm : secondary_rate_adhoc ] and
[ thm : secondary_traffic_adhoc ] , @xmath39 is lower bounded by @xmath163 w.h.p .. since there are at least @xmath164 non - outage s - d pairs , @xmath47 is lower bounded by @xmath165 w.h.p . , which completes the proof .
notice that if the secondary network knows when the primary nodes are activated in addition to their location , then @xmath166-tdma between the secondary cells in fig .
[ fig : alternative_adhoc ] can achieve the same scaling laws of theorem [ thm : secondary_throughput_adhoc ] .
specifically , each group of the secondary cells can be activated based on the @xmath63-tdma ( dotted region ) and within each group secondary cells operate @xmath63-tdma .
the indirect multihop routing in @xcite can also be adopted as a primary protocol , which provides the sum throughput of @xmath167 .
the key observation is that the construction of multihop data paths with a hop distance of @xmath168 is possible , which consists of the `` highway '' for multihop transmission . during phase 1
, each source directly transmits its packet to the closest node on the highway and , during phase 2 , the packet is delivered to the node on the highway closest to the destination by multihop transmissions using the nodes on the highway .
finally , during phase 3 , the destination directly receives the packet from the closet node on the highway .
let us assume that the transmit power of each primary tx scales according to the hop distance , that is each primary rx will receive the intended signal with a constant power . since the hop distance for phase 1 ( or 3 ) is given by @xmath169 , which is greater than @xmath170 achieved by the direct routing , the transmit power of phase 1 ( or 3 ) is greater than that of the direct routing .
the transmit power of phase 2 , on the other hand , is smaller than that of the direct routing because the hop distance is given by @xmath168 .
therefore , we can apply the previous secondary routing protocol during phase 2 of the primary indirect routing , which will cause less interference to the secondary network .
based on the analysis used for the direct routing , we derive the same results of theorems [ thm : primary_throughput_adhoc ] and [ thm : secondary_throughput_adhoc ] except now we have @xmath171 and @xmath172 .
in this section , we consider a different primary network which includes additional regularly - spaced bss . here
the primary nodes are again randomly distributed over a given area according to a p.p.p . of density @xmath0 .
in addition , the communication between the primary nodes is aided by the presence of @xmath1 bss , which may communicate at no cost in terms of scaling . in this infrastructure - supported primary model
, the secondary network continues to operate in an ad hoc fashion with nodes distributed according to a p.p.p . of density @xmath45 .
again we consider @xmath14 only .
we first outline the main results before describing the network protocols and analyzing the throughput and its asymptotic behavior for both the primary and secondary networks .
_ suppose the infrastructure - supported primary model with @xmath173 .
for any @xmath174 , the primary network can achieve the following per - node and sum throughputs w.h.p .
: @xmath175 where @xmath176 and @xmath177 .
the following per - node and sum throughputs are @xmath40-achievable w.h.p . for the secondary network : @xmath178 where @xmath179 , which converges to zero as @xmath43 .
_ compared to the throughput scalings of the ad hoc primary model , the addition of bss helps increase the scaling law of the primary network if @xmath173 , otherwise the scaling law stays unaffected @xcite .
we show here that the presence of a secondary network does not change the scaling law of this primary network for @xmath173 ( for @xmath180 , the results of the previous ad hoc primary model apply ) .
the secondary network can again achieve , with an arbitrarily small fraction of outage , the same scaling law under the direct multihop routing protocol as when the primary network is absent .
we assume the primary network uses a classical bs - based data transmission , in which sources deliver data to bss during the uplink phase and bss deliver received data to destinations during the downlink phase .
the challenge is again to prove that the secondary nodes can transmit in such a way that the primary scaling law should satisfy @xmath181 w.h.p .. we consider the primary protocol in which a source node transmits a packet to its closest bs and the destination node receives the packet from its closest bs , similar to those in @xcite and @xcite : * divide the unit area into square primary cells of area @xmath182 , where each primary cell has one bs at its center .
* during the uplink phase , each source node transmits a packet to the closest bs . * the bs that receives a packet from a source delivers it to the bs closest to the corresponding destination using bs - to - bs links . * during the downlink phase , each destination node receives its packet from the closest bs . *
a simple round - robin scheme is used for all downlink transmissions and all uplink transmissions in the same primary cell .
* at each transmission , a tx node transmits with power @xmath183 , where @xmath18 denotes the distance between the tx and its rx . under the given primary protocol ,
the sum throughput of @xmath177 is achievable , which coincides with the result of @xcite .
note that if @xmath184 , @xmath185 .
that is , when @xmath184 , using bss helps improve the throughput scaling of the primary network .
as was pointed out in @xcite , to improve throughput scaling , the number of bss should be high enough .
therefore , this primary protocol for the infrastructure - supported model is suitable for @xmath186 , while the result of the ad hoc primary model can be applied for @xmath187 .
let us consider the secondary protocol when the primary network is in the downlink phase .
since the secondary cell size is smaller than the primary cell size , the amount of interference from the secondary network to the primary network may be reduced by setting a preservation region around each primary receiving node .
however , the repeated transmissions of the same secondary packet does not guarantee a non - trivial rate for secondary transmissions since all bss are always active in the worst case for the infrastructure - supported case .
similar to the concept of preservation regions , in order to reduce the interference to the secondary nodes , in a certain region around each bs ( which are primary txs ) we insist that no secondary nodes transmit or receive in that region .
however , due to the relatively high transmit power of primary transmissions , these regions need a larger area than the previously defined preservation region .
define an _ avoidance region _ as a square containing @xmath188 secondary cells with a bs at the center , where @xmath189 is the size of the secondary cell that is the same as @xmath75 .
we also set the preservation regions around each bs consisted of @xmath190 secondary cells and around each primary node consisted of @xmath63 secondary cells .
we obtain a secondary protocol by replacing the three repeated transmissions of the previous secondary protocol by : * if a horizontal or vertical data path of each secondary s - d pair is blocked by an avoidance region , this data path is shifted horizontally ( or vertically ) to the non - blocked region . *
divide the entire time into two phases , where @xmath191 denotes the time fraction for phase @xmath120 . during phase @xmath120 ,
txs in the avoidance regions perform multihop transmissions using @xmath192 time fraction . during phase @xmath193 ,
txs outside the avoidance regions perform multihop transmissions using @xmath194 time fraction .
[ fig : secondary_routing_infra ] .
( a ) illustrates examples of shifted secondary data paths due to the avoidance regions ( for simplicity , preservation regions are not shown in this figure ) : @xmath195 illustrates the case where the hdp and vdp are not blocked , @xmath196 the case where only the hdp is blocked , @xmath197 the case where only the vdp is blocked , and @xmath198 the case where both the hdp and vdp are blocked .
[ fig : secondary_routing_infra ] .
( b ) illustrates the shifted hdp of the case @xmath196 . since the source is in the avoidance region ( but not in the preservation region )
, the multihop from the source to the first receiving node outside the avoidance region will be conducted during phase @xmath120 and the rest multihop to the destination will be conducted during phase @xmath193 .
since the area of each avoidance region is much larger than that of each preservation region , secondary cells adjacent to the avoidance regions should handle much more traffic than regular cells if we were to re - route blocked data paths using _ only _ these cells . in order to more evenly distribute the re - routed traffic , we shift an entire data path to the non - blocking region based on given mapping rule for the case when it is blocked by an avoidance region .
let us consider the details of finding a shifted secondary data path when it is blocked by an avoidance region .
define @xmath199 as the region in which extended hdps are not blocked by the avoidance regions .
this region is guaranteed to exist because of the regular bs placement , which is shown by the dotted regions in fig .
[ fig : secondary_routing_infra ] .
let us focus on the case @xmath196 , where the blocked hdp in @xmath200 is shifted to the new hdp in @xmath199 .
let @xmath201 and @xmath202 denote the @xmath144-axis of the blocked hdp and of its shifted hdp , respectively . without loss of generality , it is assumed that @xmath201 is in @xmath203 $ ] , where @xmath204 . then @xmath202 is given by @xmath205 where @xmath206 .
note that @xmath207 is half of the side length of an avoidance region , while @xmath208 is half of the length of the strips which are free of avoidance regions .
similarly , let @xmath209 denote the region in which none of vdps are blocked . we can shift a blocked vdp in @xmath210 to @xmath209 using the analogous mapping to the horizontal case .
if a hdp is shifted , it requires a series of short vertical hops to reach the shifted hdp , where we denote these vertical hops as a short vdp .
it also requires short horizontal hops to reach a destination after the vdp if that vdp is shifted , where we denote these horizontal hops as a short hdp .
let us consider the secondary protocol when the primary network is in the uplink phase .
we can also define an avoidance region at each tx ( primary node ) of the primary network . due to the irregular placement of primary nodes , however , it is hard to construct a re - routing protocol when each data path is blocked by an avoidance region .
more importantly , we can not set the area of each avoidance region as large as in the downlink case since the density of primary nodes is higher than that of bss , leading to a smaller throughput than the downlink case .
note that if we operate the secondary network during the uplink and downlink phases separately , then throughput scalings of the secondary network follow the maximum of the uplink and downlink throughputs .
therefore , overall throughput scalings follow those of the downlink phase . in this subsection , we analyze the per - node and sum throughputs of each network under given protocols and derive the corresponding scaling laws .
let us consider the per - node throughput of the primary network in the presence of the secondary network .
we first show that all primary cells may sustain a constant , non - trivial rate in lemma [ thm : primary_rate_infra ] .
we then determine the number of uplink and downlink transmissions each of these cells must support in lemma [ thm : primary_traffic_infra ] . using these results
, we obtain the primary per - node and sum throughputs in theorem [ thm : primary_throughput_infra ] .
let @xmath211 and @xmath212 denote the achievable aggregate rate of each primary cell in the presence and in the absence of the secondary network , respectively .
we define @xmath213 having a finite value for @xmath19 , which will be used to derive an upper bound on the interference power of the infrastructure - supported primary network . then the following lemma holds .
[ thm : primary_rate_infra ] suppose the infrastructure - supported model . if @xmath214 , then @xmath215 where @xmath216 and @xmath217 is given by ( [ eq : def_i ] ) .
moreover , @xmath212 is lower bounded by @xmath218 , where @xmath219 is a constant independent of @xmath1 .
the proof is in the appendix .
[ thm : primary_traffic_infra ] under the infrastructure - supported model , each primary cell needs to carry at most @xmath220 downlink and @xmath220 uplink transmissions w.h.p .. the proof is in the appendix .
[ thm : primary_throughput_infra ] suppose the infrastructure - supported model . for any @xmath51 , by setting @xmath214 , the primary network can achieve @xmath221 and @xmath222 w.h.p .
, where @xmath223 and @xmath224 the definitions of @xmath225 and @xmath218 are given in lemma [ thm : primary_rate_infra ] .
first consider the stand - alone throughput of the primary network .
let @xmath226 and @xmath227 denote the per - node throughput during downlink and uplink , respectively .
then @xmath228 , where @xmath229 arises from the fact that a source delivers a packet to its destination using one downlink and one uplink transmission .
since each primary cell can sustain a constant rate of @xmath218 ( lemma [ thm : primary_rate_infra ] ) , @xmath226 is upper bounded by @xmath218 divided by the maximum number of downlink transmissions in each primary cell .
this number of downlink transmissions is upper bounded by @xmath220 w.h.p .
( lemma [ thm : primary_traffic_infra ] ) .
therefore , @xmath226 is lower bounded by @xmath230 w.h.p .. since the same lower bound can be obtained for the case of @xmath227 , @xmath231 is lower bounded by @xmath232 w.h.p .. from the fact that there are at least @xmath98 primary s - d pairs ( lemma [ thm : node_number ] ) , @xmath49 is lower bounded by @xmath233 w.h.p .. the remaining proof about @xmath221 and @xmath222 w.h.p .
is the same as theorem [ thm : primary_throughput_adhoc ] , which completes the proof .
let us now consider the throughput scalings of the secondary network in the presence of the primary network .
we first show that the fraction of the unserved s - d pairs due to the preservation regions will be negligible w.h.p . in lemma [ thm : number_of_served_sd_infra ] .
unlike the ad hoc primary model , the overall multihop transmission of each s - d pair is divided into phases 1 and 2 depending on each tx s location . hence the per - node throughput scales as the minimum of the rate scalings related to phases 1 and 2 , respectively . we will show that although the aggregate rate of each secondary cell in the avoidance regions decreases as @xmath234 ( lemma [ thm : secondary_rate_infra ] ) , the number of data paths delivered by this cell is much less than that of each secondary cell outside the avoidance regions ( lemmas [ thm : secondary_traffic_infra ] and [ thm : secondary_traffic_infra2 ] ) .
thus the cells in the avoidance regions provide higher rate per each hop transmission than the cells outside the avoidance regions w.h.p . and , as a result
, @xmath39 and @xmath47 are determined by the transmissions outside the avoidance regions , which is phase @xmath193 .
[ thm : number_of_served_sd_infra ] under the infrastructure - supported primary model , the fraction of unserved secondary s - d pairs is upper bounded by @xmath235 w.h.p . , which converges to zero as @xmath58 .
the proof is in the appendix .
[ thm : secondary_rate_infra ] under the infrastructure - supported primary model , each secondary cell in the avoidance regions and each secondary cell outside the avoidance regions can sustain a rate of @xmath236 and @xmath237 respectively , where @xmath238 , which tends to zero as @xmath43 , and @xmath239 is a constant independent of @xmath2 .
the definitions of @xmath86 and @xmath217 are given by ( [ eq : def_i ] ) and ( [ eq : def_i ] ) , respectively .
the proof is in the appendix .
as in the ad hoc primary model , we define the secondary cells which border the preservation regions as the _ loaded _ cells and the other cells as _ regular _ cells . then , the following lemmas hold .
[ thm : secondary_traffic_infra ] suppose the infrastructure - supported primary model . each regular secondary cell and each loaded secondary cell
outside the avoidance regions need to carry at most @xmath240 and @xmath241 data paths w.h.p .
, respectively , where @xmath116 is given in corollary [ thm : clusters_of_preservation_regions ] .
the proof is in the appendix .
[ thm : secondary_traffic_infra2 ] suppose the infrastructure - supported primary model . each regular secondary cell and each loaded secondary cell in the avoidance regions need to carry at most @xmath242 and @xmath243 data paths w.h.p .
, respectively , where @xmath116 is given in corollary [ thm : clusters_of_preservation_regions ] .
the proof is in the appendix .
we can now use the previous corollaries and lemmas to obtain the per - node and sum throughputs of the secondary network in the following theorem .
[ thm : secondary_throughput_infra ] suppose the infrastructure - supported primary model
. for any @xmath51 , by setting @xmath78 within @xmath244 , the following per - node and sum throughputs are @xmath40-achievable for the secondary network w.h.p . :
@xmath245 and @xmath246 where @xmath247 , which converges to zero as @xmath43 .
the definitions of @xmath225 , @xmath237 , and @xmath116 are given in lemma [ thm : primary_rate_infra ] , lemma [ thm : secondary_rate_infra ] , and corollary [ thm : clusters_of_preservation_regions ] , respectively .
note that by setting @xmath214 , the secondary network satisfies ( [ eq : fraction_loss ] ) during its transmission .
let us first consider @xmath141 .
let @xmath248 ( similarly , @xmath249 ) denote the number of secondary s - d pairs whose original , including shifted one , or re - routed hdps are in @xmath199 ( @xmath200 ) and pass through loaded cells .
similarly , we can define @xmath250 and @xmath251 for extended vdps . to obtain an upper bound on @xmath248 ,
we consider extended hdps , which is the same as lemma [ thm : secondary_traffic_infra ] , and study the geometric scenario that requires re - routing the largest number of data paths to the loaded cells .
this worst - case scenario is obtained when the projections of all preservation regions on the @xmath144-axis are separated at a distance of at least @xmath252 and all preservation regions are in the avoidance - region free zone @xmath199 .
thus , all nodes located in the area of @xmath253 pass through loaded cells , where @xmath254 arises from the shifted hdps along with the original hdps
. therefore , an upper bound on @xmath248 follows @xmath255 .
similarly , an upper bound on @xmath249 follows @xmath256 , where we assume that all preservation regions are in @xmath200 for this case .
the vertical worst - case scenario may be similarly derived . using the same analysis from ( [ eq : prob_zero ] ) to ( [ eq : upper_m_ah_av ] ) , we obtain w.h.p .
@xmath257 where @xmath258 .
if we treat the s - d pairs passing through the loaded cell and the s - d pairs not served as outage , @xmath259 w.h.p .
, where we use the result of lemma [ thm : number_of_served_sd_infra ]
. then the achievable per - node throughput is determined by the rate of s - d pairs passing through only the regular cells .
let us consider the regular cells in the avoidance regions , which perform transmissions during phase @xmath120 .
for this case , since each cell sustains a rate of @xmath236 w.h.p .
( lemma [ thm : secondary_rate_infra ] ) , and based on lemma [ thm : secondary_traffic_infra2 ] , the rate per each hop transmission provided by these cells is lower bounded by @xmath260 w.h.p .. if we consider the regular cells outside the avoidance regions , from lemmas [ thm : secondary_rate_infra ] and [ thm : secondary_traffic_infra ] , the rate per each hop transmission is lower bounded by @xmath261 w.h.p .. since , for sufficiently large @xmath2 , the rate provided by the cells in the avoidance regions is greater than that provided by the cells outside the avoidance regions , @xmath39 is lower bounded by @xmath262 w.h.p . if @xmath141 .
let us now consider @xmath137 .
again , we obtain a lower bound on @xmath39 by considering the most heavily loaded scenario in which all served s - d pairs pass through loaded cells . then @xmath263 .
similarly , we can derive the rate per each hop transmission related to phases @xmath120 and @xmath193 from the results in lemmas [ thm : secondary_rate_infra ] to [ thm : secondary_traffic_infra2 ] . as a result , @xmath39 is lower bounded by @xmath264 w.h.p . if @xmath137 .
finally @xmath47 is lower bounded by @xmath165 w.h.p . , which completes the proof .
in this paper , we studied two co - existing ad hoc networks with different priorities ( a primary and a secondary network ) and analyzed their simultaneous throughput scalings .
it was shown that each network can achieve the same throughput scaling as when the other network is absent .
although we allow outage for the secondary s - d pairs , the fraction of pairs in outage converges to zero as node densities increase .
furthermore , these scalings may be achieved by adjusting the secondary protocol while keeping that of the primary network unchanged .
in essence , the primary network is unaware of the presence of the secondary network . to achieve this result
, the secondary nodes need knowledge of the locations of the primary nodes , and the secondary nodes need to be denser than the primary . for @xmath265 ( primary is denser than the secondary network ) , on the other hand , it seems to be more challenging to achieve similar throughput scaling results while keeping the primary unchanged , as there are many primary nodes around each secondary node .
as mentioned before , if we allow the primary protocol to adapt to the presence of the secondary network , we can achieve throughput scalings of two homogenous networks by employing tdma between two networks .
our result may be extended to more than two networks , provided each layered network obeys the same three main assumptions as in the two network case .
let @xmath267 denote the number of primary nodes in a unit area . for part ( a )
, we wish to show that @xmath268 as @xmath269 .
noting that @xmath267 is a poisson random variable with mean @xmath0 and standard deviation @xmath270 , we use chebyshev s inequality to see that @xmath271 clearly , as @xmath0 tends to infinity we can make this quantity arbitrarily small .
for part ( b ) , let @xmath272 denote the number of primary nodes in a primary cell .
then @xmath273 is given by @xmath274 therefore , the probability that there is at least one cell having no node is upper bounded by @xmath275 , where the union bound and the fact that there are at most @xmath0 primary cells are used . since @xmath276 as @xmath121 , ( b ) holds w.h.p . , which completes the proof .
suppose that at a given moment , there are @xmath277 active primary cells and @xmath278 active secondary cells , including the @xmath24-th active primary cell .
then , the rate of the @xmath24-th active primary cell is given by @xmath279 where @xmath280 indicates the loss in rate due to the @xmath63-tdma transmission of primary cells .
the rate of the @xmath24-th active primary cell in the absence of the secondary network is given by @xmath281 by setting @xmath282 .
[ fig : i_p_adhoc ] illustrates the worst case interference from the secondary interferers to the rx of the @xmath24-th active primary cell , where the dotted region denotes the preservation region around the primary rx and the shaded cells denote the active secondary cells based on the @xmath63-tdma .
because of the preservation region , the minimum distance of @xmath89 can be guaranteed from all secondary transmitting interferers to the primary rx .
thus , there exist @xmath283 secondary interferers at a distance of at least @xmath89 , and @xmath284 secondary interferers at a distance of at least @xmath285 , and so on .
then , @xmath286 is upper bounded by @xmath287 where we use the fact that @xmath288 .
then @xmath289 notice that @xmath290 is the value of @xmath78 such that the right - hand side of ( [ eq : fraction_lower ] ) is equal to @xmath60 .
thus , if we set @xmath83 , then @xmath291 . because the above inequality holds for any @xmath24 , we obtain @xmath292 .
similarly , there exist @xmath283 primary interferers at a distance of at least @xmath293 , and @xmath284 primary interferers at a distance of at least @xmath294 , and so on .
then @xmath295 where we use the fact that @xmath296 .
thus , @xmath297 therefore , lemma [ thm : primary_rate_adhoc1 ] holds .
let @xmath298 denote the number of extended hdps that should be delivered by a primary cell .
similarly , @xmath299 denotes the number of extended vdps that should be delivered by a primary cell .
when hdps are extended , the extended hdps of all primary sources located in the area of @xmath300 should be handled by the primary cell . by assuming that all primary nodes are sources , the resulting upper bound on @xmath298 follows @xmath301 . using lemma [ thm : upper_bound_poission ] , we obtain @xmath302 similarly , the extended hdps of all primary destinations located in the area of @xmath303 should be also handled by the primary cell . by assuming that all primary nodes are destinations , we obtain @xmath304 from ( [ eq : upper_n_ph ] ) and ( [ eq : upper_n_pv ] ) , we obtain @xmath305 where the last inequality comes from the union bound . therefore , the probability that there is at least one primary cell supporting more than @xmath306 extended data paths is upper bounded by @xmath307 , where the union bound and the fact that there are at most @xmath0 primary cells are used . since @xmath308 as @xmath44 , each primary cell should deliver the corresponding data of at most @xmath306 extended data paths w.h.p . , where @xmath68 .
note that the above bounds also hold for the original data paths , which completes the proof .
let @xmath309 denote the area of all preservation regions , @xmath310 denote the area of all disjoint regions due to the preservation regions except the biggest region , and @xmath311 .
define @xmath312 as the number of secondary nodes in the area of @xmath313 that follows @xmath314 .
the number of secondary s - d pairs not served is clearly upper bounded by @xmath312 . from lemma [ thm : upper_bound_poission ] , we obtain @xmath315 an upper bound on @xmath309 is obtained if we assume none of the regions overlap .
thus , as each preservation region has an area of @xmath316 and there are at most @xmath148 such regions w.h.p .
, we obtain w.h.p .
@xmath317 to derive an upper bound on @xmath310 , we assume all preservation regions form clusters having @xmath116 preservation region each ( corollary [ thm : clusters_of_preservation_regions ] ) shown in fig .
[ fig : fraction ] .
( a ) , where the shaded regions denote @xmath310 .
then the maximum disjoint area generated by a cluster of @xmath116 preservation regions is given in fig .
[ fig : fraction ] .
( b ) as a circle maximizes the area of a region for a given perimeter . because each preservation region contributes a length of at most @xmath318 to the circumference of this circle , the radius is upper bounded by @xmath319 .
thus , @xmath310 is upper bounded w.h.p .
by @xmath320 where we use the fact that the total number of clusters having @xmath116 preservation regions in each cluster is upper bounded by @xmath321 w.h.p .. from ( [ eq : upper_a_p_1 ] ) and ( [ eq : upper_a_p_2 ] ) , @xmath313 is upper bounded by @xmath322 w.h.p .. by substituting @xmath313 for its upper bound in ( @xmath323 ) , we obtain @xmath324 thus , we obtain w.h.p .
@xmath325 where @xmath326 .
since the total number of secondary s - d pairs is lower bounded by @xmath327 w.h.p . , the fraction of unserved s - d pairs is upper bounded by @xmath159 w.h.p . , which completes the proof .
since the same secondary packet is transmitted three times , the minimum distance of @xmath328 from all primary interferers to the secondary rx can be guaranteed for one out of three transmissions .
then the interference from primary interferers of that packet is upper bounded by @xmath329 where we use the same technique as in lemma [ thm : primary_rate_adhoc1 ] .
similarly , @xmath330 is lower bounded by @xmath331 .
thus , the rate of each secondary cell is lower bounded by @xmath332 where @xmath333 indicates the rate loss due to the @xmath63-tdma and repeated ( three times ) transmissions of the same secondary packet .
therefore , lemma [ thm : secondary_rate_adhoc ] holds .
let @xmath334 and @xmath335 denote the number of extended hdps including re - routed paths that should be delivered by a secondary regular cell and by a secondary loaded cell , respectively .
similarly , we can define @xmath336 and @xmath337 for extended vdps .
let us first consider a regular cell .
this regular cell delivers the corresponding data of extended hdps passing through it .
then all extended hdps of the secondary sources located in the area of @xmath338 should be handled by the regular cell , where we ignore the effect of s - d pairs not served , which yields an upper bound on the total number of hdps . by assuming that all secondary nodes are sources , the resulting upper bound on @xmath334 follows @xmath339 . from lemma
[ thm : upper_bound_poission ] , we obtain @xmath340 we obtain the same bound for @xmath336 by assuming that all secondary nodes are destinations and then @xmath341 from the union bound and the fact that there are at most @xmath2 secondary cells , each regular cell should deliver the corresponding data of at most @xmath342 extended data paths w.h.p . , where we use the fact that @xmath343 as @xmath344 .
let us now consider a loaded cell . unlike in the primary data path which has no obstacles , a secondary data path should circumvent any preservation regions which lie on its path .
therefore , the loaded cells should deliver more data paths than the regular cells w.h.p .. suppose a cluster of preservation regions located on the boundary of the network in fig .
[ fig : cluster_preservation_region ] , whose projection on @xmath144-axis has a length of @xmath345 .
then all extended hdps of the secondary sources located in the area of @xmath346 is re - routed through the dotted cells , where we ignore the effect of s - d pairs not served ( which yields an upper bound on the total number of extended hdps ) .
the other loaded cells will deliver less hdps than the dotted cells w.h.p .. recall that @xmath347 w.h.p .
( corollary [ thm : clusters_of_preservation_regions ] ) and the dotted cells need to deliver re - routing paths of at most two such clusters .
therefore , by assuming that all secondary nodes are sources , the resulting upper bound on @xmath335 follows @xmath348 .
note that the upper bound on @xmath335 is the same as the upper bound on @xmath334 except for a constant factor of @xmath349 , where @xmath350 comes from the re - routed hdps of two adjacent clusters and @xmath120 comes from the original hdps .
therefore , we can apply the same analysis used in the regular case . in conclusion
, each loaded cell should deliver the corresponding data of at most @xmath351 extended data paths w.h.p .. since the above bounds also hold for the original data paths , lemma [ thm : secondary_traffic_adhoc ] holds .
the overall procedure of the proof is similar to that of lemma [ thm : primary_rate_adhoc1 ] .
let us first consider downlink transmissions , where all primary cells are activated simultaneously at a given moment .
let @xmath352 and @xmath353 denote the interference from all primary interferers and all secondary interferers during downlink , respectively .
let @xmath354 and @xmath355 denote the downlink rates of a primary cell in the presence of the secondary network and in the absence of the secondary network , respectively .
then @xmath356 if @xmath357 . from the same bounds in ( [ eq
: upper_i_sp ] ) and ( [ eq : fraction_lower ] ) , we obtain @xmath358 for @xmath359 .
the same bound can be derived for the uplink .
thus , ( [ eq : r_p_r_alone ] ) holds .
now consider the bound on @xmath352 .
since there exist @xmath283 primary interferers at a distance of at least @xmath360 and @xmath284 primary interferers at a distance of at least @xmath361 and so on ( see fig .
[ fig : i_p_infra ] ) , we obtain @xmath362 where we use the fact that the transmit power of each bs is upper bounded by @xmath363 .
then @xmath364 in a similar manner , the rate of each primary cell during uplink is also lower bounded by @xmath218 .
therefore , we can guarantee a constant rate of @xmath218 for each primary cell during both downlink and uplink , which completes the proof .
let @xmath365 denote the number of primary nodes in a primary cell , which follows @xmath366 . from lemma [ thm : upper_bound_poission ]
, we obtain @xmath367 from the union bound , each primary cell has at most @xmath368 primary nodes w.h.p . , where we use the fact that @xmath369 as @xmath44 .
if we assume that all primary nodes are destinations ( or sources ) , the number of downlink transmissions ( or the number of uplink transmissions ) per primary cell is upper bounded by @xmath370 w.h.p .. therefore , the lemma holds . let @xmath371 denote the area of all preservation regions around bss and @xmath372 denote the number of secondary nodes in the area of @xmath371 .
then , from lemma [ thm : upper_bound_poission ] , @xmath373 since each preservation region around bs has an area of @xmath374 and there are @xmath1 such regions , which are not overlapping with each other , @xmath375 .
thus , we know @xmath376 w.h.p . ,
where @xmath377 combining this with the result of lemma [ thm : number_of_served_sd_adhoc ] , we obtain @xmath378 w.h.p .. since the number of s - d pairs not served is clearly upper bounded by @xmath379 , the fraction of unserved s - d pairs is upper bounded by @xmath380 w.h.p . , which completes the proof .
first consider the rate of a secondary cell in the avoidance regions ( but not in the preservation regions ) . due to the preservation regions around bss
, the minimum distance of @xmath381 can be guaranteed from all primary interferers .
thus , @xmath382 . similarly @xmath383 .
then the rate of each secondary cell in the avoidance regions is upper bounded by @xmath384 where @xmath385 arises from @xmath63-tdma , the time fraction of phase 1 , and the time fraction of downlink . in the case of a secondary cell outside the avoidance regions ,
the minimum distance of @xmath386 can be guaranteed from all primary interferers
. then the rate of each secondary cell outside the avoidance regions is upper bounded by @xmath387 where @xmath388 arises from @xmath63-tdma , the time fraction of phase 2 , and the time fraction of downlink .
therefore , lemma [ thm : secondary_rate_infra ] holds .
consider phase @xmath193 in which the secondary cells outside the avoidance regions are activated .
let @xmath389 and @xmath390 denote the number of extended hdps that should be delivered by a secondary regular cell and by a secondary loaded cell , respectively .
we can define @xmath391 and @xmath392 analogously for vdps .
let us first consider a regular cell in @xmath393 .
there are two types of hdps in @xmath199 : the first type is an original ( or a shifted ) hdp and the second type is a short horizontal hops in order to reach each destination . note that a short hdp only occurs if its original vdp is blocked by an avoidance region .
we assume that a short hdp always occurs regardless of its vdp and extend it to the entire horizontal line including the short hdp .
[ fig : routing_analysis_infra ] illustrates examples of original ( or shifted ) hdps ( left ) and their extended hdps ( right ) in @xmath199 .
note that the @xmath144-axis of an extended hdp from an original ( or shifted ) hdp originates from a source node .
similarly , the @xmath144-axis of an extended hdp from a short hdp originates from a destination node . as a result , under this extended traffic , all secondary nodes generate extended hdps on @xmath199 because each node is a source or a destination , where we ignore the effects of the s - d pairs not served and the s - d pairs that do not generate traffic on @xmath199 .
since a regular cell in @xmath199 delivers the corresponding data of all extended hdps passing through it , all extended hdps of the secondary nodes located in the area of @xmath394 should be delivered by the regular cell .
additionally , it should deliver the corresponding data of all nodes in the area of @xmath395 because these extended hdps are shifted to @xmath199 .
therefore , the resulting upper bound on @xmath389 follows @xmath396 , where @xmath254 . from lemma
[ thm : upper_bound_poission ] , we obtain @xmath397 the same bound can be obtained for @xmath391 . from the fact that the number of data paths that should be delivered by a regular cell in @xmath393 is given by @xmath398 , we obtain @xmath399 by the union bound and the fact that there are at most @xmath2 secondary cells , each regular cell in @xmath393 should deliver at most @xmath400 extended data paths w.h.p . , where we use the fact @xmath401 as @xmath44 .
unlike the previous case , all s - d pairs that generate hdps in @xmath200 are not vertically blocked such that only original hdps exist in @xmath200 .
then , @xmath389 is upper bounded by @xmath402 w.h.p . in this case .
therefore the regular cells in @xmath403 , @xmath404 , and @xmath405 deliver w.h.p .
less data paths compared to the regular cells in @xmath393 . in conclusion , each regular cell should deliver the corresponding data of at most @xmath406 extended data paths w.h.p .. to obtain an upper bound on @xmath390 , consider again the cluster of the preservation regions located on the boundary of the network in fig .
[ fig : cluster_preservation_region ] ( or the boundary of an avoidance region in this case ) . then all nodes located in the area of @xmath407 generate extended hdps passing through the dotted cells in @xmath199 .
additionally , all nodes located in the area of @xmath408 , belonging to @xmath200 , generate extended hdps passing through the dotted cells since they are shifted to @xmath199 .
thus , from the fact @xmath409 w.h.p . , @xmath410 w.h.p .. by applying the same bound on @xmath392
, we conclude that each loaded cell should deliver the corresponding data of at most @xmath411 data paths w.h.p .. note that the loaded cells in @xmath403 , @xmath404 , and @xmath405 deliver w.h.p .
less data paths than the loaded cells in @xmath393 .
thus , lemma [ thm : secondary_traffic_infra ] holds .
consider phase @xmath120 in which the secondary cells in the avoidance regions are activated .
since the avoidance regions are in @xmath412 , there exists no shifted data path .
the overall procedure is similar to the proof of lemma [ thm : secondary_traffic_infra ] .
let us first consider the secondary regular cells .
if we extend hdp to the line having the length of @xmath386 , which is the length of half an avoidance region side , all nodes in the area of @xmath413 generate extended hdps passing through a regular cell .
thus , the number of extended hdps delivered by each regular cell is upper bounded by @xmath414 w.h.p .. by the same analysis for vdp , each regular cell should deliver the corresponding data of at most @xmath415 extended data paths w.h.p .. similarly , each secondary loaded cell should deliver the corresponding data of at most @xmath416 extended data paths w.h.p .
, which completes the proof .
xie and p. r. kumar , `` on the path - loss attenuation regime for positive cost and linear scaling of transport capacity in wireless networks , '' _ ieee trans .
inf . theory _ ,
23132328 , june 2006 . | we study two distinct , but overlapping , networks that operate at the same time , space , and frequency .
the first network consists of @xmath0 randomly distributed _ primary users _ , which form either an ad hoc network , or an infrastructure - supported ad hoc network with @xmath1 additional base stations .
the second network consists of @xmath2 randomly distributed , ad hoc _ secondary users _ or _ cognitive users_. the primary users have priority access to the spectrum and do not need to change their communication protocol in the presence of secondary users .
the secondary users , however , need to adjust their protocol based on knowledge about the locations of the primary nodes to bring little loss to the primary network s throughput . by introducing preservation regions around primary receivers and avoidance regions around primary base stations ,
we propose two modified multihop routing protocols for the cognitive users .
base on percolation theory , we show that when the secondary network is denser than the primary network , _ both _ networks can _ simultaneously _ achieve the same throughput scaling law as a stand - alone network .
furthermore , the primary network throughput is subject to only a vanishingly fractional loss .
specifically , for the ad hoc and the infrastructure - supported primary models , the primary network achieves sum throughputs of order @xmath3 and @xmath4 , respectively . for both primary network models , for any @xmath5
, the secondary network can achieve sum throughput of order @xmath6 with an arbitrarily small fraction of outage .
thus , almost all secondary source - destination pairs can communicate at a rate of order @xmath7 .
cognitive radio , scaling law , heterogeneous networks , interference management , routing algorithm |
the pyrochlores with the chemical formula @xmath0mo@xmath1o@xmath2 ( @xmath3 being the divalent element ) exhibit a number of interesting and not completely understood phenomena,@xcite which present the main motivation for the analysis of electronic and magnetic structure of these compounds , which we undertake in this work . @xmath11 the pyrochlores @xmath0mo@xmath1o@xmath2 have rather puzzling magnetic phase diagram as a function of averaged ionic radius of @xmath3-sites , @xmath12 ( fig .
[ fig.tcexp]).@xcite large @xmath12 ( @xmath13@xmath14@xmath8@xmath15 ) stabilizes the ferromagnetic ( fm ) ground state .
typical examples of the fm pyrochlores are nd@xmath1mo@xmath1o@xmath2 and gd@xmath1mo@xmath1o@xmath2 .
the curie temperature ( @xmath16 ) is of the order of 80 k and slowly increases with @xmath12 .
smaller @xmath12 ( @xmath17@xmath14 ) give rise to the spin - glass ( sg ) behavior .
the characteristic transition temperature into the sg state is of the order of 20 k. the canonical example of the sg compounds is y@xmath1mo@xmath1o@xmath2.@xcite now it is commonly accepted that the exchange coupling between nearest - neighbor ( _ nn _ ) mo spins changes the sign around @xmath14 and becomes antiferromagnetic ( afm ) in the sg region .
the pyrochlores with afm interactions in the lattice present a typical example of geometrically frustrated systems with infinitely degenerate magnetic ground state.@xcite according to the recent experimental data,@xcite the sg behavior itself is caused by the combination of the geometrical frustrations and the disorder of the local lattice distortions .
the latter is required in order to produce an inhomogeneity in the distribution of the _ nn _ interactions and freeze the random spin configuration .
the origin of this behavior itself is very complicated , and we will not be able to address it properly in our work
. however , our main concern will be in some sense more general : we will try to understand which parameter of the crystal structure controls the sign of the _ nn _ magnetic interactions , and which part of the electronic structure is responsible for the fm and afm interactions in these compounds ?
contrary to the widespread point of view that the magnetic ground state is controlled by the mo - o - mo angle , we will argue that the key parameter is in fact the mo - mo distance , which is directly related with the unit cell volume and controls the strength of direct interactions between extended mo(@xmath6 ) orbitals . according to the experimental data , all mo pyrochlores which show the sg behavior are small - gap insulators .
however , an opposite statement is generally incorrect and in different compounds the ferromagnetism is known to coexist with the metallic as well as insulating behavior.@xcite we will show that this trend can be naturally understood in terms of the mo - mo distances and the strength of the on - site coulomb interaction @xmath7 amongst the mo(@xmath6 ) electrons . @xmath11 the spin - orbit interaction ( soi ) in the pyrochlore lattice should generally lead to a non - collinear magnetic ordering associated with the modulation of single - ion anisotropy axes.@xcite the magnitude and the main source of this non - collinearity ( whether it is primarily associated with the mo or @xmath3 sublattices , and whether it reflects the ordering of spin or orbital magnetic moments ) is largely unresolved problem . in the present work we will try to rationalize this question by focusing on the magnetic structure of the mo sublattice .
we will show that as long as the system is metallic ( that by itself depends on the strength of the coulomb interaction @xmath7 ) , the non - collinearity is very small . however , the transition into the insulating state is accompanied by the abrupt change of the spin and orbital magnetic structures , which become essentially non - collinear . @xmath11 some ferromagnetic ( or nearly ferromagnetic ) pyrochlores exhibit rather unusual anomalous hall effect ( ahe ) .
a non - vanishing hall conductivity ( @xmath18 ) emerges near @xmath16 , then monotonously increases with the decrease of the temperature @xmath19 , and reaches the maximal value ( @xmath18@xmath8@xmath20 @xmath21 for nd@xmath1mo@xmath1o@xmath2 ) near @xmath19@xmath4@xmath22 .
both magnitude and the temperature dependence of the ahe is in a drastic contrast with other oxide materials , such as colossal - magnetoresistive manganites , for example.@xcite the theory of the ahe ( originally called as the extraordinary hall effect ) has very long and rather controversial history.@xcite many controversies were related with the lack of the general transport theory and partly with the lack of good computer facilities in 1950s and 1960s , so that many discussions went around very simplified and not always justified approximations , such as the free - electron and weak scattering limit , the single - orbital tight - binding approach , the classical boltzmann transport equation , etc.@xcite the modern way to address the problem is to use the kubo formalism and relate @xmath18 with the real part of the antisymmetric off - diagonal element of the conductivity tensor.@xcite from this point of view there are no rigid constraints which would forbid the existence of @xmath18 , even for periodic systems , once the time - reversal symmetry is violated macroscopically.@xcite the main question is which mechanism leads to the finite value of @xmath18 .
the first one was proposed by karplus and luttinger,@xcite who stressed the importance of soi in the problem .
although the concrete scenario considered by karplus and luttinger was not correct,@xcite they were certainly right by arguing that the soi is essential in order to couple the spin polarization of the conduction electrons to the lattice and produce the right - left asymmetry of the stationary bloch states , which may lead to the hall current , perpendicular to both the electric field and the magnetization . however , an exceptionally large @xmath18 observed in the pyrochlore compounds urged several authors to search for a new and unconventional mechanism for the ahe .
such mechanism was proposed to be due to the spin chirality ( basically , the solid angle subtended by three non - coplanar spins ) in essentially non - collinear magnetic structure.@xcite the chirality acts as an effective field , which in the combination with the peculiar lattice geometry is believed to yield a finite @xmath18 . although the mechanism itself is rather exotic and there is a great deal of cancellations between different elements of the non - collinear spin structure,@xcite these authors were certainly right by stressing the crucial role of the inter - band transitions in the problem of ahe .
in this work we will go back to a more conventional picture and argue that the large value of @xmath18 in some mo pyrochlores can be explained by rather unique combination of three factors : ( i ) large soi associated with the mo(@xmath6 ) states in the nearly collinear fm structure ; ( ii ) strong enhancement of the soi by the on - site coulomb interactions in the @xmath5 electron systems,@xcite and ( iii ) half - metallic electronic structure of the fm pyrochlores.@xcite regarding the coulomb interaction itself , which in our work will be treated as a parameter , we will be able to present a very consistent picture by arguing that @xmath7@xmath8@xmath9-@xmath10 ev in mo pyrochlores accounts simultaneously for ( i ) metal - insulator transition ; ( ii ) fm - sg transition ; ( iii ) required enhancement for the ahe in the fm pyrochlore compounds .
the rest of the paper is organized as follows . in sec .
[ sec : main_details ] we remind the basic details of the crystal and electronic structure of @xmath0mo@xmath1o@xmath2 .
we consider three characteristic compounds corresponding to @xmath3@xmath4 y , gd and nd . in sec .
[ sec : model ] we describe the model hartree - fock ( hf ) approach , which combines the ideas of _ ab - initio _ electronic structure calculations with the physics of degenerate hubbard model on the pyrochlore lattice .
[ sec : dos ] and [ sec : exchange ] summarize results of calculations without soi ( correspondingly , the basic changes of the electronic structure and the orbital ordering , and the behavior of inter - atomic magnetic interactions ) . secs .
[ sec : nc ] and [ sec : hall ] are devoted to the analysis of soi related properties ( the non - collinear magnetic structure , the hall conductivity , and the magneto - optical effect ) . finally , in sec .
[ sec : summary ] we discuss some subtle points related with the interpretation of experimental data and give the summary of the entire work .
the pyrochlores @xmath0mo@xmath1o@xmath2 crystallize in a face - centered cubic structure with the space group @xmath23 , in which @xmath3 and mo occupy correspondingly @xmath24 and @xmath25 positions , and form interpenetrating sublattices of the corner - sharing tetrahedra .
there are two non - equivalent types of oxygen sites , and only one internal parameter which may control the properties of @xmath0mo@xmath1o@xmath2 .
that is the coordinate @xmath26 of the o @xmath27 sites .
the single mo tetrahedron is shown in fig .
[ fig.tetrahedron ] .
four mo sites are located at @xmath28@xmath4@xmath29 , @xmath30@xmath4@xmath31 , @xmath32@xmath4@xmath33 , and @xmath34@xmath4@xmath35 , in units of cubic lattice parameter @xmath36 .
each mo site has sixfold o @xmath27 coordination .
the oxygen atoms specify the local coordinate frame around each mo site .
around the site 1 , it is given by @xmath37 where @xmath38@xmath4 @xmath39 , @xmath40 , or @xmath41 .
@xmath26@xmath4@xmath42 corresponds to the perfect octahedral environment . in this case
@xmath43@xmath4@xmath44 is the standard @xmath45 rotation around the cubic @xmath46 $ ] axis .
@xmath26@xmath13@xmath42 gives rise to an additional trigonal contraction of the local coordinate frame .
similar matrices associated with the sites 2 , 3 , and 4 can be obtained by the @xmath47 rotations of @xmath43 around @xmath39 , @xmath40 , and @xmath41 , respectively .
structural parameters of @xmath0mo@xmath1o@xmath2 are listed in table [ tab : structure ] .
corresponding densities of states , obtained in the local - spin - density approximation ( lsda ) are shown in figs .
[ fig.doslsda ] and [ fig.dost2g ] . in the local coordinate frame , the mo(@xmath6 ) orbitals
are split into the triply - degenerate @xmath5 and doubly - degenerate @xmath48 states .
the splitting is of the order of 4 ev .
twelve @xmath5 bands are located near the fermi level and well separated from the rest of the spectrum , which consists of the broad o(@xmath49 ) band spreading from -8.5 to -2.5 ev and either y(@xmath6 ) or gd / nd(@xmath50 ) bands located just above the group of the @xmath5 states . the mo(@xmath48 ) states are located in the higher part of the spectrum .
interestingly enough that all three compounds are ferromagnetic , even in the lsda , that is rather unusual for the @xmath6 oxides , perhaps except the well know example of srruo@xmath51.@xcite the trigonal distortion and different hybridization with the o(@xmath49 ) states will further split the mo(@xmath5 ) states into the one - dimensional @xmath52 and two - dimensional @xmath53 representation.@xcite the crystal structure affects the mo(@xmath5 ) bandwidth via two mechanisms ( see table [ tab : structure ] ) . 1 . the mo - o - mo angle , which decreases in the direction nd@xmath54gd@xmath54y .
therefore , the interactions between mo(@xmath5 ) orbitals which are mediated by the o(@xmath49 ) states will also decrease .
this effect is partly compensated by the decrease of the mo - o bondlength for @xmath3@xmath4 gd and y. 2 .
the lattice parameter @xmath36 and the mo - mo distance , which decrease in the direction nd@xmath54gd@xmath54y by 2.6% .
this will increase direct interactions between extended mo(@xmath6 ) orbitals .
generally , these two effects act in the opposite directions and partly compensate each other .
for example , the width of the @xmath53 band is practically the same for all three compounds ( fig .
[ fig.dost2g ] ) . on the other hand , the @xmath52 orbitals , whose lobes are the most distant from all neighboring oxygen sites are mainly affected by the second mechanism , and the @xmath52 bandwidth will _ increase _ in the direction nd@xmath54gd@xmath54y . as we will show in sec .
[ sec : exchange ] , this effect plays a crucial role in the stabilization of afm interactions in y@xmath1mo@xmath1o@xmath2 , which preludes the sg behavior .
thus , despite an apparent complexity of the crystal structure , the pyrochlores @xmath0mo@xmath1o@xmath2 present rather simple example of the electronic structure and in order to understand the nature of fascinating electronic and magnetic properties of these compounds , we need to concentrate on the behavior of twelve well isolated mo(@xmath5 ) bands.@xcite
in this and subsequent sections we will further elaborate this picture , by cutting the mo(@xmath5 ) bands from the rest of the spectrum , including the on - site coulomb interactions amongst the mo(@xmath6 ) electrons , and solving this problem on the level of hf approximation .
hence , our model hamiltonian has the following form , in the basis of non - relativistic lda eigenfunctions @xmath55 ( lda is the spin - restricted version of lsda ) : @xmath56 the spin indices are already included to @xmath57 and @xmath58 , which are running from @xmath59 to @xmath60@xmath61@xmath62 .
the first term in @xmath63 is the non - magnetic lda part and @xmath64 are the lda eigenvalues .
@xmath65 is the matrix of relativistic soi , which is computed in the conventional way , @xmath65@xmath4@xmath66@xmath4@xmath67 , @xmath68 is the operator of angular momentum , @xmath69 is the vector of pauli matrices , @xmath70 is the self - consistent lda potential , and @xmath71 is the velocity of light ( @xmath71@xmath72@xmath73 in ry units ) . the inclusion of @xmath65 into the model hamiltonian is optional , that will be always specified in the text .
all magnetic effects are caused by the on - site coulomb interactions , @xmath74 , which can be written in terms of projections onto an orthonormal set of atomic - like mo(@xmath6 ) orbitals @xmath75 : @xmath76 where @xmath77 are the translation vectors in the face - centered cubic pyrochlore lattice , and @xmath78 specify different mo sites in the unit cell .
@xmath79@xmath80@xmath81 are the joint indices including the spin ( @xmath82@xmath4 @xmath83 or @xmath84 ) and orbital ( @xmath85@xmath4 @xmath86 , @xmath87 , @xmath88@xmath89@xmath90 , @xmath91 , or @xmath92@xmath89@xmath93 ) degrees of freedom
. matrix elements of the hf potential , @xmath94 , can be expressed in the most general rotationally - invariant form as:@xcite @xmath95 where @xmath96@xmath4@xmath97 are the matrix elements of the coulomb interactions .
the latter are fully specified by three radial slater integrals : @xmath98 , @xmath99 , and @xmath100 , which are assumed to be _
renormalized _ from the bare atomic values .
equivalently , one can introduce parameters of the averaged coulomb interaction , @xmath7@xmath4@xmath98 , the intra - atomic exchange coupling , @xmath101@xmath4@xmath102@xmath103@xmath104 , and the non - sphericity of the coulomb interactions between orbitals with the same spin , @xmath105@xmath4@xmath106@xmath89@xmath107 .
generally , both @xmath108@xmath4@xmath7@xmath89@xmath101 and @xmath105 contribute to the orbital polarization in solids and affect the properties caused by violation of the time - inversion symmetry.@xcite the density matrix @xmath109@xmath80@xmath110 can be obtained from the green function @xmath111@xmath80@xmath112 , using the standard relation @xmath113 where @xmath114 is the spectral representation for the green function in terms of the eigenfunctions @xmath115 and the eigenvalues @xmath116 of the hamiltonian ( [ eqn : hamiltonian ] ) .
@xmath117 is the volume of the first brillouin zone .
( [ eqn : hamiltonian ] ) , ( [ eqn : dmatrix ] ) and ( [ eqn : green ] ) are solved self - consistently .
then , the total energy is given by @xmath118 according to the constraint - lsda calculations,@xcite @xmath101 for mo is of the order of @xmath119 ev , and is not sensitive to details of the crystal environment in solids .
@xmath105 can be estimated from @xmath101 using the ratio @xmath120@xmath121@xmath122 , which holds for the slater integrals in the atomic limit .
this yields @xmath105@xmath121@xmath123 ev . the coulomb @xmath7 is treated as the parameter in order to consider different scenarios , covering both metallic and insulating behavior of @xmath0mo@xmath1o@xmath2
. the constraint - lsda calculations for mo compounds typically yield @xmath7@xmath72@xmath124 ev.@xcite this value can be further reduced by allowing for the ( proper ) @xmath48 electrons to participate in the screening of the on - site coulomb interactions between @xmath5 electrons.@xcite for the practical calculations along this line , it is convenient to use a short atomic - like basis set . for these purposes , @xmath64 and @xmath55
have been first calculated using the nearly - orthogonal version of the linearized - muffin - tin orbital method in the atomic - spheres approximation ( asa - lmto).@xcite the atomic orbitals @xmath75 have been identified with the basis orbitals of the lmto method .
atomic spheres radii have been chosen so to reproduce the electronic structure obtained by more accurate flapw method,@xcite especially for @xmath5 bands located near the fermi level .
we have also added a number of empty spheres at the @xmath125 and @xmath126 positions of the @xmath127 lattice .
all calculations have been performed on the mesh of 12@xmath6112@xmath6112 @xmath128-points in the first brillouin zone .
formally , our approach is similar to the rotationally - invariant lda@xmath103@xmath7,@xcite except that we apply it only to a limited number of bands , directly responsible for the properties of @xmath0mo@xmath1o@xmath2 compounds .
this will certainly simplify the calculations .
in addition , we get rid of several artifacts of the conventional lda@xmath103@xmath7 method , such as the substraction of the double - counting terms , which is a very ambiguous procedure by itself , and in the number of cases leads to a systematic error.@xcite
let us first discuss results without soi , assuming fm ordering of the mo spins . according to the lsda calculations ( fig .
[ fig.dost2g ] ) , the majority ( @xmath83 ) spin @xmath52 band is fully occupied and the fermi level crosses the doubly - degenerate @xmath53 band . therefore , it is clear that at some point the coulomb @xmath7 will split the @xmath53 band and form an insulating solution with the spontaneously broken @xmath23 symmetry .
such situation occurs between @xmath7@xmath4 2.0 and 2.5 ev for all considered compounds ( fig .
[ fig.gap ] , the afm alignment of the mo spins may open the band gap even for smaller @xmath7 , that well correlates with the fact that all sg compounds are insulators ) . in the metallic regime
realized for small @xmath7 , the densities of states are similar to those in lsda ( figs .
[ fig.doslsda ] and [ fig.dost2g ] ) , and we will not show them here again .
we only mention that the major effect of @xmath7 is the shift of the @xmath83-spin @xmath52 band to the low - energy part of the spectrum relative to the @xmath53 band .
corresponding orbital ordering ( the distribution of mo(@xmath6 ) electron densities ) is shown in fig .
[ fig.ooyf05 ] .
it comes exclusively from the local trigonal distortions of the oxygen octahedra and represents the alternating @xmath52 orbital densities in the background of degenerate @xmath53 orbitals .
typical densities of states in the insulating phase are shown in fig .
[ fig.dos30 ] .
the @xmath52 band has characteristic for pyrochlores three - peak structure.@xcite smaller band gap in the case of y@xmath1mo@xmath1o@xmath2 ( fig .
[ fig.gap ] ) is related with broad minority ( @xmath84 ) spin @xmath52 band , whose tail spreads below the unoccupied @xmath83-spin @xmath53 band .
the distribution of the @xmath53 states is very similar for all three compounds .
contrary to the metallic state , the orbital ordering is determined not only by local trigonal distortions , but also the form of the superexchange ( se ) interactions between _ nn _ mo sites , and minimizes the energy of these interactions.@xcite generally , it depends on the magnetic state .
two typical examples for the fm and afm ( obtained after the flip of the spin moments at the sites 2 and 3 ) phases are shown in figs .
[ fig.ooyf30 ] and [ fig.ooya30 ] , respectively .
the building blocks of the afm phase are the fm chains of the atoms ... -1 - 4 - 1 - 4- .. and ...
-2 - 3 - 2 - 3- ... , which are coupled antiferromagnetically .
as expected for the fm spin ordering,@xcite the @xmath53 orbitals tends to order antiferromagnetically and form two mo sublattices .
corresponding eigenfunctions of the occupied @xmath53 orbitals , obtained after the diagonalization of the density matrix _ in the local coordinate frame _ , are given by @xmath129@xmath121@xmath130@xmath103@xmath131@xmath103@xmath132 for the sites 1 and 4 , and @xmath133@xmath121@xmath130@xmath103@xmath134@xmath103@xmath135 for the sites 2 and 3 , with the coefficients @xmath136@xmath121@xmath137 , @xmath138@xmath121@xmath139 , and @xmath140@xmath121@xmath89@xmath141 .
clearly , this orbital ordering breaks the @xmath23 symmetry : if the sites belonging to the same sublattice can still be transformed to each other using the symmetry operations of the @xmath23 group , the sites belonging to different sublattices can not
. this will generally lead to the anisotropy of electronic properties , including the anisotropy of _ nn _ magnetic interactions , which will be discussed in the next section .
the afm spin ordering coexists with the fm orbital ordering for which @xmath136@xmath121@xmath89@xmath142 , and @xmath138@xmath4@xmath140@xmath121@xmath143 for all mo sites , in the local coordinate frame.@xcite it breaks the @xmath23 symmetry in the spin sector , but not in the orbitals one .
parameters of exchange interactions between sites @xmath78 and @xmath144 can be calculated as@xcite @xmath145 where @xmath146 is the block of matrix elements of the green function with the spin @xmath83 or @xmath84 : @xmath147 and @xmath148 is the magnetic part of the hf potential : @xmath149 @xmath150 is the trace running over the spin ( orbital ) indices .
results for _ nn _ magnetic interactions for all considered compounds are shown in fig .
[ fig.exchange ] , as a function of @xmath7 .
we note the following .
1 . @xmath151 , which are ferromagnetic for small @xmath7 , exhibit a sharp drop at the point of transition into the insulating state ( see fig . [ fig.gap ] for the behavior of the band gap ) .
2 . there is a significant difference between nd / gd and y based compounds : in the y case , the exchange parameters are almost rigidly shifted towards negative values , so that the _ nn _ coupling become antiferromagnetic in the insulating phase , while it remains ferromagnetic in the case of nd and gd .
similar conclusions can be obtained from the analysis of total energies shown in fig .
[ fig.etot].@xcite the behavior can be easily understood by considering partial @xmath52 and @xmath53 contributions to the _ nn _ exchange coupling , calculated after transformation to the local coordinate frame at each site of the system ( fig .
[ fig.jpartial ] ) .
then , the main interactions are @xmath52-@xmath52 and @xmath53-@xmath53 . the @xmath52-@xmath53 interaction is small and can be neglected .
the interaction between @xmath5 and proper @xmath48 orbitals is of the order of 1.5 - 1.8 mev and only weakly depends on @xmath7 and the type of the material .
therefore , it can be regarded as the weak fm background .
large fm @xmath53-@xmath53 interaction in the metallic regime is related with the double exchange ( de ) mechanism , which is the measure of the kinetic energy for the itinerant @xmath83-spin @xmath53 electrons ( note that for small @xmath7 all compounds are either half - metallic or nearly half - metallic fig .
[ fig.doslsda]).@xcite some increase of @xmath151 for small @xmath7 is caused by the reduction of afm se contributions , which are proportional to @xmath152 .
as long as the system is metallic , the de interactions are not sensitive to the value of @xmath7 , and the fm coupling dominates .
the transition into the insulating state is caused by the localization of the @xmath53 electrons on the atomic orbitals .
this will reduce the kinetic energy and suppress the de interactions , that explain the sharp drop of @xmath151 ( fig .
[ fig.exchange ] ) .
the main difference between y and nd / gd based compounds is related with the @xmath52-@xmath52 interactions .
since the @xmath83-spin @xmath52 band is fully occupied and the @xmath84-spin band is empty , the interactions are antiferromagnetic and the underlying microscopic mechanism is the superexchange . since the se coupling is proportional to the square of the @xmath52 bandwidth , this interaction is the largest in the case of y , that explains the afm character of the total coupling realized in this compound for large @xmath7.@xcite the anisotropy of the _ nn _ interactions , @xmath153@xmath4@xmath154@xmath89@xmath155 , caused by the orbital ordering in the fm state is of the order of 1 - 2 mev ( fig .
[ fig.exchange ] ) .
this is a small value in comparison with the sharp drop of @xmath151 caused by the coulomb interaction @xmath7 .
however , it can play an important role in the insulating regime , where @xmath153 becomes comparable with @xmath151 .
for example , one interesting question is whether this @xmath153 , of a purely coulombic origin , is sufficient to explain the appearance of the sg state .
in addition , @xmath153 has the same order of magnitude as the single - ion anisotropy energy ( sec .
[ sec : nc ] ) and the mo - gd exchange interaction ( ref . ) .
therefore , in the insulating state there is no unique interaction which would control the magnetic behavior of the @xmath0mo@xmath1o@xmath2 compounds .
all interactions are equally important , that represents the main complexity on the way of theoretical description of these materials .
the afm spin ordering changes the orbital ordering in the direction which further stabilizes the _ nn _ magnetic interactions,@xcite and in a number of cases may lead to the charge disproportionation within the fm chains.@xcite for example , @xmath7@xmath4@xmath156 ev appears to be sufficient to open the band gap ( @xmath8@xmath157 ev ) in the afm phase of y@xmath1mo@xmath1o@xmath2 .
the solution is accompanied by the small charge disproportionation , and the values of the spin magnetic moments at the four mo sites are @xmath158@xmath4@xmath89@xmath159@xmath4@xmath160 @xmath161 and @xmath89@xmath162@xmath4@xmath163@xmath4@xmath9 @xmath161 .
corresponding values of the inter - atomic exchange interactions are @xmath154@xmath4@xmath89@xmath164 , @xmath165@xmath4@xmath89@xmath166 and @xmath155@xmath4@xmath167 mev , i.e. the interactions are antiferromagnetic in the bonds 1 - 2 and 1 - 3 , and ferromagnetic in the bond 1 - 4 .
therefore , the obtained solution is locally stable .
the difference between @xmath154 and @xmath165 is caused by the charge disproportionation .
further increase of @xmath7 stabilizes the afm coupling in all mo - mo bonds , giving rise to the geometrical frustrations .
typical values of the exchange parameters obtained in the hf solution for @xmath7@xmath4@xmath124 ev , which does not show any sign of the charge disproportionation , are @xmath154@xmath4@xmath165@xmath4@xmath89@xmath168 and @xmath154@xmath4@xmath89@xmath169 mev .
in this and subsequent sections we will discuss the effects of soi on the electronic and magnetic structure of the mo pyrochlores .
the matrix element of @xmath170 calculated in the basis of mo(@xmath6 ) orbitals is of the order of @xmath171@xmath4@xmath172 mev .
it defines the characteristic energy gain after adding the soi into the model hf hamiltonian .
the equilibrium magnetic ordering depends on the competition between the isotropic exchange interactions , considered in the previous section , and the single - ion anisotropies energy .
the isotropic exchange interactions tend to align mo spins parallel to each other ( at least in the case of the fm coupling ) .
however , due to the peculiarities of the pyrochlore structure , there is no unique direction of the spin magnetization which would satisfy the single - ion anisotropy energies simultaneously for all mo sites .
these anisotropies lead to the distortion of the collinear fm ordering.@xcite in the case of fm coupling between _ nn _ spins , there are two magnetic structures , which are typically discussed in the literature and can be obtained by distorting the collinear fm ordering along either @xmath173 or @xmath174 direction ( see fig . [ fig.fstructures ] ) . 1 .
the two - in , two - out structure , where the directions of the magnetic moments at the four mo sites are given by : @xmath175@xmath4@xmath176@xmath89@xmath177 , @xmath178@xmath4@xmath179@xmath89@xmath180@xmath89@xmath177 , @xmath181@xmath4@xmath182@xmath89@xmath183@xmath89@xmath177 , and @xmath184@xmath4@xmath179@xmath89@xmath185@xmath89@xmath183@xmath89@xmath177 , in terms of the distortion @xmath186 from the perfect fm alignments . 2 .
the umbrella structure : @xmath175@xmath4@xmath187 , @xmath178@xmath4@xmath188@xmath89@xmath189@xmath103@xmath183@xmath103@xmath190 , @xmath181@xmath4@xmath188@xmath103@xmath183@xmath89@xmath189@xmath103@xmath190 , and @xmath178@xmath4@xmath188@xmath103@xmath183@xmath103@xmath183@xmath89@xmath191 . since the mo(@xmath6 ) states are strongly polarized by the on - site coulomb interactions , the main contribution to the single - ion anisotropy energy comes from the @xmath83-spin electrons . in this case
there is an one - by - one correspondence between the single - ion anisotropy energy and the magnitude of the orbital magnetization , so that the direction which maximizes the magnetization automatically minimizes the magnetic anisotropy energy.@xcite for the @xmath5 electrons affected by the trigonal distortion , the spontaneous orbital magnetization is determined by the following factors:@xcite 1 .
the interaction between occupied @xmath52 and unoccupied @xmath53 states favors the in - plane geometry of the orbital magnetization ( i.e. , perpendicular to the trigonal axes ) .
the latter is proportional to the ratio @xmath192 , where @xmath193 is the distance between the position of the @xmath52 states and the unoccupied part of the spectrum .
clearly , this interaction will vanish when the coulomb @xmath7 is large .
2 . on the contrary , the interaction between @xmath53 orbitals favors the easy axes geometry . in the pure ionic limit ,
such solution is the subject for the degenerate perturbation theory .
therefore , the wave functions and the orbital magnetization do not depend on @xmath171 .
the soi in such situation only lifts the orbital degeneracy and defines the symmetry of the @xmath53 orbitals which will be further split by @xmath7 .
the solution is accompanied by the large orbital moment of the order of @xmath59 @xmath161 , which is parallel to the trigonal axis .
thus , when @xmath7 is sufficiently large , the second mechanism will dominate and one can naturally expect the easy axes scenario for the single - ion anisotropies . in the combination with the fm coupling between _ nn _ spins , it yields the umbrella ground state.@xcite the qualitative analysis is supported by results of direct calculations shown in fig .
[ fig.canting ] . in the metallic regime ,
the isotropic double exchange interaction between mo spins clearly dominates over the single - ion anisotropy energies and the canting of spin magnetic moments is small ( both for the two - in , two - out and umbrella structure ) .
the orbital moments are not particularly strong ( @xmath17@xmath194 @xmath161 per mo site ) .
they are canted off the global fm direction by about @xmath195 . in the insulating state
, the isotropic exchange interactions are suppressed and become comparable with the single - ion anisotropy energies , which are enhanced .
this results in the large canting of the magnetic moments , which can be very different for the spin and orbital counterparts ( in the other words , the spin and orbital moments are not collinear with respect to each other).@xcite the reason for this difference is the following .
directions of the orbital magnetic moments are determined mainly by the single - ion anisotropy terms , whereas the interactions between moments located at different sites are typically much smaller .
therefore , the orbital moments always tend to align first along the crystallographic directions ( the trigonal axes , in our case ) . on the contrary , alignment of the spin magnetic moments
is strongly affected by the inter - site interactions , which generally lead to a smaller canting . for large @xmath7 ( @xmath83 ev ) , the single - ion anisotropies start to dominate over the isotropic exchange interactions . in this limit , the orbital magnetic moments are almost perfectly aligned along the trigonal axes ( that corresponds to the 55@xmath196 and 70@xmath196 canting for the two - in , two - out and umbrella structure , respectively ) .
the value of the local orbital magnetic moment is about 0.6 @xmath161 , which is smaller than 1 @xmath161 expected in the pure atomic limit . the umbrella structure has lower energy ( perhaps except the region of very small @xmath7 ) .
the energy difference between the two - in , two - out and umbrella structures is small in the metallic regime ( @xmath8@xmath197 mev per formula unit , which corresponds to the coercive field of 1 tesla ) .
however , it becomes large in the insulating state , where the interactions between the @xmath53 orbitals clearly dominate . in the case of afm coupling between _
nn _ spins , the single - ion anisotropies lift , at least partially , the high degeneracy of the magnetic ground state in the frustrated pyrochlore lattice .
the equilibrium magnetic structure realized in model hf calculations for y@xmath1mo@xmath1o@xmath2 can be obtained by the inversion of the magnetic moments at the sites 2 and 3 of the two - in , two - out structure ( fig . [ fig.astructure ] ) . for @xmath7@xmath4@xmath124
ev , the values of spin and orbital magnetic moments at the mo sites are @xmath198 and @xmath199 @xmath161 , respectively .
they are canted off the cubic ( 0,0,1 ) axis by correspondingly @xmath200 and @xmath201 .
the afm interactions are compromised with the tetrahedral geometry of the pyrochlore lattice by forming the afm coupling between two of the three projections of the _ nn _ spins in each of the mo - mo bonds , while the remaining projection is coupled ferromagnetically .
the conductivity tensor , @xmath202@xmath4@xmath203@xmath103@xmath204 , can be calculated from the eigenvalues , @xmath205 , and the eigenfunctions , @xmath206 , of the hf hamiltonian using the kubo formula,@xcite where the inter - band ( regular ) and intra - band ( drude ) contributions are given by ( in ry units ) @xmath207 \label{eqn : sinter}\ ] ] and @xmath208 respectively . in these equations
, @xmath209 is the zero - temperature fermi distribution function , @xmath210@xmath4@xmath211 is the matrix element of momentum operator , @xmath212@xmath4@xmath213@xmath89@xmath214 , @xmath215 is the unit cell volume , and @xmath216 is the phenomenological relaxation time ( @xmath217@xmath121@xmath218 mev ) introduces in order to perform summation over the discrete mesh of @xmath128-points .
the diagonal elements of @xmath204 can be also expressed through the plasma frequencies , @xmath219 , as @xmath220@xmath4@xmath221 .
the conductivity tensor has the form @xmath222 both for the two - in , two - out and umbrella structure , if the direction of the @xmath41 axis is chosen as @xmath173 and @xmath174 , respectively . the hall conductivity , defined as @xmath18@xmath4@xmath223 , is shown in fig . [ fig.hall ] as a function of @xmath7 . as expected,@xcite the effect is caused by the inter - band transitions , while all intra - band contributions vanish , even in the metallic state
@xmath18 takes the maximal value around @xmath7@xmath4@xmath9 ev , which compromises between the value of the orbital magnetization and the metallic behavior of the system . on the one hand
, @xmath7 is large enough to unquench the orbital magnetic moment of the order of 0.08 @xmath161 per mo site . on the other hand
it is small enough in order to keep the system in the metallic regime ( see fig .
[ fig.gap ] ) .
the factor - two increase of @xmath18 from @xmath7@xmath4@xmath119 to @xmath9 ev well correlates with similar increase of the orbital magnetic moment ( fig .
[ fig.canting ] ) , suggesting that both effects have relativistic origin .
the behavior coexists with nearly collinear fm ordering realized in the metallic regime .
the transition into the insulating state , which is accompanied by the large canting of the magnetic moments , results in the abrupt drop of @xmath18.@xcite therefore , we conclude that the large @xmath18 observed in the experiment is due to the orbital magnetic moments induced by the soi in the nearly collinear fm state.@xcite the absolute value of @xmath18 , @xmath8@xmath224 @xmath21 , obtained in the calculations is in a good agreement with the experimental data for nd@xmath1mo@xmath1o@xmath2.@xcite however , we would like to emphasize again that the soi alone is not sufficient to obtain such a large values of @xmath18 , and it is essential to consider the on - site coulomb interactions beyond lsda , which play a very important role in the problem and enhance the effect of the soi.@xcite according to our calculations , @xmath225 , in the geometry which specify the form of the tensor ( [ eqn : tensor ] ) , only weakly depends on the magnetic structure and the direction of the net magnetization .
large anisotropy of @xmath18 observed in the experiment@xcite is presumably related with the details of the experimental set - up and reflects the
@xmath226-law .
for example , if the electric filed is directed along the @xmath227 axis , and the current is measures along the @xmath228 axis , the effect is proportional to @xmath229 in the case of the umbrella structure , that roughly corresponds to the experimental situation .
similar arguments apply for the magneto - optical kerr rotation which will be discussed below .
if our scenario is correct , and large ahe in the mo pyrochlores can be explained by the inter - band transitions in the presence of the soi ( i.e. , by the conventional means ) , it is natural to expect similar enhancement for other phenomena related with the soi .
below we consider the complex kerr effect , which can be also regarded as a possible check for our theory in the future . the kerr rotation ( @xmath230 ) and kerr ellipticity ( @xmath231 ) can be found from the expression,@xcite @xmath232 which is formally exact and valid for arbitrary @xmath230 and @xmath231 .
@xmath233@xmath4@xmath234 are the complex refractive indices , in terms of matrix elements of the dielectric tensor , @xmath235@xmath4@xmath236@xmath103@xmath237 .
results of these calculations for gd@xmath1mo@xmath1o@xmath2 are shown in fig .
[ fig.conductivity ] , together with the @xmath238-dependence of the matrix elements of the optical conductivity .
we consider the umbrella
structure obtained for @xmath7@xmath4@xmath9 ev , which corresponds to the maximal rotation .
calculations for nd@xmath1mo@xmath1o@xmath2 and the two - in , two - out structure yield very similar results .
we also discuss partial contributions of the @xmath52 and @xmath53 states , and show the spectra obtained after removing the @xmath52 orbitals from the basis.@xcite apart from the hf model itself , we do not use any adjustable parameters .
for example , the intra - band drude contribution to the optical conductivity has been obtained in the calculations .
the diagonal part of @xmath202 comes mainly from the inter - band @xmath53@xmath54@xmath53 transitions .
the contribution of the @xmath52 states is rather small .
the absorptive ( re ) part of @xmath239 has a maximum around 1 ev .
both the peak position and its intensity roughly agree with the experimental data,@xcite tough the latter reveal an additional structure in the low-@xmath238 energy part of the spectrum , which can be due to the quantum fluctuation effects,@xcite beyond the mean - field approximation considered in our work.@xcite the contribution of the @xmath52 states to @xmath225 is very important , and can bring up to 50% of the spectral weight at certain values of @xmath238 .
however , the low-@xmath238 part of @xmath225 comes mainly from the inter - band @xmath53@xmath54@xmath53 transitions , which are responsible for almost 90% of the hall conductivity .
the large intensity of @xmath225 is due to the combination of three factors : ( i ) large soi at the mo sites ; ( ii ) the enhancement of the soi by the on - site coulomb interactions , which is especially efficient for the degenerate @xmath53 states ; ( iii ) the half - metallic electronic structure of gd@xmath1mo@xmath1o@xmath2 and nd@xmath1mo@xmath1o@xmath2 : since only @xmath83-spin states contribute to the low-@xmath238 part of @xmath225 , the cancellation between different spin channels does not take place.@xcite the kerr rotation spectrum , calculated using eg .
( [ eqn : kerr ] ) , has two structures .
the first one is close to the plasma edge ( @xmath240@xmath121@xmath157 ev ) , and is directly related with the plasma enhancement effect.@xcite the exceptionally large value of @xmath230 ( @xmath8@xmath241 ) is also related with large @xmath225 in this region , which correlates with the large value of the hall conductivity . the second structure ( @xmath230@xmath8@xmath89@xmath242 ) near @xmath243 ev
is related with the form of @xmath244 , which has a deep in this region as a joint effect of both @xmath53@xmath54@xmath53 and @xmath52@xmath54@xmath53 inter - band transitions .
smaller @xmath7 would yield somewhat smaller estimates for the rotation angles ( correspondingly @xmath245 and @xmath89@xmath246 for @xmath7@xmath4@xmath119 ev ) .
nevertheless , the fm pyrochlores @xmath0mo@xmath1o@xmath2 seem to be rather promising materials for the magneto - optical applications .
the anomalous hall and kerr effects simultaneously vanish in the afm phase , due to the cancellation of contributions coming from different mo sublattices in the macroscopic sample.@xcite
we have presented results of model hf calculations for pyrochlores @xmath0mo@xmath1o@xmath2 ( @xmath3@xmath4 y , gd , and nd ) .
the model combines details of lda electronic structure for the mo(@xmath5 ) bands located near the fermi level , the spin - orbit interaction , and the on - site coulomb interactions amongst the mo(@xmath6 ) electrons .
the main results can be summarized as follows .
@xmath11 the magnetic ground state of the @xmath0mo@xmath1o@xmath2 compounds depends on the mo - mo distance , which controls the strength of afm se interactions between the @xmath52 orbitals .
the latter competes with the fm interactions between the @xmath53 orbitals , and dominates when the mo - mo distance is small . in the frustrated pyrochlore lattice , this corresponds to the transition into the sg state , though we can not specify precisely which mechanism freeze the random configuration of the mo spins .
it could be the disorder of the mo - mo bondlengths proposed in refs . and .
another possibility is the anisotropy of nearest - neighbor magnetic interactions caused by the orbital ordering.@xcite it is also important to note that the coulomb repulsion @xmath7 should be sufficiently strong in order to open the band gap and suppress the fm double exchange interactions between the @xmath53 orbitals .
@xmath11 the metal - insulator transition in these systems does not necessary coincide with the fm - sg transitions .
all metallic compounds are expected to be nearly collinear ferromagnets , due to the double exchange mechanism . however , the inter - atomic exchange coupling in the insulating state can be both ferromagnetic and antiferromagnetic , depending on the mo - mo bondlength .
the finding is qualitatively consistent with the experimental data.@xcite the magnetic structure in the insulating state is expected to be strongly non - collinear , due to the competition between the isotropic exchange interactions and the single - ion anisotropies .
@xmath11 the large hall conductivity can be explained by the combination of three factors in the nearly collinear fm state : large soi at the mo sites , which unquenches the orbital magnetization ; strong enhancement of this effect by the on - site coulomb interaction ; and the half - metallic electronic structure of the fm pyrochlores .
we also expect large magneto - optical effect in the same compounds .
a possible way to enhance these properties is to use heavy @xmath50 elements , which have large spin - orbit coupling .
unfortunately , all known @xmath50 pyrochlores are non - magnetic , though some of them are presumably close to the point of magnetic instability.@xcite nevertheless , one can try to exploit an old idea known from the physics of intermetallic alloys,@xcite and replace part of the mo sites in @xmath0moo@xmath2 by the @xmath50 elements . then , the main question is whether the remaining mo sites will be magnetic in the new environment or not . if they will , they will also magnetize the @xmath50 states at the neighboring sites , and in this way enhance the non - reciprocal characteristics of the considered pyrochlore compounds . as a purely model example
, we have considered the ordered gd@xmath1mowo@xmath2 alloy ( using the lattice parameters of gd@xmath1mo@xmath1o@xmath2 ) . according to the lsda calculations ,
the system is expected to be ferromagnetic .
the values of ( spin only ) magnetic moments at the mo and w sites are 2.1 and 1.5 @xmath161 , respectively .
the situation is highly unusual , because the examples when @xmath6 and @xmath50 elements form a magnetic solution are rather rare .
although gd@xmath1mowo@xmath2 itself may be rather unrealistic example from the technological point of view , the idea to use mo atoms in order to magnetize the heavy @xmath50 elements in the pyrochlore lattice seems to be promising . in conclusion , the simple hf calculations allow to rationalize many unusual properties of the pyrochlore compounds @xmath0mo@xmath1o@xmath2 .
it is also important that all these properties can be explained by the universal value of @xmath7@xmath8@xmath9-@xmath10 ev , which is of the same order of magnitude as the characteristic bandwidth for the mo(@xmath5 ) electrons .
this will certainly place the considered compounds to the direct proximity of the metal - insulator transition , meaning that the quantum fluctuations beyond the mean - field hf approach will play an important role too .
this is certainly a very important direction for the extension of the present analysis .
finally , we would like to make several comments on the comparison between our calculations and the experimental magnetization@xcite and neutron diffraction@xcite data .
\(1 ) the experimental value of the mo moment ( @xmath8@xmath247-@xmath248 @xmath161 ) seems to be smaller than @xmath8@xmath160 @xmath161 obtained in our mean - field calculations for the nearly collinear fm state , even after taking into account the orbital part ( @xmath8@xmath194 @xmath161 ) , which according to the third hund s rule is antiparallel to the spin one and partly compensate the latter .
the small difference is presumably related with the quantum fluctuations at the proximity of the metal - insulator transition.@xcite \(2 ) the experimental data for nd@xmath1mo@xmath1o@xmath2 and sm@xmath1mo@xmath1o@xmath2 typically indicate at the existence of two antiferromagnetically coupled sublattices .
these sublattices are usually regarded as mo and @xmath3 ones . however , according to the electronic structure point of view , the _ spin _ coupling between these two sublattices should be ferromagnetic,@xcite being actually consistent with the experimental data for gd@xmath1mo@xmath1o@xmath2.@xcite the controversies can be reconciled with each other if the local magnetic moment at the nd / sm sites has a large orbital counterpart , which is larger than the spin one .
the afm orientation of the spin and orbital moments at the nd / sm sites is again the subject of the third hund s rule .
then , the magnetization and neutron diffraction experiments presumably probe the afm coupling between _ spin _ magnetic moments at the mo sites and the _ orbital _ magnetic moments at the nd / sm sites .
gd@xmath249 has no orbital moment , and therefore the experimental data for this compound indicate at the conventional fm coupling between two spin sublattices . then , the sharp decrease of the magnetization observed in nd@xmath1mo@xmath1o@xmath2 at low @xmath19,@xcite which is frequently tempted to connect with the anomaly of the hall conductivity is due to the orbital magnetic structure , not the spin one . the change of the _ orbital _ magnetic structure seems to be the right subject to concentrate at when trying to explain , for example , the temperature dependence of the ahe . in this respect , another interesting problem seems to be the behavior of _ spin and orbital _
magnetic moments at the mo sites , because already in the ground state these two sublattices from mutually non - collinear magnetic structures .
therefore , it is reasonable to expect very unconventional spin and orbital dynamic at the elevated temperatures .
i would like to thank lars nordstrm for collaboration on early stages of this project and kind hospitality during my stay at uppsala university .
i also benefited from discussions with n. nagaosa , y. tokura , k. terakura , and a. b. granovski .
j. ye , y. b. kim , a. j. millis , b. i. shraiman , p. majumdar , and z. teanovish , phys .
lett . * 83 * , 3737 ( 1999 ) .
y. lyanda - geller , s. h. chun , m. b. salamon , p. m. goldbart , p. d. han , y. tomioka , a. asamitsu , and y. tokura , phys .
b * 63 * , 184426 ( 2001 ) .
j. smith , physica * 21 * , 877 ( 1955 ) ; j. smith , _ ibid .
_ * 24 * , 39 ( 1958 ) ; j. m. luttinger , phys . rev . * 112 * , 739 ( 1958 ) ; j. kondo , prog . theor
. phys .
* 27 * , 772 ( 1962 ) ; f. e. maranzana , phys .
rev . * 160 * , 421 ( 1967 ) ; l. berger , phys .
b * 2 * , 4559 ( 1970 ) ; l. berger , _ ibid . _ * 5 * , 1862 ( 1972 ) ; s. k. lyo and t. holstein , phys . rev .
* 29 * , 423 ( 1972 ) ; j. smith , phys .
rev . b * 17 * , 1450 ( 1978 ) .
we would not like to go here very deeply into all details of these controversies .
briefly , karplus and luttinger in their pioneering work@xcite considered the contributions proportional to @xmath250 , where @xmath251 is the fermi distribution function .
these contributions have the form of intra - band transitions in the conductivity tensor depicted by eq .
( [ eqn : sintra ] ) .
all of them vanish in the periodic crystal , as it was correctly pointed out by smith.@xcite however , if the product of two momentum operators , @xmath252 , has an imaginary part , one can expect another contribution to @xmath18 coming from eq .
( [ eqn : sinter ] ) .
this contribution is caused by the inter - band transitions , and in this sense the considered mechanism has the same origin as the magneto - optical effect .
this term was intensively studied recently by nagaosa _
_ in the contexy of the spin - chirality problem.@xcite see also comments in ref . .
a. b. granovskiy and ye .
i. kondorskiy , fiz . metal .
metalloved .
* 39 * , 718 ( 1975 ) [ english transl . : the physics of metals and metallography * 39 * , no.4 , p.36 ( 1975 ) ] ; ye .
i. kondorskiy , a. v. vedyayev , and a. b. granovskiy , _ ibid . _ * 40 * , 455 ( 1975 ) [ english transl . :
the physics of metals and metallography * 40 * , no.3 , p.1 ( 1975 ) ] ; n. ryzhanova , a. vedyayev , a. crpieux , and c. lacroix , phys .
b * 57 * , 2943 ( 1998 ) . using expression ( [ eqn : sinter ] ) for the conductivity tensor , it is easy to show that @xmath253@xmath4@xmath254@xmath89@xmath255
. therefore , @xmath256@xmath4@xmath22 , whereas @xmath18@xmath80@xmath257 can be finite ( the author thank n. nagaosa for pointing out this property ) . on the other hand , the onsager constraint @xmath258@xmath4@xmath259@xmath89@xmath260 ( @xmath79 being the net magnetization ) requires @xmath18 to be an even function of @xmath79.@xcite strictly speaking , the effect is masked by the diagonal elements of the conductivity tensor , and since @xmath261@xmath54@xmath262 , the resistivity @xmath263@xmath4@xmath89@xmath264 , which is the measured quantity , will vanish anyway in the perfectly periodic metallic state . in this sence ,
additional scattering mechanisms@xcite are indispensable in order to have finite @xmath261 and @xmath263 .
however , it does not mean that @xmath265 itself will vanish in the periodic state .
in our work we will consider such , perhaps somewhat idealistic , situation . in the local coordinate frame ,
the @xmath52 and two @xmath53 orbitals have the following form : @xmath266@xmath4@xmath267@xmath103@xmath268@xmath103@xmath269 , @xmath270@xmath4@xmath271@xmath89@xmath269 , and @xmath272@xmath4@xmath273@xmath89@xmath274@xmath103@xmath268@xmath103@xmath275 . by considering the _ nn _ hoppings between mo(@xmath52 ) orbitals in the pyrochlore lattice , @xmath276 , it is easy to show that the electronic structure for the @xmath52 states will consist of three bands ( in units of @xmath276 ) : @xmath277@xmath4@xmath89@xmath62@xmath278 , the so - called bonding and antibonding bands , and the doubly - degenerate non - bonding band @xmath279@xmath4@xmath62 , where @xmath280@xmath4@xmath281@xmath282@xmath283@xmath103@xmath284 . in fact , there is a non - spherical ( proportional to the coulomb parameter @xmath105 ) interaction beween occupied @xmath53 and @xmath52 orbitals , which will deforms the density of the @xmath52 electrons .
this interaction has an appreciable effect in the afm state .
for example , for y@xmath1mo@xmath1o@xmath2 and @xmath7@xmath4@xmath124 ev , the occupied @xmath52 orbital at the site 1 has the form @xmath285@xmath4@xmath130@xmath103@xmath131@xmath103@xmath135 , where @xmath136@xmath121@xmath286@xmath121@xmath287 .
this will also lead to a small ( less than 1 mev ) anisotropy of the @xmath52-@xmath52 superexchange interactions .
note that @xmath151 defined by eq .
( [ eqn : exchange ] ) , which is basically the second derivative of the total energy,@xcite characterizes the _ local stability _ of the magnetic state , and provides a _ complementary _ information to the standard total energy calculations .
generally , the parameters @xmath151 depend on the magnetic state in which they are calculated and there is no simple relation between @xmath151 and the total energy . on the other hand , the total energy difference calculated for a limited number of magnetic states does not necessary guarantee that these states are locally stable to be the local minima of the total energy .
the gd(@xmath288 ) core spin polarizes the valence , mainly gd(@xmath50 ) , electrons , which can interact with the mo(@xmath6 ) electrons .
according to our estimates based on eq .
[ eqn : exchange ] , the interaction @xmath289 is ferromagnetic and is of the order of 2 mev ( in lsda , and treating the gd(@xmath288 ) states as the core states ) , in a good agreement with the experimantal data.@xcite similar conclusion can be obtained from the analysis of the total energy difference , assuming the fm and afm coupling between mo and gd spins .
results of these calculations have very transparent expanation , because gd@xmath1mo@xmath1o@xmath2 has half - metallic electronic structure and the fm coupling between mo and gd spins results in the additional energy gain for the @xmath83-spin mo(@xmath5 ) electrons ( see fig . [ fig.doslsda ] ) . for comparison ,
the direct gd(@xmath50)-gd(@xmath50 ) interaction is much smaller ( @xmath8@xmath290 mev , which is formally beyond the accuracy verge of the present calculations ) .
presumably , other factors can also play some role in explaining the large difference of @xmath52-@xmath52 interactions obtained for y@xmath1mo@xmath1o@xmath2 and nd@xmath1mo@xmath1o@xmath2 .
the standard se interaction is proportional to the ( minus ) second moment of the density of states , that actually explains the afm sign of the coupling.@xcite the @xmath83-spin @xmath52 bandwidths in y@xmath1mo@xmath1o@xmath2 and nd@xmath1mo@xmath1o@xmath2 differ by factor two ( fig . [ fig.dos30 ] )
, that should lead to the factor four difference for corresponding se interactions . however , the @xmath52 states are not fully isolated and interact with some other states , so that the expression for the se interaction should be modified as @xmath291 , where @xmath292 is the shift of the @xmath52 states due to the hybridization with other states .
this shift can be different for the @xmath83- and @xmath84-spin electrons and also depend on the material , that presumably explains an additional difference between y@xmath1mo@xmath1o@xmath2 and nd@xmath1mo@xmath1o@xmath2 .
note that at least in the mean - field approximation for the one - orbital model , the effects of the inter - site coulomb interactions , which are responsible for the charge disproportionation , can be included as a renormalization of the local ( on - site ) interaction , in the spirit of model hf calculations performed in this work : i. v. solovyev , phys .
b * 60 * , 8550 ( 1999 ) . the conclusion can be easily verified by considering the spin hamiltonian @xmath293@xmath4@xmath89@xmath294@xmath103@xmath295 , consisting of the _ nn _ exchange and anisotropy parts .
@xmath296@xmath4@xmath297 is the single - ion anisotropy tensor at the site 1 .
@xmath298 , @xmath299 , and @xmath300 are obtained by the 180@xmath196 rotations of @xmath301 around @xmath39 , @xmath40 , and @xmath41 , respectively .
then if @xmath101@xmath13@xmath22 ( the fm coupling ) and @xmath302@xmath13@xmath22 ( the easy axes scenario ) , the umbrella
structure has lower energy in comparison with the two - in , two - out structure due to the additional energy gain coming from the _ nn _ exchange interactions .
note that @xmath18@xmath303@xmath304 , where @xmath305 is the band gap .
in addition , the non - collinear alignment reduces the net magnetization and the conventional contribution to the ahe caused by the soi.@xcite c. s. wang and j. callaway , phys .
rev . b * 9 * , 4897 ( 1974 ) ; yu .
a. uspenski and s. v. khalilov , zh . eksp .
95 * , 1022 ( 1989 ) [ english transl .
jetp * 68 * , 588 ( 1989 ) ] ; p. m. oppeneer , t. maurer , j. sticht , and j. kbler , phys .
b * 45 * , 10924 ( 1992 )
. another possibility could be the sample purity .
however , according to the experimental estimates the sample quality seems to be good ( for example in y@xmath1mo@xmath1o@xmath2 the concentration of anti - site defects and deviation from the nominal oxygen stoichiometry was estimated as 4% and 1% , respectively).@xcite .structural parameters of @xmath0mo@xmath1o@xmath2 ( @xmath3@xmath4 y , nd , and gd ) : the cubic lattice parameter @xmath36 ( in ) , positions of o @xmath27 sites @xmath310 $ ] ( units of @xmath36 ) , the distances mo - mo and mo - o ( in ) , and the angles mo - o - mo and o - mo - o ( in degrees ) . [ cols="^,^,^,^,^,^,^ " , ] | being motivated by recent experimental studies , we investigate magnetic structures of the mo pyrochlores @xmath0mo@xmath1o@xmath2 ( @xmath3@xmath4 y , nd , and gd ) and their impact on the electronic properties .
the latter are closely related with the behavior of twelve mo(@xmath5 ) bands , located near the fermi level and well separated from the rest of the spectrum .
we use a mean - field hartree - fock approach , which combines fine details of the electronic structure for these bands , extracted from the conventional calculations in the local - density approximation , the spin - orbit interaction , and the on - site coulomb interactions amongst the mo(@xmath6 ) electrons , treated in the most general rotationally invariant form .
the coulomb repulsion @xmath7 plays a very important role in the problem , and the semi - empirical value @xmath7@xmath8@xmath9-@xmath10 ev accounts simultaneously for the metal - insulator ( m - i ) transition , the ferromagnetic ( fm ) - spin - glass ( sg ) transition , and for the observed enhancement of the anomalous hall effect ( ahe ) .
the m - i transition is mainly controlled by @xmath7 .
the magnetic structure at the metallic side is nearly collinear fm , due to the double exchange mechanism .
the transition into the insulating state is accompanied by the large canting of spin and orbital magnetic moments .
the sign of exchange interactions in the insulating state is controlled by the mo - mo distances .
smaller distances favor the antiferromagnetic coupling , which preludes the sg behavior in the frustrated pyrochlore lattice .
large ahe is expected in the nearly collinear fm state , near the point of m - i transition , and is related with the unquenched orbital magnetization at the mo sites .
we also predict large magneto - optical effect in the same fm compounds . |
the present paper is devoted to the phenomena of bose einstein condensation ( bec for short ) on non homogeneous networks obtained by adding density zero perturbations to homogeneous cayley trees . even if the unperturbed model already exhibit bec ( cf .
@xcite ) , the arising picture for the perturbed situation deserves attention as we have already explained in @xcite , and we are going to study in details in the sequel .
the present paper is then the second part of the previous one @xcite , in which the harmonic analysis aspects of the model were intensively studied .
the reader is also referred to @xcite for some very interesting pivotal amenable models .
we are going to investigate the physical application to the bec of the very fascinating picture which arises from the previous mathematical investigation .
the physical underlying model concerns bardeen
cooper boson pairs in networks describing arrays of josephson junctions .
the formal hamiltonian is the quartic bose
hubbard hamiltonian , given on a generic network @xmath4 by @xmath5 here , @xmath6 denotes the set of the vertices of the network @xmath4 , @xmath7 is the bosonic creator , and @xmath8 the number operator on the site @xmath9 ( cf .
@xcite ) . finally , @xmath10 is the adjacency operator whose matrix element @xmath11 in the place @xmath12 is the number of the edges connecting the site @xmath13 with the site @xmath14 ( in particular it is hermitian ) .
the reader is referred to the seminal paper @xcite in which the theory of the superconductivity ( called bcs theory in honor of the authors ) has been firstly discussed .
it has been argued in @xcite that , in the case when @xmath15 and @xmath16 are negligible with respect to @xmath17 , the hopping term dominates the physics of the system . after neglecting such terms , putting for the resulting effective coupling constant @xmath18 , in principle different from @xmath17 , @xmath19 , and finally passing to the one
particle space , the model under consideration becomes the so called pure hopping ones described by the _
pure hopping one particle hamiltonian _ which assumes the form @xmath20 here , @xmath10 is simply the adjacency of the fixed graph @xmath4 , acting on the hilbert space @xmath21 .
the constant @xmath22 is added just in order to have @xmath23 and does not affect the analysis at all .
the reader is referred to @xcite for the systematic investigation of the mathematical properties arising in non amenable cases considered here consisting of negligible additive perturbations of cayley trees .
several interesting amenable models are treated in details in @xcite .
the reader is also referred to @xcite for some promising experiments for the pure hopping model on the comb and star graphs ( see fig .
[ figd ] ) pointing out an enhanced current at low temperature which might be explained by condensation phenomena .
one of the first attempts to investigate the bec on non homogeneous amenable graphs , such as the comb graphs , was made in @xcite . in that paper , it was pointed out the appearance of a _ hidden spectrum _ in the low energy part of the spectrum , which is responsible of the finiteness of the critical density . in addition , the behavior of the wave function of the ground state , describing the spatial density of the condensate , was also computed .
some spectral properties of the comb and the star graph were also investigated in @xcite in connection with the various notions of independence in quantum probability . in that paper , it was noted the possible connection between such spectral properties and the bec . the systematic investigation of the bec for the pure hopping model on a wide class of amenable networks obtained by negligible perturbations of periodic graphs , has been started in @xcite .
the emerging results are quite surprising .
first of all , the appearance of the hidden spectrum was proven for most of the graphs under consideration .
this is due to the combination of two opposite phenomena arising from the perturbation .
if the perturbation is sufficiently large ( surprisingly , in some examples , it is enough that the perturbation is indeed finite ) , the norm @xmath24 of the adjacency of the perturbed graph becomes larger than the analogous one @xmath22 of the unperturbed adjacency . on the other hand ,
as the perturbation is sufficiently small ( i.e. zero density ) , the part of the spectrum @xmath25 in the segment @xmath26 $ ] does not contribute to the density of the states .
this allows to compute the critical density @xmath27 at the inverse temperature @xmath28 for the perturbed model by using the integrated density of the states @xmath29 of the unperturbed one , @xmath30 the resulting effect of the perturbed model exhibiting the hidden spectrum ( i.e. when @xmath31 ) is that the critical density is always finite .
the bridge between the mathematics and physics can be easily explained as follows .
consider the bose gibbs occupation number ( at the inverse temperature @xmath32 and chemical potential @xmath33 ) at low energies for the hamiltonian .
after using taylor expansion , one heuristically gets @xmath34^{-1}=[\b((\|a\|-\m)i - a)]^{-1 } \equiv \frac1{\b}r_a(\|a\|-\m)\,.\ ] ] then the study of the bec is reduced to the investigation of the spectral properties of the resolvent @xmath35 , for @xmath36 , the latter being a very familiar object for mathematicians . due to the particular form ,
as the bec phenomena are connected with the spectral properties for small energies of the hamiltonian , we can reduce the matter to the investigation of the spectral properties of the adjacency operator @xmath10 of the network close to @xmath22 .
another relevant fact connected with the introduction of the perturbation , and thus with the lack of homogeneity , is the possible change of the transience / recurrence character ( cf .
@xcite , section 6 ) of the adjacency operator .
it has to do with the possibility to construct locally normal states exhibiting bec .
as explained in @xcite , the last relevant fact is the investigation of the shape of wave function of the ground state of the model , describing the spatial distribution of the condensate on the network in the ground state of the hamiltonian . from the mathematical viewpoint , this is nothing else the perron frobenius generalized eigenvector of the adjacency ( cf .
it appears then clear that the physical and the mathematical aspects of the topological model based on the pure hopping hamiltonian are strongly related .
summarizing , the new and very surprising phenomena are the following .
first the finiteness of the critical density is connected with the appearance of the _ hidden spectrum _ above the zero of the energy .
it can be computed in a quite simple way for many amenable and non amenable models by using the _ secular equation _ , see section 6 of @xcite and section 3 of @xcite .
this leads to the fact that the bec can appear also in low dimensional cases @xmath37 .
in addition , not for all the models with finite critical density , it is possible to exhibit locally normal states ( those for which the local particle density is finite in the infinite volume limit ) describing bec . due to non homogeneity , particles condensate also in the configuration space , that is the system undergoes a `` dimensional transition '' .
this phenomenon is connected with the _ transience / recurrence _ character of the adjacency operator of the network ( the last describing the hamiltonian of the system ) .
another new aspect is the difference between the geometrical dimension of the network and the growing of the wave function of the ground state .
the last is called _ perron frobenius dimension _
@xmath38 for volume growing networks and is computed through the behavior of @xmath39norms of the finite volume perron frobenius eigenvectors , all normalized to @xmath40 at a common root . if the geometrical dimension @xmath41 is greater than @xmath38 , then it is impossible to exhibit locally normal states describing the bec , whose mean density of the particles is greater than the critical one . surprisingly , there are non homogeneous networks for which @xmath42 and @xmath43 ( i.e. the comb graph @xmath44 , see fig . [ figg ] ) .
in addition , there are also simple non homogeneous graphs exhibiting infinite critical density , but for which it is yet possible to exhibit locally normal states describing bec .
such a very intriguing situation is summarized as follows .
is the comb shaped graph whose base
point is @xmath10 , @xmath2 is the critical density , r / t denotes the recurrence / transience character of the adjacency , bec ( @xmath45bec ) denotes the existence of locally normal states exhibiting bec ( exhibiting bec at any mean density @xmath46 ) . ] [ cols="<,<,^,^,^,^,^,^",options="header " , ] a similar situation also happens for some non amenable networks .
the networks under consideration in the present paper are density zero additive perturbations of exponentially growing graphs made of homogeneous cayley trees , see fig .
the models treated in the present paper are the perturbations @xmath47 , @xmath48 , and @xmath49 , of the homogeneous cayley tree @xmath50 along a subtree isomorphic to @xmath51 , and @xmath52 respectively , see below , see also @xcite for the precise definition of such models and for further details .
quite surprisingly , the emerging mathematical situation described in @xcite for such non amenable examples , is similar to that for the amenable graphs treated in @xcite .
namely , we still find hidden spectrum , and the transience / recurrence character is determined by the transience / recurrence character of the negligible subgraph supporting the perturbation . finally , the infinite volume perron frobenius eigenvector is exponentially decreasing far from the base point supporting the perturbation .
namely , concerning the wave function of the ground state , we have precisely the same situation arising for the amenable situations treated in @xcite ( see fig . [ figd ] ) . the first part of the present paper is devoted to describe general properties of ( bounded ) hamiltonians on networks equipped with an exhaustion such that the density of the eigenvalues of the underlying hamiltonian converges in the infinite volume limit to a cumulative function called _ integrated density of the states_. it includes as particular cases those arising by negligible perturbations of cayley trees as shown in @xcite
. then we describe the general statistical properties of such models .
in particular , a careful definition of the mean density and the possible amount of the bec condensate should be done , because of the non negligible effects of the boundary terms due to the possible lack of amenability .
we pass to describe the existence of the dynamics associated to the pure hopping hamiltonian for the perturbed cayley trees .
this is a delicate issue which is done in section [ dynkmss ] for all the cases under consideration .
after that , by using the results in @xcite , in the transient case we write down the formula ( cf . ) for the natural class of states describing bec ( i.e. presenting a non negligible amount of condensate in the ground state wave function ) .
such states are kms ( i.e. equilibrium ) states for the dynamics previously investigated . as for the previous amenable models treated in @xcite , such states are infinite volume kms states whose mean density coincides with the critical one .
this simply means that the condensate is allocated on a subgraph whose volume growth is negligible with respect to that of the unperturbed one .
we end the paper by investigating the infinite volume behavior of finite volume gibbs states . due to the lack of amenability
, such an investigation presents some very delicate technical difficulties arising by the unavoidable boundary effects .
modulo such difficulties , the emerging situation seems to be in accordance with that described in @xcite , and summarized as follows .
first , relatively to the general recurrent situation , it is indeed impossible to construct locally normal states exhibiting bec at all , even if the critical density is finite .
in addition , in the all transient cases under consideration , it seems to be impossible to exhibit locally normal states describing bec at the mean density @xmath53 , @xmath54 .
finally , in the condensation regime where for the mean density @xmath1 of all such states @xmath55 , a careful choice of the sequence of the finite volume chemical potentials @xmath56 should be done . by considering the geometry of the graph @xmath4 and the @xmath39behavior of the ground state wave function @xmath57
, such a choice corresponds to the local density of the condensate @xmath58 in the infinite volume limit @xmath59 , see proposition [ 3a3c ] .
this is nothing but the naive explanation of the fact that @xmath55 for the locally normal states describing the condensation regime in all the situations under consideration .
a _ graph _ ( called also a _ network _ ) @xmath60 is a collection @xmath61 of objects , called _ vertices _ , and a collection @xmath62 of unordered lines connecting vertices , called _
edges_. usually , the edges describe the physical interaction between vertices . in our model describing the pure hopping interaction , multiple edges as well as self interactions , are also allowed .
let @xmath63 be the collection of all the edges connecting @xmath64 with @xmath65 .
we have @xmath66 .
let us denote by @xmath67_{x , y\in x}$ ] , @xmath68 , the _ adjacency matrix _ of @xmath69 : @xmath70 as explained in @xcite , all the geometric properties of @xmath69 are encoded in @xmath10 .
for example , @xmath69 is connected if and only if @xmath10 is irreducible . setting @xmath71 and @xmath72
we have @xmath73 , that is @xmath10 is bounded if and only if @xmath69 has uniformly bounded degree . for connected networks ( or on each connected component ) we can define the standard distance @xmath74 @xmath75 being the length of the path @xmath76 ( i.e. the number of the edges in @xmath76 ) .
the adjacency operator considerably differs from the laplacian @xmath77 , whenever @xmath69 is non homogeneous . on the other hand ,
it was proven in theorem 7.6 of @xcite that the value of the critical density does not change under negligible perturbations on a fixed homogeneous graph , if the hamiltonian of the model is the ( opposite of the ) laplacian .
thus , the main object for the investigation of the bec for the pure hopping model of inhomogeneous networks will be the adjacency operator . in the present paper , all the graphs are connected , countable and with uniformly bounded degree .
in addition , we deal only with bounded operators acting on @xmath78 if it is not otherwise specified .
let @xmath79 be a closed operator acting on @xmath78 , and @xmath80 the resolvent set of @xmath79 .
as usual , @xmath81 denotes the _ resolvent _ of @xmath79 . fix a bounded matrix with positive entries @xmath79 acting on @xmath78 .
such an operator is called _ positive preserving _ as it preserves the elements of @xmath78 with positive entries .
if @xmath79 is selfadjoint , @xmath79 is _ positive _ if @xmath82 for each @xmath83 .
a sequence @xmath84 is called a ( generalized ) _ perron frobenius eigenvector _ if it has positive entries and @xmath85 if such a vector is normalizable ( i.e. if it belongs to @xmath78 ) it is a standard @xmath78vector , otherwise it is a weight . in all the cases we simply call perron
frobenius eigenvector such a sequence by dropping the word generalized .
suppose for simplicity that @xmath79 is selfadjoint .
it is said to be _ recurrent _ if @xmath86 otherwise @xmath79 is said to be _ transient_. it is shown in @xcite , section 6 , that the recurrence / transience character of @xmath79 does not depend on the base point chosen in computing the limit in .
the perron
frobenius eigenvector is unique up to a multiplicative constant , if @xmath69 is finite or when @xmath79 is recurrent , see e.g. @xcite . in general
, it is not unique , see the end of the present section for the case relative to the cayley trees .
we say that an operator @xmath79 acting on @xmath78 has _ finite propagation _ if there exists a constant @xmath87 such that , for any @xmath88 , the support of @xmath89 is contained in the ( closed ) ball
@xmath90 centered in @xmath64 and with radius @xmath91 .
it is easy to show that for the adjacency operator if @xmath10 on @xmath69 , then @xmath92 has propagation @xmath93 for any integer @xmath94 .
let @xmath69 be an infinitely extended graph with an exhaustion @xmath95 which is kept fixed during the analysis .
the definition of the _ the integrated density of the states _ ( cf .
@xcite ) of a selfadjoint operator @xmath96 given in @xcite works for all the situations for which it is meaningful , and then is useful for the purposes of the present work . indeed ,
consider on @xmath97 the state @xmath98 @xmath99 being the selfadjoint projection onto @xmath100 .
define for a bounded operator @xmath79 , @xmath101 where the domain @xmath102 is precisely the linear subspace of @xmath97 for which the limit in exists .
in addition , define for a bounded selfadjoint operator @xmath79 , @xmath103 provided such a limit exists .
the domain @xmath104 is precisely the linear subspace of the unital @xmath105algebra @xmath106 generated by @xmath79 , for which the limit in exists . notice that the definition of @xmath107 , @xmath108 might depend on the chosen exhaustion which we keep fixed during the analysis .
suppose now that @xmath109 .
then it provides a state on @xmath110 and , by the riesz
markov theorem , a borel probability measure @xmath111 on @xmath112 .
thus , there exists a right continuous , increasing , positive function @xmath113 satisfying @xmath114 such that @xmath115)=f_b(x)\,,\quad x\in\br\,.\ ] ] the previous described cumulative function @xmath113 is precisely the integrated density of the states ( ids for short ) associated to @xmath79 , provided it exists for the chosen exhaustion . when the graph is amenable and the operator @xmath79 is of finite propagation , the definition and some of the main facts relative to the ids considerably simplify as the boundary effects play no role in the infinite volume limit , see theorem 2.1 of @xcite .
consider the graph @xmath116 such that @xmath117 , both equipped with exhaustions @xmath118 , @xmath119 such that @xmath120 , @xmath121 .
the graph @xmath116 is a _
negligible _ or _ density zero perturbation _ of @xmath69 if it differs from @xmath69 by a number of edges for which @xmath122 where @xmath123 denotes the symmetric difference . to simplify matters ,
we consider only perturbations involving edges , the more general case involving also vertices can be treated analogously , see @xcite .
suppose that @xmath124 that is @xmath125 admits the ids .
( proposition 1.3 of @xcite ) [ density0 ] let @xmath116 be a negligible perturbation of the tree @xmath69
. then @xmath126 , and @xmath127 if @xmath128 is sufficiently large , then it is possible to express the resolvent of @xmath116 in terms of the resolvent of @xmath69 by the krein formula .
indeed , fix @xmath69 as the reference graph and define @xmath129 , @xmath130 , where @xmath131 is the perturbation , which eventually acts on @xmath132 .
put , for @xmath133 , @xmath134 where @xmath135 is the selfadjoint projection onto @xmath136 .
the krein formula assumes the form @xmath137 the resolvent formula holds true for @xmath138 , and extends to any simply connected subset containing the point at infinity of @xmath139 made of the @xmath140 for which @xmath141 is invertible .
conversely , suppose that @xmath141 is not invertible for some positive number @xmath142 .
then the norm @xmath143 of the perturbed graph @xmath116 is greater than that @xmath144 of the unperturbed one @xmath69 .
it can be computed as @xmath145 fix now our set up which is a graph @xmath69 equipped with an exhaustion @xmath118 such that @xmath125 admits the ids .
let @xmath116 be a negligible additive perturbation of @xmath69 as explained above .
first of all , note that proposition [ density0 ] assures that @xmath146 admits the ids w.r.t .
the exhaustion @xmath119 .
in addition , we obviously have @xmath147 .
define @xmath148 denote @xmath149 , @xmath150 , where @xmath151 is the ids of the operator @xmath79 .
( corollary 2.6 of @xcite ) [ 6 ] we have @xmath152 the ids of @xmath153 and @xmath154 exist by hypothesis and by proposition [ density0 ] , respectively .
we get , again by proposition [ density0 ] , @xmath155 this leads to the assertion .
suppose that @xmath156 .
then @xmath157 in .
it has a very precise physical meaning as an effective chemical potential ( cf . ) , see proposition 7.1 of @xcite .
thus , the part of the spectrum @xmath158 $ ] does not contribute to the ids of @xmath146 .
equivalently , the perturbed pure hopping hamiltonian @xmath154 presents the hidden spectrum , see proposition [ hyspe ] .
the appearance of the hidden spectrum is then the combination of two different effects : the perturbation should be of density zero in order not to affect the ids ( proposition [ density0 ] ) , but it should be sufficiently large in order to increase the norm of the perturbed adjacency
. our framework will be the investigation of the occurrence of the bec for perturbed cayley trees , whose mathematical aspects are investigated in detail in @xcite .
namely , we consider the graphs @xmath47 with @xmath159 , and @xmath160 where @xmath161 + 1\ ] ] ( @xmath162 $ ] stands for the integer part of a positive number @xmath64 ) , and @xmath163 with @xmath164 .
such networks consist of the cayley tree @xmath50 of order @xmath165 , suitably perturbed along subtrees as explained in fig .
. one can prove ( cf .
lemma 4.1 of @xcite , together with proposition 2.2 of @xcite ) that the adjacency of @xmath50 admits the ids w.r.t .
the standard exhaustion made of all the balls of radius @xmath166 centered in a fixed root . in the above notations , the laplace transform @xmath167 $ ] ( or equivalently the _ one particle partition function _ in physical language ) of @xmath168 is given by @xmath169 this , together with proposition [ 6 ] , leads to @xmath170 , where @xmath116 is @xmath47 or @xmath163 .
as @xmath171 , they all present hidden spectrum .
we also have that the graphs @xmath47 and @xmath163 are all transient except @xmath172 , see @xcite .
we briefly report the description of the perron frobenius eigenvectors @xmath173 given in @xcite for the graphs under consideration .
let @xmath174 be the base point of the perturbed graph @xmath116 under consideration , supporting the perturbation .
the exhaustion @xmath95 has an obvious meaning by considering the part @xmath175 of @xmath174 in @xmath176 , the last being the ball of radius @xmath166 of @xmath177 , centered in a root @xmath178 , kept fixed during the analysis .
it was shown in @xcite that @xmath179 here , @xmath180 is the function given , @xmath181 is the distance given in , @xmath182 is the unique point on @xmath174 such that @xmath183 ( cf .
lemma 4.2 of @xcite ) , and finally @xmath184 is the perron
frobenius eigenvector of @xmath185 ( @xmath186 given by @xmath187 described in theorems 5.2 and 6.3 of @xcite for @xmath163 and @xmath47 , respectively . concerning the behavior of the @xmath39norms of the restrictions of @xmath173 , we have the following [ 3a3c ] for the perron
frobenius eigenvector @xmath173 described by and , we get @xmath188 first of all we note that , in all the situations , @xmath189 this leads to @xmath190 by taking into account the form of @xmath184 , we easily obtain the following estimates .
concerning @xmath49 we have @xmath191 , which leads to @xmath192 for the case @xmath47 , @xmath193 leads to @xmath194 , identically on @xmath174 . concerning the case @xmath195 ,
consider the function @xmath196 it is not difficult to see that @xmath197 and @xmath198 , @xmath199 .
this leads to @xmath200 whenever @xmath201 .
we finally get @xmath202 we end the present section by noticing that it is easily seen that the perron frobenius eigenvector is no longer unique , in general for infinitely extended networks , and in particular for the situation considered here . in fact , on @xmath177 , @xmath203 , @xmath204 , @xmath205 are perron
frobenius eigenvectors for the adjacency , for any fixed root @xmath206 .
the same happens for @xmath47 , @xmath195 where in it is enough to choose @xmath207 , @xmath208 is any fixed root , where @xmath209 the choice of the normalization at @xmath40 on a fixed root ( which is denoted by @xmath178 in our framework ) selects in a unique way the perron
frobenius vector among the class previously described . for the networks @xmath210 and @xmath172 , the perron frobenius eigenvector for the adjacency is unique ( up to a multiplicative constant ) thanks to the fact that the adjacency is recurrent .
it is expected that it is unique also for @xmath49 , thanks to the fact that @xmath211 has a unique end at infinity .
in order to investigate the statistical properties of the pure hopping model describing bardeen cooper pairs on arrays of josephson junctions on non homogeneous networks we report some standard notions useful in the sequel .
let @xmath212 be a dynamical system consisting of a non abelian @xmath105algebra and a one parameter group of @xmath213automorphism @xmath214 . .
] the state @xmath215 on the @xmath216algebra @xmath217 satisfies the kubo
martin schwinger ( kms for short ) boundary condition at inverse temperature @xmath218 w.r.t the group of automorphisms @xmath219 if * for every @xmath220 , @xmath221 , @xmath222 are continuous ; * for each @xmath223 , @xmath224 where
stands for the fourier transform and @xmath226 is the space of the infinitely often differentiable , compactly supported functions on @xmath227 .
the following facts are well known .
first , a kms is automatically invariant w.r.t .
the automorphism group @xmath228 .
second , the cyclic vector @xmath229 of the gelfand naimark
segal ( gns for short ) quadruple @xmath230 is also separating for @xmath231 . denote @xmath232 its modular group . according to the definition of kms boundary condition ,
we have @xmath233 we refer the reader to @xcite , section 5.3 for equivalent formulations of the kms condition , proofs and further details . the @xmath105algebras considered here are those arising from the _ canonical commutation relations _
( ccr for short ) .
namely , let @xmath234 be a pre hilbert space equipped with the non degenerate inner product @xmath235 . denote @xmath236 its completion . consider the following ( formal ) relations , @xmath237 it is well known that the relations can not be realized by bounded operators .
a standard way to realize them is to look at the _ symmetric fock space _
@xmath238 on which the annihilators @xmath239 and creators @xmath240 naturally act as unbounded closed , adjoint each other ( i.e. @xmath241 ) operators .
this concrete representation of the ccr is called the _
fock representation_. an equivalent description for the ccr is to put on @xmath238 , @xmath242 and define the _ weyl operators _ @xmath243 .
the weyl operators are unitary and satisfy the rule @xmath244 it is well known that the representation on @xmath238 of @xmath245 described above , is faithful .
we adopt the _ weyl algebra _ as the definition of the ccr algebra , the last being the abstract @xmath105algebra @xmath245 generated by the operators @xmath246 satisfying the relations .
sometimes , it is also useful to compute directly the expectation values of combinations of products of annihilators and creators ( equivalently field operators ) in a given state .
let @xmath247 be a self adjoint operator acting on @xmath236 .
suppose that @xmath248 .
then the one parameter group of automorphisms @xmath249 defines a one parameter group of @xmath213automorphisms @xmath228 of @xmath245 by putting @xmath250 .
the latter is called the one parameter group of _ bogoliubov automorphisms _ generated by @xmath251 .
a representation @xmath76 of the weyl algebra @xmath245 is _ regular _ if the unitary group @xmath252 is continuous in the strong operator topology , for any @xmath253 .
a state @xmath215 on @xmath245 is regular if the associated gns representation is regular .
the _ quasi free _ states on the weyl algebra are those of interest for our purposes .
such states @xmath0 are uniquely determined by the two point functions @xmath254 , @xmath255 .
a standard textbook for ccr is @xcite ( cf .
section 5.2 ) to which the reader is referred for proofs , literature and further details .
let @xmath4 be any graph . to simplify notations
, we denote the sums on the vertices @xmath256 of subgraphs @xmath257 , and of @xmath6 itself , directly by sums on @xmath174 or @xmath4 , respectively .
consider a subspace @xmath234 of @xmath258 , which contains the indicator functions @xmath259 .
a representation @xmath76 of the weyl algebra @xmath245 is said to be _ locally normal ( w.r.t . the fock representation ) _ if @xmath260 is quasi equivalent to the fock representation of @xmath261 .
a state on @xmath245 is said to be locally normal if the associated gns representation is locally normal .
a locally normal state @xmath215 does have finite local density @xmath262 even if the mean density might be infinite ( e.g. @xmath263 ) .
let @xmath264 be an _ exhaustion _ of @xmath4 , that is a sequence of finite regions invading the graph @xmath4 , together with a sequence of states @xmath265 on @xmath266 .
( lemma 3.2 of @xcite ) [ add2 ] suppose that @xmath267 for some @xmath268 .
then @xmath269 does not define any locally normal state on @xmath270 , where @xmath271 is any subspace containing the finite supported sequences .
our purposes concerns the pure hopping model on the networks @xmath47 and @xmath49 or , more generally on a uniformly bounded connected network @xmath69 equipped with an exhaustion @xmath272 for which the adjacency admits the ids ( cf .
@xcite ) . as previously explained ,
all the physical properties of the emerging quasi free multi particle model are encoded in the one particle pure hopping hamiltonian .
namely , the graph @xmath69 will be one of such networks , @xmath116 a density zero additive perturbation of @xmath69 as described above , equipped with exhaustions @xmath118 , @xmath119 .
we denote by @xmath4 one of such graphs , equipped with the finite volume exhaustion @xmath95 .
the pure hopping hamiltonian , after normalizing such that the bottom of the energy is zero , assumes the form @xmath273 the finite volume hamiltonians will be @xmath274 acting on @xmath275 , @xmath276 being the selfadjoint projection onto @xmath275 .
fix a general bounded positive hamiltonian @xmath277 admitting the ids for a fixed exhaustion as for the pure hopping one .
such that @xmath278 is the bottom of the spectrum @xmath279 of the hamiltonian @xmath247 . ]
consider the finite volume densities of the states @xmath280 relative to @xmath281 , and that @xmath282 relative to @xmath247 .
define @xmath283 where the limit in the first equation exists by lemma 3.4 of @xcite , and that in the second one is meaningful directly by the definition of the ids .
we always get @xmath284 .
[ hyspe ] if @xmath285 we say that there is a _
( low energy ) hidden spectrum _ , see e.g. @xcite . the following assertions can be shown for all the situations under consideration .
first , @xmath247 is continuous in @xmath178 ( see also @xcite , proposition 3.9 ) in all the situation considered here .
second , in the unperturbed situation the hidden spectrum does not occur ( by direct inspection of for @xmath177 , or by theorem 5.2 of @xcite in the case of periodic amenable graphs ) .
if @xmath116 is any perturbation of @xmath69 , we have for the @xmath286 appearing in , @xmath287 the opposite of @xmath288 then plays the role of a chemical potential in determining the critical density of the perturbed graphs under consideration .
indeed , consider the formula for the mean density at the inverse temperature @xmath28 of the particles ( cf .
@xcite ) for the model with hamiltonian @xmath23 , and chemical potential @xmath289 , @xmath290 and the corresponding _ critical density _ @xmath291 by taking into account proposition [ 6 ] , we get @xmath292 this means @xmath293 .
in addition , if there is hidden spectrum , @xmath294 is always finite and @xmath295 .
it can happen that , after perturbing a graph whose critical density is infinite , we get finite critical density .
as for the networks under consideration in the present paper , the unperturbed critical density is always finite ( cf .
@xcite ) , the only effect of the appearance of the hidden spectrum on the critical density is that it decreases w.r.t . the unperturbed one .
it is well known that the condensation regime is described by @xmath296 , see e.g. section 5.2 of @xcite , or section 3 of @xcite .
indeed , one starts from the gibbs grand
canonical ensemble at finite volume @xmath297 and fixed density .
one fixes an inverse temperature @xmath32 together with the particle density @xmath45 , and determines the finite volume chemical potential @xmath298 by solving @xmath299 see e.g. fig .
[ figg1 ] . , instead of the chemical potential . ]
the infinite volume limit can be investigated by means of the reference exhaustion @xmath95 . to take into account also the very different situation appearing in non homogeneous situation , we start by fixing any general sequence of chemical potential @xmath300 which we can suppose to converge ( eventually passing to a subsequence ) to some @xmath301 . in the case
when such a sequence is recovered by using , we put @xmath302 . since @xmath303
, we get @xmath289 .
the finite volume state with chemical density @xmath304 at @xmath28 is described by the two point function @xmath305 where @xmath306 .
thermodynamical states are then described as limits of the finite volume states as above .
concerning the infinite volume limit of the finite volume density , we report a result ( cf .
@xcite ) , whose proof relies upon the analogous one , proposition 3.8 of @xcite .
it describes in a rigorous way how it is possible to define the amount of the condensate in the infinite volume limit for an arbitrary sequence of finite volume chemical potential @xmath307 . due to the general character of such a result
, we suppose that @xmath247 is a bounded positive hamiltonian on a network @xmath4 equipped with an exhaustion @xmath308 such that @xmath247 admits the ids @xmath282 ( in the situation under consideration @xmath309 ) .
denote as usual @xmath310 , and @xmath311 . to avoid the almost trivial situation @xmath33 ,
we restrict the matter to the condensation regime @xmath296 .
[ ennz ] under the conditions listed above , the following assertions hold true for each sequence @xmath312 satisfying @xmath307 . *
suppose that @xmath247 admits low energy hidden spectrum .
then for each continuous mollifier @xmath29 which is identically @xmath178 in a neighborhood of @xmath178 , and @xmath40 in a neighborhood of @xmath313 , we have @xmath314 * suppose that @xmath315 . consider a sequence @xmath316 of increasing continuous mollifiers , all vanishing in a neighborhood of @xmath178 and converging almost everywhere ( w.r.t the measure determined by @xmath282 ) to @xmath40 .
then @xmath317 in addition , if @xmath318 , one can replace for the continuous mollifiers @xmath316 , the condition of monotony with uniform boundedness .
we start by noticing that @xmath319 , and all the @xmath320 are continuous in a common neighborhood of all the supports of the measures determined by the cumulative functions @xmath321 and @xmath322 .
\(i ) by the existence of the hidden spectrum , we have @xmath323 since @xmath324 , uniformly on the support of @xmath29 , and the second summand goes to zero because @xmath325 for each continuous function @xmath326 in a common neighborhood of all the supports of the involved measures @xmath327 and @xmath328 .
\(ii ) the same computation as before leads for each @xmath329 , @xmath330 the proofs follows either by the monotone or the dominated convergence theorem , as @xmath331 .
thanks to the above result , the quantity @xmath332 is well defined and independent of the particular choice of the mollifiers as above .
such a quantity describes the amount of the condensate in the ground state . indeed ,
if the sequence of chemical potentials are obtained through by fixing the mean density @xmath45 , we have @xmath333 @xmath33 if and only if @xmath334 . for any @xmath32 , @xmath33
, the sequence converges pointwise to a state @xmath0 , whose two point function is given by @xmath335 moreover , the _ density _ @xmath1 of the state @xmath0 , defined by @xmath336 satisfies @xmath337 .
the transience of @xmath10 is a necessary and sufficient condition ( see proposition [ add4 ] and remark [ suftra ] ) for the existence of locally normal states on @xmath245 describing condensation phenomena , that is when for @xmath296 , for the pure hopping model .
observe that @xmath338 if and only if @xmath339 . indeed , @xmath340 whenever @xmath33 , hence @xmath341 . as a consequence , @xmath342 , i.e. @xmath53 .
conversely , @xmath53 implies @xmath296 , hence @xmath343 .
observe that @xmath338 can be obtained only if @xmath344 .
indeed , @xmath345 is a necessary condition for the occurrence of bec .
observe that @xmath346 can occur also when the infinite volume limit of the two point function leads to non locally normal states i.e. @xmath347 .
observe that it is possible to have locally normal states exhibiting a non trivial amount of condensate even if @xmath348 , see section 10 of @xcite , or sections [ dynkmss ] , [ dynkmss1 ] below . to simplify the notations , from now on we put @xmath349 throughout the paper if it is not otherwise specified .
we also put @xmath350 , @xmath351 , and finally @xmath352 , @xmath247 being the hamiltonian of the model .
let @xmath116 be any of the perturbed graphs under investigation . in order to study states describing local properties , and in particular
those which are locally normal , we start to consider a subspace @xmath353 which contains all the canonical elements @xmath354 , and which is stable under the dynamics generated by @xmath355 . to this end
, we define @xmath356 for the algebraic span ( no closure in @xmath357 ) of all the elements of the form @xmath358 . by construction @xmath359 . for our aims , we need also that the pairing @xmath360 is well defined , when @xmath173 is the perron frobenius vector , described in @xcite , for the adjacency matrix @xmath146 in each of the cases under consideration , see and . for @xmath361
we also define @xmath362 by putting @xmath363 , @xmath364 .
[ rispf ] let @xmath365 be a function analytic in a neighborhood of @xmath366 such that @xmath367 , @xmath121
. then @xmath368 .
as the adjacency matrix has positive entries , all the involved series have positive entries .
thus , by the monotone convergence theorem we get @xmath369 the following functions , defined in @xmath370 $ ] , @xmath371 @xmath372 are useful in the sequel .
if @xmath373 , it was shown in @xcite that , for the situation under consideration , the operator @xmath374 in describing the resolvent of @xmath146 through the krein formula assumes the form @xmath375 eventually acting on @xmath376 . on which is allocated the perturbation . ] in the cases under consideration when @xmath116 is @xmath47 , @xmath49 , @xmath377 ( where @xmath378 ) and @xmath379 , respectively . in all such situations ,
is proportional to the convolution by , and for @xmath380 $ ] it assumes the form @xmath381 provided at least that the series in converges pointwise . here ,
@xmath181 is the distance function given in , and @xmath382 and @xmath383 are the analytic continuations in @xmath384 $ ] of the corresponding objects given in and , respectively .
it was shown in @xcite that is meaningful and defines an element of @xmath385 , at least when @xmath386 .
now we show that the series in converges and defines an element in @xmath385 for @xmath387 such that @xmath388 is sufficiently large .
[ rispf1 ] let @xmath382 and @xmath383 as above , and @xmath389
. then @xmath390 put @xmath391 . disregarding the constants and the quantities
whose modulus is fixed , we have to maximize and minimize on circles of fixed radii less than @xmath40 , the functions @xmath392 where @xmath184 is in the open punctured disk @xmath393 of radius @xmath40 centered in the origin of the complex plane , and @xmath394 , see fig . [ figab ] .
first of all we note that , for the analytic continuation @xmath395 of @xmath396 in the cut plane @xmath397\cup[1,+\infty)\big)$ ] , we have @xmath398 after some tedious computations , the functions @xmath29 , @xmath4 are expressed by functions @xmath326 , @xmath399 on the unit circle , given by @xmath400 here @xmath401 @xmath402 is fixed , and by , we can reduce the matter to @xmath403 .
after some straightforward computations , we get that @xmath404 and @xmath405 are never zero for @xmath406 so the the functions attain the extrema either in @xmath178 or in @xmath407 .
we get for @xmath326 , @xmath408 because @xmath409 .
concerning @xmath399 we have , @xmath410 collecting together , we get with @xmath411 , @xmath412 and the proof follows
. denote @xmath413 the closed disk centered in the origin of radius @xmath91 .
[ koro ] there exists @xmath389 such that for @xmath414 in , we get @xmath415 for each @xmath416 . thus , for @xmath416 , the resolvent of the adjacency of @xmath116 can be written by the neumann series @xmath417 let @xmath418 be the decomposition of @xmath419 in positive elements .
one has @xmath420 , @xmath421 . by taking into account and lemma [ rispf1 ]
, we get @xmath422 as @xmath423 is continuous and decreasing ( cf .
lemma 3.1 in @xcite ) with @xmath424 , there exists @xmath425 such that @xmath426 .
collecting together , we have in the complement of @xmath427 , @xmath428 in any simply connected open subset of @xmath139 containing the point at infinity for which @xmath429 is invertible , @xmath430 assumes the form @xmath431 and the proof follows from the first half , by using the neumann expansion of @xmath432 .
consider the function @xmath433 it is bounded on @xmath434 .
in addition , @xmath435 . it provides the comparison between the resolvent of @xmath10 and the functional calculus of @xmath436 associated to the bose
gibbs occupation number for the pure hopping hamiltonian .
[ add4 ] under the above notations , the following assertions hold true .
* @xmath146 is transient if and only if @xmath437 .
* for @xmath438 we have @xmath439 .
it is enough to prove the assertions for the generators @xmath440 of @xmath234 .
@xmath441 by using the splitting described by we get that @xmath184 is in @xmath442 if and only if @xmath184 is in @xmath443 .
let @xmath14 be any element of @xmath116 .
obviously , @xmath444 and , by hypothesis , @xmath445 .
so @xmath446 and @xmath146 is transient . for the reverse implication ,
we compute for @xmath447 , @xmath448 which is finite if @xmath146 is transient . thus , @xmath449
. @xmath450 fix @xmath451 , where @xmath91 is the number appearing in proposition , together with the circle @xmath452 of radius @xmath453 centered in the origin . by using the neumann expansion of @xmath432 on @xmath454 and the fact that @xmath455 itself is the convolution of the function @xmath456 ( cf .
( 7.6 ) in @xcite ) , we show that the corresponding resolvent @xmath430 is a combinations of convolutions and infinite ( absolutely converging ) sums of powers of the convolution by the function @xmath456 , see and .
let @xmath454 be counterclockwise oriented .
thus , thanks to proposition as explained above , together with lemma [ rispf ] , we get @xmath457 now we pass to describe a class of locally normal states exhibiting bec , which are indeed kms for the natural dynamics generated on @xmath245 , @xmath234 given in , by the bogoliubov automorphisms whose generator is the ( second quantization of ) the pure hopping hamiltonian . to this end , we consider the dynamical system @xmath458 , where @xmath245 is the @xmath105algebra made of the weyl ccr algebra on @xmath234 , and @xmath228 is generated by the bogoliubov automorphisms @xmath459 , @xmath460 which are well defined as @xmath461 by construction . by taking into account proposition [ add2 ] , the necessary condition for the existence of locally normal states exhibiting bec is the transience of @xmath146 . for the pure hopping model ,
the transience of the adjacency is indeed also sufficient as we are going to see . for the rest of the present section we limit our analysis when @xmath116 is either @xmath49 , @xmath462 , or @xmath47 where @xmath463 , and @xmath160 with
@xmath464 given in , see theorems 4.1 , 5.3 , 6.5 of @xcite , together with proposition [ add2 ] .
fix @xmath465 and define states on the weyl ccr algebra @xmath245 uniquely determined by the two point function @xmath466 the states @xmath467 are well defined ( cf .
@xcite , section 5.2 ) because of proposition [ add2 ] . in the case of quasi
free states on @xmath245 , it is also customary to provide the description directly in terms of their values on the weyl unitaries .
we get ( cf . section 5.2.5 of @xcite ) , @xmath468 note that for quasi free states , and then in our situation , the kms condition can be checked directly in terms of @xmath469 , at least when @xmath470 is bounded and continuous .
the boundedness automatically follows by the definition of the two point function thanks to transience .
the continuity can be easily checked by lebesgue dominated convergence theorem by taking into account again proposition [ add2 ] . for each @xmath465 ,
the states @xmath467 given in are kms ( at inverse temperature @xmath349 ) for the dynamics generated by the bogoliubov automorphisms @xmath471 .
it is easily seen by proposition [ add2 ] that @xmath472 , @xmath473 are bounded and continuous ( in @xmath474 ) .
the proof follows by an elementary change of variables in as @xmath475 .
the second addendum in describes the amount of particles which condensates in the ground state whose wave function is nothing but the perron
frobenius weight @xmath173 . in the finite region
@xmath476 , such a density of the condensate is roughly given by @xmath477 even if the model exhibits a non negligible amount of the condensate , @xmath478 as @xmath479 , see proposition [ 3a3c ] .
formula explains what happens in the non homogeneous pure hopping model .
particles condensate even in the configuration space .
the amount of particles that the system can accommodate in the subnetwork made of the support of the perturbation is governed by the ratio between the volume growth of the original network and the @xmath39growth of the wave function of the ground state . in all the situations into consideration in the present paper
, @xmath480 goes to @xmath178 as @xmath476 approaches to @xmath116 .
this is due to the quite surprising fact that the perron frobenius weight decreases exponentially far from the perturbed zone ( cf .
[ figd ] ) , see @xcite , or also @xcite for several amenable cases .
this is the naive explanation that the mean density of the particles associated to all the states @xmath481 exhibiting bec is the critical one @xmath2 .
we start with a simple general result explaining that , for the pure hopping model , the transience is a necessary and sufficient condition for the existence of locally normal states exhibiting bec if one disregards the dynamics .
[ suftra ] let @xmath4 be a uniformly bounded degree connected network .
suppose that its adjacency @xmath10 is transient .
let @xmath482 denote the algebraic span of all the @xmath483 . then there exist locally normal states on @xmath484 exhibiting bec for the pure hopping hamiltonian @xmath485 .
indeed , consider any perron
frobenius eigenvector @xmath486 which exists by compactness , see e.g. proposition 4.1 of @xcite .
fix @xmath487 and put @xmath488 by transience ( cf .
( i ) of proposition [ add4 ] which works for any network @xmath4 as above ) , is meaningful and defines a quasi
free state on @xmath484 .
such a state can be interpreted as exhibiting bec for the presence of a non trivial amount of condensate on a lowest energy wave
function @xmath486 described by the addendum @xmath489 . as explained before
( see also @xcite ) , the careful investigation of the finite volume behavior of the kms states given in plays a relevant role because of the boundary effects which can not be avoided in the case of non amenable graphs as in our situation .
due to such boundary effects , this analysis becomes very delicate .
yet , the emerging situation is surprisingly in accordance with previous results in @xcite .
the main aim of the present section is to investigate , technicalities permitting , the infinite volume limits of finite volume gibbs states arising from the grand
canonical ensemble .
we start by considering standard finite volume approximations of kms states .
we focus the attention to any uniformly bounded degree connected network @xmath4 , equipped with the exhaustion @xmath95 made of finite volume regions . fix a common root @xmath490 , @xmath121 and consider the sequence @xmath491 of perron frobenius eigenvectors of the adjacency of the finite regions @xmath308 ,
all normalized to @xmath40 at @xmath178 .
in addition , we also suppose that @xmath492 ( or equivalently the pure hopping hamiltonian @xmath309 ) admits the ids w.r.t . the exhaustion @xmath95 .
we reduce the matter to the @xmath105subalgebra @xmath493 , see section 5.2.5 of @xcite .
we denote by @xmath494 the identity on @xmath275 .
fix a sequence of finite volume chemical potentials @xmath495 , with @xmath496 converging to @xmath289 .
the finite volume gibbs states @xmath497 are given , for such a sequence of chemical potential , by @xmath498 the case when @xmath499 is quite standard ( cf .
@xcite ) and presents no further complications .
thus , we restrict ourselves to the condensation regime @xmath296 . in correspondence of such a sequence @xmath307 of chemical potential , we put @xmath500 the following result takes into account what happens when the graph @xmath4 is recurrent .
[ roec ] if @xmath4 is recurrent then for each @xmath501 , @xmath502 we consider the splitting , and @xmath503 given in .
it is readily seen that @xmath504 and @xmath505 . by recurrence and the monotone convergence theorem
, we get @xmath506 on the other hand , as @xmath507 in the strong operator topology , @xmath508 then , again by the monotone convergence theorem and fatou lemma , @xmath509 then we have just shown that the two point function @xmath510 diverges in the infinite volume limit .
the last result together with proposition [ add2 ] , tells us that , even when the critical density is finite ( which certainly happens in presence of hidden spectrum ) , it is impossible to construct locally normal states exhibiting bec when the graph @xmath4 under consideration is recurrent ( compare with the similar results in @xcite ) . in our situation ,
proposition [ roec ] covers the case @xmath172 , with @xmath462 .
now we specialize to the situation of main interest here when the network @xmath4 ( denoted symbolically by @xmath116 ) is @xmath49 and @xmath47 . in this situation
, @xmath173 denotes the perron
frobenius eigenvector for @xmath146 described in , , and @xmath511 the finite volume perron
frobenius eigenvectors of the finite regions @xmath308 .
all such perron frobenius eigenvectors are normalized at @xmath40 on the fixed common root @xmath178 .
[ faxpx111 ] let @xmath511 be the ( unique ) perron frobenius eigenvector of @xmath512 normalized to 1 at the common root @xmath490
. then @xmath513 .
in addition @xmath514 .
we start by considering the networks @xmath515 consisting of @xmath177 , perturbed along the subset @xmath516 , see fig .
[ figgk ] .
the vectors @xmath517 , also described in @xcite , are the perron
frobenius weights normalized at @xmath40 on the common fixed root .
let @xmath91 be any weak cluster point of @xmath491 which exists by compactness ( cf .
proposition 4.1of @xcite ) .
it is straightforwardly seen that there exists a collections @xmath518 of smooth universal functions ( on a sufficiently large neighborhood of @xmath143 ) such that for each @xmath519 and @xmath520 , @xmath521 here , @xmath522 is the unique element such that @xmath183 ( cf .
lemma 4.2 of @xcite ) , @xmath486 is either @xmath91 or @xmath517 , and , correspondingly , @xmath140 is @xmath143 or @xmath523 .
notice that , it was shown in @xcite that @xmath524 this first implies that there exists a unique weak cluster point , say @xmath91 , of the sequence @xmath491 .
in addition , as @xmath525 , we get for each @xmath364 , @xmath526 concerning the limits of the norms we can suppose that the @xmath511 act on @xmath385 by extending those to @xmath178 on the complement of the balls @xmath176 , respectively
. we have by fatou lemma @xmath527 which completes the proof . the precise estimate of the behavior of the @xmath39norms @xmath528 of the perron
frobenius eigenvectors @xmath511 of @xmath512 as @xmath529 is needed in the sequel .
unfortunately , even if we have a precise estimate of the behavior of @xmath530 , this does not guarantee the analogous estimate for @xmath528 .
however , it is expected that @xmath531 still holds . for the reader convenience ,
we report a sufficient condition , tested on simpler models , which implies .
suppose that @xmath532 we automatically get @xmath533 in @xmath534 .
thus , as in the proof of proposition [ 3a3c ] , we compute , @xmath535 because @xmath536 , and @xmath537 as @xmath538 in the strong operator topology . the estimate implies the weaker one @xmath539 needed in proposition [ pprroo ] .
now we pass to the transient situation by showing a preliminary result which , in this case , takes into account the case when the parameter @xmath131 describing the amount of the condensate in on the base
point subspace is zero .
[ faxpx ] for each @xmath364 we have @xmath540 by using the splitting in , we can reduce the matter to the analysis of the resolvents .
notice that @xmath541 .
thus , by monotone convergence theorem , we get @xmath542 we now recall some notations standardly used in the sequel
. put @xmath543 eventually acting on @xmath544 .
consider @xmath176 the ball of radius @xmath166 of @xmath177 , @xmath545 is the support of the perturbation inside @xmath308 ( i.e. @xmath546 for @xmath547 , and @xmath175 is isomorphic to the finite set consisting of @xmath548 points for @xmath163 ) . finally , @xmath549 is the selfadjoint projection onto @xmath550 .
as @xmath308 is the finite volume perturbation of @xmath551 , @xmath552 .
[ korila ] if @xmath553 $ ] , then @xmath554 the proof immediately follows from the same computations in lemma 3.1 of @xcite , by taking into account that the resolvent is differentiable and convex .
point to investigate the infinite volume limit for the remaining cases , that is when the underlying perturbed graph is transient , is the following [ resspr ] let @xmath555 ( i.e. @xmath504 ) such that @xmath307 ( i.e. @xmath556 ) .
then , @xmath557 where @xmath558 .
in addition , if @xmath559 then @xmath560 . by using the splitting ,
the krein formula and finally lemma 2.4 of @xcite , we easily obtain for @xmath561 the above formula with @xmath562 given by , @xmath563 where the last goes to zero by propositions [ density0 ] and [ 6 ] . by lemma [ korila ]
, the second half directly follows as @xmath564 now we specialize the situation to the transient situation for the case when finite volume gibbs states are prepared in order to describe in the infinite volume limit , a mean density @xmath46 . to this end , we chose the sequence of the chemical potential by putting @xmath565 even if it is expected that implies that the resulting limit density describes a mean density @xmath566 , in order to prove that we should show that @xmath567 the last result would follow again by the krein formula for the resolvent
. unfortunately , its proof relies by unavoidable boundary effects which seriously affect the related formulas and are not easily manageable to take the infinite volume limit .
however , for the infinite volume mean density @xmath45 , if holds true , we get by proposition [ resspr ] , @xmath568 this simply means that , if holds true , the sequence of finite volume gibbs states obtained with the choice of the chemical potentials , describes a fixed mean density @xmath46 in the infinite volume limit .
it is expected that the condition is incompatible with local normality as it is shown in the following [ pprroo ] suppose that the estimate holds true .
if the sequence of the finite volume chemical potentials satisfies , then for the two point function in , we get @xmath569 we have @xmath570 the proof easily follows collecting together and .
the last result together with proposition [ add2 ] , would mean that it is impossible to construct locally normal states exhibiting bec with mean density @xmath46 in all the situations under consideration in the present paper .
finally , we pass for the transient cases to the construction of the states by the infinite volume limits of finite volume gibbs states in with a suitable choice of the sequence of chemical potentials . also in this situation
, we assume that @xmath571 holds true . in order to obtain itself for each other point @xmath364 . ] in order to recover by the infinite volume limits of , we put @xmath572 let @xmath116 be either @xmath49 with @xmath573 , or @xmath47 where @xmath463 , @xmath160 with @xmath464 given in .
suppose that holds true , and is satisfied under the choice .
then for each @xmath364 , we get for the two point function in , @xmath574 in addition @xmath560 .
the second half directly follows from the assumption , by the second half of proposition [ resspr ] .
indeed , we get , @xmath575 concerning the first part , as usual we reduce the matter to the resolvent . as @xmath576 , for each fixed @xmath447 , we get by monotony @xmath577 by proposition [ faxpx111 ] , we get after taking the limits in @xmath166 and in @xmath578 , @xmath579 on the other hand , by using the krein formula , we obtain @xmath580 which , taking into account , leads to @xmath581 collecting together , we get the assertion .
we end by noticing that the case without condensate , but at the critical point @xmath296 , is directly covered by proposition [ faxpx ] without any further assumption .
the author would like to thank the university of pretoria , and especially a. strh , for the warm hospitality and for financial support during the period march may 2010 , when the present work has been started .
silvestrini p. , russo r. , corato v. , ruggiero b. , granata c. , rombetto s. , russo m. , cirillo m. , trombettoni a. , sodano p. _ topology induced critical current enhancement in josephson networks _ ,
phys . lett .
a * 370 * ( 2007 ) , 499503 . | we investigate the bose einstein condensation on non homogeneous non amenable networks for the model describing arrays of josephson junctions .
the graphs under investigation are obtained by adding density zero perturbations to the homogeneous cayley trees .
the resulting topological model , whose hamiltonian is the pure hopping one given by the opposite of the adjacency operator , has also a mathematical interest in itself .
the present paper is then the application to the bose
einstein condensation phenomena , of the harmonic analysis aspects , previously investigated in a separate work , for such non amenable graphs . concerning the appearance of the bose einstein condensation ,
the results are surprisingly in accordance with the previous ones , despite the lack of amenability .
the appearance of the hidden spectrum for low energies always implies that the critical density is finite for all the models under consideration .
first we show that , even when the critical density is finite , if the adjacency operator of the graph is recurrent , it is impossible to exhibit temperature states which are locally normal ( i.e. states for which the local particle density is finite ) describing the condensation at all .
a similar situation seems to occur in the transient cases for which it is impossible to exhibit locally normal states @xmath0 describing the bose
einstein condensation at mean particle density @xmath1 strictly greater than the critical density @xmath2 .
indeed , it is shown that the transient cases admit locally normal states exhibiting bose
einstein condensation phenomena . in order to construct such locally normal temperature states by infinite volume limits of finite volume gibbs states , a careful choice of the the sequence of the chemical potentials should be done . for all such states ,
the condensate is essentially allocated on the base point supporting the perturbation .
this leads to @xmath3 .
we prove that all such temperature states are kubo
martin
schwinger states for a natural dynamics .
the construction of such a dynamics , which is a delicate issue , is also done . |
_ slow - fast _ vector fields have the form @xmath0 with @xmath1 the fast variable , @xmath2 the slow variable and @xmath3 a small parameter that represents the ratio of time scales .
the pair @xmath4 will be denoted by @xmath5 and the vector field will be written @xmath6 .
simulation of these systems with explicit numerical integration algorithms is limited to time steps that are @xmath7 due to numerical instabilities .
however , invariant _ slow manifolds _ on which the motion of the system has speed that is @xmath8 are a common feature of slow - fast systems .
indeed , trajectories often spend most of their time following stable slow manifolds .
implicit `` stiff '' integration methods @xcite compute trajectories along the stable slow manifolds , taking time steps that are @xmath8 while avoiding the numerical instabilities of explicit methods . however , no initial value solver will compute forward trajectories that evolve on unstable slow manifolds because the geometric instability of these trajectories is such that nearby initial conditions diverge from one another at exponential rates commensurate with the fast time scale .
even an exact initial value solver in the presence of round - off errors of magnitude @xmath9 will amplify this round - off error to unit magnitude in a time that is @xmath10 .
trajectories on slow manifolds that expand in all normal directions can be computed by reversing time , but different strategies are needed to compute trajectories that lie on slow manifolds of saddle type .
this paper presents the first algorithms that directly compute accurate trajectories of slow manifolds of saddle type .
the existence of _ normally hyperbolic _ slow manifolds is established by _
fenichel theory _ @xcite .
the singular limit @xmath11 of system is a differential algebraic equation with trajectories confined to the _ critical manifold _
@xmath12 defined by @xmath13 . at points of @xmath14 where @xmath15 is a regular @xmath16 matrix , the implicit function theorem implies that @xmath14 is locally the graph of a function @xmath17 .
this equation yields the vector field @xmath18 for the _ slow flow _ on @xmath14 .
the geometry is more complicated at _ fold points _ of @xmath14 where @xmath15 is singular .
it is often possible to extend the slow flow to the fold points after a rescaling of the vector field @xcite . where all eigenvalues of @xmath15 have nonzero real parts , fenichel proved the existence of invariant slow manifolds @xmath19 for @xmath20 small .
these _ normally hyperbolic _ slow manifolds are within an @xmath7 distance from the critical manifold @xmath21 and the flow on @xmath19 converges to the slow flow on @xmath21 as @xmath22 .
fenichel theory is usually developed in the context of _ overflowing _ slow manifolds with boundaries which trajectories may leave through the boundaries . in this
setting , slow manifolds are not unique , but the distance between a pair of slow manifolds is `` exponentially small '' , i.e. of order @xmath23 for a suitable positive @xmath24 , independent of @xmath3 @xcite .
this section describes a collocation method for computing slow manifolds of saddle type in slow - fast systems that we call the smst algorithm .
the numerical analysis employed in the algorithm is straightforward ; the subtlety of the problem appears in the formulation of discrete systems of equations with well - conditioned jacobians .
the crucial part of the geometry is to specify boundary conditions for trajectory segments on a slow manifold that yield well - conditioned discretizations .
a trajectory segment @xmath25 \to r^{m+n}$ ] of system is determined by its initial point @xmath26 or by another set of @xmath27 boundary conditions .
trajectories that follow a slow manifold for some distance approach the manifold initially at a fast exponential rate and then diverge from the manifold at a fast exponential rate .
such trajectories will be found as solutions to a boundary value problem that imposes constraints on both @xmath26 and @xmath28 . at @xmath11 , there are specific arrival and departure points .
the singular limit of the trajectories we seek are _ candidates _ @xmath29 that consist of a fast initial segment approaching the critical manifold @xmath14 along a strong stable manifold , followed by a slow segment along @xmath14 , followed by a fast segment that leaves @xmath14 along a strong unstable manifold .
see figure [ bv_su ] .
the initial and/or final segments may be absent . for small @xmath20 ,
we seek @xmath27 boundary conditions that determine a unique trajectory near the candidate .
initial conditions that do not lie in the strong stable manifold of a point @xmath30 will diverge from the slow manifold @xmath14 at a fast exponential rate .
therefore trajectories that follow the slow manifold have initial conditions that are exponentially close to the ( unknown ) stable manifold of @xmath14 .
similarly , when trajectories depart from @xmath14 , they remain exponentially close to the unstable manifold of @xmath14 for times that are @xmath8 on the fast time scale .
consequently , to have a solvable boundary problem that is well posed , the @xmath27 boundary conditions should consist of two manifolds of dimension @xmath31 and @xmath32 , the first transverse to the stable manifold of @xmath14 and the second transverse to the unstable manifold of @xmath14 .
thus @xmath33 where @xmath34 is the dimension of the strong unstable manifolds of @xmath14 .
the fast segments of trajectories that precede and follow segments along the slow manifold are readily computed with an initial value solver ; the challenge is to locate the slow portion of the trajectory .
therefore , the algorithm presented here takes as its input a ( discretized ) trajectory @xmath35 \to c$ ] of the slow flow on the critical manifold together with two submanifolds @xmath36 and @xmath37 of dimensions @xmath31 and @xmath32 that pass close to the initial and final points @xmath38 and @xmath39 of @xmath29 .
the manifold @xmath36 is assumed to be transverse to the stable manifold of @xmath40 and the manifold @xmath37 is assumed to be transverse to the unstable manifold of @xmath40 .
see figure [ bv_su ] . .
a trajectory that approaches the slow manifold along a strong stable direction and departs along a strong unstable manifold is drawn blue .
the initial point of this trajectory lies in a two dimensional manifold @xmath36 transverse to the stable manifold of @xmath14 , and the final point lies in a one dimensional manifold @xmath37 transverse to a unstable manifold of @xmath14 .
, height=480 ] denote the mesh points in the discretization of @xmath29 by @xmath41 .
the algorithm described here is based on a set of @xmath42 equations @xmath43 for @xmath44 that yield an approximate solution to the boundary value problem that @xmath45 is a trajectory of system connecting @xmath46 to @xmath47 is allowed to vary , then the number of boundary conditions is increased by one . ] .
the discretized equations are based upon interpolation with cubic hermite splines , though higher degree splines can be used in conjunction with automatic differentiation methods that compute taylor polynomials of the vector field at mesh points @xcite . from points
@xmath48 , a @xmath49 cubic spline @xmath50 is constructed with the @xmath51 as knot points and tangent vectors @xmath52 at these points . on the mesh interval
@xmath53 $ ] , @xmath50 is a cubic curve whose coefficients are linear combinations of @xmath54 that are readily determined .
each of the @xmath55 mesh intervals @xmath53 $ ] contributes @xmath56 equations to the system @xmath43 by requiring that @xmath57 .
the values of @xmath50 and @xmath58 in these equations can be expressed as @xmath59 implicit equations for the boundary value manifolds constitute the remaining @xmath27 equations in @xmath43 .
the system @xmath43 is solved with newton s method starting with the data in @xmath29 .
the jacobian of the system @xmath43 can be computed , using the derivatives of the equations with respect to @xmath60 .
two types of error estimates are of interest for this algorithm . on each mesh interval
, there is a local error estimate for how much the spline @xmath50 differs from a trajectory of the vector field .
the spline satisfies @xmath61 at the collocation points @xmath62 and @xmath63 .
if @xmath45 is the trajectory of the vector field through one these points , this implies that @xmath64 . since this classical estimate is based upon the assumption that the norm of the vector field is @xmath8 , it is only likely to hold for intervals that are short on the fast time scale .
globally , the trajectories of the flow display a strong separation due to the normal hyperbolicity . in
_ fenichel coordinates _
@xcite , stable coordinates converge rapidly to the slow manifold while unstable coordinates diverge rapidly from the slow manifold . in the case of a one dimensional slow manifold , _
shadowing _
@xcite implies that any _ pseudo - trajectory _ pieced together from local approximations to the flow will lie close to a unique trajectory of the flow .
moreover , in this case , different choices of boundary conditions that lie in the same strong stable manifold at @xmath65 and the same strong unstable manifold at @xmath66 yield trajectories that are exponentially close to each other and to the slow manifold outside of small subintervals near the ends of the time interval @xmath67 $ ] .
consequently , the value of @xmath68 will be @xmath8 on the slow time scale and solutions of @xmath43 are expected to give quite accurate approximations to the slow manifold . rather than pursuing more careful theoretical analysis of the algorithm
, this paper tests its implementation on several examples .
this section presents four examples : 1 . a linear system for which there are explicit solutions of both the solutions of the differential equations and the boundary value solver , 2 . a three dimensional version of the morris - lecar model for bursting neurons that was used by david terman in his analysis of the transition between bursts with different numbers of spikes @xcite , 3 . a three dimensional system whose homoclinic orbits yield traveling wave profiles for the fitzhugh - nagumo model @xcite , 4 . a four dimensional model of two coupled neurons studied by guckenheimer , hoffman and weckesser @xcite .
the general solution of the linear vector field @xmath69 is @xmath70 this explicit solution provides a benchmark for evaluating the accuracy of the algorithm described above .
the slow manifold of the system is the line @xmath71 containing the trajectories @xmath72 .
the discretized equations of the algorithm can also be solved explicitly for system .
the first step in doing so is to observe that the equations for @xmath73 and @xmath74 are separable from those for @xmath75 , and this remains the case for the discretized equations of the boundary value solver . substituting the equations for the @xmath74 variable into the boundary value equations produces the equation @xmath76 on each mesh interval . if a boundary condition is imposed on one end of the time interval @xmath67 $ ] , these equations yield a solution that is a discretization of an exact solution of the differential equation .
convergence occurs in a single step .
assume now that @xmath76 and set @xmath77 to be he difference between the @xmath75 coordinate of a point and a point of the slow manifold .
the boundary value equations become @xmath78 with @xmath79 .
note that these equations are satisfied when the @xmath80 vanish , so discretizations of exact solutions along the slow manifold satisfy the boundary value equations . solving the equation for @xmath81 in terms of @xmath80 yields @xmath82 like the solutions of the differential equation , the values @xmath83 decrease exponentially as a function of time .
the ratio @xmath84 is a function of @xmath85 whose taylor expansion agrees with that of @xmath86 through terms of degree @xmath87 , and its value always lies in the interval @xmath88 .
thus the the solutions of the boundary value equation converge geometrically toward the slow manifold along its stable manifold with increasing time .
if the mesh intervals have length @xmath89 , then the relative error of the decrease satisfies @xmath90 for large values of @xmath91 , the solution is no longer accurate near @xmath92 if the boundary conditions do not satisfy @xmath93 . a similar , but
simpler argument establishes that the solution of the discretized problem converges to the slow manifold at an exponential rate with decreasing time from @xmath94 .
thus , the boundary value solver is stable and yields solutions that qualitatively resemble the exact solution for all meshes when applied to this linear problem . in particular ,
the solution of the discretized problem is exponentially close to the slow manifold away from the ends of the time interval @xmath67 $ ] .
as the mesh size decreases to zero , the algorithm has fourth order convergence to the exact solution .
action potentials are a primary means for communicating information within the nervous system @xcite .
neurons are said to burst @xcite when they fire several consecutive action potentials between `` silent '' periods free of bursts .
there is no universally accepted definition of bursts , but computational models are widely used to predict in terms of membrane channel properties when a neuron will burst .
rinzel @xcite introduced a singular perturbation perspective to the investigation of bursting in model neurons , viewing the phenomenon as a relaxation oscillation in which a system makes fast time scale transitions between slowly varying equilibrium and periodic attractors .
several classifications of bursting distinguish qualitatively different dynamics .
for example , izhikevich @xcite classifies bursts in terms of the bifurcations that mark the transitions between bursts and silent intervals .
terman @xcite studied changes in the number of spikes per burst that occur as system parameters are varied .
he gave numerical examples in a version of the morris - lecar model @xcite first analyzed by rinzel and ermentrout @xcite : @xmath95 this system has periodic bursting solutions with different numbers of spikes per burst as the parameters @xmath3 and @xmath31 vary .
figure [ terman_traj ] illustrates that there are narrow parameter ranges with two stable periodic orbits having different spike numbers .
terman described the dynamics of the transition from periodic solutions with @xmath96 spikes to those with @xmath97 , relying upon numerical simulations of trajectories in his analysis .
flow along a slow manifold of saddle type is a central aspect of this transition , but the trajectory simulations are incapable of following trajectories that remain close to this slow manifold for more than a short distance .
the boundary value solver introduced in this paper is used to compute trajectories that contain segments which follow the slow manifold of saddle type .
geometric structures involved with the transition from @xmath96 to @xmath97 spikes are visualized , and the analysis of the transition from @xmath96 to @xmath97 spikes is carried further .
the fast subsystem of is the morris - lear model for action potentials of barnacle muscle @xcite .
the morris - lecar model itself has a rich dynamical structure @xcite .
there is an interval of values for @xmath98 in which the system has three equilibrium points .
saddle - node bifurcations occur at the endpoints of this interval .
the equilibrium points of the morris - lecar model constitute the critical manifold of , and its saddle - node bifurcations ( with varying @xmath98 ) are the folds of the critical manifold .
there is also a family of periodic orbits that collapses at a subcritical hopf bifurcation and terminates at a homoclinic bifurcation .
this family of periodic orbits is folded , i.e. , there is a saddle - node of limit cycle bifurcation @xcite within the family .
the periodic orbits of the family between the fold and homoclinic bifurcations are stable .
the bursting orbits of the vector field follow a branch of the critical manifold of equilibrium points to one of its folds , jump to the family of stable periodic orbits , follow this family to its homoclinic bifurcation and then jump back to the branch of stable equilibria .
these bursting orbits occur when the value of the parameter @xmath31 is chosen so that @xmath98 increases slowly during the quiescent part of the cycle and decreases slowly during the active spiking portion of the cycle .
see figure [ terman_traj ] .
the homoclinic orbit of the singular limit @xmath99 is a transversal intersection of the stable and unstable manifolds of the branch of saddle equilibria of .
the branch of equilibria become a slow manifold @xmath14 of saddle type when @xmath100 and the homoclinic orbit persists as an intersection of the stable and unstable manifolds @xmath101 of @xmath14 .
the transition between @xmath96 and @xmath97 spikes per burst occurs when the periodic bursting cycle encounters the intersection of @xmath102 and @xmath103 .
the final spike of a periodic orbit with @xmath97 spikes follows the intersection of @xmath102 and of @xmath103 back to @xmath14 before jumping to the stable slow manifold .
figure [ terman_mflds ] visualizes @xmath104 and @xmath103 , and shows two trajectories that bracket the intersection of @xmath102 and @xmath103 . because the system is smooth and does not have an equilibrium point near the intersection of @xmath102 and @xmath103 , the transition from @xmath96 to @xmath97 spikes consists of trajectories that undergo a continuous evolution .
these trajectories contain saddle canards , segments that follow @xmath14 for varying lengths of time before leaving @xmath14 along one of its strong unstable manifolds .
trajectories lying close enough to @xmath102 turn and flow along @xmath14 when they approach it .
the distance that they travel along @xmath14 before leaving along its unstable manifold @xmath103 depends logarithmically on the initial distance of the trajectory to @xmath102 . if close enough , the trajectory will follow @xmath14 all the way to its end near a fold of the critical manifold before making a fast excursion to the stable slow manifold . accurate computation of @xmath14 is essential to understanding the details of the transition from bursts with @xmath96 spikes to bursts with @xmath97 spikes .
the critical manifold of the vector field is given by explicit formulas when parametrized by @xmath105 .
uniform meshes of values for @xmath105 were used to generate starting values for the boundary value computation of the slow manifold @xmath14 . to compute @xmath14 ,
the vector field was rescaled so that @xmath106 . with this rescaling ,
the value of @xmath98 remains constant during the newton iteration to find the solutions .
typical meshes that do not come close to the fold points result in convergence of newton s method within three or four steps .
figure [ terman_test ] illustrates the accuracy of the computations of @xmath14 and the behavior of numerical simulations of trajectories that start near @xmath14 .
a point @xmath107 on @xmath14 is chosen , and the jacobian of the fast subsystem at this point is computed to obtain approximations for the directions of its strong stable and unstable manifolds . if @xmath107 does lie on the slow manifold , then trajectories with initial conditions on opposite sides of @xmath14 on its strong unstable manifold will flow along @xmath14 but then jump in opposite directions . similarly , backward trajectories with initial conditions on opposite sides of @xmath14 on
its strong stable manifold will flow along @xmath14 but then jump in opposite directions . if @xmath107 is displaced from @xmath14 , its distance to @xmath14 can be estimated by finding the closest pairs of bracketing trajectories that do jump from @xmath14 in opposite directions .
figure [ terman_test ] displays the results of such a test .
eight pairs of trajectories displaced along the strong unstable manifold at distances @xmath108 are plotted in blue and green , and eight pairs of backward trajectories displaced along the strong stable manifold at distances @xmath108 are plotted in red and magenta .
pairs of trajectories displaced by distance @xmath109 ( not drawn in figure [ terman_test ] ) fail the test , jumping in the same direction .
this suggests that the distance from @xmath107 to the slow manifold is smaller than @xmath110 .
note also that increments in the distance that each successive pair of bracketing trajectories flows along @xmath14 are similar , consistent with the exponential separation of trajectories within the strong stable and unstable manifolds . extrapolating these increments yields the estimate that a numerically simulated trajectories starting on the slow manifold near @xmath107 will only be able to remain close to @xmath14 for time approximately @xmath111 .
this estimate is based on round - off error of the order of @xmath112 and the observation that the times at which trajectories displaced from @xmath107 by distances @xmath113 and @xmath110 appear to jump from @xmath14 are approximately @xmath114 and @xmath115 .
these crude estimates explain why initial value solvers are unable to follow the continuous evolution of trajectories in the transition from @xmath96 to @xmath97 spikes per burst .
the value of @xmath105 at @xmath107 is approximately @xmath116 and the jump from @xmath14 of numerically simulated trajectories seems to occur before @xmath105 increases to @xmath117 , but the fold of the critical manifold occurs when @xmath105 is approximately @xmath118 .
the exponential instability of @xmath14 in both forward and backward directions precludes initial value solvers from computing trajectories that flow along @xmath14 from the intersection of @xmath102 and @xmath103 to the fold of @xmath14 . along the strong stable and unstable manifolds at the point @xmath119 on the slow manifold , which is drawn as a filled black circle .
the trajectories are computed for a time interval @xmath120 and color coded so that the trajectories along the two branches of the strong unstable manifold are drawn blue and green while the trajectories along the strong stable manifold are drawn red and magenta .
parameter are @xmath121 and the objects are projected into the @xmath122 plane.,height=480 ] computation of periodic orbits with long canard segments near the slow manifold of saddle type appears to be challenging , even with continuation methods .
computation of the slow manifold @xmath14 with the boundary value solver introduced here can be coupled with the analysis of terman @xcite and lee and terman @xcite to solve this problem .
computations of the slow manifold @xmath14 are augmented with numerical forward and backward simulations of trajectories that terminate at a cross - section along the family of periodic orbits .
this extends the approach introduced by guckenheimer and lamar @xcite to efficiently compute periodic orbits containing canards . of saddle type in system
is swept out by trajectories starting close to the slow manifold .
the blue and green trajectories are followed until they intersect the cross - section @xmath123 with @xmath98 increasing .
red trajectories on the stable manifold of @xmath14 are followed backward until they intersect this cross - section .
the trajectories that are drawn reach the cross - section close to the points on the unstable manifold of @xmath14 .
parameters are @xmath124.,height=480 ] figure [ terman_umfld ] visualizes the invariant manifold @xmath103 as a collection of trajectories for parameter values @xmath124 in system .
the heavy black curve is a segment of the slow manifold @xmath14 of saddle type , and the dotted black curve is the critical manifold . at twenty initial points along @xmath14 ,
trajectories have been computed with initial conditions displaced from @xmath14 along its strong unstable manifolds by a distance @xmath125 .
the trajectories starting on one side of @xmath14 are drawn blue and the trajectories starting on the other side of @xmath14 are drawn green .
the blue trajectories make a loop around the unstable branch of the slow manifold and then flow past @xmath14 to the stable branch of the slow manifold .
the green branches flow to the stable branch of the slow manifold with @xmath105 decreasing .
both sets of branches then turn and flow along the stable branch of the slow manifold .
when they reach the fold of the slow manifold , they jump to the family of rapid oscillations .
as trajectories follow these oscillations , @xmath98 decreases .
the displayed trajectories are terminated when they reach the plane @xmath126 with @xmath98 decreasing .
the red curves displayed in figure [ terman_umfld ] are four backwards trajectories that begin at distance @xmath127 from @xmath14 along its stable manifold and end on the cross - section @xmath126 .
these trajectories were chosen on a short section of @xmath14 so that they reach the cross - section @xmath126 near the ends of the blue and green trajectories .
figure [ terman_sect](a ) shows the ends of the blue , green and red trajectories with the cross - section @xmath126 .
figure [ terman_sect](b ) and ( c ) show similar plots for the system with parameter values @xmath128 and @xmath129 . as @xmath3 varies
, these plots demonstrate that the trajectories in the unstable manifold of @xmath14 sweep across the stable manifold of @xmath14 .
figure [ terman_umfld ] supports the following procedure for finding periodic orbits containing canards .
fix a short segment @xmath130 transverse to @xmath103 . with varying @xmath3
, trajectories with initial conditions on @xmath130 sweep out a three dimensional manifold @xmath131 in @xmath132 space .
the exchange lemma @xcite implies that if @xmath131 intersects @xmath102 transversally in @xmath132 space , then part of @xmath131 will stretch along the length of @xmath14 and depart from it along @xmath103 . in particular , @xmath131 will intersect @xmath130 , giving a unique value of @xmath3 for which there is a periodic orbit intersecting @xmath130 .
figure [ terman_sect ] gives numerical evidence that @xmath131 does intersect @xmath103 , and it indicates that the value of @xmath3 will be almost constant along the family of periodic orbits containing canards .
computing trajectories with initial conditions on @xmath130 with an initial value solver will not produce these periodic orbits .
figure [ terman_retmap](a ) shows a return map , giving initial and final values for the variable @xmath105 , with 300 initial points chosen on a linear approximation to the intersection of @xmath102 with @xmath123 shown in figure [ terman_sect](a ) .
this return map has two apparent jumps .
the trajectories beginning between the two jumps make three spikes before returning to @xmath123 while the other trajectories make two spikes before returning .
none of the trajectories flows along @xmath14 with @xmath98 decreasing to a value smaller than @xmath133 .
figure [ terman_retmap](b ) plots five of the trajectories from the return map , four that bracket the jumps and one from the local maximum of the return map in figure [ terman_retmap](a ) .
the approximations of @xmath14 obtained with the boundary value solver can be used in approximating periodic orbits with canards .
the strategy we propose is illustrated by figures [ terman_umfld ] and [ terman_sect ] .
the periodic orbit will be calculated in three segments that are illustrated as black , red and blue / green curves in figure [ terman_umfld ] .
the canard segment of a periodic orbit is exponentially close to @xmath14 except at its arrival and departure points .
when the periodic orbit departs from @xmath14 , it will follow a blue or green trajectory starting at a point exponentially close to @xmath103 that is numerically indistinguishable from points on @xmath103 .
similarly , the orbit segment that arrives at @xmath14 , does so at a point that is exponentially close to @xmath102 , that is numerically approximated by backward integration beginning at a point of @xmath102 . as @xmath3 ( or another parameter )
is varied , the forward trajectory along @xmath103 and backward trajectories along @xmath102 sweep out a curve and a surface of intersection with a cross - section in @xmath132 space ( here @xmath123 ) .
a root solver can be used to locate a parameter value for which a trajectory of @xmath103 and one on @xmath102 arrive at the same point of the cross - section .
the periodic orbit will then be approximated by the union of the two trajectories and a curve that flows along @xmath14 from the chosen arrival point to the chosen departure point .
normal hyperbolicity implies that there is a unique trajectory that connects these two points .
as shown in figure [ terman_sect ] , the intersections occur for @xmath134 for all arrival and departure points .
analysis of the bifurcations and attracting limit sets of the vector field requires additional information .
numerically , it is necessary to `` fill in '' the jumps in the return map shown in figure [ terman_retmap ] , describing more carefully how the trajectories with canards return and determining the stability of trajectories containing canard segments .
figure [ terman_uv ] plots the final values of @xmath105 in trajectories on @xmath103 vs. their initial values of @xmath105 .
it is apparent that a large portion of @xmath103 contracts enough when it flows along the stable branch of the slow manifold that its intersection with the cross - section @xmath123 is very small .
the image appears to lie inside a disk centered at @xmath135 of radius @xmath136 .
the minimum return value of @xmath105 in the points plotted in figure [ terman_retmap ] is approximately @xmath137 .
thus it appears that the local minima of the return map are only a distance about @xmath138 below the lowest points plotted in this figure .
the variational equations of system can be used to estimate how much expansion takes place along canard segments of trajectories and how much contraction takes place along the stable branch of the slow manifold .
on points of the critical manifold with the same value of @xmath98 , the strong unstable eigenvalue on the middle branch has larger magnitude than the weaker stable eigenvalue on the stable branch .
if a canard segment is long enough , then the accumulated expansion will dominate the subsequent contraction that takes place on the stable branch of the slow manifold .
this suggests that the return map of the system will have an expanding direction for canards that are sufficiently long , consistent with figure [ terman_retmap ] . as the maximal canards of the return map move across the diagonal with changing parameters
, the return map is likely to have chaotic invariant sets similar to those found in the henon map @xcite .
the numerical computations reported here are insufficient to adequately determine the details of these invariant sets because the slow manifold is not computed close enough to the fold that that the maximal canards are determined with good accuracy .
terman @xcite suggests that the system has trajectories with three different spike numbers in its bursts .
the calculations here cast doubt abut whether this is possible for trajectories that lie in the forward limit set of the system .
for the parameters @xmath124 , the decrease in the value of @xmath98 between spikes along the surface of oscillations appears to be approximately @xmath139 in the region between the arrival of trajectories jumping from the fold of the stable branch of the slow manifold to the intersection of @xmath103 and @xmath102 . on the other hand , the trajectories that flow along the stable branch of the slow manifold appear to pass by the fold in a set that has a diameter at least an order of magnitude smaller than the observed separation between spikes .
our analysis of canards makes it clear that the trajectories with long canard segments all flow through a tiny region as they pass the fold .
lee and terman @xcite give asymptotic estimates of the size of these regions in terms of @xmath140 that also suggest that it is unlikely that the limit set of this system reaches the surface of oscillations in a set that is large enough to contain trajectories with three different spike numbers .
the fitzhugh - nagumo equation is a model for the electric potential @xmath141 of a nerve axon interacting with an auxillary variable @xmath142 ( see @xcite,@xcite ) : @xmath143 where @xmath144 and @xmath145 and @xmath65 are parameters . assuming a travelling wave solution with @xmath146 to ( [ eq : fhn_original ] ) we get : @xmath147 a homoclinic orbit of ( [ eq : fhn_temp ] ) corresponds to a travelling pulse solution in ( [ eq : fhn_original ] ) .
an analysis of ( [ eq : fhn_temp ] ) using numerical continuation has been carried out by champneys et al .
they fixed the parameters @xmath148 , @xmath149 , @xmath150 and investigated bifurcations in @xmath151-parameter space .
we shall fix the same values and hence write @xmath152 . to bring ( [ eq : fhn_temp ] ) into the standard form ( [ sfs ] ) set @xmath153 , @xmath154 , @xmath155 and change to the slow time scale : @xmath156
we refer to ( [ eq : fhn ] ) as `` the '' fitzhugh - nagumo equation .
our goal is to use the fast slow structure of ( [ eq : fhn ] ) and the smst algorithm to compute its homoclinic orbits .
the critical manifold @xmath14 of the fitzhugh - nagumo equation is the cubic curve : @xmath157 the two local non - degenerate extrema of @xmath158 yield the fold points of @xmath14 .
denote the local minimum by @xmath159 and the local maximum by @xmath160 .
the critical manifold @xmath14 has three normally hyperbolic components : @xmath161 fenichel s theorem provides associated slow manifolds @xmath162 , @xmath163 and @xmath164 outside neighbourhoods of the fold points .
the manifolds @xmath162 and @xmath164 are of saddle - type for @xmath165 sufficiently small .
the middle branch @xmath163 is completely unstable in the fast directions .
denote the unique equilibrium point of ( [ eq : fhn ] ) by @xmath166 .
the location of @xmath167 depends on the parameter @xmath107 and @xmath167 moves along the cubic @xmath14 . for the analysis of homoclinic orbits
we shall assume that @xmath168 .
in this case , the unstable manifold @xmath169 is one - dimensional and the stable manifold @xmath170 is two - dimensional .
this also covers the case @xmath171 by a symmetry in the fitzhugh - nagumo equation and avoids the region where @xmath167 is completely unstable @xcite,@xcite .
homoclinic orbits exist if @xmath172 .
+ we focus first on the case of relatively large wave speeds @xmath173 ( `` fast waves '' ) . the existence proof of these homoclinic orbits contructs them as perturbations of a singular trajectory consisting of four segments : a fast subsystem heteroclinic connection from @xmath167 to @xmath174 at @xmath175 , a slow segment on @xmath174 , a fast subsystem heteroclinic from @xmath174 to @xmath176 at @xmath177 for some constant @xmath178 and a slow segment on @xmath176 connecting back to @xmath167 @xcite .
we aim to compute homoclinic orbits by a similar procedure for a given small @xmath179 in several steps : 1 .
find parameter values @xmath180 such that a homoclinic orbit exists very close or exactly at @xmath180 .
this can be achieved by a splitting algorithm without computing the homoclinic orbit , even for very small values of @xmath165 @xcite .
carry out all the following compuations for @xmath181 .
[ step1 ] compute the slow manifolds @xmath182 and @xmath183 using the smst algorithm .
[ step2 ] compute the unstable manifold of the equilibrium @xmath169 by forward integration .
4 . define a section @xmath184 where the constant @xmath24 is chosen between @xmath159 and @xmath160 e.g. @xmath185 .
compute the transversal intersection of @xmath186 and @xmath187 on @xmath130 , call the intersection point @xmath188 ( see figure [ fig : hom_intersect_truncated ] ) . integrate forward and backward starting at @xmath189 to obtain trajectories @xmath190 and @xmath191 .
[ step4 ] to compute the homoclinic orbit we use the objects computed so far as approximants in different regions .
compute the closest points in @xmath169 and @xmath191 to @xmath164 and concatenate @xmath169 and @xmath191 to @xmath164 at these points .
proceed similarly with @xmath162 and @xmath190 .
remove all parts of the slow manifolds not lying between the concatenation points and past the equilibrium @xmath167 .
( green ) and @xmath187 ( magenta ) .
the manifolds are trucated at the yellow section @xmath130 and the trajectory @xmath192 started on @xmath130 at the transversal intersection point @xmath189 is shown in red.,scaledwidth=60.0% ] note that all figures for the fast wave case have been computed for @xmath193 , @xmath194 and @xmath195 .
this is a classical case @xcite for which the existence of homoclinic orbits is known . in figure
[ fig : hom_umfld_smfld ] we show the result from the smst algorithm and the unstable manifold of the equilibrium @xmath169 , i.e. the output of steps [ step1 ] and [ step2 ] . due to the exponential separation along @xmath164 the trajectory @xmath169 obtained from numerical integration can not track the slow manifold for an o(1 ) distance and escapes after following the slow manifold for a very short time .
this happens despite the fact that we have computed parameter values @xmath180 with maximal accuracy in double precision arithmetic at which we expect @xmath169 to follow @xmath164 almost up to the fold point @xmath160 .
this observation is relevant to figure [ fig : hom_orbit_fast_pieces ] where the result of step [ step4 ] is shown .
all the fast segments ( red ) had to be truncated almost immediately after they entered a neighourhood of a slow manifold .
the final output of the algorithm after interpolation near the truncation points is shown in figure [ fig : hom_orbit_fast ] .
+ now we consider the case of `` slow waves '' and work with smaller wave speeds @xmath173 .
homoclinic orbits representing slow waves should be thought of as perturbations of singular limit orbits for the fitzhugh - nagumo equation ( [ eq : fhn ] ) with @xmath196 . in this case
the fast subsystem @xmath197 is hamiltonian .
singular homoclinic orbits exist in a single fast subsystem with the y - coordinate of the equilibrium @xmath175 .
a direct application of fenichel theory implies that a perturbed singular `` slow '' homoclinic orbit persists for @xmath179 @xcite .
again it is possible to compute parameter values @xmath198 at which homoclinic orbits for @xmath179 exist @xcite . to compute the orbits themselves a similar approach as described above can be used .
we have to track when @xmath169 enters a small neighbourhood of @xmath186 respectively of @xmath162 .
figure [ fig : slow_waves ] shows two computed homoclinic orbits for @xmath199 and @xmath200 .
+ the orbits spiral around the middle branch and do not enter the vicinity of @xmath164 . this
is expected as the middle branch @xmath201 of the critical manifold consists of unstable spiral equilibria for the fast subsystems .
the hamiltonian analysis for the case @xmath196 shows that the singular slow homoclinic orbits are not close to an equilibrium on @xmath202 for values of @xmath107 approximately between @xmath203 and @xmath204 ( see @xcite ) . in figure
[ fig : slow_wave1 ] a homoclinic orbit enters the vicinity of the slow manifold @xmath162 and returns directly to @xmath167 .
figure [ fig : slow_wave2 ] shows a homoclinic orbit that makes one additional large excursion around @xmath163 after it was close to @xmath164 and then returns to @xmath167 ; hence we refer to the orbit in [ fig : slow_wave2 ] as a double pulse homoclinic orbit .
the same double pulse phenomenon exists for fast waves as well . in this case
the double pulse orbit has no additional interaction with the middle branch @xmath205 and therefore it is difficult to distinguish between different pulse types for fast waves numerically and graphically as the second loop follows the first one very closely .
this example demonstrates the use of our algorithm to compute trajectories in saddle - type slow manifolds of systems with two slow variables . the model is a caricature of a pair of neurons that are coupled with _ reciprocal inhibition _
the vector field is @xmath206 in this model , @xmath207 and @xmath208 are interpreted as the membrane potential of two neurons that are coupled synaptically through the terms involving @xmath209 .
the variables @xmath210 and @xmath211 represent the gating of membrane channels in the neurons .
the model is a caricature in that it does not incorporate the fast membrane currents which give rise to action potentials .
still more reduced models @xcite have been used to study reciprocal inhibition of a pair of neurons . reciprocal inhibition between a pair of identical neurons
has long been viewed as a mechanism for generating repetitive alternating activity in motor systems @xcite .
guckenheimer , hoffman and weckesser @xcite investigated the properties of this model when the two neurons have different parameters and therefore are not identical .
they observed that canards of several kinds were encountered while continuing periodic orbits with auto calculations .
the bifurcation mechanisms encountered in these continuation studies still have not been identified despite intensive efforts by lust @xcite to compute the multipliers of periodic orbits accurately .
our algorithm for computing invariant slow manifolds of saddle - type provides a promising new tool for investigating the bifurcations that take place in this system with two slow and two fast variables .
here we illustrate that the algorithm is indeed capable of computing trajectories that lie on these manifolds without pursuing bifurcation analysis of the system .
the periodic trajectory discussed in section 5.2 of @xcite has three different canard segments .
here we focus on the segment labelled @xmath212 in figure 6(c ) of @xcite .
the segment is a _ fold - initiated _ canard that begins as a fast trajectory , flows near a fold of the critical manifold and then moves along a saddle - type sheet of the slow manifold .
both neurons in the model have parameter values @xmath213 while @xmath214 and @xmath215 .
one of the points @xmath107 on the segment @xmath212 has coordinates @xmath216 .
we projected @xmath107 onto the critical manifold along the @xmath167 directions retaining the @xmath105 coordinates of @xmath107 and computed a trajectory @xmath217 of the slow flow on the critical manifold with this initial condition .
while the slow flow is an algebraic - differential equation , the critical manifold of is easily written as a graph of a function @xmath218 and the slow flow equations can be written as a vector field in @xmath105 .
the trajectory @xmath217 was taken as input for our algorithm .
boundary conditions were selected so that the initial point of the trajectory @xmath45 retains the same @xmath105 coordinates as @xmath107 .
figure [ rifig ] displays the trajectory @xmath45 obtained from our algorithm in black together with trajectories of its strong stable and unstable manifolds .
the distance of the initial conditions for the trajectories on the strong stable and unstable manifolds from @xmath45 is @xmath219 .
note that the first trajectories of the strong stable manifold at the bottom of the figure both flow down and to the right , reflecting that the initial points of these trajectories do not straddle the slow manifold in the strong stable direction .
similar behavior occurs at the final point of @xmath45 in the strong unstable direction .
this behavior is to be expected because the boundary conditions constrain the strong stable coordinate of the first point of @xmath45 to have a value close to that on the critical manifold rather than the invariant slow manifold . at the final point of @xmath45 ,
the strong unstable coordinate is determined by the critical manifold .
the behavior of @xmath45 is what we expect from our algorithm : the computed trajectory approaches the slow manifold of saddle - type along a strong stable direction at its beginning , flows along the slow manifold to a high degree of accuracy to near its end and then leaves the slow manifold along a strong unstable direction .
the length of @xmath45 is much longer than the segment @xmath212 shown in figure 6(c ) of @xcite . on a saddle - type slow manifold of system .
the trajectory is drawn black .
trajectories with initial conditions that are displaced by @xmath219 from @xmath45 along the direction of its strong stable manifolds are drawn red , while trajectories with initial conditions that are displaced by @xmath219 from @xmath45 along the direction of its strong unstable manifolds are drawn green .
, height=288 ] a l gorman and m v thomas ( 1978 ) changes in the intracellular concentration of free calcium ions in a pace - maker neurone , measured with the metallochromic indicator dye arsenazo iii , j physiol . 275:357 - 76 .
j. guckenheimer and d. lamar , periodic orbit continuation in multiple time scale systems .
numerical continuation methods for dynamical systems , 253267 , underst .
complex syst . , springer , dordrecht , 2007 .
e. hairer and g. wanner , solving ordinary differential equations .
stiff and differential - algebraic problems .
second edition .
springer series in computational mathematics , 14 .
springer - verlag , berlin , 1996 . | slow manifolds are important geometric structures in the state spaces of dynamical systems with multiple time scales .
this paper introduces an algorithm for computing trajectories on slow manifolds that are normally hyperbolic with both stable and unstable fast manifolds .
we present two examples of bifurcation problems where these manifolds play a key role and a third example in which saddle - type slow manifolds are part of a traveling wave profile of a partial differential equation .
initial value solvers are incapable of computing trajectories on saddle - type slow manifolds , so the slow manifold of saddle type ( smst ) algorithm presented here is formulated as a boundary value method .
we take an empirical approach here to assessing the accuracy and effectiveness of the algorithm . |
we consider an ensemble of bromotrifluoromethane ( cf@xmath0br ) molecules subjected to the light of an optical , nonresonant laser .
due to a dynamic second order stark effect , the electric field of the laser exerts a net torque on the molecular axis with the largest polarizability .
it is directed in such a way that it aligns the molecular axis with the linear polarization axis of the laser and thus aligns the molecules .
the laser intensity is too low to excite or ionize the molecules .
@xcite in combination with x rays , we have a so - called two - color problem ; the laser can be used in a pump - probe - way preceding the x rays or simultaneously with them .
the former case leads to impulsive ( transient ) alignment where a short pulse creates a rotational wavepacket which evolves freely afterwards and periodically goes through brief periods of alignment ; the latter case typically is used in conjunction with long laser pulses leading to adiabatic alignment where the molecular rotation follows the laser pulse envelope .
laser alignment has been extensively studied experimentally [ see , e.g. , ref . ] .
it was explained in the adiabatic case by friedrich and herschbach @xcite and in the case of pulsed lasers inducing rotational wavepacket dynamics leading to transient alignment by seideman .
@xcite the laser alignment of symmetric rotors was studied by hamilton _ et al . _
@xcite recently , we formulated a theory for laser - aligned symmetric - top molecules which are probed _ in situ _ by x rays ; it was applied to study bromine molecules .
@xcite in a joint experimental and theoretical investigation , the alignment of cf@xmath0br was studied for various conditions .
@xcite in this work , we would like to deepen the theoretical understanding of the x - ray probe of molecules .
especially , we are interested in trends .
resonant x - ray absorption is sensitive to molecular alignment , if the resonance has certain symmetry properties .
then selection rules can be established which lead to a different absorption for perpendicular and parallel x - ray polarization vectors with respect to the polarization axis of the laser . however , the absorption of x rays above the edge is not sensitive to molecular alignment .
rotational dynamics is conventionally specified in terms of the expectation value @xmath2 .
this value is a theoretical quantity .
experimentally , one typically has resorted to the coulomb explosion method to study alignment .
@xcite this method relies on the intricate nature of molecular strong - field ionization .
@xcite the x - ray absorption technique uses the simpler one - photon interaction and thus is physically simpler .
we address the following questions in this paper . in what way
is @xmath2 accessible from x - ray absorption measurements in a cross - correlation experiment ?
what is the achievable temporal resolution ?
we see our work in conjunction with the fascinating possibilities offered by upcoming ultrafast x - ray sources .
@xcite they will provide intriguing direct insights into the molecular rotational dynamics in the time domain . additionally , our research lays the foundation for molecular imaging of laser - aligned molecules using x - ray diffraction .
@xcite in addition to the analytical value of the x - ray absorption probe , laser - induced molecular alignment opens up a way to control x - ray absorption in an ultrafast ( picosecond ) way .
this means that by controlling the alignment of the molecules in a gas sample , we are able to control how much flux of an incident x - ray pulse is transmitted by the gas . by exploiting the available advanced optical technology to shape laser pulses ,
this offers a novel route to imprint shapes on the long x - ray pulses ( @xmath4 ) from third - generation light sources .
this paper is structured as follows .
section [ sec : theory ] discusses the x - ray absorption cross section of cf@xmath0br molecules for a two - level electronic structure model and its relation to @xmath2 ; a reduction of the cross section to experimentally easily accessible quantities ; and a model for a temperature dependent rotational period .
computational details are given in sec .
[ sec : compdet ] .
there , the dynamic dipole polarizability of cf@xmath0br and its anisotropy are also computed .
results are presented in sec .
[ sec : results ] for molecular rotational dynamics represented by @xmath2 and the x - ray absorption technique as well as the control of x rays by molecular rotation .
conclusions are drawn in sec .
[ sec : conclusion ] . in the appendix
, we show that symmetry - breaking effects are small in cf@xmath0br that modify the selection rules for resonant x - ray absorption .
our equations are formulated in atomic units , @xcite where @xmath5 is the unit of energy , @xmath6 is the unit of time , and @xmath7 is the unit of length .
the boltzmann constant @xmath8 is unity and @xmath9 is the unit of temperature .
intensities are given in units of @xmath10 and electric polarizabilities are measured in @xmath11 .
@xcite , width=188 ] our investigations are based on our theory of x - ray absorption by laser - aligned symmetric - top molecules in ref . .
there , we derive the following formula for the instantaneous x - ray absorption cross section at time @xmath12 of a molecule manipulated by a laser field : @xcite @xmath13 \ ; . \end{array}\ ] ] we denote the fine - structure constant by @xmath14 and the x - ray photon energy by @xmath15 .
the density matrix of the laser - only problem in terms of symmetric - rotor states is @xmath16 .
we use @xmath17 , @xmath18 , @xmath19 ( also with accents ) to denote the quantum numbers of symmetric - top wave functions @xcite and @xmath20 ( also with accents ) to numerate electronic states .
@xcite the electronic ground state is indicated by @xmath21 .
we assume the born - oppenheimer approximation and consider nuclear and electronic degrees of freedom to be totally separated from each other .
@xcite the matrix elements of the complex conjugate of the dipole operator @xcite in the body - fixed reference frame are indicated by @xmath22 . with @xmath23
, we denote the matrix elements of the x - ray polarization vector in the body - fixed frame in terms of symmetric - top wave functions .
the primes on the vectors indicate that they are given in terms of the spherical basis instead of the cartesian basis .
@xcite energies of symmetric - rotor states and electronic states are represented by @xmath24 , @xmath25 and @xmath26 , @xmath27 , respectively .
the finite decay width of core - excited states due to auger decay and x - ray fluorescence is described by @xmath28 .
@xcite a ) br molecule from which the x - ray absorption cross section is determined .
@xcite a ) the br@xmath29 orbital ; b ) the @xmath30 antibonding molecular orbital.,title="fig:",width=132 ] b ) br molecule from which the x - ray absorption cross section is determined .
@xcite a ) the br@xmath29 orbital ; b ) the @xmath30 antibonding molecular orbital.,title="fig:",width=132 ] the molecule cf@xmath0br ( the molecular structure in fig . [ fig : geomcf3br ] ) has an isolated resonance below the bromine @xmath18 edge in the x - ray absorption spectrum [ confer fig . 1 of ref . ] .
assuming that the x - ray photon energy is tuned to this resonance , a simplification of the electronic structure of the molecule to two levels is reasonable .
the pre - edge resonance is assigned to the transition from the br@xmath29 orbital to the @xmath30 antibonding molecular orbital .
see fig .
[ fig : twoorbs ] for a graphical representation of the two orbitals used .
resonant photoabsorption is sensitive to the alignment of cf@xmath0br molecules , because the orbitals are glued to the molecular frame and thus rotate with it .
bromotrifluoromethane has c@xmath31 symmetry ; the br@xmath29 and @xmath30 orbitals have @xmath32 symmetry . the transition is mediated by the cartesian dipole matrix element @xmath33 with @xmath34 in the molecule - fixed reference frame . due to the vanishing integrals rule , @xcite the direct product of representations @xmath35 with @xmath36 needs to contain the totally symmetric representation such that @xmath37 , i.e. , the transition @xmath38 is allowed via @xmath39 . using the product table of the c@xmath31 point group ,
@xcite we find that the @xmath40 component of @xmath41 is totally symmetric but the components @xmath42 and @xmath43 are not .
this gives rise to a dichroism with respect to the figure axis ( c br ) of cf@xmath0br : absorption for parallel x - ray polarization vector , no absorption for perpendicular x ray polarization vector . in the appendix
, we show that this result holds in spite of potential symmetry - breaking effects in cf@xmath0br . for our two - level model
, we have @xmath44 with the unit vectors @xmath45 and @xmath46 along the @xmath40 and the @xmath47 axes , respectively .
@xcite we neglect the dependence of the denominator in eq .
( [ eq : xsect ] ) on rotational quantum numbers .
this is justified because , in view of the large inner - shell decay widths , rotational states remain unresolved in x - ray absorption .
let us decompose the fraction into a real part and an imaginary part as follows : @xmath48 we tune the x rays to the resonance energy , i.e. , @xmath49 .
thus , we find from eq .
( [ eq:2levfrac ] ) @xmath50 and @xmath51 .
the resonant cross section ( [ eq : xsect ] ) of a two - level system becomes : @xmath52 with @xmath53 and @xmath54 .
the initial , thermal distribution over rotational states is sensitive to the nuclear spin statistics .
@xcite the three fluorine atoms in cf@xmath0br , are @xmath55f , the only stable fluorine isotope , with nuclear spin quantum number @xmath56 .
@xcite in this case , the statistical weights of the symmetric rotor are given by @xcite @xmath57 in most of the computations of this paper , we use a linear model for cf@xmath0br . in this case
, the cf@xmath0 portion of cf@xmath0br is assumed to be a pseudo - atom which is distinct from a bromine atom .
thus the nuclear statistical weight is in this case : @xmath58 there is no quantum number @xmath18 for a linear rotor .
@xcite the instantaneous cross section ( [ eq : xsect ] ) is not easily experimentally accessible .
however , we can use it to devise quantities which can be measured more readily .
the laser pulse is imprinted on the cross section of the two - level model @xmath59 [ eq .
( [ eq : xsecttwo ] ) ] .
we determine it for parallel ( @xmath60 ) and perpendicular ( @xmath61 ) x - ray and laser polarization vectors .
additionally , we consider the cross section without laser @xmath62 , i.e. , a thermal ensemble .
the x rays are characterized by the photon flux @xmath63 .
the cross correlation of the cross section and the x - ray pulse is defined by @xmath64 it represents the total probability of x - ray absorption for a time delay of @xmath65 between the laser and the x - ray pulse .
we define following ratios of cross correlations : [ eq : crossratios ] r _ ( ) & = & , + r _ ( ) & = & , + r _ ( ) & = & .
the ratios were studied experimentally in ref . .
let us assume ultrashort x - ray pulses which can be modeled by a @xmath66 distribution as follows : @xmath67 inserting this flux into eq .
( [ eq : crosscorr ] ) yields @xmath68 . when the ratios ( [ eq : crossratios ] ) are formed , the constant @xmath69 cancels and the ratios are given by the ratios of the participating cross sections .
hence , in the limit of ultrashort x - ray pulse duration , eq .
( [ eq : crossratios ] ) , goes over into the new ratios [ eq : crossratioscross ] r _ ( ) & = & , + r _ ( ) & = & , + r _ ( ) & = & .
we set out from the coupling matrix elements in eq .
( [ eq : xsect ] ) reading @xcite @xmath70 the x - ray polarization vector is @xmath71 and @xmath41 denotes the dipole operator .
both vectors are cartesian vectors in the laboratory reference frame . the scalar product on the left hand side of eq . ( [ eq : cplmat ] ) is formed in the spherical basis .
@xcite with @xmath72 we denote the direct product of the symmetric - rotor state @xmath73 and the electronic state @xmath74 .
let us assume the two - level model of sec .
[ sec : elstructmod ] ; the only dipole vector @xmath75 and the polarization vector @xmath71 are real vectors . the scalar product on the right hand side of eq . ( [ eq : cplmat ] ) becomes @xmath76 we use @xmath77 and @xmath78 is the angle between the two vectors @xmath75 and @xmath71 .
we choose @xmath79 along laser polarization axis . the dipole vector @xmath75here given in the laboratory frame
points along the c
br molecular axis ; in the body - fixed reference frame the c br axis was chosen to be the @xmath40 axis [ sec . [ sec : elstructmod ] ] .
thus , @xmath78 is the euler angle between the @xmath80 and @xmath40-axes of the space - fixed and molecule - fixed frame , respectively . the dependence of the denominator in eq .
( [ eq : xsect ] ) on rotational quantum numbers is disregarded as in eq .
( [ eq:2levfrac ] ) .
the sum over intermediate symmetric - rotor states can then be eliminated totally yielding @xmath81 in the numerator of eq .
( [ eq : xsect ] ) .
the remaining summation forms a trace with these matrix elements and the density matrix : @xmath82 finally , we arrive at the cross section of the two - level model ( [ eq : xsecttwo ] ) which clearly exhibits the relation to @xmath2 as follows : @xmath83 assuming an x - ray polarization vector perpendicular to the laser - polarization axis yields @xmath84 \ ; .\ ] ] the prefactor of @xmath85 in @xmath86 is only half the prefactor of @xmath2 in @xmath87 because there are two perpendicular directions to the laser - polarization axis compared to only one for parallel polarization vectors . the cross section of a thermal ensemble can be inferred easily from eq .
( [ eq : xsecttwocospar ] ) by letting @xmath88 which leads to @xmath89 given the formulas for the cross sections ( [ eq : xsecttwocospar ] ) , ( [ eq : xsecttwocosperp ] ) , and ( [ eq : xsecttwocostherm ] ) , we can write explicit expressions for the ratios in eq .
( [ eq : crossratioscross ] ) as follows : [ eq : crossratiostwo ] r _ ( ) & = & , + r _ ( ) & = & 3 ( ) , + r _ ( ) & = & [ 1 - ( ) ] .
the character of molecular alignment impulsive to adiabatic see , e.g. , fig .
[ fig : cos2t_7 ] and the ensuing discussion crucially depends on the rotational temperature @xmath90 .
@xcite conventionally , the rotational period of a linear rotor is defined by @xmath91 with the rotational constant @xmath92 .
the regimes of molecular alignment from impulsive to adiabatic are distinguished by comparing @xmath93 with the duration of the aligning laser pulse @xmath94 .
yet , as @xmath93 is independent of @xmath90 , the classification thus does not vary with @xmath90 , despite the changed alignment characteristics . [ cols=">,<,^,^,^,^,^,^ " , ] overlaid on the curves for the linear rotor are data from a symmetric rotor computation .
the deviations between both descriptions of cf@xmath0br are minute . for @xmath95 , there is no difference between both because only the rotational ground state with @xmath96 is populated .
as the interaction with the laser does not mix states with different @xmath18 quantum numbers , @xcite no difference may arise when a laser is turned on . also for higher @xmath90 , the deviation remain tiny .
we show curves for the linear rotor as solid ( black ) lines and curves for the symmetric rotor as green ( dashed ) lines .
the data are for various temperatures given by the numbers in blue .
we use laser pulses with a fwhm duration of @xmath97 . ] in fig .
[ fig : intensdep_4 ] , we investigate the influence of the laser intensity on the maximally attainable alignment .
we take the maximum of @xmath2 over the propagation time interval for a number of temperatures and laser intensities .
the computational parameters are chosen such that we are in the adiabatic regime .
for all the curves , we have the following two limits : if the laser intensity is zero , there is no alignment , only a thermal ensemble and @xmath98 .
conversely , if @xmath99 , then @xmath100 .
of course , the latter limit can never be approached closely in practice because molecules are ionized above a certain value for the intensity . with increasing temperature , the curves in fig .
[ fig : intensdep_4 ] change dramatically . for @xmath101
, we find a strong nonlinear dependence of @xmath102 on the intensity .
for @xmath103 , the dependence is essentially linear .
in analogy to the discussion of fig .
[ fig : cos2t_8 ] , we conclude that the strong temperature dependence of the functional shape of @xmath102 provides a means to determine the rotational temperature of a gas sample , if the laser intensity is accurately known .
the expectation value @xmath2 examined in the previous sec .
[ sec : lasalign ] can be inferred experimentally via the coulomb explosion technique . however , due to the complicated nature of the interaction of a laser with a molecule , this experimental technique is a somewhat intricate way to study molecular alignment ; e.g. , recently , the strong - field ionization probability of aligned molecules was measured and a nontrivial , molecule - specific angular dependence was found .
@xcite therefore , this technique is prone to systematic errors @xcite and it is highly desirable to have an alternative method for the measurement of molecular alignment .
we advocate resonant absorption of polarized x rays as a novel route to study molecular alignment .
@xcite br from fig .
[ fig : cos2t_7 ] with a picosecond x - ray source ( @xmath104 ) .
the dependence of @xmath105 [ eq . ( [ eq : crossparperp ] ) ] on the time delay @xmath65 between laser and x - ray pulses is shown . ] in fig .
[ fig : cc1ps ] , we study the four regimes of molecular alignment from fig . [
fig : cos2t_7 ] using picosecond x - ray pulses ( @xmath104 ) .
the x - ray pulse duration was chosen much shorter than the typical rotational time scale @xmath93 .
we investigate the amount of information on the rotational dynamics which is provided by the ratio @xmath105 [ eq . ( [ eq : crossparperp ] ) ] because it is an experimentally accessible quantity which has been measured successfully before by peterson _ et al .
_ @xcite the dependence of @xmath105 [ eq .
( [ eq : crossparperp ] ) ] on the laser pulse is shown for a large range of time delays @xmath65 between laser and x - ray pulses . with the picosecond pulses ,
the molecular dynamics is reproduced clearly in fig .
[ fig : cc1ps ] ; even the detailed features of fig .
[ fig : cos2t_7 ] are resolved .
this shows that short - pulse x - ray absorption is a promising novel avenue to study the details of molecular rotations .
br studied by x - ray absorption .
the setup is the one of the top panel in figs . [
fig : cos2t_7 ] and [ fig : cc1ps ] .
the ratio @xmath105 [ eq . ( [ eq : crossparperp ] ) ] is shown here .
we use @xmath106 ( top ) and @xmath107 ( bottom ) .
other computational parameters are @xmath108 and @xmath109 . ] in fig .
[ fig : cc50fs_100ps ] , we investigate the impact of the duration of the x - ray pulse on the absorption @xmath105 [ eq .
( [ eq : crossparperp ] ) ] .
we find that the curve in the top panel of fig . [
fig : cc50fs_100ps ] for @xmath106 can hardly be distinguished from the curve for @xmath105 in the top panel of fig .
[ fig : cc1ps ] for @xmath104 .
this indicates that picosecond x - ray pulses are sufficiently short to deliver snapshots without averaging noticeably over the rotation of the molecule . in other words , for cf@xmath0br picosecond x - ray pulses
can be modeled as a @xmath110 distribution ( [ eq : deltaflux ] ) and one obtains @xmath111 [ eq .
( [ eq : crossparperpcross ] ) ] . in the lower panel of fig .
[ fig : cc50fs_100ps ] , we use long x - ray pulses @xmath107 , instead , to determine @xmath105 .
we observe that the detailed information on the molecular dynamics revealed in the upper panel is basically invisible in the lower panel .
the long x - ray pulse averages out all fine details . to the cross correlation ratios @xmath112 ( dashed red ) , @xmath113 ( solid black ) , and @xmath114 ( dash - dot blue ) [ eq .
( [ eq : crossratios ] ) ] for cf@xmath0br .
the long dashed ( green ) vertical line indicates the value of a thermal ensemble .
other computational parameters are @xmath115 , @xmath97 , and @xmath116 . ] comparing the details of the curves in fig .
[ fig : cc1ps ] with the corresponding curves from fig .
[ fig : cos2t_7 ] reveals a vertical distortion .
consider , for instance , the bottom panel in fig .
[ fig : cos2t_7 ] . while the curve is basically flat around the peak , the corresponding curve in fig .
[ fig : cc1ps ] exhibits a clear peak .
the differences between the curves can be understood by examining the relation between @xmath2 and @xmath105 ( we assume a @xmath66-distribution x - ray pulse ( [ eq : deltaflux ] ) and equate @xmath12 and @xmath65 here ) . for this purposes
, we carried out absorption calculations for adiabatically aligned molecules for a series of laser intensities . in fig .
[ fig : cos2t_ratio ] , we plot @xmath105 versus @xmath2 . to this end , we use the maximum of the two quantities over the propagation interval of individual computations with varying laser intensity .
the curve reveals the nonlinear relationship between both quantities , see eq .
( [ eq : crossparperptwo ] ) . in transient alignment ,
e.g. , the top panel of fig .
[ fig : cos2t_7 ] , there is also a suppression of alignment with respect to the signal of a thermal ensemble due to antialignment ( i.e. , alignment perpendicular to the laser polarization axis ) .
we can easily add this part of the curve to fig .
[ fig : cos2t_ratio ] by realizing that at @xmath117 , the molecules are perfectly antialigned . according to sec .
[ sec : elstructmod ] , there is no absorption of x rays with a polarization parallel to the laser polarization axis .
we interpolate linearly between this point and the value unity for a thermal ensemble .
in addition to a curve for @xmath105 , we also show results for @xmath113 and @xmath118 in fig .
[ fig : cos2t_ratio ] .
these two mappings of x - ray absorption ratios to @xmath2 are linear , see eqs .
( [ eq : crossparthtwo ] ) and ( [ eq : crossperpthtwo ] ) .
they are opposite to each other : where the one is maximal , the other is minimal . yet the maximum of @xmath118 is only half the maximum of @xmath113 .
this is due to the fact that there are two axes perpendicular to the laser polarization axis but only one parallel to it .
the curves were extended into the range from @xmath119 to the value of a thermal ensemble @xmath120 in the same way as for @xmath105 . for @xmath113
, we have the two values 0 and 1 , respectively .
the curve for @xmath118 certainly assumes the value unity for a thermal ensemble as well .
however , in @xmath117 , we used the maximum value of @xmath113 divided by 2 .
the signature of molecular alignment from the ratios @xmath118 and @xmath113 is weaker than from @xmath105 .
the latter ratio is thus preferable in experiments .
( [ eq : crossparperp ] ) ] of cf@xmath0br on the rotational temperature of the molecules .
the solid ( black ) curve is for @xmath121 and @xmath107 whereas the dashed ( red ) curve is for @xmath97 and @xmath116 . for both curves ,
the peak laser intensity is @xmath122 . ] in fig .
[ fig : r1_160ps ] , we investigate the dependence of the x - ray absorption signal @xmath105 [ eq . ( [ eq : crossparperp ] ) ] on the rotational temperature . to this end , we computed the absorption of adiabatically aligned molecules for a series of temperatures . the high sensitivity of the molecular dynamics on the temperature revealed in fig .
[ fig : r1_160ps ] parallels the one found in fig .
[ fig : cos2t_8 ] .
our prediction stresses once more how important a low temperature is to observe a substantial signal .
conversely , the maximum of the x - ray absorption ratio can again be used to determine the rotational temperature of a gas sample . for this purpose ,
the precise laser and x - ray pulse parameters need to be known to tailor a figure like fig .
[ fig : r1_160ps ] for a given experimental situation .
br molecules . in the upper panel ,
we show the induced molecular dynamics subject to the sequence of 50 laser pulses with intensity @xmath123 which is shown in the lower panel .
the individual pulses have a fwhm duration of @xmath124 and are spaced by @xmath125 .
the curves in the upper panel represent instantaneous ratios of cross sections : @xmath126 [ eq . ( [ eq : crossparthcross ] ) ] ( dashed , red ) and @xmath127 [ eq . ( [ eq : crossperpthcross ] ) ] ( solid , black ) .
the rotational temperature is @xmath128 . ] in the previous sec .
[ sec : xrabsi ] , we used x - ray absorption to study the rotational molecular dynamics of a gas sample . in this subsection , we focus on the inverse process : the control of x - ray absorption by molecular alignment .
controlling x - ray pulse shapes represents a powerful tool to control inner - shell electronic processes ; it has been theoretically established in terms of electromagnetically induced transparency ( eit ) for x rays .
@xcite there we use a shaped laser pulse to cut out two short x - ray pulses from a long gaussian x - ray pulse . here
, we would like to investigate similar avenues using laser - aligned molecules .
the control of molecular rotation with appropriately shaped laser pulses has been investigated , e.g. , in ref . .
the x - ray absorption cross section of the molecules depends on the alignment [ sec .
[ sec : elstructmod ] ] .
we assume that the gas sample absorbs most x rays for randomly oriented molecules .
once the molecules are sufficiently antialigned , a large fraction of the incident x - ray flux is transmitted through the gas sample for parallel laser and x - ray polarization vectors .
let us examine examples of controlled x - ray absorption . in the upper panel fig .
[ fig : control ] , we show the ratios of cross sections @xmath126 [ eq . ( [ eq : crossparthcross ] ) ] and @xmath127 [ eq . ( [ eq : crossperpthcross ] ) ] of cf@xmath0br for parallel and perpendicular x - ray and laser polarizations , respectively .
the molecules are subject to the sequence of 50 laser pulses shown in the lower panel . as derived in sec .
[ sec : relcrosscos ] , @xmath87 [ eq . ( [ eq : xsecttwocospar ] ) ] is proportional to @xmath2 and thus is well - suited to examine the molecular motion .
the sequence imprints an oscillatory occurrence and vanishing of enhanced alignment and antialignment .
this rotational motion is enforced by the laser and only present during the pulse sequence .
afterwards , @xmath129 , we see by comparison with the upper panel in fig .
[ fig : cos2t_7 ] ( keep in mind that @xmath2 is plotted there)that the free time - evolution after a single pulse is approximately assumed .
the ratios @xmath126 and @xmath127 are complementary : the former peaks where the latter is minimal . likewise , the ratios of cross sections corresponding to the physical situation in the top panel of fig .
[ fig : cos2t_7 ] can be investigated where a single , short laser pulse creates a rotational wavepacket . about every @xmath130 ,
the molecule undergoes short time intervals of transient antialignment .
the brief periods of antialignment then lead to brief periods of transparency of the sample for x rays , i.e. , x - ray bursts of a few picoseconds may be cut out of a longer x - ray pulse . we have constructed a picosecond clock which provides us with short x - ray bursts .
for very long x - ray pulses , several bursts with a well - defined recurrence time can be generated .
control of x - ray absorption with laser - aligned molecules has advantages and disadvantages compared with the competing technique of eit for x rays .
@xcite an advantage of molecular alignment over eit is that by the choice of the molecule , x - ray absorption can be controlled for a large number of wavelengths , i.e. , by replacing cf@xmath0br with other molecules , one can shift the absorption edge over a wide range of energies .
additionally , other edges , @xmath131 , @xmath19 , @xmath132 , @xmath133 , @xmath134 , @xmath135 , can be used alternatively without any extra effort .
this adds an additional degree of freedom of available x - ray wavelengths . in the case of eit for x rays , on the other hand
, the number of suitable atoms seems to be quite limited by the requirements to sustain strong laser fields and the availability of intense laser systems with the right wavelength to couple two empty rydberg orbitals .
also with molecular alignment , absorption can be controlled for hard x rays where @xmath18-shell core - hole decay widths are large and eit for x rays is suppressed .
@xcite let us estimate the required length of cf@xmath0br gas where the laser and the x rays need to overlap for the proposed x - ray pulse shaping to work .
we approximate the cross section of cf@xmath0br on the pre - edge resonance by the @xmath136 cross section for krypton atoms @xmath137 .
@xcite let the gas jet have a number density of @xmath138 as it had in the experiment of ref . . from beer s law ,
@xcite we estimate that the overlap region needs to be @xmath139 long for the x - ray flux to drop by a factor of @xmath140 .
clearly , this is an absurd requirement in practice .
a number density of @xmath141 would be required to reduce the length of the overlap region to @xmath142 . yet
the number density in the gas jet can not be increased indefinitely because the molecules start to interact strongly with each other forming clusters and , eventually , a condensed phase .
the x - ray cross section assumed in this estimate holds for perfectly aligned molecules parallel to the x - ray polarization direction . with laser - alignment , this cross section can be reduced to zero for perfectly antialigned molecules .
note that the cross section is about a hundred times larger for neon atoms than for krypton atoms .
@xcite the estimate with respect to the fluorine pre - edge resonance is thus far more favorable than for bromine . on the down side of molecular alignment over eit
is that the time scales of the former is about three orders of magnitude slower ( picoseconds ) than the latter ( femtoseconds ) .
the molecules need to have a low rotational temperatures , too ; it is not yet clear whether this is physically possible because interactions between molecules have a considerable damping influence on molecular rotations in denser gases and condensed matter .
@xcite the latter points seem to represent a big , if not insurmountable , obstacle for control of x - ray absorption by laser - manipulated molecular rotation
. however , the ideas presented in this subsection might nevertheless prove fruitful in conjunction with recent work on the laser - alignment of molecules in solution . @xcite
this study is devoted to laser - induced rotational dynamics of bromotrifluoromethane ( cf@xmath0br ) molecules probed by x rays .
first , we investigated the molecular rotations in terms of the expectation value @xmath2 .
second , we investigate the properties of a probe of molecular rotations by x rays .
this represents an alternative to the established coulomb - explosion technique .
@xcite third , we explore possibilities to employ laser manipulation of molecular rotation to control x - ray pulse shapes , i.e. , to imprint a certain pulse shape onto a longer x - ray pulse .
we start by investigating @xmath2 . specifically , we examine its temporal evolution for varying laser - pulse durations , its temperature dependence , and its dependence on the peak laser intensity .
then , we revisit these physical situations with short - pulse x rays . additionally , we explore the dependence on the x - ray pulse duration and the relation between the maximum of x - ray absorption and the maximum of @xmath2 .
we show that @xmath2 can be directly measured by short - pulse x - ray absorption measurements .
such short pulses will be produced by the emerging ultrafast x - ray sources .
@xcite the method permits to stroboscopically photograph the steps in the time - evolution of rotational wavepackets . we can learn about the beams and the experimental conditions ; x - ray absorption is an _ in situ _ probe .
our investigations are based on our recent theory @xcite for the laser alignment of symmetric - top molecules probed by x rays . to carry out the computations in this paper , we devise a parameter - free two - level model for the electronic structure of cf@xmath0br
; it models the br@xmath143 pre - edge resonance . the x - ray absorption cross section formula from ref .
is reduced to this model .
we employ coupled - cluster linear - response techniques to determine the average dynamic dipole polarizability and the dynamic dipole polarizability anisotropy for cf@xmath0br in the laser light . finally , the usual temperature - independent criterion of adiabaticity for the laser pulse duration @xmath144 looses its predictive power because the rotational response becomes more and more adiabatic with increasing rotational temperature .
therefore , we devise a thermally averaged rotational period as a new adiabaticity criterion for rotational dynamics .
based on our work , a number of future prospects offer themselves .
absolute cross sections can be easily determined parametrizing the two - level model fully with the cross section on the br@xmath143 pre - edge resonance for molecules perfectly aligned along the x - ray polarization axis .
this parameter can be experimentally determined for clamped molecules on surfaces @xcite or computationally using _ ab initio _ methods .
our rigid - rotor model has a few limitations .
no vibrations or internal rotations were accounted for . in future work
, such effects beyond the rigid - rotor approximation should be addressed .
additionally , our formalism can be extended to three - dimensional alignment of asymmetric - top molecules .
@xcite this enables one to study a large class of molecules which basically do not rotate .
our proposal to use laser - aligned molecules to control x - ray pulse shapes offers a higher flexibility in terms of the suitable x - ray energies than electromagnetically induced transparency for x rays which has been studied only for rare - gas atoms so far .
@xcite chemical shifts of the inner - shell edges offer an added flexibility to manipulate the resonance energies of molecules compared with atoms .
to control x rays by laser - aligned molecules in practice , one has to experimentally produce a cold beam of molecules with sufficiently high number density .
however , at present , this requirement seems to inhibit a success of this technique both in principle and in practice .
further , one has to make theoretical advancements to predict the signal of such experiments . at the required number densities ,
one faces condensed - matter ; the interactions with neighboring molecules are strong and need to be incorporated in terms of a more comprehensive theory . @xcite the laser pulses need to be shaped such that a desired rotational dynamics is imposed on the molecules .
approaches to the quantum control of molecular alignment have been derived , e.g. , refs .
offering perspectives for an optimal control of the rotational dynamics of molecules and , in this way , an optimal , ultrafast ( picoseconds ) control of x - ray pulse shapes .
we would like to thank linda young for fruitful discussions .
c.b.s research was partly funded by a feodor lynen research fellowship from the alexander von humboldt foundation . c.b.s and
r.s.s work was supported by the office of basic energy sciences , office of science , u.s .
department of energy , under contract no .
de - ac02 - 06ch11357 .
in sec . [ sec : elstructmod ] we argued that because of the c@xmath31 symmetry of cf@xmath0br , the transition dipole vector between the br@xmath29 and the @xmath30 orbitals has only a nonzero @xmath40-component .
two basic mechanisms can lead to a failure of this prediction .
the first mechanism is symmetry breaking induced by molecular vibrations .
we consider this effect to be negligible for the following reason .
laser - induced alignment in the gas phase requires rotationally cold molecules .
we may therefore assume that before absorbing an x - ray photon , the molecules are in the electronic and vibrational ground state .
@xcite hence , it is only vibrations in the core - excited resonance state that could break the c@xmath31 symmetry of cf@xmath0br .
since a @xmath145 vacancy in a bromine atom has a decay width of @xcite @xmath146 and the chemical environment has only a negligible influence on this core - hole decay , the lifetime of a br@xmath143 resonance state of cf@xmath0br can be inferred to be only @xmath147 .
this time is insufficient for symmetry - breaking deformations .
the second mechanism is symmetry breaking induced by spin - orbit coupling in bromine . to estimate the impact of this effect
, we employ a simple model based on the following four classes of normalized atomic basis functions @xmath148 with @xmath149 .
we denote by @xmath150 a pauli spinor with spin projection quantum number @xmath151 . here
, @xmath152 represents a @xmath153 hybrid orbital centered on carbon and directed towards bromine .
the br@xmath154 orbitals are @xmath155 , @xmath156 , @xmath157 with orbital angular momentum projection quantum numbers @xmath158 , respectively .
let @xmath159 stand for the spin - orbit - free fock operator associated with the closed - shell electronic ground state of cf@xmath0br at equilibrium geometry .
molecular orbitals @xmath160in the absence of spin - orbit coupling have the spin - independent orbital energies @xmath161 .
we assume that the molecular orbitals of interest may be written as superpositions of the atomic orbitals @xmath162 .
specifically , for the doubly occupied @xmath163 orbital ( homo-1 ) , @xmath164 , and the unoccupied @xmath165 orbital ( lumo ) , @xmath166 , we write @xmath167 the degenerate , doubly occupied lone pairs ( homo ) @xmath168 are approximated by @xmath169 we form the effective one - electron hamiltonian @xmath170 with the spin - orbit coupling @xmath171 @xcite .
using first - order perturbation theory , the @xmath165 orbital in the presence of spin - orbit coupling is @xmath172 the notation @xmath173 shall not imply that @xmath174 is a good quantum number ; it merely indicates that @xmath173 becomes @xmath175 in the limit of vanishing spin - orbit interaction .
spin - orbit coupling is associated with the atomic potential near the bromine nucleus .
we therefore assume that the @xmath153 hybrid orbital on carbon is unaffected by spin - orbit coupling and set @xmath176 with @xmath149 and @xmath177 .
however , @xmath178 causes mixing between the br@xmath154 orbitals @xmath179 , @xmath180 , and @xmath181 . for an atom
, @xmath178 is proportional to @xcite @xmath182 , where @xmath183 and @xmath184 are the orbital and spin angular momentum operators .
thus , @xmath178 is diagonal with respect to the spin - orbit coupled states @xcite @xmath185 in this expression , @xmath186 is a clebsch - gordan coefficient , @xcite @xmath187 or @xmath188 , and @xmath189 . upon inverting eq .
( [ a13 ] ) , @xcite we obtain for the spin - orbit coupling matrix elements in the atomic basis @xmath190 where @xcite @xmath191 and @xmath192 is the fine - structure splitting in the valence shell of atomic bromine .
expression ( [ a14])in combination with eqs .
( [ a6 ] ) , ( [ a8 ] ) , and ( [ a12])justifies that terms involving matrix elements of the form @xmath193 for @xmath194 were excluded from eq . ( [ a11 ] ) . hence , upon collecting results , the spin - orbit coupled @xmath165 orbital may be written in a transparent form : @xmath195 we have introduced @xmath196 using eqs .
( [ a6 ] ) and ( [ a8 ] ) to denote the homo - lumo gap . employing standard angular momentum algebra @xcite and exploiting the fact that the br@xmath29 orbital @xmath197 has practically no overlap with the @xmath153 hybrid orbital on carbon , the x - ray absorption probability at
the br@xmath198 resonance is proportional to @xmath199 \biggr ] \ ; .
\end{array}\ ] ] here , @xmath41 is the electric dipole operator , @xmath71 is the x - ray polarization vector , @xmath200 is a reduced matrix element , @xcite and @xmath201 , @xmath202 , and @xmath203 are the cartesian components of @xmath71 in the body - fixed frame , assuming that the molecular symmetry axis is the @xmath40 axis .
x - ray polarization parallel to the cf@xmath0br symmetry axis thus corresponds to @xmath204 and @xmath205 ; x - ray polarization perpendicular to the cf@xmath0br symmetry axis corresponds to @xmath206 and @xmath207 .
it follows from eq .
( [ a18 ] ) that , assuming perfect alignment of the molecule relative to the x - ray polarization axis , the ratio between x - ray absorption in the parallel configuration ( @xmath208 ) and x - ray absorption in the perpendicular configuration ( @xmath209 ) is given by @xmath210 the spin - orbit splitting @xmath192 in a bromine atom is @xmath211 .
@xcite using dalton @xcite and the aug - cc - pvtz basis set , @xcite we find that the homo - lumo gap @xmath212 in cf@xmath0br is @xmath213 .
we may therefore conclude from eq .
( [ a19 ] ) that for perfect alignment , @xmath214 .
this upper limit is much higher than any of the values for @xmath215 calculated in sec .
[ sec : results ] . in comparison
to the restrictions imposed on @xmath215 by laser - induced alignment , the impact of spin - orbit coupling is negligible . | we present a computational study of the rotational molecular dynamics of bromotrifluoromethane ( cf@xmath0br ) molecules in gas phase .
the rotation is manipulated with an offresonant , @xmath1 laser .
the molecules are treated as rigid rotors .
frequently , we use a computationally efficient linear rotor model for cf@xmath0br which we compare with selected results for full symmetric - rotor computations .
the expectation value @xmath2 is discussed . especially , the transition from impulsive to adiabatic alignment , the temperature dependence of the maximally achievable alignment and its intensity dependence are investigated . in a next step ,
we examine resonant x - ray absorption as an accurate tool to study laser manipulation of molecular rotation .
specifically , we investigate the impact of the x - ray pulse duration on the signal ( particularly its temporal resolution ) , and study the temperature dependence of the achievable absorption .
most importantly , we demonstrated that using picosecond x - ray pulses , one can accurately measure the expectation value @xmath3 for impulsively aligned cf@xmath0br molecules .
we point out that a control of the rotational dynamics opens up a novel way to imprint shapes onto long x - ray pulses on a picosecond time scale . for our computations ,
we determine the dynamic polarizability tensor of cf@xmath0br using _ ab initio _ molecular linear - response theory in conjunction with wave function models of increasing sophistication : coupled - cluster singles ( ccs ) , second - order approximate coupled - cluster singles and doubles ( cc2 ) , and coupled - cluster singles and doubles ( ccsd ) . .
copyright 2008 american institute of physics .
this article may be downloaded for personal use only .
any other use requires prior permission of the author and the american institute of physics . ] |
recently , a chinese group has raised the proposal of constructing a high luminosity @xmath0 collider in china , the so - called super @xmath3-factory @xcite , which is similar to the gigaz program suggested by the internal linear collider collaboration @xcite but with a even higher luminosity . as for this
newly purposed super @xmath3-factory , a @xmath0 collider shall run at the energies around the mass of @xmath1-boson resonance and with a high luminosity up to @xmath2 .
many interesting topics can be studied in this super @xmath3-factory : the properties of @xmath1-boson and the precision tests of the electro - weak sector of the standard model , the properties of the heavy mesons or baryons , @xmath22 physics , the physics beyond the standard model and etc .
the @xmath16 meson has aroused people s great interests since its discovery by cdf collaboration @xcite .
the @xmath16 meson is a doubly heavy quark - antiquark bound state and carries flavors explicitly .
systematic studies of its production properties at the hadronic colliders tevatron and lhc have been done in the literature , e.g. refs.@xcite . in particular , a computer program bcvegpy for the hadronic production of the @xmath16 meson has been completed and developed in refs.@xcite , which can be conveniently implemented into pythia @xcite for simulating @xmath16 events . very recently , a first observation of the decay channel @xmath23 with the help of bcvegpy by lhcb group has been finished @xcite . at the hadronic colliders ,
there is much pollution from the hadronic background and many produced @xmath16 events shall be cut off by the trigging condition @xcite .
so , some alternative measurements would be helpful for a comprehensive study .
comparing to the hadronic colliders as the tevatron and lhc , a @xmath24 collider has three general features as cleanliness , democracy and holism @xcite , which are helpful and have some advantages to perform more precise measurements for certain processes . to seek the @xmath16 events at a @xmath0 collider
has already been tried by the lep - i experiment , however no @xmath16 events have been found there .
if the luminosity of a @xmath24 collider is raised up to @xmath25 and its colliding energy is confined to be around the @xmath1-peak , it will raise the production rate up to several orders .
this is reasonable , since as estimated in ref.@xcite , more than @xmath26 @xmath1-boson resonance can be produced for gigaz with @xmath27 .
moreover , refs.@xcite have shown that sizable @xmath16 events can be produce through the @xmath1-boson decays . then , as a rough combination , one can estimate that such a super @xmath3-factory will open new opportunities for studying the @xmath16-meson properties . and to be useful reference , it is interesting to present a more sound estimation . under the non - relativistic qcd ( nrqcd )
framework @xcite , a doubly heavy meson is considered as an expansion of fock states , and the probability of finding each fock state in the expansion is proportional to a definite power of @xmath28 , where @xmath28 is the relative velocity and @xmath29 is around @xmath30 for heavy quarkonium .
we have studied the production of @xmath31- and @xmath32- quarkonium with a photon or via initial state radiation at the @xmath1-peak in the super @xmath3-factory @xcite .
it is interesting to see whether there has enough @xmath16 events produced at such a high luminosity @xmath0 collider , which can be produced through the process @xmath33+b+\bar{c}$ ] , where @xmath6 $ ] stands for the @xmath4-quarkonium fock states .
excited @xmath4-quarkonium states , such as the color - singlet @xmath21-wave states and the color - octet @xmath34-wave states , may directly or indirectly decay to the ground state with almost 100% possibility via electromagnetic or hadronic interactions . then
, as a full estimation of the @xmath16-meson production , we shall discuss the following @xmath4-quarkonium states simultaneously ; i.e. @xmath7\rangle$ ] , @xmath8g\rangle$ ] , @xmath9\rangle$ ] , @xmath10g\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ] ( with @xmath13 ) . here
the thickened subscript of @xmath4-quarkonium denotes the color index , * 1 * for color - singlet and * 8 * for color - octet ; the relevant angular momentum quantum numbers are shown in the parentheses accordingly . because of the emergence of massive - fermion lines , the analytical expression for the squared amplitude becomes too complex and lengthy for more ( massive ) particles in the final states and for more higher - order fock states to be generated .
for example , to derive the amplitudes for the @xmath21-wave states , one also needs to get the derivative of the amplitudes over the relative momentum of the constitute quarks .
it has been found that to do the numerical calculation using the conventional squared - amplitude technology becomes time - consuming for these complex processes , since the cross - terms increases with the increment of feynman diagrams , @xmath35 , where @xmath36 stands for the number of feynman diagrams for the process .
one important way to solve this is to deal with the process directly at the amplitude level . for the purpose ,
the improved trace technology " is suggested and developed in the literature @xcite . after generating proper phase - space points ,
one first calculate the numerical value for the amplitudes , and then sum these values algebraically and square it to get the squared amplitude , @xmath37 ; through such way , numerical simulation efficiency can be greatly improved in comparison to the usual squared amplitude technology .
events through pythia by @xmath24 annihilation . however , in pythia all kinds of @xmath38-hadron ( @xmath39 , @xmath16 , @xmath40 and etc . )
events will be produced according to the proper fragmentation possibilities , among which the fragmentation possibility for the doubly heavy @xmath16 meson is quite small @xcite , thus using pythia to generate @xmath16 events is not ` economical ' ; i.e. , when simulating the @xmath16 events , too much time will be wasted by generating enormous unwanted @xmath38-hadron events .
] moreover , under the approach , many simplifications can be done at the amplitude level due to the fermion - line symmetries and the specific properties of each heavy - quarkonium fock states , then , we can even written down the analytic expressions for the amplitude .
the improved trace technology " has been adopted to derive a simplified and compact analytical expressions for the on - shell @xmath1-boson decay process , @xmath41+b+\bar{c}$ ] @xcite . at the present
, we will adopt the same method for dealing with the @xmath4-quarkonium production through the full production process , @xmath33+b+\bar{c}$ ] .
the intermediate @xmath1-boson may or may not be on - shell , and if it is off - shell , some new lorentz structures must be introduced to ensure a gauge - invariant result . in a sense
, this paper provides an extension of the improved trace technology " on how to deal with the off - shell particles .
the remaining parts of the paper are organized as follows . in sec.ii , we present the formulation for dealing with the process of @xmath42+b+\bar{c}$ ] , where improved trace technology are adopted to simplify the hard scattering amplitude . in sec.iii , we give the numerical results .
sec.iv is reserved for a summary .
all necessary expressions are put in the appendices .
+ b+\bar{c}$ ] , where @xmath6 $ ] stands for the fock states : @xmath7\rangle$ ] , @xmath8g\rangle$ ] , @xmath9\rangle$ ] , @xmath10g\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ] ( with @xmath13 ) respectively .
, scaledwidth=40.0% ] as shown in fig.([feyn ] ) , there are four feynman diagrams for the process @xmath43 ( q_3)+b(q_2)+\bar{c}(q_1)$ ] , where @xmath6 $ ] stands for the @xmath4-quarkonium fock - states : @xmath7\rangle$ ] , @xmath8g\rangle$ ] , @xmath9\rangle$ ] , @xmath10g\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ] ( with @xmath13 ) respectively . here as a comparison , the production channel via a virtual photon , which is small around the @xmath1 peak , is also presented .
according to nrqcd factorization formulae @xcite , the cross - section of the process can be written in the following factorization form : @xmath44+b+\bar{c})\langle{\cal o}^h(n)\rangle,\ ] ] where the matrix element @xmath45 is proportional to the inclusive transition probability of the perturbative state @xmath46 $ ] into the bound states .
@xmath47+b+\bar{c})$ ] stands for the short - distance cross - section ; i.e. @xmath48+b+\bar{c } ) \nonumber\\ & & \quad = \frac{1}{4\sqrt{(p_1\cdot p_2)^2-m_1 ^ 2 m_2 ^ 2 } } \overline{\sum } |{\cal m}|^{2 } d\phi_3,\end{aligned}\ ] ] where @xmath49 means we need to average over the spin states of initial particles and to sum over the color and spin of all final particles , and the three - particle phase space @xmath50 the phase space can be generated by the routine rambos @xcite , and its integration can be carried out numerically with the vegas program @xcite .
the scattering amplitude for the process can be written as , @xmath51 where @xmath52 and @xmath53 the spin - indices @xmath54 , @xmath55 , @xmath56 and @xmath57 stand for the spin projections of the initial electron and positron , and the outgoing quark and antiquark respectively .
the overall parameter @xmath58 or @xmath59 is for the production of the color - singlet or the color - octet quarkonium state respectively . for the quarkonium production through @xmath1-boson propagator : @xmath60 and @xmath61 ( @xmath62 stands for the color of the color - octet @xmath4-quarkonium state ) ; the vertex @xmath63 and the propagator @xmath64 , where @xmath65 stands for the total decay width of @xmath1 boson .
if the @xmath0 collider runs at the energies around the @xmath1 mass , it will constitute an energy scan .
the shape of the cross - section variation around the @xmath1 peak is described by a breit - wigner ansatz with an energy - dependent total width . while , for the quarkonium production through photon propagator : @xmath66 and @xmath67 ; the vertex @xmath68 and the propagator @xmath69 .
the gamma structures @xmath70 with @xmath71 can be read off from the four feynman diagrams in fig.([feyn ] ) , whose explicit expressions can be directly found in refs.@xcite .
propagator , from the correspondence between @xmath70 and @xmath72 , one only needs to replace @xmath73 in the amplitudes of refs.@xcite by @xmath74 . and
the case of photon propagator can be easily derived by changing the corresponding vertex . ]
@xmath75 and @xmath76 contain explicit strings of dirac @xmath77-matrices sandwiched by two fermion spinors , which can be done under the conventional trace technology .
however it is hard to write down the squared amplitudes for these massive processes , especially for the @xmath21-wave cases , which is too complex and lengthy . here , to simplify the analytic results and to make it more compact , we adopt the improved trace technology " @xcite .
it is noted that under suitable transformations , this approach is also useful for deriving the polarized cross - sections , which is helpful for a polarized @xmath0 collider . as an explanation for the improved trace technology " ,
we first arrange the amplitude @xmath76 into four orthogonal sub - amplitudes @xmath78 according to the spins of the outgoing quark @xmath79 with spin @xmath56 and antiquark @xmath80 with spin @xmath81 , then transform these sub - amplitudes into a trace form by properly dealing with the massive spinors with the help of an arbitrary light - like momentum @xmath82 and an arbitrary space - like momentum @xmath83 , which satisfies @xmath84 and @xmath85 .
the final results are independent of @xmath82 and @xmath83 , which provides a way to check the rightness of the derived results . and one can choose them to be those that will maximumly simply the amplitude .
and then we do the trace of the dirac @xmath77-matrix strings at the amplitude level , which shall result in explicit series over some independent lorentz - structures .
the amplitude @xmath75 can be done in a similar way .
detailed process of the approach can be found in refs.@xcite , so to short the paper , we only list all the differences for the present processes .
however , because at the present @xmath1 may or may not be on - shell , one should use these on - shell expressions carefully with suitable transformations . as for @xmath76
, it can be rearranged as the following four spin combinations @xmath86 the production through the photon propagator is similar to the case of the @xmath1-boson propagator , which can be directly read from its vector part .
so we shall only concentrate our attention on the case of the @xmath1-boson propagator .
collider within the pythia environment @xcite is in preparation , where both cases with the photon and the @xmath1-boson propagators are included @xcite . ]
@xmath87 can be expressed over the independent lorentz structures .
for example , for the production of @xmath7\rangle$]-quarkonium , we have @xmath88 } & = & { a_{1}^{[^1s_0]}}\frac{q_3^{\nu}}{m_z}+{a_{2}^{[^1s_0]}}\frac{q_2^{\nu}}{m_z}+ { a_{3}^{[^1s_0]}}\frac{k^{\nu}}{m_z } \nonumber\\ & & + { a_{4}^{[^1s_0]}}\frac{i}{m_z^3}\varepsilon(\nu , k , q_2,q_3 ) , \label{hmat}\end{aligned}\ ] ] where @xmath89 . comparing the lorentz expansions of @xmath90}$ ] with those of @xmath91}$ ] in refs.@xcite
, we obtain the same expansion coefficients for the corresponding lorentz structures .
more explicitly , by multiplying both side of eq.([hmat ] ) with the on - shell @xmath1-boson polarization vector @xmath92 , which satisfies @xmath93 , we can obtain exactly the same matrix elements i.e. @xmath94}}|_{k\cdot\epsilon=0}=m_{[^1s_0]}$ ] .
however , when the @xmath1-boson is not on - shell , @xmath95 } } { k^{\nu}}/{m_z}\right)$ ] must be kept in @xmath90}$ ] . and to be a consistent analysis for the present case , we need to take these new terms into consideration . for convenience , we present all the new lorentz structures for all the mentioned fock states together with their non - zero coefficients in appendix b. as for the leptonic vector @xmath75 , it is simpler and can be directly expressed as @xmath96 where all the non - zero coefficients @xmath97 are presented in appendix c. to be a cross check , we also use the squared amplitude approach to deal with the process , which however can be rewritten as the following more convenient form @xmath98 the leptonic tensor @xmath99 where small mass terms involving @xmath100 are neglected .
the result for the phase - space integration over @xmath101 is equal to the decay width of @xmath102+b+\bar{c}$ ] times a factor @xmath103}$ ] .
numerically , we obtain the same results for these two approaches .
the basic input parameters are chosen as the following values : @xcite : @xmath104 gev , @xmath105 gev , @xmath106 and @xmath107 .
we adopt the leading - order @xmath108 running , and take the normalization scale to be @xmath109 , which leads to @xmath110 .
the color - singlet matrix elements can be directly related to the wave functions at the origin for the @xmath34-wave states or with the first derivative of the wave functions at the origin for the @xmath21-wave states .
the @xmath34-wave color - octet ones are suppressed to the corresponding ones by a factor @xmath111 with @xmath112 @xcite . in the leading order approximation
, the color - octet or color - singlet matrix - element appears in the amplitude as a linear factor .
the uncertainties in the production from the matrix elements can be figured out straightforwardly , so throughout the paper , we will fix the values of the radial wave function at the origin and the first derivative of the radial wave function at the origin to be @xmath113 gev@xmath114 and @xmath115 gev@xmath116 @xcite which are derived under the buchmller - tye - potential model @xcite . to calculate the hard - scattering amplitude for the short - distance cross - section @xmath117 , the equations : @xmath118 , @xmath119 and @xmath120 ( @xmath121 is the @xmath4-quarkonium mass , @xmath122 with @xmath123 and @xmath124 are momenta of the constitute quarks ) must be satisfied simultaneously , and only when these equations are satisfied can the gauge invariance of the amplitude be guaranteed . in this paper , we proceed the numerical calculation by taking @xmath125 gev and @xmath126 gev as their center values .- mass for charm quark and half of the @xmath127 for bottom quark . and
they are also consistent with the bound - state parameters under the potential - model calculations @xcite . ]
the mass of the quarkonium bound - state is the sum of the two heavy quark masses ; i.e. @xmath128 gev .
we will discuss the theoretical uncertainties caused by the @xmath129- and @xmath38- quark masses by varying @xmath14 gev and @xmath15 gev .
+ b+\bar{c})$ ] versus the @xmath130 collision energy @xmath131 , where contributions from the mentioned fock states @xmath132\rangle$ ] , @xmath133\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ] have been summed up . here
the dashed , the dotted and the solid lines are for the production channel with @xmath134-propagator , @xmath1-propagator and the sum of these two channels , respectively.,scaledwidth=35.0% ] firstly , we calculate the total cross - section for the @xmath0 collider energy at the @xmath1 peak , i.e. @xmath135 .
we obtain @xmath136\rangle}&=&1.800 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 1}[^3s_1]\rangle}&=&2.575 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 1}[^1p_1]\rangle}&=&0.190 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 1}[^3p_0]\rangle}&=&0.115 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 1}[^3p_1]\rangle}&=&0.232 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 1}[^3p_2]\rangle}&=&0.256 \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 8}[^1s_0]g\rangle}&=&0.225 \times{v^4 } \;{\rm pb},\\ \sigma_{|(c\bar{b})_{\bf 8}[^3s_1]g\rangle}&=&0.322 \times{v^4 } \;{\rm pb}.\end{aligned}\ ] ] it is found that the total cross - sections of all the @xmath21-wave states can be added up to @xmath137 of that of the ground state @xmath138)$ ] .
so the @xmath21-wave @xmath4-quarkonium states should be taken into consideration , especially for the future high luminosity @xmath24 colliders .
two color - octet @xmath34-wave states shall only provide @xmath139 contribution to the ground state . - and @xmath140- distributions of the @xmath4-quarkonium , where the solid , the dashed , the dotted , the dash - dot , the triangle and the diamond lines are for @xmath7\rangle$ ] , @xmath11\rangle$ ] , @xmath141\rangle$ ] , @xmath142\rangle$ ] , @xmath9\rangle$ ] and @xmath143\rangle$ ] respectively.,title="fig:",scaledwidth=35.0% ] - and @xmath140- distributions of the @xmath4-quarkonium , where the solid , the dashed , the dotted , the dash - dot , the triangle and the diamond lines are for @xmath7\rangle$ ] , @xmath11\rangle$ ] , @xmath141\rangle$ ] , @xmath142\rangle$ ] , @xmath9\rangle$ ] and @xmath143\rangle$ ] respectively.,title="fig:",scaledwidth=35.0% ] .total cross - section of the process @xmath144+b+\bar{c}$ ] under different collision energies , where contributions from the mentioned fock states @xmath132\rangle$ ] , @xmath133\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ] have been summed up . here
@xmath125 gev , @xmath126 gev and @xmath145 . [ cols="^,^,^,^,^,^",options="header " , ] [ tevcut ] experimentally , there is no detector which can cover all the kinematics of the events , so only some of the @xmath16 events can be observed .
that is , the @xmath16 events with small @xmath146 and/or a large rapidity @xmath140 can not be detected , so they can not be utilized for experimental studies . as a final remark , it would be helpful to know how the differential cross - sections change with different @xmath146- or @xmath140- cuts . in fig.([ptvaryrap ] ) , we present the @xmath4-quarkonium @xmath146-distributions for all the considered fock states with a particular rapidity cut ( @xmath147 ) , where the solid , the dashed , the dotted , the dash - dot and the dash - dot - dot lines are for @xmath148 , @xmath149 , @xmath150 , @xmath151 and @xmath152 respectively . in fig.([rapvarypt ] ) , we present the @xmath4-quarkonium @xmath140-distributions for all the considered fock states with a particular transverse momentum cut ( @xmath153 ) , where the solid , the dashed , the doted and the dash - dot lines are for @xmath154 gev respectively . to analyze the quantitative difference of the differential distributions with regard to the parameters @xmath153 and @xmath147
, we introduce a ratio for the integrated total cross - section : @xmath155 where @xmath156 and @xmath157 are the total cross - section with and without @xmath147 respectively . in the definition of @xmath158
, contributions from all the considered fock states have been summed up .
the values of the ratio @xmath158 that correspond to different values of @xmath153 and @xmath147 are presented in tab.([tevcut ] ) .
it is found that when under the same @xmath153 , @xmath158 increases with the increment of @xmath147 , and when under the same @xmath147 , @xmath158 also increases with @xmath153 .
this shows that the rapidity distribution of the @xmath4-quarkonium shrinks with the increment of @xmath153 ; e.g. when @xmath159 gev , @xmath160 .
in the present paper , we have studied the @xmath4-quarkonium production through @xmath24 annihilation in leading @xmath108 order , but with the @xmath29-expansion up to @xmath161 . the production of the @xmath4-quarkonium in fock states @xmath7\rangle$ ] , @xmath9\rangle$ ] , @xmath11\rangle$ ] , @xmath12\rangle$ ] , @xmath8g\rangle$ ] and @xmath10g\rangle$ ]
have been studied .
higher fock states such as the @xmath21-wave states will also provide sizable contributions to the total cross - section of the @xmath16 meson .
if assuming the higher excited states decay to the ground state @xmath162\rangle)$ ] with 100% efficiency via electromagnetic or hadronic interactions , then we will obtain @xmath163 pb . according to our present calculation , one would expect to accumulate @xmath19 @xmath16 events in one operational year for the super z - factory with a high luminosity @xmath164 .
if taking the @xmath0 collision energy to run slightly off the @xmath1-peak ; i.e. @xmath165 or @xmath166 , the total production cross - section will be lowered by about one - order from its peak value .
however one can estimate @xmath167 @xmath16 events to be produced under such condition , which is still sizable . then , in addition to the hadronic colliders tevatron and lhc , the super z - factory will provide another good platform for studying the @xmath16-meson properties .
* acknowledgements : * this work was supported in part by the fundamental research funds for the central universities under grant no .
cdjxs1102209 , the program for new century excellent talents in university under grant no .
ncet-10 - 0882 and by natural science foundation of china under grant no.10805082 and no.11075225 .
we set the @xmath168-axis to be along the incoming beam . the momentum of the @xmath4-quarkonium @xmath169 , where @xmath170 , @xmath121 is the quarkonium mass , @xmath171 and @xmath172 are polar and azimuth angles respectively .
the expressions of the polarization vector can be written as @xmath173 as required , the polarization vector satisfies the following relation @xmath174 the polarization tensor @xmath175 for @xmath176-quarkonium state can be constructed in terms of @xmath177 , @xmath178 and @xmath179 as @xmath180
in the present section , we present the new lorentz structures and their non - zero coefficients for the process @xmath43 ( q_3)+b(q_2)+\bar{c}(q_1)$ ] , where @xmath182 stands for @xmath7\rangle$ ] , @xmath9\rangle$ ] , @xmath11\rangle$ ] , @xmath12\rangle$](j=1,2,3 ) respectively .
all these new lorentz structures have no correspondence to the lorentz structures introduced from the process @xmath183(q_3 ) + b(q_2)+\bar{c}(q_1)$ ] .
note , the following parameters , @xmath184 , @xmath185 , @xmath186 , @xmath187 , @xmath188 and so on , have the same meaning as those of refs.@xcite , and we shall not present them here for short .
there is only one new lorentz structure for @xmath189\rangle$ ] , @xmath190 whose coefficients @xmath191 are @xmath192 there are three new lorentz structures for @xmath193\rangle$ ] , @xmath194 whose coefficients @xmath195 ( @xmath196 ) are @xmath197,\\ a^1_3 & = & -\frac{2 l_1 { m_z}^{7/2}}{\sqrt{r_3}}(d_2-d_3),\\ a^2_1 & = & 2 d_1 l_2 { m_z}^{7/2 } r_1 \sqrt{r_3},\\ a^2_2 & = & -l_2 { m_z}^{7/2}\sqrt{r_3 } [ ( 3 d_1-d_2+d_3+d_4 ) r_1 \nonumber\\ & & -(d_1-d_2+d_3-d_4 ) r_2 ] , \\
a^2_3 & = & \frac{8 l_2 { m_z}^{7/2}}{\sqrt{r_3 } } ( d_2 t_b+d_3 t_c),\\ a^{3'}_1 & = & \frac{d_1 { m_z}^{9/2 } n_0 \sqrt{r_3}}{2 l_2}(r_1 r_3+u),\\ a^{3'}_2 & = & -\frac{{m_z}^{9/2 } n_0 \sqrt{r_3}}{4 l_2}(d_2-d_3 + 2 d_4 ( { r_3}^2 + u ) \nonumber\\ & & + d_1 ( -1 + 2 r1 r_3 + 2 x+2 y)+\nonumber\\ & & 2 ( d_2 ( r_1 r_3-x - y)+d_3 ( r_2 r_3+y))),\\ a^{3'}_3 & = & \frac{2 { m_z}^{9/2 } n_0 \sqrt{r_3}}{l_2}(d_1 t_b+d_3 t_c ) , \\
a^{4'}_1 & = & -\frac{2 d_1 { m_z}^{9/2 } n_0 \sqrt{r_3 } t_b}{l_1 } ( r_1 ^ 2-r_2 r_1+u),\\ a^{4'}_2 & = & \frac{{m_z}^{9/2 } n_0 \sqrt{r_3}}{l_1}(d_2 t_b ( -2 x-2 y+1)+\nonumber\\ & & d_1 t_b ( 2 x+2 y-1)+ t_c ( d_4 ( r_1 ^ 2-r_2 ^ 2+r_3 ^ 2
\nonumber\\ & & + 2 u)+d_3 ( 2 y-1))),\\ a^{4'}_3 & = & \frac{{m_z}^{9/2 } n_0}{2 l_1 \sqrt{r_3 } } [ ( d_2-d_3 ) ( r_1-r_2)+(d_2-d_1 ) r_3]\end{aligned}\ ] ] there are three new lorentz structure for @xmath198\rangle$ ] , which are the same as that of @xmath199 $ ] , whose coefficients @xmath195 ( @xmath196 ) are @xmath200 there are nine new lorentz structures for @xmath201\rangle$ ] , which are @xmath202 it is noted that @xmath203 is the symmetric tensor and @xmath204 is the anti - symmetric tensor , and the fact that @xmath205 . as the terms involving the following coefficients have no contributions to the square of the amplitude .
so practically we can safely set the coefficients before them to be zero : @xmath206\rangle)=0 \;\;\;\;\;\;\;\;{\rm for}\;i=(1 - 4),\\ j=(5,6,7,8,9)\\ \end{array}\ ] ] @xmath207\rangle)=0 \;\;\;\;\;\;\;\;{\rm for}\;i=(1 - 4 ) , \\
j=(1)\\ \end{array}\ ] ] @xmath208\rangle)=0 \;\;\;\;\;\;\;\;{\rm for}\;i=(1 - 4),\\ j=(1,5,6,7,8,9 ) .
\end{array}\ ] ] the coefficients @xmath209 , @xmath210,@xmath211 and @xmath212 that are the same for all three @xmath21-waves : @xmath213 remaining non - zero coefficients for @xmath214\rangle$ ] and @xmath215\rangle$ ] : @xmath216 remaining non - zero coefficients for @xmath217\rangle$ ] : @xmath218
all the non - zero coefficients of @xmath219 , @xmath220 where @xmath221 and @xmath222 .
@xmath223 can be obtained by ensuring @xmath224 to satisfy @xmath225 .
chang , j.x .
wang and x.g .
wu , phys.rev .
d*70 * , 114019 ( 2004 ) ; c.h .
chang , c.f .
qiao , j.x .
wang and x.g .
wu , phys.rev .
d*71 * , 074012 ( 2005 ) ; c.h .
chang , c.f .
qiao , j.x .
wang and x.g .
wu , phys.rev .
d*72 * , 114009 ( 2005 ) .
chang , j.x .
wang and x.g .
wu , comput .
. commun . * 174 * , 241 ( 2006 ) ; c.h .
chang , j.x .
wang and x.g .
wu , comput .
. commun . * 175 * , 624 ( 2006 ) ; x.y .
wang and x.g .
wu , comput .
. commun . * 183 * , 442 ( 2012 ) .
e. braaten , king - man cheung , t. c. yuan , phys.rev .
d*48 * , r5049 ( 1993 ) ; v.d .
barger , king - man cheung , w.y .
keung , phys.rev . d*41 * , 1541 ( 1990 ) ; king - man cheung , w.y .
keung , t.c .
yuan , phys.rev.lett . *
76 * , 877 ( 1996 ) . | considering the possibility to build an @xmath0 collider at the energies around the @xmath1-boson resonance with a planned luminosity so high as @xmath2 ( super @xmath3-factory ) , we make a detailed discussion on the @xmath4-quarkonium production through @xmath5+b+\bar{c}$ ] within the framework of non - relativistic qcd . here
@xmath6 $ ] stands for the fock - states @xmath7\rangle$ ] , @xmath8g\rangle$ ] , @xmath9\rangle$ ] , @xmath10g\rangle$ ] , @xmath11\rangle$ ] and @xmath12\rangle$ ]
( with @xmath13 ) respectively . to simplify the hard - scattering amplitude as much as possible and to derive analytic expressions for the purpose of future events simulation
, we adopt the improved trace technology " to do our calculation , which deals with the hard scattering amplitude directly at the amplitude level other than the conventional way at the squared - amplitude level .
total cross - section uncertainties caused by the quark masses are predicted by taking @xmath14 gev and @xmath15 gev . if all higher @xmath4-quarkonium states decay to the ground state @xmath16 ( @xmath7\rangle$ ] ) with @xmath17 efficiency , we obtain @xmath18 pb , which shows that about @xmath19 @xmath16 events per operation year can be accumulated in the super @xmath3-factory . if taking the collider energy runs slightly off the @xmath1-peak , i.e. @xmath20 , the total cross - section shall be lowered by about one - order from its peak value .
such a super @xmath3-factory shall provide another useful platform to study the properties of @xmath16 meson , or even the properties of its excited @xmath21-wave states , in addition to its production at the hadronic colliders tevatron and lhc . + * pacs numbers : * 12.38.bx , 12.39.jh , 14.40.pq |
in the twenty - five years since the first detection of carbon monoxide in external galaxies @xcite , astronomers have observed the 3 mm j = 10 transition of co in the centers of several hundreds of galaxies @xcite ; these observations form the basis for our statistical understanding of the molecular content of galaxies as a function of parameters like galaxy type and luminosity . despite an abundance of studies of the amount of co in galaxies ,
however , there is still a rather incomplete understanding of the distribution of co within galaxies .
this is because the overwhelming majority of co observations have been made using single - dish telescopes , the linear resolutions of which are often ten to one hundred times larger than the basic unit in which molecular gas is thought to be organized , the giant molecular cloud ( gmc ) . also , even for the roughly 200 galaxies that have been observed in multiple pointings with single - dish telescopes , all but a handful have been observed only at select positions along the major and minor axes , rather than in fully - sampled , two - dimensional maps .
so far , only a few dozen galaxies have been imaged with the high resolution achievable with an interferometer .
the largest published interferometric survey to date is by @xcite , who surveyed the central arcminute of 20 nearby spiral galaxies with 4 resolution using telescopes at the nobeyama radio observatory ( nro ) and the owens valley radio observatory ( ovro ) .
the bima survey of nearby galaxies ( bima song ) is the first systematic imaging survey of the 3 mm co j = 10 emission from the centers and disks of nearby spiral galaxies .
the basic database that we have produced is a collection of spatial - velocity data cubes for 44 nearby spiral galaxies that show the distribution and kinematics of co emission at resolutions of a few hundred parsecs ( @xmath9 6 ) and @xmath9 10 km s@xmath0 over a field of view of typically 10 kpc ( @xmath9 190 ) .
bima song was introduced in the recent study by @xcite , hereafter paper i , where we presented initial results for a subsample of 15 spiral galaxies . in this paper , we present data from the full survey of 44 spiral galaxies .
some of the notable features of the bima song survey are as follows : ( 1 ) the sources were not selected based on co brightness or infrared luminosity .
( 2 ) we mosaiced multiple fields for the 33 galaxies with the largest optical diameters , so that the full - width half - power field of view was 190 .
this field of view is a full order of magnitude larger in area than was obtained for previous high - resolution surveys .
for the 11 galaxies with smaller optical diameters , we used a single pointing of the interferometer to image a region 100 in diameter .
( 3 ) the data have very good sensitivities , with rms noise levels of 23130 mjy bm@xmath0 per 10 km s@xmath0 channel .
these are comparable to the sensitivities obtained by @xcite of 17110 mjy bm@xmath0 per 10 km s@xmath0 channel over a much smaller field of view .
the noise levels are roughly an order of magnitude lower than those of the most extensive single - dish survey , the fcrao extragalactic co survey @xcite .
( 4 ) we collected fully - sampled , on - the - fly ( otf ) single - dish maps for 24 of the brightest galaxies and incorporated these data into the song maps , so that all of the co flux is imaged for these galaxies .
these maps in themselves comprise the most extensive collection of fully - sampled , two - dimensional single - dish co maps of external galaxies to date . for the remaining 20 sources
, we collected sensitive single - dish spectra in order to evaluate the large - scale flux recovery .
( 5 ) all observations , data reduction and analysis were carried out using uniform procedures , so that systematic differences among the resultant maps should be minimized .
( 6 ) we are collecting complementary broad - band optical , infrared , h@xmath10 , hi , and radio continuum data so as to have a parallel set of observations for each galaxy .
where available , we have compiled existing data ; otherwise , we are obtaining our own new observations .
these complementary data will be presented in future song papers .
this paper is organized as follows . in 2 , we list the sample selection criteria and describe the basic properties of the sample . in 3 , we give details of the interferometric and single - dish observations . in 4 , we describe the data reduction and calibration . in 5 , we present a catalog of the bima song data in various forms , including channel maps , integrated intensity maps , spectra , and basic velocity fields . in
6 , we discuss the co detection rate for the survey . in 7 , we present direct measurements of global fluxes for the 24 otf sources in the survey . in 8 , we discuss the issue of large - scale flux recovery and present measurements for the bima song sources . in 9
, we examine the degree of central peakedness of the molecular surface density distributions . in 10 , we consider how the three local group spiral galaxies compare with the nearby spirals presented here .
we summarize the conclusions in 11 .
finally , we have included two appendices in which we give brief descriptions of some deconvolution issues at millimeter wavelengths as well as techniques for combining single - dish data with interferometric data .
we note that the data collection , reduction , and calibration procedures were described in considerable detail in paper i , which is very much a companion to the present paper .
although we present some of the same material here , we have tried to emphasize somewhat different points in the two papers , and we encourage the reader to look to both references for a complete discussion of technical details of the survey . the data from this survey are publicly available ( see 11 ) .
sources were selected without reference to co or infrared brightness . instead , hubble type sa through sd galaxies were selected with @xmath11 2000 km s@xmath0 ; if @xmath12=75 km s@xmath0 mpc@xmath0 , this corresponds to galaxies with hubble distances of @xmath13 27 mpc .
the galaxy inclinations , taken as cos @xmath14 = @xmath15 from axial ratios @xmath16 in the rc3 @xcite , were @xmath6 70 , so as not to be too edge - on to study azimuthal properties . a minimum declination of @xmath5 -20 ensured that the sources could be observed well from the hat creek radio observatory in northern california without having too elongated a synthesized beam . to limit the size of the sample , we chose galaxies with apparent total blue magnitudes @xmath8 11.0 .
finally , we used a maximum optical diameter @xmath7 70 to exclude one source , m33 , which is the subject of a separate bima survey @xcite .
the remaining sources have optical diameters @xmath7 30 . using the nasa / ipac extragalactic database ( ned )
, we identified 44 galaxies meeting all selection criteria .
table 1 lists these 44 sources , which comprise the bima song sample .
the bima song sample includes objects with a range of properties that are of interest for a variety of studies .
spirals with hubble classifications from sab to sd are represented , with roughly one - third of the sample having hubble type of sbc and the remaining objects split evenly between earlier and later types ( ab : b : bc : c : cd : d = 7:8:14:7:6:2 ) . in figure
[ typehist ] , we show the distribution of rc3 hubble types compared with the 486 galaxies from the magnitude - limited palomar survey of the northern sky @xcite .
compared to the palomar survey , the song sample has relatively few sa / sab galaxies .
this is probably a selection effect , since early - type galaxies tend to be found in clusters and are thus farther away on average than the song selection criteria allowed .
the median distance for the @xcite sample is 17.9 mpc , compared with 11.9 mpc for song .
nearly two - thirds of the sample are classified optically in the rc3 as barred galaxies , with five sb and 24 sab bars ; the remaining 15 galaxies are optically unbarred . using the nuclear classifications from the @xcite survey , we note that fourteen of the sources contain nonstellar ( seyfert or liner ) emission - line nuclei , seventeen have starburst ( h ii ) nuclei , and nine show `` transition '' spectra , with [ o i ] emission - line strengths intermediate between those of h ii nuclei and liners .
as figure [ nuchist ] shows , the nuclear types are distributed in a similar way to those from the palomar survey . in figure
[ armhist ] , we show the distribution of arm classifications ( ac ) as defined ( and classified ) by @xcite , compared with the @xcite full sample of 654 spiral galaxies
. there are 29 grand - design spirals ( ac = 512 ) , of which 19 show long arms dominating the optical disk ( ac = 9 , 12 ) ; 13 spirals are flocculents ( ac = 14 ) , exhibiting fragmented arms that typically lack regular spiral structure .
the song sample has relatively few galaxies in ac 1 5 compared with the @xcite sample ; this may be because the latter sample has more low - luminosity galaxies than does song .
the galaxies range in distance from 2.1 mpc to nearly 26 mpc , with a mean distance of 12.2 @xmath17 5.9 mpc .
the properties of the galaxies are summarized in table 1 .
we carried out bima song observations from 1997 november through 1999 december using the 10-element berkeley - illinois - maryland association ( bima ) millimeter interferometer @xcite at hat creek , ca . in all , we included data from about 160 8-hour bima song tracks as well as about 15 tracks on song sources taken by individual team members before 1997 november @xcite . before 1998 september
, the antennas were configured in the then - named c array , which included baselines as short as 7.7 m and as long as 82 m. after 1998 september , we used the current c and d configurations , which together include baselines as short as 8.1 m and as long as 87 m. for all galaxies , we observed the co j = 10 line at 115.2712 ghz .
we configured the correlator to have a resolution of 1.56 mhz ( 4 km s@xmath0 ) over a total bandwidth of 368 mhz ( 960 km s@xmath0 ) .
we observed 33/44 of the galaxies ( those with r@xmath18
@xmath3 200as well as three smaller sources ) using a 7-field , hexagonal mosaic with a spacing of @xmath19 or 44 ; this yielded a half - power field of view of about 190or 10 kpc at the average distance of galaxies in the survey .
of these , we observed some galaxies ( ngc 0628 , ngc 1068 , ngc 2903 , ngc 3627 , ngc 4736 , ngc 5033 , ngc 5194 ) with slightly different pointing spacings or additional fields .
the remaining 11/44 galaxies were observed with a single pointing , which yielded a field of view of 100 fwhm . for the multiple - pointing observations , each field
was observed for one minute before the telescope slewed to the next pointing .
thus , we returned to each pointing after @xmath98 minutes , which is within the time needed to ensure nyquist radial sampling of even the longest baselines in the @xmath20 plane ; see @xcite .
table 2 summarizes the observational parameters of the maps , including the number of fields observed , the synthesized beam size in angular and linear dimensions , the velocity binning , and the measured rms noise level in the channel maps .
we collected over 700 hours of single - dish data from 1998 april through 2000 june using the nrao 12 m telescope on kitt peak , az .
we observed orthogonal polarizations using two 256 channel filterbanks at a spectral resolution of 2.0 mhz ( 5.2 km s@xmath0 ) , and with a 600 mhz configuration of the digital millimeter autocorrelator with 0.80 mhz ( 2.1 km s@xmath0 ) resolution as a redundant backend on each polarization .
the pointing was monitored every 1 - 2 hours with observations of planets and strong quasars .
the focus was measured at the beginning of each session and after periods during which the dish was heating or cooling . at 115 ghz ,
the telescope half - power beamwidth , main beam efficiency @xmath21 , and forward spillover and scattering efficiency @xmath22 are 55 , 0.62 , and 0.72 . to convert from the recorded @xmath23 to @xmath24 , the reader should multiply @xmath23 by @xmath22/@xmath21 = 1.16 . to convert from brightness temperature @xmath23 to units of flux density ,
the reader should multiply by the antenna gain , or 33 jy k@xmath0 . for 24 galaxies ( table 2 ) with the strongest co brightness temperatures , we observed in on - the - fly ( otf ) mode @xcite , where the telescope takes data continuously as it slews across the source .
the resultant otf maps were incorporated into the bima maps to create combined bima+12 m maps that did not have large - scale emission `` resolved out '' by the interferometer - only observations .
the sources with combined maps are identified in the last column of table 2 . for the remaining 20 galaxies
, we obtained sensitive position - switched spectra in order to evaluate the large - scale flux recovery .
we describe the two observing modes in the following subsections .
the otf mode minimizes relative calibration errors and pointing errors across a map . in this mode ,
the actual telescope encoder positions are read out every 0.01 seconds and folded into the spectra , which are read out every 0.1 seconds .
we set up a typical otf map to have a total length of 8 on a side , which includes a 1 ramp - up and 1 ramp - down distance on either side of the 6 heart of the map , and we alternated scanning in the ra and dec directions for successive maps in order to minimize striping ( cf * ? ? ?
we spent about 30 seconds on each map row and observed a reference position on the sky at the beginning of every second row , calibrating often enough ( every 515 minutes ) so that the measured system temperatures changed by less than a few percent between calibrations .
the maps were of course greatly oversampled in the scanning direction ; in the orthogonal direction , we spaced the rows by 18 , which slightly oversamples the maps even by the nyquist criterion ( 22.4 at 115 ghz at the 12 m ) .
the output antenna temperatures were recorded as the standard @xmath23 = ( on - off)/off @xmath25 @xmath26 , where on is the power measured on source and off is the power measured at the reference position on the sky ( taken to be the off measurement nearest in time to a given on ) . for galaxies where the detected co emission was clearly confined to a region within the otf map boundaries
, we later created a new off reference spectrum from the average of roughly 2040 individual ( 0.1 second ) spectra on either side of each row ; although this technique may introduce a systematic , low - level error into the map , it also greatly improves the baseline stability of the spectra .
each individual otf map took @xmath27 20 minutes to complete , and we obtained anywhere from 10 to 30 otf maps to achieve the desired sensitivity ; following @xcite , we tried to spend enough time on the total - power measurements to match the signal - to - noise ratio of the interferometric observations . in practice , when smoothed to 55for direct comparison , the bima maps tended to have lower absolute noise levels than the otf maps by a factor of about two .
given reasonably stable observing conditions , the relative flux calibration across an individual otf map should be very good , so that even if the absolute flux scale drifts from map to map , the combined final 12 m map should also have very good pixel - to - pixel calibration .
an analysis of well - pointed data over many observing seasons shows that the repeatability of the flux calibration at the 12 m is accurate to better than a 1 @xmath28 uncertainty of 10% .
however , the pointing at the 12 m is a more challenging problem . for the otf data
, we were able to crosscorrelate the 12 m data with the bima data ( 4.3 ) , and we estimate that the absolute pointing accuracy was good to @xmath27 5 rms after performing the pointing crosscorrelation .
we subtracted baselines based on linear fits to the emission - free channels in the passband , then weighted and gridded the otf data to an 18 cell in aips .
the rms noise levels in the otf maps were typically 1520 mk ( 0.500.66 jy bm@xmath0 ) per 2 mhz ( 5.2 km s@xmath0 ) channel . to allow for some amount of pointing or other systematic errors in the otf maps
, we assign an overall 1 @xmath28 uncertainty of 15% to the flux calibration of the otf data .
we observed the remaining 20 galaxies in a simple position - switched mode , where the telescope tracks a specific position on the source for a given integration time ( typically 30 seconds ) , then is pointed to a nearby ( typically closer than @xmath1730 in azimuth from the source ) reference position to measure the sky brightness .
we observed each galaxy at the positions of the bima pointings , so that we obtained 7 spectra for the mosaiced bima observations and one for the single - pointing bima observations .
again , spectra of @xmath23 were recorded as described above .
we fitted and removed linear baselines from the spectra using the bell labs spectral line reduction package , comb .
the rms noise levels of the spectra were @xmath9 520 mk per 2 mhz ( 5.2 km s@xmath0 ) channel .
as with the otf data , the absolute flux calibration is presumably accurate to a 1 @xmath28 uncertainty of 10% , though this is difficult to measure directly because we were not able to correct for pointing errors , which may be as large as @xmath9 1520 .
the effect of mispointed data is essentially unpredictable without an independent way of characterizing it ( as with the bima-12 m pointing crosscorrelation for otf data described in 4.3 ) .
for example , a pointing error of 20 will cause the measured flux density of a point source to be attenuated by 12% .
however , for sources with more complicated structures , the flux density measured at the mispointed position could be higher or lower than the flux density from the desired position .
furthermore , the measured flux density is a function of the observed velocity , so that the spectral line shape may also be in error .
we therefore assign a 1 @xmath28 uncertainty of 25% to the flux calibration of the position - switched spectra , keeping in mind that this error is dominated by systematic uncertainty .
the position - switched data from the 12 m telescope are summarized in table 3 , which lists the integrated intensities measured at each position for a given source .
a description of the basic data reduction and calibration is given in considerable detail in paper i. here , we give a brief overview and detail a few points that were not discussed more fully earlier .
we reduced the bima data using the miriad package @xcite .
we removed the instrumental and atmospheric phase variations from the source visibilities by referencing the phase to observations of a nearby quasar every 30 minutes .
the antenna - based frequency dependence of the amplitudes and phases were measured and removed using the bima online passband calibration at the time of each track .
using measurements of the rms phase variation over a fixed , 100 m baseline @xcite , we applied a zeroth - order correction for the atmospheric phase decorrelation ( paper i ) .
the main effect of this correction is to create a dirty beam that more accurately represents the atmospheric - limited response of the interferometer .
we set the amplitude scale of the bima observations using mars and uranus as primary flux calibrators .
following paper i , we assign a 1 @xmath28 uncertainty to the amplitude calibration of 15% .
we weighted the visibility data by the noise variance and also applied robust weighting @xcite to produce data cubes in right ascension , declination , and lsr velocity .
we included shadowed data down to projected baselines of 5 m , after checking for false fringes in the shadowed data ; this allowed better measurements of large - scale emission with the modest penalty of @xmath27 1% attenuation of the measured source amplitude . for most sources , we produced data cubes with a velocity resolution of 10 km s@xmath0 , though we used 5 km s@xmath0 for the narrow - line galaxy ngc 0628 , and 20 km s@xmath0 for ngc 3031 , ngc 4725 , and ngc 5033 . for purposes of uniform data reduction , we treated all observations as mosaics , with the 11/44 single - pointing observations a trivial subset of the mosaiced case .
following @xcite , we briefly describe the mosaicing process we used as implemented in the miriad package .
first , the standard mapping task invert produces a `` dirty '' image of each pointing separately and then combines the separate dirty maps into a linear mosaic by taking a weighted average of pixels in the individual pointings , where the weights are chosen to minimize the theoretical noise and to correct for the primary beam attenuation of the individual pointings .
( this `` sault weighting '' scheme results in a primary beam function that is flat across most of the mosaiced image but that is attenuated at the edges of the mosaic , so that the noise level does not become excessive there . for the single - pointing observations ,
the primary beam response is taken to be the usual truncated gaussian with fwhm of 100 at 115 ghz . ) because the @xmath20 coverages of the pointings are not identical , the synthesized ( `` dirty '' )
beam also differs slightly from pointing to pointing .
together with the sault weighting scheme , this means that the point spread function ( psf ) of the mosaiced image is position - dependent .
the invert task therefore also produces a cube of beam patterns , one for each pointing , so that the deconvolution can compute the true psf at any position in the dirty image .
we deconvolved the image cubes using a steer - dewdney - ito clean algorithm @xcite ; see appendix a for a discussion of alternative deconvolution schemes and issues concerning the deconvolution .
we did not set any clean boxes so as not to introduce any _ a priori _ bias as to where we expected emission to be detected .
we allowed the deconvolution to clean deeply , to a cutoff of 1.5 times the theoretical rms in each pixel ( see appendix a , final two paragraphs ) .
the deconvolved images were restored in the usual way by convolving the clean solution with a gaussian with major and minor axes fitted to the central lobe of one plane of the synthesized beam , then adding the residuals from the clean solution to the result .
we performed an iterative , phase - only multi - channel self - calibration on most detected sources .
we checked that the source position was not affected by self - calibration by cross - correlating the final bima - only map with the _ a priori _ phase - calibrated map ; in most cases , the source moved by @xmath27 0.1after self - calibration .
instead , the absolute registration of the maps are limited by uncertainties in the phase calibration .
test measurements of two point sources separated by @xmath9 11 a typical separation for a source and phase calibrator for bima song observations showed that the absolute registration of the maps should be accurate to @xmath9 0.4 .
we detected continuum emission at 112 ghz ( the lower sideband of the local oscillator ) in three sources : ngc 1068 ( m 77 ) , ngc 3031 ( m 81 ) , and ngc 4579 .
for all three , we subtracted the continuum from the spectral line data in the visibility plane by fitting a constant to the real and imaginary parts of those channels in the passband that were free of line emission .
we list the peak flux density levels of these three galaxies in table 2 along with 3 @xmath28 upper limits for the remaining sources ; the maps of the detected galaxies are shown in figure [ continuum ] .
we tested to see whether any of the emission was resolved by fitting gaussians to the three detections . the emission from ngc
1068 was marginally resolved ; the deconvolved continuum source size was 7.8 @xmath25 2.6 along position angle 34 , and the integrated source flux density was 59.9 @xmath17 8.1 ( formal ) @xmath17 9.0 ( systematic ) mjy .
these results are consistent with previous measurements of the 3 mm continuum in ngc 1068 , which is resolved into two components , each of about 40 mjy and separated by about 5 , one to the northeast of the other ( e.g. * ? ? ?
the emission from ngc 3031 , which is known to have a bright ( @xmath9 90 mjy ) , variable , flat - spectrum radio core ( e.g. * ? ? ?
* ; * ? ? ?
* ) , was consistent with that from a point source with a flux density of 161 @xmath17 6 ( formal ) @xmath17 24 ( systematic ) mjy .
the emission from ngc 4579 , which like ngc 1068 contains a sy 2 nucleus , appears to be resolved , with a deconvolved source size of 8.7 @xmath25 5.5 along position angle -69and an integrated source flux density of 45.0 @xmath17 7.5 ( formal ) @xmath17 6.8 ( systematic ) mjy .
the implementation of otf mapping at the nrao 12 m ( 3.2.1 ) ensures that the internal pointing consistency of an individual otf map ( taken over @xmath920 min ) is excellent , even if the overall registration of the map is uncertain .
we were further able to cross - correlate either the individual otf maps or averages of several maps with each other to track slow pointing drifts , allowing otf maps taken over many hours , days , or even in different observing seasons , to have good internal pointing accuracy .
we performed the cross - correlation using the convolution theorem : the cross - correlation function of the two maps is the inverse fourier transform of the product of the fourier transforms of the two input maps .
since the input maps had many channels , we summed the resulting cross - correlation function over all channels and fitted the position of the peak using a two - dimensional gaussian .
using the sum of the cross - correlation function ensures that those channels with the highest values for the cross - correlation are weighted the most .
the absolute registration of the 12 m maps was accomplished by cross - correlating the well - pointed average 12 m map with the bima map , where the bima map was smoothed to match the 55 resolution of the 12 m at 115 ghz . in performing this crosscorrelation , we had to assume that the extended emission estimated by the deconvolution but not sampled by the interferometer ( appendix a ) that is , the `` resolved - out '' flux in the interferometer maps correctly placed the centroid of the emission . a theoretically more robust method , comparing flux density just in the region of @xmath20 overlap ,
was limited by poor s / n in the overlap region of the 12 m data . in general , the resulting pointing corrections were 5 - 10 , but some were as large as @xmath920 .
if a pointing offset of this magnitude is left uncorrected , it has the effect of inserting a phase shift into the combined map , so that there will be symmetrically placed positive and negative residuals in the final combined map .
an example of this is shown in figure [ offset ] , which shows a combined bima+12 m map of ngc 5194 before ( _ left _ ) and after ( _ middle _ ) correcting a simulated 20 offset between the 12 m and bima maps .
the difference between the uncorrected and corrected maps ( _ right _ ) emphasizes the symmetric positive and negative errors introduced in the combined map .
the peak error in the difference map is nearly 20% of the peak value in the correct map of the galaxy .
for the 24 galaxies with complete otf mapping from the nrao 12 m ( table 2 ,
3.2.1 ) , we combined the bima data with the single dish data so that the large - scale emission is accurately imaged . in appendix b
, we discuss different techniques for combining interferometric data with single dish data . for this paper
, we used the combination method described in @xcite ( with a clean - based deconvolution instead of their mem ) .
first , we created a new `` dirty '' ( undeconvolved ) map using a linear combination of the bima `` dirty '' map and the otf map , where the otf map was tapered by the bima primary beam function and resampled onto the same @xmath29 grid as the bima dirty map .
for the linear combination , we chose to weight the images in inverse proportion to the beam areas .
( in principle , one could choose any number of weighting schemes .
in practice , however , putting both images on a common temperature scale is the appropriate choice in the relatively low s / n regime of millimeter data , where there may be a significant contribution to the total flux in the residual map ; see discussion in appendix a. ) we also created a new `` dirty '' beam using the same weights for a linear combination of the bima synthesized beam and the 12 m beam , where the 12 m beam was assumed to be a truncated gaussian .
we deconvolved the new `` dirty '' map by the new `` dirty '' beam using a steer - dewdney - ito clean algorithm as described in 4.1 .
as with the bima - only maps , we restored the final images by convolving the clean solution with a gaussian ( fitted to the central lobe of one plane of the combined `` dirty '' beam cube ) and then adding the residuals from the clean solution to the result .
the full catalog of bima song figures is available online at + http://astro.berkeley.edu/ thelfer / bimasong_supplement.pdf and through the online edition of the astrophysical journal at + http://www.journals.uchicago.edu/apj/journal . for each source ,
the catalog shows the following : \1 .
( _ upper left panel _ ) : the distribution of co integrated intensity ( `` moment-0 '' ) emission is shown by red contours overlaid on an optical or infrared image of the galaxy .
the integrated intensity maps were formed in the following way .
first , we created a mask by smoothing the data cube by a gaussian with fwhm = 20 and then accepting all pixels in each channel map where the emission was detected at a level above 3 times the measured rms noise level of the smoothed cube .
second , we summed all the emission in the full - resolution data cube after accepting only those pixels that were nonzero in the mask file . finally , we multiplied by the velocity width of an individual channel to generate a map of integrated intensity in the units of jy bm@xmath0 km s@xmath0 .
this smooth - and - mask technique is very effective at showing low - level emission which is distributed in a similar way to the brighter emission in the map .
such low - level distributions are common , since the deconvolution often leaves systematic , positive residuals in the shape of the source ( appendix a ) , which is almost certainly true emission associated with the source .
another advantage of the smooth - and - mask technique is that it does not bias the noise statistics of the final image , unlike the commonly used method of clipping out low - level pixels in the channel maps .
however , the smooth - and - mask method may introduce a bias against any possible compact , faint emission that is distributed differently than the brightest emission .
for each map in the upper left panel , the contours are spaced logarithmically in units of one magnitude , or with levels at 2.51 times the previous level .
the starting contour is given in table 2 .
the solid blue contour denotes the gain=0.5 contour of the primary beam .
for the single - pointing observations ( table 2 ) , the primary beam response is unity only at the center of the map and falls off as a gaussian function at distances away from the center . for multiple pointing observations , on the other hand
, each galaxy has a uniform primary beam function inside a large region bounded by a dashed blue contour ; the gain falls from 1 to 0.5 between the dashed and solid blue contours .
( _ upper middle panel _ ) : the same co integrated intensity emission shown in the upper left panel is shown as black contours overlaid on a false - color representation of the co emission .
the contours are exponentially spaced at half - magnitude intervals , or 1.59 times the previous interval .
the starting contour is given in table 2 .
the false - color minimum is set to the value of the lowest pixel in the map , which is on average -2.5 jy bm@xmath0 km s@xmath0 .
the false - color maximum for all maps is 315 jy bm@xmath0 km s@xmath0 , so that the reader can easily determine the relative strengths of the flux density for the different sources .
the fwhm of the synthesized beam is shown near the lower left corner of this box , next to a vertical bar that shows the angular size of 1 kpc at the assumed distance to the source .
( _ upper right panel _ )
: a representative co velocity field is shown in false color .
the fields were generated by taking the mean velocity at each pixel from gaussian fits as follows .
first , a single gaussian was fit to the spectrum at each pixel in the datacube .
the initial estimates for the amplitude and velocity of the gaussian were taken from the peak of the spectrum at each pixel . the initial estimate for the fwhm of the gaussian
was taken to be 30 km s@xmath0 or 60 km s@xmath0 ( the fitting routine was run for both values , and the results for 60 km s@xmath0 were used when the 30 km s@xmath0 fit failed ) .
the gaussian fits were rejected if the velocity error exceeded 2 channels or if the integrated flux density of the gaussian was less than @xmath30 , where @xmath28 is the rms noise level in a channel map and @xmath31 is the velocity width of a channel .
finally , the same spatial mask used for the integrated intensity images ( described in `` upper left panel '' above ) was applied to the velocity fields .
the wedge at the right of each velocity field shows the range and values of the velocities for each galaxy .
velocity contours are overlaid on the false - color images ; the contours are spaced at 20 km s@xmath0 intervals , except for the narrow - line galaxies ic 342 , ngc 0628 , and ngc 3184 , which are spaced at 10 km s@xmath0 intervals , and the wide - line galaxies ngc 3368 , ngc 4258 , ngc 5005 , ngc 5033 , and ngc 7331 , which are spaced at 40 km s@xmath0 intervals .
( _ lower left panel _ ) : the individual channel maps are shown over the velocity range of emission .
the bima song data are shown as a false - color halftone , the stretch of which is shown in a wedge to the right of the last panel and which ranges from @xmath32 to 0.75 times the peak value in the channel map .
the gain=0.5 contour is shown as a solid blue contour in the upper left panel .
the lsr velocities are given for the first , second , and last panels of the map ; the velocity width of each channel may be inferred from the difference between the velocities listed in the first and second panels . where present , light gray contours show emission measured in otf mapping at the nrao 12 m telescope .
the contours start at 3 @xmath33 , where @xmath33 is the measured rms noise in the 12 m channel map ( listed in table 2 ) , and are spaced by factors of 1.5 .
( _ lower right panel _ ) : spectra comparing the bima song maps at 55 resolution are shown along with corresponding 55spectra from the nrao 12 m telescope .
for those galaxies mapped with otf mapping at the 12 m , a grid of spectra is shown overlaid on a grayscale representation of the integrated intensity map .
the spectra are spaced by 30 centered on the tracking center of the source as listed in table 1 ; the location of each spectrum is marked by a light blue cross .
the red spectrum is the bima - only data , smoothed to 55 ; the blue is the bima+12 m combined data , smoothed to 55 , and the green is the 12 m - only data , at its full resolution of 55 .
the lsr velocity and flux density scales are shown in a frame around the lower left spectrum .
for those galaxies not mapped with otf mapping at the 12 m , we show instead individual spectra taken at each of the pointing centers of the bima mosaic , at ( 0,0 ) , ( @xmath1744,0 ) , and ( @xmath1722,@xmath1738 ) .
again , the red spectrum is the bima - only data , smoothed to 55 , and the green spectrum is the 12 m - only data .
since neither co nor infrared flux density were selection criteria for the song sample , it is of interest to see what the detection rate of the survey was . of the 44 galaxies
included , only three or 7% were not detected in line emission .
these are ngc 3031 ( which was detected in the continuum ; see 4.2 ) and ngc 4699 , which were also not detected with the nrao 12 m telescope , and ngc 3992 , which was formally detected in at least one position with the 12 m ( table 3 ) .
we note that ngc 3031 ( m 81 ) is a large galaxy with known co emission at @xmath34 120 , which is outside our field of view ( e.g. * ? ? ?
this means that 42/44 = 95% of the song galaxies have known co emission .
the high detection rate for co is not unexpected for a magnitude - limited sample , since optically bright galaxies might be expected to be brighter at all wavelengths .
another contributing factor is likely to be the hubble type distribution ( figure [ typehist ] ) , since hubble types outside the sa
sd range were explicitly excluded , and there are few early - type spirals ( sa sab ) because the sample is also volume - limited and such galaxies are rarer . for comparison , @xcite detected co in 96% of sc galaxies , 88% of sa scd galaxies , and 79% of galaxies over all hubble types .
the tendency for co to be brightest in intermediate - type galaxies can be understood in terms of galactic evolution : early type galaxies such as ellipticals and s0 galaxies have converted most of their gas into stars , whereas late - type galaxies , while gas rich , are generally low in metallicity @xcite , and thus less likely to show detectable co emission .
the nrao 12 m otf data we have collected for 24 song sources comprise the most extensive collection of fully - sampled , two - dimensional single - dish maps of external galaxies to date .
we have measured the global flux densities directly in these galaxies , without having to rely on model fitting of data at selected positions and the extrapolation of these models to larger radii . for 19/24 galaxies ,
we mapped a core square region of 6@xmath256 with uniform sensitivity ; an additional 1 which was included for a ramp - up or ramp - down distance extends the total region mapped to 8@xmath258 , though the outer 1 ( square ) annulus has significantly lower sensitivity than the core region of the map . for ngc 3938 , ngc 5033 , and ngc 5247 , the core region was 5@xmath255 ;
for ngc 5457 , the core region was 7@xmath257 ; and for ngc 5194 , the core region was 8@xmath2512 .
the global flux densities , measured inside the core regions , are listed in table 4 .
the integrated flux densities are also shown as a function of the length of a square region over which the flux density was measured in figure [ boxflux ] . for most of the sources ,
the flux density at the core length @xmath35 ( denoted by the vertical dotted lines ) has achieved a plateau , which suggests that there is very little emission from co at larger regions than measured . for large sources like ic 342 , the flux density is clearly still increasing at large @xmath35 ; this is consistent with the known co distribution of this galaxy , which extends to nearly 15@xmath2515 on the sky @xcite .
some of the sources , like ngc 4569 and ngc 5005 , show a large increase in flux density outside the core region even though the emission appears to have achieved a plateau inside the core region . while it may be that the flux density for these sources truly increases outside the core region , the error bars on the points at large @xmath35 are large enough that the measured points are also consistent with the flux density measured at the core length .
we compare our measured otf global flux densities with global flux densities derived from model distributions using the fcrao survey @xcite in figure [ globflux ] . in general
, the agreement is very good ; the average ratio of the otf to fcrao flux density is 1.05 , with a standard deviation of 0.19 .
the worst discrepancies are for the galaxies with the lowest global fluxes , like ngc 3351 and ngc 4258 , where the otf measurements are both 2.2 times higher than the fcrao flux density , or ngc 3938 , where the otf measurement is 0.53 times the fcrao flux density .
this tendency was also apparent when we compared integrated intensities from the position - switched data from the 12 m telescope ( table 3 ) with those at matched positions from the fcrao survey : the weaker sources disagreed by as much as a factor of @xmath9 2 , while the stronger sources agreed to within about 20% .
table 4 also lists the co luminosity and total co mass inside the region measured for each otf source .
assuming a constant co / h@xmath36 ratio of 2 @xmath37 @xmath38 ( k km s@xmath0)@xmath0 , the calculated masses range from 2.4@xmath2510@xmath39 m@xmath40 for ngc 4736 to 7.4@xmath2510@xmath41 m@xmath40 for ngc 5248 ; the average of the measured otf masses is 3.3@xmath2510@xmath41 m@xmath40 .
an interferometer is fundamentally limited by the minimum spacing of its elements . because two elements can never be placed closer than some minimum distance @xmath42 , signals on the sky larger than some size @xmath43 @xmath44 will be severely attenuated .
this effect is commonly referred to as the `` missing flux '' problem . in this section
, we present an analysis of the issue of large - scale flux recovery as measured using a comparison of bima - only data with the total flux as measured at the nrao 12 m telescope .
we emphasize that for the 24 bima song maps that incorporate 12 m otf data into them , the final bima song maps presented in this paper are not missing any flux .
nonetheless , an analysis of the bima - only data for these sources leads to some new results on this technical issue , which are discussed below . with the bima song observations for motivation
, @xcite recently presented a study of imaging simulations of large - scale flux recovery at millimeter wavelengths .
these simulations showed that the nonlinear deconvolutions that are routinely applied to interferometric maps interpolate and extrapolate to unsampled spatial frequencies and thereby reconstruct much larger scale structures than would be suggested by an analytical treatment of the flux recovery problem .
however , @xcite also showed that fraction of flux density recovered for a given observation is a function of the s / n ratio of the map .
this is illustrated in figure [ sn ] , which shows the fractions of total and peak flux densities as a function of the s / n ratio for simulated bima observations of sources with widths of 1040 . for s / n ratios of 1040 , which are typical of individual channels for bima song maps , we could expect to reconstruct roughly 8090% of the total flux density of a 10 source and roughly 3050% of the total flux density of a 40 source , whereas for noise - free simulations , bima recovers essentially all of the total flux density of a 10 source and 80% of the total flux density of a 40 source ( see table 2 in * ? ? ? * ) .
the fraction of the peak flux density recovered ( bottom panel of figure [ sn ] ) is typically somewhat higher than the fraction of total flux density recovered . in each of the 24 galaxies for which we have complete otf mapping from the nrao 12 m telescope ( table 2 )
, we were able to make a pixel - by - pixel measurement of the flux recovery of the bima - only maps . to do this , we first divided the bima data cube by the appropriate ( mosaiced or non - mosaiced ) primary gain function ; we then smoothed the data cube to the 55 resolution of the 12 m maps .
finally we summed over the planes with emission in order to make a map of integrated intensity .
we made the corresponding integrated intensity maps of the 12 m emission and then formed ratio maps of the smoothed bima - only maps to the 12 m maps .
finally , we masked the ratio maps to show only those pixels that are non - zero in the masked , clipped moment maps shown in the bima song catalog in 5 , and we clipped outside the original gain=0.5 contour ( also shown in the maps in 5 ) .
this masking and clipping serves the purpose of eliminating pixels with the highest formal errors in the ratio maps . given the assigned 1 @xmath28 uncertainty of 15% for the flux density in the bima maps and 15% for the 12 m maps , the 1 @xmath28 error in the ratio maps is 21% .
the flux recovery ratio maps are shown in figure [ recov ] .
we can see from the figure that there is a remarkable variety in the flux recovery from source to source : for ngc 1068 , ngc 4258 , ngc 4826 , ngc 5005 , and ngc 6946 , essentially all of the single - dish flux density is recovered by the bima maps , at least in the regions of brightest emission . at the other extreme , for ngc 0628 , ngc 3938 , and ngc 5247 , only 3050% of the total single - dish flux density is recovered over most of the strongest emitting regions .
the remaining 16 sources recover intermediate amounts of flux density , with values typically about 7090% over the brightest regions . clearly , the ability of any given interferometer to reconstruct structures in a source is very dependent on the source itself .
it is clear from figure [ recov ] that the fraction of flux density recovered by a given telescope is a function of location within the map . even for a source like ngc 6946 , which has a dominant , centrally concentrated region of emission in the nuclear region which is fully recovered by the interferometer ( figure [ recov ] )
, we can see that away from the nucleus , the fraction of flux density recovered falls to 3040% .
this is also apparent in the grid of spectra shown in the catalog figure for ngc 6946 .
this trend is generally apparent ( if we exclude galaxies with ring - like emission , like ngc 3521 and ngc 7331 ) : the farther from the nucleus , the less the fraction of flux density recovered .
we can attribute this trend to three factors : first , for any given velocity in a data cube , there is a larger region of sky available to emit at locations away from the center of the galaxy rather than at the center ( that is , the velocity gradient is largest at the nucleus , which tends to confine emission near the nucleus to small widths in individual channel maps ) .
second , the tendency is for the brightest emission and best sensitivity to be near the center , so that the s / n ratio is highest in the central region . from figure [ sn ] and the description of imaging simulations above ( 8.1 ) , we would expect the deconvolution to reconstruct less of the total flux in regions of lower s / n .
third , for the bima song mosaiced galaxies , only the central pointing of the 7-pointing observations had the full benefit the @xcite effect .
@xcite argued that each @xmath20 point measured included contributions from effective baselines as short as @xmath45 , where @xmath42 is the minimum ( projected ) center - to - center separation of the dishes and @xmath46 is the dish diameter , and they showed that one could effectively extend the @xmath20 sampling inwards to @xmath45 by scanning the telescope over the source .
however , for the 26-field mosaic of ngc 5194 as well as the single - pointing observations of ngc 4414 and ngc 5005 ( and to a lesser extent , ngc 5248 and ngc 5247 ) , the fraction of flux density recovered still falls as a function of distance from the nucleus .
we may conclude that the lack of the required ekers and rots sampling for the outer fields of the bima song mosaics is somewhat less of a valid explanation for the observed trend than the other two factors described above .
the dependence of the flux recovery on the s / n ratio of the interferometric data predicted by the simulations is readily apparent in real data .
figure [ sndata ] shows the ratio of total flux recovered as a function of the s / n ratio for each of the 24 galaxies with otf maps . with the possible exception of ngc 1068
, each source clearly shows the predicted trend .
( from inspection of the ratio map of ngc 1068 in figure [ recov ] , those pixels with the highest measured ratios probably have the highest formal errors in the ratio as well . ) for the 24 galaxies shown in figure [ recov ] , we have of course incorporated the 12 m otf data in the bima song maps so that the total flux density is accurate for these maps . furthermore , structures larger than 55 should be accurately represented in the bima+12 m song maps , and structures @xmath27 20 should also be accurately represented in the maps .
it is difficult to be confident of the accuracy of structures on the intermediate size scales of 2055 . in all likelihood
, these structures are somewhat misrepresented as their attenuated bima - only structures sitting atop a smoothed - out 12 m plateau .
the remaining 20 bima - only maps typically have lower s / n ratios than the bima+12 m maps ; this is because co brightness was a selection criterion in deciding which sources to observe with otf mapping at the nrao 12 m. these galaxies were instead observed in a simple position - switched mode at the 12 m. given the lower s / n ratios , we would expect to reconstruct less flux density than for the brighter sources shown above ( 8.1 ; * ? ? ?
* ) . for the bima - only galaxies , we compared integrated intensities of spectra taken with the nrao 12 m telescope at each of the bima pointing centers with the flux density measured in the bima - only maps , smoothed to 55 .
as with the bima+12 m maps , in order to compute the ratio of flux density recovered , we first divided the bima data cube by the ( mosaiced or non - mosaiced ) primary gain .
we then summed those planes in the smoothed data cube that correspond to the velocity limits of integration in the 12 m spectra .
the 12 m and 55-smoothed bima spectra are displayed in the bima song catalog in 5 ; the ratios of flux density recovered are also listed in the last column of table 3 along with their formal uncertainties .
given the assigned 1 @xmath28 systematic uncertainties of 15% in the bima maps and 25% in the 12 m spectra ( 3.2.2 ) , the systematic 1 @xmath28 error in the ratios are 30% .
of the 97 positions observed at the 12 m and listed in table 3 , we detected 55 positions .
of these 55 positions that were detected at the 12 m , the ratios of the bima-55 flux density to 12 m flux density are measured at 21 positions with a @xmath47 2 @xmath28 confidence level .
the ratios for all 55 positions are plotted in figure [ ratio_sn ] , which shows the 21 detected ratios as open squares and the remaining 34 ratios as filled squares .
as the figure shows , there is a clear trend to detect higher ratios where the s / n of the bima data is higher .
furthermore , as with the bima+12 m maps , the ratio of flux density recovered tends to be higher in the central position than in the outer regions of the mosaic , at least for the brighter galaxies .
we may conclude that , given better s / n measurements , we would likely be able to reconstruct more flux density than has been measured here .
furthermore , following figure [ sn ] , we might expect that the peak flux density that is measured in the bima song maps is a more reliable indicator of the flux density in the source than the total flux density is .
this is exemplified in the bima and 12 m spectra shown for ngc 3184 in the bima song catalog entry ( see + http://astro.berkeley.edu/ thelfer / bimasong_supplement.pdf ) , where the ratio of peak flux density recovered for the central region is about 75% ; compare this with the ratio of total flux density recovered of 54% from table 3 .
from the results of @xcite and from the above examples , it is clear that it is very difficult to conclude whether some or all of the bima - only song maps are systematically missing large - scale emission , or whether the song maps are instead reasonable representations of the sources , albeit at low s / n
. it may be reasonable to assume that those maps of galaxies with only compact structure apparent , like ngc 3726 or ngc 4725 , may be fairly accurate representations of the sources .
maps of galaxies with clearly extended structures , like ngc 3184 or ngc 4535 , are almost certainly missing some of the larger - scale flux density .
in the fcrao extragalactic co survey @xcite , only 15% of the 193 galaxies observed at multiple positions at 45resolution showed molecular distributions that lacked a central peak . in paper
i , we showed that our single - dish data were consistent with this result , namely , that all 15 paper i galaxies showed single - dish radial profiles that rise monotonically to the nucleus .
thus , it appears that the bulk of molecular gas in nearby spiral galaxies tends to be concentrated within the central @xmath48 5 kpc . at 6 resolution , however , about half of the song subsample did not exhibit central peaks ( paper i ) , a result that was unlikely to have been caused by differences in the two samples . for the full sample
, we can quantify the degree of the central concentration by comparing the molecular surface density at the center of the galaxy to the peak molecular surface density in the galaxy .
the ratio of central to peak surface brightness should be fairly insensitive to the amount of flux resolved out , so even for the 20/44 maps which lack single - dish data , the ratio should be a valid measure of the central peakedness .
we list the peak surface density and the central surface density ( taken as the peak within the central 6 beam ) in table 5 , and we show a histogram of the distributions in figure [ centerfrac ] .
as figure [ centerfrac ] shows , just 20/44 or 45% of the song galaxies have their peak molecular surface densities within the central 6 or @xmath9 360 pc . at the other extreme , 6/44 or 14% of the galaxies lack any detectable molecular gas within the central few hundred pc .
the remaining 18/44 sources show intermediate fractions , where there is detected co at the center but at a level that is weaker than the peak surface density .
we note that two of the galaxies with particularly low fractions
ngc 0628 ( 0.20 ) and ngc 5194 ( m51 , 0.28 ) are grand - design spirals with two dominant spiral arms that may be traced all the way to the nuclear region .
the integrated intensity maps of these galaxies suggest a continuity of the large scale structure down to the 100-pc scales in the nuclear regions of these galaxies , in contrast to galaxies like the milky way ( 10 ) or song galaxies like ngc 6946 , which show central concentrations in the inner @xmath9 500 pc that are an order of magnitude higher in molecular surface density than are seen in their disks .
galaxies like the milky way and m31 produce most of the optical luminosity in the nearby universe because they have luminosities close to the knee in the luminosity function ( l@xmath49 ) ( e.g. * ? ? ?
it is reasonable then to ask the following : to what degree are their co distributions represented in the song sample ? in this section , we present a basic comparison of local group molecular morphologies and brightnesses to those from the survey .
we defer any more complex analysis of underlying causes for the morphologies , whether caused by bars or other dynamical considerations @xcite , galaxy luminosity , size , or other factors .
we first determine the characteristic sensitivity of the song survey .
the typical rms noise is about 58 mjy in a 6.1 beam when the velocity channels are averaged to 10 km s@xmath0 , a velocity resolution comparable to the fwhm linewidth of a gmc in the milky way @xcite .
( the best sensitivity achieved by song is a factor of 2.5 lower than this average ; see table 2 . )
using a co / h@xmath36 conversion factor of 2 @xmath37 @xmath38 ( k km s@xmath0)@xmath0 , this corresponds to an rms column density sensitivity of 4.6 m@xmath40 pc@xmath50 in each 10 km s@xmath0 velocity channel .
isolated points in the song maps are probably not significant at less than 3@xmath28 or 13.7 m@xmath40 pc@xmath50 , but the edges of extended emission are probably reliable at 2@xmath28 or 9.1 m@xmath40 pc@xmath50 .
the typical spatial resolution in a 6.1 beam is 350 pc at the median distance of 11.9 mpc , but is a factor of 5 smaller for the nearest galaxies . for comparison ,
the mean surface density of a gmc in the solar vicinity is 50 - 100 m@xmath40 pc@xmath50 @xcite , within an area of 2.1 @xmath51 pc@xmath52 .
the distribution of co in the milky way can be described as having a central molecular disk , a zone of small but indeterminate surface density out to about 3 kpc , increasing to a peak at 4 - 7 kpc and declining monotonically thereafter ( e.g. * ? ? ?
* ; * ? ? ?
the inner disk has a radius of about 300 pc and a surface density of about 500 m@xmath40 pc@xmath50 @xcite ; this surface density may be an overestimate if the co / h@xmath36 conversion factor is low as suggested by some studies @xcite , and it is in any event accurate only to within a factor of two .
the disk is probably oval in response to the central bar @xcite .
beyond the central disk , the surface density declines to a small value , and then increases abruptly at a distance of about 3 kpc , reaching a broad maximum of about 6 m@xmath40 pc@xmath50 at a distance of about 4 kpc from the center , thereafter declining approximately exponentially to a distance of about 15 kpc
. observed at an inclination of 45 , the azimuthally averaged column density at the peak of the distribution in the disk would be about 8.4 m@xmath40 pc@xmath50 , which is close to the detection limit of our survey .
since the co appears to be concentrated in the spiral arms , the local density might be as much as two or three times this value , and the peak of the molecular distribution in the disk would be detectable in the song survey .
thus , the milky way would appear in our survey as a bright , somewhat resolved central disk with a ring of spotty emission at a distance of 4 - 7 kpc .
we note first that at the average distance in our sample of 12 mpc , the inner disk of the milky way would contain only two or three resolution elements , and the so - called molecular ring would lie at a radius of 60- 90 . for galaxies at half
that distance , only the central disk would be observed , with a diameter of about 20 , and possibly quite elongated .
twenty of the 44 galaxies in the song have peak surface densities in excess of 300 m@xmath40 pc@xmath50 ( table 5 ) , so the large central disk in the milky way is not at all unusual , even taking into account the large uncertainty in the mass .
fewer of the galaxies have a high surface density central disk along with a large diameter , low surface density ring , but perhaps the distribution closest to that of the milky way is ngc 3351 .
it has a small inner elongated molecular structure with a peak surface density of about 560 m@xmath40 pc@xmath50 .
the inner co is nearly perpendicular to the large - scale optical bar , suggesting that the gas resides on x@xmath36 orbits , similar to what is inferred for the molecular gas in the bar of the milky way @xcite .
outside the central structure , the co is absent to about 50 ; the co becomes just detectable at a distance of about 3 kpc , similar to what is seen in the milky way .
the surface brightness of the molecular ring in this galaxy seems a bit smaller than that in the miky way , however .
another galaxy with a similar co morphology and central co disk is ngc 4535 .
its central co disk has a surface density of 610 m@xmath40 pc@xmath50 , which is similar to that of ngc 3351 , but the gas that makes up its molecular ring has a somewhat higher surface density than that in the milky way .
the co distribution of the milky way thus appears to be bracketed by these two galaxies , and is not too dissimilar from several others .
the total mass of molecular gas in the milky way is about 1.2 @xmath25 10@xmath53 m@xmath40 @xcite .
this is somewhat higher than the median mass of the song sources , 5 @xmath25 10@xmath54 m@xmath40 , though it is somewhat lower than the average of the sample ( where we inferred the non - otf masses from measurements of 15 sources by * ? ? ?
* ) , which is about 3 @xmath25 10@xmath41 m@xmath40 .
the masses of ngc 3351 and ngc 4535 are 1 @xmath25 10@xmath41 m@xmath40
( table 5 ) and 3 @xmath25 10@xmath41 m@xmath40 ( from * ? ? ?
* ) , which are comparable in magnitude to that of the milky way .
the co distribution in m31 is quite different .
it has a very low surface density of molecular gas inside a radius of 6 kpc @xcite .
like the milky way , the co declines monotonically beyond 8 kpc with a similar surface density and radial distribution . however , the peak azimuthally averaged surface density is only about 2 m@xmath40 pc@xmath50 and reaches local maxima of about 3 m@xmath40 pc@xmath50 . taking into account the increase due to inclination effects , and non - axisymmetry ,
the peak observed column density could rise to as much as 5 m@xmath40 pc@xmath50 , but probably not much more .
m31 would thus probably be undetectable , or at best , be marginally detectable as a thin , very patchy ring with a radius of 8 kpc , if the inclination angle were higher than average . in the song survey , even if such a ring were detectable , it would be observed only at radii beyond 60 in the most distant galaxies . in terms of general morphology ,
several galaxies have large diameter rings , which may be the peaks of exponentially declining distributions , without central co concentrations .
these include ngc 2841 , ngc 3953 , and ngc 7331 , though in all of these cases the ring is smaller ( @xmath9 5 kpc ) and has a larger surface density .
thus the overall co morphology of m31 is not unusual in our sample .
the closest match to m31 appears to be ngc 3953 , with a mean surface density of about 12 m@xmath40 pc@xmath50 .
however , the co mass of m31 ( @xmath9 2 @xmath25 10@xmath39 m@xmath40 , based on * ? ? ?
* ) is considerably lower than that we infer for ngc 3953 based on the fcrao survey ( 4 @xmath25 10@xmath41 m@xmath40 , * ? ? ?
several other distant galaxies have little or no molecular gas observed at our limiting surface brightness : ngc 3992 , ngc 4450 , ngc 4548 , and ngc 4699 .
the total masses we infer from the fcrao survey for ngc 4450 and ngc 4548 are @xmath9 10@xmath41 m@xmath40 ; ngc 3992 and ngc 4699 were not measured at fcrao .
some of these galaxies might be m31 analogues , but much deeper observations are needed to determine this .
m33 is not an l@xmath49 galaxy , with a luminosity of only about 20% that of the milky way ; in fact , at the mean distance of galaxies in the survey , m33 would not have satisfied the optical brightness selection criterion to be included in the bima song sample .
however , we may still ask whether it would have been detected at the mean distance of the survey . the peak co surface density is about 1.5 m@xmath40 pc@xmath50 @xcite when averaged in radial bins 500 pc in width .
although this surface density is well below the survey sensitivity , would it be possible to detect some of the brightest of the individual gmcs ?
in m33 , there is no measurable diffuse co emission ; the co is concentrated in well defined gmcs .
the brightest of these gmcs has a mass of 6.6 @xmath55 m@xmath40 and has a mean diameter of 74 pc , so that the mean surface brightness is 153 m@xmath40 pc@xmath50 . at the mean distance of our sample
, the emission would be beam diluted to be 6.8 m@xmath40 pc@xmath50 in a 350 pc beam , which is just below our sensitivity . because the emission is in well - separated clouds
, inclination effects would not be important until the inclination became quite high .
however , m33 would be clearly seen at the distance of some of the nearer galaxies . at the distance of 4 mpc , for example
, the brightest cloud would have an observed surface density of about 61 m@xmath40 pc@xmath50 , which is well above the song detection threshold . from the catalog of @xcite , which lists 146 gmcs ,
23 would be detected at the 3@xmath28 level and 31 at 2@xmath28 if m33 were at a distance of 4 mpc and observed as part of bima song .
some of these might appear to be connected if their projected separations were less than or about equal to the putative 120 pc beam at 4 mpc .
the survey contains a close analogue to m33 both in optical appearance and co distribution : ngc 2403 .
this galaxy is an sc galaxy at a distance of 4.2 mpc , and is shown in the hubble atlas @xcite on adjoining panels to emphasize their similarity .
the observed co distribution of ngc 2403 is in fact consistent with originating in individual gmcs like that in m33 .
the total mass of m33 , or 3 @xmath25 10@xmath56 m@xmath40 @xcite , is close to that of ngc 2403 , which we infer to be 7 @xmath25 10@xmath56 m@xmath40 from @xcite .
at least one song source has an inferred mass lower than m33 : ngc 2976 , with 2 @xmath25 10@xmath56 m@xmath40 . like ngc 2403
, the co emission from ngc 2976 appears to be consistent with originating in individual gmcs . in summary ,
all of the local group spiral galaxies have co morphologies that are represented in the song survey .
however , the co luminosity of the milky way is somewhat below average , and for m31 and m33 , they are well below average .
we have presented the basic millimeter - wavelength data for the bima survey of nearby galaxies , an imaging survey of the 3 mm co j = 10 emission within the centers and disks of 44 nearby spiral galaxies .
these are some of the main conclusions of the initial presentation of the data . \1 .
although the sample was not selected based on co or infrared brightness , we detected 41/44 or 93% of the sources .
together with one nondetection that has known co at radii outside the song field of view , 42/44 = 95% of the bima song galaxies have known co emission .
we detected three of the sources ( ngc 1068 , ngc 3031 , ngc 4579 ) in continuum emission at 112 ghz .
the continuum emission from two of the sources ( ngc 1068 and ngc 4579 ) appears marginally resolved , while the emission from ngc 3031 ( m81 ) is consistent with a point source . \3 .
we showed that the fraction of total flux density recovered by the interferometer alone is a function of location within each map .
as predicted by imaging simulations , the fraction of total flux density recovered is also a function of the signal - to - noise ratio of the interferometric data . \4 .
while single - dish molecular gas surface densities show that most molecular gas in nearby galaxies is confined to the central @xmath48 5 kpc , we showed that , on smaller scales , only 20/44 or 45% of the song galaxies have their peak molecular surface densities within the central 6 or @xmath9 360 pc .
this suggests that it is in fact quite common for molecular distributions _ not _ to peak at the centers of spiral galaxies .
the three local group spiral galaxies have co morphologies that are represented in the song survey . the milky way would clearly have been detected at the average distance of the survey , though it would have been below the average brightness at that distance .
m31 would have been marginally detectable at this distance , especially if its inclination angle were higher than average .
the azimuthally - averaged molecular gas surface density in m33 is far below the detection threshold of the survey , but since the emission in m33 is concentrated in well - defined clouds , we would have clearly detected emission from these clouds if m33 were placed somewhat nearer than the average distance of the survey .
the data from this survey are publicly available at the astronomy digital image library ( adil ) at + http://adil.ncsa.uiuc.edu/document/02.th.01 and at the nasa / ipac extragalactic database ( ned ) at + http://nedwww.ipac.caltech.edu/level5/march02/song/song.html .
the public data include the bima song integrated intensity images , channel maps , and gaussian fit velocity fields .
continuum images are included for the detected galaxies .
associated files including the primary beam gain functions are also included , as are the single - dish otf channel maps and integrated intensity images .
a generic informational file offers guidance to the released data .
the authors may be contacted for further information .
we thank the bima board for granting the observing time to complete this observatory project and the staffs at hat creek and the national radio astronomy observatory for their assistance with the observations .
we thank john lugten and peter teuben for extensive discussions and assistance throughout this project .
this research was supported in part by nsf grants ast 99 - 81308 ( uc berkeley ) and ast 99 - 81289 ( university of maryland ) .
tth was supported during some of the research period by nsf grant ast 99 - 00789 ( university of arizona ) .
interferometric imaging generally relies on nonlinear deconvolution algorithms to help reconstruct the true source structure on the sky .
this is because , when the imaging algorithm grids the sampled visibility data onto the @xmath20 plane , it assigns all unsampled @xmath20 cells the unlikely value of zero .
the deconvolution algorithm works to find a structure on the sky that is consistent with all the sampled visibility data but that also provides a more plausible and robust model of the unsampled visibility data . because the @xmath20 coverage is inherently undersampled , there may in principle be multiple source structures that are consistent with the sampled @xmath20 data , and the resultant deconvolved image may not be a unique solution . in practice ,
different deconvolution techniques tend to give similar results , which gives some reassurance as to the general structure on the sky . however , there are subtle differences among the techniques that are especially apparent in the relatively low signal - to - noise regime of millimeter - wavelength interferometry , and we discuss these below . for a introduction to the concept of the deconvolution as well as a detailed discussion of the algorithms and
how well they work , see the excellent review by @xcite . the maximum entropy method ( mem )
works by fitting the sampled @xmath20 data to a model image that is designed to be positive and smooth where it is not otherwise constrained . as @xcite discuss , one common pitfall in mem
is that the total flux density may be seriously biased if the signal - to - noise ratio is low . using simulated observations
, @xcite showed that mem yields a reliable flux density when the signal - to - noise ratio is above about 1000 .
however , with lower s / n ratios , mem may introduce a positive bias in the model in the form of a large - scale plateau .
this `` runaway '' flux problem gets progressively worse as the s / n ratio drops . in millimeter interferometry ,
the relatively low source strengths and relatively high random and systematic errors conspire to ensure that we are almost always working in the low signal - to - noise regime ; for bima song , typical s / n ratios for the channel maps were on the order of 1040 . in bima song data
, we saw the `` runaway '' flux feature commonly with mem , where there could be arbitrarily large flux density added into the deconvolved images , depending on what was specified as the convergence criteria .
an example of this is shown in figure [ mosmem ] , which shows bima - only maps of integrated intensity for ngc 6946 made using mem and clean , and spectra of the total flux measured in the inner 200-diameter region using the different data cubes .
as seen in the maps , the flux density in the small - scale components tends to be accurately represented , so that the final mem and clean images look very similar to one another .
but as the spectra show , the low - level flux rides systematically higher with mem than with clean ; in this example , the total flux in the mem map is over three times higher than that in the clean map .
in fact , the total flux density in the bima - only mem map is more than twice the total flux density in the 12m - only map of the same region , which is clearly an unphysical result . in general , the deeper one tries to deconvolve using mem , the worse the `` runaway '' flux problem gets .
the mem algorithm performs somewhat better on a linear combination of the bima and 12 m data , as with the @xcite method described in appendix b. in principle , one could fine - tune the convergence criteria for each channel of each map to try to get an accurate flux determination . however , there are pitfalls to doing so : first , if one cuts the deconvolution off too soon , then residual sidelobes might remain in the map ; second , specifying the convergence criteria depends on _ a priori _ knowledge of the source structure and flux density . in practice
it is difficult to fine - tune mem to work well in the low signal - to - noise regime .
the clean algorithm , first introduced by @xcite , is an iterative procedure that builds a model of the source distribution using an ensemble of point sources . @xcite
discuss the many variations on the original algorithm contributed by different authors ; in bima song tests , we used the implementation by @xcite . since clean
does not suffer from the `` runaway '' flux problem that mem is subject to , we concluded that it was a far more reliable algorithm to use for accurate flux density representation for bima song maps ; therefore , the maps in this catalog were produced using clean .
while clean does not introduce flux density errors into the map , it also does not produce an ideal deconvolution .
as @xcite showed , clean does not do as well as mem at recovering large - scale structure in the map .
( these tests were done in the high signal - to - noise regime where mem does not suffer from the `` runaway '' flux problem . )
this is because the conventional clean source model is inherently built out of point sources , so that it is effectively biased against large - scale ( and especially low - level ) emission in a map , whereas mem inherently assumes a large - scale , smooth distribution as a default model .
a problem that is common to conventional clean algorithms as well as to mem is that both algorithms tend to omit systematically distributed residuals from the source models ; the residuals , even if at a level consistent with the 1 @xmath28 noise specification , take the general shape of the source distribution in each channel map ( fig .
[ reschan ] ) .
this feature can introduce a bias for algorithms like self - calibration , which depend on the model to be an accurate representation of the source structure .
it can also lead to an inaccurate flux density determination in the final map : the final ( `` restored '' ) map is usually made by convolving the source model ( i.e. the output from the deconvolution algorithm ) with an idealized , gaussian beam , and then adding the derived residuals to the result .
the smoothed model map therefore has units of flux density per `` clean beam '' , whereas the residual map has units of flux density per `` dirty beam '' .
if one is at all fortunate , then the dirty beam has an area that is not too dissimilar to that of the clean beam , so that the convolved model and the residuals are on the same flux scale .
but if there is a significant contribution to the total flux density that remains in the residual map , as may occur when there is a broad , low - level source , and if the clean beam and the dirty beam have slightly different areas , then the total flux density in the restored map might again be inaccurately represented .
the best one can do to alleviate this problem is to clean as deeply as practical , so that as much of the flux density is represented in the source model as is possible . to summarize , ( 1 ) the clean algorithm is probably a better choice than mem for imaging sources with s / n ratios @xmath27 1000 , which covers most non - masing sources at millimeter wavelengths with current telescopes .
mem may produce images with the correct peak flux density , but it may also produce images with unphysically high total flux density because of the `` runaway '' flux problem .
( 2 ) it is extremely advantageous to get as much of the source flux density into the deconvolution _
model _ as is possible , by cleaning as deeply as is practical . that way , even if the model and the residuals are on different flux density scales , it is only the noise measurement that is ill - defined . the signal - to - noise estimates
may then be affected , but the source flux density would be properly represented .
while the idea of incorporating single dish data into interferometric data is not new ( cf emerson 1974 for the first combined image h i in m33 , and vogel et al .
1984 for the first combined mm-@xmath57 image hco@xmath58 in orion ) , the practice is far from routine . the 24 combined bima@xmath5912 m maps presented in this paper comprise by far the largest collection of single dish @xmath59 interferometric maps published to date .
however , millimeter astronomers recognize the importance of incorporating single dish data for making accurate maps of large ( @xmath1 2030 at 115 ghz ) structures : for the international atacama large millimeter array ( alma ) that is currently under development , the twin capabilities of mosaicing many fields on the sky and of incorporating single - dish data into interferometric maps together drive some of the most ambitious technical specifications for the antenna design .
@xcite has presented an excellent , detailed discussion of techniques of combining mosaiced data with total power data , most of which have been implemented and tested on high s / n data taken at centimeter wavelengths . indeed , in noise - free simulations by @xcite , we recovered all the flux density in a resolved - out source by simply adding a discrete total power point measured at ( @xmath60)=(0,0 ) and relying on the deconvolution to do the proper extrapolation between the total power point and the interferometric data . with real millimeter data , problems with pointing offsets and drifts , relative calibration between the telescopes , phase noise and atmospheric decorrelation in the interferometer data , and deformation of the telescope dish and therefore the primary beam ,
combine to place severe constraints on the deconvolution . with the added difficulty that the signal - to - noise ratios are typically much lower
, data combination tends to be much more challenging in the millimeter regime than at centimeter wavelengths . for bima song
, we tried a variety of data combination techniques to determine what worked best empirically for our data . for many of our tests ,
it was possible to fine - tune a single spectral channel to achieve seemingly good pointing or calibration agreement , but there would be a fairly random scatter of such solutions across the 20 - 40 spectral channels that together made up a data cube .
we therefore tried to achieve a solution that gave the best results for the data cube as a whole . in this technique , first demonstrated on 3 mm spectral line data by @xcite , the single - dish data are sampled at spatial frequencies 0 @xmath61 @xmath60 @xmath27 @xmath62 , where @xmath62 is the diameter of the single dish telescope . in brief , an idealized single - dish primary beam response function is deconvolved from the single - dish otf maps , to form a model of what the source looks like on the sky .
for each of the interferometer pointing centers , a model field is then generated by multiplying the sky model by the idealized interferometer primary beam at that position .
the model field is then sampled at chosen @xmath20 points . in practice
, this technique can be fine - tuned to work very well , if one has enough patience .
unfortunately , the exact results tend to depend on the details of the parameters chosen , from the method for the single - dish deconvolution to the distribution , number and weighting of the @xmath20 points specified ( which can significantly alter the synthesized beam from the interferometer - only case ) . as a check
, one can model visibilities just in the overlap region of the @xmath20 plane that both the single dish and the interferometer sample and compare the calibration and structures in the resulting maps of the overlap visibilities ( e.g. * ? ? ?
* ) ; however , in relatively low s / n observations , this comparison of the overlap region is usually unsatisfactory . in the end , one can gauge the success of the data combination by comparing the final combined map , convolved to the resolution of the single - dish telescope , to the single - dish data alone . if by inspection the agreement is not good , one can go back , adjust the calibration or try a different ( deeper or more shallow ) single - dish deconvolution , and try again . in summary , while adding single - dish data in the @xmath20 plane can be very successful given enough attention to the combination , it is typically time - consuming and hard to automate .
there are both linear and nonlinear methods of adding the single - dish data to an interferometric map in the image plane . in the linear method ( implemented as immerge in miriad )
, the interferometric data are deconvolved separately .
the separate single - dish and interferometer images are each transformed and then merged in the fourier plane , using a tapering function in the overlap region to specify the weights of each data set .
one can check the goodness of fit in the overlap region by comparing the @xmath20 data derived from each of the two input images .
however , with the s / n ratios of 1040 per spectral channel that were typical for bima song data , the calibration in the overlap region depended rather sensitively on the exact region chosen for the overlap .
one also needs to be cautious when a large amount of flux density is resolved out by the interferometer ; in this case , the interferometer - only deconvolution may run into difficulties . in practice
, @xcite used this technique successfully for co - rich bima song galaxies .
@xcite presented a nonlinear joint deconvolution algorithm ( implemented as mosmem in miriad ) that sault later modified to incorporate single dish data . in this method
, a joint maximum entropy deconvolution is performed on a linear mosaic of the `` dirty '' ( undeconvolved ) interferometer fields along with the single - dish data ; the algorithm maximizes the entropy , subject to two separate @xmath63 constraints from the two data sets .
this technique has been applied very successfully for centimeter - wavelength data and for high s / n millimeter - wavelength data ( e.g. * ? ? ?
however , given the low s /
n `` runaway '' flux problem associated with the maximum entropy deconvolution and described in appendix a , this method did not turn out to be a successful approach for bima song data .
instead , for bima song , we adopted a linear combination technique described in @xcite . in this method , we created a new `` dirty '' map by a linear combination of the interferometer map prior to deconvolving and the single dish map , where the single dish map had been weighted by an idealized form of the interferometer s primary gain of the mosaiced pointings .
we scaled the single dish map by the ratio of the area of the interferometer beam to the area of the single dish beam , and created a new `` dirty '' beam by the same linear combination of the interferometer synthesized beam and the single dish beam ( assumed to be a truncated gaussian ) .
we then deconvolved the new combined `` dirty '' map using a steer - dewdney - ito clean algorithm .
this method is easy to implement and automate , with relatively few parameters to specify .
heyer , m. h. , dame , t. m. , & thaddeus , p. 2000 , the interstellar medium in m31 and m33 .
proceedings 232 .
we - heraeus seminar , 22 - 25 may 2000 , bad honnef , germany . edited by elly m. berkhuijsen , rainer beck , and rene a. m. walterbos . shaker , aachen , 2000 , p. 29 - 36 , 29 lcccccccccccc ngc 0628 & 01 36 41.7 & @xmath5915 46 59 & 657 & 24 & 25 & sa(s)c & & 7.3 & 1 & 10.5 & 9.95 & 9 + ngc 0925 & 02 27 16.9 & @xmath5933 34 45 & 553 & 56 & 102 & sab(s)d & hii & 9.29 & 2 & 10.5 & 10.69 & 1 + ngc 1068 & 02 42 40.7 & @xmath6400 00 48 & 1136 & 33 & 13 & ( r)sa(rs)b & sy1.8 & 14.4 & 1 & 7.1 & 9.61 & 3 + ic 342 & 03 46 49.7 & @xmath5968 05 45 & 34 & 31 & 37 & sab(rs)cd & hii & 3.9 & 1 & 21.4 & 9.10 & 9 + ngc 2403 & 07 36 54.5 & @xmath5965 35 58 & 131 & 56 & 127 & sab(s)cd & hii & 4.2 & 1 & 21.9 & 8.93 & 3 + ngc 2841 & 09 22 02.7 & @xmath5950 58 36 & 638 & 64 & 147 & sa(r)b : & l2 & 14.1 & 3 & 8.1 & 10.09 & 3 + ngc 2903 & 09 32 10.1 & @xmath5921 30 02 & 556 & 61 & 17 & sab(rs)bc & hii & 6.3 & 1 & 12.6 & 9.68 & 7 + ngc 2976 & 09 47 17.1 & @xmath5967 54 51 & 3 & 63 & 143 & sac - pec & hii & 2.1 & 1 & 5.9 & 10.82 & 3 + ngc 3031 & 09 55 33.2 & @xmath5969 03 55 & @xmath6434 & 58 & 157 & sa(s)ab & sy1.5 & 3.63 & 4 & 26.9 & 7.89 & 12 + ngc 3184 & 10 18 16.7 & @xmath5941 25 27 & 592 & 21 & 135 & sab(rs)cd & hii & 8.7 & 1 & 7.4 & 10.36 & 9 + ngc 3344 & 10 43 31.1 & @xmath5924 55 21 & 586 & 24 & 0 & ( r)sab(r)bc & hii & 6.1 & 1 & 7.1 & 10.45 & 9 + ngc 3351 & 10 43 58.0 & @xmath5911 42 14 & 778 & 40 & 13 & sb(r)b & hii & 10.1 & 5 & 7.4 & 10.53 & 6 + ngc 3368 & 10 46 45.6 & @xmath5911 49 18 & 897 & 46 & 5 & sab(rs)ab & l2 & 11.2 & 6 & 7.6 & 10.11 & 8 + ngc 3521 & 11 05 49.3 & @xmath6400 02 02 & 805 & 58 & 164 & sab(rs)bc & hii / l2 : : & 7.2 & 1 & 11.0 & 9.83 & 3 + ngc 3627 & 11 20 15.1 & @xmath5912 59 22 & 727 & 63 & 176 & sab(s)b & t2/sy2 & 11.1 & 7 & 9.1 & 9.65 & 7 + ngc 3726 & 11 33 20.7 & @xmath5947 01 41 & 849 & 46 & 10 & sab(r)c & hii & 17.0 & 1 & 6.2 & 10.91 & 5 + ngc 3938 & 11 52 49.6 & @xmath5944 07 14 & 809 & 24 & 0 & sa(s)c & hii : : & 17.0 & 1 & 5.4 & 10.90 & 9 + ngc 3953 & 11 53 49.0 & @xmath5952 19 37 & 1054 & 60 & 13 & sb(r)bc & t2 & 17.0 & 1 & 6.9 & 10.84 & 9 + ngc 3992 & 11 57 36.1 & @xmath5953 22 29 & 1048 & 52 & 68 & sb(rs)bc & t2 : & 17.0 & 1 & 7.6 & 10.60 & 9 + ngc 4051 & 12 03 09.6 & @xmath5944 31 53 & 725 & 41 & 135 &
sab(rs)bc & sy1.2 & 17.0 & 1 & 5.2 & 10.83 & 5 + ngc 4258 & 12 18 57.5 & @xmath5947 18 14 & 448 & 65 & 176 & sab(s)bc & sy1.9 & 8.1 & 8 & 18.6 & 9.10 & + ngc 4303 & 12 21 54.9 & @xmath5904 28 25 & 1566 & 27 & 0 & sab(rs)bc & hii & 15.2 & 1 & 6.5 & 10.18 & 9 + ngc 4321 & 12 22 54.8 & @xmath5915 49 20 & 1571 & 32 & 154 & sab(s)bc & t2 & 16.1 & 9 & 7.4 & 10.05 & 12 + ngc 4414 & 12 26 27.2 & @xmath5931 13 24 & 716 & 55 & 159 & sa(rs)c ? & t2 : & 19.1 & 10 & 3.6 & 10.96 & 3 + ngc 4450 & 12 28 29.3 & @xmath5917 05 07 & 1954 & 41 & 175 & sa(s)ab & l1.9 & 16.8 & 1 & 5.2 & 10.90 & 12 + ngc 4490 & 12 30 36.9 & @xmath5941 38 23 & 565 & 60 & 125 & sb(s)d pec & hii & 7.8 & 1 & 6.3 & 10.22 & 1 + ngc 4535 & 12 34 20.3 & @xmath5908 11 54 & 1961 & 45 & 28 & sab(s)c & hii & 16.0 & 11 & 7.1 & 10.59 & 9 + ngc 4548 & 12 35 26.4 & @xmath5914 29 49 & 486 & 37 & 150 & sb(rs)b & l2 & 15.9 & 12 & 5.4 & 10.96 & 5 + ngc 4559 & 12 35 57.8 & @xmath5927 57 36 & 815 & 66 & 150 & sab(rs)cd & hii & 9.7 & 1 & 10.7 & 10.46 & + ngc 4569 & 12 36 49.80 & @xmath5913 09 46 & @xmath64235 & 62 & 23 & sab(rs)ab & t2 & 16.8 & 1 & 9.5 & 10.26 & 5 + ngc 4579 & 12 37 43.53 & @xmath5911 49 05 & 1519 & 37 & 95 & sab(rs)b & sy1.9/l1.9 & 16.8 & 1 & 5.9 & 10.48 & 9 + ngc 4699 & 12 49 02.33 & @xmath6408 39 55 & 1427 & 47 & 45 & sab(rs)b & & 25.7 & 1 & 3.8 & 10.41 & 3 + ngc 4725 & 12 50 26.60 & @xmath5925 30 06 & 1206 & 45 & 35 & sab(r)ab pec & sy2 : & 12.6 & 13 & 10.7 & 10.11 & 6 + ngc 4736 & 12 50 53.06 & @xmath5941 07 14 & 308 & 35 & 100 & ( r)sa(r)ab & l2 & 4.3 & 1 & 11.2 & 8.99 & 3 + ngc 4826 & 12 56 44.24 & @xmath5921 41 05 & 408 & 54 & 111 & ( r)sa(rs)ab & t2 & 4.1 & 1 & 10.0 & 9.36 & 6 + ngc 5005 & 13 10 56.23 & @xmath5937 03 33 & 946 & 61 & 65 & sab(rs)bc & l1.9 & 21.3 & 1 & 5.8 & 10.61 & 3 + ngc 5033 & 13 13 27.53 & @xmath5936 35 38 & 875 & 62 & 170 & sa(s)c & sy1.5 & 18.7 & 1 & 10.7 & 10.75 & 9 + ngc 5055 & 13 15 49.25 & @xmath5942 01 49 & 504 & 55 & 105 & sa(rs)bc & t2 & 7.2 & 1 & 12.6 & 9.31 & 3 + ngc 5194 & 13 29 52.35 & @xmath5947 11 54 & 463 & 15 & 0 & sa(s)bc pec & sy2 & 7.7 & 1 & 11.2 & 8.96 & 12 + ngc 5248 & 13 37 32.06 & @xmath5908 53 07 & 1153 & 43 & 110 & sab(rs)bc & hii & 22.7 & 1 & 6.2 & 10.97 & 12 + ngc 5247 & 13 38 02.62 &
@xmath6417 53 02 & 1357 & 29 & 20 & sa(s)bc & & 22.2 & 1 & 5.6 & 10.5 & 9 + ngc 5457 & 14 03 12.48 & @xmath5954 20 55 & 241 & 27 & 40 & sab(rs)cd & & 7.4 & 14 & 28.8 & 8.31 & 9 + ngc 6946 & 20 34 52.33 & @xmath5960 09 14 & 48 & 54 & 65 & sab(rs)cd & hii & 5.5 & 1 & 11.5 & 9.61 & 9 + ngc 7331 & 22 37 04.09 & @xmath5934 24 56 & 821 & 62 & 172 & sa(s)b & t2 & 15.1 & 15 & 10.5 & 10.35 & 3 + lccccccccc ngc 0628 & 13 & 7.2 @xmath25 5.3 & 250 @xmath25 190 & 8 & 5 & 51 & @xmath20.012 & 1.5 & 0.69 + ngc 0925 & 7 & 6.1 @xmath25 5.6 & 280 @xmath25 250 & @xmath644 & 10 & 65 & @xmath20.013 & 2.5 & + ngc 1068 & 13 & 8.9 @xmath25 5.6 & 620 @xmath25 390 & 16 & 10 & 85 & 0.041 @xmath17 0.007 & 6.0 & 0.56 + ic 342 & 7 & 5.6 @xmath25 5.1 & 110 @xmath25 97 & 34 & 10 & 129 & @xmath20.027 & 6.0 & 1.47 + ngc 2403 & 7 & 6.3 @xmath25 5.8 & 130 @xmath25 120 & 22 & 10 & 49 & @xmath20.009 & 2.0 & + ngc 2841 & 7 & 6.1 @xmath25 5.1 & 420 @xmath25 350 & 25 & 20 & 66 & @xmath20.013 & 2.5 & + ngc 2903 & 7 & 6.8 @xmath25 5.5 & 210 @xmath25 170 & 9 & 10 & 68 & @xmath20.016 & 5.0 & 0.38 + ngc 2976 & 1 & 7.8 @xmath25 6.9 & 79 @xmath25 70 & 59 & 10 & 41 & @xmath20.010 & 1.5 & + ngc 3031 & 7 & 5.9 @xmath25 5.1 & 100 @xmath25 90 & @xmath6414 & 20 & 58 & 0.161 @xmath17 0.006 & 3.0 & + ngc 3184 & 7 & 5.9 @xmath25 5.4 & 250 @xmath25 230 & 12 & 10 & 50 & @xmath20.010 & 1.5 & + ngc 3344 & 7 & 6.4 @xmath25 5.3 & 190 @xmath25 160 & 9 & 10 & 50 & @xmath20.011 & 1.3 & + ngc 3351 & 7 & 7.4 @xmath25 5.2 & 360 @xmath25 250 & 7 & 10 & 64 & @xmath20.014 & 3.0 & 0.33 + ngc 3368 & 7 & 7.7 @xmath25 5.7 & 420 @xmath25 310 & 5 & 10 & 69 & @xmath20.013 & 2.0 & + ngc 3521 & 7 & 8.8 @xmath25 5.7 & 310 @xmath25 200 & 1 & 10 & 78 & @xmath20.020 & 5.0 & 0.56 + ngc 3627 & 14 & 7.3 @xmath25 5.8 & 390 @xmath25 310 & 80 & 10 & 41 & @xmath20.009 & 5.0 & 0.43 + ngc 3726 & 1 & 6.0 @xmath25 5.0 & 500 @xmath25 420 & 11 & 10 & 32 & @xmath20.008 & 1.5 & + ngc 3938 & 7 & 5.9 @xmath25 5.4 & 480 @xmath25 440 & @xmath6433 & 10 & 59 & @xmath20.014 & 2.5 & 0.31 + ngc 3953 & 7 & 6.9 @xmath25 5.8 & 570 @xmath25 480 & 2 & 10 & 72 & @xmath20.015 & 2.0 & + ngc 3992 & 7 & 6.6 @xmath25 5.3 & 550 @xmath25 440 & 1 & 10 & 60 & @xmath20.013 & 1.5 & + ngc 4051 & 1 & 7.0 @xmath25 4.9 & 580 @xmath25 410 & @xmath642 & 10 & 23 & @xmath20.006 & 1.5 & + ngc 4258 & 7 & 6.1 @xmath25 5.4 & 240 @xmath25 210 & 4 & 10 & 60 & @xmath20.013 & 4.0 & 0.26 + ngc 4303 & 7 & 7.3 @xmath25 5.5 & 540 @xmath25 400 & @xmath644 & 10 & 47 & @xmath20.011 & 4.0 & 0.66 + ngc 4321 & 7 & 7.2 @xmath25 4.9 & 560 @xmath25 380 & 9 & 10 & 50 & @xmath20.014 & 3.0 & 0.94 + ngc 4414 & 1 & 6.4 @xmath25 5.0 & 590 @xmath25 460 & 6 & 10 & 29 & @xmath20.008 & 3.0 & 0.36 + ngc 4450 & 1 & 6.6 @xmath25 5.1 & 540 @xmath25 410 & 5 & 10 & 43 & @xmath20.013 & 1.5 & + ngc 4490 & 1 & 6.0 @xmath25 5.7 & 230 @xmath25 210 & 58 & 10 & 26 & @xmath20.007 & 0.8 & + ngc 4535 & 7 & 7.3 @xmath25 5.7 & 570 @xmath25 440 & 2 & 10 & 63 & @xmath20.015 & 2.0 & + ngc 4548 & 1 & 7.1 @xmath25 6.2 & 550 @xmath25 480 & @xmath6423 & 10 & 43 & @xmath20.011 & 2.0 & + ngc 4559 & 7 & 5.9 @xmath25 5.6 & 280 @xmath25 260 & @xmath6416 & 10 & 55 & @xmath20.013 & 2.0 & + ngc 4569 & 7 & 6.6 @xmath25 5.5 & 540 @xmath25 450 & @xmath642 & 10 & 54 & @xmath20.012 & 3.0 & 0.24 + ngc 4579 & 7 & 8.1 @xmath25 7.1 & 660 @xmath25 580 & @xmath6411 & 10 & 68 & 0.028 @xmath17 0.006 & 3.5 & + ngc 4699 & 1 & 10.3 @xmath25 6.8 & 1300 @xmath25 850 & 11 & 10 & 86 & @xmath20.019 & 3.0 & + ngc 4725 & 7 & 7.0 @xmath25 5.4 & 430 @xmath25 330 & 9 & 20 & 86 & @xmath20.017 & 3.0 & + ngc 4736 & 7 & 6.9 @xmath25 5.0 & 140 @xmath25 100 & 63 & 10 & 64 & @xmath20.017 & 3.0 & 0.38 + ngc 4826 & 7 & 7.5 @xmath25 5.2 & 150 @xmath25 100 & 3 & 10 & 82 & @xmath20.016 & 4.0 & 0.39 + ngc 5005 & 1 & 6.2 @xmath25 6.0 & 640 @xmath25 620 & 23 & 10 & 32 & @xmath20.009 & 1.5 & 0.16 + ngc 5033 & 7 & 6.1 @xmath25 5.4 & 550 @xmath25 490 & 86 & 20 & 73 & @xmath20.015 & 4.0 & 0.44 + ngc 5055 & 7 & 5.8 @xmath25 5.5 & 200 @xmath25 190 & 18 & 10 & 55 & @xmath20.011 & 5.0 & 0.45 + ngc 5194 & 26 & 5.8 @xmath25 5.1 & 220 @xmath25 190 & 4 & 10 & 61 & @xmath20.012 & 5.0 & 0.55 + ngc 5248 & 1 & 6.9 @xmath25 5.8 & 760 @xmath25 630 & 1 & 10 & 38 & @xmath20.011 & 1.0 & 0.35 + ngc 5247 & 1 & 12.2 @xmath25 4.7 & 1300 @xmath25 510 & 2 & 10 & 59 & @xmath20.016 & 3.0 & 0.32 + ngc 5457 & 7 & 5.7 @xmath25 5.4 & 210 @xmath25 190 & 59 & 10 & 45 & @xmath20.010 & 3.0 & 0.59 + ngc 6946 & 7 & 6.0 @xmath25 4.9 & 160 @xmath25 130 & 14 & 10 & 61 & @xmath20.013 & 6.0 & 0.63 + ngc 7331 & 7 & 6.1 @xmath25 4.9 & 440 @xmath25 360 & 5 & 10 & 50 & @xmath20.010 & 3.0 & 0.51 + ccccccl ngc 0925 & 0 & 0 & 0.005 & 1.10 @xmath17 0.15 & 495,640 & ( 0.07 @xmath17 0.38 ) + & 22 & 38 & 0.009 & @xmath20.87 & 453,653 & + & 44 & 0 & 0.013 & 0.76 @xmath17 0.23 & 480,540 & + & 22 & -38 & 0.005 & 1.25 @xmath17 0.14 & 430,550 & ( 0.06 @xmath17 0.30 ) + & -22 & -38 & 0.006 & @xmath20.57 & 453,653 & + & -44 & 0 & 0.012 & @xmath21.1 & 453,653 & + & -22 & 38 & 0.007 & @xmath20.69 & 453,653 & + ngc 2403 & 0 & 0 & 0.018 & 2.16 @xmath17 0.29 & 100,200 & 0.63 @xmath17 0.16 + & 22 & 38 & 0.023 & @xmath21.5 & 31,231 & + & 44 & 0 & 0.019 & 1.71 @xmath17 0.25 & 145,210 & ( 0.19 @xmath17 0.15 ) + & 22 & -38 & 0.017 & 1.14 @xmath17 0.27 & 100,200 & ( 0.55 @xmath17 0.29 ) + & -22 & -38 & 0.015 & 1.46 @xmath17 0.25 & 90,200 & 0.56 @xmath17 0.23 + & -44 & 0 & 0.020 & @xmath21.4 & 31,231 & + & -22 & 38 & 0.020 & 1.95 @xmath17 0.39 & 50,200 & ( 0.24 @xmath17 0.18 ) + ngc 2841 & 0 & 0 & 0.008 & @xmath21.4 & 290,930 & + & 22 & 38 & 0.010 & @xmath21.7 & 290,930 & + & 44 & 0 & 0.010 & @xmath21.7 & 290,930 & + & 22 & -38 & 0.011 & @xmath21.9 & 290,930 & + & -22 & -38 & 0.009 & @xmath21.7 & 290,930 & + & -44 & 0 & 0.011 & @xmath21.8 & 290,930 & + & -22 & 38 & 0.010 & @xmath21.8 & 290,930 & + ngc 2976 & 0 & 0 & 0.007 & 1.29 @xmath17 0.15 & -35,45 & ( 0.49 @xmath17 0.42 ) + ngc 3031 & 0 & 0 & 0.013 & @xmath21.3 & -134,66 & + & 22 & 38 & 0.021 & @xmath22.0 & -134,66 & + & 44 & 0 & 0.014 & @xmath21.3 & -134,66 & + & 22 & -38 & 0.014 & @xmath21.4 & -134,66 & + & -22 & -38 & 0.020 & @xmath21.9 & -134,66 & + & -44 & 0 & 0.015 & @xmath21.5 & -134,66 & + & -22 & 38 & 0.011 & 1.25 @xmath17 0.28 & 70,180 & + ngc 3184 & 0 & 0 & 0.004 & 3.96 @xmath17 0.09 & 550,660 & 0.54 @xmath17 0.07 + & 22 & 38 & 0.008 & 3.10 @xmath17 0.17 & 535,620 & 0.28 @xmath17 0.08 + & 44 & 0 & 0.007 & 2.44 @xmath17 0.14 & 560,635 & 0.30 @xmath17 0.10 + & 22 & -38 & 0.008 & 2.79 @xmath17 0.17 & 550,645 & ( 0.17 @xmath17 0.09 ) + & -22 & -38 & 0.007 & 2.43 @xmath17 0.14 & 570,650 & ( 0.16 @xmath17 0.10 ) + & -44 & 0 & 0.009 & 2.72 @xmath17 0.22 & 510,630 & ( 0.05 @xmath17 0.10 ) + & -22 & 38 & 0.010 & 2.35 @xmath17 0.24 & 500,600 & ( 0.07 @xmath17 0.11 ) + ngc 3344 & 0 & 0 & 0.015 & 3.28 @xmath17 0.45 & 475,650 & ( 0.01 @xmath17 0.11 ) + & 22 & 38 & 0.018 & 2.85 @xmath17 0.53 & 490,660 & + & 44 & 0 & 0.016 & @xmath21.1 & 500,600 & + & 22 & -38 & 0.014 & 1.90 @xmath17 0.22 & 500,545 & ( 0.02 @xmath17 0.10 ) + & -22 & -38 & 0.019 & 3.34 @xmath17 0.46 & 500,610 & ( 0.06 @xmath17 0.08 ) + & -44 & 0 & 0.010 & 2.25 @xmath17 0.30 & 475,650 & ( 0.13 @xmath17 0.15 ) + & -22 & 38 & 0.013 & 2.28 @xmath17 0.24 & 600,660 & 0.19 @xmath17 0.09 + ngc 3368 & 0 & 0 & 0.006 & 12.80 @xmath17 0.32 & 670,1200 & 0.94 @xmath17 0.06 + & 22 & 38 & 0.009 & @xmath20.90 & 797,997 & + & 44 & 0 & 0.007 & 3.04 @xmath17 0.29 & 800,1100 & 0.36 @xmath17 0.17 + & 22 & -38 & 0.007 & 3.32 @xmath17 0.28 & 800,1100 & 0.73 @xmath17 0.16 + & -22 & -38 & 0.007 & 4.23 @xmath17 0.30 & 750,1100 & 0.28 @xmath17 0.13 + & -44 & 0 & 0.008 & 3.54 @xmath17 0.32 & 680,1000 & ( 0.27 @xmath17 0.15 ) + & -22 & 38 & 0.007 & 4.58 @xmath17 0.27 & 680,1000 & 0.26 @xmath17 0.11 + ngc 3726 & 0 & 0 & 0.010 & 2.88 @xmath17 0.33 & 760,960 & ( 0.43 @xmath17 0.25 ) + ngc 3953 & 0 & 0 & 0.010 & 4.57 @xmath17 0.39 & 860,1170 & ( 0.10 @xmath17 0.16 ) + & 22 & 38 & 0.009 & 4.90 @xmath17 0.32 & 1030,1250 & ( 0.06 @xmath17 0.13 ) + & 44 & 0 & 0.011 & 3.09 @xmath17 0.41 & 950,1200 & ( 0.06 @xmath17 0.22 ) + & 22 & -38 & 0.009 & 3.99 @xmath17 0.30 & 870,1100 & ( 0.05 @xmath17 0.16 ) + & -22 & -38 & 0.009 & 4.71 @xmath17 0.25 & 850,1000 & ( 0.13 @xmath17 0.11 ) + & -44 & 0 & 0.010 & 5.21 @xmath17 0.43 & 840,1170 & ( 0.05 @xmath17 0.15 ) + & -22 & 38 & 0.009 & 5.62 @xmath17 0.33 & 1000,1240 & ( 0.06 @xmath17 0.12 ) + ngc 3992 & 0 & 0 & 0.008 & @xmath21.0 & 800,1200 & + & 22 & 38 & 0.008 & @xmath21.1 & 800,1200 & + & 44 & 0 & 0.007 & 1.54 @xmath17 0.21 & 800,950 & ( -0.22 @xmath17 0.22 ) + & 22 & -38 & 0.009 & @xmath21.3 & 800,1200 & + & -22 & -38 & 0.008 & @xmath21.1 & 800,1200 & + & -44 & 0 & 0.010 & 1.54 @xmath17 0.45 & 800,1200 & + & -22 & 38 & 0.009 & 2.06 @xmath17 0.43 & 800,1200 & + ngc 4051 & 0 & 0 & 0.012 & 8.68 @xmath17 0.42 & 615,845 & 0.34 @xmath17 0.06 + ngc 4450 & 0 & 0 & 0.011 & 4.17 @xmath17 0.46 & 1800,2150 & ( 0.18 @xmath17 0.32 ) + ngc 4490 & 0 & 0 & 0.004 & 1.63 @xmath17 0.12 & 550,720 & ( 0.27 @xmath17 0.34 ) + ngc 4535 & 0 & 0 & 0.011 & 8.26 @xmath17 0.45 & 1810,2120 & 0.65 @xmath17 0.07 + & 22 & 38 & 0.020 & 3.99 @xmath17 0.56 & 1810,1960 & 0.29 @xmath17 0.10 + & 44 & 0 & 0.019 & 5.12 @xmath17 0.64 & 1860,2090 & 0.31 @xmath17 0.09 + & 22 & -38 & 0.019 & 3.83 @xmath17 0.49 & 1950,2080 & 0.21 @xmath17 0.09 + & -22 & -38 & 0.018 & 3.86 @xmath17 0.49 & 1950,2095 & 0.31 @xmath17 0.10 + & -44 & 0 & 0.024 & 2.09 @xmath17 0.61 & 1915,2040 & 0.41 @xmath17 0.20 + & -22 & 38 & 0.019 & 3.17 @xmath17 0.50 & 1810,1945 & 0.31 @xmath17 0.12 + ngc 4548 & 0 & 0 & 0.005 & 2.77 @xmath17 0.19 & 305,625 & ( 0.31 @xmath17 0.42 ) + ngc 4559 & 0 & 0 & 0.008 & 3.09 @xmath17 0.26 & 715,910 & 0.31 @xmath17 0.12 + & 22 & 38 & 0.012 & @xmath21.1 & 715,910 & + & 44 & 0 & 0.011 & @xmath21.0 & 715,910 & + & 22 & -38 & 0.011 & 1.99 @xmath17 0.28 & 690,825 & ( 0.26 @xmath17 0.16 ) + & -22 & -38 & 0.008 & @xmath20.78 & 715,910 & + & -44 & 0 & 0.011 & 1.97 @xmath17 0.30 & 750,900 & ( -0.04 @xmath17 0.16 ) + & -22 & 38 & 0.009 & 1.80 @xmath17 0.23 & 790,910 & ( 0.22 @xmath17 0.17 ) + ngc 4579 & 0 & 0 & 0.006 & 6.33 @xmath17 0.26 & 1310,1705 & 0.32 @xmath17 0.08 + & 22 & 38 & 0.014 & 3.73 @xmath17 0.62 & 1310,1705 & ( 0.17 @xmath17 0.14 ) + & 44 & 0 & 0.013 & 2.56 @xmath17 0.36 & 1565,1705 & ( 0.13@xmath17 0.12 ) + & 22 & -38 & 0.014 & 2.09 @xmath17 0.36 & 1525,1645 & + & -22 & -38 & 0.015 & 5.62 @xmath17 0.49 & 1320,1540 & ( 0.13 @xmath17 0.07 ) + & -44 & 0 & 0.019 & 3.89 @xmath17 0.66 & 1320,1540 & + & -22 & 38 & 0.020 & 3.71 @xmath17 0.67 & 1320,1540 & + ngc 4699 & 0 & 0 & 0.011 & @xmath21.5 & 1250,1620 & + ngc 4725 & 0 & 0 & 0.007 & 2.80 @xmath17 0.32 & 1030,1400 & ( 0.60 @xmath17 0.31 ) + & 22 & 38 & 0.010 & @xmath21.3 & 1030,1400 & + & 44 & 0 & 0.018 & @xmath22.3 & 1030,1400 & + & 22 & -38 & 0.015 & @xmath21.9 & 1030,1400 & + & -22 & -38 & 0.015 & @xmath21.9 & 1030,1400 & + & -44 & 0 & 0.012 & @xmath21.5 & 1030,1400 & + & -22 & 38 & & & & + lccccc ngc 0628 & 6 & 0.57 & 1514 @xmath17 63 & 1.01 @xmath17 0.04 & 0.63 @xmath17 0.03 + ngc 1068 & 6 & 0.85 & 4102 @xmath17 217 & 10.69 @xmath17 0.57 & 6.67 @xmath17 0.35 + ic 342 & 6 & 0.28 & 9847 @xmath17 418 & 1.88 @xmath17 0.08 & 1.17 @xmath17 0.05 + ngc 2903 & 6 & 0.48 & 3254 @xmath17 282 & 1.62 @xmath17 0.14 & 1.01 @xmath17 0.09 + ngc 3351 & 6 & 0.81 & 1513 @xmath17 221 & 1.94 @xmath17 0.28 & 1.21 @xmath17 0.18 + ngc 3521 & 6 & 0.55 & 4800 @xmath17 353 & 3.13 @xmath17 0.23 & 1.95 @xmath17 0.14 + ngc 3627 & 6 & 0.66 & 4259 @xmath17 427 & 6.59 @xmath17 0.66 & 4.12 @xmath17 0.41 + ngc 3938 & 5 & 0.93 & 923 @xmath17 73 & 3.35 @xmath17 0.27 & 2.09 @xmath17 0.17 + ngc 4258 & 6 & 0.32 & 2686 @xmath17 255 & 2.21 @xmath17 0.21 & 1.38 @xmath17 0.13 + ngc 4303 & 6 & 0.92 & 2427 @xmath17 145 & 7.05 @xmath17 0.42 & 4.40 @xmath17 0.26 + ngc 4321 & 6 & 0.81 & 2972 @xmath17 319 & 9.68 @xmath17 1.04 & 6.04 @xmath17 0.65 + ngc 4414 & 6 & 1.67 & 2453 @xmath17 282 & 11.25 @xmath17 1.29 & 7.02 @xmath17 0.81 + ngc 4569 & 6 & 0.63 & 1096 @xmath17 137 & 3.89 @xmath17 0.49 & 2.43 @xmath17 0.30 + ngc 4736 & 6 & 0.54 & 2641 @xmath17 155 & 0.39 @xmath17 0.05 & 0.24 @xmath17 0.03 + ngc 4826 & 6 & 0.60 & 1845 @xmath17 217 & 0.61 @xmath17 0.04 & 0.38 @xmath17 0.02 + ngc 5005 & 6 & 1.03 & 1278 @xmath17 484 & 7.29 @xmath17 2.76 & 4.55 @xmath17 1.72 + ngc 5033 & 5 & 0.47 & 2469 @xmath17 258 & 10.85 @xmath17 1.13 & 6.77 @xmath17 0.71 + ngc 5055 & 6 & 0.48 & 3812 @xmath17 276 & 2.48 @xmath17 0.18 & 1.55 @xmath17 0.11 + ngc 5194 & 12 & 1.07 & 10097 @xmath17 267 & 7.52 @xmath17 0.20 & 4.70 @xmath17 0.12 + ngc 5248 & 6 & 0.97 & 1829 @xmath17 195 & 11.84 @xmath17 1.26 & 7.39 @xmath17 0.79 + ngc 5247 & 5 & 0.89 & 1102 @xmath17 91 & 6.82 @xmath17 0.56 & 4.26 @xmath17 0.35 + ngc 5457 & 7 & 0.24 & 3479 @xmath17 265 & 2.39 @xmath17 0.18 & 1.49 @xmath17 0.11 + ngc 6946 & 6 & 0.52 & 9273 @xmath17 241 & 3.52 @xmath17 0.09 & 2.20 @xmath17 0.06 + ngc 7331 & 6 & 0.57 & 2762 @xmath17 367 & 7.91
@xmath17 1.05 & 4.94 @xmath17 0.66 + lccc
ngc 0628 & 104 & 20 & 0.20 + ngc 0925 & 31 & 0 & 0.00 + ngc 1068 & 947 & 466 & 0.49 + ic 342 & 1267 & 1151 & 0.91 + ngc 2403 & 30 & 25 & 0.81 + ngc 2841 & 36 & 0 & 0.00 + ngc 2903 & 573 & 573 & 1.00 + ngc 2976 & 15 & 15 & 1.00 + ngc 3031 & 24 & 0 & 0.00 + ngc 3184 & 153 & 153 & 1.00 + ngc 3344 & 24 & 0 & 0.00 + ngc 3351 & 426 & 406 & 0.95 + ngc 3368 & 815 & 645 & 0.79 + ngc 3521 & 103 & 82 & 0.80 + ngc 3627 & 539 & 539 & 1.00 + ngc 3726 & 94 & 94 & 1.00 + ngc 3938 & 72 & 60 & 0.82 + ngc 3953 & 23 & 0 & 0.00 + ngc 3992 & 18 & 5 & 0.32 + ngc 4051 & 315 & 315 & 1.00 + ngc 4258 & 473 & 202 & 0.43 + ngc 4303 & 432 & 432 & 1.00 + ngc 4321 & 442 & 442 & 1.00 + ngc
4414 & 130 & 59 & 0.46 + ngc 4450 & 15 & 10 & 0.71 + ngc 4490 & 18 & 0 & 0.00 + ngc 4535 & 429 & 429 & 1.00 + ngc 4548 & 18 & 18 & 1.00 + ngc 4559 & 16 & 13 & 0.80 + ngc 4569 & 510 & 510 & 1.00 + ngc 4579 & 99 & 89 & 0.90 + ngc 4699 & 15 & 5 & 0.34 + ngc 4725 & 69 & 21 & 0.32 + ngc 4736 & 327 & 327 & 1.00 + ngc 4826 & 672 & 672 & 1.00 + ngc 5005 & 724 & 724 & 1.00 + ngc 5033 & 188 & 188 & 1.00 + ngc 5055 & 335 & 328 & 0.98 + ngc 5194 & 787 & 216 & 0.28 + ngc 5248 & 344 & 323 & 0.94 + ngc 5247 & 108 & 108 & 1.00 + ngc 5457 & 348 & 348 & 1.00 + ngc 6946 & 1854 & 1854 & 1.00 + ngc 7331 & 164 & 31 & 0.19 + 100 box ( _ top _ ) and peak flux density recovered in the map ( _ bottom _ ) for bima simulations of spiral models as a function of the log of the signal - to - noise ratio .
simulations are shown for spiral models with arms widths of 10 , 20 , 30 , and 40 .
the plots reflect the fact that it is increasingly difficult for the deconvolution to reconstruct large - scale emission as the signal - to - noise ratio decreases .
figure and caption are adapted from @xcite .
, height=480 ] | the bima survey of nearby galaxies is a systematic imaging study of the 3 mm co j = 10 molecular emission within the centers and disks of 44 nearby spiral galaxies .
the typical spatial resolution of the survey is 6 or 360 pc at the average distance ( 12 mpc ) of the sample .
the velocity resolution of the co observations is 4 km s@xmath0 , though most maps are smoothed to 10 km s@xmath0 resolution . for 33 galaxies
, multi - field observations ensured that a region @xmath1 190 ( @xmath2d@xmath3 = 10 kpc ) in diameter was imaged .
for the remaining 11 galaxies , which had smaller optical diameters and were on average farther away , single - pointing observations imaged a 100-diameter ( @xmath2d@xmath3 = 11 kpc ) region .
the sample was not chosen based on co or infrared brightness ; instead , all spirals were included that met the selection criteria of @xmath4 2000 km s@xmath0 , @xmath5 -20 , @xmath6 70 , @xmath7 70 , and @xmath8 11.0 .
the detection rate was 41/44 sources or 93% ; of the three nondetections , one ( m 81 ) is known to have co emission at locations outside the survey field of view .
fully - sampled single - dish co data were incorporated into the maps for 24 galaxies ; these single - dish data comprise the most extensive collection of fully - sampled , two - dimensional single - dish co maps of external galaxies to date .
we also tabulate direct measurements of the global co flux densities for these 24 sources . for the remaining 20 sources
, we collected sensitive single - dish spectra in order to evaluate the large - scale flux recovery .
we demonstrate that the measured ratios of flux density recovered are a function of the signal - to - noise of the interferometric data .
we examine the degree of central peakedness of the molecular surface density distributions and show that the distributions exhibit their brightest co emission within the central 6 in only 20/44 or 45% of the sample .
we show that all three local group spiral galaxies have co morphologies that are represented in song , though the milky way co luminosity is somewhat below the song average , and m31 and m33 are well below average .
this survey provides a unique public database of integrated intensity maps , channel maps , spectra , and velocity fields of molecular emission in nearby galaxies .
it also lays the groundwork for extragalactic surveys by more powerful future millimeter - wavelength interferometers like carma and alma . |
the deviations from bjorken scaling seen in deep - inelastic structure functions are usually interpreted as logarithmic scaling violations , as predicted by perturbative qcd .
there is mounting evidence @xcite , however , that part of the deviations are due to power corrections induced by higher - twist effects .
this is good news : higher - twist operators probe the non - perturbative features of hadronic bound states beyond the parton model , and they may provide valuable information on the interface between perturbative and non - perturbative physics , at least in those cases where the operator product expansion ( ope ) allows a clean separation between short- and long - distance phenomena .
the ope expresses the moments of the structure function as a series of forward hadron matrix elements of local operators with coefficients decreasing as powers of @xmath6 . in general the expansion
takes the form @xmath7 where @xmath8 and @xmath9 are the operator matrix elements and wilson coefficients , respectively , renormalized at a scale @xmath10 , with the superscript ( 2 or 4 ) denoting the twist . under ideal circumstances
the wilson coefficients can be calculated perturbatively , which leaves only the matrix elements to be computed on the lattice .
the leading - twist contribution can be written as @xmath11 where @xmath12 is the energy fraction carried by the quark of flavor @xmath13 , and @xmath14 is the charge of the quark .
the higher - twist contributions usually have no parton model interpretation . in this paper
we will consider the @xmath15 pion structure function @xcite @xmath16 this belongs to a flavor 27-plet , is purely higher twist , and receives contributions from 4-fermi operators only .
we restrict ourselves to the lowest non - trivial moment @xmath17 .
thus we will have to compute @xmath18 .
the problem splits into two separate tasks .
the first task obviously is to compute the pion matrix elements of the appropriate lattice 4-fermi operators .
the second task is to renormalize these operators at some finite scale @xmath10 .
the paper is organized as follows . in sect . 2 we identify the ( renormalized ) operator whose matrix elements we have to compute according to the ope . in sect .
3 we perform the renormalization of the 4-fermi operators .
this is done to 1-loop order in perturbation theory . in sect .
4 we present the results of the lattice calculation . finally , in sect . 5
we collect our results and conclude .
the leading twist-2 matrix element @xmath19 that enters @xmath17 is given by @xmath20 , \ ] ] @xmath21 where @xmath22 means symmetrization , and @xmath23 .
the fermion fields are 2-component vectors in flavor space , corresponding to @xmath24 and @xmath25 quarks , and the charge matrix is @xmath26 the states are normalized as @xmath27 the wilson coefficient has the form @xmath28 this contribution is the same for charged and neutral pions , and so vanishes when considering the structure function @xmath29 . the twist-4 matrix element @xmath30 receives , in general , contributions from a large variety of operators . here
we shall only be interested in 4-fermi operators , because these are the only operators that contribute to @xmath29 .
following @xcite we have @xmath31 , \ ] ] @xmath32 the corresponding wilson coefficient is given by @xcite @xmath33 the operator ( [ eq : htop ] ) is understood to be the renormalized , continuum operator .
in the following we will be working in euclidean space - time . the 4-fermi operators that we need to consider on the lattice are @xmath34 a^c_{\mu\nu } & = { \bar{\psi}}g\gamma_{\mu } \gamma_5 t^a { \psi}\ , { \bar{\psi}}g\gamma_{\nu } \gamma_5 t^a { \psi}-\mbox{trace } , \\[0.5ex ] t^c_{\mu\nu } & = { \bar{\psi}}g\sigma_{\mu\rho } t^a { \psi}\ , { \bar{\psi}}g\sigma_{\nu\rho } t^a { \psi}-\mbox{trace},\\[0.5ex ] v_{\mu\nu } & = { \bar{\psi}}g\gamma_{\mu } { \psi}\ , { \bar{\psi}}g\gamma_{\nu } { \psi}-\mbox{trace},\\[0.5ex ] a_{\mu\nu } & = { \bar{\psi}}g\gamma_{\mu } \gamma_5 { \psi}\ , { \bar{\psi}}g\gamma_{\nu } \gamma_5 { \psi}-\mbox{trace},\\[0.5ex ] t_{\mu\nu } & = { \bar{\psi}}g\sigma_{\mu\rho } { \psi}\ , { \bar{\psi}}g\sigma_{\nu\rho } { \psi}-\mbox{trace}.\\ \end{split}\ ] ] summation over repeated indices and symmetrization in @xmath10 , @xmath35 is understood . in our conventions
@xmath36 $ ] .
the @xmath15 parts of the operators are related by fierz transformations : @xmath37 the reason for considering all six operators is that they will mix under renormalization . in principle , the operators ( [ eq:6ops ] ) could also mix with gauge variant , brst invariant operators .
but there are no such 4-fermi operators with dimension six , and 2-fermi operators do not contribute .
a 1-loop calculation in the continuum and on the lattice gives for the respective operator matrix elements @xmath38 where @xmath39 are quark states in some covariant gauge and @xmath40 is the bare coupling constant .
the continuum matrix elements are understood to be renormalized in the @xmath2 scheme at the scale @xmath10 , while the lattice matrix elements are unrenormalized .
note that the tree - level matrix elements @xmath41 are the same in both cases .
the lattice and continuum matrix elements are then connected by @xmath42 & \times \,\langle p , p ' |o_j^{\rm lat}(a)| p , p ' \rangle .\\
\end{split}\ ] ] let us write @xmath43 while @xmath44 and @xmath45 depend on the state , @xmath46 is independent of the state and depends only on @xmath47 .
the renormalization constants , that take us from bare lattice to renormalized continuum numbers , are given by @xmath48 we have found it convenient to take @xmath49 .
the algebraic manipulations have been done with the help of form , using the t hooft - veltman prescription for dealing with the @xmath3 matrices .
this is the only prescription that has proven to give consistent results . integrating the t hooft - veltman prescription into form
was a non - trivial task .
the algebraic workload increased by about an order of magnitude . to check the results ,
many of the symbolic calculations were repeated by hand .
we follow the method of @xcite for regularizing the infrared divergences . the diagrams that we have calculated
are shown in fig .
[ fig : diagrams ] .
we have not made use of fierz transformations to reduce the number of diagrams , as we could do the numerical integration fast and with high precision , and also to avoid possible problems with @xmath25-dimensional fierz transformations .
[ fig : diagrams ] the 1-loop result for the renormalized 4-fermi operators in the @xmath2 scheme is @xmath50 & + & c_4 \big(\frac{n_c^2 - 1}{n_c^2 } a_{\mu\nu } + \frac{n_c^2 - 4}{n_c } a^c_{\mu\nu } \big ) - n_c c_3 t^c_{\mu\nu } \big ] , \nonumber \\[0.95ex ] v_{\mu\nu}(\mu ) & = & v_{\mu\nu } -\frac{g_0 ^ 2}{16\pi^2 } \big [ \frac{n_c^2 - 1}{2n_c } \big(c_1 + 2s\big ) v_{\mu\nu } + 4 c_4 a^c_{\mu\nu } \big ] , \\[0.95ex ] a^c_{\mu\nu}(\mu ) & = & a^c_{\mu\nu } -\frac{g_0 ^ 2}{16\pi^2 } \big [ \big ( -\frac{1}{2n_c } c^{(5)}_1 + \frac{n_c^2 - 1}{2n_c } 2s - n_c c_2 \big ) a^c_{\mu\nu } \nonumber \\[0.95ex ] & + & c_4 \big(\frac{n_c^2 - 1}{n_c^2 } v_{\mu\nu } + \frac{n_c^2 - 4}{n_c } v^c_{\mu\nu } \big ) - c_3 \big(\frac{n_c^2 - 1}{n_c^2 } t_{\mu\nu } + \frac{n_c^2 - 4}{n_c } t^c_{\mu\nu } \big)\big ] , \nonumber \\[0.95ex ] a_{\mu\nu}(\mu ) & = & a_{\mu\nu } -\frac{g_0 ^ 2}{16\pi^2 } \big [ \frac{n_c^2 - 1}{2n_c } \big(c^{(5)}_1 + 2s\big ) a_{\mu\nu } + 4 c_4 v^c_{\mu\nu } - 4 c_3 t^c_{\mu\nu } \big ] , \nonumber \end{aligned}\ ] ] where @xmath51 is the difference between the ( leg ) self - energy on the lattice , including the tadpole diagram , and in the continuum , and @xmath52 c^{(5)}_1 & = -2\log a^2\mu^2 + 5.887758 + ( 1-\alpha)\ , ( 2\log a^2\mu^2 -7.584020 ) , \\[0.5ex ]
c_2 & = \frac{1}{2}\log a^2\mu^2 -4.260157 + ( 1-\alpha)\ , ( -\log a^2\mu^2 + 3.792010 ) , \\[0.5ex ] c_3 & = 1.205379 , \\[0.5ex ] c_4 & = -\frac{1}{2}\log a^2\mu^2 + 0.094480 .
\\ \end{split}\ ] ] here @xmath53 is the covariant gauge parameter with @xmath54 corresponding to feynman gauge and @xmath55 to landau gauge .
the renormalization constants are gauge invariant .
furthermore , they are independent of @xmath10 and @xmath35 , and hence of the particular representation the operators reduce to @xcite . the anomalous dimensions of @xmath56 and @xmath57 , which are eigenvalues of the mixing matrix , agree with the result found in @xcite . later on
we will be interested in ( [ eq:1-loop ] ) for @xmath58 and @xmath59 .
for this case we have @xmath60 v_{\mu\nu}(\mu\!=\!1/a ) & = v_{\mu\nu } -g_0 ^ 2 \big ( 0.348170 \,\ , v_{\mu\nu } + 0.002393 \,\ , a^c_{\mu\nu } \big ) , \\[0.95ex ] a^c_{\mu\nu}(\mu\!=\!1/a ) & = a^c_{\mu\nu } -g_0 ^ 2 \big ( 0.291756 \,\ , a^c_{\mu\nu } + 0.000532 \,\ , v_{\mu\nu } \\ & + 0.000997 \,\ , v^c_{\mu\nu } - 0.006785 \,\ , t_{\mu\nu } - 0.012722 \,\ , t^c_{\mu\nu } \big ) , \\[0.95ex ] a_{\mu\nu}(\mu\!=\!1/a ) & = a_{\mu\nu } -g_0 ^ 2 \big ( 0.266750 \,\ , a_{\mu\nu } + 0.002393 \,\ , v^c_{\mu\nu } \\ & - 0.030533 \,\ , t^c_{\mu\nu } \big ) .\\ \end{split}\ ] ] our results will be of use to other applications of 4-fermi operators as well . for a similar calculation in the context of weak matrix elements , involving a different set of operators ,
see @xcite .
to obtain the pion matrix elements of the operators ( [ eq:6ops ] ) , we need to compute the 3- and 2-point functions @xmath61 where @xmath62 is the spatial volume of the lattice .
the pion field is @xmath63 where @xmath64 is the flavor matrix . for the different pion states the flavor matrix takes
the form @xmath65 the superscript @xmath66 in ( [ eq : pistat ] ) denotes jacobi - smeared quark fields : @xmath67 with @xmath68 being given in @xcite , and @xmath69 denoting the link variables .
the 4-fermi operators in ( [ eq:6ops ] ) can be expressed as @xmath70 where @xmath71 and @xmath72 are products of @xmath73 and @xmath74 matrices .
a transfer matrix calculation gives for the time dependence of the 2-point function @xmath75 \\ & = { \langle}0 |{\hat{\pi}}^{f}|\pi{\rangle}{\langle}\pi|{\hat{\pi}}^{f^\dagger}|0{\rangle}e^{-m_\pi t } + { \langle}0 |{\hat{\pi}}^{f^\dagger}|\pi{\rangle}{\langle}\pi|{\hat{\pi}}^{f}|0{\rangle}e^{-m_\pi ( t - t)},\\ \end{split}\ ] ] where @xmath76 is the pion ground state , @xmath77 is the time extent of the lattice , and @xmath78 is the transfer matrix . all contributions from excited states have been suppressed .
the pion states are normalized according to @xmath79 the same calculation gives for the 3-point function @xmath80 , & t\ge t\ge\tau\ge 0 , \\
{ \operatorname{tr}}[{\hat{s}}^{t-\tau}o{\hat{s}}^{\tau - t}{\hat{\pi}}^f{\hat{s}}^t{\hat{\pi}}^{f^\dagger } ] , & t\ge\tau\ge t\ge 0 \end{cases } \\[1.5ex ] & = \begin{cases } { \langle}0 |{\hat{\pi}}^f|\pi{\rangle}{\langle}\pi|o|\pi{\rangle}{\langle}\pi|{\hat{\pi}}^{f^\dagger}|0{\rangle}e^{-m_\pi t } , & t\ge t\ge\tau\ge 0 , \\ { \langle}0 |{\hat{\pi}}^{f^\dagger}|\pi{\rangle}{\langle}\pi|o|\pi{\rangle}{\langle}\pi|{\hat{\pi}}^f|0{\rangle}e^{-m_\pi ( t - t ) } , & t\ge\tau\ge t\ge 0 . \end{cases}\\ \end{split}\ ] ] thus for the ratio of 3- and 2-point functions we may expect to see a plateau in @xmath81 at @xmath82 , and a plateau at @xmath83 : @xmath84 with @xmath85 the sum is independent of @xmath86 , and finally we obtain @xmath87 we will now re - write the 3-point function ( [ eq : corr ] ) in terms of quark propagators .
two different kinds of propagators are emerging , depending on whether we start from a smeared quark field and end on a local one , or vice versa .
the local - smeared and smeared - local propagators , @xmath88 and @xmath89 , are given by @xmath90 with @xmath91 ( we have suppressed the dependence on @xmath69 . ) for @xmath15 , and after having summed over flavor indices , we obtain @xmath92\,{\rangle}_u \nonumber \\ & + & { \langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(y , z){\gamma_5}{}^{sl}\!{\delta}(z , y)\gamma { } ^{ls}\!{\delta}(y , x){\gamma_5}{}^{sl}\!{\delta}(x , y)\gamma']\,{\rangle}_u \label{eq : gen4f2 } \\ & -&{\langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(y , z){\gamma_5}{}^{sl}\!{\delta}(z , y)\gamma ' ] { \operatorname{tr}}[{}^{ls}\!{\delta}(y , x){\gamma_5}{}^{sl}\!{\delta}(x , y)\gamma]\,{\rangle}_u \nonumber \\ & -&{\langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(y , z){\gamma_5}{}^{sl}\!{\delta}(z , y)\gamma ] { \operatorname{tr}}[{}^{ls}\!{\delta}(y , x){\gamma_5}{}^{sl}\!{\delta}(x , y)\gamma']\,{\rangle}_u \big\ } .
\nonumber \end{aligned}\ ] ] one of the spatial sums can be eliminated by making use of translational invariance , finally giving @xmath93\,{\rangle}_u \nonumber \\ & + & { \langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(0,z){\gamma_5}{}^{sl}\!{\delta}(z,0)\gamma { } ^{ls}\!{\delta}(0,x){\gamma_5}{}^{sl}\!{\delta}(x,0)\gamma']\,{\rangle}_u \\ & -&{\langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(0,z){\gamma_5}{}^{sl}\!{\delta}(z,0)\gamma ' ] { \operatorname{tr}}[{}^{ls}\!{\delta}(0,x){\gamma_5}{}^{sl}\!{\delta}(x,0)\gamma]\,{\rangle}_u \nonumber \\ & -&{\langle}\,{\operatorname{tr}}[{}^{ls}\!{\delta}(0,z){\gamma_5}{}^{sl}\!{\delta}(z,0)\gamma ] { \operatorname{tr}}[{}^{ls}\!{\delta}(0,x){\gamma_5}{}^{sl}\!{\delta}(x,0)\gamma']\,{\rangle}_u \big\ } .
\nonumber \end{aligned}\ ] ] all 3-point functions can be built from @xmath94 where the first propagator can be obtained from the second one by means of ( [ eq : proprel ] ) . in terms of ( [ eq : q ] )
the 3-point function reads @xmath95 \,\rangle_u \\ & + \langle { \operatorname{tr } } [ q(-\tau ) \gamma q(t-\tau ) \gamma ' ] \,\rangle_u \\ & -\langle\ , { \operatorname{tr } } [ q(-\tau ) \gamma ' ] { \operatorname{tr } } [ q(t-\tau ) \gamma ] \,\rangle_u \\ & -\langle\ , { \operatorname{tr } } [ q(-\tau ) \gamma ] { \operatorname{tr } } [ q(t-\tau ) \gamma ' ] \,\rangle_u \big\ } .
\\ \end{split}\ ] ] by computing the propagators from a local source at @xmath96 to all @xmath97 using sink - smearing we obtain the 3-point functions for all @xmath86 and @xmath81 .
the numerical calculations are done on a @xmath98 lattice at @xmath99 in the quenched approximation .
we use wilson fermions . to be able to extrapolate our results to the chiral limit
, the calculations are done at three values of the hopping parameter , @xmath100 , 0.1530 and 0.1515 .
this corresponds to physical quark masses of about 70 , 130 and 190 mev , respectively .
for the gauge update we use a 3-hit metropolis sweep followed by 16 overrelaxation sweeps , and we repeat this 50 times to generate a new configuration .
our data sample consists of 400 configurations . in the following
we shall restrict ourselves to zero momentum pion states and operators with @xmath101 . in fig .
[ fig : plat ] we show the ratio ( [ eq : ratio ] ) for the operator @xmath102 at three different values of @xmath86 .
we find very good plateaus in @xmath81 . in our final fits
we have averaged over @xmath86 values around @xmath103 . to obtain continuum results we have to multiply each quark field by @xmath104 , and a factor of @xmath105 is needed to convert to the continuum normalization of states : @xmath106 for the reduced matrix element @xmath30 ( in ( [ eq : htwist ] ) ) we then find @xmath107 where @xmath108 is the renormalized operator , as given in ( [ eq:1-loop ] ) .
.pion masses and unrenormalized ( bare ) decay constants . [ cols="<,<,<",options="header " , ] the lattice pion masses are given in table [ tab : fpimpi ] .
in the following we shall express the dimensionful matrix elements in terms of the pion decay constant @xmath109 , whose unrenormalized values are also given in the table .
the pion masses and the decay constants are taken from @xcite , where we had a slightly higher statistics . in table
[ tab : latres ] we give our results for the unrenormalized , reduced matrix elements @xmath110 of the various @xmath15 operators .
the reader can easily check that the fierz identities ( [ eq : fierz ] ) hold identically at each value of @xmath111 . in fig .
[ fig : chifit1 ] we plot @xmath112 for the operators without color matrices , and in fig .
[ fig : chifit2 ] for the operators with color matrices , as a function of @xmath113 .
the data suggest a linear extrapolation to the chiral limit .
the result of the extrapolation is given in table [ tab : latres ] .
the critical hopping parameter is @xmath114 .
we are now ready to give results for the structure function . to minimize effects of higher order contributions to the wilson coefficient @xmath115 in @xmath116 , we shall take @xmath117 the @xmath116 in ( [ eq : htcoeff ] ) is therefore replaced by @xmath118 .
if we fix the scale by adjusting the @xmath119 mass to its physical value , we have @xcite @xmath120 .
if , instead , we take the string tension or the force parameter to set the scale , we have @xmath121 .
we calculate the renormalization constants from ( [ eq : renormres ] ) with @xmath122 . combining these with the unrenormalized lattice results in table [ tab : latres ] , we find for the @xmath15 structure function @xmath123 multiplying this number with the wilson coefficient and the kinematical factor , and expressing the result in terms of the renormalized decay constant @xmath124 with @xmath125 , computed in perturbation theory
, we finally obtain @xmath126 all numbers refer to the chiral limit .
an early calculation @xcite , based on 15 configurations , found a negative value for @xmath127 .
it is instructive to compare ( [ final ] ) with the twist-2 contribution to the structure function of ( say ) the @xmath128 .
in @xcite we found @xmath129 relative to this number ( [ final ] ) is only a small correction , except perhaps at very small values of @xmath130 . in case of the nucleon we may expect similar numbers , but with @xmath109 being replaced by the nucleon mass .
this would then result in a significant correction .
calculations of 4-fermi contributions to the nucleon structure function are in progress .
the numerical calculations have been done on the quadrics computers at desy - zeuthen .
we thank the operating staff for support . this work was supported in part by the deutsche forschungsgemeinschaft . | we present quenched lattice qcd results for the contribution of higher - twist operators to the lowest non - trivial moment of the pion structure function .
to be specific , we consider the combination @xmath0 which has @xmath1 and receives contributions from 4-fermi operators only .
we introduce the basis of lattice operators .
the renormalization of the operators is done perturbatively in the @xmath2 scheme using the t hooft - veltman prescription for @xmath3 , taking particular care of mixing effects .
the contribution is found to be of @xmath4 , relative to the leading contribution to the moment of @xmath5 .
desy 99 - 098 + hlrz 99 - 31 + hub - ep-99/33 + tpr-99 - 14 + , , , , , , and structure functions , higher twist , lattice qcd . |
we consider billiard dynamics on a rectangular domain , i.e. , point shaped `` balls '' moving with linear motion with specular reflections at the boundary , and similarly for rectangular box shaped domains in three dimensions .
we wish to determine the distribution of free path lengths of ensembles of trajectories defined by selecting a starting point and direction at random .
the question seems quite natural and interesting on its own , but we mention that it originated from the study of electromagnetic fields in `` reverberation chambers '' under the assumption of highly directional antennas @xcite . briefly , the connection is as follows ( we refer to the forthcoming paper @xcite for more details ) : given an ideal highly directional antenna and a highly transient signal , then the wave pulse dynamics is essentially the same as a point shaped billiard ball traveling inside a chamber , with specular reflection at the boundary .
signal loss is dominated by ( linear ) `` spreading '' of the electromagnetic field and by absorption occurring at each interaction ( `` bounce '' ) with the walls .
the first simple model we use in this paper neglects absorption effects , and models signal loss from spreading by simply terminating the motion of the ball after it has travelled a certain large distance .
the second model only takes into account signal loss from absorption , and completely neglects spreading ; here the motion is terminated after the ball has bounced a certain number of times .
we remark that the distribution of free path lengths is very well studied in the context of the lorentz gas here a point particle interacts with hard spherical obstacles , either placed randomly , or regularly on euclidean lattices ; recently quasicrystal configurations have also been studied ( cf . @xcite . )
let @xmath5 be large and let a rectangular @xmath0-dimensional box @xmath6 be given , where @xmath7 .
we send off a large number @xmath8 of particles , each with a random initial position @xmath9 chosen with respect to a given probability measure @xmath10 on @xmath11 , and each with a uniformly random initial direction @xmath12 , @xmath13 , for a total distance @xmath1 each .
each particle travels along straight lines , changing direction precisely when it hits the boundary of the box , where it reflects specularly .
we record the distance travelled between each pair of consecutive bounces for each particle .
( note in particular that we obtain more bounce lengths from some particles than from others . )
let @xmath14 be the uniformly distributed random variable on this finite set of bounce lengths of all the particles .
more precisely , a random sample of @xmath14 is obtained as follows : first take a random i.i.d .
sample of points ( with respect to the measure @xmath10 ) @xmath15 , and a random sample of directions @xmath16 ( with respect to the uniform measure ) .
each pair @xmath17 then defines a trajectory @xmath18 of length @xmath1 , and each such trajectory gives rise to a finite multiset @xmath19 of lengths between consecutive bounces . finally , with @xmath20 denoting the ( multiset ) union of bounce length multisets @xmath21 , we select an element of @xmath22 with the uniform distribution .
( that is , with @xmath23 denoting the integer valued set indicator function for @xmath22 , and @xmath24 we select the element @xmath25 with probability @xmath26 . )
we are interested in the distribution of @xmath14 for large @xmath27 and @xmath1 , and this turns out to be closely related to a model arising from integral geometry .
namely , let @xmath28 denote the unique ( up to a constant ) translation- and rotation - invariant measure on the set of directed lines @xmath29 in @xmath30 , and consider the restriction of this measure to the set of directed lines @xmath29 intersecting @xmath11 , normalized such that it becomes a probability measure
. denote by @xmath31 the random variable @xmath32 where @xmath29 is chosen at random using this measure .
[ thm_n ] for any dimension @xmath7 , and for any distribution @xmath10 on the starting points , the random variable @xmath14 converges in distribution to the random variable @xmath31 , as we take @xmath33 followed by taking @xmath34 , or vice versa .
the mean free path length has a quite simple geometric interpretation .
we have @xmath35 } = 2\pi \dfrac{|\mathbb s^{n-1}|}{|\mathbb s^n|}\dfrac{\operatorname{vol}(k)}{\operatorname{area}(k ) } = 2\sqrt{\pi } \cdot \dfrac{\gamma(\frac{n+1}2)}{\gamma(\frac n2)}\dfrac{\operatorname{vol}(k)}{\operatorname{area}(k ) } \label{santalo - formula } \end{gathered}\ ] ] where @xmath36 is the @xmath37-dimensional surface area of the box @xmath11 , @xmath38 is the volume of the box @xmath11 , @xmath39 is the gamma function , and where @xmath40 is the @xmath37-dimensional surface area of the sphere @xmath41 .
the formula in has been proven in a more general setting earlier ( see e.g. formula ( 2.4 ) in @xcite ) ; for further details , see section [ sec : discussion ] . for the convenience of the reader
we give a short proof of formula in our setting in section [ sec : integral - geometry - mean - value ] . throughout the paper , we will write @xmath42 and @xmath43 for the probability density function and the cumulative distribution function of @xmath44 , respectively , for random variables @xmath44 .
we next give explicit formulas for the probability density function of @xmath31 in dimensions two and three .
[ thm_2 ] for a box of dimension @xmath45 with side - lengths @xmath46 , the probability density function of @xmath31 is given by @xmath47 for @xmath48 .
we note that the probability density function in theorem [ thm_2 ] is analytic on all open subintervals of @xmath49 not containing @xmath50 or @xmath51 moreover , it is constant on the interval @xmath52 and has singularities of type @xmath53 and @xmath54 just to the right of @xmath50 and @xmath55 , respectively .
see figure [ fig : easy - rectangle ] for more details . for an explanation of these singularities , see remark [ rem : singularity - explained ] .
[ fig : easy - rectangle ] for @xmath56 .
( simulation used @xmath57 particles , each starting at the origin with a uniformly random direction , going for a total distance @xmath58 each . )
the plot is cutoff at @xmath59 since @xmath60 tends to infinity as @xmath61 and @xmath62 .
, title="fig:",width=491 ] [ thm_3 ] for a box of dimension @xmath63 with side - lengths @xmath64 , the probability density function of @xmath31 is given by @xmath65 where @xmath66 is the piecewise - defined function given by @xmath67 for @xmath68 , and by @xmath69 for @xmath70 , and by @xmath71 for @xmath72 [ fig : easy_3d ] for @xmath73 .
( simulation used @xmath57 particles , each starting at the origin with a uniformly random direction , going for a total distance @xmath58 each . ) the fact that @xmath60 is not smooth at @xmath74 is barely noticeable . , title="fig:",width=491 ] we note that the probability density function in theorem [ thm_3 ] is analytic on all open subintervals of @xmath75 not containing any of the points @xmath76 moreover , it is linear on the interval @xmath77 and has positive jump discontinuities at the points @xmath64 . at the points @xmath78 , it is continuous and differentiable .
note that the probability distribution @xmath14 gives a larger `` weight '' to some particles than others , since some particles get more bounces than others for the same distance @xmath1 .
one could also consider a similar problem where we send off each particle for a certain number @xmath79 of bounces , and then consider the limit as @xmath34 followed by taking the limit @xmath80 , where @xmath27 is the number of particles .
this would give each particle the same `` weight '' .
denote the finite version of this distribution by @xmath81 and its limit distribution as @xmath34 and then @xmath80 by @xmath82 . with regard to the previous discussion about signal loss
, we call the limit distribution @xmath31 of @xmath14 the * spreading model * and we call the limit distribution of @xmath81 the * absorption model*. determining the probability density function of the absorption model appears to be the more difficult problem , and we give a formula only in dimension two : [ thm_hard ] for a box of dimension @xmath45 with side - lengths @xmath46 , the random variable @xmath81 converges in distribution to the random variable @xmath82 , as we take @xmath34 followed by taking @xmath80 , where the probability density function @xmath83 is given by @xmath84 for @xmath85 , and by @xmath86 for @xmath87 , and by @xmath88 for @xmath89 .
[ bounce_vs_distance ] ( red line ) from theorem [ thm_2 ] vs absorption model ( black dashed line ) from theorem [ thm_hard ] , for @xmath56 . , title="fig:",width=491 ] see figure [ bounce_vs_distance ] for a comparison between the probability density functions for the two different models in dimension @xmath90 .
it is not a priori obvious that the two limit distributions should differ , and it is natural to ask how much , if at all , they differ .
we start by remarking that the expression for @xmath83 does not simplify into the expression for @xmath91 ; indeed , for @xmath56 we have @xmath92 but @xmath93 on the interval @xmath94 .
for very skew boxes , with @xmath95 and @xmath96 , it is straightforward to show that @xmath97 as @xmath96 . given a closed convex subset @xmath98 with nonempty interior it is possible to define a natural probability measure on the set of lines in @xmath99 that have nonempty intersection with @xmath100 .
the expected length of the intersection of a random line is then , up to a constant that only depends on @xmath0 , given by @xmath101 ; this is known as santalo s formula in the integral geometry and geometric probability literature ( cf .
3 ) ) . a billiard flow on a manifold @xmath27 with boundary @xmath102 gives rise to a billiard map ( roughly speaking , the phase space @xmath103 is then the collection of inward facing unit vectors @xmath104 at each point @xmath105 ) .
given @xmath106 we define the associated free path as the distance the billiard particle , starting at @xmath107 in the direction @xmath104 , covers before colliding with @xmath102 again .
as the billiard map carries a natural probability measure @xmath108 we can view the free path as a random variable , and the mean free path is then just its expected value .
remarkably , the mean free path ( again up to a constant that only depends on the dimension ) is then given by @xmath109 even for non - convex billiards .
this was deduced in the seventies at the moscow seminar on dynamical systems directed by sinai and alekseev but was never published and hence rederived by a number of researchers . for further details and an interesting historical survey ,
see chernov s paper ( * ? ? ?
* sec . 2 ) . in spirit
our methods are closely related to the ones used by barra - gaspard @xcite in their study of the level spacing distribution for quantum graphs , and this turns out to be given by the distribution of return times to a hypersurface of section of a linear flow on a torus . in particular , for graphs with a finite number of disconnected bonds of incommensurable lengths , the hypersurface of section is the `` walls '' of the torus , and the level spacings of the quantum graph is exactly the same same as the free path length distribution in our setting when all particles have the _ same _ starting velocity .
( in particular , compare the numerator in for @xmath104 fixed with ( * ? ? ?
* equation ( 49 ) ) . ) in @xcite , marklof and strmbergsson used the results by barra - gaspard to determine the gap distribution of the sequence of fractional parts of @xmath110 .
the gap distribution depends on whether @xmath111 is trancendental , rational or algebraic ; quite remarkably the density function @xmath112 for these gaps share a number of qualitative features with the density function @xmath113 for free paths in our setting .
namely , the density functions both have compact support and are smooth apart from a finite number of jump discontinuities .
further , in some cases the density function is constant for @xmath114 small ; compare figure [ fig : easy - rectangle ] ( here @xmath115 ) with ( * ? ? ?
* figure 4 ) ( here @xmath116 ) .
however , there are some important differences : for @xmath112 , left and right limits exist at the jump discontinuities , whereas for @xmath115 , the right limit of @xmath113 is @xmath117 at the jumps ( cf . figure [ fig : easy - rectangle ] . ) further , despite appearences , @xmath112 is not linear near @xmath118 ( cf .
* figure 1 ) corresponding to @xmath119 ) whereas for @xmath120 , @xmath113 is indeed linear near @xmath118 ( cf . figure [ fig : easy_3d ] ) .
we would like to thank z. rudnick for some very helpful discussions , especially for suggesting the connection with integral geometry .
we also thank j. marklof for bringing references @xcite to our attention .
was partially supported by a grant from the swedish research council ( 621 - 2011 - 5498 ) .
was partially supported by grants from the gran gustafsson foundation for research in natural sciences and medicine , and the swedish research council ( 621 - 2011 - 5498 ) .
in this section , we prove theorem [ thm_n ] . for notational simplicity ,
we give the proof in dimension three ; the general proof for @xmath7 dimensions is analogous . given a particle with initial position @xmath121 and initial direction @xmath104 ,
let @xmath122 be the number of bounce lengths we get from that particle as it has travelled a total distance @xmath5 , and let @xmath123 be the number of such bounce lengths of length at most @xmath124 .
the uniform probability distribution on the set of bounce lengths of @xmath27 particles with initial positions @xmath125 and initial directions @xmath126 has the cumulative distribution function @xmath127 ( note that the denominator is uniformly bounded from below , which follows from equation below . ) by the strong law of large numbers , the function converges almost surely to @xmath128 as @xmath34 , where @xmath129 is the probability measure with which we choose the starting points , and @xmath130 is the surface area measure on the sphere @xmath131 . by symmetry
, we may restrict the inner integrals to @xmath132 .
we now look at the limit of as @xmath33 , and we note that since the integrands are uniformly bounded , we may move the limit inside the integrals by the lebesgue dominated convergence theorem .
fix one of the integrands , and denote it by @xmath133 .
we will show that its limit @xmath134 exists for all @xmath135 and all directions @xmath136 .
moreover , if @xmath137 and @xmath138 denote random variables corresponding to an initial position and an initial direction , respectively , as above , then @xmath139 is a random variable with finite variance ( and similarly for the terms in the denominator of ; in particular recall it is uniformly bounded from below ) , and thus the strong law of large numbers gives that the limit of as @xmath33 , and then @xmath140 almost surely equals .
this shows that @xmath141 exists almost surely and is equal to @xmath142 .
consider a particle with initial position @xmath121 and initial direction @xmath143 . by `` unfolding '' its motion with specular reflections on the walls of the box to the motion along a straight line in @xmath30
see figure [ unfold_rgb ] for a 2d illustration we see that the particle s set of bounce lengths is identical to the set of path lengths between consecutive intersections of the straight line segment @xmath144 with any of the planes @xmath145 , @xmath146 .
thus we see that @xmath147 for large @xmath1 , and therefore @xmath148 as @xmath33 .
[ unfold_rgb ] now project the line @xmath149 to the torus @xmath150 where @xmath151 and let us identify the torus with the box @xmath11 ; see figure [ unfold_rgb ] .
each bounce length corresponds to a line segment which starts in one of the three planes @xmath152 , @xmath153 or @xmath154 and runs in the direction @xmath104 to one of the three planes @xmath155 or @xmath156 .
there are @xmath157 line segments which start from the plane @xmath154 , and thus the probability that a line segment starts from the plane @xmath154 is @xmath158 as @xmath33 . by the ergodicity of the linear flow on tori ( for almost all directions ) ,
the starting points of these line segments become uniformly distributed on the rectangle @xmath159\times[0,{b}]\times\{0\}$ ] for almost all @xmath160 as @xmath33 ; from here we will assume that @xmath104 is such a direction , and we will ignore the measure zero set of directions for which we do not have ergodicity .
consider one of these line segments and denote its length by @xmath161 and its starting point by @xmath162 .
for an arbitrary parameter @xmath124 , we have @xmath163 if and only if @xmath164 or @xmath165 or @xmath166 ; the starting points @xmath167\times[0,{b}]$ ] which satisfy this are precisely those outside the rectangle @xmath168\times[0,{b}-t{v_y}]$ ] assuming that @xmath169 and otherwise it is the whole rectangle @xmath159\times[0,{b}]$ ] . the area of that region is @xmath170 if @xmath171 and otherwise it is @xmath172 .
since the starting points @xmath173 are uniformly distributed in the rectangle @xmath159\times[0,{b}]$ ] as @xmath33 , it follows that the probability that @xmath163 is @xmath174 where @xmath175 is the indicator function which is @xmath176 whenever the condition @xmath177 is true , and @xmath178 otherwise .
we get analogous expressions for the case when a line segment starts in the plane @xmath152 or @xmath153 instead .
thus the proportion of all line segments with length at most @xmath135 as @xmath33 is @xmath179 which can be written @xmath180 recognizing that both integrands and are independent of the position @xmath121 , we see that the limit of as @xmath33 may be written as @xmath181 for all @xmath182 . the corresponding formula in @xmath0 dimensions
is given by @xmath183 for all @xmath182 , where the side - lengths of the box @xmath11 are @xmath184 and @xmath130 is the surface area measure on @xmath185 .
( the denominator can be given explicitly by using lemma [ integral_of_vn ] below . )
we have thus proved that the random variable @xmath14 converges in distribution to a random variable with probability density function given by as we take @xmath34 followed by taking @xmath33 , or alternatively , first taking @xmath33 followed by taking @xmath34 .
it remains to prove that this distribution agrees with the distribution of the random variable @xmath31 defined in the introduction .
we start by recalling some standard facts from integral geometry ( cf .
the set of directed straight lines @xmath29 in @xmath186 can be parametrized by pairs @xmath187 where @xmath188 is a unit vector pointing in the same direction as @xmath29 and @xmath189 is the unique point in @xmath29 which intersects the plane through the origin which is orthogonal to @xmath104 .
the unique translation- and rotation - invariant measure ( up to a constant ) on the set of directed straight lines in @xmath186 is @xmath190 where @xmath191 is the surface measure on the plane through the origin orthogonal to @xmath188 , and @xmath130 is the surface area measure on @xmath131 .
consider the set @xmath192 of directed straight lines in @xmath186 which intersect the box @xmath11 . now , since @xmath193 is the area of the projection of the box @xmath11 onto the plane @xmath194 for @xmath160 , it follows that the total measure of @xmath195 with respect to @xmath28 is @xmath196 where we used symmetry , and the integral may be evaluated by switching to spherical coordinates .
it follows that @xmath197 is a probability measure on the set of directed lines intersecting the box @xmath195 .
let @xmath29 be a random directed line with respect to this measure , and define the random variable @xmath32 , as in the introduction .
let us determine the probability that @xmath198 for an arbitrary parameter @xmath124
. by symmetry it suffices to consider only directed lines with @xmath160 .
the set of all intersection points between the rectangle @xmath159\times[0,{b}]\times\{0\}$ ] and the lines @xmath29 with @xmath198 and direction @xmath160 has area @xmath199 , as in , and its projection onto the plane @xmath194 has area @xmath200.\end{gathered}\ ] ] by symmetry it follows that the area of the set of directed lines @xmath201 with @xmath198 and direction @xmath202 projected down to @xmath194 is @xmath203 + \\ & \phantom{{\mathrel{\mathop:}=}.}{v_y}{\mathopen{}\mathclose\bgroup\oldleft}[{a}{c}-({a}-t{v_x})({c}-t{v_z})\chi({a}\geq t{v_x},{b}\geq t{v_y},{c}\geq t{v_z}){\aftergroup\egroup\oldright } ] + \\ & \phantom{{\mathrel{\mathop:}=}. } { v_z}{\mathopen{}\mathclose\bgroup\oldleft}[{a}{b}-({a}-t{v_x})({b}-t{v_y})\chi({a}\geq
t{v_x},{b}\geq t{v_y},{c}\geq t{v_z}){\aftergroup\egroup\oldright}],\end{aligned}\ ] ] and it follows that @xmath204 = \dfrac1{c_{{a},{b},{c } } } \int_{x\leq t}{\mathop{d\ell } } = \dfrac 8{c_{{a},{b},{c}}}\int_{\mathbb s^2_+ } u(v , t ) { \mathop{ds}}(v),\end{gathered}\ ] ] which we see is identical to , and we have thus proved that @xmath14 converges in distribution to @xmath31 as we take @xmath34 and then @xmath33 .
this concludes the proof of theorem [ thm_n ] .
[ sec : integral - geometry - mean - value ] we will determine the mean value of @xmath31 ; to do this we exploit the integral geometry interpretation of the random variable @xmath31 . by symmetry it suffices to restrict to directed lines @xmath29 with @xmath160 . for fixed @xmath160 , denote by @xmath205 the set of @xmath189 such that the directed line @xmath29 parametrized by @xmath187 intersects @xmath11 .
we note that @xmath206 is a volume element of the box @xmath11 for any fixed @xmath160 , and thus integrating @xmath206 over all @xmath207 yields the volume of the box .
hence the mean value is @xmath35}= \dfrac8{c_{{a},{b},{c } } } \int_{\mathbb s^2_+}\int_{q(v)}x{\mathop{da}}(q){\mathop{ds}}(v ) = \dfrac{8{a}{b}{c } } { c_{{a},{b},{c } } } \int_{\mathbb s^2_+ } { \mathop{ds}}(v ) = \dfrac{2{a}{b}{c } } { { a}{b}+{a}{c}+{b}{c}}.\end{gathered}\ ] ] in @xmath0 dimensions we get a normalizing factor @xmath208 , so with the aid of the lemma [ integral_of_vn ] in the appendix , it follows that the mean value in @xmath0 dimensions is @xmath209 } = \dfrac1{2^n \dfrac1{\pi}\dfrac{|\mathbb s^n|}{2^{n } } \frac{\operatorname{area}(k)}2 } 2^n\operatorname{vol}(k)\dfrac{|\mathbb s^{n-1}|}{2^n } = 2\pi \dfrac{|\mathbb s^{n-1}|}{|\mathbb s^n|}\dfrac{\operatorname{vol}(k)}{\operatorname{area}(k)}\end{gathered}\ ] ] where @xmath36 is the @xmath37-dimensional surface area of the box @xmath11 , and @xmath38 is the volume of the box @xmath11 .
using formula in dimension @xmath45 , we get @xmath210 we use polar coordinates @xmath211 so that @xmath212 .
then the above becomes @xmath213 _ { \cos{^{-1}}(\min(a / t , 1 ) ) } ^ { \sin{^{-1}}(\min(b / t , 1 ) ) } .
\label{two_easy } \end{gathered}\ ] ] the numerator of the second term may be written @xmath214 which can be simplified to @xmath215 inserting this into and differentiating yields theorem [ thm_2 ] .
we will evaluate the cumulative distribution function and then differentiate .
the denominator of the second term of is @xmath216 as may be evaluated by switching to spherical coordinates .
define @xmath217 by symmetry , we have @xmath218 and thus we can write the numerator in the second term of as @xmath219 exploiting the symmetries , it suffices to evaluate @xmath220 and @xmath221 ( note the order of the arguments to @xmath222 ) .
we will evaluate these integrals by switching to spherical coordinates , but first we need to parametrize the part of the sphere inside the box @xmath223 .
[ param_lemma ] fix @xmath224 .
we have @xmath225 for any integrable function @xmath226 , where @xmath227 , where @xmath228 and where we have used the shorthand @xmath229 .
we will parametrize the set of points @xmath230 on the sphere @xmath131 such that @xmath231 switch to spherical coordinates @xmath232 .
the non - negativity conditions of are equivalent to the condition @xmath233 . for such angles ,
the condition @xmath234 is equivalent to @xmath235 and the conditions @xmath236 are equivalent to @xmath237 the interval is non - empty for precisely those @xmath238 such that @xmath239 since @xmath240 thus we may restrict @xmath241 to the interval given by the inequalities @xmath242 note that we have @xmath243 for all @xmath244 since @xmath245 we conclude that we can write @xmath246 for @xmath238 , note that @xmath247 is defined precisely when @xmath248 and that @xmath249 is defined precisely when @xmath250 .
we have @xmath251 if and only if @xmath252 , and we have @xmath253 if and only if @xmath254 .
moreover we note that we always have @xmath255 $ ] .
let us rewrite the integration limits in the right - hand side of in terms of @xmath256 and @xmath257 .
a priori , we need to distinguish between the two cases @xmath258 and @xmath259 . if @xmath258 then we get @xmath260 if on the other hand @xmath261 then @xmath262 which we see is identical to . combining and we get the conclusion of the lemma .
applying lemma [ param_lemma ] we get @xmath263 an antiderivative of the integrand @xmath264 with respect to @xmath265 is @xmath266 , and thus the above is @xmath267_{{\theta_{\text{min}}}}^{\theta_{a } } + { \mathopen{}\mathclose\bgroup\oldleft } [ a^2 \frac12 \sin^2 \theta { \aftergroup\egroup\oldright}]_{\theta_{a}}^{{{\theta_{\text{max}}}}}+ { \mathopen{}\mathclose\bgroup\oldleft } [ b^2\frac12\sin^2\theta - t^2\frac14\sin^4\theta { \aftergroup\egroup\oldright}]_{\theta_{b}}^{{\theta_{\text{max}}}}{\aftergroup\egroup\oldright } ) . \label{h_explicit}\end{gathered}\ ] ] next consider @xmath268 an antiderivative of the integrand @xmath269 with respect to @xmath265 is @xmath270 , and thus the above is @xmath271_{{\theta_{\text{min}}}}^{\theta_{a } } + \dfrac{a^2}{t^2}{\mathopen{}\mathclose\bgroup\oldleft}[-\cos\theta{\aftergroup\egroup\oldright}]_{\theta_{a}}^{{{\theta_{\text{max } } } } } + { \mathopen{}\mathclose\bgroup\oldleft}[-\dfrac{b^2\cos \theta}{t^2}-\frac{\cos^3\theta}3+\cos\theta{\aftergroup\egroup\oldright}]_{\theta_{b}}^{{\theta_{\text{max}}}}{\aftergroup\egroup\oldright } ) . \label{g_explicit}\end{gathered}\ ] ] we obtain @xmath272 and @xmath273 by switching the roles of @xmath64 in .
we remark that trying to obtain @xmath272 and @xmath273 directly , by integrating @xmath274 and @xmath275 , respectively , by first integrating with respect to @xmath265 , taking the limits @xmath276 and @xmath277 , and then finding an antiderivative with respect to @xmath241 , seem to result in much more complicated expressions
. finally consider @xmath278 an antiderivative of the integrand @xmath279 with respect to @xmath265 is @xmath280 , and thus the above is @xmath281_{{\theta_{\text{min}}}}^{\theta_{a } } + { \mathopen{}\mathclose\bgroup\oldleft } [ \dfrac{-a\cos\theta}{t } { \aftergroup\egroup\oldright}]_{\theta_{a}}^{{{\theta_{\text{max } } } } } - \int_{\theta_{b}}^{{\theta_{\text{max}}}}\sqrt{1-\dfrac{b^2}{t^2 } - \cos^2\theta } \sin\theta { \mathop{d\theta } } { \aftergroup\egroup\oldright } ) \label{f_explicit}\end{gathered}\ ] ] where the last integral inside the parentheses may be written as @xmath282_{\theta_{b}}^{{\theta_{\text{max}}}}= \\ { \mathopen{}\mathclose\bgroup\oldleft } [ -\frac{1}{2 } { \mathopen{}\mathclose\bgroup\oldleft}(\cos\theta \sqrt{\sin^2\theta-\dfrac{b^2}{t^2}}+{\mathopen{}\mathclose\bgroup\oldleft}(1-\dfrac{b^2}{t^2}{\aftergroup\egroup\oldright } ) { \tan{^{-1}}}{\mathopen{}\mathclose\bgroup\oldleft}(\frac{\cos\theta}{\sqrt{\sin^2\theta-\dfrac{b^2}{t^2}}}{\aftergroup\egroup\oldright}){\aftergroup\egroup\oldright } ) { \aftergroup\egroup\oldright}]_{\theta_{b}}^{{\theta_{\text{max}}}}\label{hideous}\end{gathered}\ ] ] whenever @xmath283 , by using the fact that @xmath284 is an antiderivative of @xmath285 with respect to @xmath107 when @xmath286 is a constant .
we obtain @xmath221 and @xmath287 by switching the roles of @xmath64 in .
it remains to insert the limits @xmath288 into the antiderivatives , and above .
noting that @xmath288 are expressed in terms of piecewise - defined functions , the following manipulations will be useful .
for any function @xmath289 , we have @xmath290 where @xmath291 .
similarly , @xmath292 where @xmath293 , and @xmath294 and similarly , @xmath295 can be written as @xmath296 with this we can evaluate @xmath297_{{\theta_{\text{min}}}}^{\theta_{a } } , [ \psi]_{\theta_{a}}^{{\theta_{\text{max } } } } , [ \psi]_{\theta_{b}}^{{\theta_{\text{max}}}}$ ] .
but since we know that we will get a function symmetric with respect to the values @xmath64 , it suffices to keep only those terms with @xmath298 and @xmath299 , say , and then the other terms may be evaluated by just switching the order of @xmath64 . upon inserting the limits and differentiating ,
one obtains ( after tedious calculations ) that @xmath65 where @xmath300 .
\end{align } \end{gathered}\ ] ] rewriting @xmath66 as a piecewise function , we get theorem .
consider the distribution of the random variable @xmath81 . since we record the _ same _ number of bounces for each choice of angle @xmath265
we may replace the @xmath27-particle system with a one particle system @xmath301 as follows : randomly select , with uniform distribution , the angle @xmath265 and generate @xmath3 bounce lengths and randomly select one of these bounce lengths ( with uniform distribution ) ; by the strong law of large numbers , @xmath81 converges in distribution to @xmath301 as @xmath34 .
we now determine the limit distribution of @xmath301 .
as before , we first unfold the motion , and replace motion in a box with specular reflections on the walls with motion in @xmath302 ; see figure [ unfold_rgb ] .
the path lengths between bounces is then the same as the lengths between the intersections with horizontal or vertical grid lines . to understand the spatial distribution , we project the dynamics to the torus @xmath303 where @xmath304 is the lattice @xmath305 and we may identify the torus with the rectangle @xmath159\times[0,{b}]$ ] .
let us first consider the motion of a single particle with an arbitrary initial position , and direction of motion given by an angle @xmath265 .
taking symmetries into account , we may assume that @xmath306 $ ] .
( note that @xmath307 gives a probability measure on these angles . )
if the particle travels a large distance @xmath5 , the number of intersections with horizontal , respectively vertical , grid lines is @xmath308 , respectively @xmath309 .
thus , in the limit @xmath2 , the probability of a line segment beginning at a horizontal ( respectively vertical ) grid line is given by @xmath310 , respectively @xmath311 ( here we suppress the dependence on @xmath265 ) where @xmath312 the unfolded flow on the torus is ergodic for almost all @xmath265 , and thus the starting points of the line segments becomes uniformly distributed as @xmath33 for almost all @xmath265 .
let @xmath313 since @xmath314 , we obtain that @xmath315 let @xmath316 denote the angle of the diagonal in the box , and assume that @xmath317 .
we then observe the following regarding the line segment lengths .
first , if the segment begins at a horizontal line , it must end at a vertical line , and the possible lengths of these segment lie between @xmath178 and @xmath161 .
we find that these lengths are uniformly distributed in @xmath318 $ ] since the starting points of the segments are uniformly distributed . on the other hand ,
if the line segment begins at a vertical line , it can either end at a vertical or horizontal line .
since the starting points are uniformly distributed , the former happens with probability @xmath319 and the length of the segment is again uniformly distributed in @xmath318 $ ] , whereas the latter happens with probability @xmath320 in which case the segment is always of length @xmath161 .
now , @xmath321 $ ] implies that @xmath322 $ ] , and noting that @xmath323 we find that the probability of observing a line segment of length @xmath135 is the sum of a `` singular part '' ( the segment begins and ends on vertical lines ; note that all such segments have the _ same _ lengths ) and a `` smooth part '' ( the segment does not begin and end on vertical lines ) .
moreover , the smooth part contribution equals @xmath324 which , on inserting ( [ eq : vert - horiz - probs ] ) , equals @xmath325 on the other hand , the `` singular part contribution '' , provided @xmath326 , to the probability of a segment having length @xmath135 equals @xmath327 in case @xmath328 , a similar argument ( we simple reverse the roles of @xmath50 and @xmath55 ) shows that the smooth contribution equals @xmath329 and that the singular contribution ( if @xmath330 ) equals @xmath331 thus , if we let @xmath332 denote the `` singular contribution '' to the probability density function we find the following : if @xmath333 , then @xmath334 if @xmath335 $ ] , then @xmath336 and if @xmath337 $ ] , then @xmath338 [ rem : singularity - explained ] note that @xmath339 has a singularity of type @xmath53 just to the right of @xmath340 ( and similarly just to the right of @xmath341 ) . in a sense
this singularity arises from the singularity in the change of variables @xmath342 since @xmath343 .
the reason for the singularities in the spreading model for @xmath45 is similar , as the spreading model can be obtained from the absorption model by a smooth change of the angular measure .
similarly , the `` smooth part '' of the contribution is ( for @xmath344 $ ] ) given by @xmath345 hence the probability density function of the distribution of the segment length @xmath135 is given by @xmath346 we will now evaluate @xmath347 .
an antiderivative of @xmath348 with respect to @xmath161 for @xmath349 is @xmath350 where @xmath351 for @xmath352 .
( a quick calculation shows that @xmath353 whenever @xmath354 . ) we can rewrite as @xmath355 by lhpital s rule we have @xmath356 so the limit of as @xmath357 is @xmath358 the limit of as @xmath359 is @xmath360 thus , assuming @xmath361 , we can write @xmath362 as @xmath363 if @xmath364 , or as @xmath365 if @xmath87 or as @xmath366 if @xmath367 . adding @xmath332 to this , we get theorem [ thm_hard ] .
[ integral_of_vn ] write @xmath368 for the @xmath37-dimensional surface area of the sphere @xmath41 . then we have @xmath369 where @xmath370 is the part of the sphere @xmath371 with positive coordinates
. we may parametrize @xmath372 with @xmath373 for @xmath374 .
we have the spherical area element @xmath375 thus we get @xmath376 introducing an additional integration variable @xmath377 , we recognize the integrand as the spherical area element in @xmath378 dimensions , and thus the above is @xmath379 since @xmath380 .
m. bckstrm , s. holmin , p. kurlberg , d. mnsson .
_ randomized ray tracing for modeling uwb transients in a reverberation chamber . _ in preparation .
n. chernov .
entropy , lyapunov exponents , and mean free path for billiards . , 88(1 - 2):129 , 1997 . | we consider the distribution of free path lengths , or the distance between consecutive bounces of random particles , in an @xmath0-dimensional rectangular box . if each particle travels a distance @xmath1 , then , as @xmath2 the free path lengths coincides with the distribution of the length of the intersection of a random line with the box ( for a natural ensemble of random lines ) and we give an explicit formula ( piecewise real analytic ) for the probability density function in dimension two and three . in dimension two
we also consider a closely related model where each particle is allowed to bounce @xmath3 times , as @xmath4 , and give an explicit ( again piecewise real analytic ) formula for its probability density function .
further , in both models we can recover the side lengths of the box from the location of the discontinuities of the probability density functions . |
in @xcite lewowicz and hiraide proved that every expansive homeomorphism of a compact surface @xmath0 is conjugate with a pseudo - anosov diffeomorphism . recall that a homeomorphism @xmath1 is _ expansive _ if there is @xmath2 such that if @xmath3 for all @xmath4 then @xmath5 . in
@xcite kato introduced a generalization of expansivity called _ continuum - wise expansivity_. we say that @xmath6 is _ cw - expansive _ if there is @xmath2 such that if @xmath7 is a continuum ( compact connected ) and @xmath8 for all @xmath4 then @xmath9 is a singleton . in the works of kato on cw - expansivity we find several generalizations of results holding for expansive homeomorphisms .
also , he found new phenomena as for example a cw - expansive homeomorphism with infinite topological entropy . in this paper
we investigate the possibility of extending results from @xcite for a cw - expansive surface homeomorphism .
a key concept in dynamical systems is that of the stable set of a point . given a homeomorphism @xmath1 and @xmath10 we define the @xmath11-_stable set _ of a point @xmath12 as @xmath13 for a hyperbolic set it is well known that local stable sets are embedded submanifolds ( the invariant manifold theorem ) . in the papers
@xcite they prove that if @xmath1 is expansive then the connected component of @xmath14 in @xmath15 is a locally connected set .
this implies the arc - connection of this components and allows them to prove that each local stable set is a finite union of arcs . in some sense
it is an invariant manifold theorem for expansive homeomorphisms of surfaces .
after this , they prove the conjugacy with a pseudo - anosov diffeomorphism , giving a complete classification of such dynamics .
some cw - expansive homeomorphisms of surfaces are not expansive , see @xcite . in these examples
the components of local stable sets are locally connected .
the purpose of this paper is to construct a cw - expansive homeomorphism of a compact surface with a point whose local stable set is connected but it is not locally connected .
the example is a variation of those in @xcite .
we start defining a homeomorphism of @xmath16 such @xmath17 as a fixed point and its stable set is not locally connected .
then , this anomalous saddle is _ inserted _ in a derived from anosov diffeomorphism of the torus .
finally , this anomalous derived from anosov system is connected via a wandering tube with a usual derived from anosov to obtain our example .
first , we will construct a plane homeomorphism @xmath6 with a fixed point at the origin whose local stable set is connected but not locally connected .
the homeomorphism will be defined as the composition of a piece - wise linear transformation @xmath18 and a time - one map of a flow @xmath19 .
this flow will have a non - locally connected set @xmath20 of fixed points .
we start with the linear part of the construction .
let @xmath21 , for @xmath22 , be the linear transformations defined by @xmath23 , @xmath24 , @xmath25 , @xmath26 .
define the piece - wise linear transformation @xmath27 as @xmath28 in figure [ sillaloca ] we illustrate the definition of @xmath18 .
now we define the non - locally connected plane continuum @xmath20 .
some care is needed in order to be able of relate this set with the transformation @xmath18 .
define the sets : @xmath29 also consider the non - locally connected continuum @xmath30\times\{0\})$ ] shown in figure [ conjuntoe ] .
now we will define a flow related with the set @xmath20 .
consider the continuous function @xmath31 defined by @xmath32 and the vertical vector field @xmath33 defined as @xmath34 since @xmath35 for all @xmath36 , we have that @xmath37 is lipschitz .
therefore , by picard s theorem , @xmath38 has unique solutions and we can consider the flow @xmath39 induced by @xmath38 . since @xmath40 for all @xmath41 we have that every solution is defined for all @xmath42 .
consider @xmath52 such that @xmath51 .
since @xmath38 is a vertical vector field we have that @xmath57}(p)\subset r_1 $ ] . for @xmath58 , if @xmath59 then @xmath60 therefore , @xmath61 for @xmath58 and consequently for @xmath62 .
now take a point @xmath70\times[0,1]$ ] . for @xmath71 , the set defined in ( [ eqr1 ] ) , it is easy to see that @xmath72 as @xmath73 .
assume that @xmath74 .
we will show that @xmath75 .
it is sufficient to show that for some @xmath76 the point @xmath77 is not in @xmath78 . by contradiction , assume that @xmath79 for all @xmath80 .
then , by lemma [ lemar1 ] we know that @xmath81 notice that @xmath82 . then , it only rests to prove that @xmath83 for some @xmath76 .
but this is easy because the velocity of @xmath84 is @xmath85 and this velocity increases with @xmath86 .
we start recalling some properties of what is known as a derived from anosov diffeomorphisms .
the interested reader should consult @xcite*section 8.8 for a construction of such a map and detailed proofs of its properties .
a derived from anosov is a @xmath87 diffeomorphism @xmath88 of the two - dimensional torus such that : it satisfies smale s axiom a and its non - wandering set consists of an expanding attractor and a repeller fixed point @xmath89 .
the expanding attractor is locally a cantor set times an arc , and it has two hyperbolic fixed points of saddle type @xmath90 and @xmath91 as in figure [ figda ] .
now we will _ insert _ the anomalous saddle in the derived from anosov .
let @xmath90 be the hyperbolic fixed point shown in figure [ figda ] .
consider a topological rectangle @xmath100 covering a half - neighborhood of @xmath90 as in figure [ figdarect ] .
consider the homeomorphism with an anomalous saddle fixed point defined in section [ secirregular ] . call this homeomorphism @xmath101 ( to avoid confusion with the derived from anosov @xmath6 ) .
denote by @xmath102 its fixed point ( the origin of @xmath16 ) and take a rectangle @xmath103 , similar to @xmath104 , as in figure [ figdarect ] .
now we can _ replace _ @xmath100 with @xmath105 and define what we call a _ derived from anosov with an anomalous saddle _ as in figure [ figanoda ] . in this section
we finish the construction with ideas from @xcite .
consider @xmath106 and @xmath107 two disjoint copies of the torus @xmath108 .
let @xmath109 , @xmath110 , be two homeomorphisms such that : * @xmath111 is the derived from anosov with an anomalous saddle from the previous section , denote by @xmath112 the source fixed point of @xmath111 , * @xmath113 is the inverse of the derived from anosov ( the usual one ) with a sink fixed point at @xmath114 .
consider the open disk @xmath121 and the compact annulus @xmath122 define @xmath123 as the inversion @xmath124 .
the pull - back of the unstable foliation on @xmath107 by @xmath125 on the annulus @xmath126 is shown in figure [ figfols ] .
on the disjoint union @xmath127\cup [ s_2\setminus \varphi_2(d_{1/2})]$ ] consider the equivalence relation generated by @xmath128 for all @xmath129 .
denote by @xmath130 $ ] the equivalence class of @xmath14 .
the surface @xmath131 is the genus two surface if equipped with the quotient topology .
consider the homeomorphism @xmath1 defined by @xmath132)=\left\ { \begin{array}{ll } \left[f_1(x)\right ] & \hbox { if } x\in s_1 \setminus \varphi_1(d_{1/2})\\ \left[f_2(x)\right ] & \hbox { if } x\in s_2 \setminus \varphi_2(d_2)\\ \end{array } \right.\ ] ] define @xmath135 $ ] the annulus on @xmath0 corresponding to @xmath126 .
we will perturb the homeomorphism @xmath6 defined above on the annulus @xmath136 .
first note that the non - wandering set of @xmath6 is expansive and dynamically isolated , i.e. there is a neighborhood @xmath137 of the non - wandering set @xmath138 such that if @xmath139 for all @xmath4 then @xmath140 .
also note that for every wandering point @xmath12 there is @xmath4 such that @xmath141 .
therefore , it is sufficient to prove that there is a homeomorphism @xmath142 such that @xmath143 and there is @xmath144 such that for each @xmath145 the intersection @xmath134 is totally disconnected . in figure [ figfols ] we have the picture of the unstable foliation on @xmath136 ( or in local charts ) .
the problem is that the stable sets do not make a foliation , this is because there is an anomalous saddle .
then , it is convenient to consider the stable partition , i.e. , the partition defined by the equivalence relation of being positively asymptotic .
this partition is illustrated in figure [ figstpart ] .
we know that the unstable leaves are circle arcs , as in figure [ figfols ] .
therefore , it is sufficient to consider a @xmath146 perturbation @xmath101 of @xmath6 supported on @xmath136 , such that the stable partition of @xmath101 in the annulus contains no circle arc , in local charts . see comments below . by the previous comments this implies that @xmath101 is cw - expansive . since @xmath101 coincides with @xmath6 outside @xmath136 , we have that @xmath101 has an anomalous saddle with non - locally connected stable set .
this finishes the proof .
the example has further properties that we wish to remark .
given @xmath147 we say that @xmath6 is @xmath148-_expansive _
@xcite if there is @xmath2 such that @xmath149 for all @xmath12 , where @xmath150 stands for the cardinality of the set @xmath126 .
we have that the example of the previous proof is not @xmath148-expansive for all @xmath147 because there are point with @xmath151 for arbitrarilly small values of @xmath152 .
we say that a probability measure @xmath153 on @xmath0 is an _ expansive measure _
@xcite if there is @xmath2 such that @xmath154 for all @xmath12 .
obviously , if @xmath153 is an expansive measure then @xmath155 for all @xmath12 , i.e @xmath153 is non - atomic . in @xcite
it is shown that every non - atomic probability measure is expansive if and only if there is @xmath2 such that @xmath156 for all @xmath12 .
this property is called _ countable - expansivity _ and our example satisfies this condition . in the generalized pseudo - anosov shown in @xcite
there is a finite number of _ spines _ ( or 1-_prongs _ ) , i.e. points whose local stable sets do not separate arbitrarilly small neighborhoods .
this is a cw - expansive homeomorphism on the two - sphere .
our example has a countable set of spines , namely , the points in the set @xmath20 of figure [ conjuntoe ] in the line @xmath157 give rise to spines in the example .
as explained in @xcite the generalized pseudo - anosov of the two - sphere has points with its local stable set non - locally connected .
but the components are arcs .
our example has connected components not being locally connected .
it seems to be the case that if we start with a set like the graph of @xmath158 , in place of the set @xmath20 , we can obtain an anomalous saddle with no arc - connected stable set .
notice that the set @xmath20 is arc - connected . | we prove that the genus two surface admits a cw - expansive homeomorphism with a fixed point whose local stable set is not locally connected . |
we propose a simple stirring experiment to generate quantized ring currents and solitary excitations in bose - einstein condensates in a toroidal trap geometry .
simulations of the 3d gross - pitaevskii equation show that pure ring current states can be generated efficiently by adiabatic manipulation of the condensate , which can be realized on experimental time scales .
this is illustrated by simulated generation of a ring current with winding number two .
while solitons can be generated in quasi-1d tori , we show the even more robust generation of hybrid , solitonic vortices ( svortices ) in a regime of wider confinement .
svortices are vortices confined to essentially one - dimensional dynamics , which obey a similar phase - offset velocity relationship as solitons .
marking the transition between solitons and vortices , svortices are a distinct class of symmetry - breaking stationary and uniformly rotating excited solutions of the 2d and 3d gross - pitaevskii equation in a toroidal trapping potential .
svortices should be observable in dilute - gas experiments
. persistent currents , quantized vortices , and solitons in bose - einstein condensates ( bec ) are high - energy excitations predicted by mean - field theory . quite recently , both solitonic waves @xcite and singly quantized vortices @xcite have been unambiguously identified experimentally in 3d harmonic trap geometries .
the next step in increasing our understanding of quantized currents and solitons is to study their stability properties and measure their intrinsic lifetimes thereby testing the validity of mean - field theory in a so - far unexplored regime .
the torus presents an ideal geometry in this context as solitons can travel without reaching an end of the trap and ring currents are stabilized by the ring geometry as opposed to vortices where the core can drift @xcite .
excited states in toroidal traps also have the potential to provide useful applications .
e.g. drummond _ et .
al .
_ @xcite have suggested constructing a mode - locked atom laser using a dark soliton with a nodal plane ( also called a black soliton or kink configuration ) in a ring geometry .
they demonstrated the creation of a black soliton by stirring a condensate in a 1d ring with a paddle at constant angular velocity and damping additionally generated excitations by raman outcoupling .
other simulations involving nonlinear excitations in toroidal geometries can be found in references @xcite . in this letter
we propose an adiabatic stirring technique to impose circulation onto a bec in a 3d toroidal geometry . without any need for external cooling or damping mechanisms ,
adiabatic manipulations allow one to generate pure ring - current states with essentially arbitrary circulation .
a variation of the stirring technique can produce solitons if the transverse trap confinement in the toroidal tube is quasi-1d . for
slightly less tight confinement new nonlinear objects are created , which have properties of both vortices and solitons .
we call these objects solitonic vortices , or svortices , and briefly discuss their properties . in our simulations ,
the condensate is stirred with a rotating paddle which can easily be realized with a blue - detuned laser - generated light sheet .
being purely optical in nature , this method does not rely on specific magnetic field configurations and is suitable for single - component as well as spinor condensates .
toroidal trap geometries have already been used in early bec experiments @xcite and in the recent jila vortex experiments @xcite .
quasi-1d confinement in a toroidal trap certainly presents a challenge for experimentalists but seems realistic after promising results in linear geometries @xcite and with the prospect of a tightly confining purely optical toroidal trap @xcite .
we model the dynamics of the bec with the time - dependent gross - pitaevskii or nonlinear schrdinger equation ( nlse ) , which provides the zero - temperature mean - field theory @xcite . for a bec of @xmath0 atoms of mass @xmath1 , confined in an external potential @xmath2
, the nlse reads : @xmath3\psi , \ ] ] where @xmath4 is the coupling constant .
the @xmath5-wave scattering length @xmath6 is assumed to be positive , relating to repulsive interparticle interactions .
the order parameter or condensate wavefunction @xmath7 is normalized to @xmath8d*r * @xmath9 and may be written as @xmath10*r*@xmath11 , where the square root of the density @xmath12 and the phase @xmath13 are real functions of * r * and @xmath14 .
we have solved the time - dependent nlse ( [ nlse ] ) numerically on a grid using a pseudo - spectral fft representation and a 4th - order variable - step runge - kutta integrator in box boundary conditions .
the trap potential @xmath15 mimics a torus by a 3d harmonic - oscillator potential @xmath16 in the shape of an oblate disk pierced by a gaussian blue - detuned laser beam modelled by the core potential @xmath17 resembling the trap used in early mit experiments @xcite .
a time - dependent `` paddle '' potential @xmath18 mimicking an ellipsoidal gaussian light sheet is used to stir the condensate in the torus with @xmath19 ; the core - potential is @xmath20 .
the paddle potential has the form of a gaussian ellipsoid centred at a radius @xmath21 and initially reads @xmath22 . during the simulation
it is accelerated to a uniform rotation around the @xmath23-axis and retracted by pulling out to the side .
the following parameters @xmath24hz , @xmath25nk , @xmath26 m , @xmath27nk , @xmath28 m , @xmath29 m , @xmath30 m , were used in the simulation and are comparable to currently achievable experiments . ] . for a condensate of @xmath31 atoms of na with @xmath32 nm ,
our simulations relate to a torus of radius @xmath33 m ( distance of the potential minimum from the symmetry axis of the torus ) .
the healing length @xmath34*r*@xmath35 , which depends on the spatially inhomogeneous density @xmath36 , sets the relevant length scales for the size of solitons ( @xmath37 ) and the transverse size of vortex filaments ( @xmath38 ) @xcite . from the numerically obtained ground state wavefunction in the torus
, we find a value of @xmath39 m for the size of the healing length at peak density .
the transverse confinement of the condensate in the toroidal tube is best characterized by the number of healing lengths , which we define by the line integral @xmath40 , taken across the toroidal tube along a radial line @xmath41 embedded in the symmetry plane of the torus .
in other geometries , the most sensible curve of integration @xmath41 is dictated by the confinement .
the dimensionless confinement parameter @xmath42 can be computed directly form the numerical wavefunction and does not rely on the thomas - fermi approximation , which is often used to characterize the extent of a trapped condensate . in our simulation the confinement in the radial direction is about the same as in the axial direction and amounts to @xmath43 .
figure [ fig_b48_vort_gen ] shows the result of the simulation of a stirring experiment to generate a ring current with winding number @xmath44 .
the ground state of the condensate in the toroidal trap intersected by a narrow ellipsoidal gaussian paddle potential was found by imaginary - time propagation . during the simulation
the paddle was accelerated to azimuthal rotation `` stirring '' the condensate and subsequently retracted in the direction of the torus axis . .
parts a ) and b ) show gray - scale coded 2d slices of the density @xmath45 and the phase @xmath13 respectively in the @xmath46 plane of the trap .
rows c ) and d ) show the density and the phase on the line of minimal potential parameterized by the angle @xmath47 .
note that the phase is plotted within an arbitrary offset and modulo @xmath48 .
the columns show time slices of our simulation .
the first column shows the condensate ground state in the toroidal trap intersected by a resting stirring paddle .
columns 2 and 3 show how a phase gradient builds up while the stirring paddle is accelerated at a constant rate between @xmath49 and @xmath50s .
afterwards the condensate is stirred with constant angular velocity @xmath51hz . at this stage ,
the density notch goes down to about 1% of the maximum density and a phase step of @xmath52 has built up across the paddle .
the fifth column shows how the paddle is retracted between @xmath53s and @xmath54s .
the last column shows the @xmath44 ring current that has established in the toroidal trap .
the transverse confinement in the torus is @xmath55 healing lengths . ] in order to elucidate the effect of adiabatic stirring on the condensate wavefunction and to interpret the result of the simulation shown in figure [ fig_b48_vort_gen ] , we start with an idealized 1-d picture , neglecting the transverse dimensions of the torus : when blocked by a paddle , the torus can be treated as a 1d box with hard - wall boundary conditions .
the ground state @xmath56 of the condensate in a 1d box is a jacobi elliptic function of type @xmath57 @xcite , which is real - valued and therefore has a constant phase .
when the paddle is moving , the wavefunction acquires a phase ramp .
this can be understood by considering the galilean invariance of the nlse : the boosted ground state @xmath58 $ ] satisfies the boundary conditions in the moving frame and is the `` ground state '' of the translated box .
a phase gradient in the wavefunction is equivalent to a supercurrent with velocity @xmath59 .
a state like this can either be reached by an adiabatic transition between the resting and the moving box through accelerating the paddle slowly or one can prepare the condensate right from the start in a torus with a rotating paddle .
removal of the stirring paddle reconnects both ends of the `` 1d box '' to a ring and can be modelled in 1d as decreasing a narrow potential barrier to zero height . in the adiabatic limit of this process ,
the phase step of @xmath60 acquired by stirring reconnects at the nearest integer multiple of @xmath48 while the density is still low . in this way
phase gradients are smoothed out before the density notch fills in completely leaving the condensate in an azimuthally symmetric current - carrying state thus creating a pure toroidal ring current . in this idealized picture
there is no upper limit for the current velocities . above a critical velocity , however , the supercurrent is instable against phase slips on resting obstacles which might arise through imperfections in the trap @xcite .
it is not obvious that the simple 1d picture of adiabatic stirring sketched above should be transferable to experimental conditions in 3d tori well out of the quasi-1d limit .
the simulation of figure [ fig_b48_vort_gen ] , however , shows that an essentially pure ring - current state can be generated by realistic choices of time scales and parameters . in order to get a rough estimate for the timescale of adiabaticity , consider the time @xmath61 it takes for a sound wave to travel around the ring of radius @xmath62 once .
we use the bogoliubov speed of sound @xmath63 @xcite , which amounts to @xmath64 at the peak density of the numerically obtained ground - state wavefunction , to estimate this time to @xmath65s .
it is further useful to compare the acceleration rate of the stirring paddle to @xmath66 , which is the acceleration that brings the velocity up to the speed of sound in the time that it takes a sound wave to travel around the ring .
we find that acceleration rates of about @xmath67 work well . alternatively , instead of starting with the ground state of an intersected toroidal trap , the light sheet potential barrier can also be ramped up slowly .
our simulations show that ramping up the paddle linearly within about @xmath68 is sufficiently slow in order to make no appreciable difference for the generation of ring currents .
further , retracting the core potential @xmath17 , as realized in reference @xcite , leads to the generation of vortex filaments .
we observe that essentially pure vortex states can be generated by ramping down the core - potential on a timescale of about @xmath69 .
a variation of the stirring experiment as described above can be used to generate solitons . in the non - adiabatic limit , when the intersecting light sheet is removed suddenly , the imposed phase step and density notch set an initial condition for the dynamics of the condensate .
it has been discussed in other places , that the combination of a phase step @xcite and a density notch is a very efficient way to start density - notch solitons @xcite .
for the example of a linear quasi-1d trap it is discussed by carr _ et .
al .
_ @xcite how the combination of a density notch generated by sudden removal of a gaussian potential barrier in combination with a phase step can lead to the generation of one or more solitons with different velocities depending on the size of the phase step and the width of the potential barrier .
we have performed simulations in tightly - confined toroidal traps with a transverse confinement in terms of healing lengths @xmath70 between 2 and 5 , where we observe a quasi-1d type of behaviour and the discussion in reference @xcite applies .
our 3d simulations show the generation of band solitons upon sudden removal of the paddle .
however , in situations where the confinement is less tight and the curvature of the torus becomes apparent the phenomenology changes : we observe the robust generation of a unique class of solitary excitations in a torus with transverse confinement in terms of healing lengths @xmath71 .
these excitations have properties of both solitons and vortices , as will be elaborated below , and therefore we call them solitonic vortices or _ svortices_. figure [ fig_sol_gen ] shows a simulated stirring experiment in a condensate of @xmath31 atoms of na where a phase step of approximately @xmath72 is generated by stirring na atoms : @xmath73hz , @xmath74nk , @xmath75 m ; @xmath76nk ; @xmath77 m ; @xmath78 m .
the toroid radius in this simulation is @xmath79 m . at peak density of the condensate ground state in the torus
we find a bogoliubov sound speed of @xmath80 and healing length of @xmath81 m . ] .
ramping down the paddle potential within 66ms ( which is roughly @xmath82 with @xmath83s in this simulation ) , a solitary wave is created that has characteristics of both solitons and vortices .
while showing the phase singularity signature of a vortex , the density profile is squeezed by the tranverse confinement ( see figure [ fig_svort_iso ] ) .
after the stirring potential is turned off , the svortex moves at constant angular velocity around the ring .
our simulations show that the velocity of the svortex is sensitive to the details of the stirring process like the paddle acceleration rate and timescale of paddle retraction .
similar to solitons , the velocity of svortices relative to the condensate is related to a phase offset . and the phase @xmath13 of the condensate at different times as in figure [ fig_b48_vort_gen ] .
the first and the second column show the acceleration phase which lasts from @xmath84 to @xmath85 .
afterwards the condensate is stirred with constant angular velocity @xmath86hz .
the paddle is contracted between @xmath87 and @xmath88 . due to the transverse confinement being about 9 healing lengths ,
a dark soliton is not completely stabilized but transforms into a vortex by the influence of curvature .
a state is generated which is on the transition between a dark soliton and vortex . ] in the xy plane .
the condensate wavefunction was prepared in a stirring experiment described in the text below .
the solitary density deformation and phase singularity preserve their shape while rotating clockwise about the axis of the torus with roughly half the angular velocity of a ring current carrying a single quantum of vorticity .
the current plot shows the wavefunction at @xmath89s . ]
the velocity of solitons relative to the condensate is given by @xmath90 @xcite , where @xmath91 is the phase step across the density notch relating to standing black solitons ( kink states ) at @xmath92 and to increasingly shallow gray solitons moving with the speed of sound @xmath93 as @xmath91 approaches zero .
the experiments of reference @xcite have confirmed this relation for dark solitons in a bec .
we find that svortices obey a similar relationship .
figure [ fig_velo ] shows the velocity phase - offset relation for a svortex and a band soliton , which have been created by imaginary - time propagation in a long thin rectangular 2d box ( of dimensions @xmath94 ) with hard - wall boundary conditions @xcite .
we find that the svortex velocity depends linearly on small deviations of the phase offset from @xmath72 resembling the relation ( [ velooffset ] ) for solitons .
the proportionality coefficient , however , is below the value for solitons , given by the bogoliubov sound speed @xmath95 .
simulations in differently sized boxes show that this coefficient depends strongly on the transverse confinement .
we observe the discussed behaviour for transverse confinements @xmath70 between 6 and 12 healing lengths .
it is this velocity phase - offset relation , which makes the vortex confined to a thin waveguide a solitonic vortex or svortex .
a detailed analysis of the svortex s properties including the collisional properties , which are considerably more complicated than those of solitons , will be published elsewhere @xcite .
we note that the term `` vortex soliton '' or `` optical vortex soliton '' has been used before in the nonlinear optics community in a much broader context @xcite whereas we introduce the term solitonic vortices referring to transversely confined vortices with specific solitonic properties . .
a stationary black band soliton ( nodal - plane state ) and a stationary svortex were set in motion by imprinting a phase step .
the deviation of the measured phase offset @xmath91 from @xmath72 is approximately half the imprinted phase step . ] in conclusion , we have shown that ring currents and solitonic structures can be generated very efficiently in toroidal trap geometries by stirring the condensate with a blue detuned laser light sheet and consequently retracting the stirring paddle .
pure ring currents can be generated this way for a wide range of transverse confinement parameters .
density notch dark and gray solitons can be generated in quasi-1d transversely confined tori . in a regime of slightly wider confinement , a novel hybrid object with properties of solitons and vortices is generated .
these solitonic vortices show a phase - offset velocity relationship similar to solitons .
we like to thank l. carr and s. mckinney at the university of washington and d. feder and j. denschlag at nist for helpful discussions and comments .
this work was partially supported by nsf chemistry and physics .
j. b. acknowledges support from the alexander von humboldt foundation through a feodor lynen research fellowship . |
|
the study of the solutions corresponding to the electrical impedance equation @xmath0 where @xmath1 represents the conductivity and @xmath2 denotes the electric potential , is fundamental for the proper understanding of a wide variety of boundary value problems , that posses special relevance in different branches of mathematical physics . among many important problems , it is particularly interesting the one corresponding to the electrical impedance tomography , because of its importance in applied physics and engineering . as a matter of fact , only taking into account the medical imaging applications , no doubts remain about its relevance . the study of this inverse problem , correctly posed in mathematical terms by a.p .
calderon @xcite , is based upon iterative methods that employ solutions of the forward problem , attempting to fulfil a certain boundary condition ( usually upcoming from physiological measurements ) , by introducing variations in the conductivity function at every step , in order to reduce the difference between the theoretical @xmath2 and the measured one ( see e.g. , the classical work @xcite ) . yet
, the mathematical complexity of ( [ int:00 ] ) posed so strong challenges , that the electrical impedance tomography remained only as an alternative medical imaging technique , seldom considered among the basic clinical applications .
as a matter of fact , the electrical impedance tomography is still considered an ill posed problem . in this direction ,
the mathematical foundations dedicated to this topic , perhaps received one of the most important contributions when v. kravchenko in 2005 @xcite , and independently k. astala and l. pivrinta in 2006 @xcite , discovered that the two - dimensional case of ( [ int:00 ] ) was completely equivalent to a special kind of vekua equation @xcite , which had been deeply studied in a variety of works , been the most important those published by l. bers @xcite and by i. vekua @xcite .
the list of novel works arising after such discovering , is as long as it is interesting .
still , it shall be pointed out that just some of them are dedicated to the engineering applications , since it is not clear how to model a wide variety of physical phenomena , in order to make them susceptible to be analysed by means the modern theory of pseudoanalytic functions ( see e.g. @xcite ) .
this work intends to make a positive contribution in this direction .
already in @xcite was posed a basic idea for interpolating values of conductivity within bounded domains , in order to obtain a certain class of analytic representations : two - dimensional separable - variables functions , one of the keys to fully applied the novel mathematical methods into engineering problems . in this work ,
a more general methodology is posed , based upon the main conjecture arising from @xcite . starting with some examples for which the conductivity functions are known in exact form , the elements of the pseudoanalytic function theory are used , for the first time , to approach solutions of the forward boundary value problem for ( [ int:00 ] ) , considering conductivity distributions upcoming from geometrical figures .
even the full set of examples is provided for a circular domain , the results can be extended to a wide variety of bounded domains .
thus , the methods provided along these pages , could well be ready for studying images corresponding already to physical experimental models .
following @xcite , let the complex - valued functions @xmath3 satisfy the condition @xmath4 where @xmath5 denotes the complex conjugation of @xmath6 : @xmath7 , and @xmath8 is the standard imaginary unit @xmath9 .
thus , any complex - valued function @xmath10 can be expressed by means of the linear combination of @xmath6 and @xmath11 : @xmath12 where @xmath13 and @xmath14 are purely real functions .
two complex functions that fulfil ( [ pre:00 ] ) shall be called a _ generating pair _ @xmath15 .
bers @xcite introduced the @xmath15-derivative of the function @xmath10 according to the expression : @xmath16 but this derivative will exist if and only if @xmath17 here @xmath18 notice that these operators are classically introduced with the factor @xmath19 , but it will result somehow more convenient to omit it int his work . introducing the functions @xmath20 the expression ( [ pre:01 ] ) of the @xmath15-derivative will turn into @xmath21 as well the condition ( [ pre:02 ] ) can be written as @xmath22 the functions defined in ( [ pre:03 ] ) are called the _
characteristic coefficients _ of the generating pair @xmath15 , and the functions @xmath10 satisfying the condition ( [ pre:05 ] ) are named @xmath15-pseudoanalytic .
indeed , the equation([pre:05 ] ) is know as the _ vekua equation _
@xcite , and in many senses is the foundation of the present work .
the following sentences were originally presented in @xcite and @xcite , and they have been slightly adapted for the purposes of this paper .
[ th:00 ] the elements of the generating pair @xmath15 are @xmath15-pseudoanalytic : @xmath23 let @xmath24 be a non - vanishing function within a bounded domain @xmath25 .
the functions @xmath26 constitute a generating pair , whose characteristic coefficients are @xmath27 let @xmath28 and @xmath29 be two generating pairs of the form ( [ pre:07 ] ) , and let their characteristic coefficients satisfy the relation @xmath30 thus the pair @xmath31 will be called a successor of the pair @xmath28 , whereas @xmath28 will be called a predecessor of @xmath31 .
let @xmath32 be a set of generating pairs , and let every @xmath33 be a successor of @xmath34 .
thus , the set @xmath35 is called a generating sequence . moreover , if there exist a number @xmath36 such that @xmath37 the generating sequence is said to be periodic , with period @xmath36 . finally ,
if @xmath38 , the generating pair @xmath39 will be embedded into the generating sequence @xmath35 .
[ th:01 ] let @xmath39 be a generating pair of the form ( [ pre:06 ] ) , and let @xmath24 be a separable - variables function : @xmath40 where @xmath41 .
thus @xmath39 is embedded into a periodic generating sequence , with period @xmath42 , such that @xmath43 when @xmath44 is even , and @xmath45 when @xmath44 is odd . moreover , if in particular @xmath46 , it is easy to see that the generating sequence in which @xmath39 is embedded will be also periodic , but it will posses period @xmath47 .
l. bers also introduced the concept of the @xmath28-integral of a complex - valued function @xmath10 .
the detailed conditions for its existence can be found in @xcite .
let @xmath28 be a generating pair of the form .
its adjoin generating pair @xmath48 is defined according to the formulas @xmath49 the @xmath28-integral of a complex - valued function @xmath10 ( when it exists @xcite ) is defined as : @xmath50 where @xmath51 is a rectifiable curve within a domain @xmath52 .
specifically , if we consider the @xmath28-integral of the @xmath28-derivative of @xmath10 , we will have : @xmath53 here @xmath54 , and @xmath55 is a fixed point in the complex plane . but according to the theorem [ th:00 ] , the @xmath28-derivative of @xmath6 and of @xmath11 vanish identically , thus the expression ( [ pre:09 ] ) can be considered as the @xmath28-antiderivative of @xmath56 .
[ def:00 ] the formal power @xmath57 belonging to the generating pair @xmath34 , with formal degree @xmath58 , complex constant coefficient @xmath59 , center at @xmath55 , and depending upon @xmath54 , is defined according to the expression : @xmath60 where @xmath61 and @xmath62 are complex constants that fulfil the condition : @xmath63 the formal powers with higher degrees are calculated according to the recursive formulas : @xmath64 where @xmath65 notice the integral operators in the right side of the last expression are @xmath34-antiderivatives .
the formal powers posses the following properties : 1 .
every @xmath66 is an @xmath34-pseudoanalytic function . 2 .
let @xmath67 , where @xmath68 and @xmath69 are real constants .
the following relation holds @xmath70 let @xmath10 be an @xmath34-pseudoanalytic function .
then it can be expressed by means of the so - called taylor series in formal powers : @xmath71 since any @xmath34-pseudoanalytic function @xmath10 accepts this expansion , this is an analytical representation for the general solution of the vekua equation ( [ pre:07 ] ) .
let us consider the two - dimensional case of the equation ( [ int:00 ] ) : @xmath72 as it has been shown in several previous works ( see e.g. @xcite and @xcite ) , if @xmath1 can be expressed by means of a separable - variables function @xmath73 introducing the notations @xmath74 the equation ( [ int:00 ] ) will turn into the vekua equation @xmath75 for which the functions @xmath76 conform a generating pair . from ( [ eie:00 ] )
it is easy to see that this pair is embedded into a generating sequence with period @xmath42 , because @xmath24 is separable - variables , according to theorem [ th:01 ] .
the possession of an explicit generating sequence , allows the construction of the formal powers ( [ pre:10 ] ) , so we can approach any solution for ( [ eie:01 ] ) , which will be closely related with the solutions of ( [ int:00 ] ) according to the relations ( [ eie:00 ] ) . as a matter of fact ,
one special and very important relation between the solutions of ( [ int:00 ] ) and of ( [ eie:01 ] ) was elegantly posed in @xcite , and this idea will play a central role in the present work .
@xcite let us consider the set of formal powers @xmath77 corresponding to the generating pair ( [ eie:02 ] ) , and let @xmath78 be a bounded domain such that @xmath79 , but @xmath80 .
then the set of functions valued on @xmath51 : @xmath81 conform a complete system for approaching solutions of the dirichlet boundary value problem corresponding to ( [ int:00 ] ) .
the last statement implies that any boundary condition @xmath82 , imposed for the solutions @xmath2 of ( [ int:00 ] ) , can be approached according to the expression : @xmath83 where @xmath84 and @xmath85 are real constants .
summarizing , when a separable - variables conductivity function @xmath1 is given within a bounded domain @xmath86 , and a boundary condition @xmath82 is imposed for the solution of ( [ int:00 ] ) , it will be always possible to construct a finite set of functions , subset of ( [ eie:03 ] ) , such that @xmath87^{2}dl<\epsilon , \label{eie:04}\ ] ] where @xmath88 and @xmath89 .
one of the main objectives of this work is to contribute into the construction of a new theory for the electrical impedance tomography problem .
hence it is natural to search for the mathematical tools that will allow us to fully apply the modern pseudoanalytic function theory into the analysis , e.g. , of medical images .
this means that it is necessary to introduce interpolation methods that , given a set of conductivity values defined into a bounded domain on the plane , can reach separable - variables functions .
one of the first approaches in this direction was posed in @xcite , and it was properly analysed in @xcite . indeed
, the last reference is completely dedicated to prove the following assessment .
[ con:00 ] @xcite let @xmath1 be a function defined within a bounded domain @xmath90 , possessing discontinuities only of the first kind .
then it is possible to approach @xmath1 by means of a piecewise separable - variables function of the form : @xmath91 .
\end{array } \right.\ ] ] @xmath92 this separable - variables function can be employed for numerically constructing a finite set of formal powers of the form ( [ eie:03 ] ) , in order to approach solutions for the dirichlet boundary value problem of the two - dimensional electrical impedance equation ( [ int:00 ] ) , in the sense of ( [ eie:04 ] ) , when a boundary condition @xmath82 is imposed .
a simplified illustration of the procedure for the construction of such piecewise function , is provided along in figure 1 , and a brief explanation of the central ideas will be provided now . for simplicity , we will consider hereafter the domain @xmath86 as the unitary circle , noticing that all postulates can be extended to a wide set of bounded domains , that are of special interest in many branches of applied mathematics , physics , and engineering .
the first step is to divide the domain @xmath86 into a number @xmath93 of subsections , as shown in the figure [ fig : eie00a ] . along a strait line crossing every subsection , parallel to the @xmath94-axis , we will collect a number @xmath95 of conductivity values , in order to introduce an interpolating function @xmath96 depending only upon the spatial variable @xmath94 .
we will posses then a set of @xmath93 interpolating functions , one for every subsection , as shown in figure [ fig : eie00b ] .
finally , we state that the conductivity function in every subregion will have the separable - variables form @xmath97 where @xmath98 is the number of the subregion , @xmath99 is the common @xmath100-coordinate of all points collected in the @xmath98-subregion , and @xmath101 is a positive real constant such that @xmath102 .
the postulate remains a conjecture because not any formal extension of the theorem posed in @xcite , about the completeness of the set ( [ eie:03 ] ) , is known for the case of piecewise separable - variables functions within bounded domains .
yet , the full set of examples presented in @xcite shows that the representation ( [ eie:05 ] ) is useful for solving the dirichlet boundary value problem of ( [ int:00 ] ) , since the numerical calculations succeed to approach the boundary conditions provided by known exact solutions , for a variety of conductivities that were both separable - variables and non separable - variables by definition .
this property will be now extended for analysing conductivities upcoming from geometrical distributions , precisely as those employed in physical measurements @xcite . to achieve this
, we will consider a limit case of the conjecture [ con:00 ] that will result specially useful for these classes of conductivities .
in behalf of simplicity , @xmath86 will be considered as the unitary circle , but it is possible to verify that the next statements can be generalized for a wide variety of bounded domains .
[ pro:00 ] every conductivity function @xmath1 , defined within a bounded domain @xmath25 , and possessing only discontinuities of the first kind , can be considered the limit case of a piecewise separable - variables conductivity function @xmath103 of the form ( [ eie:05 ] ) , when the number @xmath93 of subsections , and the number @xmath95 of collected values at every subsection , tend to infinity .
this is @xmath104 moreover , since @xmath105 where @xmath98 is the number of the subsection , it follows from theorem [ th:01 ] , that the generating sequence corresponding to this limit case , employed for numerically approaching the formal powers , will be periodic with period @xmath47 .
let us consider that @xmath86 has been sectioned into @xmath93 subdomains @xmath106 , by employing the set of equidistant @xmath94-axis parallel lines @xmath107 since @xmath86 coincides with the unitary circle , it can be described by the union of @xmath93 subdomains , defined according to the expressions
@xmath108,\ x^{2}+y^{2}\leq 1\right\rbrace . \nonumber\end{aligned}\ ] ] let us consider also the set of lines @xmath109 it is clear that @xmath110 and in consequence , for @xmath111 ; we will have that
@xmath112 it immediately follows that @xmath113 this implies that every @xmath114 will be confined into a segment of a parallel line to the @xmath94-axis .
indeed , when the @xmath115 , @xmath86 can be represented by the set of the line segments @xmath116 beside , since the number @xmath95 of conductivity values , collected along the line @xmath117 within every @xmath114 , also tends to infinite , not any interpolation method will be required for obtaining the @xmath118 functions .
they will simply coincide with the original values of the function @xmath1 over the elements of the set ( [ eie:08 ] ) .
finally , from ( [ eie:07 ] ) , it follows that @xmath119 which , according to the last sentence of the theorem [ th:01 ] , will provoke that the corresponding generating sequence , employed to numerically approach some of the formal powers ( [ eie:03 ] ) , will be periodic with period @xmath47 .
we now analyse a selected set of analytic conductivity functions for which exact solutions are known , in order to examine the effectiveness of the method posed in proposition [ pro:00 ] .
thereafter , we will consider conductivity distributions upcoming from geometrical distributions , whose analytical representation is , in general , unknown , imposing certain boundary conditions that will help us to appreciate the behaviour of the technique in this special and important cases .
a detailed description of the numerical methods used to approach the solution , that fulfil the boundary condition of every example , can be found in @xcite .
since we are considering the unitary circle , and taking into account the validity of the expression ( [ eie:04 ] ) , that as a matter of fact is a lebesgue integral - type operator , let us introduce an inner product for the elements of the finite set @xmath120 according to the formula @xmath121 where @xmath89 , and @xmath122 it follows that we can obtain a set of @xmath123 orthonormal functions @xmath124 , such that we can approach an imposed boundary condition @xmath82 according to the expression @xmath125 where @xmath126 are real constant coefficients .
notice that the apparent lose of one base function is because , by virtue of the definition [ def:00 ] , we have @xmath127 thus @xmath128 .
it should be also noticed that the orthonormalization procedure has been performed considering , first , the subset of @xmath129 functions @xmath130 [ ns:04 ] followed by the subset of @xmath131 functions @xmath132 this remark is important for adequately examining the illustrations where the absolute values of the coefficients @xmath126 are displayed .
let @xmath133 then the function @xmath134 is a particular solution of the electrical impedance equation ( [ int:00 ] ) .
this case was selected because of the variations of conductivity that take place within the unitary circle .
unfortunately , for avoiding the indetermination of the tangent functions contained into the particular solution ( [ ns:02 ] ) , we could only consider the case @xmath135 .
an illustration of this conductivity is given in the figure 2 .
within the unitary circle.,title="fig : " ] [ fig : ns00 ] the number of points located around @xmath51 , when imposing the boundary condition , does not necessarily has to coincide with the number @xmath136 of orthonormal functions , as it is shown in @xcite , where the collocation method is employed . nevertheless , on behalf of simplicity , this work will consider @xmath137 equally distributed points around the perimeter of the unitary circle , and @xmath138 base functions @xmath139 .
the boundary condition will be obtained by evaluating the solution @xmath2 , presented in ( [ ns:02 ] ) , over this set of points .
the absolute error @xmath140 is defined as the classical lebesgue norm @xmath141 where @xmath142 represents the solution ( [ ns:02 ] ) valued on the boundary @xmath51 , and @xmath143 .
the result of this integral will be approached using the standard trapezoidal method over @xmath144 equally distributed points on the segment @xmath145 $ ] . the figure [ fig : ns01a ] displays a logarithmic plot of the absolute magnitudes corresponding to every coefficient @xmath126 used for approaching the boundary condition
. we shall remember that the first @xmath146 coefficients correspond to orthonormal system obtained from the set ( [ ns:04 ] ) , whereas the remaining @xmath147 correspond to the set ( [ ns:05 ] ) .
the figure [ fig : ns01b ] displays a comparison between the imposed boundary condition ( [ ns:02 ] ) valued on @xmath51 , drew in blue , and the approached solution , in red .
it is not possible to detect any difference at first sight in this case . beside , the obtained error @xmath148 ,
thus it is possible to assess the approach is adequate .
the table [ tab : ns00 ] show the numerical values of some of the most relevant coefficients @xmath126 used in the reconstruction .
it is important to pay attention to the number @xmath136 selected , since , in general , they are not presented in consecutive order .
.[tab : ns00]values of the coefficients @xmath126 corresponding to the boundary value problem with @xmath149 . [ cols="^,^,^,^,^,^,^,^",options="header " , ]
on of the most important contributions of the present work is the full opening of the path for applying the modern theory of pseudoanalytic functions , into the analysis of wide class of conductivity functions upcoming from physical problems , by virtue of the proposition [ pro:00 ]
. tested in a variety of examples , the numerical analysis based upon the cited proposition , succeed to approach the imposed boundary conditions with considerable accuracy , presenting the highest divergence only in a kind of problem that is well known for its complexity when analysing boundary value problems for the two - dimensional electrical impedance equation .
an immediate implication of these results is the possibility of analysing most classes of images corresponding to classical applications of the electrical impedance tomography , as it is the medical imaging clinical monitoring . in this precise direction
, it is possible to assure that the search for patterns of change in the boundary electric potentials , when changes in the conductivity within the domain of interest are taking place , is completely viable by applying the techniques posed before .
it is important to remark that all presented results can be extended without mayor complications to a wide class of bounded domains , beside the unitary circle , that on behalf of simplicity was considered in these pages . indeed , most problems of mathematical physics , closely related with the modern pseudoanalytic function theory , as those elegantly studied in @xcite , as well as in @xcite , where a special case of the fokker - planck equation is posed , could well be susceptible for this kind of analysis .
a. bucio r. , r. castillo - perez , m.p .
ramirez t. ( 2011 ) , _ on the numerical construction of formal powers and their application to the electrical impedance equation _ , 8th international conference on electrical engineering , computing science and automatic control , ieee catalog number : cfp11827-art , isbn:978 - 1 - 4577 - 1013 - 1 , pp .
769 - 774 .
h. m. campos , r. castillo - perez , v. v. kravchenko ( 2011 ) , _ construction and application of bergman - type reproducing kernels for boundary and eigenvalue problems in the plane _ , complex variables and elliptic equations , 1 - 38 .
r. castillo - perez . , v. kravchenko , r. resendiz v. ( 2011 ) , _ solution of boundary value and eigenvalue problems for second order elliptic operators in the plane using pseudoanalytic formal powers _
, mathematical methods in the applied sciences , vol .
34 , issue 4 .
v. v. kravchenko ( 2005 ) , _ on the relation of pseudoanalytic function theory to the two - dimensional stationary schrdinger equation and taylor series in formal powers for its solutions _ , journal of physics a : mathematical and general , vol .
3947 - 3964 .
m. p. ramirez t. ( 2010 ) , _ on the electrical current distributions for the generalized ohm s law , applied mathematics and computation _ , elsevier ( submitted for publication ) , available in electronic at http://arxiv.org m. p. ramirez t. , r. a. hernandez - becerril , m. c. robles g. ( 2011 ) , _ first characterization of a new method for numerically solving the dirichlet problem of the two - dimensional electrical impedance equation _ , available in electronic at http://arxiv.org | using a conjecture that allows to approach separable - variables conductivity functions , the elements of the modern pseudoanalytic function theory are used , for the first time , to numerically solve the dirichlet boundary value problem of the two - dimensional electrical impedance equation , when the conductivity function arises from geometrical figures , located within bounded domains . |
most of the notation used in this paper is standard ( see @xcite or @xcite for any undefined terminology ) .
a _ perfect matching _ of a graph @xmath2 is a spanning subgraph of @xmath2 in which each vertex has degree precisely one . in this paper
we will only deal with _
cubic bridgeless _
graphs , that is graphs in which each vertex has degree 3 and such that each component is 2-edge - connected .
we are interested in the following conjecture of berge and fulkerson @xcite .
[ berge - fulkerson ] if @xmath2 is a cubic bridgeless graph , then there exist six perfect matchings of @xmath2 such that each edge of @xmath2 belongs to exactly two of them .
this conjecture is equivalent to the existence of five perfect matchings of @xmath2 such that any three of them have empty intersection .
therefore , a weaker statement is the following conjecture of fan and raspaud @xcite .
[ fan - raspaud ] if @xmath2 is a cubic bridgeless graph , then there exist three perfect matchings @xmath10 , @xmath11 and @xmath12 of @xmath2 such that @xmath13 . in this paper we will also consider the following middle step between these two conjectures .
[ 4pms ] if @xmath2 is a cubic bridgeless graph , then there exist four perfect matchings such that any three of them have empty intersection . following the notation introduced in @xcite we define @xmath14 as the maximum fraction of edges in @xmath2 that can be covered by @xmath3 perfect matchings and we define @xmath0 as the infimum of @xmath14 over all cubic bridgeless graphs @xmath2
. the second author @xcite proved that the berge - fulkerson conjecture is equivalent to the conjecture that the edge - set of every cubic bridgeless graph can be covered by 5 perfect matchings , i.e. @xmath5 .
kaiser , krl , and norine @xcite proved that the infimum @xmath15 is a minimum , attained by the petersen graph ( i.e. @xmath16 ) , and patel @xcite conjectured that the values of @xmath17 and @xmath18 are also attained by the petersen graph .
in other words : [ pat3 ] @xmath7 .
[ pat4 ] @xmath6 .
patel proved @xcite that the berge - fulkerson conjecture implies conjectures [ pat3 ] and [ pat4 ] .
these implications are summed up in figure [ fig : implicationsa ] .
( berge ) berge - fulkerson ; ; ( conj3 ) conjecture [ 4pms ] ; ( fan ) fan - raspaud ; ( m5 ) @xmath5 ; ( m4 ) @xmath6 ; ( m3 ) @xmath7 ; ( c1 ) @xcite ; ( c2 ) @xcite ; ( c3 ) @xcite ; ( berge)(conj3 ) ; ( berge)(m5 ) ; ( berge)(m4.west ) ; ( berge)(m3.west ) ; ( conj3)(fan ) ; tang , zhang , and zhu @xcite recently showed that weighted versions of conjectures [ pat3 ] and [ pat4 ] imply the fan - raspaud conjecture : for instance , they prove that if for any cubic bridgeless graph @xmath2 and any edge - weighting of @xmath2 , a weighted version of @xmath19 is at least @xmath20 , then the fan - raspaud conjecture holds . in section
[ sec : impl ] , we show that it is enough to consider only the edge - weighting where every edge has weight 1 .
more precisely , we show that @xmath6 implies conjecture [ 4pms ] and @xmath21 implies the fan - raspaud conjecture .
we also show that @xmath6 implies @xmath21 .
tang , zhang , and zhu @xcite conjectured that for any real @xmath22 , determining whether a cubic bridgeless graph @xmath2 satisfies @xmath23 is an np - complete problem . in section [ sec : np ] we prove this conjecture together with similar statements for @xmath24 and @xmath25 .
given two cubic graphs @xmath2 and @xmath26 and two edges @xmath27 in @xmath2 and @xmath28 in @xmath26 , the _ glueing _ , or _ 2-cut - connection _ , of @xmath29 and @xmath30 is the graph obtained from @xmath2 and @xmath26 by removing edges @xmath27 and @xmath28 , and connecting @xmath31 and @xmath32 by an edge , and @xmath33 and @xmath34 by an edge .
we call these two new edges the _ clone edges of @xmath27 or @xmath28 _ in the resulting graph . note that if @xmath2 and @xmath26 are cubic and bridgeless , then the resulting graph is also cubic and bridgeless . in the present paper @xmath26
will always be @xmath35 or the petersen graph , which are both arc - transitive ( for any two pairs of adjacent vertices @xmath36 and @xmath37 , there is an automorphism that maps @xmath38 to @xmath39 and @xmath40 to @xmath41 ) . in this case
the choice of @xmath28 and the order of each pair @xmath42 and @xmath43 are not relevant , so we simply say that _ we glue @xmath26 on the edge @xmath27 of @xmath2_. in what follows , we will need to glue several graphs on each edge of a given graph @xmath2 .
this has to be understood as follows : given copies @xmath44 ( @xmath45 ) of @xmath35 or the petersen graph , _ glueing @xmath44 on the edge @xmath46 of @xmath2 _ means glueing @xmath47 on some clone edge of @xmath46 in the glueing of @xmath48 on the edge @xmath46 of @xmath2 ( see figure [ fig : example ] ) . is obtained by glueing 3 copies of @xmath35 on the edge @xmath46 of @xmath2 .
the clone edges of @xmath46 in @xmath49 are drawn as thick edges.,width=302 ] note that each perfect matching in the graph @xmath49 resulting from the glueing of @xmath44 ( @xmath50 ) on some edge @xmath46 of @xmath2 either contains all clone edges of @xmath46 , or none of them ( since each pair of such edges forms a 2-edge - cut in @xmath49 , and the intersection of every perfect matching with an edge - cut has the same parity as the edge - cut ) .
it follows that each perfect matching @xmath51 of @xmath49 corresponds to a perfect matching @xmath52 of @xmath2 and perfect matchings @xmath53 of @xmath54 , for each @xmath55 .
we call each of these perfect matchings the _ restriction _ of @xmath52 to @xmath56 , respectively .
the following is a well - known property of the petersen graph .
[ lem : petersen ] the petersen graph @xmath57 has exactly 6 distinct perfect matchings . moreover the following properties hold : * every edge of @xmath57 is covered by exactly two distinct perfect matchings ; * every two distinct perfect matchings of @xmath57 intersect in exactly one edge ( and therefore cover exactly 9 edges ) ; * every three distinct perfect matchings of @xmath57 cover exactly 12 edges ; * every four distinct perfect matchings of @xmath57 cover exactly 14 edges .
it was proved in @xcite that a fractional version of conjecture [ pat3 ] ( @xmath7 ) implies the fan - raspaud conjecture .
we strengthen this result by showing that not only the integral conjecture itself implies the fan - raspaud conjecture , but it implies an even stronger statement ( which will be needed in what follows ) . [ the : m3_fr ] conjecture [ pat3 ] implies that any cubic bridgeless graph @xmath2 has three perfect matchings @xmath58 such that @xmath59 and @xmath60 . in particular , conjecture [ pat3 ] implies the fan - raspaud conjecture .
let @xmath2 be a cubic bridgeless graph , and let @xmath49 be the graph obtained by glueing @xmath61 copies of the petersen graph on each edge of @xmath2 ( see figure [ fig : glueingp ] ) .
the number of edges of @xmath49 can be easily computed as @xmath62 .
if conjecture [ pat3 ] holds , then @xmath49 has a set @xmath63 of perfect matchings such that @xmath64 contains at least @xmath65 edges of @xmath49 .
let @xmath66 be the restriction of @xmath67 to @xmath2 .
we now prove that @xmath60 and @xmath59 .
is obtained by glueing multiple copies of the petersen graph on each edge of @xmath2 .
the thick edges are the clone edges of @xmath46 in @xmath49.,width=340 ] note that by lemma [ lem : petersen ] , at most 12 edges are covered by the restriction of @xmath67 to each of the @xmath68 copies of the petersen graph glued on the edges of @xmath2 . since @xmath64 contains at least @xmath69 edges of @xmath49 , it follows that @xmath60 . by lemma [ lem : petersen ] , for each edge @xmath46 of @xmath2 lying in @xmath70 , at most 9 edges are covered by the restriction of @xmath67 to each of the @xmath61 copies of the petersen graph glued on @xmath46 .
if @xmath71 , this implies that @xmath64 contains at most @xmath72 edges , a contradiction .
let @xmath73 be a set of perfect matchings of a graph @xmath2 . for @xmath74
, we define the _ depth _ of @xmath46 in @xmath73 , denoted by @xmath75 , as the number of perfect matchings of @xmath76 containing @xmath46 .
the _ edge - depth _ of @xmath76 , denoted by @xmath77 , is the maximum depth of an edge of @xmath2 in @xmath76 .
the same proof as that of theorem [ the : m3_fr ] , considering four perfect matchings instead of three , shows the following connection between conjecture [ pat4 ] ( @xmath6 ) and conjecture [ 4pms ] .
[ the : m4_4pms ] conjecture [ pat4 ] implies that any cubic bridgeless graph @xmath2 has a set @xmath78 of four perfect matchings such that @xmath79 and @xmath80 .
in particular , conjecture [ pat4 ] implies conjecture [ 4pms ] .
we now use theorem [ the : m4_4pms ] to show that @xmath6 implies @xmath7 .
[ the:141545 ] conjecture [ pat4 ] implies conjecture [ pat3 ] .
assume that conjecture [ pat4 ] holds and let @xmath2 be any cubic bridgeless graph .
by theorem [ the : m4_4pms ] , @xmath2 has a set @xmath81 of four perfect matchings such that @xmath79 and @xmath82 . for @xmath83 , let @xmath84 be the fraction of edges of @xmath2 covered precisely @xmath85 times .
consider the set @xmath86 of edges covered exactly once by @xmath76 , and remove from @xmath87 the perfect matching , say @xmath10 , containing the smallest number of edges of @xmath86 .
then the fraction of edges of @xmath2 covered by @xmath88 is at least @xmath89 .
since @xmath90 and every perfect matching contains a third of the edges , we have @xmath91 . combining this with the assumption that @xmath92
, we obtain @xmath93 , which concludes the proof .
( berge ) berge - fulkerson ; ; ( conj3 ) conjecture [ 4pms ] ; ( fan ) fan - raspaud ; ( m5 ) @xmath5 ; ( m4 ) @xmath6 ; ( m3 ) @xmath7 ; ( c1 ) theorem [ the : m4_4pms ] ; ( c2 ) theorem [ the : m3_fr ] ; ( c3 ) theorem [ the:141545 ] ; ( berge)(conj3 ) ; ( berge)(m5 ) ; ( m5)(m4 ) ; ( m4)(m3 ) ; ( conj3)(fan ) ; ( m4)(conj3 ) ; ( m3)(fan ) ; we summarize in figure [ fig : implicationsb ] the implications which follow from the results of this section .
an important remark is that in order to prove that a given cubic bridgeless graph @xmath26 satisfies @xmath94 , in theorem [ the:141545 ] we really use the assumption that @xmath95 _ for every cubic bridgeless graph @xmath2_. we do not know how to prove the stronger statement that if @xmath96 , then @xmath97 . in the remainder of this section
, we provide weaker bounds relating @xmath98 and @xmath99 , for any @xmath45 . if @xmath2 is a cubic bridgeless graph , then @xmath100 .
let @xmath58 be perfect matchings of @xmath2 covering a fraction of @xmath19 of the edges of @xmath2 .
for @xmath83 , let @xmath84 be the fraction of edges of @xmath2 covered precisely @xmath85 times by @xmath58 .
we have @xmath101 @xmath102 and @xmath103 @xmath104 .
the combination @xmath105 gives @xmath106 .
we remove from @xmath58 the perfect matching , say @xmath10 , containing the least number of edges covered precisely once by @xmath58 .
it follows that the fraction of edges of @xmath2 covered by @xmath107 is at least @xmath108 .
[ th : mk ] if @xmath2 is a cubic bridgeless graph and @xmath109 is an integer , then @xmath110 .
let @xmath111 be perfect matchings of @xmath2 covering a fraction of @xmath98 of the edges of @xmath2 .
we say that a vertex of @xmath2 has type @xmath112 if the three edges incident to @xmath34 are covered @xmath31 , @xmath33 , and @xmath113 times ( respectively ) by @xmath111 .
let @xmath114 be the fractions of vertices of type @xmath115 , @xmath116 , @xmath117 , @xmath118 , @xmath119 , and @xmath120 , respectively .
any vertex of @xmath2 is of one of these six types , so we have @xmath101 @xmath121 .
let @xmath122 and @xmath123 denote the number of vertices and edges of @xmath2 , respectively .
then by definition of @xmath98 , the number of edges covered by @xmath111 is @xmath124 . since @xmath2 is cubic , @xmath125 and so , @xmath103 @xmath126 .
similarly , the fraction of edges covered precisely once by @xmath111 is equal to @xmath127 , so we can remove one of the perfect matchings , say @xmath10 , in such way that @xmath128 cover a fraction of at least @xmath129 of the edges of @xmath2 .
note that the combination @xmath130 gives @xmath131 , so it follows that @xmath110 .
in particular , this shows that any cubic bridgeless graph @xmath2 whose edge - set can be covered by 4 perfect matchings satisfies @xmath132 .
it follows that the conjecture stating that every cubic bridgeless graph @xmath2 satisfies @xmath97 only needs to be verified for graphs whose edge - set can not be covered by 4 perfect matchings ( some results on this class of graphs can be found in @xcite and @xcite ) .
recall that kaiser , krl , and norine @xcite proved that every cubic bridgeless graph @xmath2 satisfies @xmath133 .
an interesting problem is to characterize graphs for which equality holds ( the petersen graph is an example ) .
the next theorem implies that , again , these graphs are such that their edge - set can not be covered by 4 perfect matchings ( indeed , if the edge - set of a cubic bridgeless graph @xmath2 can be covered by 4 perfect matchings , then @xmath134 , and theorem [ th : m2m4 ] below implies that @xmath135 ) .
[ th : m2m4 ] if @xmath2 is a cubic bridgeless graph , then @xmath136 .
let @xmath137 be perfect matchings of @xmath2 covering a fraction of @xmath25 of the edges of @xmath2 .
let @xmath114 be the fractions of vertices defined in the proof of theorem [ th : mk ] , with @xmath138 . observe that @xmath139 , thus we have @xmath101 @xmath140 and @xmath103 @xmath141 .
let @xmath84 be the fraction of edges of @xmath2 covered precisely @xmath85 times by @xmath142 .
note that after the removal of two perfect matchings @xmath53 and @xmath143 ( @xmath144 ) from @xmath137 , the edges that were covered only by @xmath53 , or only by @xmath143 , or only by @xmath53 and @xmath143 are not covered anymore .
if we sum the fractions of edges of these three types , for any of the six pair @xmath145 we obtain @xmath146 ( every edge covered exactly once is counted three times ) .
note that @xmath147 and @xmath148 , so @xmath149 . thus we can choose a pair @xmath145 , such that after the removal of @xmath53 and @xmath143 from @xmath137 , a fraction of at least @xmath150 of the edges of @xmath2 is still covered .
the combination @xmath151 gives @xmath152 , so it follows that @xmath153 .
a cubic bridgeless graph @xmath2 satisfies @xmath154 if and only if it is 3-edge - colorable , and deciding this is a well - known np - complete problem ( see @xcite ) .
it was proved in @xcite that deciding whether a cubic bridgeless graph @xmath2 satisfies @xmath134 is also an np - complete problem . in this section , we prove that determining whether @xmath24 , @xmath19 , and @xmath25 are more than any given constant ( lying in some appropriate interval ) is also an np - complete problem . in the case of @xmath19 ( see theorem [ th : m3 ] ) , this solves a conjecture of tang , zhang , and zhu @xcite .
[ th : m2 ] for any constant @xmath155 , deciding whether a cubic bridgeless graph @xmath2 satisfies @xmath156 ( resp .
@xmath157 ) is an np - complete problem .
the proof proceeds by reduction from the 3-edge - colorability of cubic bridgeless graphs , which is a well - known np - complete problem @xcite .
note that our problem is clearly in np , since any set of two perfect matchings whose union covers a fraction of more than ( resp .
at least ) @xmath9 of the edges is a certificate that can be checked in polynomial time . for any cubic bridgeless graph @xmath2 with @xmath123 edges ,
we construct ( in polynomial time ) a cubic bridgeless graph @xmath49 ( of size polynomial in @xmath123 ) , such that @xmath2 is 3-edge - colorable if and only if @xmath158 ( resp .
@xmath159 ) .
let @xmath160 and @xmath161 .
it can be checked that @xmath162 the graph @xmath49 is obtained from @xmath2 by glueing @xmath163 copies of @xmath35 and @xmath164 copies of the petersen graph on every edge of @xmath2 .
note that @xmath49 has precisely @xmath165 edges .
assume first that @xmath2 is 3-edge - colorable .
then @xmath2 has two perfect matchings @xmath10 and @xmath11 such that @xmath166 .
we construct two perfect matchings @xmath167 and @xmath168 of @xmath49 by combining @xmath10 and @xmath11 , respectively , with perfect matchings of each copy of @xmath35 and the petersen graph glued on the edges of @xmath2 . since no edge of @xmath2 is contained both in @xmath10 and @xmath11 , these two perfect matchings can be combined with perfect matchings covering precisely 4 edges of each copy of @xmath35 ( see figure [ fig : m2 ] , left and center ) and 9 edges of each copy of the petersen graph ( by lemma [ lem : petersen ] ) .
it follows that @xmath49 contains two perfect matchings @xmath167 and @xmath168 covering @xmath169 edges of @xmath49 .
therefore , we have @xmath170 , as desired
. glued on some edge of @xmath2 .
edges are labelled @xmath85 if they are covered by @xmath171 in @xmath49.,width=453 ] assume now that @xmath2 is not 3-edge - colorable , and take any two perfect matchings @xmath167 and @xmath168 of @xmath49 .
let @xmath10 and @xmath11 be the restriction of @xmath167 and @xmath168 to @xmath2 , respectively . since @xmath2 is not 3-edge - colorable , @xmath10 and @xmath11 have non - empty intersection , so @xmath2 has an edge @xmath46 lying both in @xmath10 and @xmath11 .
we consider the union of the restriction of @xmath167 and the restriction of @xmath168 to each copy of @xmath35 or the petersen graph glued on the edges of @xmath2 .
as before , this union covers at most 4 edges in each copy of @xmath35 , and at most 9 edges in each copy of the petersen graph .
however , it can be checked that since @xmath46 is contained in @xmath172 , at most 2 edges can be covered in each copy of @xmath35 glued on @xmath46 ( see figure [ fig : m2 ] , right ) .
it follows that @xmath173 covers at most @xmath174 edges of @xmath49 .
therefore , @xmath175 , as desired .
[ th : m3 ] for any constant @xmath176 , deciding whether a cubic bridgeless graph @xmath2 satisfies @xmath177 ( resp .
@xmath23 ) is an np - complete problem .
the proof follows the lines of the proof of theorem [ th : m2 ] . here
we define @xmath178 , where @xmath123 is the number of edges of @xmath2 , and @xmath179
. it can be checked that @xmath180 the graph @xmath2 is 3-edge - colorable if and only if it contains 3 perfect matchings with pairwise empty intersection .
it follows that as above , if @xmath2 is 3-edge - colorable we can find 3 perfect matchings of @xmath49 such that their restrictions cover 6 edges in each copy of @xmath35 ( see figure [ fig : m3 ] , left ) and @xmath181 edges in each copy of the petersen graph ( by lemma [ lem : petersen ] ) glued on the edges of @xmath2 .
glued on some edge of @xmath2 .
edges are labelled @xmath85 if they are covered by @xmath171 in @xmath49.,width=302 ] if @xmath2 is not 3-edge - colorable , then for any 3 perfect matchings of @xmath49 there is an edge of @xmath2 covered at least twice by their restrictions to @xmath2 .
this implies that at most 4 edges are covered in each copy of @xmath35 glued on this edge ( see figure [ fig : m3 ] , right ) and the result follows .
[ th : m4 ] for any constant @xmath182 , deciding whether a cubic bridgeless graph @xmath2 satisfies @xmath183 ( resp .
@xmath184 ) is an np - complete problem . instead of reducing from 3-edge - colorability , we have to make a small modification here .
the proof proceeds by reduction from the problem of deciding whether a cubic bridgeless graph @xmath2 satisfies @xmath134 , which is an np - complete problem @xcite .
we define @xmath160 , where @xmath123 is the number of edges of @xmath2 , and @xmath185
. it can be checked that @xmath186 as above , given a cubic bridgeless graph @xmath2 , we construct @xmath49 by glueing @xmath163 copies of @xmath35 and @xmath164 copies of the petersen graph on every edge of @xmath2 .
if @xmath134 , then @xmath2 contains 4 perfect matchings such that any edge of @xmath2 is covered by at least 1 of the 4 perfect matchings , and avoided by at least 2 of the 4 perfect matchings .
it follows that @xmath49 contains 4 perfect matchings such that their restrictions cover 6 edges in each copy of @xmath35 ( see figure [ fig : m4 ] , left and center ) and @xmath187 edges in each copy of the petersen graph ( by lemma [ lem : petersen ] ) glued on the edges of @xmath2 .
glued on some edge of @xmath2 .
edges are labelled @xmath85 if they are covered by @xmath171 in @xmath49.,width=453 ] if @xmath188 , then for any 4 perfect matchings of @xmath49 there is an edge of @xmath2 avoided by each of their restrictions to @xmath2 .
this implies that at most 4 edges are covered in of each copy of @xmath35 glued on this edge ( see figure [ fig : m4 ] , right ) and the result follows .
recall that the berge - fulkerson conjecture is equivalent to @xmath5 ( see @xcite ) , so if this conjecture is true we can not prove hardness results similar to theorems [ th : m2 ] , [ th : m3 ] , and [ th : m4 ] , for @xmath189 when @xmath190 .
let @xmath191 be the family of cubic bridgeless graphs @xmath2 for which @xmath192 .
a nice problem left open by the results in the previous section is the following . by theorem
[ th : m2m4 ] , the edge - set of a graph of @xmath191 can not be covered by 4 perfect matchings . using arguments similar to that of @xcite
, it can be proved that any graph @xmath194 has a set @xmath76 of at least three perfect matchings , such that for any @xmath195 there is a set @xmath196 of at least three perfect matchings of @xmath2 satisfying the following : for any @xmath197 , @xmath198 .
however , this necessary condition is not sufficient : it is not difficult to show that it is satisfied by the dodecahedron and by certain families of snarks . similarly
, we do not know if it can be decided in polynomial time whether a cubic bridgeless graph @xmath2 satisfies @xmath200 ( resp .
@xmath96 ) , but the questions seem to be significantly harder than for @xmath15 ( since it is not known whether @xmath7 and @xmath6 ) .
t. kaiser , d. krl and s. norine , _ unions of perfect matching in cubic graphs _ , in `` topics in discrete mathematics '' ( m. klazar , j. kratochvl , j. matousek , r. thomas , p. valtr , eds . ) , springer , 2006 , 225230 . | a conjecture of berge and fulkerson ( 1971 ) states that every cubic bridgeless graph contains 6 perfect matchings covering each edge precisely twice , which easily implies that every cubic bridgeless graph has three perfect matchings with empty intersection ( this weaker statement was conjectured by fan and raspaud in 1994 ) .
let @xmath0 be the supremum of all reals @xmath1 such that for every cubic bridgeless graph @xmath2 , there exist @xmath3 perfect matchings of @xmath2 covering a fraction of at least @xmath4 of the edges of @xmath2 .
it is known that the berge - fulkerson conjecture is equivalent to the statement that @xmath5 , and implies that @xmath6 and @xmath7 . in the first part of this paper , we show that @xmath6 implies @xmath7 , and @xmath7 implies the fan - raspaud conjecture , strengthening a recent result of tang , zhang , and zhu . in the second part of the paper , we prove that for any @xmath8 and for any real @xmath9 lying in some appropriate interval , deciding whether a fraction of more than ( resp .
at least ) @xmath9 of the edges of a given cubic bridgeless graph can be covered by @xmath3 perfect matching is an np - complete problem .
this resolves a conjecture of tang , zhang , and zhu .
[ theorem]lemma [ theorem]definition [ theorem]example [ theorem]corollary [ theorem]conjecture [ theorem]observation [ theorem]problem _ keywords : berge - fulkerson conjecture , perfect matchings , cubic graphs .
+ msc(2010):05c15 ( 05c70 ) _ |
neutral hydrogen in the intergalactic medium ( igm ) produces a plethora of absorption lines in the spectra of high redshift quasars .
numerical n body / hydrodynamics simulations of structure formation in the igm , within the framework of cold dark matter ( cdm ) dominated cosmologies ( cen 1994 ; zhang , anninos , & norman 1995 ; miralda escud 1996 ; hernquist 1996 ; zhang 1998 ; theuns 1998a , b ; dav 1999 ; bryan 1999 ) , have recently provided a definite picture for the origin of the forest , one of an interconnected network of sheets and filaments with virialized systems ( halos ) located at their points of intersection . at @xmath7 , the lowest column density absorbers ( @xmath8 ) arise in this context in the underdense ( @xmath9 ) minivoids between the filaments , absorbers that give rise to moderate column density lines ( @xmath10 ) correspond to modestly overdense ( @xmath11 ) filaments , and high column density absorbers ( @xmath12 ) arise in highly overdense ( @xmath13 ) structures at the intersection of filaments ( zhang 1998 ) .
most simulations to date assume that the igm is photoionized and photoheated by a uv background close to that inferred from quasars ( as computed by haardt & madau 1996 , hereafter hm96 ) .
the overall normalization of the column density distribution depends approximately on the parameter @xmath14 , where @xmath15 is the hydrogen ionization rate , @xmath16 is the hubble constant in units of @xmath17 , and @xmath18 is the baryon density parameter .
the width of the lines , as measured by the @xmath19 parameter of a voigt profile , is set by thermal broadening , peculiar velocities , and hubble expansion across the filaments .
while the first simulations showed good agreement with the observed line statistics , recent detailed studies at higher numerical resolution have revealed a serious conflict with the data : the models predict median values for the @xmath19 distribution of about @xmath20 at @xmath21 , compared with the observed median of @xmath22 ( theuns 1998b ; bryan 1999 ) .
one possible way to restore the agreement with observations is to increase the temperature of the igm , adding to the thermal broadening .
as theuns ( 1999 ) have pointed out , this may be achieved by increasing the baryon density @xmath23 , or by heating the gas for longer by increasing the age of the universe ( e.g. by lowering the total matter density @xmath24 ) .
however , the gas temperature at late times is determined ( approximately ) by the balance between adiabatic cooling and photoelectric heating and varies as @xmath25 ( hui & gnedin 1997 ) .
thus the needed increase in temperature ( a factor of 2 ) is difficult to accomplish without conflicting with big bang nucleosynthesis for reasonable values of @xmath24 and @xmath16 ( theuns 1999 and 3 of this paper ) . delaying the reionization of helium
may also help ( haehnelt & steinmetz 1998 ) , but this appears to boost the temperature by only a small factor .
other mechanisms that could increase the line
width include photoelectric dust heating ( nath , sethi , & shchekinov 1999 ) and radiative transfer effects during the reionization of helium ( miralda escud & rees 1994 ; abel & haehnelt 1999 ) . in this _
letter _ we point out that a potentially important heating term has been neglected in all previous numerical studies , namely compton heating of electrons by hard x ray background photons .
we show that this effect will dominate over photoelectric heating of hydrogen at redshifts @xmath26 in popular models for the redshift evolution of the extragalactic background radiation and that it will drive the igm towards a temperature of about @xmath27 k at @xmath6 .
in an igm that interacts solely with a background of uv ionizing photons ( assumed to originate from quasars and/or star forming galaxies ) , the dominant heat input is due to photoelectric heating of hydrogen and helium atoms .
the energy input from hydrogen photoionization is given by @xmath28 where @xmath29 is the mean neutral hydrogen density , @xmath30 the specific intensity of the uv radiation background , @xmath31 the photoionization cross section , and @xmath32 the energy of the lyman edge .
equation ( [ photo ] ) can be integrated to give @xmath33 where @xmath34 is the ( proper ) energy density of the metagalactic flux . in a quasar
dominated background , photoheating by will be larger than @xmath35 by a factor 1.52 .
the main cooling processes are compton cooling against the microwave background at high redshifts and adiabatic cooling at redshifts @xmath36 .
allow now the intergalactic gas to exchange energy with the xrb through compton heating at a rate @xmath37 where @xmath38 is the mean electron density , @xmath39 the specific intensity of the xrb , and @xmath40 the thomson cross section .
energy is transferred from photons to electrons when @xmath41 , and in this limit equation ( [ compt ] ) becomes @xmath42 where @xmath43 is the mean photon energy and @xmath44 the energy density of the xrb . for photons above 100 kev relativistic effects
become significant , and compton scattering takes place in the klein
nishina regime .
the relativistic generalization of equation ( [ compt ] ) can be written as ( cf .
blumenthal 1974 ) @xmath45 \label{comptkn}\ ] ] ( for @xmath46 ) , where @xmath47 is the photon energy in units of @xmath4 .
all the energy of the primary electrons will be deposited into the medium as heat ( through coulomb collisions ) , since energy losses due to collisional excitation and ionization of h and he are negligible in such a highly ionized igm ( e.g. shull & van steenberg 1985 ) . the xrb spectrum is characterized by a narrow peak in spectral power , @xmath48 , at 30 kev ; a simple analytical fit to the observed energy flux in units of @xmath49 yields @xmath50 from 3 to 60 kev ( boldt 1987 ) . at higher energies , the _ heao _ 1 a4 data can be fitted with a power
law , @xmath51 from 80 to 400 kev ( kinzer 1997 ) .
the preliminary diffuse @xmath52ray spectrum from 800 kev to 30 mev measured with comptel ( kappadath 1996 ) obeys another power
law , @xmath53 ( kribs & rothstein 1997 ) .
the total energy density in the 0.00330 mev band is @xmath54 .
the mean , `` effective '' ( as it includes a large correction due to klein
nishina effects ) photon energy is @xmath55 kev . note that , because of the decline at high energies of the klein
nishina cross section , the ( x@xmath56)ray heating rate is numerically equivalent to the rate one would derive by assuming pure thomson scattering of photons in the range 3180 kev only . with the adopted spectrum , about half of the total energy exchange rate at the present epoch is provided by photons above 110 kev and about 14% is contributed by photons above 1 mev .
we can now estimate the relative importance of compton versus photoelectric heating of hydrogen .
the qso contribution to the local metagalactic flux at the lyman edge is @xmath57 @xmath58 ( hm96 ) . assuming a spectral index of @xmath59 , from equations ( [ photo ] ) and ( [ compt ] ) we get @xmath60 at @xmath61 .
the equation of ionization equilibrium for an igm with mean hydrogen density @xmath62 @xmath63 @xmath64 and temperature @xmath65 k gives @xmath66 , hence @xmath67 . at the present
epoch , photoelectric heating is then the dominant contribution . at @xmath68 , the uv background from optically
selected quasars increases to @xmath69 @xmath58 ( hm96 ) , the gas temperature reaches @xmath70k or higher , and ionization equilibrium now yields @xmath71 . if the bulk of the xrb is produced at high redshifts , @xmath44 increases as @xmath72 , and the rate of gain in thermal energy per baryon due to compton scattering goes up as @xmath73 , where @xmath74 is less than unity because klein
nishina corrections become more important at high redshifts .
the same exercise then yields @xmath75 at @xmath68 .
the relative strength of the two competing heating terms will obviously depend on the redshift evolution of ionizing and x ray extragalactic radiation .
two cosmological effects may tend to make compton scattering of hard x ray photons the dominant source of heating at high redshifts in a highly ionized , low density igm : * at fixed ionization state and energy density ratio @xmath76 , the ratio @xmath77 increases with @xmath43 as @xmath78 , where @xmath74 for @xmath79 ; * at fixed ratio between the quasar uv and x ray volume emissivities , the radiation energy density @xmath44 increases with redshift relative to @xmath34 , as the mean free path of photons at @xmath80 becomes very small , sources at higher redshifts are severely attenuated , and the uv background is largely `` local '' beyond a redshift of @xmath81 ( e.g. madau 1992 ) . in the absence of any other cooling or heating mechanisms , compton cooling off cosmic microwave background ( cmb ) photons and compton heating by the xrb will drive the thermal state of the igm towards the compton temperature @xmath82 independently of the electron density . at @xmath61 , @xmath83 k.
in this section we will calculate numerically the thermal history of a uniform igm of primordial composition within a cosmological context .
the code that we use includes the relevant cooling and heating processes ( as described in theuns 1998b , and references therein ) and follows the non equilibrium evolution of hydrogen and helium ionic species .
the gas is allowed to interact with the cmb through compton cooling , with the xrb through compton heating , and with the time dependent qso ionizing background as computed by hm96 .
it is widely believed that the diffuse xrb is due to the summed emission from agns ( see fabian & barcons 1992 for a review ) .
there is insufficient information , however , to uniquely infer from the data the evolution with cosmic time of the hard x ray volume emissivity . according to the findings of boyle ( 1994 ) from _ rosat _ data
( see also page 1996 ) , the model that best fits the cosmological evolution of soft x ray selected agns in an einstein
desitter universe is one in which the comoving volume emissivity increases as @xmath84 for @xmath85 ( in qualitative agreement with the evolution of the qso optical luminosity function ) , and remains roughly constant at higher redshifts , up to @xmath86 ( miyaji , hasinger , & schmidt 1998 ) . assuming that these simple prescriptions also describe the evolution of hard x ray sources , the energy density of the xrb then increases approximately as @xmath87 at low redshifts , and as @xmath88 at @xmath89 . to account for the existence of a high redshift cut
off , we will parameterize the evolution of the x ray energy density as @xmath90 figure 1 shows the ensuing compton heating rate per baryon as a function of redshift for @xmath91 , compared with the photoionization heating rate from optically selected quasars .
figure 2 shows the thermal history of an igm with @xmath92 ( burles and tytler 1998 ) for two cosmological models : an einstein
desitter universe and a spatially flat universe dominated by a cosmological constant with parameters close to those inferred from type ia supernovae and cosmic microwave background anisotropies ( see efstathiou 1999 , and references therein ) . in both cases , we assume a photoionizing background as computed by hm96 and a hubble constant of @xmath93 ( see freedman 1998 ) .
the dotted lines show the temperature evolution if the x ray background is ignored , as has been assumed in all previous work .
the solid lines include compton heating by the xrb according to the model of equation ( [ ux ] ) . including the x ray background with a redshift cut
off parameter of @xmath91 causes the temperature of the igm to rise to @xmath94k at @xmath95 and @xmath96k at @xmath21 .
the igm temperatures are even higher in the @xmath97dominated model .
these results are extremely insensitive to the amplitude of the photoionizing background and relatively weakly dependent on @xmath98 and @xmath16 .
they are , however , sensitive to the assumed redshift evolution of the x ray background and to the hard x ray spectrum at photon energies above @xmath99 kev , which is poorly known but dominate the heating rate .
the results of figure 2 and the numerical simulations of theuns ( 1999 ) suggest that heating of the igm by the hard xrb may well be able to account for the @xmath19parameter distribution of the forest .
furthermore , our results suggest that the observed @xmath19parameter distribution may provide a useful probe of the evolution of x ray diffuse radiation at high redshift .
compton heating will also tend to flatten the equation of state of the intergalactic gas , since the compton heating per unit mass is independent of density .
however , for our model with @xmath91 , this effect is small , even for low density regions .
we have shown that compton heating of electrons by hard x ray background photons provides an important heating source of the igm . for plausible parameters ,
heating by the xrb will be comparable to photoelectric heating at redshifts @xmath26 .
this effect has been ignored in most numerical simulations of the forest , yet it can have a significant impact on the temperature of the igm , raising it to temperatures as high as @xmath100k at redshifts @xmath101 .
the increase in igm temperature caused by this effect may explain why the @xmath19parameter distributions derived from numerical simulations ( theuns 1998b ; bryan 1999 ) are narrower than those observed .
theuns ( 1999 ) have previously suggested that the temperature of the igm can be raised by invoking a high baryon density in a universe with low total matter density .
however , the density required , @xmath102 , is well above the value implied by primordial nucleosynthesis and the measurements of the deuterium abundances from quasar spectra ( @xmath103 , burles and tytler 1998 ) .
compton heating by hard x ray photons may help to resolve this discrepancy .
the origin and main production epoch of the xrb are still largely unsolved problems .
the @xmath19parameter distribution of the lines could provide important information on the evolution of x ray background radiation at high redshifts .
finally , it is interesting to note that , as another consequence of compton heating of the igm , there may exist an x ray proximity effect , in analogy with the usual uv proximity effect ( e.g. bechtold 1994 and references therein ) .
this could result in a hotter igm , and consequently broader lines , in the immediate vicinity of a long lived x ray emitting quasar ( lifetime @xmath104 the compton heating timescale ) .
we have benefited from useful discussions with g. bryan , a. meiksin , j. miralda escud , m. rees , and t. theuns .
we are especially grateful to jordi miralda
escud for pointing out the importance of relativistic corrections at high redshift .
support for this work was provided by nasa through atp grant nag54236 , and by nsf through grant phy9407194(pm ) .
gpe thanks pparc for the award of a senior fellowship . | high resolution hydrodynamics simulations of the forest in cold dark matter dominated cosmologies appear to predict line
widths that are substantially narrower than those observed . here
we point out that compton heating of the intergalactic gas by the hard x ray background ( xrb ) , an effect neglected in all previous investigations , may help to resolve this discrepancy .
the rate of gain in thermal energy by compton scattering will dominate over the energy input from hydrogen photoionization if the xrb energy density is @xmath0 times higher than the energy density of the uv background at a given epoch , where @xmath1 is the hydrogen neutral fraction in units of @xmath2 and @xmath3 is the mean x ray photon energy in units of @xmath4 .
the numerical integration of the time
dependent rate equations shows that the intergalactic medium approaches a temperature of about @xmath5 k at @xmath6 in popular models for the redshift evolution of the extragalactic background radiation .
the importance of compton heating can be tested experimentally by measuring the line width distribution as a function of redshift , thus the forest may provide a useful probe of the evolution of the xrb at high redshifts . # 1to 0pt#1 |
[ [ section ] ] the riemann moduli space @xmath1 of surfaces of genus @xmath2 with @xmath3 marked points has become a central object in mathematical physics . introduced in algebraic geometry , there is a natural compactification @xmath4 of these spaces ; their importance was emphasized by grothendieck in his famous _
esquisse dun programme _
@xcite . the special case when @xmath5 , the space @xmath6 of @xmath3 punctures on the sphere @xmath7 , is a building block leading to higher genera .
they play a crucial role in the theory of gromov - witten invariants and symplectic geometry .
furthermore , they appear in the work of kontsevich on quantum cohomology , and are closely related to the operads of homotopy theory .
we look at the _ real _ points @xmath0 of this space ; these are the set of fixed points under complex conjugation
. this moduli space will provide the motivation to study certain combinatorial objects .
the real _ deligne - mumford - knudsen _ moduli space @xmath0 is a compactification of the configurations of @xmath3 labeled points on @xmath8 quotiented by the action of @xmath9 .
@xmath0 is a manifold without boundary of real dimension @xmath10 .
the moduli space is a point when @xmath11 and @xmath8 when @xmath12 . for @xmath13 ,
these spaces become non - orientable .
figure [ m05 ] shows @xmath14 ( shaded ) as the connected sum of five copies of @xmath15 , with the four hexagonal ` boundaries ' and the outer circle carrying antipodal action . in general
, @xmath0 can be described as the iterated blow - ups of real projective spaces ( see @xcite , @xcite ) , getting extremely complicated as @xmath3 increases .
the diagram shown in figure [ m05 ] is first found in the work of brahana and coble in @xmath16 @xcite , relating to possibilities of maps with twelve five - sided countries .
[ [ section-1 ] ] a beautiful fact about the moduli space is its tiling by the convex polytope known as the associahedron @xmath17 .
it is the geometric realization of the poset of all meaningful bracketings on @xmath3 variables .
originally defined for use in homotopy theory by stasheff @xcite , it was later given a convex polytope realization by lee @xcite .
we use an alternative definition : the _ associahedron _
@xmath17 is a convex polytope of dim @xmath18 with codimension @xmath19 faces corresponding to using @xmath19 sets of non - intersecting diagonals on an @xmath20-gon .
the associahedron @xmath21 is a point .
figure [ k3k4k5 ] shows the associahedra @xmath22 , @xmath23 , and @xmath24 .
for instance , the polytope @xmath23 is a pentagon whose sides ( codim one faces ) are labeled with 5-gons with one diagonal and whose vertices ( codim two faces ) are labeled with 5-gons with two diagonals .
we redraw our pictures of dissected polygons so that every region of the dissection becomes a regular polygon _ without _ diagonals . in the nomenclature of @xcite , we denote the resulting object as a _ cluster _ , where each regular polygon of the cluster is called a _
cell_. thus the polygon shown in figure [ cluster]a with five diagonals is redrawn as a cluster in figure [ cluster]b having six cells .
clearly dissected polygons and clusters are equinumerous , and it is sometimes easier to work with clusters rather than with dissections , as the following important property of the associahedron demonstrates .
[ p : product ] each face of @xmath17 is a product of lower dimensional associahedra .
we look at the top ( codim one ) faces of @xmath24 .
the three - dimensional @xmath24 corresponds to a 6-gon , which has two distinct ways of adding a diagonal .
one way , in figure [ k5codim1]a , shows the @xmath25-gon as a cluster of two @xmath26-gon cells ; each cell corresponds to the line segment @xmath22 , where the product of two such is a square . alternatively , figure [ k5codim1]b illustrates the other type of face , which is the product of @xmath21 ( point ) and @xmath23 ( pentagon ) .
[ h ] [ [ section-2 ] ] it was shown in @xcite that @xmath0 is tessellated by exactly @xmath27 copies of @xmath28 .
for example , observe from figure [ m05 ] that @xmath14 is tiled by twelve @xmath23 pentagons .
the reasoning is as follows : there are @xmath29 possible ways of labeling the sides of an @xmath3-gon , each corresponding to an associahedron . however , since there exists a copy of the dihedral group @xmath30 of order @xmath31 in @xmath9 , and since @xmath0 is defined as a quotient by @xmath9 , two labeled polygons ( associahedra ) are identified by an action of @xmath30 .
therefore , in some sense , each associahedral domain of @xmath0 corresponds to a labeled @xmath3-gon up to rotation and reflection .
let @xmath32 be the set of equivalence classes of _ labeled _ regular @xmath3-gons with @xmath19 diagonals up to the action of @xmath30 .
let @xmath33 be an element of @xmath32 and @xmath34 be a diagonal of @xmath33 .
the _ twist _ of @xmath33 along @xmath34 , denoted by @xmath35 , is the polygon in @xmath32 obtained by separating @xmath33 along @xmath34 into two parts , ` twisting ' ( reflecting ) one of the pieces , and gluing them back ( figure [ twist ] ) .
it does not matter which piece is twisted since the two results are identified by an action of @xmath30 .
[ h ] two polygons , @xmath36 , representing codim @xmath19 faces of associahedra , are identified in @xmath37 if there exist diagonals @xmath38 of @xmath39 such that @xmath40 [ t : glue ] the previous statement shows how copies of associahedron glue together to form the moduli space .
figure [ m04 ] demonstrates @xmath41 tiled by three labeled @xmath22 s .
[ [ section-3 ] ] recall that faces of associahedra can be represented by dissected polygons .
we give the following definitions that categorize the faces of @xmath17 .
two polygons @xmath39 and @xmath42 are of the same _ dimension _ if their corresponding faces are of the same dimension .
_ type _ if their corresponding faces are the same polytope . _ class _ if they are identified under the actions of @xmath30 and twisting .
dimension , type and class are in increasing order of finer partitions . in figure [ types ] , we note the difference between the categories of polygons belonging to @xmath24 .
recall that the number of diagonals of a polygon is equal to the codim of its face ; polygons ( a ) and ( b ) are of dim two , and ( c ) through ( e ) are of dim zero .
in particular , polygons ( c ) through ( e ) are of the same type ( points ) whereas ( a ) and ( b ) are not ; the polytope ( a ) is a square and ( b ) a pentagon .
only ( d ) and ( e ) are of the same class ; that is , twisting along a diagonal of ( d ) will result in ( e ) .
[ h ] it makes sense to decompose the faces into dimensions and polytopes , but what does it mean to talk about classes ? the two different classes of points discussed above correspond to the following : at one class ( c ) , three pentagons meet ; there are two such points in @xmath24 .
the other class ( d ) is where two pentagons and one square meet ; there are twelve such points .
therefore , classes categorize _ the different ways polytopes intersect _ in @xmath17 .
this idea can be extended to better understand the structure of @xmath0 .
[ [ section-4 ] ] the combinatorics involving dimensions are handled in @xcite .
we recall the number of faces in @xmath17 of codim @xmath19 is @xmath43 in other words , given the @xmath44-vector of the polytope @xmath17 to be @xmath45 , the number above is @xmath46 .
[ l : cayley ] this was originally proved by professor a. cayley in 1891 in finding the number of partitions of an @xmath20-gon into @xmath19 parts @xcite .
the recent work of przytycki and sikora gives a short and elementary proof while providing relations to knot theory @xcite .
the number of faces in @xmath17 of codim @xmath18 is the well - known catalan number @xmath47 .
that is , there are @xmath48 vertices in the associahedron @xmath17 .
it is interesting to note that stanley shows sixty - six ways to represent the catalan numbers ( * ? ? ?
* exercise 6.19 ) .
these are the three fundamental problems addressed in this paper : _ p1 .
_ how many different types / classes of codim @xmath19 are there in @xmath17 ?
_ what are the different types / classes in @xmath17 of codim @xmath19 ?
_ how many faces of @xmath17 belong to a given type / class ? in some sense , faces of the associahedron correspond to polygons and faces of the moduli space to _ labeled _ polygons .
therefore , since types and classes were defined regardless of labeling , the answers to _ p1 _ and _ p2 _ both carry over to the moduli space .
this is not the case for _
p3_. however , we find in @xcite a formula ( that follows from theorem [ t : glue ] ) which addresses this dilemma .
let @xmath49 denote the number of codim @xmath19 faces in the cw - complex @xmath50
. then @xmath51 [ p : k - to - m ] [ [ section-5 ] ] in what follows , we find solutions to our fundamental problems regarding the classification of types .
we try to clarify the situation with the following example .
looking at the case of @xmath52 , lemma [ l : cayley ] shows there to be @xmath53 codim one faces .
figure [ k6codim1 ] shows there to be two different types of such faces ( answers _ p1 _ ) , one being @xmath52 and the other a pentagonal prism ( answers _ p2 _ ) . out of the 14 faces , there turn out to be seven faces of each type ( answers _ p3 _ ) .
[ h ] polygons are of the same type if their corresponding associahedral faces have identical polytopal structure .
as proposition [ p : product ] indicates , the polytope structure is solely determined by the number of different cells , not the way these cells are glued to form the cluster .
the following notation is well - defined : a face is of type @xmath54 if it corresponds to a cluster that has @xmath55 number of @xmath56-sided cells .
the superscript @xmath57 as well as the terms with @xmath58 are omitted .
for example , the three types in figure [ types ] are , respectively , @xmath59 define @xmath60 to be the number of partitions of @xmath61 into exactly @xmath19 integers @xmath62 , where @xmath63 and @xmath64 .
there is no closed form for @xmath60 but the following recurrence relation provides a means to calculate it : @xmath65 [ solution to _ p1 _ and _ p2 _ ] the number of different types in @xmath17 of codim @xmath19 is @xmath66 .
the different partitions correspond to the different types , where @xmath67 is associated with the polytope @xmath68 .
an @xmath20-gon with @xmath19 diagonals ( that is , a cluster containing @xmath69 cells ) represents a codim @xmath19 face of @xmath17 .
the type of a face depends on the polytope structure ; this is soley determined by the cells forming the cluster , _ not _ the way in which they are arranged to form the @xmath20-gon .
associating an @xmath20-gon to the integer @xmath70 , observe that partitioning the polygon into @xmath69 cells is identical to partitioning the integer into @xmath69 parts .
[ solution to _ p3 _ ] when @xmath71 and @xmath72 , the number of faces in @xmath17 of type @xmath54 is @xmath73 this is proven by goulden and jackson ( * ? ? ?
* exercise 2.7.14 ) .
their argument is similar to the proof given below for the enumeration of @xmath74-clusters ( see [ ss : aclusters ] ) , but without the twist operation and with the extra variables @xmath55 in the generating function to mark the numbers of @xmath56-sided regions . from the recursive equation for the generating function
, the coefficients ( functions of @xmath55 ) are obtained by lagrange s theorem and then expanded using the multinomial theorem .
values computed are shown in table [ t : diss ] using equation [ e : part ] . for each size of polygon , and each distribution of sizes of the cells in the dissection , it gives the number of polygons which have that distribution .
an example of an entry for the partition of a 10-gon shows that there are 660 polygons of type @xmath75 .
that is , there are 660 faces in @xmath76 that have the polyhedral structure of @xmath77 .
[ [ ss : aclusters ] ] unlike types , the problem with regard to classes is no longer classical . counting the number of dissected polygons up to rotation and reflection
was studied in @xcite .
the difference now comes from imposing the additional equivalence relation of twisting .
the method used to enumerate classes naturally follows as an extension of @xcite .
a cluster is said to be _ rooted at an outside edge _ if one edge ( the _ root _ ) of its perimeter is distinguished from the others .
figure [ rootedge]a shows the root edge as a thicker line . for convenience ,
we refer to clusters rooted at an outside edge as _
@xmath74-clusters_. let @xmath78 be the number of @xmath74-clusters having @xmath79 cells and @xmath3 outside edges not counting the root edge .
define the generating function @xmath80 where for the moment the limits for @xmath79 and @xmath3 are unspecified . for an @xmath74-cluster
, there will be one cell with the root edge on its boundary ; we can call this the _
root cell_. assume this cell has @xmath19 unrooted edges .
at each of these edges we have the option of gluing the root of some @xmath74-cluster to it .
each of these glued clusters contributes its values of @xmath79 and @xmath3 to those of the resulting @xmath74-cluster .
for the edges of the root cell at which nothing was glued , the contributions to @xmath79 and @xmath3 are 0 and 1 respectively .
hence the generating function @xmath81 will contain the term @xmath82 . in the rest of @xmath81 , @xmath79 goes from 1 upwards and @xmath3 from @xmath83 upwards .
we now have a typical plya - class problem .
there are @xmath19 _ sites _ around the root cell at each of which we can place a figure , an @xmath74-cluster or just an edge , and the parameters @xmath79 and @xmath3 are additive .
the figure counting series is the function @xmath81 just defined . to accommodate the twisting operation
, we must regard an @xmath74-cluster as being equivalent to the one obtained by reversing the root edge as in figure [ rootedge]a and [ rootedge]b .
this reversal interchanges the sites in pairs if @xmath19 is even ; if @xmath19 is odd then one site remains fixed while the others interchange in pairs .
hence the cycle - index for this problem is @xmath84 by plya s theorem , the generating function for @xmath74-clusters for which the root cell has @xmath19 unrooted edges is obtained by putting @xmath85 in the appropriate formula above . summing for @xmath19 from 2 upwards we get a function of @xmath86 and @xmath87 , @xmath88 which reduces to @xmath89.\ ] ] by substituting @xmath85 in this expression , multiplying by @xmath90 to account for the root cell , and adding the term @xmath82 which we know must be present , we recover the generaing function @xmath81 for @xmath74-clusters .
this gives the recursive equation @xmath91 .
\label{e : axy}\ ] ] the occurrence of @xmath92 in this equation makes it impossible to obtain a closed form for the coefficients @xmath78 but they can be successively computed as far as one wishes ( see table [ a - mn ] ) . another proof of lemma [ l : cayley ]
is obtained by computing the corresponding result above when twisting is _ not _ allowed .
then the relevant permutation group is the identity group , with cycle - index @xmath93 , and the analog of equation [ e : axy ] is @xmath94 where the numbers @xmath95 are the counterparts to the @xmath78 under the new conditions . in this case ,
as was shown in @xcite , a closed result is possible : @xmath96 that is , @xmath95 is the number of codim @xmath97 faces in @xmath17 ; values are given in table [ v - mn ] .
[ [ section-6 ] ] we now turn to clusters rooted at a cell , denoted as @xmath98-clusters .
figure [ rootcell ] shows a typical @xmath98-cluster , having a distinguished cell ( shown shaded ) called the _ root cell_. a @xmath98-cluster is obtained by gluing @xmath74-clusters ( including possibly the empty @xmath74-cluster ) by their roots to the edges of the root cell .
if the root cell is @xmath19-sided , we have a plya - class problem for which the figure counting series is @xmath99 and the permutation group is the dihedral group @xmath100 . since the twisting operation is allowed , the way that the @xmath74-clusters are attached is not important . for this reason
we are spared the tiresome necessity of distinguishing between symmetric and asymmetric clusters which had greatly complicated the enumerations performed in @xcite .
the cycle - index of @xmath100 is @xmath101 the @xmath98-clusters for a given value of @xmath19 are enumerated by substituting @xmath99 in this cycle - index to get the generating function denoted by @xmath102 .
hence the set of all @xmath98-clusters is enumerated by @xmath103 there seems to be no way of simplifying this expression , but the coefficients can be easily computed as far as desired . some values for @xmath104 are given in table [ b - mn ] .
[ [ section-7 ] ] the next enumeration is for clusters rooted at an inside edge which are asymmetrical about that edge ; that is , for which the two clusters at the edge are different ( figure [ irootedge ] ) .
clearly these @xmath105-clusters , as they can be called , are formed by gluing together two different @xmath74-clusters , neither of which is the empty cluster , at their root edges .
plya s theorem shows the generating function for these clusters would be obtained by substituting @xmath106 in the cycle - index of the symmetric group @xmath107 , namely @xmath108 , but for the fact that we want the two @xmath74-clusters to be different .
a variant of plya s theorem , which is well - known but in any case easily verified , provides the required result by substituting instead in the expression @xmath109 . calling the resulting generating function @xmath110
, we have @xmath111.\ ] ] the final step , going from the generating function for cell - rooted clusters to _ free _ clusters ( those not rooted at all ) , is carried out by means of a theorem originally due to otter @xcite ( also see @xcite , @xcite ) .
it states that the generating function for these free clusters , call it @xmath112 , is obtained by subtracting @xmath110 from @xmath113 .
the number @xmath114 of clusters with @xmath79 cells and @xmath3 outside edges , is found from @xmath115 some values for @xmath114 are given in table [ f - mn ] .
we have just proved [ solution to _
_ ] the number of different classes of @xmath17 of codim @xmath19 is @xmath116 .
[ [ section-8 ] ] as a help in the study of the various classes , we prepared a set of drawings of the corresponding dissected polygons up to ten vertices .
figures 22 - 26 illustrate the different classes , labeled as @xmath117 to denote an @xmath3-gon with @xmath19 diagonals where @xmath56 ranges from one to @xmath118 .
the computer algorithm for doing this was nothing out of the ordinary , but a few comments on how the drawings were produced may be of some interest .
we exploited the connection between dissected polygons and trees as illustrated in figure [ f : pct ] .
trees dual to dissected polygons are _ homeomorphically irreducible _ , having no vertices of degree two .
henceforth , mention of trees will imply homeomorphically irreducible ones . to each dissected polygon
, we associate a tree imbedded in the plane . for each diagonal of the polygon , there is a tree edge connecting the two vertices on either side .
the twist operation corresponds to reflecting one of the two subtrees at the ends of a given edge .
classes correspond to planar imbeddings of trees up to twisting .
in other words , for a given class , each vertex of the tree is given the addition information of preserving the cyclic order of edges around that vertex .
the problem was therefore to produce one such tree for each class . since the trees needed were quite small ,
it was not worth devising a clever algorithm ; both elegance and efficiency were sacrificed for ease of programming .
rooted trees , the root being a vertex of the tree , were generated step by step according to height .
there is one tree of height zero , and those of height one are the stars with three or more edges at the root .
those of height two were obtained by grafting a subtree of height at most one , by its root , to the terminal vertices of these stars . in general ,
trees of height @xmath19 were obtained by grafting trees of height at most @xmath119 . at each stage , if two trees were reflections of each other , only one was retained , thus taking care of the twist operation .
this corresponds to producing clusters rooted at an edge , which made possible a check that the right numbers of trees were being produced .
the resulting lists of rooted trees would usually contain several copies of the same unrooted tree .
the unwanted duplicates were weeded out by computing for each tree a _ code _ similar to that described in @xcite and eliminating all but one of those with the same code .
the actual program contained some refinements to the procedure just described but we shall skip the details .
it was a fairly simple task to go from trees to dissected polygons .
the diagrams were then produced automatically using programs that had already been developed for the production of the book _ an atlas of graphs _
@xcite .
the different classes correspond to the different pictures in figures 22 - 26 . at a later stage we decided to compute for each class the number of fixed dissections of that class .
had we thought of this earlier , it could have been done more easily along with the generation of the classes , but as it was , we derived these numbers directly from the data on the dissected polygons . for each such polygon , we produced the fixed polygons obtainable from it by rotations or by combinations of the twist operation .
note that twisting about some chords leaves the diagram unchanged , in which case there is no point in performing the twist .
for example , those chords which have a single polygon on one side fall into this category , and are easily recognised .
for ease of programming , only the chords of this kind were exempted from the twist operation ; no attempt was made to recognize any others . as a result , duplicates could , and were , produced . however , for polygons with at most ten sides there can be at most five such chords , so the number of duplicates was small .
they were eliminated by a suitable coding procedure , similar to that mentioned above for trees .
it is reassuring that by summing these numbers for the appropriate sets of classes we recover the numbers given in table [ t : diss ] .
the number of faces of @xmath17 of a particular class is indicated in the top right - hand corner of each picture in figures 22 - 26 .
[ p : cp3 ] [ [ section-9 ] ] as mentioned above , there is a duality between clusters and trees .
clearly each cell of a cluster corresonds to a star in the tree .
a face of the associahedron of a given type only recognizes the set of stars , not the way in which they are grafted to form the tree .
however , a given class not only sees the underlying tree , but the additional cyclic order of edges around each vertex .
hence , faces of associahedra can be categorized by types , trees , and classes , in increasing order of refinement .
figure [ phylo ] shows three different trees and their corresponding ( dual ) polygons .
all of them are of the same type @xmath120 $ ] and all fall into different classes .
note , however , that the first two correspond to the same tree , whereas the third does not .
by not preserving the cyclic order of edges , we obtain a distinction that falls between the categories of type and class .
the enumeration of all trees ( not just homeomorphically irreducible ones ) by their degree partitions is given by harary and prins @xcite .
they provide a method for computing the generating function @xmath121 where @xmath122 is the number of trees having @xmath3 vertices altogether , with @xmath123 vertices of degree @xmath56 . to get the number of _
homeomorphically irreducible _
trees @xmath124 partitioned by degree , we remove the possibility of any degree being two ( letting @xmath125 ) .
this involves only standard enumerative procedures and the details are omitted .
we give below the equations that determine the required generating function for homeomorphically irreducible trees.$ ] in place of @xmath126 . ]
@xmath127 + t_4 h_3[p ] + \ldots ) \\ r & = & x(t_1 p + t_3 h_3[p ] + t_4 h_4[p ] + \ldots ) \\ h & = & r - a_2[p]\end{aligned}\ ] ] the number of different trees in @xmath17 of codim @xmath19 is given by the above generating function @xmath128 where @xmath129 , @xmath130 , and @xmath131 .
we note in passing that all homeomorphically irreducible trees up to @xmath132 vertices are depicted in @xcite together with their degree partitions .
[ [ section-10 ] ] roughly speaking , faces of the associahedron correspond to polygons and faces of the moduli space to labeled polygons .
both types ( polytope structures ) and classes ( intersections of polytopes ) are natural objects in the associahedron and the moduli space .
we extend our idea of trees by labeling their external vertices .
unfortunately , _ labeled trees _ are of little use to @xmath0 ; the trees shown in figure [ phylolabel ] are identical considered as labeled trees and yet distinct as faces in @xmath0
. however , labeled trees are _
exactly _ what are needed to understand the points of the _ complex _ moduli space @xmath6 .
a rough sketch for this is as follows : the real moduli space @xmath0 comes from a configuration of labeled points on @xmath8 ( viewed as a circle ) .
keeping track of the cyclic ordering of these points is crucial , since two adjacent labeled points can switch places only by first colliding with each other .
points on a circle give rise to the planarity of trees , and the collision / permutation of adjacent points are kept track of with the twist operation .
the situation is different for the complex case .
@xmath6 is based on a configuration of labeled points on @xmath7 ( viewed as a sphere ) .
there are diffeomorphisms of the sphere which can permute any two labeled points without their need to collide , giving rise to non - planar trees .
this is the fundamental reason that the underlying structures are classes for @xmath0 and trees for @xmath6 .
we point out a few observations about moduli spaces and trees .
the codim @xmath19 faces of @xmath6 correspond to labeled trees with @xmath3 external vertices and @xmath69 internal vertices .
in particular , both @xmath0 and @xmath6 have the same number of zero dim cells , enumerated by binary trees with labeled external vertices .
in other words , the twist operation on planar binary trees removes the restriction of planarity .
there are operad structures related to the moduli spaces @xmath0 @xcite and @xmath6 @xcite , classically described in terms of labeled trees @xcite .
the associahedron is related to @xmath133 structures , whereas @xmath0 seems to play a similar role in fukaya s lagrangian floer cohomology .
kapranov @xcite has defined a natural double cover of the moduli space @xmath0 by fixing one of the @xmath3 distinct labeled points in @xmath8 to be @xmath134 .
similar to @xmath0 , this double cover also has an underlying structure of labeled trees @xcite with two exceptions .
first , the trees are no longer free but rooted at the label @xmath134 .
second , twisting is allowed except along the edge containing the root .
the reason for these moduli spaces to be characterized by trees without degree two vertices is due to the stability condition coming from geometric invariant theory @xcite .
git gives a natural compactification of the space of genus zero algebraic curves which are _ stable _ in the sense of having only finitely many automorphisms ; that is , the curves can not have just two distinguished points .
[ [ section-11 ] ] there is a natural action of the symmetric group @xmath135 on @xmath0 which permutes the labelings of the @xmath3-gon .
this action is certainly not free ; therefore , for a given face @xmath44 in @xmath0 , its isotropy group @xmath136 need not be trivial , where @xmath137 given two distinct faces of @xmath0 which are of the same class , their isotropy groups are _ isomorphic _ ; in fact , they are conjugate subgroups in @xmath135 .
we wish to determine , up to isomorphism , the isotropy group of a given class ; we interchangably write @xmath138 to denote the isotropy group of @xmath44 and its isomorphism class .
let @xmath44 be the face of @xmath139 given in figure [ isotropy]a ; its class is @xmath140 .
its isotropy group is generated by the reflections @xmath141 . since the reflections commute , @xmath138 is isomorphic to the direct product @xmath142 .
as discussed above , all isotropy groups of class @xmath140 will be isomorphic . in particular , a face @xmath2 of the same class , as shown in figure [ isotropy]b , will have a conjugate subgroup @xmath143 .
[ e : isotropy ] for a given class @xmath117 , let @xmath144 be the number of faces in @xmath28 of this class . the number of faces in @xmath0 of the class @xmath117 is @xmath145 this immediately follows from propositions [ p : k - to - m ] and [ p : cp3 ] .
this can be obtained using another approach : choose a dissected polygon representing the class @xmath117 . under all possible labelings ,
there are @xmath29 such polygons , many of which are identified in @xmath0 .
since the isotropy group maps the polygon to itself in the moduli space , we quotient @xmath29 by the order of the isotropy group @xmath146 , where @xmath44 is any face in @xmath0 of the given class .
this provides another answer to for the corollary above ; combining the results , we find a solution to _ p3 _ alternate to the one given in proposition [ p : cp3 ] .
the number of faces in @xmath28 of the class @xmath117 is @xmath147 [ p : new ] [ [ section-12 ] ] the following sketches one possible algorithm for finding the group @xmath138 .
this is done _ locally _ ( looking at each cell of the cluster under twisting ) and _ globally _ ( looking at the entire dissected polygon under the dihedral group ) . for a connected graph @xmath33 , the _ distance _ between two vertices @xmath148 is the minimum of the lengths of paths containing @xmath90 and @xmath82 .
eccentricity _ of a vertex is the maximum of its distances to all other vertices .
the _ center _ of @xmath33 is the subgraph induced by the vertices of minimum eccentricity .
it is a classic result of graph theory that the center of a tree is either one vertex or one edge . taking the dual ,
the center of the dissected polygon is either a cell or a diagonal . _
non - central cells : _ for each non - central cell of the polygon , observe if there exists an automorphism of the cell obtained by twisting along diagonals .
if such a map exists , then this contributes @xmath149 to @xmath138 .
figure [ localcell]b shows the class @xmath150 as a cluster , with the central cell shaded . out of the four non - central cells , three of them have a @xmath149 action ( figure [ localcell]c ) , contributing @xmath142 to the isotropy group
. _ central diagonal : _ if the center is a diagonal , there are two possibilites : observe whether or not there exists a ( global ) reflection along the central diagonal .
figure [ centerdiag ] shows class @xmath151 with reflection and class @xmath152 without .
_ central cell : _ if the center is a cell , there are three possibilites .
figure [ centercell]a shows class @xmath153 with the central cell ( shaded ) having no symmetry .
the cell could also have just a reflection action ( class @xmath154 ) , as in figure [ centercell]b .
finally , for an @xmath3-gon , there could be a dihedral action @xmath100 on the central cell , where @xmath19 divides @xmath3 .
figure [ centercell]c shows class @xmath155 with the @xmath156 group action . rotation symmetry ,
reflection is also possible due to twisting . ] since there are no intersecting diagonals , the different twists and reflections , along with any dihedral group action , all commute .
therefore , to find the isotropy group , we look at the contribution of each local and global action and simply take their direct product .
for example , the isotropy groups ( up to isomorphism ) for the polygons of figure [ centercell ] are @xmath157 , @xmath157 , and @xmath158 , respectively .
using proposition [ p : new ] , we calculate @xmath144 for each of the classes to be @xmath159 , @xmath160 , and @xmath161 , respectively , matching the values obtained from proposition [ p : cp3 ] .
each class corresponds to a polytope . by definition ,
each element of the isotropy group maps the polytope to itself ; but what type of map is it ? using figure [ isotropy]a as an example of a face @xmath44 , we show how the generators of @xmath138 act on the polytope .
note that the polytope structure of @xmath44 is a cube , the product of three line segments , as labeled in figure [ isocube ] .
the actions of the three generators of @xmath138 given in example [ e : isotropy ] are the reflections along the shaded planes .
[ [ section-13 ] ] we first point out a few interesting similarities to classical combinatorics @xcite . as noted above , partitioning of an @xmath3-gon using @xmath19 diagonals corresponds to using @xmath19 sets of brackets meaningfully on @xmath70 _ non - commutative _ variables ( see figure [ k4 ] ) .
the _ generalized bracketing problem of schrder _ is to find the number @xmath162 of different bracketings of @xmath3 variables using an arbitrary number of brackets ( where each bracket must contain at least two elements ) .
lemma [ l : cayley ] gives us the number of different bracketing for a fixed set of @xmath19 brackets .
summing up , we find @xmath163 the _ wedderburn - etherington commutative bracketing problem _ is to find the number @xmath164 of ways of using @xmath18 sets of brackets meaningfully on @xmath3 _ commutative _ variables .
in other words , it is the number of different _ bracketings _ that arise from commuting binary subproducts .
it is not too hard to see that @xmath164 gives the number of @xmath20-gons with maximal diagonals rooted at an outside edge , identified up to reflection along the root edge and twisting along diagonals , but _ not _ rotation .
it follows that @xmath165 ( see table [ a - mn ] ) .
figure [ f : wedder ] shows @xmath166 and @xmath167 as polygon and bracketing representations .
the coefficients of the generating function of equation [ e : gena ] generalize the wedderburn - etherington numbers : the possible ways of using @xmath19 brackets on @xmath3 commutative variables is @xmath168 .
we mention two occurrences outside mathematics that encode information using labeled trees .
the first is in biology ; in particular , the areas of genetics and evolution @xcite .
the theory of evolution conjectures that there exist links between certain species to common ancestors using the evidence from protein sequences .
an evolutionary tree , or _
phylogeny _ , is a means of organizing this data : the external ( labeled ) vertices represent species for which data is available whereas the internal ( unlabeled ) vertices are the hypothetical ancestors for which no direct evidence is known .
phylogenies usually have no vertices of degree two .
the following emumerates the number of phylogenies having a fixed set of leaves .
\2 . a _ rooted _ phylogeny is a rooted labeled tree where the root does not receive a label .
the root of a phylogeny represents the ancestor of all organisms of the tree . for @xmath3
labeled external vertices , the number @xmath172 of distinct rooted phylogenies is @xmath173 @xcite .
there exist many possible metrics on phylogenetic trees @xcite .
it was also noted in @xcite that one can give lengths to the edges of labeled trees , making them _ weighted _ , which hold data on the time estimate of evolution taking place between two species of the tree .
recently , a space of phylogenetic trees was constructed , closely related to the ` dual ' of @xmath0 @xcite ; the metric on this tree space inherits some properties from the natural metric on @xmath0 @xcite .
another appearance of labeled trees is in physics ; in particular , the quantum theory of angular momentum . given a system of @xmath3 independent parts , each with its own angular momentum ,
the theory deals with constructing the total angular momentum by coupling the given @xmath3 momenta sequentially in pairs using @xmath174 wigner coefficients .
biedenharn and louck further relate this to the racah - wigner algebra and the 6-@xmath175 symbols @xcite .
enumerating the various _ binary coupling schemes _ gives a total of @xmath48 possibilities , thought of as counting maximal bracketings on @xmath3 variables ( zero dim cells of @xmath17 ) .
permuting two angular momenta implies a phase change in their _
state vectors_. considering binary coupling schemes over all such permutations yield @xmath176 possibilites ( zero dim cells in @xmath29 copies of @xmath17 ) .
however , in quantum physics it is not possible to distinguish state vectors differing by phase factors ; schemes are identified by commuting binary subproducts .
this yields @xmath177 possible coupling schemes ( zero dim cells in @xmath178 ) .
as noted , the idea of binary coupling closely follows the gluing of associahedra to form the moduli space , at least in the zero dimensional case . although combinatorially the idea of coupling momenta can be generalized beyond binary pairs ( by possibly counting the codim @xmath19 cells of moduli spaces )
, the meaning of this in the context of theoretical physics is not yet well understood .
the authors would like to acknowledge their indebtedness to , and admiration for , the work of professor arthur cayley , who not only addressed a dissection problem similar to the one in this paper , but also laid the foundations for the enumerative study of tree - like structures in general .
this pioneering work is all the more remarkable in that it was only a small part of the mathematical output of this eminent victorian . | the polytope structure of the associahedron is decomposed into two categories , types and classes .
the classification of types is related to integer partitions , whereas the classes present a new combinatorial problem .
we solve this , generalizing the work of @xcite , and incorporate the results into properties of the moduli space @xmath0 studied in @xcite .
connections are discussed with relation to classic combinatorial problems as well as to other sciences .
[ subsection ] [ thm]proposition [ thm]corollary [ thm]lemma [ thm]conjecture [ thm]definition [ thm]example |
this paper is a capsule summary of our approach to invariants of three manifolds via right integrals on finite dimensional hopf algebras .
the aim of this paper is to quickly show the reader how natural it is to consider right integrals in this context .
the defining property of a right integral is a categorical algebra translation of the handle - sliding move in the kirby calculus .
we show in @xcite that invariants defined in terms of right integrals , as considered in this paper , are distinct from the invariants of reshetikhin and turaev @xcite .
we show that the hennings invariant ( defined via these integrals ) is non - trivial for the quantum group @xmath0 when @xmath1 is an fourth root of unity .
the reshetikhin turaev invariant is trivial at this quantum group and root of unity .
the hennings invariant distinguishes all the lens spaces @xmath2 from one another at this root of unity .
this proves that there is non - trivial topological information in the non - semisimplicity of @xmath0 .
this non - triviality result has also been obtained by ohtsuki @xcite .
the reader interested in comparing the approach of this paper with other ways to look at quantum link invariants will enjoy looking at the references @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
in particular , the method we use to write link invariants directly in relation to a hopf algebra is an analog of the construction in @xcite and it is a generalization of the formalism of @xcite and @xcite .
the papers @xcite , @xcite , @xcite , @xcite consider categorical frameworks that also use right integrals on hopf algebras . the paper is organized as follows .
section 2 recalls the hopf algebra background .
section 3 recalls the category associated with a hopf algebra .
section 4 weaves in the tangle category and constructs the functor from the tangle category to the hopf algebra category .
section 5 describes the invariants of three manifolds .
* acknowledgement*i thank the national science foundation for support of this research under nsf grant dms-9205277 and the nsa for partial support under grant number mspf-96g-179 .
recall that a hopf algebra @xmath3 @xcite is a bialgebra over a commutative ring @xmath4 that has an associative multiplication and a coassociative comultiplication and is equipped with a counit , a unit and an antipode .
the ring @xmath4 is usually taken to be a field .
@xmath3 is an algebra with multiplication @xmath5 .
the associative law for m is expressed by the equation @xmath6 where @xmath7 denotes the identity map on a. the coproduct @xmath8 is an algebra homomorphism and is coassociative in the sense that @xmath9 the unit is a mapping from @xmath4 to @xmath3 taking @xmath10 in @xmath4 to @xmath10 in @xmath3 , and thereby defining an action of @xmath4 on @xmath11 it will be convenient to just identify the units in @xmath4 and in @xmath3 , and to ignore the name of the map that gives the unit .
the counit is an algebra mapping from @xmath3 to @xmath4 denoted by @xmath12 the following formulas for the counit dualize the structure inherent in the unit : @xmath13 it is convenient to write formally @xmath14 to indicate the decomposition of the coproduct of @xmath15 into a sum of first and second factors in the two fold tensor product of @xmath3 with itself .
we shall often drop the summation sign and write @xmath16 the antipode is a mapping @xmath17 satisfying the equations @xmath18 and @xmath19 where 1 on the right hand side of these equations denotes the unit of @xmath4 as identified with the unit of @xmath11 it is a consequence of this definition that @xmath20 for all @xmath15 and @xmath21 in a. a quasitriangular hopf algebra @xmath3 @xcite is a hopf algebra with an element @xmath22 satisfying the following equations : \1 ) @xmath23 where @xmath24 is the composition of @xmath25 with the map on @xmath26 that switches the two factors .
\2 ) @xmath27 @xmath28 the symbol @xmath29 denotes the placement of the first and second tensor factors of @xmath30 in the @xmath31 and @xmath32 places in a triple tensor product .
for example , if @xmath33 then @xmath34 these conditions imply that @xmath30 has an inverse , and that @xmath35 it follows easily from the axioms of the quasitriangular hopf algebra that @xmath30 satisfies the yang
baxter equation @xmath36 a less obvious fact about quasitriangular hopf algebras is that there exists an element @xmath37 such that @xmath37 is invertible and @xmath38 for all @xmath15 in @xmath11 in fact , we may take @xmath39 where @xmath40 .
this result is due to drinfeld @xcite .
an element @xmath41 in a hopf algebra is said to be _ grouplike _ if @xmath42 and @xmath43 ( from which it follows that @xmath41 is invertible and @xmath44 ) .
a quasitriangular hopf algebra is said to be a _ ribbon hopf algebra _
@xcite , @xcite if there exists a grouplike element @xmath41 such that ( with @xmath37 as in the previous paragraph ) @xmath45 is in the center of @xmath3 and @xmath46 .
we call g a _ special grouplike _ element of a. since @xmath47 is central , @xmath48 for all x in a. therefore @xmath49 we know that @xmath50 thus @xmath51 for all @xmath15 in @xmath11 similarly , @xmath52 thus the square of the antipode is represented as conjugation by the special grouplike element in a ribbon hopf algebra , and the central element @xmath47 is invariant under the antipode .
we will recall in capsule form the method for obtaining invariants of knots , links and 3manifolds that we used in @xcite , @xcite and @xcite .
this method produces a functor @xmath53 where @xmath54 is the category of unoriented tangles taken up to regular isotopy , and @xmath55 is a natural category associated with a quasi - triangular hopf algebra .
we first recall the structure of @xmath55 , then show how right integrals on the hopf algebra lead to invariants of three manifolds . in this recollection , we start by describing the association of a category with an algebra @xmath3 , not necessarily a hopf algebra , and then add structure to the category corresponding to properties of a hopf algebra . by the end of the discussion , the category will have been built . at the outset
@xmath3 is an algebra with a commutative ground ring @xmath56 , so that @xmath3 is a module over @xmath57 @xmath55 is an ( associative tensor ) category whose objects are generated under tensor product by two objects @xmath4 and @xmath58 such that @xmath59 and @xmath60 and whose morphisms are generated using composition and tensor product , from basic morphisms that will be defined in this section .
the object @xmath58 is strictly abstract .
the object @xmath4 will be identified with the ground ring @xmath56 as described below .
functors on this category will often associate to @xmath58 a representation space of the algebra @xmath11 to each element @xmath61 of the algebra @xmath3 there will be a generating morphism ( still denoted by @xmath61 ) @xmath62 multiplication in the algebra corresponds to composition of the corresponding morphisms .
the identity morphism @xmath63 corresponds to the identity element @xmath10 in the algebra @xmath3 in the same way , elements of the tensor powers @xmath64 of the algebra correspond to morphisms with domain and range @xmath65 morphisms in @xmath55 corresponding to elements of the algebra are denoted by vertical upward pointing arrows ( with the domain below the range ) .
we shall customarily denote the presence of the algebra element @xmath61 by a bead " ( a node ) on the arrow with the label @xmath66 an unlabelled arrow without a bead corresponds to the identity morphism @xmath67 see figure 1 .
-3mm-5 mm the algebra @xmath3 has a base ring @xmath56 ( usually a field ) .
the object @xmath4 in the category is identified with the base ring @xmath56 .
if @xmath68 denotes the unit in the algebra ( ie , @xmath69 where @xmath15 is in @xmath4 and @xmath10 is the unit in the algebra ) then elements of @xmath70 form a special set of morphisms in the category since @xmath71 for any @xmath15 in @xmath4 and @xmath61 and @xmath72 in @xmath3 .
accordingly , we shall denote a morphism in @xmath55 that corresponds to an element @xmath73 in @xmath70 by an arrow _ without a bead _ but labelled with @xmath73 . in this way we can move the label @xmath73 freely and when it leaves a given arrow , that arrow ( if otherwise unlabelled ) reverts to an identity arrow as it should
. see figure 1 .
we are free to utilize the internal structure of the base ring @xmath4 when that is appropriate .
it is a curious feature of this category that one of its objects is also the holder of an important subcategory ( the morphisms generated by @xmath70 ) . in other respects
the base object @xmath4 can be regarded in a purely external categorical way .
every hopf algebra has an antipode @xmath74 with the defining property that @xmath75 where @xmath76 is the multiplication in the algebra , @xmath77 is the counit , and @xmath78 is the coproduct .
a key property of the antipode is that it is an antimorphism of the algebra to itself .
it reverses the order of multiplication so that @xmath79 for all @xmath61 and @xmath72 in @xmath11 the structure of the antipode is reflected in the category @xmath55 by new morphisms that we now describe as @xmath80 and @xmath81 the key property of these morphisms is their behaviour with respect to the antipode : @xmath82 @xmath83 these morphisms and all the remaining morphisms that we construct in this category are represented by ( immersed ) curves in the plane .
see figure 2 for the diagrammatic depiction of @xmath84 , @xmath85 and these bead sliding " properties .
it is convenient to think of these identities as permissions to slide the beads past maxima or minima in the diagrams for the morphisms .
upon sliding across a maximum in a counterclockwise direction , a bead has the antipode applied to it .
the properties of @xmath84 and @xmath85 are augmented with the _ topological _ identities @xmath86 @xmath87 see figure 3 .
these identities assert that @xmath84 and @xmath85 behave just as topological maxima and minima : when composed without any intervening morphisms they cancel to the identity .
-2 mm -2 mm it follows from the antipodal and topological behaviour of @xmath84 and @xmath85 that the morphism corresponding to @xmath88 is represented by composing on the top and the bottom of @xmath61 with a @xmath84 and a @xmath85 , as shown in figure 4 . a proof by bead sliding is shown in that figure . formally
, we have the result : let @xmath61 be an element of @xmath3 viewed as a morphism from @xmath58 to @xmath58 .
then @xmath89 .
@xmath90 @xmath91 @xmath92 @xmath93 in fact , the cup and cap morphisms provide an interpretation of @xmath94 for any morphism @xmath95 in the category by the replacement of @xmath61 by @xmath96 in the diagram in figure 4 , and in the formula of this lemma .
thus @xmath97 one last type of morphism completes that catalog for @xmath98 this morphism is a formal permutation @xmath99 with the property @xmath100 we represent @xmath101 by a pair of crossed arrows as shown in figure 5 .
it is assumed that @xmath101 satisfies the usual axioms for permutation plus a compatibility condition with respect to the @xmath84 and the @xmath85 as shown in figure 6 .
-2 mm -8 mm the hopf algebra @xmath3 is endowed with a coproduct @xmath102 this map is indicated algebraically via the formula @xmath103 where an implicit summation is sometimes indicated without the presence of the summation symbol . in diagrams the line with a bead labelled @xmath61 on it
is replaced by a two parallel lines ( for the tensor product ) with the left line s bead labelled @xmath104 and the right line s bead labelled @xmath105 see figure 7 .
-4 mm clearly @xmath25 must simply double all diagrammatic lines to be defined on the structural morphisms in @xmath55 .
this is illustrated in figure 8 . before going to the tangle category it is useful to delineate special morphisms in @xmath55 that we call curls " .
a _ curl _ is a morphism of the form @xmath106 or @xmath107 .
the second curl is denoted by @xmath108 because the equations @xmath109 hold via a regular homotopy of the corresponding plane curves .
( the regular homotopy is implicitly shown in figure 9 . )
these equations are the categorical analog of the so - called whitney trick " .
the whitney graustein theorem @xcite tells us that any morphism from @xmath58 to @xmath58 with no elements of @xmath3 in its factorization is equivalent to an integer power of @xmath41 .
the element @xmath41 is a formal grouplike element in the sense that @xmath110 as illustrated in figure 8 . in terms of the morphism @xmath41 we can represent the square of the antipode in @xmath55 by the formula @xmath111 where this formula denotes a composition of morphisms in @xmath98 in many cases the element @xmath41
can be identified with a specific grouplike element in the hopf algebra @xmath3 .
the category takes the lead here in expressing a preference for such a formula .
see figure 9 .
-2 mm -5 mm it is now easy to see the following : every morphism @xmath15 in @xmath55 with source and target both @xmath58 can be written in the form @xmath112 where @xmath113 and @xmath114 is an integer .
this result is accomplished by sliding all the algebra on the immersed curve that represents the morphism to the bottom of the curve .
the element @xmath61 is the algebraic element so concentrated at the top of the curve . the curve itself is regularly homotopic ( fixing the ends at top and bottom ) to @xmath115 for a unique integer @xmath114 ( the whitney degree of that curve ) .
since regular homotopy is produced by the topological axioms of the category , this shows that @xmath116 as desired . using the method of this proof
, we define @xmath117 and @xmath118 so that @xmath119
for the detailed definition of @xmath54 we refer the reader to @xcite .
the objects will again be formally generated by two ( formal ) symbols @xmath4 and @xmath58 as for @xmath98 the morphisms of @xmath54 are generated by @xmath84 , @xmath85 , the identity line ( and its tensor products ) and the _ right and left crossings _ , maps of @xmath120 to itself .
the crossings are denoted @xmath56 and @xmath121 as shown in figure 10 .
@xmath56 denotes a crossing of two arcs so that the overcrossing line goes upward from right to left .
@xmath121 denotes a crossing of two arcs so that the overcrossing line goes upward from left to right .
all diagrams of morphisms in @xmath54 are drawn in the plane with no intersections of lines other than the formalized crossings .
the crossings are intended to be interpreted in sense of the knot theorist .
that is , one can _ interpret _ such a diagram as representing an embedding in three dimensional space of a collection of curves .
the crossing indicates that one curve passes over another from the point of view of a planar projection .
note that the permutation @xmath101 of the category @xmath55 is absent from @xmath122 the crossings can be regarded as generalizations of the permutation .
we will shortly use this idea in making a functor @xmath123 from @xmath54 to @xmath98 -2 mm -4 mm every morphism in the tangle category is a composition of the elementary morphisms @xmath84,@xmath85 , @xmath56 and @xmath124 the objects in @xmath54 are identical to the objects in @xmath98 while the objects in the tangle category are very simple , the morphisms are quite complex .
each morphism in the tangle category consists in a link diagram with free ends which is transverse with respect to a given direction in the plane .
( this special direction will be called the _ vertical _ direction . )
the transversality of the diagram to this vertical direction means that any given line perpendicular to the vertical direction intersects the diagram either tangentially at a maximum or a minimum , at non - zero angle for any other strand .
we shall further assume that any given perpendicular intersects the diagram at most at one crossing . with these stipulations
the free ends of the diagram occur at either its top or its bottom ( top and bottom taken with respect to the designated vertical direction).we shall assume that all the top ends occur along the same perpendicular , and that all the bottom ends occur along another perpendicular to the vertical .
to each of these two rows of diagram ends is assigned a tensor product of copies of @xmath58 , one for each end . in the tangle category ,
the diagram is a morphism from the lower tensor product to the upper tensor product . if the top of a diagram has no free ends , then its range is @xmath4 .
if the bottom of a diagram has no free ends , then its domain is @xmath4 .
a diagram is said to be _ closed _ if it has no free ends .
thus a closed diagram is a morphism from @xmath4 to @xmath4 . if @xmath3 and @xmath125 are morphisms in the tangle category with @xmath126 , then the composition of @xmath3 and @xmath125 is denoted @xmath127 . in discussing @xmath54
we shall continue to use the topological terminology _ tangle _ for a morphism in the category .
@xmath128 tangle _ is a tangle with @xmath114 inputs and @xmath129 outputs .
thus a @xmath130 tangle is any map from @xmath58 to @xmath58 in @xmath54 , and a @xmath131 tangle is any knot or link arranged with respect to the vertical to give a morphism from @xmath4 to @xmath4 . if @xmath132 is an @xmath128 tangle in @xmath54 we define @xmath133 to be the @xmath134 tangle obtained by replacing every strand of @xmath132 by two parallel copies of that strand .
let @xmath3 be a quasi - triangular hopf algebra .
let @xmath136 denote the yang
baxter element for @xmath3 and write @xmath30 symbolically in the form @xmath137 .
we wish to define a functor from @xmath54 to @xmath55 .
it suffices to define @xmath123 on the generating morphisms @xmath56,@xmath121,@xmath84 and @xmath85 .
we define @xmath138 @xmath139 @xmath140 @xmath141 diagrammatically , it is convenient to to picture @xmath142 as a flat crossing with beads _ above _ the crossing labeled @xmath143 and @xmath144 from left to right , with the summation indicated by the double appearance of the letter @xmath143 .
similarly , @xmath145 is depicted as a crossing with beads _ below _ the crossing and labeled @xmath146 and @xmath144 .
see figure 11 .
-2 mm -5 mm proof that this functor is well - defined is found in @xcite and @xcite . knowing that @xmath123 is a functor on the tangle category , means that we have implicitly defined many invariants of knots and links , since regularly isotopic tangles will have equivalent images under @xmath123 .
we can immediately state the following theorem for @xmath130 tangles , giving invariants with values in the hopf algebra @xmath3 .
let @xmath132 be a single - stranded , @xmath130 tangle .
that is , @xmath132 is a knot on a string " . then @xmath147g^{d(t)}$ ] is a regular isotopy invariant of @xmath132 .
here @xmath148=w(f(t))$ ] is the element of @xmath3 defined in the last section by concentrating all the algebra in @xmath149 in the lower part of the tangle , and @xmath150 is the whitney degree of the plane curve underlying @xmath132 .
in fact @xmath148 \in a$ ] is itself a regular isotopy invariant of @xmath132 , as is @xmath150 . for the proof see @xcite .
* example*in figure 12 we illustrate how a topological curl gives rise under the functor @xmath123 to the special central element @xmath47 of a ribbon hopf algebra .
this element is of particular use in defining the invariants described in the next section .
-2 mm -2 mm
the structure we have built so far can be used to construct invariants of 3manifolds presented in terms of surgery on framed links .
we sketch here our technique that simplifies an approach to 3manifold invariants of mark hennings @xcite .
recall that an element @xmath151 of the dual algebra @xmath152 is said to be a _
right integral _
if @xmath153 for all @xmath15 in @xmath11 for a unimodular @xcite , @xcite finite dimensional ribbon hopf algebra @xmath3 there is a right integral @xmath151 satisfying the following properties for all x and y in a : the upshot of this theorem is that for a unimodular finite dimensional hopf algebra there is a natural trace defined via the existent right integral .
remarkably , this trace is just designed to behave well with respect to handle sliding @xcite , @xcite .
handle sliding is the basic transformation on framed links that leaves the corresponding 3manifold obtained by framed surgery unchanged .
see @xcite .
this means that a suitably normalized version of this trace on framed links gives an invariant of 3manifolds . for a link @xmath165 ,
we let tr(k ) denote the functional on links , as described in the previous section , defined via @xmath162 as above . to see how the condition on handle sliding and the property of being a right integral are related in our category , we refer the reader to figure 13 where the basic form of handle sliding is illustrated and its algebraic counterpart is shown .
the algebraic counterpart arises when we concentrate all the algebra in a given link component in one place on the diagram .
the component is then replaced by a circle and formally its evaluation is @xmath166 for a suitable @xmath15 in the hopf algebra where @xmath166 denotes the evaluation of the morphism corresponding to a circle labelled with @xmath15 as in the figure .
as the diagram shows , if we let @xmath167 , then invariance under handle sliding is implicated by @xmath151 being a right integral on the hopf algebra .
a proper normalization of @xmath168 gives an invariant of the 3manifold obtained by framed surgery on @xmath169 more precisely ( assuming that @xmath170 and @xmath171 are non - zero ) , let @xmath172 where @xmath173 denotes the number of components of k , and @xmath174 denotes the signature of the matrix of linking numbers of the components of @xmath165 ( with framing numbers on the diagonal ) . then @xmath175 is an invariant of the 3manifold obtained by doing framed surgery on @xmath165 in the blackboard framing .
this is our reconstruction of hennings invariant @xcite in an intrinsically unoriented context . , _ on kauffman s knot invariants arising from finite dimensional hopf algebras _ , from : `` advances in hopf algebras '' , lecture notes in pure and appl . math .
148 , marcel dekker , new york ( 1992 ) 205266 | this paper gives a summary of our approach to invariants of three manifolds via right integrals on finite dimensional hopf algebras and their relation to the kirby calculus .
_ it gives the author great pleasure to dedicate this paper to rob kirby on his sixtieth birthday . _ |
the main problem that has motivated the study in this paper is _ approximate range counting_. in a typical example , one is given a set @xmath1 of points in the plane , and the goal is to preprocess @xmath1 into a data structure which supports efficient approximate counting of the number of points of @xmath1 that lie inside a query halfplane .
the hope is that approximate counting can be done more efficiently than exact counting .
this is an instance of a more general and abstract setting . in general , we are given a _ range space _ @xmath2 , where @xmath3 is a set of @xmath4 objects and @xmath5 is a collection of subsets of @xmath3 , called _
ranges_. in a typical geometric setting , @xmath3 is a subset of some infinite ground set @xmath6 ( e.g. , @xmath7 and @xmath3 is a finite point set in @xmath8 ) , and @xmath9{0cm}{0.4cm}}\right . } } \right\}}$ ] , where @xmath10 is a collection of subsets ( _ ranges _ ) of @xmath6 of some simple shape , such as halfspaces , simplices , balls , etc .
( to simplify the notation , we will use @xmath5 and @xmath10 interchangeably . ) the _ measure _ of a range @xmath11 in @xmath3 is the quantity @xmath12 given @xmath2 as above , and a parameter @xmath13 , the goal is to preprocess @xmath3 into a data structure that supports efficient queries of the form : given @xmath14 , compute a number @xmath15 such that @xmath16 we refer to the estimate @xmath17 as an _
@xmath18-approximate count _ of @xmath19 .
the motivation for seeking approximate range counting techniques is that exact range counting ( i.e. , computing the exact value @xmath20 , for @xmath21 ) is ( more ) expensive .
for instance , consider the classical _ halfspace range counting _
problem @xcite , which is the main application considered in this paper . here , for a set @xmath1 of @xmath4 points in @xmath8 , for @xmath22 , the best known algorithm for exact halfspace range counting with near - linear storage takes @xmath23 time @xcite . as shown in several recent papers , if we only want to approximate the count , as in eq .
( [ equation : guarantee ] ) , there exist faster solutions , in which the query time is close to @xmath24 ( and is polylogarithmic in two and three dimensions ) @xcite .
notice that the problem of approximate range counting becomes more challenging as the size of @xmath19 decreases . at the extreme ,
when @xmath25 , we must produce the count _
exactly_. in particular , we need to be able to detect ( without any error ) whether a given query range @xmath26 is _ empty _ , i.e. , satisfies @xmath27 .
thus , approximate range counting ( in the sense defined above ) is at least as hard as range emptiness detection .
we make the standard assumption that the range space @xmath2 ( or , in fact , @xmath28 ) has _ finite vc - dimension _
@xmath29 , which is a constant independent of @xmath4 .
this is indeed the case in many geometric applications . in general , range spaces involving semi - algebraic ranges of _ constant description complexity _ , i.e. , semi - algebraic sets defined as a boolean combination of a constant number of polynomial equations and inequalities of constant maximum degree , have finite vc - dimension .
halfspaces , balls , ellipsoids , simplices , and boxes are examples of ranges of this kind ; see @xcite for definitions and more details .
[ [ known - notions - of - approximations . ] ] known notions of approximations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a standard and general technique for tackling the approximate range counting problem is to use @xmath18-approximations
( absolute - error ) @xmath18-approximation _ for @xmath2 is a subset @xmath30 such that , for each @xmath11 , @xmath31 . in the general case
, it is known that a random sample of size @xmath32 is an @xmath18-approximation with at least some positive constant probability @xcite , and improved bounds are known in certain special cases ; see below for details .
another notion of approximation was introduced by brnnimann_et al . _
a subset @xmath33 is a _
sensitive @xmath18-approximation _ if for all ranges @xmath34 , we have @xmath35 .
brnnimann_et al . _
present a deterministic algorithm for computing sensitive approximations of size @xmath36 , in deterministic time @xmath37-sample_. for given parameters @xmath38 , a sample @xmath33 is a @xmath39-sample if , for any range @xmath11 , we have @xmath40 , where @xmath41 .
et al . _
gave the currently best known upper bound on the size of a sample which guarantees this property , showing that a random sample of @xmath3 of size @xmath42 is a @xmath39-sample with probability at least @xmath43 ; see below for more details .
[ [ relative - pvarepsilon - approximation . ] ] relative @xmath0-approximation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this paper , we consider a variant of these classical structures , originally proposed a few years ago by cohen _
et al . _
@xcite , which provides _ relative - error approximations_. ideally , we want a subset @xmath30 such that , for each @xmath11 , we have @xmath44 this `` definition '' suffers however from the same syndrome as approximate range counting ; that is , as @xmath45 shrinks , the absolute precision of the approximation has to increase . at the extreme , when @xmath46 , @xmath19 must also be empty ; in general , we can not guarantee this property , unless we take @xmath47 , which defeats the purpose of using small - size @xmath18-approximations to speed up approximate counting .
for this reason , we refine the definition , introducing another fixed parameter @xmath48 .
we say that a subset @xmath30 is a _ relative @xmath0-approximation _ if it satisfies eq .
( [ equation : rel - app ] ) for each @xmath11 with @xmath49 . for smaller ranges
@xmath26 , the error term @xmath50 in eq .
( [ equation : rel - app ] ) is replaced by @xmath51 .
as we will shortly observe , relative @xmath0-approximations are equivalent to @xmath52-samplings , with appropriate relations between @xmath53 , @xmath18 , and @xmath54 , @xmath55 ( see theorem [ theo : equivalent : sampling ] ) , but this new notion provides a better working definition for approximate range counting and for other applications . [ [ known - results . ] ] known results .
+ + + + + + + + + + + + + + as shown by vapnik and chervonenkis @xcite ( see also @xcite ) , there always exist absolute - error @xmath18-approximations of size @xmath56 , where @xmath57 is an absolute constant .
moreover , a random sample of this size from @xmath3 is an @xmath18-approximation with constant positive probability .
this bound has been strengthened by li _
et al . _
@xcite ( see also @xcite ) , who have shown that a random sample of size @xmath58 is an @xmath18-approximation with probability at least @xmath43 , for a sufficiently large ( absolute ) constant @xmath57 .
( interestingly , until very recently , this result , worked out in the context of machine learning applications , does not seem to have been known within the computational geometry literature . ) @xmath18-approximations of size @xmath59 can also be constructed in deterministic time @xmath60 @xcite . as shown in @xcite
, there always exist smaller ( absolute - error ) @xmath18-approximations , of size @xmath61 where @xmath62 is the exponent of either the _ primal _ shatter function of the range space @xmath2 ( and then @xmath63 ) or the _ dual _ shatter function ( and then @xmath64 ) . the time to construct these improved @xmath18-approximations is roughly the same as the deterministic time bound of @xcite stated above , for the case of the dual shatter function . for the case of the primal shatter function ,
the proof is only existential .
consider next relative @xmath0-approximations .
one of the contributions of this paper is to show that these approximations are in fact just an equivalent variant of the @xmath52-samplings of li _
et al . _
@xcite ; see section [ sec : relations ] . as a consequence
, the analysis of @xcite implies that there exist relative @xmath0-approximations of size @xmath65 , where @xmath57 is an absolute constant .
in fact , any random sample of these many elements of @xmath3 is a relative @xmath0-approximation with constant probability .
success with probability at least @xmath43 is guaranteed if one samples @xmath66 elements of @xmath3 , for a sufficiently large constant @xmath57 @xcite . to appreciate the above bound on the size of relative @xmath0-approximations , it is instructive to observe that , for a given parameter @xmath53 , any absolute error @xmath67-approximation @xmath68 will approximate `` large '' ranges ( of measure at least @xmath53 ) to within _ relative _ error @xmath18 , as in eq .
( [ equation : rel - app ] ) , as is easily checked ( and the inequality for smaller ranges is also trivially satisfied ) , so @xmath68 is a relative @xmath0-approximation .
however , the vapnik - chervonenkis bound on the size of @xmath68 , namely , @xmath69 , as well as the improved bound of @xcite , are larger by roughly a factor of @xmath70 than the improved bound stated above .
the existence of a relative @xmath0-approximation @xmath68 provides a simple mechanism for approximate range counting : given a range @xmath21 , count @xmath71 exactly , say , by brute force in @xmath72 time , and output @xmath73 as a ( relative ) @xmath18-approximate count of @xmath19 .
however , this will work only for ranges of size at least @xmath74 .
aronov and sharir @xcite show that an appropriate incorporation of relative @xmath0-approximations into standard range searching data structures yields a procedure for approximate range counting that works , quite efficiently , for ranges of any size .
this has recently been extended , by sharir and shaul @xcite , to approximate range counting with general semi - algebraic ranges .
[ [ our - results . ] ] our results .
+ + + + + + + + + + + + in this paper , we present several constructions and bounds involving relative @xmath0-approximations .
we first consider samplings in general range spaces of finite vc - dimension , and establish relations between several different notions of samplings , including @xmath52-samplings , relative @xmath0-approximations , and sensitive @xmath18-approximations .
our main observations are : \(i ) the notion of @xmath0-approximation is equivalent to that of @xmath52-sample , by choosing @xmath54 to be proportional to @xmath53 and @xmath55 proportional to @xmath18 ; see theorem [ theo : equivalent : sampling ] .
\(ii ) a sensitive @xmath75-approximation is also a relative @xmath0-approximation .
\(iii ) the result of li _ et al . _
@xcite is sufficiently powerful , so as to imply known bounds on the size of many of the different notions of samplings , including @xmath18-nets , @xmath18-approximations , sensitive @xmath18-approximations , and , as just said , relative @xmath0-approximations .
some of these connections have already been noted earlier , in @xcite and in @xcite .
we offer this portion of section [ sec : relations ] as a service to the computational geometry community , which , as already noted , is not as aware of the results of @xcite and of their implications as the machine learning community .
next , we return to geometric range spaces , and study two cases where one can construct relative @xmath0-approximations of smaller size .
the first case involves planar point sets and halfplane ranges , and the second case involves point sets in @xmath8 , @xmath76 , and halfspace ranges . in the planar case , we show the existence , and provide efficient algorithms for the construction , of relative @xmath0-approximations of size @xmath77 .
our technique also shows in this case the existence of sensitive @xmath18-approximations with improved quality of approximation .
specifically , for a planar point set @xmath3 , we show that there exists a subset @xmath33 of size @xmath78 , such that , for any halfplane @xmath26 , we have @xmath79 .
( this new error term is indeed an improvement when @xmath80 and is the same as the standard term otherwise . ) in the planar case , the construction is based on an interesting generalization of spanning trees with small crossing number , a result that we believe to be of independent interest . specifically , we show that any finite point set @xmath1 in the plane has a spanning tree with the following property : for any @xmath81 , any _
@xmath82-shallow _ line ( a line that has at most @xmath82 points of @xmath1 in one of the halfplanes that it bounds ) crosses at most @xmath83 edges of the tree .
in contrast , the classical construction of welzl @xcite ( see also @xcite ) only guarantees this property for @xmath84 ; i.e. , it yields the uniform bound @xmath85 on the crossing number .
we refer to such a tree as a _ spanning tree with low relative crossing number _ , and show how to use it in the construction of small - size relative @xmath0-approximations .
things are more complicated in three ( and higher ) dimensions . we were unable to extend the planar construction of spanning trees with low relative crossing number to @xmath86 ( nor to higher dimensions ) , and this remains an interesting open problem .
( we give a counterexample that indicates why the planar construction can not be extended `` as is '' to 3-space . ) instead , we base our construction on the shallow partition theorem of matouek@xcite , and construct a set @xmath68 of size @xmath87 , which yields an absolute approximation error of at most @xmath51 for halfspaces that contain _ at most _ @xmath74 points .
note that this is the `` wrong '' inequality to guarantee small relative error we need this to hold for all ranges with _ at least _
@xmath74 points . to overcome this difficulty
, we construct a _ sequence _ of approximation sets , each capable of producing a relative @xmath18-approximate count for ranges that have roughly a fixed size , where these size ranges grow geometrically , starting at @xmath74 and ending at roughly @xmath4 .
the sizes of these sets decrease geometrically , so that the size of the first set ( that caters to ranges with about @xmath74 points ) , which is @xmath87 , dominates asymptotically the overall size of all of them .
we output this sequence of sets , and show how to use them to obtain an @xmath18-approximate count of any range with at least @xmath74 points .
the situation is somewhat even more complicated in higher dimensions .
the basic approach used in the three - dimensional case can be extended to higher dimensions , using the appropriate version of the shallow partition theorem . however , the bounds get somewhat more complicated , and apply only under certain restrictions on the relationship between @xmath18 and @xmath53 .
we refer the reader to section [ sec : higher ] , where these bounds and restrictions are spelled out in detail .
the paper is organized as follows : in section [ sec : relations ] we survey the sampling notions mentioned above , and show how they relate to each other . in section [ sec : plane ] , we describe how to build a small relative @xmath0-approximation in the planar case , by first showing how to construct a spanning tree with low relative crossing number . in section [ sec : higher : sec ] , we extend the result to higher dimensions . in section [ sec : approximate : counting ] , we revisit the problems of halfplane and 3-dimensional halfspace approximate range counting , and provide algorithms whose query time is faster than those in the previous algorithms .
this section is somewhat independent of the rest of the paper , although we do use relative @xmath0-approximations for the 3-dimensional case .
we conclude in section [ sec : conclusions ] with a brief discussion of the results and with some open problems .
[ sec : relations ] in this section we study relationships between several commonly used notions of samplings in abstract range spaces . in particular , we show that many of these notions are variants or special cases of one another .
combined with the powerful result of li _
et al . _
@xcite , this allows us to establish , or re - establish , for each of these families of samplings , upper bounds on the size of samples needed to guarantee that they belong to the family ( with constant or with high probability ) .
[ [ definitions . ] ] definitions .
+ + + + + + + + + + + + we begin by listing the various kinds of samplings under consideration . in what follows , we assume that @xmath2 is an arbitrary range space of finite vc - dimension @xmath29 . for a given parameter @xmath13
, a subset @xmath33 is an _
@xmath18-net _ for @xmath2 if @xmath88 , for every @xmath11 such that @xmath89 . for a given parameter @xmath13
, a subset @xmath33 is an _
@xmath18-approximation _ for @xmath2 if , for each @xmath11 , we have @xmath90 . for given parameters @xmath91 , a subset @xmath92 is a _ relative @xmath0-approximation _ for @xmath2 if , for each @xmath11 , we have a. @xmath93 , if @xmath94
. b. @xmath95 , if @xmath96 . for a given parameter @xmath97 ,
a subset @xmath33 is a _
sensitive @xmath18-approximation _ for @xmath2 if , for each @xmath11 , we have @xmath98
. finally , for a parameter @xmath99 , consider the distance function between nonnegative real numbers @xmath100 and @xmath101 , given by @xmath102 @xmath103 satisfies the triangle inequality @xcite , and is thus a metric ( the proof is straightforward albeit somewhat tedious ) . for given parameters @xmath104 and @xmath99 , a subset @xmath105 is a _
@xmath39-sample _ for @xmath2 if , for each range @xmath11 , we have @xmath106 .
( note that @xmath107 is uninteresting , because @xmath108 is always at most @xmath109 . )
[ [ equivalence - of - relative - pvarepsilon - approximations - and - nualpha - samples . ] ] equivalence of relative @xmath0-approximations and @xmath52-samples .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we begin the analysis with the following easy properties .
the first property is a direct consequence of the definition of @xmath108 .
let @xmath110 be non - negative real numbers , with @xmath111 .
then @xmath112 if and only if @xmath113 for any non - negative real numbers , @xmath114 , with @xmath115 , put @xmath116 then we have : \(i ) if @xmath117 then @xmath112 .
\(ii ) if @xmath118 then @xmath119 .
[ cor : d : nu : as : interval ] if @xmath120 is a @xmath39-sample for @xmath2 , and @xmath121 is a @xmath39-sample for @xmath122 , then @xmath68 is a @xmath123-sample for @xmath2 . [
lemma : transitive : sample ] an immediate consequence of the triangle inequality for @xmath124 .
the following theorem is one of the main observations in this section .
let @xmath2 be a range space as above .
a. if @xmath33 is a @xmath39-sample for @xmath2 , with @xmath125 and @xmath126 , then @xmath68 is a relative @xmath127-approximation for @xmath2 .
b. if @xmath68 is a relative @xmath52-approximation for @xmath2 , with @xmath128 and @xmath126 then @xmath68 is a @xmath52-sample for @xmath2 .
[ theo : equivalent : sampling ] \(i ) by corollary [ cor : d : nu : as : interval](ii ) , we have , for each @xmath21 , @xmath129 thus , if @xmath130 then @xmath131 , and if @xmath132 then @xmath133 , establishing the claim .
\(ii ) if @xmath130 then @xmath134 which implies , by corollary [ cor : d : nu : as : interval](i ) , that @xmath135 . if @xmath132 then @xmath136 and the claim follows . for a range space @xmath2 ,
if @xmath120 is a relative @xmath52-sample for @xmath2 , and @xmath121 is a relative @xmath39-sample for @xmath122 , with @xmath137 and @xmath126 , then @xmath68 is a relative @xmath138-approximation for @xmath2 .
[ cor:29iii ] by theorem [ theo : equivalent : sampling](ii ) and lemma [ lemma : transitive : sample ] , @xmath68 is a @xmath139-sample for @xmath2 . by theorem [ theo :
equivalent : sampling](i ) , it is then a relative @xmath140-approximation for @xmath2 .
we next recall the bound established in @xcite , and then apply it to theorem [ theo : equivalent : sampling ] . specifically ,
we have : \(i ) * ( li _ et al . _
@xcite ) * a random sample of @xmath3 of size @xmath141 for an appropriate absolute constant @xmath57 , is a @xmath52-sample for @xmath2 with probability at least @xmath43 .
\(ii ) consequently , a random sample of @xmath3 of size @xmath142 for another absolute constant @xmath143 , is a relative @xmath0-approximation for @xmath2 with probability at least @xmath43 .
[ theo : l : l : s ] we next observe that @xmath18-nets and @xmath18-approximations are special cases of @xmath52-samples , where the second observation has already been made in @xcite .
the bound on the size of @xmath52-samples ( theorem [ theo : l : l : s](i ) ) then implies the known bounds on the size of @xmath18-nets ( see @xcite ) and of @xmath18-approximations ( see @xcite ) .
specifically , we have : let @xmath2 be a range space , as above , and let @xmath144 .
a. for any @xmath145 and @xmath146 , a @xmath52-sample from @xmath147 is an @xmath18-net for @xmath2 .
consequently , a random sample of @xmath3 of size @xmath148 , with an appropriate choice of the constant of proportionailty , is an @xmath18-net for @xmath2 with probability at least @xmath43 .
b. if @xmath149 and @xmath150 , then a @xmath52-sample from @xmath147 is an @xmath18-approximation for @xmath2 .
consequently , a random sample of @xmath3 of size @xmath151 , with an appropriate choice of the constant of proportionailty , is an @xmath18-approximation for @xmath2 with probability at least @xmath43 .
[ theo : carnival ] \(i ) we need to rule out the possibility that , for some range @xmath21 , we have @xmath152 and @xmath153 .
since @xmath68 is an @xmath154-sample , we must then have @xmath155 which is impossible , since the fraction is at least @xmath156 .
\(ii ) with this choice of parameters , we have , for any range @xmath21 , @xmath157 as desired .
as noted , the bounds on the sample sizes follow theorem [ theo : l : l : s](i ) .
we also note the following ( weak ) converse of theorem [ theo : carnival](ii ) : if @xmath92 is an @xmath158-approximation for @xmath2 then @xmath68 is a @xmath52-sample for @xmath2 .
indeed , we have already noted in the introduction that an @xmath158-approximation for @xmath2 is also a relative @xmath52-approximation , so the claim follows by theorem [ theo : equivalent : sampling](ii ) .
this is a weak implication , though , because , as already noted in the introduction , the bound that it implies on the size of @xmath52-samples is weaker than that given in @xcite ( see theorem [ theo : l : l : s ] ( i ) ) . [ [ sensitive - approximations . ] ] sensitive approximations .
+ + + + + + + + + + + + + + + + + + + + + + + + + we next show that the existence of sensitive @xmath18-approximations of size @xmath159 can also be established using @xmath52-samples .
the proof is slightly trickier than the preceding ones , because it uses the fact that a sample of an appropriate size is ( with high probability ) a @xmath160-sample , for an entire sequence of pairs @xmath160 .
the bound yielded by the following theorem is in fact ( slightly ) better than the bound established by @xcite , as mentioned in the introduction .
let @xmath2 be a range space , as above , and let @xmath144 . a random sample from @xmath3 of size @xmath161 is a sensitive @xmath18-approximation , with probability @xmath162 .
[ theo : sensitive ] put @xmath163 , @xmath164 , for @xmath165 .
note that @xmath166 for each @xmath167 .
let @xmath68 be a random sample of size @xmath168 theorem [ theo : l : l : s ] implies that , with an appropriate choice of the constant of proportionality , the following holds : for each @xmath167 , @xmath68 is a @xmath160-sample , with probability at least @xmath169 .
hence , with probability at least @xmath170 , @xmath68 is a @xmath160-sample for every @xmath167 .
now consider any range @xmath11 , and put @xmath171 , @xmath172 .
let @xmath167 be the index satisfying @xmath173 .
assume first that @xmath174 , so we have @xmath175 , and thus @xmath176 since @xmath68 is a @xmath160-sample , we have @xmath177 if @xmath178 then this implies @xmath179 so sensitivity holds in this case . otherwise , if @xmath180 , then @xmath181 since @xmath182 .
hence , by eq .
( [ equation : stupid ] ) , we have @xmath183 so sensitivity holds in this case too .
finally , assume @xmath184 , so @xmath185 .
in this case we have @xmath186 if @xmath187 then @xmath188 as required . if @xmath189 then we have @xmath190 showing that sensitivity holds in all cases .
it is ineresting to note that the bound on the size of sensitive @xmath18-approximations can not be improved ( for general range spaces with bounded vc - dimension ) .
this is because a sensitive @xmath18-approximation is also an @xmath191-net , and there exist range spaces ( of any fixed vc - dimension @xmath29 ) for which any @xmath191-net must be of size @xmath192 @xcite .
[ [ from - sensitive - to - relative - approximations . ] ] from sensitive to relative approximations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our next observation is that sensitive approximations are also relative approximations , with an appropriate calibration of parameters . in a way , this can be regarded as a converse of theorem [ theo : sensitive ] , which shows that a set which is simultaneously a relative approximation ( i.e. , a @xmath193-sample ) for an entire appropriate sequence of pairs of parameters is a sensitive approximation .
specifically , we have : let @xmath194 be given parameters , and set @xmath195 . then , if @xmath92 is a sensitive @xmath196-approximation for @xmath2 , it is also a relative @xmath0-approximation for @xmath2 .
[ theo : sensitive : to : relative ] we are given that @xmath197 for each @xmath11 .
now let @xmath21 be a range with @xmath198 , that is , @xmath199 , for some @xmath107
. then @xmath200 similarly , if @xmath96 , then @xmath201 hence , @xmath68 is a relative @xmath0-approximation .
this observation implies that one can compute relative @xmath0-approximations efficiently , in a deterministic fashion , using the algorithms in @xcite for deterministic construction of sensitive approximations .
we thus obtain the following result .
let @xmath202 be a range space with finite vc - dimension @xmath29 , where @xmath203 , and let @xmath204 be given parameters .
then one can construct a relative @xmath0-approximation for @xmath2 of size @xmath205 , in @xmath206 deterministic time .
[ lemma : r : e : build ]
[ sec : plane ] in this section , we present a construction of smaller - size relative @xmath0-approximations for the range space involving a set of points in the plane and halfplane ranges
. the key ingredient of the construction is the result of the following subsection , interesting in its own right .
we derive a refined `` weight - sensitive '' version of the classical construct of _ spanning trees with small crossing number _ , as obtained by chazelle and welzl @xcite , with a simplified construction given in @xcite .
we believe that this refined version is of independent interest , and expect it to have additional applications . in accordance with standard notation used in the literature ,
we denote from now on the underlying point set by @xmath1 .
we first recall the standard result : let @xmath1 be a set of @xmath4 points in @xmath8 .
then there exists a straight - edge spanning tree @xmath207 of @xmath1 such that each hyperplane in @xmath8 crosses at most @xmath23 edges of @xmath207 .
[ theo : crossing ] let @xmath1 be a set of @xmath4 points in the plane .
for a line @xmath208 , let @xmath209 ( resp . ,
@xmath210 ) be the number of points of @xmath1 lying above ( resp . , below or on ) @xmath208 , and define the _ weight _ of @xmath208 , denoted by @xmath211 , to be @xmath212 .
let @xmath213 be the intersection of all closed halfplanes that contain at least @xmath214 points of @xmath1 .
note that , by the centerpoint theorem ( see @xcite ) , @xmath215 is not empty for @xmath216 .
moreover , @xmath215 is a convex polygon , since it is equal to the intersection of a finite number of halfplanes .
( indeed , it is equal to the intersection of all halfplanes containing at least @xmath214 points of @xmath1 and bounded by lines passing through a pair of points of @xmath1 . )
the region @xmath215 can be interpreted as a level set of the _ tukey depth _ induced by @xmath1 ; see @xcite .
let @xmath1 be a set of @xmath4 points in the plane .
( i ) any line @xmath208 that avoids the interior of @xmath215 has weight @xmath217 .
( ii ) any line @xmath208 that intersects the interior of @xmath215 has weight @xmath218 .
[ lemma : weight : line ] \(i ) translate @xmath208 in parallel until it supports @xmath215 . the new line @xmath219 must pass through a vertex @xmath220 of @xmath215 which is the intersection of two lines bounding two respective closed halfplanes , each having @xmath82 points in its complement . thus , the union of the complements of these two halfplanes contains at most @xmath221 points , and it contains @xmath219 and @xmath208 .
thus , @xmath208 has at most @xmath221 points on one of its sides .
\(ii ) the second claim is easy : if the weight of @xmath208 were at most @xmath82 then , by definition , the interior of @xmath215 would be completely contained on one side of @xmath208 .
the set @xmath222 can be covered by pairwise openly disjoint triangles @xmath223 , each containing at most @xmath221 points of @xmath224 , such that _ any _ line intersects at most @xmath225 of these triangles .
moreover , @xmath226 , for each @xmath227 .
[ lemma : cover ] we construct polygons @xmath228 iteratively , as follows .
let @xmath229 and @xmath230 be the two vertical lines supporting @xmath215 on its left and on its right , respectively .
the polygon @xmath231 ( resp . , @xmath232 ) is the halfplane to the left ( resp .
, right ) of @xmath229 ( resp . , @xmath230 ) .
the construction maintains the invariant that the complement of the union of the polygons @xmath233 constructed so far is a convex polygon @xmath234 that contains @xmath215 and each edge of the boundary of @xmath234 passes through some vertex of @xmath215 , so that @xmath235 consists of pairwise disjoint connected `` pockets '' .
( initially , after constructing @xmath231 and @xmath232 , we have two pockets the regions lying respectively above and below @xmath215 , between @xmath229 and @xmath230 . ) and the decomposition of its complement . ]
each step of the construction picks a pocket that contains more than @xmath221 points of @xmath1 , finds a line @xmath208 that supports @xmath215 at a vertex of the pocket , and subdivides the pocket into two sub - pockets and a third piece that lies on the other side of @xmath208 .
the line @xmath208 is chosen so that the two resulting sub - pockets contain an equal number of points of @xmath1 . the third piece , which clearly contains at most @xmath221 points of @xmath1 ( see lemma [ lemma : weight : line ] ) , is taken to be the next polygon @xmath236 , and the construction continues in this manner until each pocket has at most @xmath221 points .
we refer to the polygons @xmath237 constructed up to this point as _ non - terminal_. we then terminate the construction , adding all the pockets to the output collection of polygons , referring to them as _ terminal _ polygons . note that each non - terminal @xmath228 is a ( possibly unbounded ) triangle , having a `` base '' whose relative interior passes through a vertex of @xmath215 , and two other sides , each of which is a portion of a base of an earlier triangle .
the terminal polygons are _ pseudo - triangles _
, each bounded by two straight edges and by a `` base '' which is a connected portion of the boundary of @xmath215 , possibly consisting of several edges . see figure [ cuts ] .
we claim that each line @xmath208 intersects at most @xmath238 polygons . for this , define the _ weight _ @xmath239 of a non - terminal polygon @xmath228 , for @xmath240 , to be the number of points of @xmath1 in the pocket that was split when @xmath228 was created ; the weight of each terminal polygon is the number of points of @xmath1 that it contains , which is at most @xmath221 .
define the _ level _ of @xmath228 to be @xmath241 .
it is easily checked that @xmath208 crosses at most two terminal polygons ( two if it crosses @xmath215 and at most one if it misses @xmath215 ) , and it can cross both ( non - base ) sides of at most one ( terminal or non - terminal ) polygon .
any other polygon @xmath228 crossed by @xmath208 is such that @xmath208 enters it through its base , reaching it from another polygon whose level is , by construction , strictly smaller than that of @xmath228 .
since there are only @xmath242 distinct levels , the claim follows .
it is easy to verify that the convex hulls @xmath243 are triangles with pairwise disjoint interiors , which have the required properties .
let @xmath244 be a prespecified parameter .
one can construct a spanning tree @xmath207 for @xmath245 , such that each line intersects at most @xmath246 edges of @xmath207 .
[ lemma : level : k ] construct the decomposition of @xmath222 into @xmath247 covering triangles @xmath248 , using lemma [ lemma : cover ] . for each @xmath227
, construct a spanning tree @xmath249 of @xmath250 with crossing number @xmath251 , using theorem [ theo : crossing ] .
in addition , connect one point of @xmath250 to an arbitrary vertex of @xmath252 .
since @xmath253 and @xmath254 are disjoint , for @xmath255 , it follows that no pair of edges of any pair of ( the plane embeddings of the ) trees among @xmath256 cross each other .
let @xmath257 be the planar straight - line graph formed by the union of @xmath258 , plus the connecting segments just introduced , and let @xmath259 be a spanning tree of @xmath257 ; the vertex set of @xmath259 contains @xmath222 .
let @xmath208 be a line in the plane .
the proof of the preceding lemma implies that @xmath208 intersects at most @xmath238 of the polygons @xmath260 .
hence , @xmath208 crosses at most two edges of @xmath261 , at most @xmath262 of the connecting segments , and it can cross edges of at most @xmath262 trees @xmath249 , for @xmath227 . since @xmath208 crosses at most @xmath251 edges of each such tree , we conclude that @xmath208 crosses at most @xmath246 edges of @xmath259 . finally , we get rid of the extra `` steiner vertices '' of @xmath259 ( those not belonging to @xmath222 ) in a straightforward manner , by making @xmath259 a rooted tree , at some point of @xmath222 , and by replacing each path connecting a point @xmath263 to an ancestor @xmath264 , where all inner vertices of the path are steiner points , by the straight segment @xmath265 .
this produces a straight - edge spanning tree @xmath207 of @xmath222 , whose crossing number is at most that of @xmath259 . given a set @xmath1 of @xmath4 points in the plane , one can construct a spanning tree @xmath207 for @xmath1 such that any line @xmath208 crosses at most @xmath266 edges of @xmath207 . the tree @xmath207 can be constructed in @xmath267 ( deterministic ) time , for any fixed @xmath144 .
[ theo : good : partition ] we construct a sequence of subsets of @xmath1 , as follows .
put @xmath268 . at the @xmath167thstep , @xmath269 ,
consider the polygon @xmath270 , and let @xmath271 .
we stop when @xmath272 becomes empty . by construction ,
the @xmath167th step removes at least @xmath273 points from @xmath274 , so @xmath275 , and the process terminates in @xmath276 steps . for each @xmath167 , construct a spanning tree @xmath249 for @xmath277 , using lemma [ lemma : level : k ] ( with @xmath278 ) .
connect the resulting trees by straight segments into a single spanning tree @xmath207 of @xmath1 .
we claim that @xmath207 is the desired spanning tree .
indeed , consider an arbitrary line @xmath208 of weight @xmath82 .
observe that @xmath208 can not cross any of the polygons @xmath279 , for @xmath280 , since any line that crosses such a polygon must be of weight at least @xmath281 , by lemma [ lemma : weight : line](ii ) .
thus @xmath208 crosses only the first @xmath6 layers of our construction .
hence , the number of edges of @xmath207 that @xmath208 crosses is at most @xmath282 as is easily verified .
this establishes the bound on the crossing number of @xmath283 .
[ [ running - time . ] ] running time .
+ + + + + + + + + + + + + computing @xmath279 can be done in the dual plane , by constructing the convex hulls of the levels @xmath273 and @xmath284 in the arrangement of the lines dual to the points of @xmath274 .
we use the algorithm of matouek @xcite , which constructs the convex hull of a level in @xmath285 time , for a total of @xmath286 time .
this also subsumes the time needed to construct @xmath272 from @xmath274 , for all @xmath167 .
next , we carry out the constructive proof of lemma [ lemma : cover ] for @xmath287 , which can be implemented to run in @xmath288 time .
finally , for each set in the cover , we apply the algorithm of @xcite to construct the corresponding subtree with low crossing number .
this takes @xmath289 time for a set of size @xmath290 , for any fixed @xmath144 .
we continue in this fashion , as described in the first part of the proof .
it is now easy to verify that the resulting construction takes overall @xmath267 time , for any fixed @xmath144 .
the main technical step in the construction of theorem [ theo : good : partition ] is the partition of the set of _
points of @xmath1 , namely , those that are contained in some halfplane with at most @xmath82 points of @xmath1 , into subsets , each containing at most @xmath221 points , so that any line crosses the convex hulls of at most @xmath291 of these subsets .
it is natural to try to extend this construction to three ( or higher ) dimensions .
however , as we show next , there are examples of point sets in @xmath86 where no such partition exists , even when the points are in convex position . to see this , let @xmath290 be an integer , and put @xmath292 .
define @xmath293 that is , @xmath1 is the set of the vertices of the @xmath294 integer grid in the @xmath295-plane , lifted to the standard ( convex ) paraboloid @xmath296 .
thus , all points of @xmath1 are in convex position ( and are thus @xmath109-shallow ) .
let @xmath297 be an arbitrary parameter , and consider any partition @xmath298 of @xmath1 into @xmath299 sets , where @xmath300 , for @xmath227 .
let @xmath260 denote the two - dimensional convex hull of the projection of @xmath279 onto the @xmath295-plane , for @xmath301 .
the sum of the @xmath302-span and the @xmath303-span of @xmath260 is at least @xmath304 ( or else there would be no room for @xmath279 to contain all its points ) .
hence , the total length of these spans of @xmath248 is at least @xmath305 .
consider the @xmath306 vertical planes @xmath307 , and @xmath308 .
clearly , the overall number of intersection points between the boundaries of the @xmath260 s and these planes is proportional to the sum of their @xmath302-spans and @xmath303-spans .
hence , there is a plane in this family that intersects @xmath309 sets among @xmath248 , and thus it intersects the @xmath310 corresponding convex hulls among @xmath311 .
note that a similar argument can be applied to the set of the vertices of the @xmath312 integer lattice . in this case , for any partition of this set of the above kind , there always exists a plane that crosses the convex hulls of at least @xmath313 of the subsets .
( this matches the upper bound in the partition theorem of matouek @xcite . ) of course , here , ( most of ) the points are not shallow . to summarize , there exist sets @xmath1 of @xmath4 points in convex position in 3-space ( so they are all @xmath109-shallow ) , such that any partition of @xmath1 into sets of size ( roughly ) @xmath82 will have a plane that crosses at least @xmath314 sets in the partition .
( without the convex position , or shallowness , assumption , there exist sets for which this crossing number is at least @xmath313 . ) we do not know whether this lower bound is worst - case tight .
that is , can a set @xmath1 of @xmath4 @xmath82-shallow points in 3-space be partitioned into @xmath315 subsets , so that no plane separates more than @xmath316 subsets ?
if this were the case , applying the standard construction of spanning trees with small crossing numbers to each subset would result in a spanning tree with crossing number @xmath317 .
note that this still leaves open the ( more modest ) possibility of a partition which is `` depth sensitive '' , that is , a partition into subsets of size roughly @xmath82 , with the property that any halfspace that contains @xmath290 points crosses at most ( or close to ) @xmath318 sets in the partition .
we can turn the above construction of a spanning tree with small relative crossing number into a construction of a relative @xmath0-approximation for a set of points in the plane and for halfplane ranges , as follows .
let @xmath1 be a set of @xmath4 points in the plane , and let @xmath207 be a spanning tree of @xmath1 as provided in theorem [ theo : good : partition ] .
we replace @xmath207 by a perfect matching @xmath319 of @xmath1 , with the same relative crossing number , i.e. , the number of pairs of @xmath319 that are separated by a halfplane of weight @xmath82 is at most @xmath83 .
this is done in a standard manner
we first convert @xmath207 to a spanning path whose relative crossing number is at most twice larger than the crossing number of @xmath207 , and then pick every other edge of the path .
we now construct a coloring of @xmath1 with low discrepancy , by randomly coloring the points in each pair of @xmath319 .
specifically , each pair is randomly and independently colored either as @xmath320 or as @xmath321 , with equal probability .
the standard theory of discrepancy ( see @xcite ) yields the following variant .
given a set @xmath1 of @xmath4 points in the plane , one can construct a coloring @xmath322 , such that , for any halfplane @xmath323 , @xmath324 the coloring is balanced
each color class consists of exactly @xmath325 points of @xmath1 .
[ lemma : coloring ] as shown in @xcite , if a halfplane @xmath323 crosses @xmath15 edges of the matching , then its discrepancy is @xmath326 , with high probability . as shown above ,
@xmath327 , for @xmath328 , so the discrepancy of @xmath323 is , with high probability , @xmath329 .
we need the following fairly trivial technical lemma . for any @xmath330 , @xmath331 , and @xmath332
, we have @xmath333 .
[ lemma : stupid : inequality ] observe that @xmath334 .
as we next show , the improved discrepancy bound of lemma [ lemma : coloring ] leads to an improved bound on the size of @xmath52-samples for our range space , and , consequently , for the size of relative @xmath0-approximations . given a set @xmath1 of @xmath4 points in the plane , and parameters @xmath128 and @xmath335
, one can construct a @xmath336-sample @xmath337 of size @xmath338 .
[ theo : sample : half : planes ] following one of the classical constructions of @xmath18-approximations ( see @xcite , we repeatedly halve @xmath1 , until we obtain a subset of size as asserted in the theorem , and then argue that the resulting set is a @xmath52-sample .
formally , set @xmath268 , and partition @xmath274 into two equal halves , using lemma [ lemma : coloring ] ; let @xmath272 and @xmath339 denote the two halves ( consisting of the points that are colored @xmath340 , @xmath341 , respectively ) .
we keep @xmath272 , remove @xmath342 , and continue with the halving process .
let @xmath343 denote the size of @xmath272 .
for any halfplane @xmath323 , we have @xmath344 where @xmath57 is some appropriate constant . recalling that @xmath345 and that @xmath346 , this can be rewritten as @xmath347 since @xmath348 , we have @xmath349 since @xmath350 , we have @xmath351 applying lemma [ lemma : stupid : inequality ] , with @xmath352 , @xmath353 , and @xmath354 , the last expression is at most @xmath355
this implies that @xmath356 .
the triangle inequality then implies that @xmath357 for @xmath358 .
the theorem then follows by taking @xmath68 to be the smallest @xmath272 which still satisfies this size constraint .
using theorem [ theo : equivalent : sampling ] , we thus obtain : given a set @xmath1 of @xmath4 points in the plane , and parameters @xmath13 and @xmath48 , one can construct a relative @xmath0-approximation @xmath337 of size @xmath359 .
[ cor : relative : error : half : planes ] * remark : * one can speed up the construction of the relative @xmath0-approximation of corollary [ cor : relative : error : half : planes ] , by first drawing a random sample of slightly larger size , which is guaranteed , with high probability , to be a relative approximation of the desired kind , and then use halving to decimate it to the desired size .
implemented carefully , this takes @xmath360 time , and thus produces a relative @xmath0-approximation , with high probability , of the desired size @xmath359 .
in fact , the preceding analysis leads to an improved bound for sensitive approximations for our range space .
the improvement is in terms of the quality of the `` sensitivity '' of the approximation , which is achieved at the cost of a slight increase ( by a sublogarithmic factor ) in its size , as compared to the standard bound , provided in theorem [ theo : sensitive ] .
that is , we have : let @xmath1 be a set of @xmath4 points in the plane , and let @xmath361 be a parameter .
one can compute a subset @xmath362 of size @xmath363 , such that for any halfplane @xmath323 , we have @xmath364 .
fix parameters @xmath128 and @xmath99 , and apply the construction of theorem [ theo : sample : half : planes ] until we get a subset @xmath68 of size @xmath365 ; the constant of proportionality will be determined by the forthcoming considerations .
the key observation is that @xmath68 is a @xmath52-sample for any @xmath128 and @xmath99 that satisfy @xmath366 , because the construction is oblivious to the individual values of @xmath55 and @xmath54 , and just requires that the size of the sample remains larger than the above threshold . using this observation
, we proceed to show that @xmath68 satisfies the property asserted in the theorem .
so let @xmath323 be a halfplane .
suppose first that @xmath367 . by corollary [ cor :
relative : error : half : planes ] , @xmath68 is a relative @xmath368-approximation to @xmath1 , with an appropriate choice of the constant of proportionality ( note the change of roles of `` @xmath53 '' and `` @xmath18 '' ) .
this implies that @xmath369 .
suppose then that @xmath370 , and set @xmath371 .
observe that @xmath372 thus , with an appropriate choice of the constant of proportionality for @xmath290 , @xmath68 is a relative @xmath373-approximation , which implies that @xmath374 as asserted .
this compares favorably with the result of brnnimann_et al . _
@xcite , which in this case implies that there exists a subset of size @xmath375 such that , for each halfplane @xmath323 , we have @xmath376 . our bound is smaller when @xmath377 and is the same otherwise .
[ sec : higher : sec ] the construction in higher dimensions is different from the planar one , because of our present inability to extend the construction of spanning trees with low relative crossing number to three or higher dimension .
for this reason we use the following different strategy .
we say that a hyperplane @xmath323 _ separates _ a set @xmath378 if @xmath323 intersects the interior of @xmath379 ; that is , each of the open halfspaces that @xmath323 bounds intersects @xmath380 .
the main technical step in the construction is given in the following theorem .
let @xmath1 be a set of @xmath4 points in @xmath86 , and let @xmath13 , @xmath48 be given parameters .
then there exists a set @xmath337 , of size @xmath381 , such that , for any halfspace @xmath323 with @xmath382 , we have @xmath383 [ theo : a3d ] * remark : * let us note right away the difference between eq .
( [ equation : a1 ] ) and the situation in the preceding sections .
that is , up to now we have handled ranges of measure _ at least _
@xmath53 , whereas eq . (
[ equation : a1 ] ) applies to ranges of measure _ at most _ @xmath53 .
this issue requires a somewhat less standard construction , that will culminate in a _ sequence _ of approximation sets , each catering to a different range of halfspace measures .
nevertheless , the overall size of these sets will satisfy the above bound , and the cost of accessing them will be small .
put @xmath384 , and apply the shallow partition theorem of matouek @xcite , to obtain a partition of @xmath1 into @xmath385 subsets @xmath386 , each of size between @xmath387 and @xmath221 , such that any _ @xmath82-shallow _
halfspace @xmath323 ( namely , a halfspace that contains at most @xmath82 points of @xmath1 ) separates at most @xmath388 subsets , for some absolute constant @xmath57 .
( note that if @xmath323 meets any @xmath272 , it has to separate it , because @xmath323 is too shallow to fully contain @xmath272 . ) without loss of generality , we can carry out the construction so that the size of each @xmath272 is even .
we then construct , for each subset @xmath272 , a spanning tree of @xmath272 with crossing number @xmath389 @xcite , and convert it , as in the preceding section , to a perfect matching of @xmath272 , with the same asymptotic bound on its crossing number , which is the maximum number of pairs in the matching that a halfspace separates .
we combine all these perfect matchings to a perfect matching of the entire set @xmath1 . we then color each matched pair independently , as above , coloring at random one of its points by either @xmath341 or @xmath340 , with equal probabilities , and the other point by the opposite color .
let @xmath390 be the set of points colored @xmath341 ; we have @xmath391 . with high probability , the discrepancy of any halfspace @xmath323 is at most @xmath392 , where @xmath393 is the crossing number of @xmath323 ( see @xcite ) ; we may assume that the coloring does indeed have this property .
( if we do not care about the running time , we can verify that the constructed set has the required property , and if not regenerate it . ) hence , if @xmath323 is a @xmath82-shallow halfspace , then , by construction , @xmath394 , because @xmath323 separates @xmath395 subsets and crosses @xmath389 edges of the spanning tree of each of them .
hence the discrepancy of any @xmath82-shallow halfspace @xmath323 is @xmath396 .
we continue recursively in this manner for @xmath397 steps , producing a sequence of subsets @xmath398 , where @xmath399 is obtained from @xmath400 by applying the partitioning of @xcite with a different parameter @xmath401 , and then by using the above coloring procedure on the resulting perfect matching .
we take @xmath402 , where @xmath403 is the constant derived in the following lemma .
( the bound asserted in the lemma holds with high probability if we do not verify that our colorings have small discrepancy , and is worst - case if we do verify it . )
there exists an absolute constant @xmath403 such that any @xmath74-shallow halfspace satisfies , for any @xmath404 , @xmath405 where @xmath397 is the largest index satisfying @xmath406 , where @xmath407 .
[ lemma : relative : higher : dim ] delegated to appendix [ apnd : proof : a ] .
the lemma implies that @xmath323 is @xmath408-shallow in each of the subsets @xmath409 , so we can use the above bound on the discrepancy of @xmath323 with respect to each of these subsets .
( the reader can note the similarity between the forthcoming analysis and the proof of lemma [ lemma : relative : higher : dim ] . )
we thus have @xmath410 substituting @xmath411 , for each @xmath167 , and adding up the inequalities , it is easily checked that the last right - hand side dominates the sum ( compare with the analysis in the proof of theorem [ theo : good : partition ] ) , so we obtain , using the triangle inequality , @xmath412 we choose @xmath397 to be the largest index for which this bound is at most @xmath413 .
that is , @xmath414 .
note that this can be rewritten as @xmath415 , which implies that @xmath406 , as required in lemma [ lemma : relative : higher : dim ] , provided that @xmath18 is smaller than some appropriate absolute constant .
hence , since @xmath397 was chosen as large as possible , the size of @xmath416 is @xmath417 taking @xmath418 completes the proof of theorem [ theo : a3d ] .
[ [ construction . ] ] construction .
+ + + + + + + + + + + + + the preceding construction used a fixed @xmath419 , and assumed that the query halfspace is _ at most _ @xmath82-shallow .
however , our goal is to construct a subset that caters to all halfspaces whose measure is _ at least _ some given threshold . while unable to meet this goal exactly , with a single subset , we almost get there , in the following manner .
let @xmath53 be the given threshold parameter .
we consider the geometric sequence @xmath420 , where @xmath421 ; the last ( largest ) element is @xmath422 , and its index is @xmath423 .
for each @xmath15 , we construct a relative @xmath424-approximation @xmath425 for @xmath1 , as in theorem [ theo : a3d ] , where @xmath57 is a sufficiently small constant , whose value will be determined later . clearly , the overall size of all these sets is dominated by the size of the first set , namely , it is @xmath426 we output the entire sequence @xmath427 , as a substitute for a single relative @xmath0-approximation , and use it as follows .
[ [ answering - a - query . ] ] answering a query .
+ + + + + + + + + + + + + + + + + + let @xmath323 be a given halfspace , so that @xmath428 .
let @xmath429 be the index ( initially unknown ) for which @xmath430 .
thus @xmath323 is @xmath431-shallow , and is also @xmath432-shallow , for every @xmath433 .
hence , if we use the set @xmath434 , for each @xmath433 , to approximate @xmath435 , we get , by theorem [ theo : a3d ] , a count @xmath436 , which satisfies @xmath437 . in other words
, we have @xmath438for each @xmath433 . to answer the query , we access the sets @xmath439 , in decreasing order , and find the largest index @xmath101 satisfying @xmath440 we return @xmath441 as the desired approximate count . [ [ analysis . ] ] analysis .
+ + + + + + + + + we claim that eq .
( [ equation : c : s : hold ] ) must hold at @xmath442 , assuming @xmath443 . indeed , since @xmath444 , we have @xmath445 hence , @xmath446 . on the other hand , @xmath447 . combining these two inequalities ,
we obtain @xmath448 if @xmath443 , as assumed .
our choice of @xmath101 thus satisfies @xmath433 .
moreover , we have @xmath449 this determines @xmath435 , up to a factor of @xmath450 , which thus determines the correct index @xmath15 ( up to @xmath451 ) .
in fact , as our query answering procedure actually does , we do not have to find the exact value of @xmath15 , because we use a smaller value of @xmath18 in the construction of the sets @xmath434 . specifically , with an appropriate choice of @xmath57 , we have @xmath452 , so @xmath453 and , similarly , @xmath454 so @xmath441 is an @xmath18-approximate count of @xmath455 , establishing the correctness of our procedure .
( the specific choice of @xmath57 , which we do not spell out , can easily be worked out from the preceding analysis . )
note that our structure also handles halfspaces @xmath323 with @xmath456 .
specifically , if we find an index @xmath101 satisfying eq .
( [ equation : c : s : hold ] ) then @xmath441 is an @xmath18-approximate count of @xmath455 , as the preceding analysis shows .
if no such @xmath101 is found then we must have @xmath457 ( otherwise , as just argued , there would exist such an @xmath101 and the procedure would find it ) . in this case
we have @xmath458 , and we return @xmath459 with the guarantee that ( a ) @xmath460 , and ( b ) @xmath458 . note that this constitutes a somewhat unorthodox approach we have logarithmically many sets instead of a single one ( although their combined size is asymptotically the same as that of the largest one ) , and we access them sequentially to find the one that gives the best approximation . an interesting useful feature of the construction is that , if the given halfspace @xmath323 has weight @xmath435 that satisfies @xmath461 , then the approximate counting mechanism accesses sets whose overall size is only @xmath462 .
that is , the larger @xmath435 is , the faster is the procedure . to summarize , we have shown : given a set @xmath1 of @xmath4 points in @xmath86 , and two parameters @xmath13 , @xmath48 , we can construct @xmath463 subsets of @xmath1 , @xmath464 , of total size @xmath465 , so that , given any halfspace @xmath323 containing @xmath466 points of @xmath1 , we can find a set @xmath425 that satisfies @xmath467 the ( brute - force ) time it takes to search for @xmath425 and obtain the count @xmath468 is @xmath469 [ sec : higher ] the preceding construction can be generalized to higher dimensions , with some complications .
we first introduce the following parameters : @xmath470 note that , for @xmath471 , @xmath472 ( and tends to @xmath473 as @xmath474 increases ) , and @xmath475 ( and tends to @xmath473 as @xmath474 increases ) .
the analogous version of theorem [ theo : a3d ] is : let @xmath1 be a set of @xmath4 points in @xmath8 , @xmath471 , and let @xmath13 , @xmath48 be given parameters .
then there exists a set @xmath337 , of size @xmath476 , such that , for any @xmath74-shallow halfspace @xmath323 , we have @xmath477 provided that @xmath478 .
[ theo : add ] as above , put @xmath479 , and apply matouek s shallow partition theorem @xcite , to obtain a partition of @xmath1 into @xmath480 subsets @xmath386 , each of size between @xmath387 and @xmath221 , such that any @xmath82-shallow halfspace separates at most @xmath481 subsets , for some absolute constant @xmath57 , where @xmath482 .
( as above , if @xmath323 meets any @xmath272 , it has to separate it . ) also , we may assume that the size of each @xmath272 is even .
we then construct , for each subset @xmath272 , a spanning path of @xmath272 with crossing number @xmath483 @xcite , for @xmath484 , convert it to a perfect matching of @xmath272 , with the same asymptotic crossing number , and combine all these matchings to a perfect matching of @xmath1 .
we then apply the same coloring scheme as in the three - dimensional case , and let @xmath390 be the set of points colored by @xmath341 ; we have @xmath485 . with high probability
, the discrepancy of any halfspace @xmath323 is at most @xmath486 , where @xmath393 is the number of pairs in the matching that @xmath323 separates @xcite . if @xmath323 is @xmath82-shallow then , by construction , @xmath487 .
hence the discrepancy of @xmath323 is @xmath488 we continue recursively in this manner for @xmath397 steps , producing , as above , a sequence of subsets @xmath398 , where @xmath399 is obtained from @xmath400 by applying the partitioning of @xcite with a different parameter @xmath401 , and then by using the above coloring procedure on the resulting perfect matching .
we take @xmath402 , where @xmath403 is the constant derived in the following lemma .
( as in the 3-dimensional case , if we want the proof of the theorem to be constructive , we either verify that each half - sample has the desired low discrepancy , and then the bounds are worst - case , or else the bounds hold with high probability . ) there exists an absolute constant @xmath403 such that any @xmath74-shallow halfspace satisfies , for any @xmath404 , @xmath405 where @xmath397 is the largest index satisfying @xmath489 , where @xmath407 .
[ lemma : k : o : k : d ] delegated to appendix [ apnd : k : o : k : d ] . as in the 3-dimensional case ,
the lemma justifies the following chain of inequalities ( note again the similarity between the proof of the lemma and the analysis below ) . @xmath490
substituting @xmath491 , for each @xmath167 , and adding up the inequalities , the last right - hand side dominates , so we obtain @xmath492 substituting @xmath419 and the values of @xmath302 and @xmath303 , this is equal to @xmath493 we choose the first @xmath397 so that this bound is at most @xmath51 .
that is , @xmath494 @xmath495 we note that the choice of @xmath496 satisfies the lower bound constraint in lemma [ lemma : k : o : k : d ] .
hence , taking @xmath418 completes the proof . *
obtaining a relative approximate count * is done exactly as in the three - dimensional case , producing a sequence of approximations , and searching through the sequence for the approximation which caters for the correct range of the size of @xmath455 .
we thus have the following result .
given a set @xmath1 of @xmath4 points in @xmath8 , and two parameters @xmath13 , @xmath48 , we can construct @xmath463 subsets of @xmath1 , @xmath464 , of total size @xmath497 , so that , given any halfspace @xmath323 containing @xmath466 points of @xmath1 , we can find a set @xmath425 that satisfies @xmath498 the ( brute - force ) time it takes to search for @xmath425 and obtain the count @xmath468 is @xmath499 [ [ discussion . ] ] discussion .
+ + + + + + + + + + + we have two competing constructions , the `` traditional '' one , with @xmath500 elements ( see section [ sec : relations ] ) , and the new one , with @xmath501 elements . the new construction is better , in terms of the size of the approximation , when @xmath502 ( for simplicity , we ignore the constants of proportionality ) . for further simplicity , assume that @xmath53 is not much larger than @xmath18 , so that @xmath503 and @xmath504 are roughly the same , up to some constant factor .
then we replace the above condition by @xmath505 substituting the values of @xmath506 and @xmath507 , and simplifying the expressions , this is equivalent to @xmath508 this establishes a lower bound for @xmath53 , above which the new construction takes over . for example , for @xmath509 , @xmath53 has to be @xmath510 .
[ sec : approximate : counting ] in this section we slightly deviate from the main theme of the paper . since approximate range counting is one of the main motivations for introducing relative @xmath0-approximations , we return to this problem , and propose efficient solutions for approximate halfspace range counting in two and three dimensions . the solution to the 3-dimensional problem uses relative approximations , whereas the solution to the 2-dimensional problem is simpler and does not require such approximations .
both solutions pass to the dual plane / space , construct a small subset of levels , of small overall complexity , in the arrangement of the dual lines / planes , and search through them with the point dual to the query halfplane / halfspace to retrieve the approximate count .
we first present two solutions for the planar case , and then consider the 3-dimensional case .
let @xmath1 be a set of @xmath4 points in the plane in general position , and @xmath144 be a prescribed parameter .
the task at hand is to preprocess @xmath1 for halfplane approximate range counting ; that is , given a query halfplane @xmath323 , we wish to compute a number @xmath511 that satisfies @xmath512 .
we can reformulate the problem in the dual plane , where the problem is to preprocess the set @xmath513 of lines dual to the points of @xmath1 for approximate vertical - ray range counting queries ; that is , given a vertical ray @xmath514 , we want to count ( approximately , within a relative error of @xmath18 ) the number of lines of @xmath513 that intersect @xmath514 . without loss of generality , we only consider downward - directed rays .
let @xmath515 denote the @xmath167thlevel in the arrangement @xmath516 ; this is the closure of the set of all the points on the lines of @xmath513 that have exactly @xmath167 lines of @xmath513 passing below them .
each @xmath515 is an @xmath302-monotone polygonal curve , and its _ combinatorial complexity _ ( or just _ complexity _ ) is the number of its vertices . for integers @xmath517 , the total complexity of the levels of @xmath516 in the range @xmath518 $ ]
is @xmath519 . in particular , the average complexity of a level in this range is @xmath520 .
[ lemma : levels ] this result is a strengthening of a similar albeit weaker bound due to welzl @xcite , and is implicit in @xcite .
it was recently rederived , in a more general form , in an unpublished m.sc .
thesis by kapelushnik @xcite .
we sketch the proof for the sake of completeness .
consider the primal setting , and connect two points @xmath521 by an edge , if the open halfplane bounded by the line through @xmath247 and @xmath220 and lying below that line contains exactly @xmath397 points of @xmath1 ( we refer to @xmath522 as a _ @xmath397-set _ ) , where @xmath523 $ ] .
let @xmath524 denote the resulting set of edges .
all the edges of @xmath524 that are @xmath397-sets can be decomposed into @xmath525 concave chains ( see , e.g. , @xcite ) .
similarly , they can be decomposed into @xmath526 convex chains .
overall , the edges of @xmath524 can be decomposed into at most @xmath527 concave chains , and into at most @xmath528 convex chains .
each pair of a convex and a concave chain can intersect in at most two points .
this implies that the segments of @xmath524 can cross each other at most @xmath529 times .
on the other hand , consider the ( straight - edge plane embedding of the ) graph @xmath530 .
it has @xmath4 vertices , and @xmath531 edges . by the classical crossing lemma ( see @xcite ) , it has @xmath532 crossing pairs of edges ( assuming @xmath533 ) .
we thus have @xmath534 or @xmath535 .
the second claim in the lemma is an immediate consequence of this bound .
\(i ) for each @xmath536 , for @xmath13 , and for any fixed positive constant @xmath403 , let @xmath82 be an integer chosen randomly and uniformly in the range @xmath537 $ ] .
then the expected complexity of @xmath538 is @xmath539
. this bound also holds , with an appropriate choice of the constant of proportionality , with probability @xmath540 .
\(ii ) if @xmath538 has complexity @xmath247 then it can be replaced by an @xmath302-monotone polygonal curve with @xmath541 edges , which lies between the two curves @xmath542 and @xmath543 .
[ claim : easy ] the first claim is an immediate consequence of lemma [ lemma : levels ] by setting @xmath544 and @xmath545 .
the second claim is well known : the curve is obtained by shortcutting @xmath538 in `` jumps '' of @xmath546 vertices ; see , e.g. , @xcite .
we present two variants of an algorithm for the problem at hand , which differ in the dependence of their performance on @xmath18 .
the first has @xmath547 query time , but requires @xmath548 storage , while the second one uses only @xmath549 storage ( no dependence on @xmath18 ) , but its query time is @xmath550 .
we first compute the union @xmath551 of the first @xmath552 levels of @xmath516 . as is well known ( see , e.g. , @xcite ) , the overall complexity of @xmath551 is @xmath553 .
next , set @xmath554 , for @xmath555 .
let @xmath403 be a constant that satisfies @xmath556 for @xmath13 ( @xmath557 would do ) .
we pick a random level with index in the range @xmath558 ; by claim [ claim : easy](i ) , most of these levels have complexity @xmath539 .
we thus assume that the chosen level has this complexity ( or else we resample ; since the probability of success is at least @xmath156 , this does not affect the expected running time ) .
we then simplify each such level , using claim [ claim : easy](ii ) ( with @xmath559 instead of @xmath18 ) .
the resulting polygonal curve @xmath560 is easily seen to lie ( strictly ) between @xmath561 and @xmath562 , and its complexity is @xmath563 in particular , the total complexity of the curves @xmath564 is @xmath565 , and these curves are pairwise disjoint .
together with the segments in @xmath551 , they form a planar subdivision @xmath380 with @xmath548 edges , which we preprocess for efficient point location . using kirkpatrick s algorithm @xcite ,
this can be done with @xmath548 preprocessing time and storage , and a query can be answered in @xmath566 time . we also store ( with no extra asymptotic cost ) a count @xmath567 with each edge @xmath568 of @xmath380 .
it is equal to the level of @xmath568 if @xmath568 is an edge of @xmath551 , and to @xmath496 if @xmath568 is an edge of @xmath569 .
now , given a query point @xmath570 ( i.e. , a downward - directed ray emanating from @xmath570 ) , we locate @xmath570 in @xmath380 and retrieve the count @xmath567 , where @xmath568 is the edge lying directly below @xmath570 .
it is easy to verify that @xmath567 is indeed an @xmath18-approximation of the number of lines below @xmath570 .
we have thus shown : given a set @xmath1 of @xmath4 points in the plane , and a parameter @xmath13 , one can build , in @xmath571 expected time , a data - structure that uses @xmath548 space , so that , given a query halfplane @xmath323 , one can approximate @xmath572 within relative error @xmath18 , in @xmath573 time .
[ theo : approximate:2:fast ] the construction is described above , and we only bound the running time .
computing the bottom @xmath319 levels takes @xmath574 time @xcite .
computing the remaining randomly chosen @xmath575 levels requires @xmath576 time per level , using the ( very involved ) dynamic convex hull algorithm of @xcite ( or simpler earlier algorithms with a slight ( logarithmic or sub - logarithmic ) degradation in the running time ) .
overall , the running time is @xmath577 .
let @xmath319 be a random integer in the range @xmath578 $ ] .
by lemma [ lemma : levels ] , the expected complexity of the level @xmath579 is @xmath549 . by markov
s inequality , the complexity of @xmath579 is @xmath549 with probability at least @xmath156 , with an appropriate choice of the constant of proportionality .
thus , redrawing the index @xmath319 if necessary ( without affecting the expected asymptotic running time ) , we may assume that @xmath579 does have linear complexity .
next , we define the curves @xmath580 as above , with the new value of @xmath319 as the starting index .
note that each @xmath560 is a shortcutting of a random level in the range @xmath581 $ ] , where @xmath582 .
more precisely , we can regard the random choice of the level from which @xmath560 is produced as a 2-step drawing , where we first draw @xmath319 and then draw @xmath82 in the `` middle '' of the range @xmath583 $ ] , as above .
the combined drawing is not exactly uniform , but is close enough to make lemma [ lemma : levels ] and claim [ claim : easy ] hold in this scenario too .
we assume @xmath584 , and here we have @xmath585 , but the lemma continues to hold in this case too , as is easily checked . ]
hence , the expected complexity of the level corresponding to @xmath560 is @xmath549 for each @xmath167 .
thus , the overall expected complexity of the shortcut curves @xmath586 is now only @xmath587 ( with a constant of proportionality independent of @xmath18 ) .
we construct the collection of these curves , and assume ( using resampling if necessary ) that their overall complexity is indeed linear .
we preprocess the planar map formed by these curves , and by the edges of @xmath579 , for fast point location , as above , and store with each curve @xmath560 , for @xmath588 , the level @xmath589 that it approximates .
in addition , we sweep @xmath579 from left to right , and store , with each of its edges @xmath568 , the ( fixed ) set of lines passing below ( any point on ) @xmath568 .
this can be done with only @xmath549 storage , using persistence @xcite .
now , given a query point @xmath570 , we locate it in the planar map .
if it lies above @xmath590 , then the index stored at the segment lying directly below @xmath570 is an @xmath18-approximation of the number of lines below @xmath570 .
if @xmath570 lies below @xmath590 , we find the edge @xmath568 of @xmath579 lying above or below @xmath570 , retrieve the set of lines stored at @xmath568 ( using the persistent data structure ) , and search it , in @xmath591 time , to count ( exactly ) the number of lines below @xmath570 .
we thus have shown : given a set @xmath1 of @xmath4 points in the plane , and a parameter @xmath13 , one can build , in @xmath592 time , a data - structure that uses @xmath549 space , so that , given a query halfplane @xmath323 , one can approximate @xmath572 , within relative error @xmath18 , in @xmath593 time .
[ theo : approximate:2:small : space ] the construction requires the computation of @xmath594 levels , each of expected complexity @xmath549 .
thus , this takes @xmath595 time ( or slightly worse , as in the comment in the preceding proof ) .
the query time and space complexity follow from the discussion above . observe that theorem [ theo : approximate:2:fast ] and theorem [ theo : approximate:2:small : space ]
improve over the previous results in @xcite , which have query time @xmath596 .
it would also be interesting to compare these results to the recent technique of aronov and sharir @xcite ; as presented , this technique caters only to range searching in four and higher dimensions , but it can be adapted to two or three dimensions too . we can extend the above algorithms to three dimensions . after applying duality , the input is a set @xmath597 of @xmath4 planes in 3-space , which we want to preprocess for approximate vertical ray range counting .
the general idea is very similar : ( i ) compute a sequence of levels of @xmath598 , whose indices form roughly a geometric sequence .
( ii ) replace each level by a simplified @xmath295-monotone polyhedral surface which approximates it well .
( iii ) find the belt between two consecutive surfaces which contains the query point @xmath570 ( the apex of the query vertical ray ) , and thereby obtain the desired approximate count .
implementing step ( ii ) is considerably harder in three dimensions than in the plane , and we do it using an appropriate relative @xmath0-approximation . for the sake of simplicity of presentation , we do not attempt to optimize the choice of parameters , and just describe the general technique .
concrete and improved versions can be worked out by the interested reader .
[ [ approximating - a - specific - level . ] ] approximating a specific level .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider first the problem of approximating a specific level @xmath290 of @xmath598 .
consider the range space that has @xmath597 as the ground set , whose ranges are induced by vertical downward - directed rays , where the range associate with a ray @xmath514 is the subset of planes of @xmath597 crossed by @xmath514 .
this range space has finite vc - dimension , so we can apply to it the analysis of section [ sec : relations ] .
put @xmath599 , and construct a @xmath600-relative approximation @xmath601 , by taking a random sample of size @xmath602 from @xmath597 ( see theorem [ theo :
l : l : s ] and @xcite ) ; with high probability , the sample is indeed such an approximation .
set @xmath603 . by construction , the @xmath604thlevel @xmath605 of @xmath606
is guaranteed to lie between the levels @xmath607 and @xmath608 of @xmath598 , so it provides an adequate approximation to the @xmath290thlevel of @xmath598 . since @xmath54 is `` small '' , we can compute @xmath605 in time @xmath609 , using , e.g. , the algorithm of @xcite .
[ [ approximating - all - levels . ] ] approximating all levels .
+ + + + + + + + + + + + + + + + + + + + + + + + + we first compute explicitly all the @xmath610 bottom levels of @xmath598 .
their overall complexity is @xmath611 @xcite , and their construction takes @xmath612 time @xcite .
next , we approximate each of the levels @xmath613 , up to relative error of @xmath614 , using the algorithm described above , for @xmath615 .
this results in a sequence of @xmath616 pairwise disjoint @xmath295-monotone polyhedral surfaces in @xmath86 ( i.e. , the exact @xmath617 bottom levels , and the additional @xmath618 approximated levels ) .
we need to store these surfaces so that , given a query point @xmath570 , the two surfaces which lie directly above and below @xmath570 can be found efficiently .
this is done using binary search through the sequence of surfaces , where each step of the search is implemented by locating the @xmath295-projection @xmath619 of @xmath570 in the @xmath295-projection of a surface ( which is a planar map ) , and then by testing @xmath570 against the plane inducing the face containing @xmath619 .
thus the cost of a query is @xmath620 .
the index of the surface directly below @xmath570 ( namely , either its exact level if it is one of the first bottom @xmath621 levels , or the index of the level of @xmath598 that it approximates ) yields the desired approximate count .
the preceding analysis is easily seen to imply that the overall storage and preprocessing cost of the algorithm are both @xmath622 .
( concrete and reasonably small values of the powers of the polylogarithmic factor and of the factor @xmath623 can be easily worked out , but we skip over this step . ) hence we obtain the following result . given a set @xmath1 of @xmath4 points in three dimensions and a parameter @xmath624 , one can build a data - structure , in @xmath625 time and space , so that , given a query halfspace @xmath323 , one can approximate @xmath626 , up to relative error of @xmath627 .
the query time is @xmath620 .
[ theo : approximate:3:fast ] as in the planar case , theorem [ theo : approximate:3:fast ] improves over the previous results @xcite , which require @xmath628 time to answer a query .
however , in a subsequent work , afshani and chan @xcite managed to obtain an improved solution . specifically , they show that , with @xmath629 expected preprocessing time , one can build a data structure of expected size @xmath630 which can answer approximate 3-dimensional halfspace range counting queries in @xmath631 expected time , where @xmath632 is the actual value of the count , and @xmath633 hides constant factors that are polynomial in @xmath610
. it would also be interesting to compare our result to the appropriate variant of the technique of @xcite .
[ sec : conclusions ] in this paper we first established connections between the @xmath52-samples of li _
et al . _
@xcite and relative @xmath0-approximations ( and other notions of approximations ) . this has allowed us to establish sharp upper bounds on the size of relative @xmath0-approximations in arbitrary range spaces of finite vc - dimension .
we then turned to study geometric range spaces , and gave a construction of even smaller - size relative approximations for halfplane ranges , by revisiting the classical construction of spanning trees with low crossing number , and by modifying it to be `` weight - sensitive '' .
we then gave similar constructions of `` almost '' relative approximations for halfspace ranges in three and higher dimensions , using a different approach .
finally , we have also revisited the approximate halfspace range - counting problem in two and three dimensions , and provided better algorithms than those previously known .
there are several interesting open problems for further research .
the main one is to extend the construction of spanning trees with small relative crossing number to three and higher dimensions .
another open problem is to improve theorem [ theo : good : partition ] .
a minor further improvement of theorem [ theo : good : partition ] is possible by plugging the construction of theorem [ theo : good : partition ] into the construction of lemma [ lemma : level : k ] .
this still falls short of the desired spanning tree with crossing number @xmath634 , for a line @xmath283 of weight @xmath635 .
we leave this as an open problem for further research .
interestingly , the partition of lemma [ lemma : cover ] can be interpreted as a strengthening of the shallow partition theorem of matouek@xcite in two dimensions .
it is quite possible that a similar ( but probably weaker ) strengthening is possible in three and higher dimensions .
the authors thank the anonymous referees for their insightful comments .
the authors also thank boris aronov , ken clarkson , edith cohen , haim kaplan , yishay mansour and shakhar smorodinsky for useful discussions on the problems studied in this paper .
p. k. agarwal , b. aronov , t. m. chan and m. sharir , on levels in arrangements of lines , segments , planes , and triangles , _ discrete comput .
_ 19:315331 , 1998 .
p. afshani and t. m. chan , on approximate range counting and depth , _ discrete comput .
_ 41:321 , 2009 .
n. amenta , m. bern , d. eppstein , and s .- h .
teng , regression depth and center points , _ discrete comput . geom . _ , 23:305323 , 2000 .
a. andrzejak , _ on @xmath82-sets and their generalizations _ , ph.d .
thesis , inst .
fr theoretische informatik , eth zurich , 2000 .
a. andrzejak , b. aronov , s. har - peled , r. seidel , and e. welzl , results on @xmath82-sets and @xmath397-facets via continuous motion arguments , _ proc .
14th annu .
acm sympos .
_ , 1998 , pages 192199 .
b. aronov and s. har - peled , on approximating the depth and related problems , _ siam j. comput .
_ , 38:899921 , 2008 .
b. aronov and m. sharir , approximate halfspace range counting , _
siam j. comput .
_ , to appear .
h. brnnimann , b. chazelle , and j. matouek , product range spaces , sensitive sampling , and derandomization , _ siam j. comput .
_ , 28:15521575 , 1999 .
g. s. brodal and r. jacob .
dynamic planar convex hull . in _ proc .
43th annu .
ieee sympos . found .
_ , pages 617626 , 2002 .
h. brnnimann , _ derandomization of geometric algorithms _ , ph.d .
thesis , dept .
princeton university , princeton , nj , may 1995 .
t. m. chan , random sampling , halfspace range reporting , and construction of @xmath636-levels in three dimensions , _ siam j. comput .
_ , 30(2):561575 , 2000 .
b. chazelle , _ the discrepancy method : randomness and complexity _ , cambridge university press , new york , 2001 .
b. chazelle , the discrepancy method in computational geometry , chapter 44 , in _ handbook of discrete and computational geometry _ , 2nd edition , j.e .
goodman and j. orourke , eds .
, crc press , boca raton , 2004 , pages 983996 . b. chazelle and e. welzl , quasi - optimal range searching in spaces with finite vc - dimension , _ discrete comput .
_ , 4:467490 , 1989 .
k. l. clarkson and p. w. shor , applications of random sampling in computational geometry , ii , _ discrete comput geom .
_ 4:387421 , 1989 .
e. cohen , h. kaplan , y. mansour and m. sharir , approximations with relative errors in range spaces of finite vc - dimension , manuscript , 2006 .
t. dey , improved bounds for planar @xmath82-sets and related problems , _ discrete comput .
_ 19:373382 , 1998 .
h. everett , j .-
robert , and m. van kreveld .
an optimal algorithm for the @xmath637-levels , with applications to separation and transversal problems .
, 6:247261 , 1996 . s.
har - peled , carnival of samplings : nets , approximations , relative and sensitive , manuscript , 2008 .
http://arxiv.org/abs/0908.3718 and http://valis.cs.uiuc.edu/~sariel / papers/08/sampling_survey/. revised version appears as chapter 6 , `` yet even more on sampling '' in har - peled s class notes .
s. har - peled and m. sharir , relative @xmath0-approximations in geometry , manuscript , 2009 .
available from http://valis.cs.uiuc.edu/~sariel/papers/06/relative .
d. haussler , decision theoretic generalizations of the pac model for neural nets and other learning applications , _ inform .
_ , 100:78150 , 1992 .
d. haussler and e. welzl , epsilon nets and simplex range queries , _ discrete comput .
_ , 2:127151 , 1987 .
l. kapelushnik , _ computing the @xmath82-centrum and the ordered median hyperplane _
thesis , school of computer science , tel aviv university , 2008 .
d. g. kirkpatrick , optimal search in planar subdivisions , _ siam j. comput .
_ 12:2835 , 1983 .
h. kaplan , e. ramos and m. sharir , range minima queries with respect to a random permutation , and approximate range counting , _ discrete comput .
_ , to appear . also in http://www.math.tau.ac.il/~michas/ , 2008
h. kaplan , e. ramos and m. sharir , the overlay of minimization diagrams in a randomized incremental construction , http://www.math.tau.ac.il/~michas/ , 2008 .
h. kaplan and m. sharir , randomized incremental construction of three - dimensional convex hulls and planar voronoi diagrams , and approximate range counting , _ proc .
17th acm - siam sympos . discrete algorithms _ pages 484493 , 2006 .
j. komls , j. pach and g. woeginger , almost tight bounds for @xmath18-nets , _ discrete comput .
_ 7:163173 , 1992 .
y. li , p. m. long , and a. srinivasan , improved bounds on the sample complexity of learning , _ j. comput .
_ , 62:516527 , 2001 .
j. matouek , construction of epsilon - nets , _ discrete comput .
_ , 5:427448 , 1990 .
j. matouek , computing the center of planar point sets , in _ computational geometry : papers from the dimacs special year _ ( j.e .
goodman , r. pollack and w. steiger , eds . ) , ams , providence , ri , 1991 , pages 221230 .
j. matouek , reporting points in halfspaces , _ comput .
theory appl .
_ , 2:169186 , 1991 .
j. matouek , efficient partition trees , _ discrete comput .
_ , 8:315334 , 1992 .
j. matouek , _ geometric discrepancy _ ,
algorithms and combinatorics , vol .
18 , springer verlag , heidelberg , 1999 .
j. matouek , _ using the borsuk - ulam theorem _ , universitext , springer - verlag , berlin , 2003 , lectures on topological methods in combinatorics and geometry , written in cooperation with anders bjrner and gnter m. ziegler .
j. matouek , e. welzl and l. wernisch , discrepancy and approximations for bounded vc - dimension , _ combinatorica _ , 13:455466 , 1993 .
j. pach and p. k. agarwal , _ combinatorial geometry _ , wiley interscience , new york , 1995 .
d. pollard , rates of uniform almost - sure convergence for empirical processes indexed by unbounded classes of functions , manuscript , 1986 .
h. shaul and m. sharir , semi - algebraic range reporting and emptiness searching with applications , manuscript , 2009 .
an earlier incomplete version in _ proc .
16th acm - siam sympos . discrete algorithms _ , pages 525534 , 2005 .
n. sarnak and r. e. tarjan , planar point location using persistent search trees , _ commun .
, 29:669679 , 1986 .
m. talagrand , sharper bounds for gaussian and empirical processes , _
annals probab .
_ 22:2876 , 1994 .
vapnik and a. ya .
chervonenkis , on the uniform convergence of relative frequencies of events to their probabilities , _ theory of probability and its applications _ , 16:264280 , 1971 .
e. welzl , more on @xmath82-sets of finite sets in the plane , _ discrete comput .
_ 1:95100 , 1986 .
e. welzl , on spanning trees with low crossing numbers , in _ data structures and efficient algorithms , final report on the dfg special joint initiative _ ,
volume 594 of _ lect .
notes in comp .
_ , pages 233249 , springer - verlag , 1992 .
[ apnd : proofs ] | we re - examine the notion of _ relative @xmath0-approximations _ , recently introduced in @xcite , and establish upper bounds on their size , in general range spaces of finite vc - dimension , using the sampling theory developed in @xcite and in several earlier studies @xcite .
we also survey the different notions of sampling , used in computational geometry , learning , and other areas , and show how they relate to each other .
we then give constructions of smaller - size relative @xmath0-approximations for range spaces that involve points and halfspaces in two and higher dimensions .
the planar construction is based on a new structure spanning trees with small _ relative crossing number _ , which we believe to be of independent interest .
relative @xmath0-approximations arise in several geometric problems , such as approximate range counting , and we apply our new structures to obtain efficient solutions for approximate range counting in three dimensions
. we also present a simple solution for the planar case . |
this research was supported by the sistema nacional de investigadores ( mxico ) . 99 cerkaski , m. , r. g. nazmitdinov , and a. puente , physical review e 91.3 ( 2015 ) : 032312 a.a .
berezin , nature * 315 * , 104 ( 1985 ) l.t . wille and j. vennik ,
a : math . gen . * 18 * , l1113 ( 1985 ) k.j .
nurmela , j. phys .
a : math . gen . * 31 * , 1035 ( 1998 ) s. erkoc and h. oymak , phys . rev.*e * 62 , r3075-r3076 ( 2000 ) s. erkoc and h. oymak , physics letters a 290.1 ( 2001 ) : 28 - 34 a. worley , arxiv : physics/0609231 ( 2006 ) mughal , a. and m. a. moore , physical review e 76.1 ( 2007 ) : 011606
thomson , philos . mag . * 7 * , 237 ( 1904 ) m. calef , w. griffiths and a. schultz , j. stat .
phys . * 160 * , 239 - 253 ( 2015 ) | we have found that the minimum energy configuration of @xmath0 charges confined in a disk and interacting via the coulomb potential , reported by cerkaski et al . in ref .
@xcite is not a global minimum of the total electrostatic energy .
we have identified a large number of configurations with lower energy , where defects are present close to the center of the disk ; thus , the formation of a hexagonal core and valence circular rings for the centered configurations , predicted by the model of ref .
@xcite , is not supported by numerical evidence and the configurations obtained with this model can not be used as a guide for the numerical calculations , as claimed by the authors . in a recent paper , ref .
@xcite , cerkaski et al . studied the problem of a finite number of equal charges , interacting via the coulomb potential and confined inside a disk .
this problem has been previously studied by several authors in a series of papers , refs .
@xcite , and it can be regarded as a generalization of the well known thomson problem @xcite ( finding the configurations of minimum energy of @xmath1 equal charges on the surface of a sphere ) . despite the apparent simplicity , both problems provide a serious computational challenge , of increasing difficulty with @xmath1 : in particular , the number of local minima of the total electrostatic energy grows very fast with @xmath1 ( for the case of the thomson problem see for example the discussion in ref .
@xcite ) . as a result
, the identification of the global minimum of a system of @xmath1 charges typically requires extensive numerical calculations : in the absence of a formal criterium to establish whether a given configuration of equilibrium is a global minimum , one has to repeat the numerical calculations several times , keeping @xmath1 fixed , and regard the configuration with lowest energy among those obtained as a probable candidate for a global minimum . for the case of the disk , erkoc and oymak @xcite
have observed the tendency , for systems with modest number of charges ( @xmath2 ) , to accomodate the charges on concentric rings , empirically deducing the rules for the distribution of charges on the disk ( incidentally , most of the energies reported by these authors in tables @xcite do not correspond to global minima ) . the analysis performed by cerkaski et al . in ref .
@xcite is a refinement of the work of erkoc and oymak @xcite , and it relies on the hypothesis that charges arrange on concentric rings , for configurations of minimum energy . in this way ,
the original problem is reduced to the much simpler problem of calculating the electrostatic energy due to @xmath3 rings , each carrying an appropriate number of charges , uniformely distributed over the ring ; the equilibrium configuration in this case is obtained by solving a system of two equations ( their eqs .
( 16 ) and ( 17 ) ) . to test their model
the authors have performed numerical ( molecular dynamics , md for short ) calculations for systems up to @xmath4 charges . based on this analysis
they conclude that their approach allows one `` to determine with high accuracy the equilibrium configurations of a few hundred charged particles '' . in particular , for @xmath5 their approach `` predicts the formation of the hexagonal core and valence circular rings for the centered configurations '' , with `` an increasing sequence of rings , starting from the center , matching the regular hexagonal pattern '' . for the case of @xmath0 , discussed at length in ref .
@xcite the authors observe the formation of an hexagonal structure with rings @xmath6 . fig .
2b of ref .
@xcite , that reports a comparison between the model and the numerical md calculations for @xmath0 charges , displays an excellent agreement between the two , with only a small mismatch just outside the hexagonal structure ( the green region in the figure is used to highlight the hexagonal structure ) . the energy reported by the authors for this configuration , which is expected to be
a global minimum of the total energy , is @xmath7 , compared to the energy @xmath8 , obtained with their model , with an error of just @xmath9 . with the purpose of veryfying the results of ref .
@xcite we have carried out extensive numerical calculations , in particular for the case of @xmath10 charges .
the approach that we have implemented allows one to generate configurations with the desired number of charges on the border : in this way we have verified that the lower energies occurr when @xmath11 charges are disposed on the border of the disk , in agreement with ref .
@xcite .
we have thus generated @xmath12 configurations with @xmath11 , starting from initial configurations where the internal charges are randomly distributed , and we have found that @xmath13 of them have energy lower that the value reported in ref .
@xcite , @xmath14 .
the lowest energy among those that we have calculated ( possibly a global minimum ) is @xmath15 .
the histogram in fig .
[ fig_1 ] illustrates these points .
interestingly , we have also found that even the configuration with largest energy has an energy slightly lower that the value predicted by the model of ref .
@xcite , @xmath16 ( see fig . [ fig_1 ] ) . in fig .
[ fig_2 ] we display the configuration with the lowest energy among those calculated ( an hexagonal grid is also plotted , to facilitate the identification of a centered hexagonal structure ) ; note that the color of the vertices , representing the charges , depends on the number of nearest neighbors . studying this figure
, we observe the presence of defects very close to the center of the disk , and of a single , slightly deformed , hexagonal cell ( the yellow region ) , centered at the origin , in sharp contrast with the numerical and theoretical observations of ref .
@xcite .
additionally , we have also found that similar behaviors are also observed for the configurations with slightly larger energy .
we summarize our main findings : * the occurrence of defects , even very close to the center of the disk , may help to lower the total energy , while disrupting the hexagonal structure ; * all the configurations with @xmath11 that we have calculated have energy lower than the theoretical value obtained in ref .
@xcite and about @xmath17 of them has energy lower than the numerical md value of ref .
@xcite ; * the use of the model of ref .
@xcite as a guide for the numerical md calculation ( which is also claimed to cut the cpu times by a factor @xmath18 ) , as suggested by the authors , is definitively unjustified . using this approach
, we may expect that the solutions not only will correspond to local minima of the energy , but they will also be strongly biased ( i.e. the more symmetric structures could be favored ) ; * the number of configurations grows very fast with @xmath1 , thus requiring an efficient numerical approach : our program allows to generate configurations with a desired number of charges on the border .
we have found that in this way the performance is drastically improved ; in light of this findings , the validity of the model of ref .
@xcite must be questioned , particularly for @xmath5 ; it also appear clear the inadeguacy of the numerical calculation of ref .
@xcite , which has failed to identify a very large number of configurations with energy lower than the one reported by the authors ( about @xmath17 of the configurations that we have calculated have lower energy than the one of ref .
@xcite ! ) .
we are not sure whether this problem has been triggered by using the output of the model as a guide for the numerical calculation or if the program used by the authors produced a small number of configurations with @xmath11 .
charges with @xmath11.,width=264 ] charges with energy @xmath19,width=302 ] |
heavy - ion induced fusion reactions have been studied extensively for more than forty years , especially since the discovery of the sub - barrier enhancement phenomenon @xcite .
coupled - channels descriptions were shown to explain the phenomenon successfully @xcite .
recently , evidence was found for a strong hindrance of the fusion process at extreme sub - barrier energies , an effect for which there is no satisfactory explanation in present model calculations @xcite .
a systematic survey of existing data from the literature shows that there is a regular pattern to the energy dependence for the appearance of fusion hindrance @xcite . for stiff , closed - shell colliding systems ,
a significant maximum in the @xmath3-factor is present as a function of the beam energy .
this maximum signals the onset of sub - barrier hindrance , and the energy , @xmath4 , at which it is located can be described well by an empirical formula @xcite : @xmath6 where @xmath7 . for softer systems , this formula provides an upper limit for the energy at which the @xmath3-factor has its maximum . fusion hindrance at extreme sub - barrier energies is relevant , not only for understanding the dynamics of reactions between complex systems , but also for astrophysics and for the synthesis of superheavy elements .
the influence of nuclear structure on this hindrance behavior was first studied in a detailed comparison @xcite for the colliding systems : @xmath8ni + @xmath8ni @xcite , @xmath8ni + @xmath9ni @xcite , @xmath8ni + @xmath0ni @xcite , and @xmath0ni + @xmath0ni @xcite , where the systems are arranged in order of decreasing stiffness . for @xmath0ni + @xmath0ni , an open - shell colliding system ,
the measured location in energy of the @xmath3-factor maximum is about @xmath10 lower than the value expected from eq .
[ eq1 ] @xcite , while the other systems are more in line with the systematics .
the aim of the present paper is to further investigate the hindrance phenomenon by measuring fusion evaporation for the open - shell system @xmath0ni + @xmath11mo .
compared to @xmath0ni , @xmath11mo is a transitional nucleus with two protons outside the closed proton shell .
two earlier measurements of fusion excitation functions for @xmath0ni + @xmath11mo can be found in the literature @xcite .
the minimum cross section measured in these two experiments is about 0.4 mb . since the hindrance behavior is expected ( from eq .
[ eq1 ] ) to occur at much lower energies , the main aim of the present measurement was an extension of the excitation function into the nb region in order to localize and quantify this hindrance .
the experiment was performed with @xmath0ni beams in the energy range of 196 - 262 mev from the superconducting linear accelerator atlas at argonne national laboratory .
the maximum beam current used was @xmath2 60 pna .
the high melting point of the target material , metallic molybdenum evaporated on a 40@xmath12g/@xmath13 carbon foil , prevented damage to the target by the relatively high beam current .
the target thickness was constant during the experiment , as monitored with si detectors .
thin targets with thicknesses of 8 or 18@xmath12g/@xmath13 were used in order to reduce the correction for target thickness in the energy regime of steep fall - off of the excitation function .
the isotopic abundance of @xmath11mo was 97.42@xmath14 , with the remainder coming from @xmath15mo ( 0.96% ) , @xmath16mo ( 0.28% ) , @xmath17mo ( 0.34% ) , @xmath18mo ( 0.29% ) , @xmath19mo ( 0.18% ) , and @xmath20mo ( 0.53% ) .
the selection of the beam and target combination excluded the possibility of background from fusion reactions coming from beam or target contaminants @xcite .
two surface - barrier si detectors , located at @xmath21 43@xmath22 with respect to the beam direction , served as monitors .
the absolute cross sections for fusion - evaporation were determined by using elastic scattering measured with the monitors .
the evaporation residues were identified and measured with the fragment mass analyzer ( fma ) @xcite , which has been upgraded with the installation of a split - anode in the first electric - dipole @xcite .
the background , originating mostly from scattered beam , was greatly suppressed after this upgrade . in order to push the cross section measurements to the lowest level ,
a new focal - plane detector with the configuration pgac - tic - pgac - tic - pgac - ic , was used in the experiment . here , the symbols pgac stand for @xmath23-@xmath24 position sensitive parallel grid avalanche counters @xcite , tic for transmission ionization chambers @xcite , and ic for a large volume multi - anode ionization chamber .
the first pgac was mounted at the horizontal ( @xmath23-direction ) focal - plane of the fma .
the @xmath24-focus condition occurs about 70 cm downstream of the focal - plane , which is nearly at the middle of the last ionization chamber .
the three sets of position signals @xmath25 and @xmath26 were measured with the three pgac s .
the flight - times @xmath27 and @xmath28 obtained from pgac@xmath29-pgac@xmath30 and pgac@xmath29-pgac@xmath31 were also recorded .
seven @xmath32 signals were measured with the ionization chambers ( the first four @xmath33-@xmath34 in the two transmission ionization chambers and the last three @xmath35-@xmath36 in the final ionization chamber , ic ) .
another time - of - flight signal , @xmath37 , measured the time difference between the rf - system of the accelerator and the first pgac .
three additional @xmath32 signals from the pgac s were also recorded .
the three pgac s and two tic s operated at a fixed pressure of 3 torr of isobutane .
the last ionization chamber had an adjustable pressure of 22 - 30 torr of isobutane .
there was one 0.13 mg/@xmath13 mylar pressure foil in front of the first pgac , and another one with a thickness of 0.22 mg/@xmath13 located between the third pgac and the last ionization chamber .
this setup allowed for full tracking of each particle detected in the system , and provided very good separation between evaporation residues and background events .
a more detailed description of the new detector system will be published elsewhere @xcite . for most settings of the fma
, two charge states of the residues were collected simultaneously .
for the energies @xmath38 = 260.5 , 245.8 , 209.1 , 207.1 and 202.2 mev , full charge state distributions were measured , while for most other energies , two fma settings , _
i.e. _ four charge states were recorded . at the four lowest energies ,
only two charge states were measured . from the full charge state distributions ,
charge state fractions were determined for extrapolation to all other energies .
the energy distributions and angular distributions of evaporation residues were calculated with the statistical code pace @xcite .
total angular distributions for fusion - evaporation have been measured in ref .
@xcite for @xmath0ni + @xmath11mo . in order to check the pace code ,
calculations were compared with these experimental angular distributions of ref .
@xcite and good agreement was found as long as the total calculated angular distributions were taken as a weighted sum of the angular distributions of the different masses from our @xmath39 measurements and folded with multiple scattering . whereas in ref .
@xcite rather thick targets were used , the corrections from multiple scattering are small in the present experiment .
the transport efficiencies of the fma were calculated with these angular distributions together with monte carlo simulations , using a modified version of the gios code @xcite .
the large momentum acceptance , @xmath21 10 @xmath14 , and the large angular acceptance , @xmath40 2.3@xmath41 , of the fma result in a high detection efficiency for the residues .
the flight - time @xmath28 between pgac@xmath29 and pgac@xmath31 , together with a @xmath32 signal measured in a counter far behind the focal plane ( e.g. @xmath42 ) , was found to give the best separation of the residues from background events .
two - dimensional plots of @xmath42 vs. @xmath28 are shown in fig .
[ fig2 ] , indicating the excellent separation achieved with these two quantities . down to a cross section level of @xmath2 300
nb , the evaporation residues could be identified on the basis of these two parameters alone .
the events that fall within the expected window for evaporation residues are shown as open circles in fig .
events shown by small points originate from beam particles , scattered from the beam pipe or from the area of the `` beam stop '' at the first split anode . the events in fig .
[ fig2]a correspond to a cross section of 242 nb obtained in a 12 hours run .
a two - dimensional plot of @xmath35 vs. @xmath28 , corresponding to the same events is shown in fig .
[ fig4]a . at this
energy evaporation residues ( open circles ) are again well separated from background events ( small points ) . at the lowest energy , @xmath43=196.0 mev
, we find that 23 events fall within the expected evaporation residue window as illustrated in fig .
most of these events are suspiciously close to an intense band of scattered beam particles which borders the fusion evaporation residue window . by examining the @xmath35 vs. @xmath28 spectrum shown in fig .
[ fig4]b , it is evident that only one of these events falls within the expected range of the @xmath35 signal .
however , this event does not have the correct @xmath39-value as will be demonstrated below .
an @xmath39 spectrum containing the 23 events that fall inside the @xmath44 window is given in fig .
this spectrum is spread out over the whole @xmath39 range and does not appear to be associated with fusion evaporation residue events seen at higher energies to be concentrated near channels 103 and 145 ( see fig .
[ fig3]a ) .
the only candidate event at @xmath45=196.0 mev , which satisfies the conditions for the @xmath46 , and @xmath28 signals , appears in channel 114 as seen in fig .
[ fig3]b . by further checking all the detector signals for this event
, it was concluded that it is a background event , probably arising from scatterings of a beam particle at several locations through the spectrometer .
hence , only an upper limit for the measured cross section could be determined at this beam energy .
these results show that a background suppression factor of about 4@xmath4710@xmath48 can be achieved with the upgraded fma and the present detector system .
the uncertainties in the reported evaporation cross sections arise mainly from corrections for the charged state distribution , the fma transport efficiency , the detector efficiency , and counting statistics .
the total uncertainties for the evaporation cross sections are around 10 - 17@xmath14 , except for the measurements at the two lowest energies where only upper limits can be given .
the cross sections are listed in table i. the fusion - fission cross sections for @xmath0ni + @xmath11mo have not been measured previously .
they were , however , calculated in ref .
@xcite with the code cascade @xcite .
similar calculations with the same parameters were performed for the present experiment .
the total fusion cross sections are also listed in table i , together with the fusion - fission cross sections .
rather large uncertainties were given for the fusion - fission contributions resulting in somewhat larger errors for the total cross sections at the highest beam energies .
the experimental results for the total fusion cross sections , spanning eight orders of magnitude , are presented as a function of laboratory energy in fig .
[ fig5 ] ( solid circles ) .
the incident energies have been corrected for target thickness and for the steep energy dependence ; these corrections are small because rather thin targets were used . for the lowest two energies
no evaporation residue was observed .
the results are shown as upper limits corresponding to one count in each case .
the two earlier measurements @xcite are also shown in fig .
[ fig5 ] as open circles and open squares , respectively .
statistical errors are smaller than the size of the symbols .
the three measurements are generally in good agreement within the quoted uncertainties , except for the lowest energy point of ref .
ccccc @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 + ( mev ) & & ( mb ) & ( mb ) & ( mb ) + 158.8 & 14 & 264 @xmath21 35 & 275 & 539 @xmath21 111 + 149.9 & 9 & 210 @xmath21 25 & 80 & 290 @xmath21 42 + 141.1 & 4 & 80.0 @xmath21 8.8 & 2 & 82.0 @xmath21 9.8 + 136.1 & 5 & 29.2 @xmath21 3.0 & 0 & 29.0 @xmath21 3.0 + 131.2 & 4 & 6.80 @xmath21 0.71 & 0 & 6.80 @xmath21 0.71 + 129.2 & 4 & 2.87 @xmath21 0.30 & 0 & 2.87 @xmath21 0.30 + 127.5 & 12 & 0.92 @xmath21 0.10 & 0 & 0.92 @xmath21 0.10 + 126.2 & 6 & 0.35 @xmath21 0.04 & 0 & 0.35 @xmath21 0.04 + 125.0 & 4 & 0.109 @xmath21 0.012 & 0 & 0.109 @xmath21 0.012 + 123.9 & 4 & 0.0253 @xmath21 0.0029 & 0 & 0.0253 @xmath21 0.0029 + 123.3 & 8 & 0.0132 @xmath21 0.0014 & 0 & 0.0132 @xmath21 0.0014 + 122.9 & 4 & 7.4@xmath54 & 0 & 7.4 @xmath54 + 121.7 & 2 & 1.10@xmath55 & 0 & 1.10@xmath55 + 121.2 & 2 & 2.42@xmath56 & 0 & 2.42@xmath56 + 120.2 & 2 & @xmath57 2.0@xmath58 & + 119.5 & 2 & @xmath57 4.6@xmath59 & + the coupled - channels calculations of ref .
@xcite reproduced the excitation function quite well down to the 0.5 mb level .
it was found that with the same coupling parameters , the calculations can not reproduce the new data at lower beam energies .
in fact , in the literature there are not many such calculations that can reproduce successfully experimental data for heavy systems .
since the nucleus @xmath11mo is rather soft , multi - phonon states and large coupling effects should be included in the calculations and this represents a challenge . in the following , two sets of coupled - channels calculations are presented , including two- and three - phonon states , respectively . cccccc nucleus & @xmath60 & e@xmath61 & b(e@xmath62 ) & @xmath63 & @xmath64 + & & ( mev ) & ( w.u . ) & & + @xmath0ni & @xmath65 & 1.346 & 8.6 & 0.165 & 0.185 + & 2ph(@xmath65 ) & 2.869 & - & 0.165 & 0.185 + & @xmath66 & 3.560 & 12 & 0.193 & 0.200 + @xmath11mo & @xmath65 & 0.536 & 37 & 0.231 & 0.254 + & 2ph(@xmath65 ) & 1.002 & 68 & 0.222 &
0.244 + & 3ph(@xmath65 ) & 1.671 & 90 & 0.206 & 0.215 + & @xmath66 & 1.908 & 32 & 0.210 & 0.230 + & 2ph(@xmath67 & 3.816 & - & 0.210 & 0.230 + ccccc nucleus & & + @xmath68 & @xmath69 & @xmath70&@xmath69 & @xmath70 + & ( mev ) & ( w.u . ) & ( mev ) & ( w.u . )
+ @xmath71 & 2.867 & 110 & 0.695 & 92 + @xmath72 & 2.277 & 0 & 1.064 & 51 + @xmath73 & 2.610 & @xmath5737 & 1.136 & 69 + effective-2ph & 2.87 - 2.75&22 - 41 & 1.002 & 68 + @xmath74&- & 0.177 - 0.242 & - & 0.222 + the nuclear structure input for the calculations is given in table ii .
the structure input for @xmath0ni is the same as used in ref .
@xcite to analyze the @xmath0ni + @xmath0ni fusion - evaporation excitation function .
the nuclear quadrupole coupling , @xmath64 , for @xmath0ni was set @xmath75 higher than the value for coulomb coupling , @xmath63 , ( for reasons discussed in ref .
@xcite ) . for simplicity
, it was assumed that the quadrupole mode corresponds to a perfect vibrator .
this is not unreasonable , considering the uncertainty of the predicted one- and two - phonon couplings shown in table iii . the structure input for @xmath11mo is also given in table ii . here
, the nuclear @xmath76-values were also set @xmath75 higher than the coulomb @xmath76-values .
the two - phonon ( 2ph ) calculations include the @xmath65 and @xmath66 one - phonon states in both nuclei , as well as mutual excitations and the two - phonon states listed in table ii .
the two - phonon octupole excitation of @xmath11mo was also included .
the two - phonon states were treated as perfect vibrational states .
this gives a total number of 14 coupled channels .
the three - phonon ( 3ph ) calculations additionally include all mutual excitations of the states considered in the 2ph calculations up to two mutual two - phonon states .
moreover , the estimated three - phonon excitation of the @xmath65 state in @xmath11mo shown in table ii was also included .
however , states above 5.5 mev were not included , so the total number of channels is 31 .
the ion - ion potential was parametrized as a woods - saxon well with a depth of @xmath77 = -82.9 mev , diffuseness @xmath78 = 0.686 fm , and nuclear radius @xmath79 = 10.19 + @xmath80 fm .
if the value of @xmath80=0.06 fm was used in the two - phonon calculations , the result is similar to the calculations in ref .
@xcite , and it agrees quite well with the data for the excitation function above the 0.5 mb level .
this two - phonon calculation is shown in fig .
[ fig6 ] as the dotted - dashed curve and it is evident that it does not reproduce the data well at low energies .
we have also tried the recipe of a larger diffuseness ( up to @xmath81=5 fm ) inside the coulomb barrier , which was described in ref .
@xcite , but this did not result in any significant improvement of the fit to the data . a value of @xmath80 = 0.21 fm was needed to reproduce the present data in the cross section region of 0.1 to 100 @xmath12b . the potential with this @xmath80 value produces a coulomb barrier of 134 mev , and a pocket inside the coulomb barrier at 112.9 mev , which is about 20 mev higher than the ground state energy of the compound nucleus at 92.3 mev .
the value of @xmath80 = 0.21 fm was also adopted in the 3ph calculations .
the result is shown as the solid curve in fig .
[ fig6 ] , which is seen to provide a better fit to the data in the 0.1 to 100 @xmath12b cross section range .
the dashed curve shows as a reference the results obtained in a one - dimensional calculation , _
i.e. _ , without any couplings ( with @xmath80 = 0.21 fm ) .
it should be emphasized that the difference between the two sets of calculations , using the same @xmath80 = 0.21 fm , is rather modest .
hence , the 2ph calculation with @xmath80 = 0.21 fm is not shown for clarity .
these two calculations reproduce the experimental data in the highest energy region and in the 0.1 to 10 @xmath12b range , but the cross sections in the region around e=130 mev are overpredicted .
these calculations also overpredict the cross sections at the lowest energies , which is the main topic of the present study .
the coupled - channels calculations shown in fig . [ fig6 ] exhibit essentially the same energy dependence ( slope ) as the one - dimensional calculation when the cross sections are small ( @xmath57 10 @xmath12b ) .
the coupled - channels calculations are just shifted to lower energies relative to the one - dimensional calculation . in the present case
the shift is about 7 mev ( for the same @xmath80 = 0.21 fm ) .
this is a general feature of coupled - channels calculations and it is , therefore , very unlikely that any minor adjustment in the coupled - channels calculations would reproduce the steep falloff that the data exhibit at extreme sub - barrier energies .
thus , it appears that the fusion hindrance behavior , which now has been observed for many systems , is also present in the new data for @xmath0ni+@xmath11mo .
this will be shown more convincingly in the next section , where other representations of the fusion cross section are discussed . from the experimental data and calculations
alike , one can infer a rather broad , about 18 mev wide , barrier distribution ( not shown here ) .
an elaborate coupling scheme is therefore needed in the coupled - channels calculations . at energies above the barrier , one - dimensional calculations result in higher cross sections when compared to coupled - channels calculations .
this behavior is caused by the long - range coulomb - excitation / polarization of the low - lying quadrupole states , as pointed out in ref .
as expected , the suppression of the coupled - channels fusion cross sections is rather strong in this region for the `` soft '' @xmath11mo nucleus .
the logarithmic derivative , @xmath82 , originally introduced in ref .
@xcite , is shown in fig .
[ fig7]a for the @xmath0ni + @xmath11mo system .
the solid circles were obtained directly from two successive data points , whereas the stars were derived from least - squares fits to three neighboring data points .
the lower limit of the logarithmic derivative was derived from the upper limit on the cross section at @xmath83=120.2 mev and the data point at @xmath83=121.2 mev ( see table i ) .
the present data are compared to those of refs .
@xcite , which are represented by open circles and solid triangles in fig .
[ fig7]a , respectively .
only two - point derivatives are shown for these data .
the three - phonon coupled - channels and the one - dimensional barrier penetration calculations are shown as solid and dashed curves , respectively , while the thick solid , nearly horizontal line corresponds to a constant @xmath3-factor expression derived in ref .
the present experimental results just reach the constant @xmath3-factor line , implying that the experimental data have reached a maximum value for the @xmath3-factor .
the energy , @xmath4 , representing the intersection between the experimental logarithmic derivatives and the constant @xmath3-factor line corresponds to a value of @xmath4=120.6 mev . the dashed - dotted line in fig .
[ fig7]a is an extrapolation obtained under the assumption that the logarithmic derivative is a straight line near the crossing point .
this method was first introduced in ref .
@xcite in order to obtain the extrapolated values of @xmath4 for the systems @xmath8ni + @xmath9ni and @xmath8ni + @xmath0ni .
the calculated logarithmic derivatives are seen to saturate around @xmath84=1.5 - 2 mev@xmath85 ( or start to oscillate ) below @xmath83=124 mev , whereas the corresponding experimental values continue to grow with decreasing energies .
this saturation behavior has already been noted in refs .
@xcite .
cccccccccccc system & @xmath86 & @xmath4 & @xmath87 & @xmath5 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & ref .
+ & & ( mev ) & ( mev ) & ( mev ) & & ( mb ) & ( mev ) & ( mev ) & & + @xmath8ni+@xmath8ni & 4222 & 94.0 & & 93.0 & 1.01 & 0.049 & 66.12&102.0 & 2 + 2 & @xcite + @xmath8ni+@xmath9ni & 4258 & 92 & & 93.5 & 0.98 & 0.040 & 62.69&101.3 & 2 + 4 & @xcite + @xmath8ni+@xmath0ni & 4325 & 89 & & 94.5 & 0.94 & 0.077 & 53.04&100.0 & 2 + 4 & @xcite + @xmath0ni+@xmath0ni & 4435 & 87.7 & & 96.1 & 0.91 & @xmath93&48.78 & 98.1 & 4 + 4 & @xcite + @xmath9ni+@xmath94y & 6537&122.9 & & 124.5&0.99 & @xmath95&90.50 & 136.5 & 4 + 1 & @xcite + @xmath8ni+@xmath20mo & 7014 & & 132.9&130.5 & & 0.17 & 108.0&148.6 & 2 + 2 & @xcite
+ @xmath0ni+@xmath20mo & 7225 & & 132.1&133.1 & & 2.6 & 100.6&146.0 & 4 + 2 & @xcite + @xmath8ni+@xmath11mo&7125 & & 128.9&131.8 & & 0.72 & 90.39&143.3 & 2 + 10 & @xcite + @xmath0ni+@xmath11mo&7346&120.6 & & 134.5&0.90 & @xmath96&92.29&136.5 & 4 + 10 & present + @xmath8ni+@xmath97ge & 5109 & 98.5 & & 105.6&0.93 & 0.037 & 62.03&113.4 & 2 + 6 & @xcite + @xmath0ni+@xmath97ge & 5249 & 97.5 & & 107.5&0.91 & 0.013 & 58.48&111.3 & 4 + 6 & @xcite + @xmath8ni+@xmath98zr & 6652 & & 127.3&125.9 & & 0.35 & 97.24&141.1 & 2 + 0 & @xcite + @xmath8ni+@xmath99zr & 6666 & & 126.4&126.1 & & 0.25 & 94.18&140.8 & 2 + 1 & @xcite + @xmath8ni+@xmath19zr & 6708 & & 122.7&126.6 & & 0.35 & 86.94&139.8 & 2 + 4 & @xcite + @xmath0ni+@xmath20zr & 6881 & & 119.0&128.8 & & @xmath100&91.45&137.8 & 4 + 2 & @xcite + @xmath0ni+@xmath17zr & 6940 & & 120.2&129.5 & & 0.15 & 86.48&136.6 & 4 + 6 & @xcite + @xmath8ni+@xmath101sn&8801 & & 149.4&151.8 & & 0.19 & 112.4&170.3 & 2 + 8 & @xcite + @xmath0ni+@xmath101sn&9096 & & 155.0&155.1 & & 23 . & 117.5&167.3 & 4 + 8 & @xcite + the @xmath3-factor representation for the @xmath0ni + @xmath11mo system is presented in fig .
[ fig7]b . as the experimental logarithmic derivatives
only just reach the constant @xmath3-factor curve , the @xmath3-factor maximum is not fully developed .
additional measurements would be required to clearly delineate the maximum in the @xmath3-factor .
the dashed - dotted curve corresponds to the straight line extrapolation of the logarithmic derivative in fig .
the two- and three - phonon calculations are shown in fig .
[ fig7]b as dashed and solid curves , respectively .
it is evident that coupled - channels calculations overpredict the fusion cross section at extreme sub - barrier energies .
the experimental fusion data involving ni projectiles and compound nuclei in the a=100 - 200 region , are summarized in table iv . for four systems , namely the @xmath8ni+@xmath8ni , @xmath0ni+@xmath0ni , @xmath9ni+@xmath94y and the present @xmath0ni+@xmath11mo reactions ,
the cross sections have been measured to sufficiently low energies to determine the energy , @xmath4 of the maximum of the @xmath3-factor representation .
previously , we have obtained the @xmath4 values for the @xmath8ni+@xmath9ni and @xmath8ni+@xmath0ni systems @xcite .
the values of @xmath4 are listed in column three of table iv .
two additional systems , namely @xmath8ni+@xmath97ge and @xmath0ni+@xmath97ge from ref.@xcite , have been measured down to levels close to where the @xmath3-factor maximum occurs .
the logarithmic derivative for these systems is shown in fig .
the location of the @xmath3-factor maximum was obtained by performing a small extrapolation ( solid line ) of the logarithmic derivative to where it crosses the constant @xmath3-factor line ( dashed line ) .
the resulting values are listed in table iv . because of the uncertainties inherent in these extrapolations , errors of 2@xmath14 ( @xmath22 mev ) were assigned to the extrapolated @xmath4 data .
table iv also lists other fusion systems involving ni beams for which the cross section was not measured down to a level that allows for determination of the @xmath3-factor maximum . for these systems
only the lowest measured energy is listed .
the values of @xmath4 are plotted as a function of the parameter @xmath102 in fig .
[ fig8 ] and compared to the empirical formula ( solid curve ) , eq . [ eq1 ] , obtained from a fit to all available fusion data involving stiff nuclei .
obviously , only two systems , @xmath8ni+@xmath8ni and @xmath9ni+@xmath94y follow the systematics ; all other systems fall below the curve .
previously , it has been pointed out @xcite that there is a rather compelling correlation between the stiffness of the interacting nuclei and the location of the @xmath3-factor maximum , @xmath4 , relative to the empirical trend for the ni+ni systems .
the addition to the systematics of the data points for @xmath0ni+@xmath11mo and for the two ni+ge systems , which all involve soft nuclei , appears to corroborate this observation .
thus , for the @xmath0ni+@xmath11mo system , the value of @xmath5 predicted from eq .
[ eq1 ] is @xmath5=134.5 mev whereas the measured value of @xmath4=120.6 mev is only 90@xmath14 of this value .
the reason for this reduction is presumably that @xmath0ni + @xmath11mo should be viewed as an open - shell colliding system so that strong coupling effects broaden the effective barrier distribution and push the energy where the hindrance behavior occurs down to even lower energies . the deviation of the measured or extrapolated values of @xmath4 from the expected @xmath5 value , seen in fig .
[ fig8 ] , thus appears to depend on the stiffness of the fusing nuclei .
a quantitative relation between the stiffness and the deviation from @xmath5 is not yet known . as a first attempt
, we associate the stiffness of a nucleus to its proximity to closed proton or neutron shells and define the number of `` valence nucleons '' , @xmath92 , as the sum of particles and holes outside the nearest closed shells . here , @xmath0ni is considered to have four holes in the n=40 neutron shell rather than eight particles outside n=28 .
the values of @xmath92 are listed in table iv and , in fig .
[ fig10 ] , the ratio @xmath103 is plotted as a function of this parameter .
there is a general trend of decreasing values of @xmath104 with increasing values of @xmath92 .
we observe that for the other systems in table iv , the data for @xmath87 ( upper limits ) are not in contradiction with figs .
[ fig8 ] and [ fig10 ] .
it should be noted that all of the systems shown in figs .
[ fig8 ] and [ fig10 ] and table iv have rather large negative fusion @xmath105values .
one may also compare the @xmath4-values to the height of the coulomb barrier ( obtained from the bass prescription ) , which is listed in table iv .
the ratio @xmath104 exhibit a stronger dependency on the value of @xmath92 than that obtained for the ratio @xmath106 indicating that the observed effect does not just depend on the change in the coulomb barrier height with addition of neutrons to the interacting nuclei .
the above observations are all phenomenological in nature as there is at present no satisfactory understanding of the fusion hindrance at extreme sub - barrier energies .
many authors have tried to explain this new phenomenon @xcite .
one suggestion is to use a large diffuseness of the ion - ion potential in the coupled - channels calculations .
this recipe is sometimes used to explain high - precision fusion data @xcite .
it has been argued that the failure to reproduce the steep fall - off is caused by the hill - wheeler approximation @xcite . in ref .
@xcite it was shown , however , that these limitations of the analysis are not responsible for the observed sub - barrier fusion hindrance phenomenon .
dasso and pollarolo tried to reduce the fusion cross sections by using a shallow well inside the barrier @xcite .
giraud et al . investigated the effect of `` friction '' @xcite .
more discussions about this phenomenon can be found in refs .
all of these suggestions may improve the fit of calculations to the experimental data in some cases , but they do not provide a convincing explanation of the observed suppression in all systems .
these studies are presently at the stage of raising questions and discovering weak points in the existing models .
the data are most likely still insufficient to lead to the correct explanation .
more precision sub - barrier fusion measurements are required to further explore which modifications in the theoretical models are relevant for a correct description of the phenomenon .
the phenomenon of sub - barrier fusion hindrance was first observed in systems involving stiff nuclei and a simple expression @xcite was derived for the energy at which the hindrance of fusion between such nuclei occurs .
furthermore , a study of ni+ni fusion involving different ni isotopes @xcite has shown that the onset of the fusion hindrance deviates from these systematics depending on the `` stiffness '' of the interacting nuclei . in the present work ,
we have measured the fusion excitation function for the system @xmath0ni+@xmath11mo down to a cross section level of @xmath25 nb , _
i.e. _ about 12% below the bass barrier .
we observe that the fusion process is hindered at the lowest energies relative to expectations based on coupled - channels calculations .
the present study of the @xmath0ni+@xmath11mo system shows that the fusion hindrance for this system , as well as two other soft systems , deviates strongly from the systematics , and it thus corroborates the earlier observation in ni+ni systems .
it is furthermore shown that this deviation depends monotonically on a parameter @xmath92 , which is the sum of nucleons ( holes ) outside of closed shells of the fusing nuclei suggesting that this parameter is a good measure of the `` stiffness '' of the interacting nuclei . in conclusion , an interesting nuclear structure dependence of the fusion hindrance has been observed .
the origin of this effect is still unknown .
it occurs at relatively high excitation energies ( for systems in fig .
[ fig8 ] and [ fig10 ] , they are around @xmath107 30 - 40 mev ) , where the natural width of compound levels is larger than their spacing .
a lack of available final states thus appears to be ruled out as an explanation .
m. beckerman , physics report , * 129 * , 145 ( 1985 ) .
m. beckerman , rep .
phys . * 51 * , 1047 ( 1988 ) .
r. vandenbosch , annu .
nuc . part .
sci . * 42 * , 447 ( 1992 ) .
a. b. balantekin and n. takigawa , rev .
phys . * 70 * , 77 ( 1998 ) .
m. dasgupta , d.j .
hinde , n. rowley and a.m. stefanini , annu .
nucl . part .
sci . * 48 * , 401 ( 1998 ) .
k. hagino and n. takigawa , phys .
c * 55 * , 276 ( 1997 ) .
n. rowley , g.r .
satchler and p.h .
stelson , phys . lett . *
b254 * , 25 ( 1991 ) .
et al . , _ phys .
lett . * 89 * , 052701 ( 2002 ) .
jiang , h.esbensen , b.b .
back , r.v.f .
janssens and k.e .
rehm , phys .
c. * 69 * , 014604 ( 2004 ) .
et al . , _ phys .
* 93 * , 012701 ( 2004 ) .
m. beckerman , j. ball , h. enge , m. salomaa , a. sperduto , s. gazes , a. dirienzo and j.d .
molitois , phys .
c * 23 * , 1581 ( 1982 ) .
a.m. stefanini _ et al .
* 74 * , 864 ( 1995 ) .
m. beckerman _
et al . , _ phys . rev .
c * 25 * , 837 ( 1982 ) .
d. ackermann _ et al .
a609 * , 91 ( 1996 ) .
rehm , h. esbensen , j. gehring , b. glagola , d. henderson , w. kutschera , m. paul , f. soramel and a.h .
wuosmaa , phys . lett .
* b317 * , 31 ( 1993 ) .
halbert , j.r .
beene , d.c .
hensley , k. honkanen , t.m .
semkow , v. abenante , d.g .
sarantites , and z. li , phys .
c. * 40 * , 2558 ( 1989 ) .
davids and j.d .
larson , nucl .
instr . and meth .
* b40/41 * , 1224 ( 1989 ) ; c.n .
davids , b.b . back , k. bindra , d.j .
henderson , w. kutschera , t. lauritsen , y. nagame , p. sugathan , a.v .
ramayya and w.b .
walters , nucl . instr . and
* b70 * , 358 ( 1992 ) .
davids , anl phys . div .
annual report * anl-03 - 23 * , ( 2003 ) , p. 104 .
kovar and d.j .
henderson , anl phys . div .
annual report * anl-90 - 18 * , ( 1990 ) , p. 108 .
pennington , d.j .
henderson , d.seweryniak , k.e .
rehm , c.l .
jiang , c.n .
davids , c.j .
lister , b.j .
zabransky and b. blank , anl phys . div .
annual report * anl-03 - 23 * , ( 2003 ) , p. 105 .
et al . , _ to be submitted to nuclear instrument and methods a. a. gavron , phys .
c * 21 * , 230 ( 1980 ) .
jiang and c.n .
davids , anl phys . div .
annual report * anl-95/14 * , ( 1995 ) , p. 74 . f. phlhofer , nucl . phys . * a280 * , 267 ( 1977 ) .
j. charbonneau , n.v .
de castro faria , j. lecuyer and d. vitoux , bull .
* 16 * , 625 ( 1971 )
. f. videbaek , p.r .
christensen , o. hansen and k. ulbaek , nucl .
a256 * , 301 ( 1976 ) .
m. r. braunstein , j.j .
kraushaar , r.p .
michel , j.h .
mitchell , r.j .
peterson , h.p .
blok , h. de vries , phys .
c * 37 * , 1870 ( 1988 ) .
b. singh , at .
data nucl .
data tables * 81 * , 1 ( 1997 ) .
k. hagino , n. rowley and m. dasgupta , phys .
c * 67 * , 054603 ( 2003 ) .
r. bass , nucl .
a231 * , 45 ( 1974 ) .
f. scarlassara , s. beghini , f. soramel , s. signorini , l. corradi , g. montagnoli , d.r .
napoli , a.m. stefanini and z.c .
li , z. phys .
a * 338 * , 171 ( 1991 ) .
a.m. stefanini , l. corradi , h. moreno , l. mller , d.r .
napoli , p. spolaore and e. adamides , phys .
b * 252 * , 43 ( 1990 ) .
janssens , r. holzmann , w. henning , t.l .
khoo , k.t .
lesko , g.s.f .
stephans , d.c . radford and a.m. van den berg , phys .
b * 181 * , 16 ( 1991 ) . w. henning _
et al . _ , private communication ( 1994 ) .
freeman , h. ernst , d.f .
geesaman , w. henning , t.j .
humanic , w. khn , g. rosner , j.p .
schiffer , b. zeidman and f.w .
prosser , phys .
lett * 50 * , 1563 ( 1991 ) . c. j. lin , phys .
lett . * 91 * , 229201 ( 2003 ) . c. l. jiang _ et al .
lett . * 91 * , 229202 ( 2003 ) .
dasso and g. pollarolo , phys .
c * 68 * , 054604 ( 2003 ) .
giraud , s. karataglidis , k.amos and b.a .
robson , phys .
c * 69 * , 064613 ( 2004 ) .
procedings of the international conference fusion03 : from a tunneling nuclear microscope to nuclear processes in matter , nov . 12 - 15 , 2003 , matsushima , japan , prog
. theor . phys .
154 , 2004 .
hinde , p. 1 ; h. esbensen , p. 11 ; c.l .
jiang , p.61 ; c.j .
lin , p. 184 ; g. pollaroro , p. 201 ; m. dasgupta , p. 209 ; d. brink , p. 268 ; a.b .
balantakin , p. 465 .
_ , conference book of international conference fusion03 : from a tunneling nuclear microscope to nuclear processes in matter , nov .
12 - 15 , 2003 , matsushima , japan , p. 143 | the excitation function for the fusion - evaporation reaction @xmath0ni + @xmath1mo has been measured down to a cross - section of @xmath2 5 nb .
extensive coupled - channels calculations have been performed , which can not reproduce the steep fall - off of the excitation function at extreme sub - barrier energies .
thus , this system exhibits a hindrance for fusion , a phenomenon that has been discovered only recently . in the @xmath3-factor representation introduced to quantify the hindrance , a maximum is observed at @xmath4=120.6 mev , which corresponds to 90% of the reference energy @xmath5 , a value expected from systematics of closed - shell systems .
a systematic analysis of ni - induced fusion reactions leading to compound nuclei with mass a=100 - 200 is presented in order to explore a possible dependence of the fusion hindrance on nuclear structure . |
the miniboone experiment has performed a search for @xmath2 oscillations at large @xmath3 @xcite , an oscillation signature that had been observed by the lsnd experiment , with 3.8@xmath4 significance @xcite .
this oscillation interpretation relies on the existence of a fourth , sterile neutrino mass eigenstate , with @xmath5 0.1 - 100 ev@xmath6 .
mixing via this fourth mass eigenstate can lead to a small probability amplitude for @xmath2 and @xmath7 oscillations at @xmath8/e[mev]\sim$ ] 1 .
miniboone has previously searched for this type of oscillation using a neutrino beam @xcite , and , under the assumption of cpt conservation , has excluded the lsnd interpretation 98% confidence level ( cl ) @xcite . at the same time , the search revealed an excess of @xmath9 events at low energy @xcite , which is inconsistent with the lsnd excess under the a single sterile neutrino oscillation hypothesis ; however , extensions of this model @xcite offer the possibility of reconsiling the miniboone neutrino results with the lsnd antineutrino result .
these models involve large cp violation which leads to different probabilities for @xmath7 as opposed to @xmath2 oscillations . other models @xcite have also been suggested as explanations , some of which offer predictions for antineutrino running at miniboone . in order to provide another handle on the low - energy excess ,
miniboone was approved in 2007 for extended antineutrino running @xcite , which also enabled miniboone to perform a direct test of the lsnd oscillation interpretation , using antineutrinos .
the miniboone experiment uses 8 gev protons incident on a beryllium target in order to produce mesons which subsequently decay to generate the neutrino beam .
a magnetic field is used at the target to focus positively charged mesons in the forward direction , and defocus negatively charged mesons , to produce a neutrino beam .
reversing the polarity of the magnetic field allows miniboone to switch from a neutrino to an antineutrino beam . the flux @xcite consists primarily of @xmath10 and @xmath11 .
the low @xmath9 and @xmath12 content of the beam minimizes the background to the oscillation search , ensuring sensitivity to small - amplitude ( of order 10@xmath13 ) oscillations .
the @xmath10 flux has a mean energy of @xmath14600 mev .
the miniboone detector @xcite is located at l=541 m from the proton target .
this establishes an @xmath15 similar to lsnd , and therefore sensitivity to @xmath5 1 ev@xmath6 .
the detector is a spherical tank , 12 meters in diameter , filled with mineral oil and lined with photomultiplier tubes ( pmts ) .
the particle detection and identification method relies on the detection of cherenkov and scintillation light emitted by outgoing charged particles which are produced in neutrino interactions .
the antineutrino oscillation analysis @xcite employs the same analysis chain that was implemented in neutrino mode @xcite .
the analysis relies on differentiation between a majority of @xmath10 charged - current quasi - elastic ( ccqe ) events , and @xmath12 ccqe events , which are the signal .
a track - based event reconstruction is implemented , which uses pmt hit topology and timing to identify electron - like or muon - like cherenkov rings from the corresponding ccqe interactions .
the @xmath12 and @xmath10 ccqe spectra are fitted simultaneously as a function of reconstructed antineutrino energy , @xmath16 , and the oscillation parameters @xmath3 and @xmath17 are extracted .
the @xmath10 ccqe prediction is used in the fit in order to provide a constraint to the @xmath18 ccqe prediction , as both spectra are correlated through flux and cross section systematics . the @xmath12 ccqe background prediction , for 3.39@xmath010@xmath1 pot ,
is shown in fig .
[ fig : nue ] .
the background is dominated at low energy by mis - identified @xmath10 events , such as neutral - current ( nc ) @xmath19 interactions .
that is because miniboone can not differentiate between a single photon or a single electron produced in the detector . at high energy ,
the dominant background is ccqe interactions of intrinsic @xmath12 from the beam , produced in @xmath20 and @xmath21 decays .
the sensitivity to @xmath2 oscillations is shown in fig .
[ fig : lim ] . the miniboone sensitivity provides substantial coverage of the lower @xmath3 region allowed at 90% cl by a joint analysis of lsnd and karmen @xcite data @xcite . the antineutrino sensitivity and limit to lsnd - allowed @xmath2 oscillations , from fits to 200 @xmath22 3000 mev .
the miniboone antineutrino dataset corresponds to 3.39@xmath010@xmath1 pot .
also shown is the limit from karmen @xcite.,width=264 ]
the reconstructed energy spectrum of @xmath12 ccqe data is shown in fig .
[ fig : nue ] , overlaid on the predicted @xmath12 ccqe background . at energies above 475 mev
, the data agree with the background prediction within systematic and statistical uncertainties .
the 475 - 675 mev energy region shows a 2.8@xmath4 data fluctuation above background prediction .
this fluctuation forces the miniboone limit , shown in fig .
[ fig : lim ] , to be significantly worse than the sensitivity at lower @xmath3 .
the miniboone best oscillation fit corresponds to ( @xmath23 4.4 ev@xmath6 , @xmath24 0.004 ) .
interestingly , the low - energy region ( 200 - 475 mev ) shows no evidence of an excess . in this range , miniboone observes 61 events , compared to a background expectation of [email protected] ( sys+stat ) events .
table [ tab : excess ] shows the probability ( from a two - parameter fit to the data ) that each of the following hypotheses explains the observed number of low - energy neutrino and antineutrino events : 1 ) same @xmath4 : same nc cross section for neutrinos and antineutrinos .
2 ) @xmath19 scaled : scaled to the number of nc @xmath19 events .
3 ) pot scaled : scaled to number of pot .
4 ) bkgd scaled : scaled to the total number of background events .
5 ) cc scaled : scaled to the number of cc events .
5 ) kaon scaled : scaled to the number of low - energy @xmath20 events .
6 ) neutrino scaled : scaled to the number of neutrino events .
the same @xmath4 , pot scaled , and kaon scaled hypotheses are disfavored as explanations of the miniboone low - energy excess .
the most preferred model is that where the low - energy excess is contributed from only neutrinos in the beam .
.the @xmath26-probability that each hypothesis explains the observed number of low - energy neutrino and antineutrino events , assuming only statistical , fully correlated systematic , and fully uncorrelated systematic errors .
a proper treatment of systematic correlations is in progress . [
cols="<,^,^,^",options="header " , ]
miniboone has performed a blind analysis for @xmath2 oscillations .
the @xmath12 data is found in agreement with the background prediction as a function of @xmath16 .
no strong evidence for oscillations in antineutrino mode has been found , given the current statistics .
interestingly , there is no evidence of significant excess at low energy in antineutrino mode .
this has already placed constraints to various suggested low - energy excess interpretations .
miniboone is currently collecting more antineutrino data , for a total of 5.0@xmath010@xmath1 pot , and has been approved for further running , to collect a total of 10.@xmath010@xmath1 pot .
this will improve sensitivity to oscillations , and allow further investigation of the neutrino - mode low - energy excess .
additional information will be provided by the numi - beam neutrinos detected at miniboone @xcite . | these proceedings summarize the first miniboone electron antineutrino appearance search results , corresponding to a data sample collected for 3.39@xmath010@xmath1 protons on target ( pot ) .
the search serves as a direct test of the lsnd oscillation signature , and provides complementary information which can be used in studies addressing the miniboone neutrino - mode low - energy excess . |
one of the most striking empirical indications for the approximate chiral symmetry of the strong interaction is the existence of the two complete low mass vector meson nonets with opposite parities .
an important phenomenological implication of the vector mesons is their role in the electromagnetic structure of the baryons .
it was realized early on that the @xmath4 and @xmath5 mesons can be used to parametrize the isovector and isoscalar electromagnetic formfactors of the nucleon in terms of ( di)pole forms ( vector meson dominance vmd ) @xcite @xcite .
the vmd idea has also been extended to the parametrization of the axial isovector form factor of the nucleon @xcite .
this is illustrated e.g. by the phenomenological vector meson lagrangian proposed in ref.@xcite , which suggests that the charged weak vector and axial vector nucleon form factors should be expressed as pole terms that contain the masses @xmath6 and @xmath7 of the @xmath4 and @xmath8 mesons and their respective couplings @xmath9 , and @xmath10 to the external @xmath11 weak current .
we here extend the vmd hypothesis to the isoscalar axial form factor of the nucleon along the lines of refs .
@xcite , @xcite and present a model for the @xmath12 coupling constant .
the paper is organized as follows . in section 2
we calculate the weak decay couplings of the @xmath2 ( subsequently denoted by @xmath13 ) and @xmath14 ( subsequently denoted by @xmath15 ) mesons and exploit vmd to derive relations between hadronic and semileptonic couplings . in section 3 we calculate the contribution of the @xmath16 triangle diagram to the @xmath17 vertex and demonstrate its importance for the coupling of the nucleon to an external isoscalar axial vector current .
the paper ends with a short summary .
the neutral weak axial current of the nucleon @xmath18 is determined by the matrix element of the corresponding isotriplet ( @xmath19 ) , singlet ( @xmath20 ) , and octet ( @xmath21 ) quark currents as @xmath22 where @xmath23 is the flavour triplet @xmath24 of up , down and strange quark fields .
the vmd hypothesis implies the following field
current identities : @xmath25 where @xmath26 , @xmath27 , and @xmath28 , which denote the fields of the isotriplet , the octet singlet , and the octet isoscalar axial vector mesons , respectively and @xmath29 are the corresponding decay constants .
the vmd hypothesis allows derivation of interesting relations between various hadronic and semileptonic couplings .
for example , in describing the @xmath30 annihilation coupling constant @xmath31 by means of the chain @xmath32 ( fig .
1a ) the following expression is obtained : @xmath33 another intriguing relation arises between the pion weak decay coupling constant @xmath34 and the coupling constant @xmath35 of the ( hypothetical ) weak decay reaction @xmath36 , which is of interest for the evaluation of short range meson exchange currents in nuclear muon capture reactions of the type displayed in fig .
1b , @xmath37 to obtain this equation we use the lagrangian density @xmath38 in eq .
( 2.4 ) @xmath39 is the external weak axial lepton current . in the limit of equal meson masses
the nonrelativistic quark model leads to the following relation between these decay constants : @xmath40 the numerical value of @xmath41 is determined by the second weinberg sum rule @xcite as @xmath42 , while @xmath9 is given by the ksfr relation as @xmath43 , where @xmath34 is the pion decay constant .
( 2.2 ) allows expression of the couplings @xmath44 and @xmath45 in terms of @xmath41 .
this leads to the following numerical values : @xmath46 according to the universal current - current coupling model the meson - nucleon vertices are described by effective lagrangians of the type @xmath47 in the quark model the wave functions of the low lying vector mesons are described as linear combinations of quark antiquark ( @xmath48 ) pairs in ( three dimensional ) flavour space .
the physical singlet and isoscalar states of both the vector and axial vector meson nonets are moreover predicted to be mixed by the so called ideal mixing angle @xmath49 with the consequence that the states @xmath1 and @xmath50 decouple completely .
this is equivalent to a suppression of @xmath51 transitions ( the okubo
iizuka rule ) .
consequently the decay of , say , the @xmath52 meson into three pions via the chain @xmath53 is represented in qcd by nonplanar diagrams , which are known to be suppressed relative to the leading planar graphs by @xmath54 .
the constants @xmath55 and @xmath56 corresponding to the ideally mixed @xmath13 and @xmath15 mesons are then given as @xmath57 in this way the couplings of the isoscalar axial vector mesons to the external lepton current equal ( up to the sign ) the corresponding couplings of the isocalar vector mesons for which the following relations have been obtained within the quark model @xcite , @xmath58 , and @xmath59 .
the experimental values for the physical mixing angle ( subsequently denoted by @xmath60 ) , which have been extracted from the empirical meson masses with linear or quadratic mass formulae , disagree however with the ideal one .
this difference is comparatively small in the case of the vector meson nonet ( about @xmath61 ) .
even so , the small observed @xmath62 mixing has a non
negligible impact on the weak vector strangeness radius of the nucleon a result obtained in @xcite , @xcite under the assumption of vmd of the isoscalar weak vector current .
the axial vector meson nonet is clearly also non - ideal , as the empirical mixing value for it is @xmath63 @xcite .
the wave functions of the isoscalar axial vector mesons that take into account the mixing between the strange and non strange quarkonia are @xmath64 the substantial deviation of the d e mixing from the ideal mixing angle is explained qualitatively in qcd as being due to strong nonperturbative effects in the isoscalar axial channel @xcite . in the case of the mixed isoscalar axial vector mesons eqs .
( 2.9)(2.10 ) are replaced by @xmath65 the isoscalar axial vector mesons d(1285 ) and e(1420 ) are expected to be important as intermediate states in @xmath66 annihilation as well as in neutrino scattering processes . in this context the weak decay couplings of the @xmath67 mesons can be associated with the quantities @xmath68 .
the interpretation of the latter as the strong meson
nucleon couplings is , however , not obvious .
this follows from the observation that eq .
( 2.1 ) in fact reduces to @xmath69 where @xmath70 is the weak proton axial coupling constant , and @xmath71 denotes the so called weak isoscalar nucleon coupling ( @xmath72 @xcite ) .
this equation suggests that the coupling constants in the @xmath12 and @xmath73 vertices ( in turn denoted by @xmath74 , and @xmath75 ) will be determined by the @xmath1 components of the corresponding wave functions according to @xmath76 here @xmath77 denotes the coupling of the purely strange quarkonia to the nucleon . by the parity invariance ( chiral symmetry ) of the strong interaction
one can assume that the value @xmath78 from the axial vector meson nonet equals the empirical value of @xmath79 , which is extracted from data on the vector meson nonet @xcite .
this yields the following relation @xmath80 in the following section we show that by attributing the f@xmath81(1285)nn coupling to the @xmath3 triangle diagram , which is the dominant one loop contribution to the isoscalar weak axial nucleon coupling , the empirical value for the coupling of the @xmath1 quarkonium to the nucleon is well understood .
the @xmath83 triangle diagram contribution to the @xmath13-nucleon coupling @xmath74 ( fig .
2 ) and the corresponding form factor can be calculated using the @xmath84 lagrangian ( 2.5 ) and the following lagrangians for the @xmath85 and @xmath86 couplings @xmath87 here @xmath88 and @xmath89 in turn denote the pseudovector @xmath90nn and the scalar @xmath91 coupling constants , for which we adopt the values @xmath92 and @xmath93 , respectively .
these values are implied by the semiphenomenological bonn one boson exchange model for the nucleon - nucleon interaction @xcite .
we define the effective d - nucleon vertex by the scalar product between the weak isoscalar axial vector current of the nucleon ( @xmath18 ) and the @xmath2 field ( denoted by @xmath94 ) as @xmath95 the product @xmath96 equals now the quantity @xmath74 introduced by eq .
( 2.16 ) with @xmath97 . to regularize the integral in the triangle diagrams in fig.2
we introduce the same monopole formfactors at the @xmath90nn and @xmath98nn vertices as in the potential model @xcite .
the two triangle diagrams then give the following contribution to the dnn vertex : @xmath99 here @xmath100 and @xmath101 are the mass parameters in the monopole vertex factors , for which we use the values 1.3 gev and 2.0 gev respectively , and the function @xmath102 is defined as @xmath103 the dnn coupling constant is obtained by setting @xmath104 in eq . ( 3.4 ) .
note that as @xmath105 the expression ( 3.4 ) has a small imaginary part when the d - meson is on shell , which we shall ignore .
using for @xmath106 the value 5.7 from the experimental partial decay width @xcite , the real part of ( 3.4 ) at @xmath104 is then 1.9 , which we shall interpret as the value for the dnn coupling constant , i.e. @xmath107 the @xmath108 dependence of @xmath74 is shown in fig.3 .
the @xmath108-dependence corresponds approximately to that of a monopole form with a mass scale parameter @xmath109 2 gev .
it has long been known that the @xmath110nn coupling is strongly underpredicted in the quark model as seen from the unexpectedly large value for the @xmath111nn coupling extracted from data analyses in @xcite .
comparison of our result from eq .
( 3.6 ) shows that the numerical value for @xmath74 associated with the @xmath3 triangle diagram agrees fairly well with the empirical estimate given in eq .
( 2.17 ) . consequently , in spite of the small decay width of the @xmath2 meson the coupling constant in the @xmath12 vertex seems to be completely exhausted by the contribution of the @xmath3 triangle diagram .
+ the result that the empirical value for that coupling constant can be explained by the @xmath112 triangle diagram alone shows that the meson model is able to predict realistic results for reactions that involve the coupling of the @xmath113 system to nucleons .
+ using the experimental value of @xmath71 from @xcite the value of @xmath114 is obtained as @xmath115 which is compatible with the size of @xmath116 obtained by means of the ksfr relation .
this suggests that the vector meson dominance model may be applied in the case of isoscalar axial vector currents . + * acknowledgements * 99 y. nambu , phys . rev .
* 106 * ( 1957 ) 1366 w. r. frazer and j. r. fulco , phys
* 117 * ( 1960 ) 1609 j. j. sakurai , ann .
* 11 * ( 1960 ) 1 s. i. dolinski et al .
, phys . rep .
* 202 * ( 1991 ) 101 v.i ogievetsky and b. m. zupnik , nucl .
* b24 * ( 1970 ) 612 m. hjorth
jensen , m. kirchbach , k. tsushima , and d. o. riska , nucl .
a563 * ( 1993 ) 525 m. kirchbach and h. arenhvel , proc .
int . conf . on physics with gev particle beams , jlich , 2225 august , 1994 , germany , world scientific ( 1995 ) , in press s. weinberg , phys .
18 * ( 1967 ) 507 m. m. nagels , t. a. rijken and j. j. de swart , phys . rev .
* d12 * ( 1975 ) 744 r. l. jaffe , phys.lett . * b229 * ( 1989 ) 275 t. d. cohen , h. forkel , and m. nielsen , phys.lett . * b316 * ( 1993 ) 1 t. bolton et al . , phys . lett .
* b278 * ( 1992 ) 495 m. a. shifman et al .
, nucl . phys . * b147 * ( 1979 ) 448 j. ellis and m. karliner , phys . lett .
* b313 * ( 1993 ) 131 c. elster , k. holinde , and r. machleidt , phys.rept .
* 149 * ( 1987 ) 149 review of particle properties , phys
* d50 * ( 1994 ) 1173 h. genz and g. hhler , phys . lett . * 61b * ( 1976 ) 389 fig .
1 ( a ) @xmath13 exchange contribution to @xmath117 , ( b)@xmath90 exchange contribution to @xmath118 . + fig .
2 the @xmath112 triangle diagram contribution to the @xmath13-nucleon coupling . + fig .
3 the momentum dependence of the @xmath112 triangle diagram contribution to the @xmath17 coupling @xmath119 .
the units of @xmath108 are @xmath120 . | the weak decay as well as the strong nucleon coupling constants of the isoscalar axial vector meson @xmath0 are estimated within the octet quark model .
it is shown that the empirical value for the coupling of @xmath1 quarkonium to nucleons can be understood by attributing the @xmath2-nucleon coupling to the @xmath3 triangle diagram .
-2 cm the coupling of the @xmath2 meson to the isoscalar + axial current of the nucleon m. kirchbach institut fr kernphysik , th darmstadt , d64289 darmstadt , germany and d.o .
riska department of physics , sf00014 university of helsinki , finland |
recently , mcduff has proven a powerful theorem concerning when one four - dimensional _ symplectic ellipsoid _ on @xmath5 @xmath6 embeds into another . to state mcduff s theorem ,
let @xmath7 be the @xmath8 smallest element in the matrix of numbers @xmath9 the number @xmath7 is the @xmath10 _ embedded contact homology _
( ech ) capacity defined for any symplectic four - manifold .
the ech capacities obstruct symplectic embeddings , see @xcite for a summary .
] of @xmath11 .
mcduff showed that the ech capacities give sharp obstructions to symplectic embeddings of ellipsoids : ( * ? ? ?
1.1 ) [ thm : mcduffhofer ] @xmath12 if and only if @xmath13 for all @xmath14 . here
, the symbol @xmath15 means that the embedding is symplectic , while @xmath16 denotes the interior of @xmath11 .
theorem [ thm : mcduffhofer ] connects lattice point enumeration with symplectic geometry . in the present work ,
we explore some of these connections .
specifically , in @xcite mcduff and schlenk found that embeddings of a four - dimensional ellipsoid @xmath3 into a ball are partly determined by an infinite fibonacci staircase , " and in @xcite frenkel and mller studied embeddings of @xmath3 into a scaling of @xmath0 and discovered another staircase involving the pell numbers exactly when @xmath3 symplectically embeds into a ball and frenkel and mller also determine when @xmath3 symplectically embeds into a scaling of @xmath0 for all @xmath17 . ] . in this paper , we apply theorem [ thm : mcduffhofer ] to give new proofs of these results . we also show a surprising relationship between these staircases and a purely combinatorial phenomenon of independent interest concerning ehrhart quasipolynomials of rational polytopes called period collapse . " by exploiting this phenomenon , we prove that another infinite staircase appears when considering symplectic embeddings into scalings of @xmath1 .
the ehrhart quasipolynomial of a @xmath18-dimensional rational polytope @xmath19 is the counting function @xmath20 let @xmath21 be the smallest @xmath22 such that the vertices of @xmath23 are integral .
this is called the _ denominator _ of @xmath19 .
the ehrhart quasipolynimal is always a degree @xmath18 polynomial in @xmath24 with periodic coefficients of period @xmath21 .
the minimal period of @xmath25 is called the _ period _ of @xmath19 , and _ period collapse
_ refers to any situation where the period of @xmath19 is less than the denominator of @xmath19 .
there has been considerable interest in understanding when exactly period collapse occurs , see for example @xcite .
the link we establish between the staircases from @xcite and @xcite and period collapse comes from another theorem explaining when various families of triangles of symplectic interest have the same ehrhart quasipolynomial .
specifically , for positive real numbers @xmath26 and @xmath27 , denote the triangle with vertices @xmath28 and @xmath29 by @xmath30 , and call two triangles _ ehrhart equvialent _ if they have the same ehrhart quasipolynomial . for positive integers
@xmath31 with @xmath32 and @xmath33 relatively prime , if @xmath34 and @xmath35 are ehrhart equivalent we must have @xmath36 the equation comes from equating the linear terms of the corresponding ehrhart quasipolynomials and its short derivation will be presented in [ sec : eqfds ] .
we will have reason to restrict our attention to pairs @xmath37 for which both @xmath14 and @xmath38 divide @xmath39 .
the only such values with @xmath40 are @xmath41 and unless explicitly stated otherwise , we will assume this holds .
our first theorem states that for these values , is the _ only _ obstruction to ehrhart equivalence : [ thm : periodcollapse ] suppose @xmath42 and assume that @xmath32 and @xmath33 are relatively prime
. then @xmath34 and @xmath35 are ehrhart equivalent if and only if @xmath43 satisfies . to explain the relevance of theorem [ thm : periodcollapse ] to symplectic embeddings of ellipsoids , for positive real numbers @xmath44 and @xmath24 , let @xmath45 by theorem [ thm : mcduffhofer ] , @xmath46 if and only if @xmath47 for all @xmath24 .
now assume @xmath48 and @xmath18 are integers .
then @xmath49 is always an integer , so it suffices to check @xmath47 for @xmath24 a positive integer . since @xmath50 for such @xmath24 , we have proven the following in view of theorem [ thm : mcduffhofer ] : [ clm : theclm ] let @xmath48 and @xmath18 be integers
. then @xmath51 if and only if @xmath52 note that by scaling , to determine when one rational ellipsoid symplectically embeds into another , it suffices to consider integral ellipsoids .
now define the function @xmath53 by scaling , the function @xmath54 completely determines when one four - dimensional ellipsoid symplectically embeds into another .
we can determine the solutions of , and so lemma [ clm : theclm ] can be combined with theorem [ thm : periodcollapse ] to better understand the function @xmath54 .
specifically , define the sequence @xmath55 by @xmath56 and @xmath57 for example , the @xmath58 are the odd - index fibonacci numbers familiar from the work of mcduff and schlenk @xcite and the @xmath59 are related to the pell numbers . set @xmath60 . in [ sec : classification ] , we show that the solutions of are precisely the pairs @xmath61 for @xmath62 . also define the sequences a(k , l)_n:=\ { & n , + & n , . and
@xmath63 finally , define the positive real number @xmath64 we always have @xmath65 by combining lemma [ clm : theclm ] with theorem [ thm : periodcollapse ] , we can deduce the following staircase " theorem concerning the function @xmath54 : [ thm : setheorem ] suppose @xmath42 . for @xmath17 in the interval
@xmath66 $ ] , @xmath67 , \\
\frac{a}{\sqrt{\frac{k}{l } a(k , l)_{n } } } & \textrm{if } a \in [ a(k , l)_n , b(k , l)_n ] , \\ \sqrt{\frac{l}{k}a(k , l)_{n+1 } } & \textrm{if } a \in [ b(k , l)_n , a(k , l)_{n+1 } ] .
\end{cases}\ ] ] thus , for @xmath68 the graph of @xmath69 begins with an infinite staircase , see figure [ fig:3_2 ] . for @xmath70 ,
theorem [ thm : setheorem ] is a restatement of ( * ? ? ?
1.1.i ) and for @xmath71 it is a restatement of ( * ? ? ?
1.1.i ) , although as previously mentioned the proofs of ( * ? ? ?
1.1.i ) and ( * ? ?
1.1.i ) do not use theorem [ thm : mcduffhofer ] . for a survey of the methods from @xcite ,
see @xcite ; the methods in @xcite are similar . for @xmath68
an argument using theorem [ thm : mcduffhofer ] ( and lemma [ clm : theclm ] ) shows that if @xmath72 then the only obstruction to symplectically embedding @xmath3 into a scaling of @xmath73 is the volume .
this is explained in the appendix .
the relevance of theorem [ thm : periodcollapse ] to period collapse comes from the observation that if @xmath14 and @xmath38 are relatively prime then the period of @xmath35 is @xmath74 .
in fact , for @xmath75 we can explain how to classify all such triangles for which this period collapse occurs .
specifically , we show : [ thm : periodcollapse2 ] assume that @xmath76 and @xmath77 are relatively prime .
a. if @xmath78 , then the ehrhart quasipolynomial of @xmath34 has period @xmath74 if and only if for some @xmath62 , @xmath79 b. the ehrhart quasipolynomial of @xmath80 has period @xmath81 if @xmath82 for example , theorem [ thm : periodcollapse2 ] gives examples of triangles with arbitrarily high denominator and period @xmath83 , compare ( * ? ? ? * ex .
. one could ask whether the very strong condition that @xmath14 and @xmath38 both divide @xmath39 in theorem [ thm : periodcollapse ] can be replaced by something weaker .
this would have implications for symplectic embeddings of ellipsoids along the lines of theorem [ thm : setheorem ] . without some extra condition on @xmath14 and @xmath38 , theorem [ thm : periodcollapse ]
certainly does not hold .
for example , for @xmath84 there are many examples of triangles @xmath34 that are not ehrhart equivalent to @xmath35 even if @xmath43 satisfies .
moreover , experimental evidence suggests that this condition on @xmath14 and @xmath38 can not be replaced by any weaker condition in theorem [ thm : periodcollapse ] .
it is also interesting to ask whether other infinite staircases appear in the graph of @xmath54 .
work of frenkel and schlenk @xcite implies that @xmath85 is equal to the volume obstruction except on finitely many intervals for which it is linear , and it is suspected by both the authors and schlenk @xcite that the graph of @xmath86 never contains an infinite staircase for even integer @xmath87 .
a simple characterization of period collapse for the family @xmath30 along the lines of theorem [ thm : periodcollapse2 ] is also open . ]
[ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + we are indebted to matthias beck , michael hutchings , dusa mcduff and felix schlenk for helpful conversations and for their encouragement throughout the project , and we thank michael hutchings , felix schlenk , and david frenkel for patiently reading over earlier drafts of this work .
this project grew out of conversations with daniel bragg , peter cheng , caitlin stanton , and olga vasileva during the @xmath88 uc berkeley geometry , topology , and operator algebras rtg summer research program for undergraduates .
this program was funded by the national science foundation .
we begin by developing the combinatorial machinery that will be used in the proof of theorem [ thm : periodcollapse ] . given a triangle of the form
@xmath89 one can use generating functions to compute the ehrhart quasipolynomial of @xmath90 .
this is explained in @xcite . we will only need to consider the special case where @xmath91 , @xmath92 , @xmath93 , and @xmath94 , for @xmath95 positive integers with @xmath96 and @xmath97 relatively prime . in this case , ( * ? ? ?
2.10 ) gives : @xmath98 where @xmath99 denotes the _ fourier - dedekind sum _
@xmath100 here and in the following sections , @xmath101 denotes the primitive @xmath102 root of unity @xmath103 . since theorem [ thm : periodcollapse ]
is an equivalence of quasipolynomials , we must have equality in the coefficients of each power of @xmath24 . equating the linear terms given by yields the diophantine equation , and thus gives the `` only if '' direction of theorem [ thm : periodcollapse ] . in
[ sec : classification ] we completely classify the solutions of when @xmath37 satisfies . in view of , the hard part of theorem [ thm : periodcollapse ] is evaluating the expression @xmath104
so for @xmath43 satisfying , consider the sum @xmath105 writing @xmath106 for @xmath107 , @xmath108 gives @xmath109 the sum @xmath110 can be similarly rewritten to obtain @xmath111 the inner sums of and are _ convolutions _ that can be evaluated explicitly using the fourier transform .
to evaluate the convolutions in and , we will apply the following general lemma : [ lem : keylemma ] let @xmath112 @xmath113 and @xmath48 be integers such that @xmath113 divides neither @xmath114 nor @xmath115 .
then @xmath116 where @xmath117 we will be most interested in the lemma when in addition @xmath113 divides @xmath118 but @xmath119 does not divide @xmath120 . in this case , lemma [ lem : keylemma ] implies that the sum in is equal to @xmath121 .
our proof is given in three steps .
_ this step summarizes the inputs from finite fourier analysis .
if @xmath122 is a function with period @xmath113 , recall that its _ fourier transform _ is the function @xmath123 the convolution of two periodic functions @xmath124 with period @xmath113 is given by @xmath125 a version of the _ convolution theorem _
7.10 ) for the fourier transform says that @xmath126 _ step 2 .
_ we can compute the fourier transform of the family of functions that are relevant to the proof of lemma [ lem : keylemma ] explicitly : [ lemma : hat ] fix positive integers @xmath44 and @xmath48 such that @xmath113 does not divide @xmath17 .
let @xmath127 be the periodic function of period @xmath48 given by @xmath128 then for integers @xmath129 , @xmath130 for notational simplicity , throughout this proof let @xmath131 . for @xmath132
, we have @xmath133 we can break the last line up into two sums and interchange the order of summation in the last sum to get @xmath134 the innermost sum on the right hand side of is @xmath121 if @xmath48 does not divide @xmath135 .
since @xmath136 , when @xmath129 the sum always vanishes and @xmath137 to simplify the summation , let @xmath138 and note that the @xmath139 are the roots of the degree-@xmath48 polynomial @xmath140 .
hence @xmath141 equating the coefficient of @xmath142 on each side of gives @xmath143 and lemma [ lemma : hat ] now follows by combining and . _ step 3 . _
the sum in is @xmath144 , so by and lemma [ lemma : hat ] , @xmath145 the sum in the last line evaluates to @xmath48 if @xmath146 , and otherwise we have @xmath147 this completes the proof .
certain expressions with fourier - dedekind sums can be evaluated directly .
the proof of theorem [ thm : periodcollapse ] uses the following result from ( * ? ? ?
8.8 ) : ( rademacher reciprocity ) [ lem : reciprocity ] let @xmath148 .
then @xmath149 there is another reciprocity statement for @xmath150 : ( * ? ? ?
8.7 ) : [ lem : reciprocity2 ] @xmath151
we will now apply the machinery from the previous section to prove theorem [ thm : periodcollapse ] .
we showed the `` only if '' direction in [ sec : eqfds ] , so it remains to show the `` if '' direction .
assume @xmath14 and @xmath38 both divide @xmath39 , that @xmath96 and @xmath97 are relatively prime , and that @xmath43 satisfies .
the proof that @xmath152 for all positive integers @xmath24 follows in four steps .
_ step 1 . _ both @xmath153 and @xmath154 are quadratic quasipolynomials in @xmath24 . by both
have the same coefficient of @xmath155 , and by both have the same coefficient of @xmath24 .
it remains to show that they have the same constant term .
_ step 2 . _ to evaluate the relevant fourier - dedekind sums , the following elementary fact will be useful : [ lem : usefulfact ] @xmath77 is relatively prime to @xmath156 and @xmath76 is relatively prime to @xmath157 .
since @xmath37 satisfies we can argue case by case .
if @xmath70 , then lemma [ lem : usefulfact ] follows by reducing @xmath158 . if @xmath71 , then reducing mod @xmath159 shows that @xmath76 and @xmath157 are relatively prime , and reducing mod @xmath160 shows that @xmath77 and @xmath156 are relatively prime
finally , if @xmath161 then reducing mod @xmath162 shows that @xmath77 and @xmath156 are relatively prime , and reducing mod @xmath163 shows that @xmath76 and @xmath157 are relatively prime .
_ step 3 . _
we now begin the computation of the fourier - dedekind sums . by , , and we have @xmath164 @xmath165 by we always have @xmath166 , and by lemma [ lem : usefulfact ] , @xmath167 if and only if @xmath168 .
so applying lemma [ lem : keylemma ] with @xmath169 and @xmath170 gives @xmath171 an identical argument gives @xmath172 now @xmath173 and
@xmath174 by , so combining and gives @xmath164 @xmath175 _ step 4 . _ by , we must show @xmath176 for all @xmath177 .
the right hand side of is periodic in @xmath24 with period @xmath74 , and by , the left hand side is as well . for @xmath178 , by the right hand side is equal at @xmath179 and @xmath180 , and is equal at @xmath181 and @xmath182 .
thus , when @xmath183 we can assume @xmath184 , when @xmath185 we can assume @xmath186 or @xmath83 , and when @xmath178 we can assume @xmath187 .
when @xmath186 , we can apply lemma [ lem : reciprocity2 ] to both sides of to get the desired equality . for other @xmath24 , we can apply rademacher reciprocity to evaluate @xmath188 as long as @xmath189 .
this holds for all @xmath190 when @xmath191 , and we can always assume @xmath24 lies in this range by the previous paragraph . for these @xmath24
lemma [ lem : reciprocity ] gives @xmath192 since also holds for @xmath193 , and because @xmath194 by , theorem [ thm : periodcollapse ] follows in this case as well by .
in this section we prove the following : [ prop : classification ] suppose @xmath42 .
then @xmath195 is a solution to if and only if @xmath61 for some @xmath196 . to prove theorem [ thm : periodcollapse ]
, we only need the if " direction of proposition [ prop : classification ] . when @xmath197 , the only if " direction will be used in the proof of theorem [ thm : periodcollapse2 ] .
we include the @xmath161 case here for completeness in view of theorem [ thm : periodcollapse ] .
fix @xmath42 , and consider the pair of congruence relations @xmath198 since @xmath14 and @xmath38 both divide @xmath39 , if @xmath195 satisfies then @xmath199 is a solution to .
we will show that the converse holds , so it suffices to classify the solutions of , and we will then show this is precisely the set of @xmath200 .
we first solve .
the key observation is that if @xmath195 is a solution of , then @xmath201 are integers and @xmath202 and @xmath203 are also solutions to .
motivated by this , we define the involutions @xmath204 we claim that if @xmath205 then either @xmath206 or @xmath207 decreases a coordinate .
suppose that @xmath208 and @xmath209 . then @xmath210 .
if @xmath211 then @xmath212 , if @xmath213 then @xmath214 so @xmath215 , and if @xmath216 then @xmath217 . by examining each of these cases
separately we see that if @xmath40 , @xmath218 , @xmath219 , then @xmath220 in particular , if we assume in addition that @xmath37 satisfies , then @xmath193 .
now define the sequence @xmath221 by @xmath222 , @xmath223 if @xmath195 satisfies then @xmath224 for some @xmath196 .
this follows by induction after applying either @xmath206 or @xmath207 .
another induction using shows that @xmath225 satisfies .
thus , the solutions of and are the same . to show the @xmath226 for all @xmath196 , we induct using and to get @xmath227 we can then apply a final induction using the recurrence relations , and .
there are several basic properties of @xmath69 that significantly simplify the proof of theorem [ thm : setheorem ] : [ lem : axioms ] fix @xmath14 and @xmath38 .
then the function @xmath69 satisfies : a. ( continuity ) @xmath69 is a continuous function of @xmath17 .
b. ( monotonicity ) @xmath69 is a monotonically nondecreasing function of @xmath17 . c. ( subscaling ) @xmath228 when @xmath229 . to prove statement ( i ) , note that for any @xmath230 , if @xmath231 is close enough to @xmath17 then @xmath232 and @xmath233 . statement ( i ) then follows from the observation that if @xmath234 then @xmath235 .
statement ( ii ) follows from the fact that if @xmath236 then @xmath237 .
statement ( iii ) follows because we have @xmath238 for any @xmath229 , and we know that @xmath239 our strategy for proving theorem [ thm : setheorem ] is to calculate @xmath240 , bound @xmath241 from below , and apply lemma [ lem : axioms ] .
we first claim that @xmath240 is always equal to the volume obstruction .
to simplify the notation , we now let @xmath243 and @xmath244 denote @xmath245 and @xmath55 for fixed @xmath37 satisying . by definition , @xmath246 and @xmath247 . to show @xmath248 it suffices by lemma [ clm : theclm ] to show that @xmath249 when @xmath250 by induction , and show that that @xmath251 and @xmath252 are relatively prime . since @xmath253 for @xmath42
, induction also shows that @xmath254 , @xmath255 . then follows from , and theorem [ thm : periodcollapse ] . by lemma [ lem : axioms ] and ,
to prove theorem [ thm : setheorem ] it remains to show that @xmath257 we will show that for the index @xmath258 we have @xmath259 we begin with the proof of .
we have @xmath260 where the last line follows from the well - known identity @xmath261 for @xmath262 .
the fact that @xmath263 follows from an induction using and . since @xmath264
we have that @xmath265 , and follows .
we next prove .
we have that @xmath266 where the second line follows from and the last line follows from .
similarly , by and , @xmath267 by , @xmath268 for @xmath196 even , this is equal to @xmath269 if @xmath270 similarly , holds for @xmath196 odd if @xmath271 induction on and gives @xmath272 , @xmath273 and @xmath274 . by direct computation , and hold for each @xmath42 .
this completes the proof of theorem [ thm : setheorem ] .
we conclude by proving theorem [ thm : periodcollapse2 ] .
assume throughout that @xmath37 satisfies , and continue the notation of the previous section by letting @xmath244 denote @xmath55 .
we first prove the if " statements of theorem [ thm : periodcollapse2 ] . if @xmath275 then , as explained in [ sec : classification ] , @xmath96 and @xmath97 are relatively prime and @xmath43 satisfies .
thus , by theorem [ thm : periodcollapse ] @xmath276 is ehrhart equivalent to @xmath35 and so @xmath276 has period @xmath74 . similarly , if @xmath277 then for @xmath278 @xmath279 and @xmath280 are relatively prime and @xmath281 satisfies .
hence , by theorem [ thm : periodcollapse ] , @xmath282 is ehrhart equivalent to @xmath35 .
thus , @xmath276 has period @xmath74 , since @xmath276 is ehrhart equivalent to @xmath282 .
assume now in addition that @xmath197 .
we now complete the proof of theorem [ thm : periodcollapse ] by proving the only if " statements .
if @xmath96 and @xmath97 are relatively prime and @xmath34 has period @xmath74 , then by we must have @xmath283 we know in addition that @xmath284 .
hence , we can apply lemma [ lem : reciprocity ] and lemma [ lem : reciprocity2 ] to to conclude that @xmath43 satisfies , so the only if " direction of theorem [ thm : periodcollapse2 ] follows by proposition [ prop : classification ] .
if @xmath96 and @xmath97 are not relatively prime , then we must have @xmath71 and it must also be the case that @xmath77 is divisible by @xmath163 and @xmath76 is not divisible by @xmath163 .
define @xmath285 .
we know that @xmath286 and @xmath287 are relatively prime .
moreover , @xmath288 is ehrhart equivalent to @xmath289 . if @xmath289 has period @xmath163 then by we must have @xmath290 since @xmath291 we can apply lemma [ lem : reciprocity ] and lemma [ lem : reciprocity2 ] to to conclude that @xmath292 satisfies
. theorem [ thm : periodcollapse2 ] again then follows by proposition [ prop : classification ] .
the purpose of this appendix is to explain how theorem [ thm : mcduffhofer ] implies the following : [ rm : whythosekl ] the method in the proof of theorem [ thm : appendixtheorem ] can be adapted to establish equations like for other @xmath14 and @xmath38 .
as @xmath14 and @xmath38 vary , all one needs in the proof is a bound like , which can often be found by direct computation using . to place theorem [ thm : appendixtheorem ] in its appropriate context , note that mcduff and schlenk prove @xmath296 for all @xmath297 , and frenkel and mller prove @xmath298 for all @xmath299 .
both proofs use methods that differ from ours . in (
1.3 ) , buse and hind show that for any @xmath300 with @xmath40 , holds for all @xmath301 when theorem [ thm : appendixtheorem ] applies , it gives sharper bounds than .
in particular , for @xmath161 , gives the bound @xmath302 , while theorem [ thm : appendixtheorem ] gives the bound @xmath303 .
if @xmath304 and @xmath305 are two sequences ( indexed with @xmath14 starting at @xmath121 ) , define a new sequence @xmath306 this is the _ sequence sum _ operation originally defined by hutchings in @xcite .
let @xmath307 be the sequence whose @xmath10 term is @xmath7 .
for the sequences @xmath307 , the sequence sum operation satisfies the identity @xmath308 for @xmath309 and @xmath48 any positive integers .
this is proven by an elementary argument in ( * ? ? ?
it follows , see @xcite , that if @xmath17 is rational then there is a finite sequence of positive rational numbers @xmath310 associated to @xmath17 , called a _ weight sequence _ for @xmath17 , such that @xmath311 @xmath312 and @xmath313 the weight sequence is closely related to the continued fraction expansion for @xmath17 , see @xcite . by
, to prove theorem [ thm : appendixtheorem ] , it suffices to show that @xmath314 whenever @xmath315 are nonnegative integers satisfying @xmath316 for @xmath317 , we know by direct computation that @xmath318 if @xmath319 then follows by applying and the cauchy - schwarz inequality .
thus , if @xmath315 satisfies , we can assume that @xmath320 then by , @xmath321 since @xmath322 if @xmath323 theorem [ thm : appendixtheorem ] follows .
99 m. beck and s. robins , _ computing the continuous discretely : integer point enumeration in polyhedra , _ springer @xmath324 .
m. beck , s. sam , and k. woods , _ maximal periods of ( ehrhart ) quasi - polynomials _ , journal of combinatorial theory ser .
a 115 , no . 3 ( 2008 ) , 517 - 525 .
p. biran , _ symplectic packing in dimension @xmath160 _ , geometric and functional analysis , * 7 * ( 1997 ) , 420 - 47 .
mr1466333 zbl 0892.53022 .
o. buse and r. hind , _ ellipsoid embeddings and symplectic packing stability _ , compositio mathematica , ( 2013 ) , 1 - 14 .
k. cieliebak , h. hofer , and f. schlenk , _ quantitative symplectic geometry _ ,
dynamics , ergodic theory , and geometry , 1 - 44 , math .
. publ . * 54 * , cambridge university press , 2007 .
l. guth , symplectic embeddings of polydisks , _ inven .
_ * 172 * ( 2008 ) , 477 - 489 .
mr2393077 , zbl 1153.53060 . c. haase and t. mcallister , _ quasi - period collapse and gl@xmath325-scissors congruence in rational polytopes , _ to appear in proceedings of integer points in polyhedra m. hutchings , _ quantitative embedded contact homology _ , j. diff
* 88 * ( 2011 ) , 231266 .
m. hutchings , _ lecture notes on embedded contact homology _ , arxiv:1303.5789 .
m. hutchings , _ recent progress on symplectic embedding problems in four dimensions _ , pnas 108 ( 2011 ) , 8093 - 8099 .
d. frenkel and d. mller , _ symplectic embeddings of @xmath160-dimensional ellipsoids into cubes _ , arxiv:1210.2266 d. frenkel and f. schlenk , _ the embedding capacity of a 4-dimensional symplectic ellipsoid into integral polydiscs _ , in preparation .
m. gromov , _ pseudo - holomorphic curves in symplectic manifolds _ , inventiones math .
* 82 * ( 1985 ) 307 - 347 .
t. mcallister and k. woods , _ the minimum period of the ehrhart quasipolynomial of a rational polytope , _ j. comb . theory ,
a 109(2 ) : 345 - 352 ( 2005 ) .
d. mcduff , _ symplectic embeddings of @xmath160-dimensional ellipsoids , _
j. topol . 2 ( 2009 ) , 122 .
d. mcduff , _ the hofer conjecture on embedding symplectic ellipsoids ,
_ j. differential geom .
* 88 * ( 2011 ) , 519 - 532 .
d. mcduff , _ symplectic embeddings and continued fractions : a survey _
, japanese journal of mathematics december 2009 , volume 4 , issue 2 , pp 121 - 139 .
d. mcduff and f. schlenk , _ the embedding capacity of @xmath160-dimensional symplectic ellipsoids , _ ann . of math .
* 175 * ( 2012 ) , 1191 - 1282 . f. schlenk , private communication .
f. schlenk , _ embedding problems in symplectic geometry _ , de gruyter expositions in mathematics 40 .
c. h. taubes , _ embedded contact homology and seiberg - witten floer cohomology i v _ , geom .
* 14 * ( 2010 ) , 2497 - 3000 .
k. woods , _ computing the period of an ehrhart quasi - polynomial _ , electron .
j. combin . ,
* 12 * 2005 . | mcduff and schlenk determined when a four - dimensional ellipsoid can be symplectically embedded into a ball , and found that part of the answer is given by a `` fibonacci staircase . '' similarly , frenkel and mller determined when a four - dimensional ellipsoid can be symplectically embedded into the ellipsoid @xmath0 and found that part of the answer is given by a `` pell staircase . '' ech capacities give an obstruction to symplectically embedding one four - dimensional ellipsoid into another , and mcduff showed that this obstruction is sharp .
we use this result to give new proofs of the staircases of mcduff - schlenk and frenkel - mller , and we prove that another infinite staircase arises for embeddings into the ellipsoid @xmath1 .
our proofs relate these staircases to a combinatorial phenomenon of independent interest called period collapse " of the ehrhart quasipolynomial . in the appendix
, we use mcduff s theorem to show that for @xmath2 the only obstruction to embedding an ellipsoid @xmath3 into a scaling of @xmath1 is the volume , and we also give new proofs of similar results for embeddings into scalings of @xmath4 and @xmath0 . |
in this talk i will describe some recent work together with michael dinsdale concerning the relative size of non - perturbative power corrections for qcd event shape observables @xcite . for @xmath1 event shape _ means _
the delphi collaboration have found in a recent analysis that , if the next - to - leading order ( nlo ) perturbative corrections are evaluated using the method of effective charges @xcite , then one can obtain excellent fits to data without including any power corrections @xcite .
in contrast fits based on the use of standard fixed - order perturbation theory in the @xmath2 scheme with a physical choice of renormalization scale equal to the c.m .
energy , require additional power corrections @xmath3 with @xmath4 .
power corrections of this size are also predicted in a model based on an infrared finite coupling @xcite , which is able to fit the data reasonably well in terms of a single parameter . given the delphi
result it is interesting to consider how to extend the method of effective charges to event shape _
distributions _ rather than means . +
consider an @xmath1 observable @xmath5 , e.g. an event shape observable- thrust or heavy - jet mass , @xmath6 being the c.m .
@xmath7 here @xmath8 .
normalised with the leading coefficient unity , such an observable is called an _
effective charge_. the couplant @xmath9 satisfies the beta - function equation @xmath10 here @xmath11 and @xmath12 are universal , the higher coefficients @xmath13 , @xmath14 , are rs - dependent and may be used to label the scheme , together with dimensional transmutation parameter @xmath0 @xcite . the _ effective charge _
@xmath15 satisfies the equation @xmath16 this corresponds to the beta - function equation in an rs where the higher - order corrections vanish and @xmath17 , the beta - function coefficients in this scheme are the rs - invariant combinations @xmath18 eq.(3 ) for @xmath19 can be integrated to give @xmath20 + \int_{0}^{{\cal{r}}(q)}{dx}\left[\frac{b}{\rho(x)}+\frac{1}{{x}^{2}(1+cx)}\right]\;.\ ] ] the dimensionful constant @xmath21 arises as a constant of integration .
it is related to the dimensional transmutation parameter @xmath22 by the exact relation , @xmath23 here @xmath24 with @xmath25 , is the nlo perturbative coefficient .
eq.(5 ) can be recast in the form @xmath26 the final factor converts to the standard convention for @xmath0 . here
@xmath27 is the _ universal _
function @xmath28 and @xmath29 is @xmath30 here @xmath31 is the nnlo ech rs - invariant . if only a nlo calculation is available , as is the case for @xmath1 jet observables , then @xmath32 , and @xmath33 eq.(10 ) can be used to convert the measured data for the observable @xmath15 into a value of @xmath34 bin - by - bin .
such an analysis was carried out in ref .
@xcite for a number of @xmath1 event shape observables , including thrust and heavy jet mass which we shall focus on here . it was found that the fitted @xmath0 values exhibited a clear plateau region , away from the two - jet region , and the region approaching @xmath35 where the nlo thrust distribution vanishes .
the result for 1-thrust corrected for hadronization effects is shown in fig . 1 .
obtained for hadronization corrected 1-thrust data .
@xcite ] another way of motivating the effective charge approach is the idea of `` complete renormalization group improvement '' ( corgi ) @xcite .
one can write the nlo coefficient @xmath36 as @xmath37 hence one can identify scale - dependent @xmath38-logs and rs - invariant `` physical '' uv @xmath6-logs .
higher coefficients are polynomials in @xmath39 .
@xmath40 given a nlo calculation of @xmath39 , parts of @xmath41 are `` rg - predictable '' .
one usually chooses @xmath42 then @xmath43 is @xmath6-independent , and so are all the @xmath44 . the @xmath6-dependence of @xmath5 then comes entirely from the rs - dependent coupling @xmath45 .
however , if we insist that @xmath38 is held constant _ independent of @xmath6 _ the only @xmath6-dependence resides in the `` physical '' uv @xmath6-logs in @xmath43 .
asymptotic freedom then arises only if we resum these @xmath6-logs to _ all - orders_. given only a nlo calculation , and assuming for simplicity that that we have a trivial one loop beta - function @xmath46 so that @xmath47 the rg - predictable terms will be @xmath48 summing the geometric progression one obtains @xmath49 \nonumber \\ & = & 1/b{\ln}(q/{\lambda}_{\cal{r}})\;,\end{aligned}\ ] ] the @xmath38-logs `` eat themselves '' and one arrives at the nlo ech result @xmath50 . +
as we noted earlier , and as will be discussed by klaus hamacher in his talk @xcite , use of nlo effective charge perturbation theory ( renormalization group invariant ( rgi ) perturbation theory ) leads to excellent fits for @xmath1 event shape _ means _
consistent with zero power corrections , as illustrated in figure 2 . taken from ref.@xcite . for means of @xmath1 event shape observables taken from ref.@xcite .
the quality of the `` pure rgi '' fits on the right is noteworthy . ]
given this result it would seem worthwhile to extend the effective charge approach to event shape _ distributions_. it is commonly stated that the method of effective charges is inapplicable to exclusive quantities which depend on multiple scales .
however given an observable @xmath51 depending on @xmath52 scales it can always be written as @xmath53 here the @xmath54 are _ dimensionless _ quantities that can be held fixed , allowing the @xmath55 evolution of @xmath15 to be obtained as before . in the 2-jet region for @xmath1 observables large logarithms @xmath56 arise and need to be resummed to all - orders .
event shape distributions for thrust ( @xmath57 ) or heavy - jet mass ( @xmath58 ) contain large kinematical logarithms , @xmath59 , where @xmath60 .
@xmath61 here @xmath62 , @xmath63 , denote leading logarithms , next - to - leading logarithms , etc . for thrust and heavy - jet mass the distributions _ exponentiate _
@xcite @xmath64 here @xmath65 contains the ll and @xmath66 the nll .
@xmath67 is independent of @xmath68 , and @xmath69 contains terms that vanish as @xmath70 .
it is natural to define an effective charge @xmath71 so that @xmath72 this effective charge will have the expansion @xmath73 here @xmath74 , and the higher coefficients @xmath75 have the structure @xmath76 usually one resums these logarithms to all - orders using the known closed - form expressions for @xmath77 and @xmath78 , where @xmath79 is taken to be the @xmath80 coupling with a `` physical '' scale choice @xmath25 ( @xmath80ps ) .
instead we want to resum logarithms to all - orders in the @xmath81 function ( ech ) .
the form of the @xmath82 rs - invariants ( eq.(4 ) ) means that the @xmath82 have the structure @xmath83 one can then define all - orders rs - invariant @xmath62 and @xmath63 approximations to @xmath81 , @xmath84 the resummed @xmath85 can then be used to solve for @xmath86 by inserting it in eq.(5 ) .
notice that since @xmath21 involves the _ exact _ value of @xmath87 there is no matching problem as in the standard @xmath2ps approach .
the resummed @xmath88 can be straightforwardly numerically computed using @xmath89 with @xmath79 chosen so that @xmath90 .
the same relation with @xmath91 suffices for @xmath92 , although in this case one needs to remove @xmath93 terms , e.g. an @xmath94 term which would otherwise be included in @xmath31 .
this can be accomplished by numerically taking limits @xmath95 with @xmath96 fixed .
+ as we have noted a crucial feature of the effective charge approach is that it resums to all - orders _ rg - predictable _ pieces of the higher - order coefficients , thus the nlo ech result ( assuming @xmath97 for simplicity ) corresponds to an rs - invariant resummation ( c.f .
eq.(13 ) . )
@xmath98 thus even at fixed - order without any resummation of large logs in @xmath81 a _ partial _ resummation of large logs is automatically performed .
furthermore one might expect that the ll ech result contains already nll pieces of the standard @xmath80ps result .
+ ps . the dotted curve is an unexponentiated nlo ech fit .
delphi data at @xmath99 are plotted .
@xmath100 is assumed . ]
ps coefficient at o(@xmath101 ) `` predicted '' from the ll ech result to the exact result ( diamonds ) .
the triangles show the `` prediction '' from the nlo ech result . ] in figure 3 we show various nlo approximations .
notice that the solid curve , which corresponds to the exponentiated nlo ech result , is a surprisingly good fit even in the 2-jet region , whereas the dashed curve which is the nlo @xmath2ps result , has a badly misplaced peak .
the all - orders partial resummation of large logs in eq.(15 ) gives a reasonable 2-jet peak .
figure 4 shows that the nll @xmath80ps coefficients `` predicted '' from the ll ech result by re - expanding it in the @xmath2ps coupling are in good agreement with the exact coeffiecients out to o(@xmath102 ) .
we now turn to fits simultaneously extracting @xmath34 and the size of power corrections @xmath3 from the data . to facilitate this we use the result that inclusion of power corrections effectively shifts the event shape distributions , which can be motivated by considering simple models of hadronization , or through a renormalon analysis @xcite .
thus we define @xmath103 this shifted result is then fitted to the data for 1-thrust and heavy jet mass .
@xmath1 data spanning the c.m .
energy range from @xmath104 gev was used ( see @xcite for the complete list of references ) . the resulting fits for 1-thrust and heavy - jet mass
are shown in figures 5 .
and 6 .. + the ech fits for thrust and heavy jet mass show great stability going from nlo to ll to nll , presumably because at each stage a partial resummation of higher logs is automatically performed .
the power corrections required with ech are somewhat smaller than those found with @xmath80ps , but we do not find as dramatic a reduction as delphi find for the means . this may be because their analysis corrects the data for bottom quark mass effects which we have ignored .
the fitted value of @xmath34 for ech is much smaller than that found with @xmath80ps , ( @xmath105 ( thrust ) and @xmath106 ( heavy - jet mass ) ) .
similarly small values are found with the dressed gluon exponentiation ( dge ) approach @xcite .
a problem with the effective charge resummations is that the @xmath81 function contains a branch cut which limits how far into the 2-jet region one can go .
we are limited to @xmath107 in the fits we have performed .
this branch cut mirrors a corresponding branch cut in the resummed @xmath108 function .
similarly as @xmath109 approaches @xmath110 the leading coefficient @xmath111 vanishes and the effective charge formalism breaks down . we need to restrict the fits to @xmath112 . from the `` rg - predictability '' arguments we might expect that these difficulties would also become apparent for a nnll @xmath80ps resummation .
one will be able to check this expectation when a result for @xmath113 becomes available . and @xmath114 .
solid @xmath115 error ellipses are for ech , dashed are @xmath80ps .
the arrows show the effect of varying the scale between @xmath116 . ]
event shape means have also been studied in dis at hera @xcite . for such processes
one has a convolution of proton pdf s and hard scattering cross - sections , @xmath117 there is no way to directly relate such quantities to effective charges . the dis cross - sections will depend on a _ factorization scale _ @xmath118 , and a renormalization scale @xmath38 at nlo . in principle
one could identify unphysical scheme - dependent @xmath119 and @xmath120 , and physical uv @xmath6-logs , and then by all - orders resummation get the @xmath118 and @xmath38-dependence to `` eat itself '' .
the pattern of logs is far more complicated than the geometrical progression in the effective charge case , and a corgi result for dis has not been derived so far .
instead one can use the principle of minimal sensitivity ( pms ) @xcite , and for an event shape mean @xmath121 look for a stationary saddle point in the @xmath122 plane @xcite .
it turns that there are large cancellations between the nlo corrections for quark and gluon initiated subprocesses .
one can distinguish between two approaches , @xmath123 where one seeks a saddle point in the @xmath122 plane for the sum of parton subprocesses , and @xmath124 where one introduces two separate scales @xmath125 and @xmath126 and finds a saddle point in @xmath127 .
@xmath123 gives power corrections fits comparable to @xmath80ps with @xmath128 .
@xmath124 in contrast gives substantially reduced power corrections .
this is shown in figure 7 for a selection of hera event shape means .
given large cancellations of nlo corrections rg - improvement should be performed _ separately _ for the @xmath129 and @xmath130-initiated subprocesses , and so @xmath124 which indeed fits the data best , is to be preferred .
, and the solid line to the physical scale choice @xmath131 .
the dotted line is @xmath124 and is in much better agreement with the data points .
event shape means in @xmath1 annihilation are well - fitted by nlo perturbation theory in the effective charge approach , without any power corrections being required . with the usual @xmath80ps approach power corrections @xmath132
are required with @xmath133 gev .
similarly sized power corrections are predicted in the model of ref.@xcite .
it would be interesting to modify this model so that its perturbative component matched the effective charge prediction , but this has not been done .
we showed how resummation of large logarithms in the effective charge beta - function @xmath134 could be carried out for @xmath1 event shape distrtibutions .
if the distributions are represented by an exponentiated effective charge then even at nlo a partial resummation of large logarithms is performed .
as shown in figure 3 this results in good fits to the 1-thrust distribution , with the peak in the 2-jet region in rough agreement with the data .
in contrast the @xmath80ps prediction has a badly misplaced peak in the 2-jet region , and is well below the data for the realistic value of @xmath135 mev assumed .
we further showed in figure 4 that the ll ech result contains already a large part of the nll @xmath80ps result .
we found unfortunately that @xmath134 contains a branch point mirroring that in the resummed @xmath77 function .
this limited the fit range we could consider .
we fitted for power corrections and @xmath34 to the 1-thrust distribution and heavy - jet mass distributions , finding somewhat reduced power corrections for the ech fits compared to @xmath80ps , with good stability going from nlo to ll to nll .
the suggestion of the `` rg - predictability '' manifested in figure 4 would be that the nll ech result contains a large part of the nnll @xmath80ps result .
this suggests that the branch point problem which limits the ability to describe the 2-jet peak , would also show up given a nnll analysis .
this can be checked once the @xmath136 function becomes available .
recent work on event shape means in dis was briefly mentioned and seemed to indicate that greatly reduced power corrections are found when a correctly optimised pms approach is used .
mrinal dasgupta , yuri dokshitzer and gavin salam are thanked for their painstaking organisation of this stimulating and productive workshop .
dinsdale and c.j .
maxwell , nucl .
b713 * ( 2005 ) 465 .
g. grunberg , phys . lett .
* b95 * ( 1980 ) 70 ; phys . rev .
* d29 * ( 1984 ) 2315 .
delphi collaboration ( j. abdallah et .
j. * c29 * ( 2003 ) 285 .
dokshitzer and b.r .
webber , phys .
b404 * ( 1997 ) 321 .
p.m. stevenson , phys . rev .
* d23 * ( 1981 ) 2916 .
burby and c.j .
maxwell , nucl .
b609 * ( 2001 ) 193 .
maxwell and a. mirjalili , nucl .
b611 * , 423 ( 2001 ) .
klaus hamacher , invited talk at this workshop .
[ arxiv hep - ex/0605123 ] s. catani , g. turnock , b.r .
webber and l. trentadue , phys . lett . *
b263 * ( 1991 ) 491 .
webber [ hep - ph/9411384 ] .
e. gardi and j. rathsman , nucl . phys .
* b638 * ( 2002 ) 243 . c. adloff _ et al . _
[ h1 collaboration ] , eur .
j. c * 14 * ( 2000 ) 255 [ erratum - ibid .
c * 18 * ( 2000 ) 417 ] [ arxiv : hep - ex/9912052 ] .
dinsdale [ arxiv : hep - ph/0512069 ] . | we introduce and motivate the method of effective charges , and consider how to implement an all - orders resummation of large kinematical logarithms in this formalism
. fits for qcd @xmath0 and power corrections are performed for the @xmath1 event shape obesrvables 1-thrust and heavy - jet mass , and somewhat smaller power corrections found than in the usual approach employing the `` physical scale '' choice . |
braneworld models of the universe in which the observable universe is a four dimensional timelike hypersurface ( brane ) embedded in a higher dimensional ( bulk ) space - time have attracted much recent attention .
this is partly due to the fact that superstring / m - theory seems to require the existence of extra dimensions and the braneworld approach may be one way of reconciling our 3 + 1 dimensional universe with these higher dimensional theories @xcite .
another reason for the current popularity of the braneworld construct is due to the fact that brane cosmology is usually accompanied by new features and is therefore , in principle , falsifiable @xcite .
the simplest randall sundrum ( rs ) braneworld , for instance , gives rise to an evolutionary equation for the brane which differs from standard general relativity at _ early times _
this leads to several interesting consequences .
for instance , the very early universe expands as @xmath5 , instead of the more familar @xmath6 in standard cosmology .
the changed expansion rate causes a scalar field to experience greater damping , which , in turn , allows inflation to occur for a broader class of initial conditions and potentials @xcite .
if the fifth dimension is timelike then the universe generically _ bounces _ and avoids the initial big bang singularity which plagues standard cosmology @xcite .
( the behaviour of anisotropies in the rs scenario can also be very different from that in standard general relativity @xcite . ) a complementary approach to braneworld cosmology pioneered by the dgp model @xcite , allows the universe to accelerate at _ late times _ thus providing a geometrical answer to the riddle posed by dark energy .
models which unify the rs and dgp approaches also lead to several new features @xcite .
for instance ( i ) the phenomenon of dark energy can be transient so that the universe accelerates for a while before settling back into matter dominated expansion , ( ii ) the effective equation of state of dark energy can be _ phantom - like _
( @xmath7 ) , ( iii ) new cosmological singularities can arise in such models @xcite .
such alternative cosmological models provide reasonable fits to the current cosmological data @xcite . in this paper
we address the issue of cosmological evolution on a brane in a theory of gravity whose action includes , in addition to the familiar einstein term , a gauss bonnet contribution .
bonnet terms arise naturally in superstring theories @xcite and their cosmological effects have been discussed in several papers @xcite .
the present paper deals with this issue in greater generality , we examine both cases : when the bulk dimension is spacelike as well as timelike . we also a develop a new pictorial method of analysis which provides qualitative insights into the evolution of the universe in this potentially important new model of gravity .
we begin with the following @xmath8-dimensional ( @xmath9 ) action : @xmath10 \ , , \ ] ] where @xmath11 is the @xmath8-dimensional ricci scalar , @xmath12 is the @xmath8-dimensional cosmological constant , and @xmath13 , where @xmath14 is the @xmath8-dimensional gravitational constant .
the gauss bonnet term @xmath15 is a combination of the ricci scalar , the ricci tensor @xmath16 and the riemann tensor @xmath17 : @xmath18 the constant @xmath19 in ( [ action ] ) is the coupling constant of the gauss bonnet term and for @xmath20 our model reduces to the familiar randall sundrum model @xcite .
the action ( [ action ] ) can be obtained in the low - energy limit of heterotic superstring theory @xcite , in which case @xmath19 can be regarded as the inverse string tension and is positive - definite .
we , therefore , assume @xmath21 throughout this paper .
( we shall explicitely be assuming @xmath9 since for @xmath22 the gauss bonnet term is a topological invariant and does not contribute to the field equations . ) the gravitational equations which result from the action ( [ action ] ) are @xmath23 where @xmath24 - { 1\over 2 } g_{\mu\nu } { l}_{gb}\ , .\end{aligned}\ ] ] the @xmath8-dimensional vacuum solution can be obtained as a product manifold @xmath25 with the line element @xmath26 where @xmath27 is an @xmath28-dimensional space of constant curvature with unit metric @xmath29 . in the equations which follow , @xmath30 denotes the curvature of @xmath27 and takes the values @xmath31 ( positive curvature ) , @xmath32 ( zero curvature ) , and @xmath33 ( negative curvature ) .
the value of the constant @xmath34 determines whether the ( bulk ) fifth dimension is spacelike ( @xmath35 ) or timelike ( @xmath36 ) . in the former case , @xmath37 is a lorenzian manifold , whereas in the latter case , it is a euclidean manifold .
the basic equations of the theory under consideration are @xmath38\frac{dh^2}{dr^2 } \nonumber \\ & & + 2(n-3)r\biggl[2\alpha(n-4)(n-5)h-\varepsilon \biggl\{r^2 + 2\alpha k(n-4)(n-5)\biggl\}\biggl]\frac{dh}{dr } \nonumber \\ & & + 2\alpha r^2(n-3)(n-4)\biggl(\frac{dh}{dr}\biggl)^2+\alpha(n-3)(n-4)(n-5)(n-6)h^2 \nonumber \\ & & -\varepsilon ( n-3)(n-4)h\biggl[r^2 + 2\alpha k ( n-5)(n-6)\biggl ] \nonumber \\ & & -2\lambda r^4+k(n-3)(n-4)r^2+\alpha k^2(n-3)(n-4)(n-5)(n-6)\ , , \\
\nonumber \\ 0&=&(n-2)r\biggl[2\alpha(n-3)(n-4)h-\varepsilon \biggl\{r^2 + 2\alpha k(n-3)(n-4)\biggl\}\biggl]\frac{dh}{dr } \nonumber \\ & & + \alpha(n-2)(n-3)(n-4)(n-5)h^2 \nonumber \\ & & -\varepsilon ( n-2)(n-3)h\biggl[r^2 + 2\alpha k ( n-4)(n-5)\biggl ] \nonumber \\ & & -2\lambda r^4+k(n-2)(n-3)r^2+\alpha k^2(n-2)(n-3)(n-4)(n-5)\ , , \end{aligned}\ ] ] where the former is the @xmath39 component of eq .
( [ beq ] ) , while the latter is the @xmath40 or @xmath41 component acting as a constraint .
the general solution of these equations is @xmath42 where @xmath43 is a constant .
our solution for @xmath44 has two branches , which correspond to the two signs in front of the square root in eq .
( [ horizon ] ) .
we call the family with the minus ( plus ) sign the minus - branch ( plus - branch ) solution . * for @xmath45 , the minus - branch solution has the general relativistic limit as @xmath46 , while there is no general relativistic limit for the plus - branch solution .
( the global structures of this solution were presented in @xcite . ) * for @xmath47 , the plus - branch solution has the general relativistic limit as @xmath46 , while the minus - branch solution does not .
hereafter , we shall be considering a five - dimensional bulk spacetime , for which the metric ( [ eq : product_manifold ] ) reduces to @xmath48\ , , \\
h(r ) & = & \varepsilon k+\frac{r^2}{4\alpha}\biggl(\varepsilon\mp \sqrt{1+\frac{\alpha \mu}{r^{4}}+\frac43\alpha\lambda}\biggr ) \ , , \label{horizon-2}\end{aligned}\ ] ] where @xmath49 , @xmath50 , @xmath51 , and @xmath52 . in this spacetime
, there are two classes of singularities when @xmath53 .
one is the central singularity at @xmath54 and the other is the _ branch singularity _ at @xmath55^{1/4 } > 0 $ ] , when the term inside the square - root in eq .
( [ horizon-2 ] ) vanishes .
the branch singularity exists if @xmath43 is negative , or if @xmath56 for positive @xmath43 .
the position of the three - brane is described by the functions @xmath57 and @xmath58 parametrized by the proper time @xmath59 on the brane .
the tangent vector to the brane is written as @xmath60 where a dot denotes the differentiation with respect to @xmath59 .
the normalization condition @xmath61 leads to @xmath62 and the induced metric of the three - brane @xmath63 is given by @xmath64\ , .\ ] ] the unit normal 1-form to the three - brane @xmath65 is given by @xmath66 where @xmath67 and @xmath68 are satisfied .
the extrinsic curvature of the three - brane is obtained from @xmath69 , where @xmath70 .
we have @xmath71 and @xmath72 then , we obtain the non - zero component of @xmath73 as @xmath74 where a prime denotes differentiation with respect to @xmath75
. the junction condition at the brane is given by @xcite @xmath76_\pm - \delta^a{}_b [ k]_\pm + 2\alpha \left ( 3\varepsilon[j^a{}_b]_\pm -\varepsilon\delta^a{}_b [ j]_\pm - 2 p^a{}_{dbf}[k^{df}]_\pm \right ) = - \varepsilon\kappa_5 ^ 2 \tau^a{}_b \ , , \ ] ] where @xmath77d}+2h_{d[b}r_{f]a } + rh_{a[b}h_{f]d } \ , .\end{aligned}\ ] ] the energy - momentum tensor @xmath78 on the brane is given by @xmath79 where @xmath2 and @xmath80 are the energy density and pressure of a perfect fluid on the three - brane , and the constant @xmath1 is the brane tension .
we have introduced the notation @xmath81_\pm:= x^+-x^- \ , , \ ] ] where @xmath82 is the quantity @xmath83 evaluated either on the @xmath84 or @xmath85 side of the brane , and @xmath86 is the divergence - free part of the riemann tensor , i.e. , @xmath87 where @xmath88 is the covariant derivative on the brane . from the @xmath89 and @xmath39 components of eq .
( [ j - condition ] ) and eq .
( [ norm ] ) , we obtain @xmath90 ^ 2\ , ,
\label{gb - f}\ ] ] where @xmath91 . here
, we have assumed @xmath92-symmetry of reflection with respect to the brane .
this generalized friedmann equation reduces to that obtained by davis @xcite for @xmath45 .
differentiating eq .
( [ gb - f ] ) with respect to @xmath59 and using eq . ( [ norm ] ) and the @xmath89 and @xmath39 components of eq .
( [ j - condition ] ) , we obtain @xmath93 which is the energy - conservation equation on the three - brane .
let us assume that the perfect fluid on the three - brane obeys @xmath94 where we assume that the equation of state of matter on the brane lies within the zeldovich interval @xmath95 due to the dominant energy condition ( equivalently , @xmath96 , where @xmath97 is the equation of state ) . from eq .
( [ em - cons ] ) , we then obtain @xmath98 where @xmath99 is a positive constant , so that @xmath2 is a monotonically decreasing function of @xmath75 for @xmath100 . in this paper
, we shall consider only those solutions of ( [ horizon-2 ] ) and ( [ gb - f ] ) which possess the general - relativistic limit since other solutions may describe physically inadmissible evolution of our brane .
the minus- and plus - branch solutions in ( [ horizon-2 ] ) have the general relativistic limits for @xmath45 and @xmath33 in ( [ gb - f ] ) , respectively .
as mentioned earlier , the action ( [ action ] ) contains the randall sundrum model as a subclass . setting @xmath46 in eq .
( [ gb - f ] ) , one gets the generalised randall sundrum ( rs ) model @xmath101 from this equation and from eq .
( [ energy ] ) , we obtain @xmath102 the randall sundrum model corresponds to @xmath35 , while the dual model with @xmath47 was discussed in @xcite . permitted values of the expansion factor must clearly satisfy @xmath103 .
an interesting consequence of ( [ bounce ] ) is the possibility of singularity - free solutions when @xmath47 @xcite .
we saw in the previous section that the evolution equation for the gauss bonnet brane can be quite complicated and , therefore , difficult to analyze analytically . in this section
, we present a general method of analysis which allows one to study pictorially the behaviour arising from the generic cosmological equation ( [ gb - f ] ) .
we notice that equation ( [ gb - f ] ) , describing the cosmological evolution of the gauss bonnet brane , always has the form of a _ cubic curve _ in the @xmath104 plane : @xmath105 where @xmath106 , @xmath107 and @xmath108 are functions of @xmath75 , @xmath109 is a positive constant and the @xmath110 sign corresponds to @xmath111
. the value of cosmological constant @xmath1 can be positive , negative or zero .
the right - hand side of equation ( [ cubic_curve ] ) has exactly three real zeros in @xmath112 , two of which coincide , namely , @xmath113 , and @xmath114 .
only part of this cubic curve lies in the physical domain @xmath115 , @xmath116 , and it is in this domain that the evolution of the brane takes place .
consequently , the evolution of our brane - universe can be pictured as a point moving along this cubic curve in the physical domain @xmath116 , @xmath115 .
this pictorial representation is very useful in appreciating the full gamut of possibilities for cosmic evolution of this brane . for comparison ,
it is helpful to note that cosmological evolution in general relativity ( gr ) is described by h^2 = _ - , [ eq : quadratic ] where we have set the proportionality term @xmath117 to unity . equation ( [ eq : quadratic ] ) describes a _
quadratic curve _ in the @xmath118 plane .
another example is the randall sundrum brane , which , for every value of @xmath75 , is described by a _ straight line _ in the @xmath104 plane : [ eq : rs_bounce ] h^2 = + - + , where @xmath111 , and we have set the term @xmath119 in ( [ bounce ] ) to unity . before commencing our discussion on the subtleties of cosmological evolution on the gauss bonnet brane , it will be helpful to first consider the different evolutionary possibilities in a spatially flat universe described by the more familiar general - relativistic equation ( [ eq : quadratic ] ) with @xmath120 , where @xmath1 acts as a cosmological constant . in this case the expansion of the universe can proceed in three distinct ways , corresponding to the cases @xmath121 , @xmath122 and @xmath123 .
all three possibilities correspond to motion along the quadratic curve in fig .
[ fig : gr ] .
notice that expansion along the _ entire curve _ from the top ( early times ) to the origin ( late times ) takes place only if @xmath124 . in the absence of a cosmological constant ( @xmath123 ) the origin @xmath125 is reached at @xmath126 .
in contrast , for @xmath122 , the origin is reached in a finite interval of time when the matter density has dropped to @xmath127 . at this point @xmath128 , in other words expansion
ceases and the universe begins to contract .
evolution thereafter proceeds upward along the same curve in _ reverse _ fashion . finally , if @xmath129 , evolution does not proceed all the way to the origin but terminates at some point @xmath130 along the curve . at this point , @xmath131
so that @xmath132 and @xmath133 . the end point @xmath130 of evolution corresponds to the universe s asymptotic approach towards de sitter space .
( this , for instance , would be the case for a spatially flat @xmath12cdm universe which accelerates at late times . ) the evolution of a spatially flat frw universe in gr proceeds along this curve .
the downward arrow indicates _ expansion _ while the upward arrow indicates _
contraction_. the latter is only possible if @xmath122 . for @xmath129 the expansion of the universe terminates at the point @xmath130 at which @xmath134 .
at this point , the universe expands exponentially . for @xmath123 , the origin ( @xmath135 , @xmath136 ) marks the end point of evolution .
the scale of the @xmath137 and @xmath138-axis is arbitrary . ]
let us now discuss the evolution on the gauss bonnet brane in greater detail .
for a spacelike extra dimension , the cosmological equation ( [ gb - f ] ) has the form @xmath139 where @xmath140 and @xmath141 in general , are functions of the scale factor @xmath75 . as mentioned earlier , equation ( [ plus ] ) has the form of a cubic curve in the @xmath142 plane .
the two signs in ( [ ab - plus ] ) correspond to the two different ways of embedding the brane in the bulk space .
in this paper we only consider the upper sign , which has the gr limit .
as discussed in the previous section , the evolution of the braneworld is described by a point moving along the cubic curve in the @xmath118 plane , in the physical domain @xmath115 , @xmath116 , with the parameters of the cubic curve simultaneously changing with time due to the dependence of @xmath107 and @xmath108 on the scale factor ( see below ) .
the evolution can proceed in three distinct ways which are summarized below .
all three cases correspond to @xmath143 in equation ( [ ab - plus ] ) , and the first two also have @xmath144 . 1 . the behaviour of the universe is shown in the left panel of fig .
[ fig - space - pp ] .
the point @xmath145 corresponds to @xmath146 .
during the course of expansion , the motion along the curve is initially downwards from the initial big bang ( bb ) singularity towards @xmath145 .
however , for @xmath145 to be reachable in a finite time interval the brane tension @xmath1 must be negative since only then is @xmath147 permitted .
the point @xmath145 marks a turning point for the evolution along the curve : after this point , the energy density of the universe keeps decreasing while the quantity @xmath148 is increasing . in the case
@xmath149 , we also have @xmath150 at the point @xmath145 . in this case , the hubble parameter passes through an inflection point at @xmath145 .
since = h + h^2 , it follows that @xmath151 when @xmath127 . in other words , @xmath152 for some length of time during the upward motion along the curve _ away from _ @xmath145 .
thus the universe _ accelerates at late times_. note that @xmath153 in @xmath12cdm and @xmath152 is usually associated with a _
phantom _ equation of state @xmath3 in dark - energy models .
( this qualitative behaviour will take place also for sufficiently small values of @xmath154 reached in the course of expansion in the neighbourhood of the point @xmath145 , which will make @xmath107 almost constant in time . )
+ the growth in @xmath155 , however , can not continue indefinitely since @xmath131 eventually , and @xmath156 ( corresponding to the point @xmath130 ) , which implies @xmath157 .
this means that the universe approaches a de sitter - like attractor ( @xmath130 ) at very late times .
we therefore conclude that our gauss bonnet brane can display phantom - like features and super - accelerate at late times , before approaching @xmath158 in the distant future .
note that the big - rip future singularity ( at which @xmath159 ) is absent in this case , which is one of the appealing features of this scenario .
( other braneworld models with this property have been discussed in @xcite . ) 2 . for @xmath129
, the point @xmath145 can never be reached , and expansion proceeds along @xmath160 , culminating in de sitter - like asymptotic expansion at @xmath130 .
+ 3 .
the case with @xmath143 and @xmath161 is shown in the right panel of fig .
[ fig - space - pp ] .
the point @xmath162 is the point of recollapse . at this point
the ( spatially flat ) universe ceases to expand and begins to contract .
the point @xmath162 is reachable either if the brane tension is negative , or if it is positive with the value of @xmath163 lying below the point @xmath162 .
one should note that , theoretically , the scale - factor dependent parameter @xmath164 can change sign during the course of evolution , so that the curve along which the evolution takes place can continuously evolve from that in the right panel of fig .
[ fig - space - pp ] to that in its left panel , and vice versa .
this introduces an obvious modification to the description of the evolution , which does not change in any significant way . the complete set of figures showing the @xmath165 plane are shown in fig .
[ fig : spacelike - all ] of the appendix .
also in this case , there is only one branch of the generic cosmological equation ( [ gb - f ] ) having the gr limit which has the form ( [ cubic_curve ] ) , namely , @xmath166 where @xmath167 and @xmath168 clearly , the theory makes sense only for @xmath161 ( for which the branch singularity does not appear ) , hence , also @xmath143 . a typical graph illustrating the case @xmath169 ( equivalently @xmath170 ) is shown in fig .
[ fig - time1](a ) .
the point @xmath145 corresponds to @xmath171 . the graph corresponding to @xmath172 , which is equivalent to @xmath173 is shown in fig .
[ fig - time1](b ) .
the end points @xmath162 and @xmath145 in fig .
[ fig - time1](a ) are the reverse points of evolution .
the point @xmath174 is the position of a sudden ` quiescent ' singularity of the type described in @xcite .
indeed , the evolution of the universe can not be continued beyond this point because the quantity @xmath175 should change in the same direction ( decrease ) , which is physically impossible .
note that the value of @xmath155 is finite and nonzero at this point , while @xmath176 is divergent .
( this can easily be seen by writing @xmath177 where @xmath178 and noting that @xmath179 at @xmath174 . )
the kretschmann invariant on the brane @xmath180 is given by @xmath181~,\ ] ] and diverges as the _ quiescent _ singularity is approached .
consider now in more detail the evolution of the gauss bonnet brane suggested by fig .
[ fig - time1](a ) . the big bang singularity which featured prominently in fig .
[ fig - space - pp ] has effectively been replaced by the sudden singularity @xmath174 .
the following four possibilities for evolution immediately suggest themselves : 1 .
expansion commences at @xmath174 and proceeds to @xmath162 , which marks a turning point at which @xmath135 .
thereafter , the universe ceases to expand and begins to contract .
the contracting trajectory ends at @xmath174 .
the sudden singularity as @xmath174 marks both the beginning and end of evolution in this scenario .
( the possibility that quantum effects might modify cosmological evolution in the vicinity of such a singularity has been discussed in @xcite ; see also @xcite . ) 2 .
the universe _ contracts _ from the singularity at @xmath174 until it reaches @xmath162 , where it bounces , then expands back to @xmath174 . in this case
, the brane tension must be negative @xmath182 since that is a necessary condition for moving along the trajectory @xmath183 during contraction . in the vicinity of @xmath184 ,
the map @xmath185 is bivalued ( see fig . [ fig - map ] ) , which allows @xmath186 to increase both when @xmath2 increases as well as decreases .
this ambiguity is responsible for the two possibilities discussed above .
the trajectory @xmath187 describes a super - accelerating universe expanding from the singularity @xmath174 , since it suggests that @xmath112 increases while @xmath2 decreases .
( in fact , @xmath188 at the point @xmath174 . ) if @xmath149 , then @xmath152 throughout this phase , and it is unlikely that @xmath189 in this case can describe the real universe . if @xmath190 , then , in the course of the evolution , super - acceleration may be replaced by the `` usual '' acceleration .
if the brane tension is negative , then the point @xmath145 is reached , after which the evolution turns back to the @xmath191 path .
then , depending on the value of the brane tension , it either reaches the singularity @xmath174 again or asymptotically approaches the de sitter state at an intermediate point between @xmath174 and @xmath145 .
4 . for a brane with negative tension ,
one also has the time - reversal of the previous case , which describes a universe contracting either from the de sitter state at an intermediate point between @xmath174 and @xmath145 or from the singularity @xmath174 , proceeding to @xmath145 and then to singularity @xmath174 . for negative values of the brane tension ( @xmath122 ) two values of the matter density @xmath2 map onto a single value of @xmath192 , as illustrated in this figure . ]
we remember that during the evolution the parameters @xmath107 and @xmath108 describing the cubic curve change their values , and it may happen that the curve changes its shape during the evolution , that some critical points leave the physical domain @xmath115 , @xmath193 or , on the contrary , enter this domain .
all such possibilities are quite easy to investigate case by case , but we will not do this in this paper . the complete set of figures showing the @xmath165 plane are shown in fig . [
fig : timelike - all ] of the appendix . in order to address the issue of a bounce in the gauss bonnet brane in more detail ,
let us first consider this issue within the context of the randall sundrum model ( which presents a limiting case of our braneworld ) .
as mentioned earlier , cosmological evolution of the rs brane is described by equation ( [ eq : rs_bounce ] ) which represents a straight line in the @xmath104 plane .
we show this line in the left panel of fig .
[ fig - time3 ] for a time - like extra dimension ( @xmath47 ) .
one can see a close qualitative similarity between the curve on the left panel of this figure and the curve in fig .
[ fig - time1](b ) .
the bouncing scenario in figures [ fig - time1](b ) and [ fig - time3 ] proceeds as follows : the universe begins to contract from the point @xmath145 at which @xmath134 and @xmath194 . in other words , both the starting point and end point of evolution correspond to de sitter space .
an increase in the value of the matter density brings us to the point @xmath195 at which latexmath:[$\rho = @xmath122 is a prerequisite of this model , since otherwise the point @xmath195 can not be reached . ) the universe contracts further from @xmath195 to @xmath162 and in this segment both @xmath2 as well as @xmath197 increase . at @xmath162
the density has reached a finite maximum value while the hubble parameter has declined to zero .
the universe therefore bounces at @xmath162 , then re - expands and evolves in reverse fashion along @xmath198 .
braneworld cosmology has attracted considerable interest during the past decade .
this is partly due to the fact that such models may play an important role in the low energy limit of m - theory / string theory .
another reason for the growing interest in brane dynamics is associated with the new features which some of these models possess and which , in turn , can lead to new cosmological predictions and scenario s .
our attempt in this paper has been to develop a completely general qualitative approach to determine the salient features of a brane embedded in a five dimensional bulk and evolving according to the precepts of einstein gauss bonnet gravity . for this purpose
we show that the 3 + 1 dimensional equations of motion of several popular cosmological models can be depicted as simple curves in the @xmath0 plane .
( here @xmath155 is the hubble parameter , @xmath2 the density and @xmath1 the brane tension . ) for instance , the spatially flat frw universe in gr has the form of a _
quadratic curve _ while the randall - sundrum model describes a _ straight line _ in the @xmath0 plane .
the gauss bonnet brane , on the other hand , describes a _ cubic curve _ in the @xmath199 plane see equations ( [ eq : rs_bounce ] ) , ( [ eq : quadratic ] ) and ( [ cubic_curve ] ) .
this _ pictorial depiction of dynamics _ permits us to discover the salient features of cosmic evolution very simply . applying this approach to the gauss
bonnet brane we discover the following interesting properties : 1 . for a finite region in parameter space the gauss bonnet brane _ accelerates _ at late times .
acceleration can be _ phantom - like _
( @xmath3 ) but does not lead to the eventual destruction of the universe in a _ big - rip _ future singularity . instead , at very late times the expansion of the universe approaches de sitter space and becomes exponential ( @xmath158 ) .
( the possibility that the current expansion of the universe may be phantom - like has evoked much recent interest and discussion ; see @xcite for a non - exhaustive list of papers discussing this issue and @xcite for a summary of recent observational results . ) 2 .
the expansion of the universe may commence from or terminate in a ` sudden ' quiescent singularity , at which the hubble parameter and the density of matter remain finite but @xmath200 diverges .
3 . the universe can evade the initial big bang singularity and _ bounce_. ( this possibility is realized if the fifth dimension is timelike . ) whether any of these properties of the gauss bonnet braneworld is realised in practice is currently an open question which can be answered by : ( i ) a deeper understanding of the embedding of this cosmology within a more fundamental theoretical framework , ( ii ) a comparison with observations .
hm would like to thank iucaa , where this work was conceived and formulated , for warm hospitality .
the work of hm was supported by the grant no .
1071125 from fondecyt ( chile ) and the grant - in - aid for scientific research fund of the ministry of education , culture , sports , science and technology , japan ( young scientists ( b ) 18740162 ) .
cecs is funded in part by an institutional grant from millennium science initiative , chile , and the generous support to cecs from empresas cmpc is gratefully acknowledged . vs and
yus acknowledge the support of the indo - ukrainian program of cooperation in science and technology sponsored by the department of science and technology of india and ministry of education and science of ukraine .
yus is also supported in part by the intas grant no .
05 - 1000008 - 7865 and by grant
5 - 20 of the `` cosmomicrophysics '' programme of the ukrainian academy of sciences .
99 p. hoava and e. witten , nucl .
b * 475 * , 94 ( 1996 ) ; a. lukas , b. a. ovrut , k. s. stelle and d. waldram , phys .
d * 59 * , 086001 ( 1999 ) ; a. lukas , b. a. ovrut and d. waldram , phys .
d * 60 * , 086001 ( 1999 ) .
n. arkani - hamed , s. dimopoulos , and g. dvali , phys .
b * 429 * , 263 ( 1998 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , phys .
b * 436 * , 257 ( 1998 ) .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) ; phys .
lett . * 83 * , 4690 ( 1999 ) .
r. maartens living rev .
* 7 * , 7 ( 2004 ) .
v. sahni , _ cosmological surprises from braneworld models of dark energy _ , jgrg14 : proceedings . edited by w. hikida , m. sasaki , t. tanaka and t. nakamura .
pp . 95115 , arxiv : astro - ph/0502032v3 .
e. j. copeland , a. r. liddle and j. e. lidsey , phys . rev .
d * 64 * , 023509 ( 2001 ) .
y. shtanov and v. sahni , phys .
* b557 * , 1 ( 2003 ) .
r. maartens , v. sahni and t. d. saini , phys .
rev . d. * 63 * , 063509 ( 2001 ) .
g. dvali , g. gabadadze and m. porrati , phys .
b * 485 * , 208 ( 2000 ) ; g. dvali and g. gabadadze , phys .
d * 63 * , 065007 ( 2001 ) ; g. dvali , g. gabadadze , and m. shifman , phys .
d * 67 * , 044020 ( 2003 ) .
v. sahni and yu .
shtanov , jcap * 0311 * , 014 ( 2003 ) ; int . j. mod
d * 11 * , 1515 ( 2000 ) .
y. shtanov and v. sahni , class .
quantum grav .
* 19 * , l101 ( 2002 ) .
u. alam and v. sahni , phys .
d * 73 * , 084024 ( 2006 ) ; r. maartens and e. majerotto , phys .
d * 74 * , 023004 ( 2006 ) ; r. lazkoz , r. maartens and e. majerotto , phys .
d * 74 * , 083510 ( 2006 ) ; r. lazkoz and e. majerotto , jcap * 0707 * , 015 ( 2007 ) .
d. j. gross and j. h. sloan , nucl .
b * 291 * , 41 ( 1987 ) ; m. c. bento and o. bertolami , phys .
* b368 * , 198 ( 1996 ) .
s. nojiri and s. d. odintsov , jhep * 0007 * , 049 ( 2000 ) ; h. m. lee , _ gauss - bonnet interaction in randall - sundrum compactification _ ,
arxiv : hep - th/0010193 ; b. abdesselam and n. mohammedi , phys .
d * 65 * , 084018 ( 2002 ) ; j. e. lidsey , s. nojiri and s. d. odintsov , jhep * 0206 * , 026 ( 2002 ) ; s. nojiri , s. d. odintsov and s. ogushi , phys . rev .
d * 65 * , 023521 ( 2001 ) ; s. nojiri , s. odintsov and s. ogushi , int . j. mod
a * 17 * , 4809 ( 2002 ) ; i. p.
neupane , class .
quantum grav .
* 19 * , 1167 ( 2002 ) ; i. p.
neupane , class .
quantum grav .
* 19 * , 5507 ( 2002 ) ; c. germani and c. f. sopuerta , phys . rev . lett . * 88 * , 231101 ( 2002 ) ; j. e. lidsey and n. j. nunes , phys . rev .
d * 67 * , 103510 ( 2003 ) ; c. charmousis and j. dufaux , phys .
b * 562 * , 118 ( 2003 ) ; s. c. davis , phys .
d * 67 * , 024030 ( 2003 ) ; j. p. gregory and a. padilla , class .
quantum grav . *
20 * , 4221 ( 2003 ) ; n. deruelle and c. germani , nuovo cim . * 118b * , 977 ( 2003 ) ; n. deruelle and j. madore , _ on the quasi - linearity of the einstein``gauss bonnet '' gravity field equations _ , arxiv : gr - qc/0305004 ; c. barcelo , c. germani and c. f. sopuerta , phys . rev .
d * 68 * , 104007 ( 2003 ) ; k. maeda and t. torii , phys .
d * 69 * , 024002 ( 2004 ) ; b. c. paul and m. sami , phys .
d * 70 * , 027301 ( 2004 ) ; g. kofinas and e. papantonoupoulos , jcap * 0412 * , 011 ( 2004 ) ; m. sami , n. savchenko and a. toporensky , phys . rev .
d * 70 * , 123528 ( 2004 ) ; m. sami and v. sahni , phys . rev .
d * 70 * , 083513 , ( 2004 ) ; s. nojiri , s. d. odintsov , and m. sasaki , phys .
d * 71 * , 123509 ( 2005 ) ; g. calcagni , s. tsujikawa and m. sami , class .
quantum grav .
* 22 * , 3977 ( 2005 ) ; g. calcagni , b. de carlos and a. de felice , nucl .
b * 752 * , 404 ( 2006 ) ; l. amendola , c. charmousis and s. c. davis , jcap * 0612 * , 020 ( 2006 ) ; b. m. n. carter and i. p.
neupane , jcap * 0606 * , 004 ( 2006 ) ; t. kobayashi , t. shiromizu and n. deruelle , phys .
d * 74 * , 104031 ( 2006 ) ; t. kobayashi , t. shiromizu and n. deruelle , phys .
d * 74 * , 104031 ( 2006 ) ; t. koivisto and d. f. mota , phys .
d * 75 * , 023518 ( 2007 ) ; g. cognola , e. elizalde , s. nojiri , s. d. odintsov and s. zerbini , _ string - inspired gauss
bonnet gravity reconstructed from the universe expansion history and yielding the transition from matter dominance to dark energy _ ,
arxiv : hep - th/0611198 ; r. a. brown , _ brane world cosmology with gauss bonnet and induced gravity terms _ , phd thesis , arxiv : gr - qc/0701083 ; m. heydari - fard and h. r. sepangi , phys .
d * 75 * , 064010 ( 2007 ) ; b. m. leith and i. p. neupane , jcap * 0705 * , 019 ( 2007 ) ; l. amendola , c. charmousis and s. c. davis , _ solar system constraints on gauss - bonnet mediated dark energy _
, arxiv:0704.0175 ; p. s. apostolopoulos , n. brouzakis , n. tetradis and e. tzavara , _ modified brane cosmologies with induced gravity , arbitrary matter content and a gauss - bonnet term in the bulk _ , arxiv:0708.0469 ; e. elizalde , j. q. hurtado and h. i. arcos , _
de sitter cosmology from gauss bonnet dark energy with quantum effects _ , arxiv:0708.0591 ; shao - feng wu , auttakit chatrabhuti , guo - hong yang and peng - ming zhang , _ crossing the phantom divide in brane cosmology with curvature corrections and brane - bulk energy transfer _
, arxiv:0708.1038 .
e. gravanis and s. willison , phys .
b * 562 * , 118 ( 2003 ) .
s. c. davis , phys .
d * 67 * , 024030 ( 2003 ) . c. garraffo , g. giribet , e. gravanis , and s. willison , j. math .
* 49 * , 042502 ( 2008 ) .
t. torii and h. maeda , phys .
d * 71 * , 124002 ( 2005 ) .
j. d. barrow , class .
quantum grav . * 21 * , l79 ( 2004 ) .
p. tretyakov , a. toporensky , yu .
shtanov and v. sahni , class .
quantum grav .
* 23 * , 3259 ( 2006 ) .
s. nojiri and s. d. odintsov , phys .
b * 595 * , 1 ( 2004 ) .
a. a. starobinsky , , 757 ( 1998 ) ; t. chiba , t. okabe and m. yamaguchi , , 023511 ( 2000 ) ; s. hellerman , n. kaloper and l. susskind , jhep * 0106 * , 003 ( 2001 ) ; r. r. caldwell , , 23 ( 2002 ) ; b. mcinnes , jhep * 0208 * , 029 ( 2002 ) ; r. r. caldwell , m. kamionkowski and n. n. weinberg , phys .
lett . * 91 * , 071301 ( 2003 ) ; s. m. carroll , m. hoffman and m. trodden , , 023509 ( 2003 ) ; p. frampton , , 139 ( 2003 ) ; p. frampton and t. takahashi , , 135 ( 2003 ) ; p. singh , m. sami and n. k. dadhich , , 023522 ( 2003 ) ; v. b. johri , , 041303(r ) ( 2004 ) ; m. p. dabrowski , t. stachowiak , and m. szydowski , , 103519 ( 2003 ) ; s. nojiri and s. d. odintsov , phys.lett .
b * 562 * , 147 ( 2003 ) ; s. nojiri and s. d. odintsov , phys . rev .
d * 70 * , 103522 ( 2004 ) ; m. c. b. abdalla , s. nojiri and s. d. odintsov , class .
quantum grav .
* 22 * , l35 ( 2005 ) ; c. csaki , n. kaloper and j. terning , annals phys . *
317 * , 410 ( 2005 ) ; s. k. srivastava , phys .
b * 619 * , 1 ( 2005 ) ; s. nesseris and l. perivolaropoulos , , 123529 ( 2004 ) ; a. lue and g. d. starkman , , 101501(r ) ( 2004 ) ; j. m. cline , s. jeon and g. d. moore , , 043543 ( 2004 ) ; j. sola and h. stefancic , phys .
b * 624 * , 147 ( 2005 ) ; r .- g .
cai , h .- s .
zhang and a. wang , commun .
. phys . * 44 * , 948 ( 2005 ) ; r. a. brown , _ brane universes with gauss bonnet - induced - gravity _ , arxiv : gr - qc/0602050 ; a. a. andrianov , f. cannata and a. y. kamenshchik , j. phys .
a * 39 * , 9975 ( 2006 ) ; l. p.
chimento , r. lazkoz , r. maartens and i. quiros , jcap * 0609 * , 004 ( 2006 ) ; l. p. chimento and r. lazkoz , phys .
b * 639 * , 591 ( 2006 ) ; f. cannata and a. yu . kamenshchik , _ networks of cosmological histories , crossing of the phantom divide line and potentials with cusps _
, arxiv : gr - qc/0603129 ; j. c. fabris and s. v. b. goncalves , phys .
d * 74 * , 027301 ( 2006 ) ; h. wei and r .-
cai , phys .
b * 634 * , 9 ( 2006 ) ; s. cotsakis and i. klaoudatou , geom .
57 * , 1303 ( 2007 ) ; e. i. guendelman and a. b. kaganovich , phys .
d * 75 * , 083505 ( 2007 ) ; e. i. guendelman and a. b. kaganovich , _ initial singularity , lambda - problem and crossing the phantom divide in scale invariant tmt model _
, arxiv : gr - qc/0703079 .
v. sahni and a. a. starobinsky , int .
d * 15 * , 2105 ( 2006 ) ; u. alam , v. sahni and a. a. starobinsky , jcap * 0702 * , 011 ( 2007 ) .
figures [ fig : spacelike - all ] and [ fig : timelike - all ] describe the evolution of our brane with a spacelike and a timelike extra dimension , respectively . these figures supplement those appearing in the main body of the paper .
the @xmath201 plane is shown for the gb brane with a spacelike extra dimension for the following values of the parameters @xmath107 , @xmath108 in ( [ ab - plus ] ) with the upper sign : ( a ) @xmath202 , ( b ) @xmath203 , ( c ) @xmath204 , ( d ) @xmath205 , ( e ) @xmath206 , and ( f ) @xmath207 ( at the branch singularity ) . here ,
@xmath208 and @xmath209 .
note that the region with @xmath210 or @xmath211 is nonphysical . ]
the @xmath201 plane is shown for the gb brane with a timelike extra dimension for the following values of the parameters @xmath107 , @xmath108 in ( [ ab - minus ] ) : ( a ) @xmath212 , ( b ) @xmath213 , and ( c ) @xmath214 .
the strange singularity characterized by @xmath205 and @xmath215 appears in the extremal case of ( a ) .
here , @xmath216 and @xmath217 .
note that the region with @xmath210 or @xmath211 is nonphysical . ] | we discuss the cosmological evolution of a braneworld in five dimensional gauss bonnet gravity .
our discussion allows the fifth ( bulk ) dimension to be space - like as well as time - like .
the resulting equations of motion have the form of a cubic equation in the @xmath0 plane , where @xmath1 is the brane tension and @xmath2 is the matter density .
this allows us to conduct a comprehensive pictorial analysis of cosmological evolution for the gauss bonnet brane .
the many interesting properties of this braneworld include the possibility of accelerated expansion at late times . for a finite region in parameter space
the accelerated expansion can be _ phantom - like _ so that @xmath3 . at late times , this branch approaches de sitter space ( @xmath4 ) and avoids the _ big - rip _ singularities usually present in phantom models . for a time - like extra dimension the gauss bonnet brane can bounce and avoid the initial singularity . |
numerical computation of miscible displacement in porous media has attracted much attention in recent decades due to its applications in reservoir simulations and exploration of underground oil ; see @xcite .
the model describes the motion of a miscible fluid of two ( or more ) components in porous media , where the velocity of the fluid is given by darcy s law @xmath0 in the last equation , @xmath1 denotes the pressure of the fluid mixture , @xmath2 denotes the permeability of the porous media , and @xmath3 is the viscosity of the fluid depending on the concentration @xmath4 of the first component .
the incompressibility of the fluid is described by @xmath5 where @xmath6 and @xmath7 are given injection and production sources .
the concentration @xmath4 is governed by a convection - diffusion equation @xmath8 where @xmath9 denotes the porosity of the media and @xmath10 denotes the diffusion - dispersion tensor , which is given by @xcite @xmath11 .\end{aligned}\ ] ] in this formula , @xmath12 is a function of the local molecular peclet number @xmath13 , where @xmath14 , @xmath15 , @xmath16 , @xmath17 and @xmath18 are positive constants related to the porous media .
it is straightforward to verify that @xmath19 for some positive constants @xmath20 , @xmath21 and @xmath22 .
existence of weak or semiclassical solutions for the miscible displacement equations was studied in @xcite , and numerical analysis of the model has been done by many authors . in particular , a galerkin fem was studied by ewing and wheeler @xcite , and a galerkin - mixed fem was analyzed by douglas et al @xcite , where the galerkin method was used to solve the parabolic concentration equation and a mixed fem was applied to solve the elliptic pressure equation . for both methods ,
a linearized semi - implicit euler scheme was used for the time stepping and optimal error estimates were presented roughly under the restriction @xmath23 . in @xcite , a characteristic method was applied to the parabolic concentration equation and the mixed fem was used to solve the elliptic pressure equation .
optimal error estimates were established under the same condition , i.e. @xmath24 .
more recently , a galerkin method combined with a post - process technique was studied in @xcite , an euler lagrange localized approximation method was studied in @xcite and a modified method of characteristics combined with mixed fem was studied in @xcite . in all these works ,
error estimates were derived with certain restrictions on the grid ratio . to remove these restrictive conditions ,
a new approach was introduced in @xcite to decouple the discretization errors from the temporal and spatial directions , and optimal error estimates of a galerkin - mixed fem was established without restriction on the grid ratio .
the methodology of @xcite was later successfully applied to other nonlinear parabolic equations , such as the nonlinear schrdinger equation @xcite , the thermistor equations @xcite and the navier - stokes equations @xcite . however , all the analyses presented in these works rely on the global @xmath25 regularity of the `` time - discrete solution '' ( the solution of the linearized pdes ) , which requires the permeability and porosity to be globally smooth in the miscible displacement model . in engineering computations , due to the existence of fault , filling - type karst caves or complex geological composition , the permeability and porosity are often discontinuous across some interfaces .
it is desirable to solve the miscible displacement equations with discontinuous permeability and porosity by stable and accurate numerical methods . for this purpose ,
numerical methods for flow in porous media with discontinuous permeability have been studied by many authors based on linear models .
for example , see @xcite on the approximation of the elliptic pressure equation and see @xcite on the approximation of a parabolic pressure equation from the compressible model .
convergence of finite element methods for general linear elliptic and parabolic interface problems can also be found in @xcite . in this paper , we study stability and convergence of fully discrete fems for the full model of miscible displacement in porous media , where the permeability and porosity are discontinuous across some interfaces .
mathematically , we assume that the domain @xmath26 is partitioned into @xmath27 separated by the interfaces @xmath28 , @xmath29 , as shown in figure [ fig01 ] , and we consider the nonlinear elliptic - parabolic interface problem @xmath30 \left[c\right]=0,\quad \left[d(\u , x)\nabla c\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma_m,\\[10pt ] d(\u , x)\nabla c\cdot{\bf n}=0&\mbox{on}~~\partial\omega,\\[8pt ] c(x,0)=c_0(x)~~ & \mbox{for}~~x\in\omega , \end{array } \right.\label{e - fuel-1}\\[8pt ] & \left\ { \begin{array}{ll } \displaystyle \nabla\cdot\u = q_i - q_p & \mbox{in}~~\omega_m,\\[8pt ] \displaystyle\u=-\frac{k(x)}{\mu(c)}\nabla p & \mbox{in}~~\omega_m,\\[10pt ] \displaystyle \left [ p \right]=0,\quad \left[\u\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma_m,\\[8pt ] \displaystyle \u\cdot{\bf n}=0 & \mbox{on}~~\partial\omega .
\end{array } \right .
\label{e - fuel-3}\end{aligned}\ ] ] in each subdomain @xmath31 , the pressure @xmath1 , the velocity @xmath32 and the concentration @xmath4 are governed by the partial differential equations , and jump conditions are specified across the interfaces .
the permeability @xmath2 and porosity @xmath9 are assumed to be constant in each subdomain @xmath31 but are discontinuous across the interfaces @xmath28 .
clearly , the diffusion - dispersion tensor @xmath33 is an unbounded function of @xmath32 . due to this strong nonlinearity and the coupling of equations ,
previous error estimates presented for the linear interface problems can not be extended here .
a direct application of the traditional error estimates requires undesired restrictions on the grid ratio to control the numerical velocity . in order to avoid any restrictive conditions on the grid ratio
, one has to use the error - splitting technique introduced in @xcite .
however , due to the discontinuity of the permeability and porosity across the interfaces , the solution of - is not globally smooth .
instead , they are at most piecewise smooth @xcite , as assumed in this paper . in this case , the analysis for the galerkin - mixed fem presented in @xcite does not work . in this paper , we show that a decoupled and linearized mixed fem is stable for the nonlinear interface problem by proving that the time - discrete solution is piecewise smooth enough in each subdomain separated by the interfaces .
optimal error estimates are established without restriction on the grid ratio .
we believe that the methodology of this paper , together with lemma [ lemhk0]-[lemhkp1 ] introduced here , can also be applied to other nonlinear parabolic interface problems in engineering and physics .
suppose that the smooth domain @xmath26 is partitioned into @xmath34 , where @xmath31 is enclosed by a smooth interface @xmath28 for @xmath29 , and @xmath35 . for any integer @xmath36 and a subdomain @xmath31 ,
we let @xmath37 and @xmath38 denote the usual sobolev spaces of functions defined on the domain @xmath31 ; see @xcite .
let @xmath39 denote the abbreviations of @xmath40 and define @xmath41 as the subspace of @xmath39 equipped with the norm @xmath42 therefore , the functions in @xmath43 are in @xmath37 for each subdomain @xmath31 , but may not be continuous in the whole domain @xmath26 . to simplify the notations ,
we define @xmath44 , @xmath45 and @xmath46 for any banach space @xmath47 and a function @xmath48 , we define the norm @xmath49 \displaystyle{\rm ess}\!\!\sup_{t\in ( 0,t)}\|g(t)\|_x , & p=\infty
. \end{array } \right.\ ] ] let @xmath50 be a uniform partition of the time interval @xmath51 $ ] with @xmath52 and denote @xmath53 for any sequence of functions @xmath54 , we define @xmath55 .
let @xmath56 denote a quasi - uniform partition of @xmath57 into triangles ( or tetrahedrons ) @xmath58 , @xmath59 . for a triangle @xmath58 with two vertices on the boundary @xmath60 or an interface @xmath28 , we define @xmath61 to be a triangle with one curved side which fit the boundary or the interfaces exactly , with the same vertices as @xmath58 .
let @xmath62 denote the mesh size , and let @xmath63 denote the space of polynomials of degree @xmath64 .
we define the discontinuous finite element space @xmath65 let @xmath66 denote the space of functions in @xmath67 restricted to @xmath28 . to simplify the notations ,
we define @xmath35 , @xmath68 , and define @xmath69 as the space of functions on @xmath70 whose restriction to @xmath28 is in @xmath66 .
let @xmath71 be the space of vector - valued functions @xmath72 such that @xmath73 , @xmath74 on @xmath60 and @xmath75=0 $ ] on @xmath28 , @xmath29 .
let @xmath76 denote the raviart
thomas mixed finite element subspace of @xmath71 introduced in @xcite , which coincides with an element of @xmath77 in each triangle @xmath58 .
moreover , we require that the functions @xmath78 satisfy the boundary condition @xmath79 , @xmath80 , on each boundary edge @xmath81 and the jump condition @xmath82\ , \chi_h\d s=0 $ ] , @xmath80 , on each interface edge @xmath81
. then we have @xmath83 for @xmath78 .
to approximate @xmath1 , @xmath4 , @xmath32 and @xmath84 , we look for @xmath85 and @xmath86 which satisfy the equations @xmath87 & \big(\nabla\cdot \u_h^{n } , \ , \varphi_h\big ) = \big(q_i^{n}-q_p^{n},\ , \varphi_h\big ) , \label{e - fem-2}\\[3pt ] & \big(d ( \u^{n}_h , x)^{-1}{\bf w}_h^{n+1},\,\overline { \bf v}_h\big ) = \big({\cal c}_h^{n+1 } , \ , \nabla \cdot \overline { \bf v}_h \big ) , \label{e - fem-3}\\[3pt ] & \big(\phi(x ) d_\tau { \cal c}_h^{n+1 } , \,\overline\varphi_h\big ) + \big(\nabla\cdot { \bf w}_h^{n+1 } , \ , \overline\varphi_h \big ) - \big(d(\u^n_h , x)^{-1}\u^{n}_h\cdot { \bf w}_h^{n+1},\ , \overline \varphi_h\big ) { \nonumber}\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad = \big(\hat c^{n+1}q_i^{n+1}-{\cal c}_h^{n+1}q_i^{n+1 } , \ , \overline\varphi_h\big ) , \label{e - fem-4}\end{aligned}\ ] ] for any @xmath88 and @xmath89 , where @xmath90 , and the initial data @xmath91 is chosen as the lagrangian interpolation of @xmath92 .
for the initial - boundary value problem ( [ e - fuel-1])-([e - fuel-3 ] ) to be well - posed , we require the compatibility condition @xmath93 and the physical restrictions @xmath94 , @xmath95 , @xmath96 and @xmath97 .
moreover , we assume that the solution of the initial - boundary value problem ( [ e - fuel-1])-([e - fuel-3 ] ) exists and possesses certain piecewise regularity such as @xmath98 for some positive constant @xmath99 .
the main result of this paper is the following theorem .
[ mainthm ] _ under the assumptions - , there exists a positive constant @xmath100 such that when @xmath101 the finite element system ( [ e - fem-1])-([e - fem-4 ] ) admits a unique solution @xmath102 , @xmath103 , which satisfies that @xmath104 where @xmath105 is some positive constant independent of @xmath106 and @xmath107 . _
the proof of theorem [ mainthm ] is presented in section [ section3]-[section5 ] . in section 3 ,
we introduce an intermediate problem , a system of elliptic interface problems , whose finite element solution coincides with @xmath102 , @xmath90 .
then we prove that the solution of the system of elliptic interface problems is piecewise smooth enough in each subdomain separated by the interfaces , and the piecewise regularity is uniform with respect to @xmath106 ( as @xmath108 ) . in section 4 , we present error estimates for the finite element discretization of the elliptic interface problems and prove the boundedness of the finite element solution based on the error estimates . in section 5 , we prove the error estimates in theorem [ mainthm ] based on the boundedness of the finite element solution .
our analysis in section 3 relies on the following two lemmas concerning the piecewise regularity of some elliptic and parabolic interface problems , which are generalizations of the results in @xcite to problems with nonsmooth coefficients , with more precise dependence on the regularity of the coefficients .
the proofs of the lemmas are given in section [ lemhkp000 ] .
[ lemhk0 ] _ if @xmath109 satisfies that @xmath110 for @xmath111 and @xmath112 , and @xmath113 is a solution of @xmath114 \displaystyle [ \phi]=0,\quad \big[a\nabla \phi\cdot{\bf n}\big]=g_m & \mbox{on}~~\gamma_m,~~m=1,\cdots , m,\\[10pt ] \displaystyle a\nabla \phi\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right .
\end{aligned}\ ] ] then @xmath115 where the constant @xmath116 is independent of @xmath106 .
_ [ lemhkp1 ] _ suppose that @xmath117 and @xmath118 , @xmath119 , are solutions of @xmath120 \displaystyle [ \phi^{n+1}]=0,\quad \big[a^{n+1}\nabla \phi^{n+1}\cdot{\bf n}\big]=[{\bf g}^{n+1}\cdot{\bf n } ] & \mbox{on}~~\gamma_m,~~m=1,\cdots , m,\\[10pt ] \displaystyle a^{n+1}\nabla \phi^{n+1}\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \\[8pt ] \phi^0=0&\mbox{in}~~\omega . \end{array } \right . \end{aligned}\ ] ] then we have @xmath121 where the constant @xmath122 @xmath123dependent on @xmath124 @xmath125 is independent of @xmath106 , and @xmath126 _ in the rest part of this paper , we denote by @xmath116 a generic positive constant and by @xmath124 a small generic positive constant , which are independent of @xmath127 , @xmath106 and @xmath107 .
we introduce @xmath128 , @xmath90 , as the solution of an iterative system of linear elliptic interface problems : @xmath129
\displaystyle \left[p^{n}\right]=0,\quad \left[\frac{k(x)}{\mu({\cal c}^{n})}\nabla p^{n}\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma_m,\\[8pt ] \displaystyle \frac{k(x)}{\mu({\cal c}^{n})}\nabla p^{n}\cdot{\bf n}=0&\mbox{on}~~\partial\omega , \end{array } \right .
\label{tde - fuel-2}\\[10pt ] & \left\ { \begin{array}{ll } \phi(x ) d_\tau { \cal c}^{n+1 } -\nabla\cdot(d(\u^{n},x)\nabla { \cal c}^{n+1 } ) + \u^{n } \cdot\nabla{\cal c}^{n+1}= \hat c^{n+1 } q_i^{n+1}-{\cal c}^{n+1}q_i^{n+1 } & \mbox{in}~~\omega_m,\\[10pt ] \left[{\cal c}^{n+1}\right]=0,\quad \left[d(\u^{n},x)\nabla { \cal c}^{n+1}\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma_m,\\[10pt ] d(\u^{n},x)\nabla { \cal c}^{n+1}\cdot{\bf n}=0&\mbox{on}~~\partial\omega , \end{array } \right .
\label{tde - fuel-3}\end{aligned}\ ] ] with the initial condition @xmath130 and the normalization condition @xmath131 . existence and uniqueness of the solution for the linear elliptic interface problems - follow iteratively , and it is easy to see that @xmath132 and @xmath133 at the initial time step . with this definition ,
the fully discrete solution @xmath102 , @xmath90 , defined in - can also be viewed as the finite element solution of - . in this section ,
we establish the uniform piecewise regularity of @xmath134 with respect to @xmath106 .
the following proposition is the main result of this section .
[ erresttdsol ] _ there exists a positive constant @xmath135 such that when @xmath136 , we have @xmath137 _ the importance of this proposition is that the constant @xmath116 does not depend on @xmath106 , which is the key to our error estimates in the next section .
_ proof of proposition [ erresttdsol ] _
let @xmath138 , @xmath139 and @xmath140 .
comparing - with - , we see that @xmath141 , @xmath142 and @xmath143 satisfy the equations @xmath144 , & \mbox{in}~~\omega_m,\\[10pt ] \displaystyle [ e_p^{n}]=0,\quad -\bigg[\frac{k(x)}{\mu({\cal c}^{n})}\nabla e^{n}_p\cdot{\bf n}\bigg]=\bigg[\biggl(\frac{k(x)}{\mu({\cal c}^{n } ) } -\frac{k(x)}{\mu(c^{n})}\biggl)\nabla p^{n}\cdot{\bf n}\bigg ] & \mbox{on}~~\gamma_m , \\[10pt ] \displaystyle \frac{k(x)}{\mu({\cal c}^{n})}\nabla e^{n}_p\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right .
\end{aligned}\ ] ] @xmath145 \displaystyle [ e_c^{n+1}]=0,\quad \big[d(\u^{n},x)\nabla e_c^{n+1}\cdot{\bf n}\big ] = [ ( d(\u^{n+1},x)-d({\bf u}^{n},x))\nabla c^{n+1}\cdot{\bf n } ] & \mbox{on}~~\gamma_m , \\[10pt ] \displaystyle d(\u^{n},x)\nabla e_c^{n+1}\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \\[8pt ] e_c^0=0&\mbox{in}~~\omega , \end{array } \right .
\end{aligned}\ ] ] and @xmath146 respectively , where @xmath147 denotes the truncation error due to the time discretization . from the regularity assumption
for @xmath4 in ( [ strongsolest ] ) we can see that @xmath148 integrating ( [ tderr - fuel-1 ] ) against @xmath149 , we get @xmath150 which together with ( [ tderr - fuel-2 ] ) gives @xmath151 then we integrate ( [ tderr - fuel-3 ] ) against @xmath152 and obtain @xmath153 where we have used the inequality @xmath154 by applying gronwall s inequality to , we see that there exists a positive constant @xmath155 such that when @xmath156 there holds @xmath157 the last inequality , together with ( [ tderr1fuel-1])-([tderr1fuel-2 ] ) , implies that @xmath158 let @xmath159 be the constant defined in lemma [ lemhkp1 ] .
we proceed with a mathematical induction on @xmath160 which clearly holds when @xmath161 ( as @xmath162 and @xmath163 ) .
we shall assume that the above inequality holds for @xmath164 and prove that it also holds for @xmath165 . with
, we can apply lemma [ lemhk0 ] to for @xmath164 and obtain @xmath166\bigg\|_{\overline l^2 } + c\bigg\|\left[\biggl(\frac{k(x)}{\mu({\cal c}^{n } ) } -\frac{k(x)}{\mu(c^{n})}\biggl)\nabla p^{n}\right]\bigg\|_{h^{1/2}(\gamma ) } \leq c\|e^{n}_c\|_{\overline h^1 } , \\ & \|e^{n}_p\|_{\overline h^3 } & \leq c\bigg\|\nabla\cdot\biggl[\biggl(\frac{k(x)}{\mu({\cal c}^{n } ) } -\frac{k(x)}{\mu(c^{n})}\biggl)\nabla p^{n}\biggl]\bigg\|_{\overline h^1 } + c\bigg\|\left[\biggl(\frac{k(x)}{\mu({\cal c}^{n } ) } -\frac{k(x)}{\mu(c^{n})}\biggl)\nabla p^{n}\right]\bigg\|_{h^{3/2}(\gamma ) } \leq c\|e^{n}_c\|_{\overline h^2 } , \end{aligned}\ ] ] and from we see that @xmath167 as a consequence , by the sobolev embedding inequality , we have @xmath168 applying the difference operator @xmath169 to the equation , we obtain @xmath170 \displaystyle [ d_\tau e_p^{n+1}]=0,\quad -\bigg[\frac{k(x)}{\mu({\cal c}^{n+1})}\nabla d_\tau e^{n+1}_p\cdot{\bf n}\bigg]=\big[{\bf f}^{n+1}\cdot{\bf n}\big ] & \mbox{on}~~\gamma_m , \\[10pt ] \displaystyle \frac{k(x)}{\mu({\cal c}^{n+1 } ) } \nabla d_\tau e^{n+1}_p\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right .
\end{aligned}\ ] ] where @xmath171 .\end{aligned}\ ] ] for @xmath164 , from and we can derive that @xmath172 \|d_\tau e_c^{n+1}\|_{l^2 } & \leq c\|e_c^{n+1}\|_{\overline h^2 } + c\|e_u^{n}\|_{\overline h^2}+c\|{\cal e}^{n+1 } \|_{l^2}\\ & \leq c\|e_c^{n+1}\|_{\overline h^2 } + c\|e_c^{n}\|_{\overline h^2}+c\|{\cal e}^{n+1 } \|_{l^2},\end{aligned}\ ] ] from we see that @xmath173 and from we derive that @xmath174 with - , we let @xmath175 and apply lemma [ lemhkp1 ] to for @xmath164 . then we derive that , for @xmath176 , @xmath177 which reduces to @xmath178 by applying gronwall s inequality , there exists a positive constant @xmath179 such that when @xmath180 @xmath181 from the last inequality we see that @xmath182 and from we see that @xmath183 with @xmath184 given by , we can apply lemma [ lemhk0 ] to again and obtain @xmath185 and so @xmath186 the last five inequalities imply that there exists a positive constant @xmath187 such that when @xmath188 we have @xmath189 the mathematical induction on is completed . thus - hold for @xmath190 with the same constant @xmath116 , provided @xmath191 . with the regularity @xmath192 , as shown in , by applying lemma [ lemhk0 ]
to we obtain @xmath193 the proof of proposition [ erresttdsol ] is completed .
based on the finite element discretization of the elliptic interface problems , we prove the following proposition in this section . [ bdunh ] _ there exist positive constants @xmath195 and @xmath196 such that when @xmath197 and @xmath198 , the finite element system ( [ e - fem-1])-([e - fem-4 ] ) admits a unique solution @xmath199 , @xmath103 , such that @xmath200 _ before we prove this proposition , we define some notations below .
let @xmath201 denote the piecewise linear lagrange interpolation operator onto the finite element space @xmath202 .
let @xmath203 denote the @xmath204 projection onto the finite element space @xmath202 , i.e. @xmath205 and let @xmath206 denote the @xmath204 projection onto the finite element space @xmath69 satisfying @xmath207 let @xmath208 be a projection satisfying ( see @xcite for the construction of such a projection operator ) @xmath209 for any edge @xmath81 in the triangulation .
then we have @xmath210 let @xmath211 and , for any fixed integer @xmath212 , let @xmath213 be the finite element solution of the equation @xmath214 ( d(\u^{n},x)^{-1}(\overline { \bf w}_h^{n+1}-{\bf w}^{n+1}),{\bf v}_h ) = ( \overline{\cal c}^{n+1}_h-{\cal c}^{n+1},\nabla\cdot { \bf v}_h),&\forall~ { \bf v}_h\in { \bf h}_h^r , \end{array } \right.\end{aligned}\ ] ] with @xmath215 for the uniqueness of solution , where we define @xmath216 . the pair @xmath217 can be viewed as the ritz projection of @xmath218 by the mixed fem .
we require @xmath136 so that proposition [ erresttdsol ] holds . with the regularity of @xmath219 and @xmath220 given in proposition [ erresttdsol ] , by the theory of mixed fem for linear elliptic equations @xcite , we have @xmath221 & \|\phi-\pi^\gamma_{h}\phi\|_{l^2(\gamma)}\leq ch^{k+1/2}\|\phi\|_{h^{k+1/2}(\gamma ) } , \quad \forall~\phi\in \overline h^{k+1/2}(\gamma)~~\mbox{with
$ 1\leq k\leq r$ } , \label{p - apap}\\[5pt ] & \| { \bf w}^{n+1}-l_h{\bf w}^{n+1}\|_{l^2 } + \|{\bf w}^{n+1}-\overline { \bf w}_h^{n+1}\|_{l^2 } \leq ch^2 , \\[5pt ] & \|{\cal c}^{n+1}-\overline { \cal c}_h^{n+1}\|_{l^2}\leq ch^2 .
\label{p - app}\end{aligned}\ ] ] therefore , by the inverse inequality , we have @xmath222 which implies the existence of a positive constant @xmath223 such that when @xmath224 there holds @xmath225 moveover , we need the following two lemmas in the proof of proposition [ bdunh ] .
[ dtauch ] _ under the regularity of @xmath219 and @xmath226 proved in proposition [ erresttdsol ] ,
we have@xmath227 _ _ proof _ from we derive that @xmath228 where implies that @xmath229 . by choosing @xmath230 in
, we derive that @xmath231 and so , by the inverse inequality , @xmath232 [ lembf41 ] _ if @xmath239 with @xmath240 , then @xmath241\big)_{\gamma_m}\big|\leq c\|g\|_{\overline h^{k+1}}\|{\bf v}_h\|_{l^2}h^{k+1 } .\end{aligned}\ ] ] _ _ proof _ for simplicity , we only prove the 2d case .
the 3d case can be proved in the same way . for a triangle @xmath58 on the boundary ( or an interface ) , we denote by @xmath242 its edge with at two vertices on the boundary ( or an interface ) and denote by @xmath81 the curved edge on the boundary ( or an interface ) . via a rigid rotation , we assume that @xmath242 is on the @xmath243-axis and @xmath81 is parametrized by @xmath244 .
let @xmath245 be a finite element function whose restriction to @xmath242 coincides with @xmath246 , i.e. @xmath247 , satisfying @xmath248 .
then we have , with @xmath249 , @xmath250 which implies that @xmath251 the estimate of @xmath252\d s $ ] on an interface @xmath28 is similar .
firstly , we take @xmath256 in ( [ erre - fem-1 ] ) and get @xmath257\big)_{\gamma_m}\\ & \leq c\|p^n\|_{\overline h^3}\| \u^{n}_h - q_h \u^{n}\|_{l^2}h^{2}\end{aligned}\ ] ] which implies that @xmath258 in particular , we have @xmath259 and so , by the inverse inequality , @xmath260 as a result , there exists a positive constant @xmath261 such that when @xmath262 we have @xmath263 secondly , we proceed with a mathematical induction on @xmath264 which is already proved for @xmath161 . in the following ,
we assume that it holds for @xmath164 and try to prove that it also holds for @xmath165 . since @xmath268 , by applying and gronwall s inequality
, there exists a positive constant @xmath187 such that when @xmath188 we have @xmath269 which , together with , gives @xmath270 from we further derive that @xmath271 which implies that @xmath272 and so , by the inverse inequality , @xmath273 in view of the last inequality , there exists a positive constant @xmath274 such that when @xmath275 we have @xmath276 the mathematical induction on is completed , and proposition [ bdunh ] is proved with @xmath277 and @xmath278 .
in this section , we prove theorem [ mainthm ] based on the boundedness of the fully discrete solution proved in proposition [ bdunh ] . similar as the last section
, for any fixed integer @xmath279 we introduce the ritz projection @xmath280 of @xmath281 as the finite element solution of @xmath282 ( d(\u^{n+1},x)^{-1}(\overline { \bf w}_h^{n+1}-{\bf w}^{n+1}),{\bf v}_h ) = ( \overline c^{n+1}_h - c^{n+1},\nabla\cdot { \bf v}_h),&\forall~ { \bf v}_h\in { \bf h}_h^r , \end{array } \right.\end{aligned}\ ] ] with @xmath283 for the uniqueness of solution .
then there exists a positive constant @xmath284 such that when @xmath285 the following inequalities hold : @xmath286 & \max_{0\leq n\leq n}(\|c^{n}-\overline c_h^{n}\|_{l^2 } + \|{\bf u}^{n}-q_h { \bf u}^{n}\|_{l^2 } + \|{\bf w}^{n}-\overline { \bf w}_h^{n}\|_{l^2 } ) \leq ch^{r+1 } , \\ & \max_{0\leq n\leq n } ( \|{\bf w}^{n}\|_{l^\infty } + \|\overline { \bf w}_h^{n}\|_{l^\infty } ) \leq c , \label{p - app3}\\ & \bigg(\sum_{n=0}^{n-1}\tau \|d_\tau ( c^{n+1}-\overline c_h^{n+1})\|_{l^2}^2 \bigg)^{1/2}\leq ch^{r+1 } .\label{p - apap3}\end{aligned}\ ] ] firstly , by choosing @xmath287 in and @xmath288 in , we derive that @xmath289 in order to make use of , we define @xmath290 as the solution of @xmath291=0,\quad [ \widetilde g^n\cdot{\bf n}]=0 & \mbox{on}~~\gamma_m,\\ \nabla \widetilde g^n\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right.\end{aligned}\ ] ] and substitute @xmath292 into
. then we obtain @xmath293 which together with implies that @xmath294 when @xmath295 , by the inverse inequality we have @xmath296 then we choose @xmath297 in and @xmath298 in . with the boundedness of @xmath299
, we derive that @xmath300 applying gronwall s inequality , there exists a positive constant @xmath301 such that when @xmath302 and @xmath295 we have @xmath303 taking @xmath306 in ( [ erre - fem-13 ] ) , we get @xmath307\big)_{\gamma_m},\\ & \leq c\|p^n\|_{\overline h^{r+1}}\| \u^{n}_h - q_h \u^{n}\|_{l^2}h^{r+1 } , \end{aligned}\ ] ] which implies that @xmath308 to estimate @xmath315 , we define @xmath316 as the solution of @xmath317=0,\quad [ \nabla g^{n}\cdot{\bf n}]=0 & \mbox{on}~~\gamma_m,\\ \nabla g^{n}\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right.\end{aligned}\ ] ] and substitute @xmath318 into . since @xmath319
, it follows that @xmath320 which gives @xmath321 finally , when @xmath322 and @xmath295 we see that - give @xmath323 from - we see that theorem [ mainthm ] holds .
in this section , we prove lemma [ lemhk0 ] and lemma [ lemhkp1 ] , which were used in section [ section3 ] to prove the uniform piecewise regularity of the solution of the linearized pdes .
we shall use the notation @xmath324 , with @xmath325 .
before we prove lemma [ lemhk0 ] , we need to introduce some lemmas below .
[ lemapp ] _ let @xmath326 , @xmath327 , @xmath328 .
let @xmath329 and @xmath330 .
suppose that @xmath331 satisfies the strong ellipticity condition @xmath332 and @xmath333 is a solution of @xmath334 \displaystyle [ \phi]=0,\quad \big[a\nabla \phi\cdot{\bf n}\big]=g & \mbox{on}~~\gamma_{2r } .
\end{array } \right.\end{aligned}\ ] ] then @xmath335 where @xmath336 for any @xmath337 and nonnegative integer @xmath338 . _ _ proof _ to simplify the notations , we relax the dependence on @xmath339 in the generic constant , and set @xmath340 , @xmath341 . differentiating
the equation with respect to @xmath342 for some fixed @xmath343 and denote @xmath344 , we obtain that @xmath345 \displaystyle [ \phi_j]=0,\quad \big[a\nabla \phi_j\cdot{\bf n}\big ] = \partial_jg-\big[\partial_ja\nabla \phi\cdot{\bf n}\big ] & \mbox{on}~~\gamma_{2r } .
\end{array } \right.\end{aligned}\ ] ] where @xmath346 denote the upward unit normal vector on @xmath70 .
if we define @xmath347 as a smooth cut - off function satisfying @xmath348 , @xmath349 in @xmath350 and @xmath351 outside @xmath352 , then times @xmath353 gives @xmath354 which reduces to @xmath355 and from we see that @xmath356 the last two inequalities imply .
the above lemma can be easily extended to the case that @xmath359 is a smooth surface defined by @xmath360 for some smooth function @xmath361 .
[ lemm73 ] _ let @xmath362 , @xmath363 , @xmath364 , and @xmath365 .
suppose that @xmath366 satisfies that @xmath367 and assume that @xmath368 is a solution of @xmath369 \displaystyle [ \phi]=0,\quad \big[a\nabla \phi\cdot{\bf n}\big]=g & \mbox{on}~~\gamma_{2r}(\varphi ) .
\end{array } \right . \end{aligned}\ ] ] then @xmath370 & \|\phi\|_{\overline h^3(s_r(\varphi ) ) } \leq c_r\big(\|f\| _ { \overline h^1(s_{2r}(\varphi ) ) } + \| g\|_{h^{3/2}(\gamma_{2r}(\varphi ) ) } + \|\phi\|_{\overline w^{2,4}(s_{2r}(\varphi ) ) } \big ) , \label{trmmp13}\end{aligned}\ ] ] where @xmath371 for any @xmath372 and nonnegative integer @xmath338 .
_ _ proof _ let @xmath373 denote the coordinates transformation @xmath374 and @xmath375
. under this coordinates transformation , the problem is converted to @xmath376 \displaystyle [ \widetilde\phi]=0,\quad \big[\widetilde a(y ) \nabla_y \widetilde\phi\cdot{\bf n}\big]=\widetilde g(y ) & \mbox{on}~~\gamma_{2r } , \end{array } \right .
\end{aligned}\ ] ] where @xmath377 , @xmath378 , @xmath379 and @xmath380 . by applying lemma [ lemapp ] to the problem , we get @xmath381 transforming back to the @xmath382-coordinates , the last two inequalities imply -
. _ proof of lemma [ lemhk0 ] _ without loss of generality
, we can assume that the functions @xmath383 , @xmath384 and @xmath234 are sufficiently smooth so that the problem has a piecewise @xmath385 solution @xcite .
if we can prove with a constant @xmath386 which does not depend on the extra smoothness of @xmath383 , @xmath384 and @xmath234 , then a compactness argument gives for the nonsmooth @xmath383 , @xmath384 and @xmath234 under the condition of lemma [ lemhk0 ] .
first , multiplying the equation by @xmath387 , it is easy to derive the basic @xmath388 estimate : @xmath389 secondly , by a `` partition of unity '' , there exist a finite number of cylinders @xmath390 , @xmath59 , such that @xmath391 covers @xmath28 , @xmath29 .
moreover , each @xmath392 only intersects one interface @xmath28 and in each @xmath392 , up to a rotation , the interface @xmath28 can be expressed as @xmath393 for some smooth function @xmath394 .
then , by applying lemma [ lemm73 ] , we derive that @xmath395 let @xmath396 .
it is well - known that , by the interior estimates of elliptic equations , there hold @xmath397 the last four inequalities imply that @xmath398 where @xmath124 can be arbitrarily small . finally , by choosing @xmath124 small enough and using the basic @xmath388 estimate ,
the last two inequalities imply . integrating against @xmath399 , it is easy to derive that @xmath400 then , by setting @xmath401 , integrating against @xmath402 gives @xmath403 which further reduces to @xmath404 + \big\|\phi^{-1/2}\nabla\cdot\big(a^{n+1}\nabla \phi^{n+1}\big)\big\|_{\overline l^2}^2\\ & \leq c\|f^{n+1}\|_{\overline l^2}^2+c\|\nabla\cdot {
\bf g}^{n+1}\|_{\overline l^2}^2 + c\big(d_\tau a^{n+1}\nabla\phi^{n}\cdot\nabla \phi^{n } ) -\big(\nabla\phi^{n},d_\tau{\bf g}^{n+1}\big)\\ & \leq c\|f^{n+1}\|_{\overline l^2}^2+c\|{\bf g}^{n+1}\|_{\overline h^1}^2 + c\|d_\tau a^{n+1}\|_{\overline l^2}\|\nabla\phi^{n}\|_{\overline l^4}^2 + \|\nabla\phi^{n}\|_{\overline l^2}\|d_\tau{\bf g}^{n+1}\|_{\overline l^2}d_{n,0}\\ & \leq c\|f^{n+1}\|_{\overline l^2}^2+c\|{\bf g}^{n+1}\|_{\overline h^1}^2 + c_\epsilon\|\nabla\phi^{n}\|_{\overline l^2}^2 + \epsilon(\|\nabla\phi^{n}\|_{\overline h^1}^2 + \|d_\tau{\bf g}^{n+1}\|_{\overline l^2}^2d_{n,0 } ) .\end{aligned}\ ] ] from lemma [ lemhk0 ] we know that @xmath405\|_{h^{1/2}(\gamma)}\\ & \leq c\big\|\phi^{-1/2}\nabla\cdot \big(a^{n+1}\nabla \phi^{n+1}\big)\big\|_{\overline l^2 } + c\|{\bf g}^{n+1}\|_{\overline h^1}\end{aligned}\ ] ] the last two inequalities imply that @xmath404 + c^{-1}\|\phi\|_{\overline h^2}^2\\ & \leq c\|f^{n+1}\|_{\overline l^2}^2+c\|{\bf g}^{n+1}\|_{\overline h^1}^2 + c_\epsilon\|\nabla\phi^{n}\|_{\overline l^2}^2 + \epsilon(\|\nabla\phi^{n}\|_{\overline h^1}^2 + \|d_\tau{\bf g}^{n+1}\|_{\overline l^2}^2d_{n,0}).\end{aligned}\ ] ] summing up the last inequality for @xmath406 , we obtain @xmath407 which further reduces to .
the proof of lemma [ lemhkp1 ] is completed .
in this section , we present numerical examples to support our theoretical error analysis . the computations are performed with the software freefem++ @xcite .
we solve the problem @xmath408 \left[c\right]=0,\quad \left[d(\u , x)\nabla c\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma , \\[10pt ] d(\u , x)\nabla c\cdot{\bf n}=0&\mbox{on}~~\partial\omega,\\[8pt ] c(x,0)=c_0(x)~~ & \mbox{for}~~x\in\omega , \end{array } \right.\\[8pt ] & \left\ { \begin{array}{ll } \displaystyle \nabla\cdot\u= g & \mbox{in}~~\omega_0\cup \omega_1,\\[8pt ] \displaystyle\u=-\frac{k(x)}{\mu(c)}\nabla p & \mbox{in}~~\omega_0\cup \omega_1,\\[10pt ] \displaystyle \left [ p \right]=0,\quad \left[\u\cdot{\bf n}\right]=0 & \mbox{on}~~\gamma , \\[8pt ] \displaystyle \u\cdot{\bf n}=0 & \mbox{on}~~\partial\omega , \end{array } \right.\end{aligned}\ ] ] in the unit ball @xmath409 which is separated by the interface @xmath410 into two subdomains @xmath411 for simplicity , we choose @xmath412 , @xmath413 and choose the permeability and porosity @xmath414 which are smooth in each subdomain but discontinuous across the interface @xmath415 .
let @xmath416 so that @xmath417 . the functions @xmath384 , @xmath234 and the initial data @xmath418 are chosen corresponding to the exact solution @xmath419 0 & \mbox{for}~~x\in \omega\backslash\omega_k,\\ \end{array}\right .
\\[5pt ] & c(x , t)=0.5 + 50\ , p(x , t ) \cos(0.4\ , x_1)\sin(0.4 \ , x_2)\sin(4 t ) , \end{aligned}\ ] ] which satisfy the jump conditions on the interface @xmath415 and the boundary conditions on @xmath60 , while @xmath420 and @xmath421 are discontinuous across the interface @xmath415 . , @xmath422 and @xmath423 .
] , @xmath422 and @xmath423 . ] , @xmath422 and @xmath423 . ]
we partition the domain @xmath26 into quasi - uniform triangles with @xmath424 nodes on the boundary @xmath60 and @xmath425 nodes on the interface @xmath70 with @xmath426 , as shown in figure [ figure1 ] . for simplicity , we set @xmath427 and solve the system by the proposed method using the raviart
thomas linear finite elements up to the time @xmath428 . to test the convergence rate of the proposed method
, we solve the problem for different @xmath106 and @xmath107 , and present the errors of the numerical solutions in table [ extab1 ] , where the convergence rates of @xmath429 and @xmath430 are calculated by the formulas @xmath431 at the finest two meshes . from table
[ extab1 ] we see that the convergence rate of the numerical solution is about second order , which is consistent with our numerical analysis . 0.1 in -0.2 in 0.1 in [ extab1 ] .errors of the linearized mixed fem with @xmath432 . [ cols="<,<,^,^,^,^",options="header " , ]
we have studied the convergence of a linearized mixed fem for a nonlinear elliptic - parabolic interface problem from the model of incompressible miscible flow in porous media .
we showed that the solution of the linearized pdes is piecewise uniformly regular in each subdomain separated by the interfaces if the solution of the original problem is piecewise regular , and established optimal - order error estimates for the fully discrete solution without restriction on the grid ratio .
the analysis presented in this paper , together with lemma [ lemhk0][lemhkp1 ] , may be extended to other nonlinear parabolic interface problems with other time - stepping schemes .
b. amaziane and m. el ossmani , _ convergence analysis of an approximation to miscible fluid flows in porous media by combining mixed finite element and finite volume methods _ , numer .
methods partial differential eq . , 24 ( 2008 ) , pp .
799832 .
h. chen , z. zhou and h. wang , _ an optimal - order error estimate for an @xmath388-galerkin mixed method for a pressure equation in compressible porous medium flow _
, int . j. numer .
modeling , 9 ( 2012 ) , pp .
132 - 148 .
j. douglas , jr . , r. ewing and m.f .
wheeler , _ a time - discretization procedure for a mixed finite element approximation of miscible displacement in porous media _ , rairo anal .
, 17 ( 1983 ) , pp .
249 - 265 .
ewing and j. shen , _ superconvergent error estimates for a class of discontization methods for coupled first - order system with discontinuous coefficients _ , numer .
methods partial diff .
15 ( 1999 ) , pp .
267 - 283 .
raviart and j.m .
thomas , _ a mixed finite element method for @xmath434nd order elliptic problems _ , mathematical aspects of finite element methods , lecture notes in math .
606 , springer - verlag , 1977 , pp .
292 - 315 .
b. rivire , m.f .
wheeler and v. girault , _ improved energy estimates for interior penalty , constrained and discontinuous galerkin methods for elliptic problems .
part i _ , computational geosciences , 3 ( 1999 ) , pp .
337 - 360 .
t. sun and y. yuan , _ an approximation of incompressible miscible displacement in porous media by mixed finite element method and characteristics - mixed finite element method _ , j. comput .
, 228 ( 2009 ) , pp .
391 - 411 . | in this paper , we study the stability and convergence of a decoupled and linearized mixed finite element method ( fem ) for incompressible miscible displacement in a porous media whose permeability and porosity are discontinuous across some interfaces .
we show that the proposed scheme has optimal - order convergence rate unconditionally , without restriction on the grid ratio ( between the time - step size and spatial mesh size ) .
previous works all required certain restrictions on the grid ratio except for the problem with globally smooth permeability and porosity .
our idea is to introduce an intermediate system of elliptic interface problems , whose solution is uniformly regular in each subdomain separated by the interfaces and its finite element solution coincides with the fully discrete solution of the original problem . in order to prove the boundedness of the fully discrete solution , we study the finite element discretization of the intermediate system of elliptic interface problems . |
the following inequality is well known in the literature as the hermite - hadamard integral inequality:@xmath1 where @xmath2 is a convex function on the interval @xmath3 of real numbers and @xmath4 with @xmath5 in @xcite , v.g .
mihean introduced the class of @xmath6convex functions as the following : the function @xmath7\rightarrow % tcimacro{\u{211d } } % % beginexpansion \mathbb{r } % endexpansion $ ] is said to be @xmath6convex , where @xmath8^{2 } , $ ] if for every @xmath9 $ ] and @xmath10 $ ] we have@xmath11 note that for @xmath12 one obtains the following classes of functions : increasing , @xmath13starshaped , starshaped , @xmath14convex , convex and @xmath15convex .
denote by @xmath16 the set of all @xmath6convex functions on @xmath17 $ ] which @xmath18 for recent results and generalizations concerning @xmath14convex and @xmath0 convex functions see @xcite , @xcite and @xcite . in @xcite , m. emin zdemir ,
merve avci and erhan set used the following lemma in order to establish some inequalities for @xmath14 convex functions .
[ lemma 1.1 ] let @xmath19 be a twice differentiable mapping on @xmath20 with @xmath21 and @xmath22.$ ] then the following equality holds:@xmath23 in the same paper @xcite , zdemir et al . discussed the following new results connecting with @xmath14convex functions : [ teo 1.1 ] let @xmath24 where @xmath25 be a twice differentiable function on @xmath20 with @xmath21 and suppose that @xmath26.$ ] if @xmath27 is @xmath28 convex on @xmath29 $ ] for some fixed @xmath30 and @xmath31 $ ] then the following inequality holds:@xmath32where @xmath33 [ co 1.1 ] with the above assumptions given that @xmath34 on @xmath29,$ ] and @xmath35 we have the inequality@xmath36 in @xcite , sarikaya and aktan obtained the following result concerning hermite - hadamard s inequality for functions whose second derivative in absolute value is convex as follows : [ teo 1.2]let @xmath37 be an open interval , @xmath4 with @xmath21 and @xmath38 be twice differentiable mapping such that @xmath39 is integrable and @xmath40 if @xmath41 is a convex function on @xmath29,$ ] then the following inequalities hold:@xmath42 , \\ \\
\frac{\left ( b - a\right ) ^{2}\left ( 3\lambda -1\right ) } { 48}\left [ \left\vert f^{\prime \prime } ( a)\right\vert + \left\vert f^{\prime \prime } ( b)\right\vert \text { } \right ] , \text { \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ for } \frac{1}{2}\leq \lambda \leq 1.% \end{array}% \right.\end{aligned}\ ] ] in theorem [ teo 1.2 ] , if we choose @xmath43 we have@xmath44 .
\label{0}\ ] ] the aim of this paper is to establish some inequalities like those given in @xcite , but now for @xmath6convex functions .
that is , this study is a continuation of @xcite . in order to obtain our results ,
we modified lemma [ lemma 1.1 ] given in the @xcite@xmath45
[ lemma 2.1 ] let @xmath2 be a twice differentiable mapping on @xmath47 where @xmath4 with @xmath48 and @xmath31.$ ] if @xmath49,$ ] then the following equality holds:@xmath50 [ teo 2.1 ] let @xmath51\rightarrow % tcimacro{\u{211d } } % % beginexpansion \mathbb{r } % endexpansion $ ] be a differentiable mapping on @xmath52 such that @xmath26 $ ] where @xmath53 @xmath54 with @xmath55 @xmath56 if @xmath57 is @xmath6convex on @xmath58 $ ] for @xmath59^{2},$ ] @xmath60 then the following inequality holds:@xmath61 ^{\frac{1}{q}}.\end{aligned}\ ] ] suppose that @xmath62 from lemma [ lemma 2.1 ] and using the @xmath6convexity of @xmath63 we have@xmath64 dt \notag \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\left [ \left\vert f^{\prime \prime } ( a)\right\vert \frac{1}{\left ( \alpha + 2\right ) \left ( \alpha + 3\right ) } % + m\left\vert f^{\prime \prime } ( a)\right\vert \left ( \frac{1}{6}-\frac{1}{% \left ( \alpha + 2\right ) \left ( \alpha + 3\right ) } \right ) \right ] \notag\end{aligned}\]]which completes the proof for @xmath62 suppose now that @xmath65 from lemma [ lemma 2.1 ] and using the hlder s integral inequality for @xmath66 we have@xmath67 ^{1-\frac{1}{q}}% \left [ \int_{0}^{1}\left ( t - t^{2}\right ) \left\vert f^{\prime \prime } ( ta+m(1-t)b)\right\vert ^{q}dt\right ] ^{\frac{1}{q } } \label{2}\end{aligned}\]]where @xmath68 since @xmath27 is @xmath6 convex on @xmath29 $ ] we know that for every @xmath10$]@xmath69from ( [ 1])-([3 ] ) we have@xmath70 ^{1-\frac{1}{q}}\left [ \int_{0}^{1}\left ( t - t^{2}\right ) \left\vert f^{\prime \prime } ( ta+m(1-t)b)\right\vert ^{q}dt% \right ] ^{\frac{1}{q } } \\ & \leq & \frac{\left ( mb - a\right ) ^{2}}{2}\left [ \int_{0}^{1}\left ( t - t^{2}\right ) dt\right ] ^{1-\frac{1}{q } } \\ & & \times \left [ \int_{0}^{1}\left ( t - t^{2}\right ) \left [ t^{\alpha } \left\vert f^{\prime \prime } ( a)\right\vert ^{q}+m\left ( 1-t^{\alpha } \right ) \left\vert f^{\prime \prime } ( b)\right\vert ^{q}\right ] dt\right ] ^{% \frac{1}{q } } \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\left ( \frac{1}{6}\right ) ^{1-\frac{1}{q}% } \\ & & \times \left [ \left\vert f^{\prime \prime } ( a)\right\vert ^{q}\frac{1}{% \left ( \alpha + 2\right ) \left ( \alpha + 3\right ) } + m\left\vert f^{\prime \prime } ( b)\right\vert ^{q}\left ( \frac{1}{6}-\frac{1}{\left ( \alpha + 2\right ) \left ( \alpha + 3\right ) } \right ) \right ] ^{\frac{1}{q}}\end{aligned}\]]which is the required . from lemma [ lemma 2.1 ] and
using the hlder s integral inequality , we obtain@xmath74 dt\right ) ^{\frac{1}{q } } \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\left ( \frac{2^{-1 - 2p}\sqrt{\pi } \gamma \left ( 1+p\right ) } { \gamma \left ( \frac{3}{2}+p\right ) } \right ) ^{\frac{1}{p}% } \left ( \left\vert f^{\prime \prime } ( a)\right\vert ^{q}\frac{1}{\alpha + 1}% + m\left\vert f^{\prime \prime } ( b)\right\vert ^{q}\left ( \frac{\alpha } { % \alpha + 1}\right ) \right ) ^{\frac{1}{q } } \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\frac{\left ( \sqrt{\pi } \right ) ^{\frac{1% } { p}}}{2^{\frac{1}{p}}2^{2}}\left ( \frac{\gamma \left ( 1+p\right ) } { \gamma \left ( \frac{3}{2}+p\right ) } \right ) ^{\frac{1}{p}}\left ( \left\vert f^{\prime \prime } ( a)\right\vert ^{q}\frac{1}{\alpha + 1}+m\left\vert f^{\prime \prime } ( b)\right\vert ^{q}\left ( \frac{\alpha } { \alpha + 1}\right ) \right ) ^{\frac{1}{q}}\end{aligned}\]]and since @xmath75 we have@xmath76where @xmath68 [ teo 2.3 ] let @xmath51\rightarrow % tcimacro{\u{211d } } % % beginexpansion \mathbb{r } % endexpansion $ ] be a differentiable mapping on @xmath52 such that @xmath26 $ ] where @xmath53 @xmath54 with @xmath55 @xmath56 if @xmath57 is @xmath6 convex on @xmath58 $ ] for @xmath59^{2},$ ] @xmath66 @xmath84 then the following inequality holds:@xmath85 from lemma [ lemma 2.1 ] and using the well known hlder s integral inequality we have@xmath86 dt\right ) ^{% \frac{1}{q } } \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\left ( \frac{1}{p+1}\right ) ^{\frac{1}{p}% } \\ & & \times \left ( \left\vert f^{\prime \prime } ( a)\right\vert ^{q}\beta \left ( \alpha + 1,q+1\right ) + m\left\vert f^{\prime \prime } ( b)\right\vert ^{q}\left ( \frac{1}{q+1}-\beta \left ( \alpha + 1,q+1\right ) \right ) \right ) ^{% \frac{1}{q}}.\end{aligned}\]]since @xmath87 we obtain@xmath88 [ teo 2.4 ] let @xmath51\rightarrow % tcimacro{\u{211d } } % % beginexpansion \mathbb{r } % endexpansion $ ] be a differentiable mapping on @xmath52 such that @xmath26 $ ] where @xmath53 @xmath54 with @xmath55 @xmath56 if @xmath57 is @xmath6 convex on @xmath58 $ ] for @xmath59^{2},$ ] and for some fixed @xmath60 then the following inequality holds:@xmath89 from lemma [ lemma 2.1 ] and using the well - known power - mean integral inequality we have@xmath90 dt\right ) ^{% \frac{1}{q } } \\ & = & \frac{\left ( mb - a\right ) ^{2}}{2}\left ( \frac{1}{2}\right ) ^{1-\frac{1}{q}% } \\ & & \times \left ( \left ( \left\vert f^{\prime \prime } ( a)\right\vert ^{q}\beta \left ( \alpha + 2,q+1\right ) + m\left\vert f^{\prime \prime } ( b)\right\vert ^{q}\left
( \frac{1}{\left ( q+1\right ) \left ( q+2\right ) } -\beta \left ( \alpha + 2,q+1\right ) \right ) \right ) \right ) ^{\frac{1}{q}}\end{aligned}\]]which completes the proof .
bakula , m.e .
zdemir and j. peari , hadamard type inequalities for @xmath14convex and @xmath0convex functions , j. inequal . pure and appl . math .
, 9(2008 ) , article 96 .
[ online : http://jipam.vu.edu.au ] m. klarii bakula , j. peari and m. ribii , companion inequalities to jensen s inequality for @xmath14convex and @xmath91convex functions , j. inequal .
pure and appl .
, 7(2006 ) , article 194 .
[ online : http://jipam.vu.edu.au ] | in this paper , we establish some integral inequalities for functions whose second derivatives in absolute value are @xmath0 convex
. corresponding author ] |
reaction - diffusion equations with spatially distributed hysteresis were first introduced in @xcite to describe the growth of a colony of bacteria ( salmonella typhimurium ) and explain emerging spatial patterns of the bacteria density . in @xcite , numerical analysis of the problem
was carried out , however without rigorous justification .
first analytical results were obtained in @xcite ( see also @xcite ) , where existence of solutions for multi - valued hysteresis was proved .
formal asymptotic expansions of solutions were recently obtained in a special case in @xcite .
questions about the uniqueness of solutions and their continuous dependence on initial data as well as a thorough analysis of pattern formation remained open . in this paper , we formulate sufficient conditions that guarantee existence , uniqueness , and continuous dependence of solutions on initial data for systems of reaction - diffusion equations with discontinuous spatially distributed hysteresis .
analogous conditions for scalar equations have been considered by the authors in @xcite .
denote @xmath0 , where @xmath1 .
let @xmath2 and @xmath3 ( @xmath4 ) be closed sets .
we assume throughout that @xmath5 , @xmath6 , @xmath7 .
we consider the system of reaction - diffusion equations @xmath8 with the initial and boundary conditions @xmath9 here @xmath10 is a positive - definite diagonal matrix ; @xmath11 is a hysteresis operator which maps an initial configuration function @xmath12 ( @xmath13 ) and an input function @xmath14 to an output function @xmath15 . as a function of @xmath16
, @xmath17 takes values in a set @xmath18 ( @xmath19 ) .
now we shall define this operator in detail .
let @xmath20 be two disjoint smooth manifolds of codimension one without boundary ( hysteresis `` thresholds '' ) . for simplicity
, we assume that they are given by @xmath21 and @xmath22 with @xmath23 and @xmath24 , respectively , where @xmath25 and @xmath26 are @xmath27-smooth functions ( in the general situation , atlases can be used ) . denote @xmath28 , @xmath29 , @xmath30 .
assume that @xmath31 and @xmath32 ( fig .
[ fig1 ] ) .
next , we introduce locally hlder continuous functions ( _ hysteresis `` branches '' _ ) @xmath33 we fix @xmath1 and denote by @xmath34 the space of functions continuous on the right in @xmath35 . for any @xmath36 ( _ initial configuration _ ) and @xmath37;\mathcal u)$ ] ( _ input _ ) , we introduce the _ configuration _ function @xmath38;\mathcal u)\to c_r[0,t),\quad \zeta(t)=\zeta(\zeta_0,u_0)(t)\ ] ] as follows .
let @xmath39:u_0(s)\in\gamma_\alpha\cup \gamma_\beta\}$ ] .
then @xmath40 if @xmath41 , @xmath42 if @xmath43 , @xmath44 if @xmath45 ; for @xmath46 $ ] , @xmath47 if @xmath48 , @xmath49 if @xmath50 and @xmath51 , @xmath52 if @xmath50 and @xmath53 ( fig . [ fig1 ] )
. , width=226 ] now we introduce the _ hysteresis operator _
@xmath54;\mathcal u)\to c_r[0,t ) $ ] by the following rule ( cf .
@xcite ) . for any initial configuration @xmath36 and input @xmath55;\mathcal u)$
] , the function @xmath56\to\mathcal w$ ] ( _ output _ ) is given by @xmath57 where @xmath58 is the configuration function defined above .
assume that the initial configuration and the input function depend on spatial variable @xmath59 .
denote them by @xmath12 and @xmath60 , where @xmath61;\mathcal u)$ ] . using and treating @xmath59 as a parameter
, we define the _ spatially distributed hysteresis _
@xmath62 where @xmath63 is the _ spatial configuration_. denote by @xmath64 , @xmath65 , the standard lebesgue space and by @xmath66 with natural @xmath67 the standard sobolev space . for a noninteger @xmath68 ,
denote by @xmath66 the sobolev space with the norm @xmath69}(0,1)}+\left(\int_0 ^ 1dx\int_0 ^ 1 \dfrac{|{{\varphi}}^{([l])}(x)-{{\varphi}}^{([l])}(y)|^q}{|x - y|^{1+q(l-[l])}}dy\right)^{1/q},\ ] ] where @xmath70 $ ] is the integer part of @xmath67 .
introduce the anisotropic sobolev spaces @xmath71 with the norm @xmath72 and the space @xmath73 of @xmath74-valued functions continuously differentiable on @xmath75 $ ] with the norm @xmath76 denote by @xmath77 , @xmath78 , the hlder space . for the vector - valued functions
, we use the following notation .
if , e.g. , @xmath6 and each component of @xmath79 belongs to @xmath80 , then we write @xmath81 . throughout
, we fix @xmath82 and @xmath83 such that @xmath84 and @xmath85 this implies that @xmath86 for @xmath87 ( see lemma 3.3 in ( * ? ? ?
2 ) ) . to define the space of initial data , we use the fact that
if @xmath88 , then the trace @xmath89 is well defined and belongs to @xmath90 ( see lemma 2.4 in ( * ? ? ?
we denote the norm in the latter space by @xmath91 .
moreover , one can define the space @xmath92 as the subspace of functions from @xmath93 with the zero neumann boundary conditions .
we assume that @xmath94 and @xmath95 in .
[ defsolution ] a pair @xmath96 is a _ solution of problem , _
if @xmath17 is measurable with respect to @xmath16 and , hold .
we will deal with the case where @xmath12 has one discontinuity point .
generalization to finitely many discontinuity points is straightforward .
[ conda1 ] 1 . for some @xmath97 , we have @xmath98 2 . for @xmath99 $ ] , we have @xmath100 or , equivalently , @xmath101 .
3 . for @xmath102 $ ] , we have @xmath103 or , equivalently , @xmath104 .
4 . if @xmath105 , then @xmath106
. it follows from condition [ conda1 ] that the hysteresis in at the initial moment equals @xmath107 for @xmath108 and @xmath109 for @xmath110 .
items 2 and 3 in condition [ conda1 ] are necessary for the hysteresis to be well - defined at the initial moment , while item 4 is an essential assumption .
we refer to item 4 as the _ spatial transversality _ and say that @xmath111 _ is transverse with respect to the spatial configuration @xmath12_. this means that if @xmath112 , then the vector @xmath113 is transverse to the manifold @xmath114 at this point .
consider time - dependent functions @xmath60 such that @xmath115 .
we say that a function @xmath79 is _ transverse _ _ on _ @xmath75 $ ] @xmath116with respect to a spatial configuration @xmath117 if , for every fixed @xmath118 $ ] , either @xmath119 has no discontinuity points for @xmath120 , or it has one discontinuity point and the function @xmath121 is transverse with respect to the spatial configuration @xmath119 .
a function @xmath79 _ preserves spatial topology @xmath116of a spatial configuration @xmath117 on @xmath75 $ ] _ if there is @xmath122 such that , for @xmath118 $ ] , there is a continuous function @xmath123 such that @xmath124 for @xmath125 and @xmath126 for @xmath127 .
the solution from definition [ defsolution ] is called _ transverse _ ( _ preserving spatial topology _ ) if the function @xmath60 is transverse ( preserving spatial topology ) .
the function @xmath128 defining discontinuity of @xmath129 plays a role of free boundary , which has much in common with free boundary in parabolic obstacle problems ( see , e.g. , @xcite and the references therein ) .
however , in our case , the behavior of @xmath128 is defined differently .
first , we assume the following .
[ condregularity ] the functions @xmath130 and @xmath131 are locally lipschitz continuous in @xmath132 .
next , we formulate dissipativity conditions for @xmath133 and @xmath134 . in the following condition ,
we denote by @xmath135 , @xmath136 , closed parallelepipeds in @xmath137 with the edges parallel to respective coordinate axes such that @xmath138 for all @xmath139 $ ] .
[ conddissipativity - f ] there is a parallelepiped @xmath140 and , for each sufficiently small @xmath141 , there is a parallelepiped @xmath142 and a locally lipschitz continuous function @xmath143 such that 1 .
@xmath144 converges to @xmath145 uniformly on compact sets in @xmath146 as @xmath147 , 2 . at each point @xmath148 , @xmath149 , the vector
@xmath150 points strictly inside @xmath140 .
3 . at each point
@xmath151 , @xmath149 , the vector @xmath152 points strictly inside @xmath135 for all @xmath153 . to formulate the assumption on @xmath134 , we fix @xmath154 satisfying condition [ conddissipativity - f ] and set @xmath155 [ conddissipativity - g ] for any @xmath156 , there is a compact @xmath157 such that @xmath158 @xmath159 and the cauchy problem @xmath160 has a solution @xmath161 satisfying @xmath162 whenever @xmath163 it follows from ( * ? ? ?
* theorem 1 , p. 111 ) that system has a unique solution @xmath161 for a sufficiently small @xmath156 .
condition [ conddissipativity - g ] additionally guarantees the absence of blow - up . in particular
, the uniform boundedness of @xmath164 holds if @xmath165 , where @xmath166 and @xmath167 are bounded on compact sets ( see example [ exbacteria ] ) .
however , if @xmath168 , one must additionally check that @xmath164 never leaves @xmath169 . to fulfill condition [ conddissipativity - g ] , one could alternatively assume the existence of invariant parallelepiped for @xmath134 ( similarly to condition [ conddissipativity - f ] ) .
[ exbacteria ] the hysteresis operator and the right - hand side in the present paper apply to a model describing a growth of a colony of bacteria ( salmonella typhimurium ) on a petri plate ( see @xcite ) .
let @xmath170 and @xmath171 denote the concentrations of diffusing buffer ( ph level ) and histidine ( nutrient ) , respectively , while @xmath172 denote the density of nondiffusing bacteria .
these three unknown functions satisfy the following equations : @xmath173 supplemented by initial and no - flux ( neumann ) boundary conditions . in , @xmath174 are given constants and @xmath17 is the hysteresis operator . in this example
, we have @xmath175 , @xmath176 , @xmath177 .
the hysteresis thresholds @xmath114 and @xmath178 are the curves on the plane given by @xmath179 and @xmath180 , respectively , where @xmath181 are some constants ( fig .
[ fig1 ] ) ; the hysteresis `` branches '' are given by functions @xmath182 @xmath183 and @xmath184 @xmath185 .
in what follows , we assume that conditions @xmath186@xmath187 hold .
[ thlocalexistence ] there is a number @xmath1 such that 1 .
there is at least one solution of problem , in @xmath188 2 .
any solution in @xmath189 is transverse and preserves spatial topology .
[ tcont ] let @xmath190 be a transverse topology preserving solution of problem , in @xmath189 for some @xmath1 .
then it can be continued to an interval @xmath191 , where @xmath192 has the following properties .
1 . for any @xmath193 , the pair @xmath194 is a transverse solution of problem , in @xmath195 .
either @xmath196 , or @xmath197 and @xmath190 is a solution in @xmath198 , but @xmath199 is not transverse with respect to @xmath200 .
[ tcontdepinitdata ] assume the following . 1 .
there is a number @xmath1 such that problem , with initial functions @xmath201 and initial configuration @xmath12 defined by its discontinuity point @xmath202 admits a unique transverse topology preserving solution @xmath190 in @xmath203 for any @xmath204 .
2 . let @xmath205 , @xmath206 , @xmath207 , be a sequence of other initial functions such that @xmath208 , @xmath209 as @xmath210 .
3 . let @xmath211 , @xmath207 , be a sequence of other initial configurations defined by their discontinuity points @xmath212 similarly to and @xmath213 as @xmath210 .
then , for all sufficiently large @xmath214 , problem , with initial data @xmath215 has at least one transverse topology preserving solution @xmath216 .
each sequence of such solutions satisfies @xmath217}\to 0,\\ \sup\limits_{t\in[0,t]}\left(\|v_n(\cdot , t)- v(\cdot , t)\|_{l_q((0,1);{{\mathcal v}})}+\|v_{nt}(\cdot , t)- v_t(\cdot , t)\|_{l_q((0,1);{{\mathcal v}})}\right)\to 0 \end{gathered}\ ] ] as @xmath218 where @xmath128 and @xmath219 are the respective discontinuity points of the configuration functions @xmath129 and @xmath220 .
if one a priori knows that all @xmath221 are transverse on some interval @xmath75\subset[0,t_{max})$ ] , then one can prove that @xmath221 approximate @xmath79 on @xmath75 $ ] even if @xmath79 is not topology preserving on @xmath75 $ ] .
now we discuss the uniqueness of solutions .
we strengthen the assumption about local hlder continuity of @xmath222 .
let @xmath140 be the set from condition [ conddissipativity - f ] .
[ conduniq ] there are numbers @xmath223 and @xmath224 such that @xmath225 we refer readers to @xcite for the discussion about functions satisfying this condition .
[ tuniqueness ] assume additionally that condition @xmath226 holds .
let @xmath190 and @xmath227 be two transverse solutions of problem , in @xmath189 for some @xmath1
. then @xmath228 .
in this section , we prove theorems [ thlocalexistence][tcontdepinitdata ] . throughout the section , we fix @xmath154 satisfying condition [ conddissipativity - f ] and @xmath229 given by .
next , we fix some @xmath230 $ ] and then @xmath231 satisfying condition [ conddissipativity - g ] .
the following result is straightforward . [ lab ] 1 .
let @xmath232 , @xmath233 $ ] , and @xmath234 } a(s)$ ] .
then @xmath235 $ ] and @xmath236}\le\|a\|_{c^\lambda[0,t]}. $ ] 2 .
if @xmath237 $ ] and @xmath238 } a_j(s)$ ] , @xmath239 , then @xmath240}\le \|a_1-a_2\|_{c[0,t]}. $ ] for some @xmath241 , @xmath242 , and @xmath243 $ ] such that @xmath244 ( @xmath245 ) and @xmath246 ( @xmath118 $ ] ) , we define the function @xmath247 by @xmath248 here we assume @xmath249 to be extended to @xmath154 without loss of regularity . [ lu0b0v0 ] let @xmath250 be functions with the above properties , and let @xmath251 be functions with the same properties .
let @xmath252 be defined by and @xmath253 by with @xmath254 and @xmath255 instead of @xmath256 and @xmath257 , respectively .
then , for any @xmath258 , @xmath259 where @xmath260 is a hlder exponent for @xmath261 and @xmath262 depends on @xmath154 and @xmath263 , but does not depend on @xmath264 .
we fix some @xmath118 $ ] and assume that @xmath265 for this @xmath266 .
then , using and omitting the arguments of the integrands , we have @xmath267 using the hlder continuity and the boundedness of @xmath268 for @xmath269 and integrating with respect to @xmath266 from @xmath145 to @xmath270 , we complete the proof .
now we introduce sets that `` measure '' the spatial transversality .
denote by @xmath271 , @xmath19 , the set of triples @xmath272 such that @xmath273 , @xmath274 , @xmath12 is of the form , and the following hold : 1 . @xmath275 $ ] , 2 .
@xmath276 for @xmath277 $ ] , 3 .
@xmath278 for @xmath279 $ ] , 4 .
if @xmath280 $ ] and @xmath281 $ ] , then @xmath282 , 5 .
@xmath283 and @xmath284 .
it is easy to check that @xmath285 .
moreover , one can show ( lemma 2.25 in @xcite ) that the union of all sets @xmath271 coincides with the set of all data satisfying condition @xmath186 . from now on ,
we fix @xmath286 such that @xmath287 .
the next lemma follows from the implicit function theorem and lemma [ lab ] .
[ loneroot ] let @xmath288 , @xmath289 , @xmath290 for some @xmath291 , @xmath292 , and @xmath293 .
then there is @xmath294 and a natural number @xmath295 which do not depend on @xmath296 such that the following is true for any @xmath297 $ ] . 1
. the equation @xmath298 for @xmath299 $ ] has no more than one root .
if this root exists , we denote it by @xmath300 otherwise , we set @xmath301 .
one has @xmath302 $ ] , @xmath303 $ ] .
2 . the hysteresis @xmath304 and its configuration function @xmath305 have exactly one discontinuity point @xmath306 moreover , @xmath307}a_1(s)$ ] , @xmath308 $ ] . consider functions @xmath242 and @xmath309 such that @xmath310 for some @xmath1 .
define the functions @xmath311 consider the auxiliary problem @xmath312 set @xmath313 and @xmath314 , where @xmath315 .
the next result follows from the standard estimates for solutions of linear parabolic equations @xcite , from conditions [ condregularity][conddissipativity - g ] combined with the principle of invariant rectangles @xcite , and from lemma [ loneroot ] .
[ ltnp0 ] 1 . for any @xmath241
, problem has a unique solution @xmath316 and @xmath317}\|u(\cdot , t)\|_{q } \le c_1(\|{{\varphi}}\|_{q}+f_u),\\ \notag \|v\|_{w_\infty^{0,1}(q_t;{{\mathcal v}})}\le \|\psi\|_{l_\infty((0,1);{{\mathcal v}})}+2g_u,\\ \label{eqnp02 * } \|u\|_{c^\gamma({{\overline q}}_t;{{\mathcal u}})}+\|u_x\|_{c^\gamma({{\overline q}}_t;{{\mathcal u } } ) } \le c_2 ( \|{{\varphi}}\|_{q}+f_u),\end{gathered}\ ] ] where @xmath318 depend only on @xmath319 .
2 . if @xmath320 , @xmath207 , are solutions of problem , with @xmath321 replaced by @xmath322 @xmath116with the same properties@xmath323 and @xmath324 then @xmath325 3 . there is @xmath326 and a natural number @xmath327 such that , for any @xmath328 $ ] , conclusions ( 1 ) and ( 2 ) from lemma @xmath329 hold for @xmath60 , for the corresponding `` root '' function @xmath330 , for the configuration function @xmath129 of the hysteresis @xmath331 , for its discontinuity point @xmath128 , and for @xmath332 instead of @xmath333 .
furthermore , @xmath334 .
let us prove the first assertion . 1.1 .
fix @xmath336 in lemma [ loneroot ] such that @xmath337 .
fix @xmath338 from lemma [ ltnp0 ] .
set @xmath339 , where @xmath340 is the embedding constant such that @xmath341 . set @xmath342 , where @xmath343 are defined in lemmas [ loneroot ] , [ ltnp0 ]
let @xmath344 be the set of functions @xmath60 such that @xmath345 , @xmath346 the set @xmath344 is a closed convex subset of the banach space endowed with the norm given by the left - hand side in .
similarly , we define @xmath347
we construct a map @xmath348 .
take any @xmath349 and define @xmath350 and @xmath351 according to lemma [ loneroot ] . then define @xmath247 by and , using this @xmath252 , define @xmath352 by .
finally apply lemma [ ltnp0 ] and obtain a solution @xmath190 of auxiliary problem .
we now define @xmath353 .
the operator @xmath354 is continuous .
indeed , it is not difficult to check that the mapping @xmath355 is continuous from @xmath344 to @xmath356 $ ] .
thus , the continuity of @xmath354 follows by consecutively applying lemmas [ lab ] ( part 2 ) , [ lu0b0v0 ] , [ ltnp0 ] ( part 2 ) , and the continuity of the embedding @xmath357 . furthermore , due to and the choice of @xmath358 , the operator @xmath354 maps @xmath344 into itself . as an operator acting from @xmath344 into itself , it is compact due to and the compactness of the embedding @xmath359 .
therefore , applying the schauder fixed - point , we conclude the proof of the first assertion of the theorem .
note that @xmath354 is not lipschitz continuous ( mind the exponent @xmath360 in ) .
hence , the contraction principle does not apply .
we prove uniqueness separately in sec .
[ secuniqueness ] .
the second assertion follows by applying the principle of invariant rectangles ( see @xcite ) and lemma [ ltnp0 ] .
theorem [ tcont ] follows from part 3 of lemma [ ltnp0 ] and from the following fact ( see lemma 2.25 in @xcite ) .
assume ( 1 ) @xmath362 , @xmath363 ; ( 2 ) @xmath364 and @xmath365 as @xmath366 for some @xmath367 and @xmath368 ; ( 3 ) @xmath369 as @xmath366 for some @xmath370 $ ] .
then @xmath371 or @xmath111 is not transverse with respect to @xmath12 , where @xmath12 is given by .
\1 . it suffices to prove the theorem for a sufficiently small time interval . since @xmath287 , it is easy to show that there is @xmath373 such that @xmath374 for all @xmath375 .
hence , by theorem [ thlocalexistence ] , there is @xmath376 $ ] for which problem , has transverse topology preserving solutions @xmath190 and @xmath216 with the corresponding initial data .
moreover , any solution of problem , in @xmath377 is transverse and preserves topology . we introduce the functions @xmath330 and
@xmath378 corresponding to @xmath79 and @xmath221 as described in part 3 of lemma [ ltnp0 ] .
then the discontinuity points of the corresponding configuration functions @xmath129 , @xmath220 are given by @xmath234}a(s)$ ] and @xmath379}a_n(s)$ ] .
assume that there is @xmath380 such that @xmath381 for some subsequence of @xmath221 , which we denote @xmath221 again .
theorem [ thlocalexistence ] implies that @xmath221 and @xmath382 are uniformly bounded in @xmath383 and @xmath384 $ ] , respectively .
hence , we can choose subsequences of @xmath221 and @xmath382 ( which we denote @xmath221 and @xmath382 again ) such that @xmath385}\to 0,\quad n\to\infty \label{eq1.1 - 9}\end{gathered}\ ] ] for some function @xmath386 with @xmath387 and some @xmath388 $ ]
. set @xmath389}\hat a(s)$ ] .
due to , , and lemma [ lab ] , we have @xmath390}\to 0,\quad n\to\infty , \\ \label{eqcontdepw } w(\xi_0(x),\hat u(x,\cdot))(t)=\begin{cases } w_1(\hat u(x , t ) ) , & 0\le x\le \hat b(t),\\ w_2(\hat u(x , t ) ) , & \hat b(t)<x\le 1 . \end{cases}\end{gathered}\ ] ] \3 .
now we show that @xmath391}\left(\|v_n(\cdot , t)-\hat v(\cdot , t)\|_{l_q((0,1);{{\mathcal v}})}+\|v_{nt}(\cdot , t)-\hat v_t(\cdot , t)\|_{l_q((0,1);{{\mathcal v}})}\right)\to 0,\quad n\to\infty,\ ] ] for some @xmath392 .
take an arbitrary @xmath393 .
it follows from the assumptions of the theorem , from , , , and from lemma [ lu0b0v0 ] that @xmath394 provided @xmath395 are large enough .
estimates , the second equation in , and the local lipschitz continuity of @xmath134 yield @xmath396 where @xmath397 does not depend on @xmath395 .
hence , by gronwall s inequality , @xmath398 where @xmath399 does not depend @xmath400 and @xmath118 $ ] .
a similar inequality for the time derivative of @xmath401 follows from , , and from the second equation in . since @xmath393 is arbitrary ,
does hold .
now consider , with the subsequence @xmath402 and pass to the limit as @xmath210 . due to the uniqueness assumption , @xmath403 .
therefore , is not true and we have the convergence for the whole sequence @xmath216 .
in this section , we prove theorem [ tuniqueness ] . for the clarity of exposition , we restrict ourselves to the case where initial data satisfy the equality @xmath404 in addition to condition [ conda1 ] .
( the case @xmath405 can be treated easily because then the hysteresis @xmath406 remains constant on some time interval . )
we fix @xmath408 such that the conclusions of lemma [ loneroot ] are true for the @xmath79 , @xmath409 on @xmath410 .
let @xmath330 , @xmath128 , @xmath411 , @xmath412 be the functions defined in lemma [ loneroot ] for @xmath79 and @xmath409 , respectively .
we fix @xmath413 and @xmath393 such that the following hold for @xmath118$]@xmath414 @xmath415,\\ \label{lem4.0.5 } \gamma_{\beta}(u(x , t ) ) < 0 , \quad x\in[0,b(t)],\end{gathered}\ ] ] and the analogous inequalities hold for @xmath409 . \1 .
denote @xmath417 , @xmath418 .
the functions @xmath419 , @xmath420 satisfy the equations @xmath421 and the zero boundary and initial conditions , where @xmath422 obviously , @xmath423 .
the function @xmath419 can be represented via the green function @xmath424 of the heat equation with the neumann boundary conditions : @xmath425 let us prove this inequality for the function @xmath431 , assuming that @xmath432 .
( the cases of @xmath433 and @xmath434 are treated analogously . ) since @xmath133 is locally lipschitz , @xmath435 where @xmath436 and the constants @xmath437 below do not depend on @xmath438 $ ] .
inequality implies that @xmath441,@xmath442 on the closed set @xmath443,\ s\in[0,t]\}$ ] .
hence , the values @xmath444 and @xmath445 are separated from 0 . therefore , using condition [ conduniq ] , we obtain @xmath446 2.2 .
boundedness of @xmath447 and @xmath184 for @xmath448 and lemma [ lab ] imply @xmath449}\le k_5 \|a-\hat a\|_{c[0,t]}.\ ] ] using , we obtain for any @xmath118 $ ] the inequalities @xmath450 where @xmath451 is a respective lipschitz constant for @xmath452 and hence does not depend on @xmath453 .
inequalities and yield @xmath454 . a mathematical model for bacterial growth described by a hysteresis operator .
recent advances in nonlinear analysis proceedings of the international conference on nonlinear analysis .
hsinchu , taiwan , 2006 .
green s matrices of boundary value problems for petrovski parabolic systems of general form , ii , mat .
, * 114(156 ) * , no .
4 , 523565 ( 1981 ) ; english transl . in math .
ussr sbornik , * 42 * 461489 ( 1982 ) . : _ pavel gurevich _ , free university berlin , arnimallee 3 , berlin , 14195 , germany ; peoples friendship university , mikluho - maklaya str
. 6 , moscow , 117198 , russia ; e - mail : ` [email protected] ` .
_ sergey tikhomirov _ , chebyshev laboratory , saint - petersburg state university , 14th line of vasilievsky island , 29b , saint - petersburg , 199178 , russia ; max planck institute for mathematics in the sciences , inselstrasse 22 , 04103 , leipzig , germany ; e - mail : ` [email protected] ` . | we study systems of reaction - diffusion equations with discontinuous spatially distributed hysteresis in the right - hand side .
the input of hysteresis is given by a vector - valued function of space and time .
such systems describe hysteretic interaction of non - diffusive ( bacteria , cells , etc . ) and diffusive ( nutrient , proteins , etc . )
substances leading to formation of spatial patterns .
we provide sufficient conditions under which the problem is well posed in spite of the discontinuity of hysteresis .
these conditions are formulated in terms of geometry of manifolds defining hysteresis thresholds and the graph of initial data . |
since the discovery of the first hot jupiter system @xcite and the subsequent realization that such systems are ubiquitous @xcite , astronomers have struggled to observe these planets and to compare these observations to theoretical predictions about their surface temperature variations , composition , albedo , and internal structure .
the observation of extrasolar planetary transits , starting with hd 209458b @xcite , have allowed for the determination of the mass and size of these planets , but it was only with the observation of secondary eclipses , starting with hd 209458b @xcite and tres-1 @xcite , that it has been possible to measure the flux from these planets ( for a review of transiting exoplanet science see * ? ? ?
the depth of the secondary eclipse , however , only characterizes the day - side flux from the planet . to fully characterize the planet s longitudinal temperature profile ,
observations must be made at a variety of planetary phases .
the first such observations were reported by @xcite and @xcite . in this paper
we present similar observations of three other planetary systems .
we briefly review models of hot jupiter phase variations in 1.2 and 1.3 , and the observational constraints in 1.4 . in
2 we present our observations and in 3 we explain how they constrain the properties of hot jupiters .
we state our conclusions in 4 .
for the purposes of this paper , the phase function of a planetary system is taken to be the relative flux contribution from the planet as a function of its orbital position . here
we describe a toy model of hot jupiter phase functions , based on a planet in a circular orbit with two isothermal hemispheres and neglecting the planet s remnant heat of formation , an effective temperature of @xmath14 5075 k @xcite .
the bond albedo , @xmath15 , is the fraction of the incident stellar flux that is reflected by the planet .
a planet s energy budget is largely determined by its bond albedo .
hence , if we treat the host star as a blackbody , the equilibrium temperature of a planet s day and night faces is , respectively @xmath16 and @xmath17 where @xmath18 and @xmath19 are the star s radius and effective temperature , @xmath20 is the planet s semi - major axis and @xmath21 quantifies the portion of the absorbed stellar energy advected to the planet s night - side ( @xmath22 for no redistribution ; @xmath23 for full redistribution ) @xcite .
the depth of secondary eclipse ( for edge - on systems ) is approximately the ratio of the planet s day - side flux , @xmath24 , to the stellar flux , @xmath25 . in light of the models of @xcite
, we assume that the planet s day and night - side radiate as blackbodies ( @xmath26 ) and that the host star emits with a mid - ir brightness temperature @xmath27 ( we use @xmath28 at 8 @xmath0 m based on the models of * ? ? ?
the resulting planet - star flux ratio is : @xmath29 where @xmath30 is the planetary radius and the ratios @xmath31 and @xmath32 can be well - constrained if the planet transits its host star .
the wavelength - dependent geometric albedo , @xmath33 , is the ratio of the total brightness of the planet at full phase to that of a lambertian disk with the same cross - section
. the second term ( thermal emission from the planet ) can be neglected at optical wavelengths . in the infrared , we may neglect the contribution from reflected starlight .
the peak - to - trough amplitude of the phase function for a transiting systems then becomes : @xmath34 where @xmath35 is the flux from the planet s night - side and where we have neglected the planet s flux contribution ( @xmath36% ) in the denominator .
note that the absolute flux from the planet is a minimum during the secondary eclipse , irrespective of phase variations , since the planet is entirely hidden from view at that time .
the normalized phase variations , @xmath37 , for a system with inclination @xmath38 is : @xmath39 where the @xmath40 is the planet s orbital phase ( @xmath41 at inferior conjunction ; @xmath42 at superior conjunction ) and @xmath43 is the time - averaged flux of the system .
measurements of secondary eclipse depth thus constrain a combination of @xmath15 , @xmath33 and @xmath21 in this toy model .
observations of the phase function amplitude constrain these same parameters , as well as @xmath44 and @xmath30 in the case of non - transiting systems .
in addition to the toy model outlined above , there have been numerous more sophisticated attempts to model the atmospheres of hot jupiters and thus to predict the depth of secondary eclipses as well as the full phase function of specific extrasolar systems .
we summarize the most pertinent aspects of these models here ( for a more thorough background see * ? ? ?
@xcite determine energy recirculation and predict global temperature contrasts of @xmath45500 k at the photosphere .
they also predict a noticeable offset between the phase function peak and the time of opposition .
@xcite construct models with either no or full recirculation and find that the latter is a better fit to secondary eclipse observations .
@xcite , @xcite and @xcite also run simulations with different amounts of recirculation and find that significant redistribution of heat must be occurring to explain the observations .
@xcite run models to calculate emitted spectra as a function of orbital phase for both equilibrium and non - equilibrium chemistry .
they find that non - equilibrium chemistry leads to significantly smaller phase variations and that current observations can be explained with a blackbody - like planetary spectrum .
finally , @xcite use the factor @xmath21 to characterize the efficiency of energy recirculation and construct models for five different values of the parameter .
they find that current observations rule out models with no recirculation and are in fact consistent with full recirculation ( although they admit that current observations are not precise enough to determine the exact degree of recirculation ) .
secondary eclipses have been reported for four short - period extrasolar planetary systems and a phase function has been reported in a different , non - transiting system .
we summarize these results below .
@xcite placed an upper limit of @xmath46 on the relative secondary eclipse depth at 2.2 @xmath0 m using the spex instrument at the nasa infrared telescope facility . the secondary eclipse of hd 209458 , observed with the multiband imaging photometer for spitzer ( mips ) yielded a relative depth at 24 @xmath0 m of @xmath47 @xcite .
similarly , the secondary eclipse of tres-1 was observed in the irac 4.5 and 8 @xmath0 m bands , yielding eclipse depths of @xmath48 and @xmath49 , respectively @xcite .
the latest system to have been detected in secondary eclipse is hd 189733 @xcite .
observations with the spitzer infrared spectrograph ( irs ) peak - up imager at 16 @xmath0 m give an eclipse depth of @xmath50 .
in addition to infrared detections , the non - detection of a secondary eclipse for hd 209458 at optical wavebands has constrained the 500 nm secondary eclipse depth to be less than @xmath51 at the @xmath6 level and less than @xmath52 at the @xmath53 level @xcite .
ground - based k - band ( 2.2 @xmath0 m ) observations of ogle - tr-113b tentatively show a secondary eclipse depth of @xmath54 @xcite .
mips 24 @xmath0 m imaging of @xmath55 andromeda has yielded a phase function with peak - to - trough amplitude of @xmath56 and weak evidence of a small phase offset @xcite .
this surprisingly large amplitude suggests that the planet has parameters @xmath57 , @xmath58 and @xmath59 .
continuous irac 8 @xmath0 m monitoring of hd 189733 for half of a planetary orbit led to the detection of a phase function with a peak - to - trough amplitude of 0.00121(4 ) @xcite .
this modest phase function , roughly one third the relative depth of secondary eclipse , indicates that much of the incident stellar energy is being recirculated to the planet s night side .
@xcite used irs to obtain a mid - ir spectrum for hd 189733b , which showed no evidence for absorption bands due to either water or methane . @xcite
obtained the relative mid - ir spectrum of hd 209458b and similarly concluded that such absorption features could be excluded ; they also identified candidate emission features at two wavelengths .
m observation of science target , 51 peg , and flux standard , hd 217636 , showing the flux ramp - up .
this observation began at mjd = 53728.447 , or @xmath60 .
each point represents 64 exposures , each of 0.1 sec .
the error bars , which have an average value of @xmath61 , correspond to the standard deviation in the counts from the 64 images . for comparison
, one would expect uncertainties of roughly @xmath62 in the poisson noise regime.,width=317 ]
we acquired spitzer photometry with the irac 3.6 @xmath0 m or 4.5 @xmath0 m , and 8 @xmath0 m detectors in sub - array mode ( 32@xmath6332 pixels , or [email protected] arcseconds ) to avoid saturation @xcite .
observations were made at eight different phases for each planetary system with follow - on time constraints .
each observation consists of four nods to the target star interleaved with four nods to a nearby flux standard of comparable infrared brightness .
the number and duration of exposures for each of our targets are summarized in table [ results ] .
we convert fluxes from the spitzer units of specific intensity ( mjy / sr ) to photon counts , and use aperture photometry to recover the counts for the science and standard stars .
we determine the psf centroid using a centre - of - light function , then use the idl routine aper to extract the source flux .
after testing many combinations of aperture and sky annulus radii for robustness , we settle on a 7 pixel aperture radius and a sky annulus centred on the target star , with an inner radius of 10 pixels and an outer radius of 15 pixels .
a single observation lasts only 20 minutes , compared to planetary orbital periods of days , so the flux is not expected to vary noticeably within a single observation .
unfortunately , the measured signal in the 3.6 @xmath0 m and 4.5 @xmath0 m bands exhibits large ( few % ) fluctuations .
the pointing drifts between nods are sub - pixel yet other researchers have found that the irac 4.5 @xmath0 m insb array shows significant intra - pixel sensitivity variations @xcite .
the 3.6 @xmath0 m detector , also an insb array , likely suffers from the same problem .
we do not have sufficient images to properly map out the position - sensitivity relation , and this severely limits the quality of our 3.6 @xmath0 m and 4.5 @xmath0 m data .
the 8 @xmath0 m observations of 51 peg and hd 179949 show a smooth increase in gain , with a similar shape for both the science target and flux standard , indicating an instrumental effect , as shown in figure [ 51peg_ch4_observation_2 ] .
hd 209458 , a somewhat fainter source , does not exhibit such a marked ramp - up .
the illumination - dependent nature of this instrumental effect , described in greater detail in @xcite , is consistent with charge trapping , in that pixel sensitivities are initially deficient but asymptote to a constant value as the pixels are illuminated repeatedly . for each observation , we determine the total counts by multiplying the median count by the number of frames , thus ignoring the minority of frames affected by cosmic rays .
we repeat this median averaging for all the exposures within each nod ( with duration @xmath64 ) and take the standard deviation in the counts from each exposure multiplied by the square - root of the number of exposures per nod to be the uncertainty in the total counts .
we find these uncertainties to be 45 times larger than the expectation from poisson noise at 3.6 @xmath0 m and 4.5 @xmath0 m , and roughly twice poisson noise at 8 @xmath0 m . at the shorter wavebands , as well as for the 8 @xmath0 m observations of hd 209458 ,
the points from different nods show considerable scatter but no obvious trend , so we add up the counts from all the nods separately for the science and reference star , then take the ratio of the two sums to determine the relative flux at that epoch . noticing that the ramp - up at 8 @xmath0 m for 51 peg and hd 179949 levels off towards the end of any given observation
, we take the ratio of the summed counts for the two final nods as the relative flux at that epoch . in all cases
we compute the uncertainty in the flux ratio by taking the quadratic sum of the relative uncertainties in the science and standard counts .
the average flux measured for each star can be found in table [ results ] . [ cols="<,^,^,^,^,^,^",options="header " , ] the wavelength of the observations is @xmath65 , @xmath66 is the duration of a single exposure , @xmath67 is the number of exposures per nod , @xmath64 is the total time on target per nod , @xmath68 is the average flux of the system and the associated error , while @xmath69 is the @xmath1 upper limit , or the best - fit peak - to - trough amplitude and associated @xmath6 uncertainty for phase variations , as appropriate .
m light curves for the science target , 51 peg , and the flux standard , hd 217636 .
the top panel shows the _ relative _ light curve ( the ratio of the two lower plots ) .
the phase of the planet ( @xmath40 from 1.2 ) is 0 at inferior conjunction , and @xmath70 at superior conjunction .
the @xmath6 error bars have an average size of @xmath71 and represent the standard deviation in the mean of the binned counts .
the curve is the best - fit sinusoid ( with fixed period and no phase offset ) , as determined by a @xmath72-step markov chain monte carlo.,width=317 ] m light curves for the science target , hd 179949 , and the flux standard sao 187890 .
the top panel shows the _ relative _ light curve at for hd 179949 .
the phase of the planet ( @xmath40 from 1.2 ) is 0 at inferior conjunction , and @xmath70 at superior conjunction .
the @xmath6 error bars have an average size of @xmath73 and represent the standard deviation in the mean of the binned counts .
the curve is the best - fit sinusoid ( with fixed period and no phase offset ) , as determined by a @xmath72-step markov chain monte carlo.,width=317 ] m light curves for the science target , hd 209458 , and the flux standard star , hd 210483 .
the top panel shows the _ relative _ light curve .
the phase of the planet ( @xmath40 from 1.2 ) is 0 at inferior conjunction , and @xmath70 at superior conjunction .
the greyed - out areas correspond to a transit or secondary eclipse .
the @xmath6 error bars have an average size of @xmath74 and represent the standard deviation in the mean of the binned counts .
the curve is the best - fit sinusoid ( with fixed period and no phase offset ) , as determined by a @xmath72-step markov chain monte carlo.,width=317 ]
we use ephemerides from @xcite , for which projected errors at the epochs of our observations are less than 60 min .
for all three systems .
we neglect the correction for the light travel time across the solar system , since the size of this correction ( less than 8 minutes ) is less than the uncertainty in the orbital phase . by matching the planet phases with the relative flux at that epoch
, we construct a relative light curve for each system , shown in the top panel of figures [ 51peg_ch4_light_curve][hd209458_ch4_light_curve ] ( the 3.6 @xmath0 m and 4.5 @xmath0 m light curves are not plotted due to prohibitively large systematic uncertainties ) .
hd 179949 and 51 peg are non - transiting systems , while the observations of hd 209458 were taken outside of transit or secondary eclipse ( represented by shaded areas in figure [ hd209458_ch4_light_curve ] ) .
we use a @xmath75-step markov chain monte carlo with metropolis - hastings algorithm @xcite to determine the best - fit sinusoid , which is shown as the curve in the top panel of figures [ 51peg_ch4_light_curve][hd209458_ch4_light_curve ] .
the amplitude and constant offset of the sinusoid are allowed to vary , while the period is fixed at @xmath76 and the phase offset is set to zero , in accordance with 1.2 , as well as the results of @xcite .
none of the 3.6 @xmath0 m and 4.5 @xmath0 m light curves , or indeed the 8 @xmath0 m light curve for 51 peg or hd 209458 show a variation that is better fit by a sinusoid than a straight line .
we do , however , detect an 8 @xmath0 m phase function with peak - to - trough amplitude of @xmath3 for hd 179949 , compared to an amplitude of @xmath77 we would expect for an edge - on planet with a radius of @xmath78 with an albedo of @xmath79 and a recirculation parameter of @xmath22 .
for the two systems where we do not detect a phase function , the markov chain produces the cumulative distribution functions shown in figures [ 51peg_ch4_exclusion_plot ] and [ hd209458_ch4_exclusion_plot ] .
these plots show the full range of theoretically allowed peak - to - trough phase function amplitudes on the x - axis , while the y - axis shows the probability that the system has an amplitude _ less than _ a given amplitude .
the 1 , 2 and 3@xmath80 upper limits are marked for convenience .
the @xmath1 upper - limits on phase variations at 8 @xmath0 m are @xmath10 for 51 peg ( compared to an amplitude of @xmath81 for an edge - on planet with @xmath82 with an albedo of @xmath79 and a recirculation parameter of @xmath22 ) , and @xmath11 for hd 209458 ( compared to an amplitude of @xmath83 if the planet has an albedo of @xmath79 and a recirculation parameter of @xmath22 ) .
our complete results are summarized in table [ results ] .
we recognize that @xmath55 andromeda and hd 179949 are the only two planet - bearing systems studied by @xcite to exhibit chromospheric activity in phase with the planetary orbit .
it is therefore possible that the mid - ir phase variations that have been observed in these systems are a consequence of stellar activity and not the diurnal flux contrast of the planet .
we do not believe that chromospheric activity is responsible for the phase function we observe in hd 179949 , however , as the cycle in chromospheric activity @xcite observed in this system was @xmath84 out of phase with the planetary orbit , and this pattern persisted for at least the year they observed it .
if the phase variation of hd 179949 is taken at face value , it constrains the properties of this system , as explained in section [ toy ] . constraining the planetary radius to within @xmath4 , the observed phase variation places a @xmath6 upper - limit of @xmath85 . further constraining the albedo to @xmath86 , consistent with the value of @xmath87 that @xcite measured for hd 209458b
, we obtain the exclusion plot shown in figure [ hd179949_ch4_exclusion_plot_2 ] .
our model thus places a @xmath6 limit of @xmath88 on the mass of hd 179949b . because of the edge - on configuration of hd 209458
, our upper - limit can be combined with the constraint on the planet / star radius ratio of @xcite , yielding constraints on the albedo and recirculation of the hot jupiter s atmosphere ( s.f .
[ toy ] ) .
in particular , the @xmath89 , @xmath22 configuration of our toy model is excluded at the @xmath53 level , as shown in figure [ hd209458_ch4_exclusion_plot_2 ] .
more importantly , we exclude the much more realistic atmospheric models of @xcite for @xmath90 by more than @xmath1 , since their models predict stronger phase variations than our toy model for the same @xmath21 .
the x - axis spans the full range of theoretically allowed amplitudes for this system.,width=317 ] m images .
the x - axis spans the full range of theoretically allowed amplitudes for this system.,width=317 ] , and recirculation , @xmath21 , of hd 179949b based on its 8 @xmath0 m phase function .
@xmath22 corresponds to no redistribution , while @xmath23 corresponds to full redistribution .
lower masses correspond to inclinations near @xmath91 , while greater masses correspond to more face - on configurations .
the radius of planet is assumed to be in the range @xmath4 , and its bond albedo is set to @xmath79 .
the @xmath6 and @xmath1 confidence intervals for large phase function amplitudes lie on top of each other
so the former is omitted for clarity.,width=317 ] , and recirculation , @xmath21 of hd 209458b based on its 8 @xmath0 m phase function .
@xmath22 corresponds to no redistribution , while @xmath23 corresponds to full redistribution .
solid lines correspond to upper limits in this work .
the dotted lines corresponds to constraints from the secondary eclipse measurement ( d. charbonneau , personal communication ) , while the dashed line is from the @xcite albedo upper limit .
the bottom - left corner of the plot ( the no albedo , no recirculation configuration ) is excluded at the @xmath53 level by our data , while the shaded region in the top left of the figure is within all the @xmath6 contours.,width=317 ]
we have acquired spitzer irac 3.6 @xmath0 m or 4.5 @xmath0 m , and 8 @xmath0 m images of three known hot jupiter systems .
the use of standard stars has allowed us to achieve better than 0.05% relative photometry with multi - epoch irac 8 @xmath0 m observations .
intra - pixel sensitivity variations limit our photometry at 3.6 @xmath0 m and 4.5 @xmath0 m .
we detect a 8 @xmath0 m phase function for hd 179949 and determine upper limits for the amplitude of phase variations in the two other observed systems .
we analyze these light curves using a toy model of the planets and constrain the properties of the planets .
our observations suggest that hot jupiters do not all exhibit the large mid - infrared diurnal flux contrast detected by @xcite . even within our modest sample of three planets
, it appears that hd 179949b and hd 209458b exhibit rather different degrees of recirculation . the data for hd 179949 and @xmath55 andromeda provide the most uncertain constraints , however ,
while the two transiting systems with well - constrained phase functions ( hd 209458 and hd 189733 , * ? ? ?
* ) both point toward a large value of @xmath92 , so it remains to be seen if the observations of the non - transiting planets hold up under further scrutiny .
the authors would like to acknowledge mark claire and daryl haggard for their insightful comments , and greg laughlin for his help tracking down ephemerides .
n.b.c . has been supported in this research by the natural sciences and engineering research council of canada , le fonds qubcois de la recherche sur la nature et les technologies , and the u.w .
astrobiology program .
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 a contract with nasa .
support for this work was provided by nasa through an award issued by jpl / caltech . | we present results from spitzer space telescope observations of the mid - infrared phase variations of three short - period extrasolar planetary systems : hd 209458 , hd 179949 and 51 peg .
we gathered infrared array camera ( irac ) images in multiple wavebands ( 3.6 @xmath0 m or 4.5 @xmath0 m , and 8 @xmath0 m ) at eight phases of each planet s orbit .
we find the uncertainty in relative photometry from one epoch to the next to be significantly larger than the photon counting error at 3.6 @xmath0 m and 4.5 @xmath0 m .
we are able to place @xmath1 upper limits of only @xmath2% on the phase variations at these wavelengths . at 8 @xmath0
m the epoch - to - epoch systematic uncertainty is comparable to the photon counting noise and we detect a phase function for hd 179949 which is in phase with the planet s orbit and with a relative peak - to - trough amplitude of @xmath3 . assuming that hd 179949b has a radius @xmath4 , it must recirculate less than @xmath5% of incident stellar energy to its night side at the @xmath6 level ( less than @xmath7% at the @xmath1 level , where @xmath8% signifies full recirculation ) .
if the planet has a small bond albedo , it must have a mass less than @xmath9 ( @xmath6 ) .
we do not detect phase variations for the other two systems but we do place the following @xmath1 upper limits : @xmath10 for 51 peg , and @xmath11 for hd 209458 . due to its edge - on configuration , the upper limit for hd 209458 translates , with appropriate assumptions about bond albedo , into a lower limit on the recirculation occuring in the planet s atmosphere .
hd 209458b must recirculate at least @xmath12% of incident stellar energy to its night side , at the @xmath6 level ( at least @xmath13% at the @xmath1 level ) , which is consistent with other constraints on recirculation from the depth of secondary eclipse depth at 8 @xmath0 m and the low optical albedo .
these data indicate that different hot jupiter planets may experience different recirculation efficiencies .
methods : data analysis stars : planetary systems techniques : photometric |
throughout the paper @xmath0 is a simply - connected simple compact lie group and @xmath1 is an orientable smooth closed 4-manifold . given a principal @xmath0-bundle @xmath6 , its associated gauge group @xmath7 is the group of @xmath0-equivariant automorphisms of @xmath8 which fix @xmath1 and the pointed gauge group @xmath9 is the subgroup containing those automorphisms which restrict to the identity on the fiber over the basepoint . in @xcite , atiyah and bott show that @xmath10 is weakly homotopy equivalent to @xmath11 , the component of the space of maps from @xmath1 to @xmath12 that contains the map inducing @xmath8 .
similarly there is a weak homotopy equivalence between @xmath13 and the pointed mapping space @xmath14 . in what follows we do not distinguish between @xmath10 and @xmath11 .
the isomorphism class of @xmath8 is classified by @xmath15 $ ] . in the cases we consider
@xmath15 $ ] is isomorphic to @xmath16 and each principal @xmath0-bundle @xmath8 corresponds to some integer @xmath17 .
denote its associated gauge group by @xmath18 .
in @xcite theriault shows that when @xmath1 is a simply - connected spin 4-manifold with @xmath19 of rank @xmath20 there is a homotopy equivalence @xmath21 this reduces the calculation of the homotopy type of @xmath18 to that of @xmath22 . in the non - spin case the homotopy equivalence still holds when localized away from 2 .
the goal of this paper is to study the homotopy types of @xmath18 for non - simply - connected @xmath1 .
it turns out that under certain conditions the structure of @xmath23 determines a decomposition of @xmath18 . in section 2
we develop lemmas which lead to a statement relating a splitting of @xmath23 to a decomposition of @xmath18 .
more precisely , if there is a homotopy equivalence @xmath24 such that @xmath25 and the @xmath26 s are cw - complexes satisfying the conditions described in proposition [ x seq ] , then @xmath18 will admit a homotopy decomposition @xmath27 where @xmath28 is the connected component containing the basepoint . in section 3
we apply this result and calculate @xmath23 and @xmath18 for three specific cases of non - simply - connected 4-manifolds @xmath1 , in which the @xmath26 s are spheres and moore spaces , and @xmath25 is either @xmath29 or @xmath30 according to the structure of @xmath23 .
in this section we develop some useful lemmas that apply more generally to an orientable 4-dimensional cw - complex @xmath31 with one 4-cell . for @xmath32 let @xmath33 be its @xmath34-skeleton and
let be the attaching map of the 4-cell .
we begin by showing that the isomorphism class of principal @xmath0-bundles over @xmath31 is classified by @xmath16 .
[ isom class of p_k ] there is an isomorphism @xmath35\cong{\mathbb{z}}$ ] .
the homotopy cofibration sequences @xmath36 give the following diagram @xmath37}\ar[d]^-{(\sigma q)^ * } & & & \\ { [ \sigma(x_3 ) , bg]}\ar[r]^-{(\sigma f)^*}\ar[d ] & { [ s^4 , bg]}\ar[r]^-{p^ * } & { [ x , bg]}\ar[r ] & { [ x_3 , bg]}\\ { [ \sigma(x_2 ) , bg ] } & & & } \ ] ] where the row and the column are exact sequences .
we will show that @xmath38 is an isomorphism .
since @xmath12 is 3-connected , @xmath39 $ ] and @xmath40 $ ] are zero .
therefore @xmath38 is an epimorphism and any map @xmath41 $ ] has a lifting @xmath42 through @xmath43 to @xmath44 $ ] .
the latter implies that @xmath45 .
note that the composition @xmath46 is the differential map of the 4-cell in the cellular chain complex of @xmath31 .
since @xmath47 is @xmath16 , this implies that @xmath46 is null homotopic .
therefore @xmath48 and so @xmath49 .
hence @xmath38 is also an injection .
lemma [ isom class of p_k ] says that any principal @xmath0-bundle over @xmath31 corresponds to some integer @xmath17 so our notation @xmath50 is justified .
when @xmath31 is simply - connected , it admits a minimal cell structure so @xmath51 is homotopy equivalent to a wedge of 2-spheres @xmath52 .
the homotopy equivalence in ( [ simply_conn_decomp ] ) is obtained as a consequence of the attaching map @xmath53 of the 4-cell in @xmath31 having the property that @xmath54 is null homotopic .
so if we can find a subcomplex @xmath55 such that the attachment of @xmath53 onto @xmath56 vanishes after suspension , then @xmath50 can be decomposed into a mapping space of @xmath56 and other factors following the same argument .
[ suspended f as sum ] if @xmath57 , then @xmath54 is homotopic to where @xmath58 is the composition @xmath59 .
since @xmath54 is a co - h - map , there is a homotopy commutative diagram @xmath60 ^ -{\sigma f}\ar[d]^-{\sigma } & \sigma(x_3)\ar[r]^-{g+h}\ar[d]^-{\sigma } & \sigma(x_3)\\ s^4\vee s^4\ar[r]^-{\sigma f\vee\sigma f } & \sigma(x_3)\vee\sigma(x_3)\ar[r]^-{g\vee h } & \sigma(x_3)\ar[u]_-{\triangledown } } \ ] ] where @xmath61 is the comultiplication and @xmath62 is the folding map .
the map along the bottom row is homotopic to @xmath63 , so we have @xmath64 .
go back to our case .
observe that @xmath65 is the identity map in @xmath66 .
therefore we have @xmath67 . [ gaugedecomp ]
suppose that there exist a cw - complex @xmath56 of dimension at most 3 and a map @xmath68 such that @xmath69 has a left homotopy inverse .
then @xmath70 , where @xmath71 is the homotopy cofibre of @xmath72 , and there is a homotopy equivalence @xmath73 where @xmath28 is the connected component containing the basepoint .
the homotopy cofibration sequence @xmath74 induces a long exact sequence @xmath75\overset{(\sigma\phi)^*}{\longrightarrow}[\sigma y , bg]\overset{\epsilon^*}{\longrightarrow}[x ' , bg]\overset{q^*}{\longrightarrow}[x , bg]\overset{\phi^*}{\longrightarrow}[y , bg].\ ] ] since @xmath69 has a left homotopy inverse , the map @xmath76 is null homotopic and @xmath77 is zero . by hypothesis ,
the dimension of @xmath56 is at most 3 and @xmath12 is 3-connected , so @xmath78 $ ] is zero .
therefore by exactness @xmath79\to[x , bg]$ ] is an isomorphism .
fix an integer @xmath17 and consider evaluation fibrations @xmath80 { \textrm{map}}^*_k(x ' , bg)\rightarrow{\textrm{map}}_k(x ' , bg)\rightarrow bg .
& \end{array}\ ] ] since @xmath81 is an isomorphism , by naturality there is commutative diagram @xmath82\ar@{=}[d ] & { \textrm{map}}^*_k(x ' , bg)\ar[r]\ar[d]^-{q^ * } & { \textrm{map}}_k(x ' , bg)\ar[r]\ar[d]^-{q^ * } & bg\ar@{=}[d]\\ g\ar[r ] & { \textrm{map}}^*_k(x , bg)\ar[r ] & { \textrm{map}}_k(x , bg)\ar[r ] & bg } \]][1-conn naturality ] as in @xcite , from the left square in ( [ 1-conn naturality ] ) we obtain a homotopy commutative diagram @xmath83 ^ -{\cong}\ar[d ] & { \textrm{map}}^*_k(\sigma x , bg)\ar[d]\\ { \mathcal{g}}_k(x')\ar[r]^-{q^*}\ar@{=}[d ] & { \mathcal{g}}_k(x)\ar[r]^-{h}\ar[d ] & { \textrm{map}}^*_{k'}(\sigma y , bg)\ar[d]^-{\epsilon^*}\\ { \mathcal{g}}_k(x')\ar[r ] & g\ar[r]\ar[d ] & { \textrm{map}}^*_k(x ' , bg)\ar[d]\\ & { \textrm{map}}^*_k(x , bg)\ar@{=}[r ] & { \textrm{map}}^*_k(x , bg ) } \ ] ] here @xmath84 is some connected component .
it is known that all connected components of @xmath85 are homotopy equivalent as @xmath86 is a co - h - space @xcite , so we may replace it by @xmath87 . since @xmath76 is null homotopic , @xmath77 is trivial and therefore @xmath88 has a right homotopy inverse .
the group multiplication in @xmath50 then gives a homotopy equivalence @xmath89 and therefore @xmath50 is homotopy equivalent to @xmath90 . by cellular approximation
, we may assume that the image of @xmath72 is contained in @xmath51 .
let @xmath91 be the map induced by restricting the range of @xmath72 to @xmath51 . if @xmath69 has a left homotopy inverse @xmath92 , then the composition @xmath93 is a left homotopy inverse of @xmath94 .
therefore @xmath66 is homotopy equivalent to @xmath95 , where @xmath96 is the homotopy cofibre of @xmath97 , and the composition @xmath98 is null homotopic .
conversely , the latter two conditions imply that @xmath69 has a left homotopy inverse
. then we can rephrase lemma [ gaugedecomp ] as follows .
[ gaugedecomp][sigma f factors ] suppose that there exist a cw - complex @xmath56 of dimension at most 3 and a map @xmath91 such that : ( 1 ) @xmath99 and ( 2 ) the composition @xmath100 is null homotopic .
then there are homotopy equivalences @xmath101 moreover , @xmath54 factors through @xmath102 where @xmath103 is the attaching map of the 4-cell in @xmath71 . from the above discussion , we only need to show that @xmath54 factors through @xmath102 .
expand the commutative diagram @xmath104\ar[d]^-{\phi ' } & y\ar[d]^-{\phi}\\ x_3\ar[r ] & x } \ ] ] to a homotopy cofibration diagram @xmath104\ar[d]^-{\phi ' } & y\ar[r]\ar[d]^-{\phi } & { * } \ar[r]\ar[d ] & \sigma y\ar[d]^-{\sigma\phi'}\\ x_3\ar[r]\ar[d]^-{q ' } & x\ar[r]\ar[d]^-{q } & s^4\ar[r]^-{\sigma f}\ar@{=}[d ] & \sigma(x_3)\ar[d]^-{\sigma q'}\\ x'_3\ar[r ] & x'\ar[r ] & s^4\ar[r]^-{\sigma f ' } & \sigma(x'_3 ) } \ ] ] which defines the maps @xmath105 and @xmath106 .
let be a left homotopy inverse of @xmath69 and let @xmath107 be the suspension of the skeletal inclusion .
by lemma [ suspended f as sum ] we have @xmath108 since @xmath109 is null homotopic , @xmath110 factors through @xmath102 .
[ q isom on h free ] assume the conditions in proposition [ gaugedecomp ] .
if @xmath111 is torsion , then the quotient map @xmath112 induces an isomorphism @xmath113 preserving cup products on the free parts .
consider the long exact sequence of reduced cohomology groups @xmath114 again , use a left homotopy inverse and get an isomorphism @xmath115 since @xmath111 is torsion , @xmath81 is an isomorphism on the free part of second cohomology groups and the naturality of cup products implies the statement .
now we go back to calculate @xmath50 when @xmath31 is simply - connected .
the 3-skeleton @xmath51 is homotopy equivalent to @xmath116 and the attaching map @xmath53 is an element of @xmath117 . by the hilton - milnor theorem
it can be written as @xmath118 , where @xmath119 is an integer , the map @xmath120 is the composition of the hopf map @xmath121 with the inclusion and @xmath122 is a sum of whitehead products .
if @xmath54 is not trivial , then some of the @xmath119 s are odd .
let @xmath123 be the number of odd @xmath119 s , where @xmath124 .
relabelling the spheres if necessary , we may assume that @xmath119 is odd for @xmath125 and is even otherwise .
we can imagine that @xmath53 is a vector , @xmath119 s are coordinates and the pinch map @xmath126 is a coordinate function because @xmath58 kills @xmath122 and @xmath127
. in linear algebra , the expression of a vector can be simplified by change of basis .
the following lemma realizes this idea geometrically .
suppose that @xmath128 is a wedge of 2-spheres and @xmath118 .
then there is a map @xmath129 such that its homotopy cofibre @xmath130 is homotopy equivalent to @xmath31 and it satisfies @xmath131 and @xmath132 , where @xmath133 and @xmath134 .
if all @xmath119 s are even , then @xmath54 is zero in @xmath135 and set @xmath136 to be @xmath53 .
otherwise @xmath123 , the number of odd @xmath119 s , is greater than zero . for each @xmath137 , define a map @xmath138 as follows .
on @xmath139 , @xmath140 is the composition @xmath141 where @xmath61 is a comultiplication and @xmath62 is a folding map of @xmath142 and @xmath143 . on the remaining spheres , @xmath140 is the identity .
extend @xmath140 to the whole of @xmath31 and call it @xmath144 . since @xmath144 induces an isomorphism on @xmath145 and @xmath31
is simply - connected , @xmath144 is a homotopy equivalence .
let @xmath146 be @xmath147 and @xmath148 be its homotopy cofibre .
then there is a homotopy commutative diagram @xmath149 ^ -{f}\ar@{=}[d ] & x_3\ar[r]\ar[d]^-{\phi_k } & x\ar[d]^-{\tilde{\phi_k}}\\ s^3\ar[r]^-{\tilde{f}_k } & x_3\ar[r ] & \tilde{x}_k } \ ] ] by the five lemma , @xmath150 is an isomorphism on their cohomology groups , so by whitehead s theorem @xmath150 is a homotopy equivalence .
further , @xmath151 where @xmath152 and @xmath153 .
the composition @xmath154 is homotopic to @xmath155 because is homotopic to @xmath156 and @xmath157 .
set @xmath136 to be @xmath158 and set @xmath130 to be its homotopy cofibre .
then @xmath136 and @xmath130 satisfy the statement requirements .
if @xmath54 is trivial , then the 5-cell attaches to @xmath66 at the basepoint and @xmath159 is homotopy equivalent to @xmath160 .
set @xmath56 to be @xmath52 and set @xmath161 to be the identity .
then lemma [ gaugedecomp ] recovers equation ( [ simply_conn_decomp ] ) , @xmath162 if @xmath54 is not trivial , then we may replace @xmath31 by @xmath130 and @xmath53 by @xmath136 in lemma [ simplify s2 ] such that @xmath163 .
the 5-cell only attaches to @xmath164 in @xmath66 .
set @xmath56 to be @xmath165 and set @xmath166 to be the inclusion .
the quotient @xmath167 is the mapping cone of @xmath168 where @xmath169 is odd .
denote it by @xmath170 .
we adopt this notation because @xmath170 is @xmath30 when @xmath171 or @xmath172 .
one may ask when @xmath173 can be replaced by @xmath174 .
[ cp_a ] for @xmath169 odd , @xmath175 is homotopy equivalent to @xmath176 .
let @xmath177 be the composition @xmath178 and let @xmath179 be its homotopy cofibre .
by lemma [ simplify s2 ] , we can assume that @xmath180 is @xmath181 and the composition is null homotopic . set @xmath56 to be @xmath182 and use lemma [ gaugedecomp ] to get @xmath183 swap @xmath184 and @xmath182 and repeat the process .
then we have @xmath185 comparing the two equations gives the asserted homotopy equivalence .
[ simplify s2 ] assume that @xmath51 is homotopy equivalent to @xmath116 and @xmath145 satisfies poincar duality . if @xmath54 is null homotopic , then there are homotopy equivalences @xmath186 and if @xmath54 is non - trivial , then there are homotopy equivalences @xmath187 the case when @xmath54 is trivial is already discussed above .
assume that @xmath54 is not trivial .
if @xmath51 is @xmath188 , then poincar duality will imply that the cup square product of the generator of @xmath189 must be 1 or -1 .
therefore @xmath31 is homotopy equivalent to @xmath30 and the statement holds . otherwise , write @xmath118 where @xmath190 and @xmath122 is a sum of whitehead products .
at least one of the @xmath119 s is odd and it may be assumed to be @xmath191 without loss of generality . by lemma
[ simplify s2 ] we may assume @xmath192 .
since the 5-cell only attaches to @xmath164 , @xmath159 is homotopy equivalent to @xmath193 .
set @xmath194 to be the wedge of 2-spheres except @xmath184 .
then the composition @xmath195 is null - homotopic and lemma [ gaugedecomp ] implies that there is a homotopy equivalence @xmath196 since @xmath20 is great than 1 , there is at least one copy of @xmath197 in the equation .
therefore we can replace @xmath198 by @xmath174 by lemma [ cp_a ] .
lemma [ simplify s2 ] is a refinement of equation ( [ simply_conn_decomp ] ) for the non - spin case in @xcite .
it uses lemma [ cp_a ] which says that for @xmath169 odd , @xmath173 is homotopy equivalent to @xmath174 in the presence of @xmath197 .
in fact after localizing away from 2 , the attaching maps in @xmath30 and @xmath170 are trivial , so @xmath174 and @xmath173 are both homotopy equivalent to @xmath199 and hence homotopy equivalent to each other . moreover , since @xmath200 is homotopic to @xmath201
, there is a homotopy equivalence .
the evaluation fibrations @xmath202
\omega{\textrm{map}}_k(\sigma{\mathbb{c}}{\mathbb{p}}^2_a)\to\omega g\to{\textrm{map}}^*_k(\sigma{\mathbb{c}}{\mathbb{p}}^2_a)\to{\textrm{map}}_k(\sigma{\mathbb{c}}{\mathbb{p}}^2_a)\to g & \end{array}\ ] ] and the naturality of @xmath203 give a homotopy commutative diagram @xmath204\ar[r ] & \omega g\ar@{=}[d]\ar[r ] & { \textrm{map}}_k^*(\sigma{\mathbb{c}}{\mathbb{p}}^2_a , bg)\ar[d]^-{\varphi^*}\\ \omega{\mathcal{g}}_k({\mathbb{c}}{\mathbb{p}}^2)\ar[r ] & \omega g\ar[r ] & { \textrm{map}}_k^*(\sigma{\mathbb{c}}{\mathbb{p}}^2 , bg ) . }
\ ] ] so @xmath205 and @xmath206 are homotopy equivalent .
all these facts suggest that @xmath174 and @xmath173 may be homotopy equivalent , but we can not determine the answer exactly . we summarizing this section in the following proposition .
[ x seq ] suppose that @xmath31 is an orientable 4-dimensional cw - complex with one 4-cell .
let @xmath207 be a sequence of cw - complexes such that ( 1 ) @xmath208 and ( 2 ) for , where @xmath209 and @xmath26 is a cw - complex of dimension at most 3 .
if for each triple @xmath210 satisfies the conditions in lemma [ gaugedecomp ] , then there are homotopy equivalences @xmath211 in addition , if ( 3 ) all @xmath212 s are torsion , ( 4 ) the 3-skeleton of @xmath71 is homotopy equivalent to @xmath52 and ( 5 ) @xmath145 satisfies poincar duality , then we have @xmath213 when @xmath54 is null homotopic and @xmath214 when @xmath54 is non - trivial .
the first two homotopy equivalences clearly come from applying lemma [ gaugedecomp ] to each triple @xmath210 .
so we only need to prove the second part .
suppose that the sequence satisfies conditions ( 3 ) ( 5 ) .
let @xmath215 be the attaching maps of the 4-cell in @xmath216 and let @xmath217 be the quotient map .
since all the @xmath212 s are torsion , the composition @xmath218 induces an isomorphism which preserves cup products on the free part by lemma [ q isom on h free ] and hence @xmath219 satisfies poincar duality .
moreover , each @xmath220 factors through @xmath221 by lemma [ sigma f factors ] , so @xmath222 is trivial if and only if @xmath54 is .
applying lemma [ simplify s2 ] to @xmath25 implies the statement .
from now on we assume that @xmath1 is an orientable smooth closed 4 manifold . using morse theory , in ( * ? ? ?
* theorem 3.35 ) it is shown that @xmath1 admits a cw - structure with one 4-cell .
let @xmath223 be the 3-skeleton of @xmath1 and let @xmath224 be the attaching map of the 4-cell . in this section we will calculate the homotopy types of @xmath23 and @xmath18 when @xmath2 is : ( 1 ) a free group @xmath3 ; ( 2 ) a cyclic group @xmath4 where @xmath5 is an odd prime ; ( 3 ) a free product of the previous two types @xmath225 , where @xmath226 is a power of an odd prime .
when @xmath2 is a free group , there is a homotopy equivalence @xcite @xmath228 for @xmath229 , let @xmath230 be the @xmath34th copy of the @xmath123-sphere and let @xmath231 be the composition . by lemma [ suspended f as sum ] ,
@xmath54 is homotopic to the sum @xmath232 . for each @xmath233 ,
the map @xmath234 is null homotopic because @xmath235 is zero .
note that @xmath236 is the differential map of the 4-cell to @xmath237 in the cellular chain complex .
it is zero since @xmath1 is orientable and @xmath238 .
therefore each @xmath239 is null homotopic for @xmath240 and @xmath241 .
set @xmath31 to be @xmath1 , set @xmath56 to be @xmath242 and set @xmath243 to be the inclusion .
then the composition @xmath244 is null homotopic .
note that @xmath111 is zero and @xmath245 is @xmath246 .
using proposition [ x seq ] we have the following theorem .
[ gauge decomp_free ] suppose @xmath247 .
if @xmath54 is null homotopic , then there are homotopy equivalences @xmath248 \displaystyle{{\mathcal{g}}_k(m)\simeq{\mathcal{g}}_k(s^4)\times\prod^m_{i=1}\omega^3_0g\times\prod^d_{j=1}\omega^2g\times\prod^m_{k=1}\omega g. } \end{array}\ ] ] if @xmath54 is non - trivial , then there are homotopy equivalences @xmath249 \displaystyle{{\mathcal{g}}_k(m)\simeq{\mathcal{g}}_k({\mathbb{c}}{\mathbb{p}}^2)\times\prod^m_{i=1}\omega^3_0g\times\prod^{d-1}_{j=1}\omega^2g\times\prod^m_{k=1}\omega g. } \end{array}\ ] ] for @xmath251 , recall that an @xmath252-dimensional moore space @xmath253 is defined to be the mapping cone of the degree @xmath17 map .
its reduced homology groups are zero except .
denote its generator by @xmath254 . with @xmath255 coefficients
, the universal coefficient theorem implies that @xmath256 is generated by @xmath254 , @xmath257 is generated by @xmath258 , the bockstein homomorphism image of @xmath254 , and @xmath259 is zero for @xmath260 . for any space @xmath31 , define the mod-@xmath17 homotopy group @xmath261 to be @xmath262 $ ] .
the comultiplication of @xmath253 induces a group multiplication in @xmath261 for @xmath263 , which becomes abelian for @xmath264 since @xmath265 .
moreover , there is a mod-@xmath17 hurewicz homomorphism analogous to the integral hurewicz homomorphism @xmath266 it is natural and compatible with bockstein homomorphisms @xcite @xmath267 & \pi_{n+1}(x;{\mathbb{z}}_k)\ar[r]^-{\beta_\pi}\ar[d]^-{\bar{\varphi } } & \pi_n(x)\ar[r]^-{k}\ar[d]^-{\varphi } & \pi_n(x)\ar[r]\ar[d]^-{\varphi } & \pi_n(x;{\mathbb{z}}_k)\ar[r]\ar[d]^-{\bar{\varphi } } & \cdots\\ \cdots\ar[r ] & h_{n+1}(x;{\mathbb{z}}_k)\ar[r]^-{\beta } & h_n(x)\ar[r]^-{k } & h_n(x)\ar[r ] & h_n(x;{\mathbb{z}}_k)\ar[r ] & \cdots } \ ] ] where @xmath268 is the standard hurewicz homomorphism and @xmath269 is the bockstein homomorphism on homotopy groups . let @xmath270 be the kernel of @xmath268 .
the mod-@xmath17 hopf invariant is defined in a similar way to the integral hopf invariant . for
a map @xmath271 , let @xmath272 be the mapping cone of @xmath273 , and let @xmath123 and @xmath274 be generators of @xmath275 and @xmath276 .
then the mod-@xmath17 hopf invariant @xmath277 is defined by the formula @xmath278 .
* corollary 11.12)[mod p hurewicz ses ] let @xmath263 and @xmath5 be an odd prime .
then there is a short exact sequence @xmath279 where @xmath280 and @xmath281 for @xmath264 .
moreover , if @xmath282 , then it is null homotopic if and only if its mod-@xmath283 hopf invariant @xmath284 .
suppose that @xmath2 is @xmath4 .
poincar duality and the universal coefficient theorem imply that @xmath1 has homology groups @xmath285 in order to apply proposition [ x seq ] to calculate @xmath18 , we need to study the structure of @xmath286 . after localizing away from @xmath5 ,
any @xmath289 is contractible so @xmath287 and @xmath288 are @xmath5-torsion .
localize at @xmath5 .
consider the long exact sequence of homotopy groups for the pair @xmath290 @xmath291 the pair @xmath290 is 3-connected , so @xmath292 is @xmath16 by the hurewicz theorem .
since @xmath293 at odd primes , @xmath294 is an injection .
but @xmath287 is torsion , so this injection only makes sense if @xmath295 .
now we calculate @xmath288 .
it is equal to @xmath296 , where @xmath297 is the homotopy fibre of the pinch map @xmath298 . by (
* proposition 11.7.1 ) , there is a @xmath5-local homotopy equivalence @xmath299 where @xmath300 is the fibre of the degree @xmath301 map @xmath302 and the dimension @xmath303 is either 3 or higher than 4 . on the right side , @xmath304 and @xmath305 since @xmath306 is 3-connected for each @xmath307 . for the remaining factor of @xmath308 , consider the string of isomorphisms @xmath309 the first isomorphism is obtained from the hilton - milnor theorem using dimension and connectivity considerations .
the second isomorphism holds since only one @xmath310 equals 3 while the rest are strictly larger than 4 .
the third isomorphism holds by adjunction , and the isomorphism with zero holds by the first assertion of the lemma .
therefore @xmath311 and hence so is @xmath288 .
[ 2 moore sp p ] if @xmath2 is @xmath4 for @xmath5 an odd prime , then there is a homotopy equivalence @xmath312 in addition , there exists a map @xmath313 such that @xmath314 is an isomorphism and @xmath315 has a left homotopy inverse .
after suspension , @xmath286 is simply - connected and admits a minimal cell structure . by ( * ? ? ?
* proposition 4h.3 ) @xmath286 is homotopy equivalent to the mapping cone of @xmath316 where @xmath88 induces a trivial homomorphism @xmath317 .
now we claim that @xmath88 is null homotopic . since @xmath318\cong[p^3(p^r ) , p^3(p^r)]\oplus(\oplus^d_{i=1}[s^2_i , p^3(p^r)]),\ ] ] we may write @xmath319 , where @xmath320 $ ] and @xmath321 $ ] . consider the commutative diagram @xmath322 & k_3(p^r)\ar[r ] & \pi_3(p^3(p^r);{\mathbb{z}}_{p^r})\ar[r]^-{\varphi}\ar[d]^-{\beta_\pi } & h_3(p^3(p^r);{\mathbb{z}}_{p^r})\ar[r]\ar[d]^-{\beta_h } & 0\\ & & \pi_2(p^3(p^r))\ar[r]^-{\varphi } & h_2(p^3(p^r ) ) & } \ ] ] where the first row is the exact sequence in lemma [ mod p hurewicz ses ] and the square is due to the compatibility of mod-@xmath283 hurewicz homomorphisms and bockstein homomorphisms . since @xmath323 is trivial , so is @xmath324 and therefore @xmath325 .
note that @xmath326 is an isomorphism in this case , so @xmath327 is in the kernel @xmath328 .
denote the mapping cone of @xmath327 by @xmath329 .
since @xmath330 is a suspension , @xmath331 has trivial cup products and so does @xmath332 .
therefore the mod-@xmath283 hopf invariant @xmath333 is zero and @xmath327 is null homotopic by lemma [ mod p hurewicz ses ] . for each @xmath34 , there is a commutative diagram @xmath334 ^ -{[\,h''_i]\,}\ar[d]_-{\varphi } & \pi_2(p^3(p^r))\ar[d]_-{\varphi}\\ h_2(s^2_i)\ar[r]^-{(h''_i ) _ * } & h_2(p^3(p^r ) ) } \ ] ] both hurewicz homomorphisms @xmath268 are isomorphisms by the hurewicz theorems .
the induced homomorphism @xmath335 is trivial and the diagram implies that @xmath336 is null homotopic .
now we prove the second part of the lemma .
the 1-skeleton of @xmath1 is a wedge of spheres @xmath337 .
let @xmath338 and @xmath339 be inclusions .
their induced homomorphisms @xmath340 is a surjection and @xmath341 is a reduction map .
therefore at least one of the @xmath342 s sends the identity @xmath343 to the generator of @xmath2 . without loss of generality
we assume that it is @xmath344 .
extend @xmath345 to @xmath346 through the pinch map @xmath274 .
@xmath347 ^ -{p^r } & s^1\ar[r]^-{\beta}\ar[d]^-{j_1 } & p^2(p^r)\ar@{-->}[dl]^{\epsilon}\\ & m & } \ ] ] by cellular approximation the range of @xmath76 can be restricted to @xmath223 .
note that both @xmath348 are reduction maps and @xmath349 , so @xmath314 is an isomorphism . by the hurewicz theorem ,
the induced homology group homomorphism is an isomorphism and so is @xmath350 .
therefore the composition @xmath351 induces an isomorphism in homology and so is a homotopy equivalence .
[ gauge decomp_torsion ] suppose that @xmath2 is @xmath4 where @xmath5 is an odd prime .
if @xmath54 is null homotopic , then there are homotopy equivalences @xmath356 \displaystyle{{\mathcal{g}}_k(m)\simeq{\mathcal{g}}_k(s^4)\times\omega^3_0g\{p^r\}\times\prod^d_{i=1}\omega^2g\times\omega^2g\{p^r\}. } \end{array}\ ] ] if @xmath54 is non - trivial , then there are homotopy equivalences @xmath357 \displaystyle{{\mathcal{g}}_k(m)\simeq{\mathcal{g}}_k({\mathbb{c}}{\mathbb{p}}^2)\times\omega^3_0g\{p^r\}\times\prod^{d-1}_{i=1}\omega^2g\times\omega^2g\{p^r\ } , } \end{array}\ ] ] where @xmath358 is the connected component containing the basepoint . in the following
we construct a sequence @xmath207 of cw - complexes satisfying the conditions in proposition [ x seq ] for @xmath1 .
let @xmath359 be the 3-skeleton in @xmath216 and let @xmath215 be the attaching map of the 4-cell .
_ step 1 _ : by lemma [ 2 moore sp p ] , there is a map @xmath313 such that @xmath360 has a left homotopy inverse and @xmath314 is an isomorphism .
set @xmath361 to be @xmath1 , set @xmath362 to be @xmath346 and set @xmath363 to be @xmath76 .
then the composition @xmath364 is null homotopic since @xmath365 by lemma [ pi_pm ] .
then we use lemma [ gaugedecomp ] and get @xmath366 _ step 2 _ : set @xmath367 to be @xmath368 .
then @xmath367 is simply connected and admits a minimal cell structure , that is @xmath369 and @xmath370 .
set @xmath371 to be @xmath372 and set to be the inclusion .
the composition @xmath373 is null homotopic since @xmath287 is zero by lemma [ pi_pm ] .
then we use lemma [ gaugedecomp ] and get @xmath374 _ step 3 _ : set @xmath375 to be @xmath376 .
the 3-skeleton @xmath377 is homotopy equivalent to @xmath378 .
note that @xmath212 is torsion for @xmath379
. then proposition [ x seq ] implies the asserted statement .
suppose that @xmath381 , where @xmath382 is a power of an odd prime .
it is known that for some number @xmath383 there is a diffeomorphism @xmath384 where @xmath385 and @xmath386 s are orientable smooth closed 4-manifolds with @xmath387 and ( * ? ? ? * theorem 1.3 ) .
using the fact that the 3-skeleton of a connected sum @xmath388 is a wedge of the 3-skeletons of @xmath389 and @xmath390 , we can calculate the suspensions and gauge groups for both sides of ( [ klt decomp ] ) . by induction
we only need to show the lemma for @xmath392 .
the 3-skeleton of is @xmath393 .
there is a map @xmath394 which collapses the connecting cylinder between @xmath1 and @xmath395 .
so we have a homotopy commutative diagram @xmath396\ar@{=}[r ] & m_3\vee s^2\vee s^2\ar[d]\\ m\#(s^2\times s^2)\ar[d]\ar[r]^-{q } & m\vee(s^2\times s^2)\ar[d]\\ s^4\ar[d]^-{\sigma f_\#}\ar[r]^-{\sigma } & s^4\vee s^4\ar[d]^-{\sigma f\vee\sigma f_s}\\ \sigma(m_3)\vee s^3\vee
s^3\ar@{=}[r ] & \sigma(m_3)\vee s^3\vee s^3 } \ ] ] where @xmath397 and @xmath398 are the attaching maps in @xmath399 and @xmath395 respectively . set @xmath56 to be @xmath400 and set @xmath401 to be the inclusion .
since @xmath398 is a whitehead product , the composition @xmath402 is null homotopic and the statement follows from lemma [ gaugedecomp ] . [ gauge decomp _ * ] if @xmath381 where @xmath226 is a power of an odd prime , then there exists an integer @xmath383 such that the suspension of @xmath403 is homotopy equivalent to either @xmath404 \displaystyle{\sigma{\mathbb{c}}{\mathbb{p}}^2\vee(\bigvee^m_{i=1}s^4)\vee(\bigvee^n_{j=1}p^4(q_j))\vee(\bigvee^{d+2l-1}_{k=1}s^3)\vee(\bigvee^n_{j'=1}p^3(q_{j'}))\vee(\bigvee^m_{i'=1}s^2 ) . } & \end{array}\ ] ] in the first case , we have @xmath405 in the second case , we have @xmath406 by ( [ klt decomp ] ) there exists a diffeomorphism @xmath407 so it is equivalent to calculate @xmath408 and @xmath409 . in the following
we construct a sequence @xmath207 of cw - complexes satisfying the conditions in proposition [ x seq ] for @xmath410 .
let @xmath359 be the 3-skeleton of @xmath216 and let @xmath215 be the attaching map of the 4-cell .
_ step 1 _ : set @xmath361 to be @xmath411 .
its 3-skeleton @xmath412 is a wedge sum of the 3-skeletons of @xmath385 and the @xmath386 s . in particular
, the 3-skeleton of @xmath385 is @xmath413 @xcite .
set to be the wedge of 1-spheres and 3-spheres and set to be the inclusion . for each @xmath233 ,
the composition @xmath414 is null homotopic since @xmath235 is zero . and for each @xmath237 , the composition @xmath415 is the differential map of the 4-cell in the cellular chain complex , and so is zero because @xmath416 .
then we use lemma [ gaugedecomp ] and get @xmath417 \displaystyle{{\mathcal{g}}_k(x^1)\simeq{\mathcal{g}}_k(x^1/y^1)\times\prod^m_{i=1}\omega^3_0g\times\prod^m_{i'=1}\omega g. } & \end{array}\ ] ] _ step 2 _ : set @xmath367 to be @xmath418 .
its 3-skeleton @xmath419 is a wedge sum of 2-spheres and the 3-skeletons of the @xmath386 s .
denote the 3-skeleton of @xmath386 by @xmath420 . by lemma [ 2 moore sp p ] , @xmath421 is homotopy equivalent to @xmath422 and there exists a map @xmath423 such that @xmath424 has a left homotopy inverse and @xmath425 is an isomorphism . set @xmath371 to be @xmath426 and set @xmath427 to be the composition @xmath428 for each @xmath429 , the composition @xmath430 is null homotopic since @xmath431 is zero by lemma [ pi_pm ] .
then we use lemma [ gaugedecomp ] and get @xmath432 _ step 3 _ : set @xmath375 to be @xmath376 . by the van kampen theorem @xmath433 is an isomorphism and @xmath375 is simply connected .
therefore it admits a minimal cell structure and @xmath434 .
set @xmath435 to be @xmath436 and set @xmath437 to be the inclusion . for each @xmath429 ,
the composition @xmath438 is null homotopic since @xmath439 is zero by lemma [ pi_pm ] .
then we use lemma [ gaugedecomp ] and get @xmath440 _ step 4 _ : set @xmath441 to be @xmath442 .
its 3-skeleton @xmath443 is homotopy equivalent to @xmath444 .
since @xmath212 is torsion free for @xmath445 , proposition [ x seq ] implies that @xmath446 when @xmath447 is null homotopic and @xmath448 when @xmath447 is non - trivial .
the proofs of theorem [ gauge decomp_free ] and [ gauge decomp_torsion ] are also valid for any orientable 4-dimensional cw - complex with one 4-cell , while theorem [ gauge decomp _ * ] requires the smoothness of @xmath1 for the stable diffeomorphism splitting in @xcite .
one can ask under what condition the stabilizing factor @xmath453 can be cancelled so that the factors @xmath454 can be removed from the equation . if this can be achieved , theorem [ gauge decomp_free ] and [ gauge decomp_torsion ] will be corollaries of theorem [ gauge decomp _ * ] when @xmath1 is smooth
. m. atiyah and r. bott , _ the yang - mills equations over riemann surfaces _ , philosophical transactions of the royal society of london .
series a , mathematical and physical sciences , * 308*(1505 ) , ( 1983 ) , 523 615 .
m. kreck , w. lck and p. teichner , stable prime decompositions of 4-manifolds _ prospects in topology _ edited by f. quinn , annuals of mathematics studies , * 138 * , princeton university press , princeton , ( 1995 ) , 251 269 . | let @xmath0 be a simply - connected simple compact lie group and let @xmath1 be an orientable smooth closed 4-manifold . in this paper
we calculate the homotopy type of the suspension of @xmath1 and the homotopy types of the gauge groups of principal @xmath0-bundles over @xmath1 when @xmath2 is : ( 1 ) a free group @xmath3 , ( 2 ) a cyclic group @xmath4 where @xmath5 is an odd prime , or ( 3 ) a free product of the previous two types .
[ multiblock footnote omitted ] |
high redshift luminosity distance measurements @xcite and wmap measurement @xcite of cosmic microwave background ( cmb ) interpreted in the framework of standard frw cosmological models have strongly disfavored a matter dominated universe , and strongly supported a dominant dark energy component , corresponding to a positive cosmological acceleration , which we will denote as @xmath2 ( not to be confused with the scale factor @xmath3 ) . as an alternative to dark energy
, it has been proposed @xcite that we may be at the center of a inhomogeneous isotropic universe described by a lemaitre - tolman - bondi ( ltb ) solution of einstein s fields equations , where spatial averaging over one expanding and one contracting region is producing a positive averaged acceleration @xmath0 .
another more general approach to map luminosity distance as a function of redshift @xmath1 to ltb models has been recently proposed @xcite , showing that an inversion method can be applied successfully to reproduce the observed @xmath1 .
the main point is that the luminosity distance is in general sensitive to the geometry of the space through which photons are propagating along light geodesics , and therefore arranging appropriately the geometry of a given cosmological model it is possible to reproduce a given @xmath1 . for frw models this
correspond to set constraints on @xmath4 and @xmath5 and for ltb models it allows to determine the functions @xmath6 .
the averaged acceleration @xmath0 on the other side is not directly related to @xmath2 , since this is obtained integrating the position dependent cosmological redshift along the light geodesics , while @xmath0 is the result of spatial averaging , and has no relation to the causal structure of the underlying space .
this crucial difference can make @xmath0 unobservable to a central observer @xmath7 when the scale of the spatial averaging is greater than its event horizon .
lemaitre - tolman - bondi solution can be written as @xcite as : @xmath8 where @xmath9 is a function of the time coordinate @xmath10 and the radial coordinate @xmath11 , @xmath12 is an arbitrary function of @xmath11 , and @xmath13 denotes the partial derivative of @xmath9 with respect to @xmath11 .
einstein s equations give : @xmath14 with @xmath15 being an arbitrary function of @xmath11 and the dot denoting the partial derivative with respect to @xmath10 .
the solution of eq .
( [ eq2 ] ) can be written parametrically by using a variable @xmath16 , as follows @xmath17 \ , , \\
\label{eq5 } t(\eta , r ) & = & \frac{m(r)}{- 2 e(r ) } \left [ \eta -\frac{1}{\sqrt{-2 e(r ) } } \sin \left(\sqrt{-2 e(r ) } \eta \right ) \right ] + t_{b}(r ) \
, , \end{aligned}\ ] ] where @xmath18 has been introduced to make clear the distinction between the two functions @xmath19 and @xmath20 which are trivially related by @xmath21 and @xmath22 is another arbitrary function of @xmath11 , called bang function , which corresponds to the fact that big - bang / crunches happen at different times in this space .
this inhomogeneity of the location of the singularities is the origin of the possible causal separation between the central observer and the spatially averaged region for models with positive @xmath23 .
we can introduce the following variables @xmath24 so that eq . ( [ eq1 ] ) and
the einstein equations ( [ eq2 ] ) and ( [ eq3 ] ) can be written in a form which is more similar to frw models @xmath25 \ , , \ ] ] @xmath26 the solution in eqs .
( [ eq4 ] ) and ( [ eq5 ] ) can now be written as @xmath27 \
\label{ltb soln2 t } t(\tilde{\eta},r ) & = & \frac{\rho_0(r)}{6k(r ) } \left [ \tilde{\eta } -\frac{1}{\sqrt{k(r ) } } \sin \left(\sqrt{k(r ) } \ , \tilde{\eta } \right ) \right ] + t_{b}(r ) \
, , \end{aligned}\ ] ] where @xmath28 .
following the standard averaging procedure @xcite we define the volume for a spherical domain , @xmath29 , as @xmath30 and the length associated to the domain as @xmath31 via which the deceleration parameter @xmath32 and the average acceleration @xmath0 ( not to be confused with the scale factor @xmath3 ) are defined as : @xmath33 the models studied in @xcite correspond to the following functions @xmath34 : @xmath35 after exploring the 9 parameters space they give three examples of ltb solutions with positive @xmath32 as shown in table i. .three examples of the domain acceleration . [
cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] is plotted for the model corresponding to row 2 of table i from @xcite . ] since @xmath36 and @xmath37 we can fix the following initial condition for @xmath19 : @xmath38 which clearly shows why @xmath39 is called bang function .
defining @xmath40 as the time in the first column of table i , i.e. the time at which @xmath41 , with @xmath32 being the value in the last column of the same table , we solved the light geodesics equation @xmath42 imposing the following initial conditions @xmath43 eq.([ini1 ] ) is the natural way to map these models into luminosity distance observations for a central observer which should receive the light rays at the time @xmath40 at which the averaged acceleration is positive . in other words
the time dependence of @xmath44 corresponds to the choice of initial conditions for the light geodesics used to compute the luminosity distance . solving the equation @xmath45 for the model corresponding to the second row of table
i we obtain @xmath46 , where @xmath47 is the upper limit of the volume averaging integral . in this context @xmath48
can be thought of as the comoving horizon , the maximum radial coordinate from which photons can reach the central observer @xmath7 at time @xmath40 . as it can be seen in fig .
[ fig : m2rrr ] at @xmath49 there is a singularity along the geodesic , corresponding to the fact that light can not reach the central observer at time @xmath40 from points at radial coordinate @xmath50 .
in fact , since @xmath39 is a decreasing function of r , photons emitted at @xmath50 will arrive at @xmath49 at time @xmath51 when a local singularity develops because @xmath52 since regions of the universe at radial coordinates greater than @xmath48 have never been in causal contact with @xmath7 at time @xmath40 , the scale at which @xmath32 is defined is beyond the region causally connected to @xmath7 .
therefore @xmath32 can not be detected from local observation of the luminosity distance which is used to define @xmath2 .
this result is not surprising since the spatial averaging procedure is insensitive to the causal structure of the underlying space .
the other models studied in @xcite do not show this causal behavior , but solving the geodesic equations and computing the luminosity distance using the same initial conditions adopted above it can be shown that they do not reproduce correctly the observed luminosity distance , but this goes beyond the scope of the present article , which is the study of the local observability of the averaged acceleration .
we will study the luminosity distance of these and other ltb models with positive @xmath0 , and the relation between @xmath32 and @xmath53 in another work . and
@xmath54 are plotted for model 2 of @xcite.at about @xmath48=0.03298 there is a singularity .
, title="fig : " ] and @xmath54 are plotted for model 2 of @xcite.at about @xmath48=0.03298 there is a singularity .
, title="fig : " ] is plotted for the model corresponding to row 2 of table i from @xcite . ] and the dashed curve shows the bang function @xmath39 . for @xmath55 @xmath39
is so small that it can not be distinguished from the horizontal zero axis . ]
this example shows how @xmath0 may not even be causally related to the local observation of @xmath1 , and gives a reverse example of the results obtained in @xcite , where it was studied a ltb model fitting the observed luminosity distance , consistent with a positive @xmath2 , but without positive averaged acceleration @xmath0 .
our results do not rule out ltb models as alternatives to dark energy since , since the inversion method @xcite allows to obtain the observed luminosity distance without any averaging , and some concrete examples derived independently have already been proposed @xcite .
we can conclude that the luminosity distance @xmath1 contains more information than the spatially averaged acceleration @xmath0 because the first is sensitive to the causal structure of the entire space - time while the second is the result of averaging only the spatial part of the geometry , making the relation between them in general not one - to - one .
our results confirm the general argument @xcite that the spatial averaging must be performed on comoving domains whose size coincides with the length scale at which homogeneity sets in , and not on domains which may be as large as or larger than the observable universe .
it is possible that writing the geodesics equation in a form consistent with the buchert averaged formalism could overcome the problem , but this would be just a formal solution , since light is propagating along the geodesics of the inhomogeneous space , not of the averaged one .
we showed that ltb models with positive averaged acceleration can require averaging on scales beyond the event horizon of the central observer . in these cases
@xmath0 is causally unrelated to @xmath2 which is obtained from the observed luminosity distance , showing that a positive averaged acceleration @xmath0 in ltb models is in general not equivalent to a positive @xmath2 .
this example shows how the averaging buchert @xcite formalism is not preserving the causal structure of space - time and can lead to the definition of locally unobservable averaged quantities .
the study of the constraints on the local observability of averaged quantities will be addressed in a more general way , not only in the context of ltb models , in a future work .
r. a. vanderveld , e. e. flanagan and i. wasserman , phys .
d * 74 * , 023506 ( 2006 ) [ arxiv : astro - ph/0602476 ] .
s. apostolopoulos , n. brouzakis , n. tetradis and e. tzavara , jcap * 0606 * , 009 ( 2006 ) [ arxiv : astro - ph/0603234 ] . t. kai , h. kozaki , k. i. nakao , y. nambu and c. m. yoo , arxiv : gr - qc/0605120 . c. h. chuang , j. a. gu and w. y. hwang , arxiv : astro - ph/0512651 .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astron .
j. * 116 * , 1009 ( 1998 ) [ arxiv : astro - ph/9805201 ] . s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 517 * , 565 ( 1999 ) [ arxiv : astro - ph/9812133 ] .
d. j. h. chung and a. e. romano , phys .
d * 74 * , 103507 ( 2006 ) [ arxiv : astro - ph/0608403 ] .
e. w. kolb , s. matarrese and a. riotto , arxiv : astro - ph/0506534 .
d. palle , nuovo cim . * 117b * , 687 ( 2002 ) [ arxiv : astro - ph/0205462 ] . j. l. tonry _ et al . _
[ supernova search team collaboration ] , astrophys .
j. * 594 * , 1 ( 2003 ) [ arxiv : astro - ph/0305008 ] .
r. a. knop _ et al . _
[ the supernova cosmology project collaboration ] , astrophys .
j. * 598 * , 102 ( 2003 ) [ arxiv : astro - ph/0309368 ] .
b. j. barris _ et al .
_ , astrophys .
j. * 602 * , 571 ( 2004 ) [ arxiv : astro - ph/0310843 ] .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astrophys .
j. * 607 * , 665 ( 2004 ) [ arxiv : astro - ph/0402512 ] . c. l. bennett
_ et al . _ ,
astrophys .
j. suppl .
* 148 * , 1 ( 2003 ) [ arxiv : astro - ph/0302207 ] ; d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) [ arxiv : astro - ph/0302209 ] .
d. n. spergel _ et al .
_ , arxiv : astro - ph/0603449 . | we show that positive averaged acceleration @xmath0 obtained in ltb models through spatial averaging can require integration over a region beyond the event horizon of the central observer .
we provide an example of a ltb model with positive @xmath0 in which the luminosity distance @xmath1 does not contain information about the entire spatially averaged region , making @xmath0 unobservable .
since the cosmic acceleration @xmath2 is obtained from fitting the observed luminosity distance to a frw model we conclude that in general a positive @xmath0 in ltb models does not imply a positive @xmath2 . |
throughout the paper , @xmath2 denotes a very general quintic hypersurface , and @xmath21 is a stable rank two vector bundle of degree one . for the present considerations , bundles of degree @xmath0 are more practical in terms of hilbert polynomial .
we apologize for this inconvenience , but luckily the indexation by second chern class stays the same .
indeed , if @xmath21 has degree @xmath0 then @xmath22 as can be seen for example on the bundle @xmath23 with @xmath24 .
thus , the moduli space of stable bundles @xmath3 we look at here is isomorphic to @xmath25 considered in @xcite . ] with determinant @xmath26 such that @xmath27 .
the moduli space of stable bundles in general has been the subject of much work @xcite , but the special case @xmath11 considered here goes into a somewhat new and uncharted territory .
note that @xmath28 is generated by @xmath29 .
the canonical bundle is @xmath20 . for any @xmath16 we have @xmath30 . for @xmath31
the map @xmath32 is an isomorphism .
the hilbert polynomial of @xmath21 is @xmath33 .
in particular @xmath34 .
we will be assuming that @xmath21 is general in some irreducible component of the moduli space . from the previous paper @xcite using some techniques for bounding the singular locus which had also been introduced in @xcite and @xcite , it follows that @xmath21 is unobstructed , so if @xmath35 denotes the trace - free part of @xmath36 then @xmath37 .
note however that @xmath38 , indeed it is dual to @xmath39 .
thus @xmath40 the dual bundle is given by @xmath41 , so duality says that @xmath42 .
the dimension of any irreducible component of the moduli space is the expected one , @xmath43 .
the subspace of bundles @xmath21 with @xmath44 has dimension @xmath45 , see our previous paper @xcite , so a general @xmath21 has @xmath46 .
it follows from duality that @xmath47 and by @xmath34 we get @xmath48 . throughout the paper ( except at one place in section [ onquintic ] )
, we consider only bundles with @xmath46 .
duality says that @xmath49 is dual to @xmath50 . since @xmath51 , if we set @xmath52 then @xmath53 .
in particular there are at least @xmath54 linearly independent sections of @xmath55 which may be viewed as maps @xmath56 or equivalently @xmath57 .
note that the zero set of @xmath58 has to be of codimension @xmath4 , as any codimension - one component would be a divisor integer multiple of the hyperplane class but @xmath59 so this ca nt happen .
if we choose one such map @xmath58 , then we get the _ standard exact sequence _ @xmath60 and its twisted versions such as @xmath61 as a notational matter , @xmath62 denotes the ideal of @xmath63 in @xmath19 or in @xmath64 .
which one it is should be clear from context , for example it is the ideal of @xmath65 in the above sequences , and we choose not to weigh down the notation with an extra subscript .
calculation of the chern class @xmath27 shows that @xmath65 is a subscheme of length @xmath43 .
it is a union of possibly nonreduced points , which are locally complete intersections i.e. defined by two equations .
furthermore , as is classically well - known @xcite @xcite , @xmath63 satisfies the cayley - bacharach condition for @xmath66 which we denote by @xmath67 , saying that any subscheme @xmath68 of colength @xmath0 imposes the same number of conditions as @xmath63 on sections of @xmath66 .
the extension class is governed by an element of @xmath69 , and we have the exact sequence @xmath70 to get a locally free @xmath21 , the extension class should be nonzero on each vector coming from a point in @xmath63 , the existence of such being exactly @xmath67 . note that @xmath71 , and define @xmath72 . then @xmath73 ( the extension ca nt be split , indeed this is part of the @xmath67 condition ) , and @xmath74 .
the `` well - determined '' case is when @xmath75 .
then the extension class is well - defined up to a scalar multiple which does nt affect the isomorphism class of @xmath21 , and the existence of the nonzero class in @xmath76 is expected to impose @xmath77 conditions on the @xmath43 points , giving the expected dimension of the hilbert scheme of such subschemes @xmath78 to be @xmath79 . in section [ seminatural ] , we will show @xmath80 and in that case the bundle @xmath21 has seminatural cohomology .
here are a few techniques which will often be useful throughout the paper .
if a zero - dimensional subscheme @xmath65 satisfies @xmath81 then it satisfies @xmath82 for any @xmath83 .
suppose @xmath68 has colength @xmath0 .
choose a section @xmath84 nonvanishing at all points of @xmath63 , then if @xmath85 we have @xmath86 .
by @xmath81 , @xmath87 vanishes on @xmath63 , but @xmath88 is a unit near any point of @xmath63 so @xmath89 vanishes on @xmath63 , proving @xmath82 . the results of @xcite allow us to estimate the dimension of the hilbert scheme of @xmath90-dimensional subschemes of a curve , as was used in some detail in @xcite .
mainly , as soon as the curve is locally planar , the space of subschemes of length @xmath91 has dimension @xmath92 .
if @xmath93 is a divisor and @xmath63 is a zero - dimensional subscheme , we obtain the _ residual subscheme _ @xmath94 of @xmath63 with respect to @xmath95 , such that @xmath96 . it is characterized by the property that sections of @xmath97 which vanish on @xmath94 , map to sections of @xmath98 vanishing on @xmath63 . if @xmath63 is reduced then @xmath94 is just the union of those point of @xmath63 which are not in @xmath95 ; if @xmath63 contains some nonreduced schematic points then the structure of @xmath94 may be more complicated .
[ inquadric ] if @xmath63 satisfies @xmath99 and @xmath100 is a subscheme of colength @xmath4 , suppose @xmath101 is contained in a quadric
. then @xmath63 is contained in the same quadric .
the residual subscheme @xmath94 for the quadric has length @xmath102 .
if it is nonempty , we can choose a linear form containing a subscheme of colength @xmath0 of @xmath94 , corresponding to a subscheme @xmath103 of colength @xmath0 .
applying @xmath99 to the product of the quadric and the linear form is a contradiction , so @xmath104 and we re done .
suppose @xmath105 is a hyperplane , and let @xmath106 . by the genericity assumption on @xmath19 in particular @xmath107 generated by @xmath29
, we get that @xmath108 has to be reduced and irreducible .
its canonical sheaf is @xmath109 .
when @xmath108 is smooth , then , it is a plane curve of degree @xmath54 and genus @xmath110 .
we have an exact sequence @xmath111 from the vanishing of @xmath112 it follows that @xmath113 ( but this is also clear from duality ) , and @xmath114 @xmath115 suppose @xmath116 is a line .
a generic @xmath19 does nt contain any lines , so @xmath117 is a finite subscheme of length @xmath118 .
we claim that for a general plane @xmath119 containing @xmath120 , the intersection @xmath121 is smooth .
this holds by bertini s theorem away from the base locus of the linear system of planes passing through @xmath120 , so we just have to see that it also holds at a point @xmath122 .
note that @xmath123 is a one - dimensional subspace .
a general @xmath119 will have tangent space which is a general plane in @xmath124 containing @xmath125 .
thus , a general plane @xmath119 containing @xmath120 has tangent space @xmath126 which does nt contain @xmath127 ; in particular the intersection @xmath128 is transverse .
this implies that @xmath129 is smooth at @xmath130 .
this works for all the finitely many points @xmath122 , so the general section @xmath121 is smooth .
it is therefore a smooth plane curve of degree @xmath54 and genus @xmath110 .
notice that @xmath131 so @xmath132 . pick @xmath108 as in the previous paragraph ,
suppose @xmath133 is a finite subscheme of length @xmath134 , and suppose that @xmath135 is a section vanishing on @xmath136 .
then @xmath137 is a section of @xmath138 vanishing on @xmath139 . as @xmath108 is smooth ,
the finite subscheme @xmath136 is a cartier divisor .
the section @xmath137 corresponds to a map @xmath140 which , since it vanishes on @xmath136 , gives a map @xmath141 let @xmath142 be the divisor of zeros of @xmath58 , in particular @xmath143 , and @xmath58 extends to a strict map , i.e. an inclusion of a sub - vector bundle @xmath144 the quotient line bundle is @xmath145 where the notation here combines @xmath146 which is three times the hyperplane divisor ( which has degree @xmath54 on @xmath108 ) , with the divisor @xmath147 .
in particular @xmath145 is a line bundle of degree @xmath148 .
we obtain an exact sequence @xmath149 leading to the long exact sequence of cohomology
. this construction will be used many times in section [ baseloci ] .
another useful construction is the following .
write @xmath150 , possibly with some of the points being the same .
take a linear form containing @xmath151 as an isolated zero , and divide by the equation of @xmath120 .
this gives a meromorphic function whose polar divisor is @xmath152 .
equivalently , @xmath153 has a section nonvanishing at the points @xmath154 .
this will be used often without too much further notice below .
[ snhyp ] assume that @xmath18 , and that @xmath155 for @xmath156 .
recall that the second part is true for any @xmath21 general in its irreducible component as discussed above .
the goal of this section is to show that [ snhyp ] implies @xmath157 and @xmath21 has seminatural cohomology , which in this case means @xmath158 for @xmath159 , @xmath160 for @xmath161 , and @xmath162 for all @xmath16 .
our main theorem [ main ] is the statement that there is only a single irreducible component corresponding to such bundles , so hypothesis [ snhyp ] will be in effect throughout the rest of the paper .
[ sncase ] if @xmath18 then @xmath157 , in other words @xmath163 .
if @xmath164 has scheme of zeros @xmath63 , then saying @xmath18 is equivalent to requiring that @xmath165 , and saying that all @xmath166 is equivalent to requiring that @xmath167 .
as discussed above , @xmath168 so the fact that @xmath51 gives the first statement . for the second statement , use the fact that @xmath30 for all @xmath16 , and the long exact sequences of cohomology for the extension @xmath55 of @xmath169 by @xmath170 and similarly @xmath21 of @xmath171 by @xmath172 .
a first part of the seminatural condition is easy to see .
[ seminat02 ] under our hypothesis [ snhyp ] , @xmath158 for @xmath159 , and @xmath160 for @xmath161 .
since @xmath46 it follows that @xmath158 for all @xmath159 , and for @xmath161 , @xmath173 is dual to @xmath174 .
the main step towards the seminatural condition is the next twist : we also have @xmath175 . if @xmath121 is a smooth plane section , we claim @xmath176 . if not , then we would get an inclusion @xmath177 , hence @xmath178 .
however , @xmath108 is a curve of genus @xmath110 and @xmath179 so @xmath180 has dimension @xmath110 .
this gives @xmath181 .
consider the exact sequence @xmath182 the fact that @xmath48 implies that @xmath138 surjects onto @xmath183 , so @xmath184 .
this is a contradiction to @xmath18 , showing that @xmath176 . to show that @xmath175 , it suffices by duality to show that @xmath185 .
consider the exact sequence @xmath186 again by duality from lemma [ sncase ] , @xmath187 , so the long exact sequence gives an isomorphism between @xmath188 and @xmath189 . from the previous paragraph we obtain @xmath185 .
this proves the proposition .
[ seminatcor ] under hypothesis [ snhyp ] , @xmath21 has seminatural cohomology : @xmath162 for all @xmath16 . by duality it suffices to consider @xmath161 and
we have already done @xmath190 .
consider the case @xmath191 .
this could be done by continuing as in the previous proposition but here is another argument .
choose an inclusion @xmath164 , and let @xmath63 be the subscheme of zeros of @xmath58 .
choose a general plane section @xmath121 such that @xmath119 passes through one point @xmath192 in a general direction .
then @xmath137 has a zero at @xmath193 , of multiplicity @xmath194 with @xmath195 .
indeed , @xmath63 can not contain a @xmath110-fold fat point whose length is @xmath196 , because @xmath63 has length @xmath43 .
thus the multiplicity of a general plane section of @xmath63 at any point @xmath193 is @xmath197 .
the section @xmath58 restricted to @xmath108 therefore induces a strict inclusion of vector bundles from @xmath198 to @xmath55 , hence an exact sequence , of the form @xmath199 note that the sub - line bundle has degree @xmath200 and the quotient line bundle has degree @xmath201 , so both of these have vanishing @xmath202 by duality .
it follows that @xmath203 . for any @xmath204 a similar argument ( but @xmath108 does nt even need to pass through a point of @xmath63 )
shows that @xmath162 .
it follows that @xmath75 , which is to say that for any inclusion @xmath164 , if @xmath63 is the subscheme of zeros of @xmath58 then @xmath205 .
choose an inclusion @xmath58 and consider the exact sequence @xmath206 notice that @xmath207 .
the long exact sequence of cohomology then reads @xmath208 since @xmath209 and @xmath30 for all @xmath16 .
the previous conclusion says the term on the left @xmath210 vanishes , so @xmath211 .
it is generated by the nonzero extension class governing the exact sequence corresponding to @xmath58 .
on the other hand we have @xmath212 so the map @xmath213 has cokernel @xmath76 of dimension @xmath0 .
it follows that the kernel @xmath214 has dimension @xmath77 .
[ snunique ] pick a section @xmath215 and let @xmath63 be its subscheme of zeros .
the extension class defining @xmath21 as an extension of @xmath171 by @xmath172 is unique up to a scalar . recall from where @xmath216 was defined that the space of extensions , @xmath76 , has dimension @xmath217 .
thus , the condition @xmath75 means that this is a line : the extension is unique up to scalars , and for a given subscheme @xmath63 there is a unique bundle extension @xmath21 up to isomorphism .
[ h0eyzero ] if @xmath121 is a plane section , then @xmath218 .
also , @xmath219 and @xmath220 .
consider the exact sequence @xmath221 from @xmath46 and @xmath187 we get @xmath218 .
similarly , the exact sequence @xmath222 together with @xmath155 gives @xmath223 .
let @xmath224 be the subset of points where all sections of @xmath138 vanish .
let @xmath225 be the subset of points @xmath130 such that the image of @xmath226 has dimension @xmath227 ( in particular @xmath228 ) .
these are the _ base loci _ of sections of @xmath55 . in this section ,
we obtain some information about these base loci , which will allow us to to deduce , in section [ pstructure ] , that the zero - scheme of a general section @xmath58 has some fairly strong general position properties .
[ b2onepoint ] the subset @xmath229 has at most one point and if it exists , then the sections of @xmath138 define this reduced point as a subscheme .
suppose @xmath230 are two points of @xmath229 .
then all sections of @xmath55 vanish at @xmath231 and @xmath232 .
consider a plane section @xmath121 such that @xmath233 , but @xmath108 general for this property , in particular @xmath108 is smooth ( section [ planesection ] ) .
the map @xmath234 is injective .
furthermore it is surjective since @xmath220 .
let @xmath120 denote the line through @xmath231 and @xmath232 .
it intersects @xmath108 in a divisor denoted @xmath235 .
some of the points @xmath236 may be equal or equal to @xmath231 or @xmath232 .
we have an exact sequence @xmath237 and on the other hand , the exact sequence @xmath238 gives @xmath239 .
hence the image of @xmath240 has codimension @xmath54 , and since @xmath241 is a finite subscheme of length @xmath54 , @xmath242 so the image has dimension @xmath54 too .
we may impose the condition of vanishing at two points @xmath243 and obtain a nonzero section @xmath244 .
this has the required meaning when some of the points coincide , using the previous paragraph .
however , the section @xmath58 then does nt vanish at the third point @xmath151 , otherwise we would get a section in @xmath245 contradicting corollary [ h0eyzero ] .
this section generates a sub - line bundle @xmath246 , with @xmath247 for an effective divisor @xmath248 not passing through @xmath151 .
note that @xmath249 has a nonzero section , corresponding to the quotient of a linear form ( on the plane @xmath119 ) vanishing at @xmath151 but not along @xmath120 , divided by a linear form vanishing along @xmath120 .
if @xmath248 does nt contain both @xmath231 and @xmath232 then we would get a section of @xmath55 nonvanishing at one of those points , contradicting our assumption @xmath250 .
therefore @xmath251 .
it follows that @xmath252 .
the same reasoning works for @xmath253 and @xmath254 too , so @xmath236 are three points distinct from @xmath231 or @xmath232 .
our section @xmath58 comes from a section in @xmath138 corresponding to @xmath255 , and the subscheme of zeros @xmath63 contains @xmath256 .
these are four points on the line @xmath120 , so any cubic form vanishing at @xmath63 has to vanish along @xmath120 .
in particular , elements of @xmath257 vanish at @xmath151 .
this implies that elements of @xmath138 evaluate at @xmath151 to elements in the line @xmath258 .
thus @xmath259 .
this reasoning holds even if @xmath151 coincides say with @xmath254 ; it means that all elements of @xmath260 have to vanish in the tangent direction corresponding to the additional point @xmath151 glued onto @xmath254 , which still gives a rank one condition on the values of sections of @xmath55 at the point @xmath151 .
the same reasoning holds for @xmath253 and @xmath254 .
if at least two of the points @xmath236 are distinct , then we obtain this way at least two points of @xmath261 along the line @xmath120 .
then , vanishing at these two points consists of two conditions , so we can impose further vanishing at the third point ( even if it is a tangential point at one of the other two ) and obtain a nonzero section which vanishes at all five points . as before this yields a nonzero section of @xmath262 contradicting corollary [ h0eyzero ] .
it remains to consider the case when all three points are the same , that is to say @xmath263 with @xmath264 , and choosing a section vanishing at @xmath265 and two times at @xmath253 generates a subbundle @xmath266 with @xmath248 an effective divisor distinct from @xmath267 , and @xmath268 .
recall that if either @xmath269 or @xmath270 then this would give a section of @xmath55 nonvanishing at @xmath231 or @xmath232 contradicting our assumption @xmath250 .
as above , we have @xmath271 .
therefore , choosing a section in @xmath272 represents only @xmath273 conditions rather than @xmath134 , hence there are two linearly independent such sections @xmath274 .
we claim that the values of these two sections , in @xmath275 , are linearly independent . indeed ,
otherwise a combination of the two would vanish again at @xmath253 and this would give a section of @xmath276 which also vanishes at @xmath277 .
this would give a nonzero element of @xmath262 which ca nt happen .
let @xmath278 and @xmath279 be the smallest possible orders of vanishing of @xmath280 at @xmath231 and @xmath232 respectively , and by linear combinations we can assume that both of them vanish to those orders .
they give maps @xmath281 @xmath282 and the resulting map @xmath283 has image of rank @xmath4 at the point @xmath253 by the previous paragraph .
therefore it is injective .
it follows that @xmath284 .
suppose @xmath285 is the smaller of the two , then we get @xmath286 .
we may also by symmetry assume @xmath287 .
write @xmath288 and @xmath289 .
there are three possibilities : @xmath290 @xmath291 or @xmath292 in each case , let @xmath293 be the quotient bundle .
recall that @xmath294 and @xmath295 .
we have an exact sequence @xmath296 the rightmost map is dual to @xmath297 notice however that @xmath298 so @xmath299 .
hence our rightmost map is dual to @xmath300 this map is surjective ; indeed , the condition @xmath250 means that all sections of @xmath301 must vanish at @xmath231 and @xmath232 , and sections of @xmath302 are in particular sections of @xmath301 , so every element of @xmath303 must come from an element of @xmath304 .
this surjectivity translates by duality to the statement that the rightmost map in the above exact sequence , is injective .
it follows that @xmath305 is surjective . in other words ,
the values of global sections of @xmath306 at @xmath231 and @xmath232 span a two - dimensional space .
since on the other hand the values of sections of @xmath307 must vanish at @xmath231 and @xmath232 , this implies from the exact sequence @xmath308 that we have @xmath309 .
consider now the three cases , the first case being @xmath310 , with @xmath311 , so @xmath309 implies that @xmath312 .
vanishing at @xmath313 imposes two conditions which leaves @xmath314 .
these sections must vanish at @xmath231 and @xmath232 , so we get @xmath315 .
now , our two independent sections of @xmath316 can not vanish at both @xmath231 and @xmath232 because @xmath108 is not @xmath317 so there are no functions with a single nontrivial pole at @xmath318 .
we get a section of @xmath316 whose value at one of @xmath231 or @xmath232 is nonzero . multiplying this by the section of @xmath319 nonvanishing at @xmath231 and @xmath232 , gives a section of @xmath301 nonvanishing at @xmath231 or @xmath232 , a contradiction which treats the first case . in the next case ,
@xmath320 with @xmath321 so @xmath309 gives @xmath322 .
as usual , sections of @xmath301 have to vanish at @xmath231 and @xmath232 so @xmath323 . but notice that @xmath324 .
the map @xmath325 is an isomorphism ; and @xmath326 is generated by its global sections .
hence , vanishing of a section at @xmath253 imposes a nontrivial condition , giving @xmath327 .
this contradicts the previous estimation of @xmath328 .
this contradiction completes this case . in the last case ,
@xmath329 with @xmath330 so @xmath309 gives @xmath312 .
this is similar to the first case .
vanishing at @xmath313 imposes two conditions , and then the sections must further vanish at @xmath231 and @xmath232 , which leaves @xmath331 .
if we have a section here which is nonzero at either @xmath231 or @xmath232 , then multiplying it by the section of @xmath319 nonvanishing at @xmath231 , gives a section of @xmath301 nonvanishing at @xmath231 or @xmath232 , a contradiction . therefore , both sections in @xmath332 have to vanish further at @xmath231 and @xmath232 . this would give @xmath333 .
that can happen only if @xmath108 is a hyperelliptic curve . but a smooth plane curve of degree @xmath54 is never hyperelliptic . if @xmath108 were hyperelliptic , for a general @xmath334 let @xmath335 be the conjugate by the involution ; there is a meromorphic function @xmath89 with polar divisor @xmath336 .
the line @xmath120 through @xmath337 and @xmath335 meets @xmath108 in @xmath54 different points , otherwise the map sending @xmath120 to the point of higher multiplicity would be a @xmath338 . write @xmath339 .
a linear form vanishing at one of the other points , say @xmath151 , divided by the equation of @xmath120 , gives a meromorphic function @xmath88 whose polar locus is @xmath340 .
then @xmath341 $ ] provides a degree @xmath134 map to @xmath342 . by looking at the genus , it ca
nt be injective , also it spans the plane so it is nt a degree @xmath134 map to a line .
the only other case would be a degree @xmath4 map to a conic .
but in that case , a linear combination of @xmath89 and @xmath88 would have polar divisor @xmath343 .
doing the same for the other possibilities gives a @xmath273 dimensional space of functions with poles @xmath344 , but @xmath108 ca nt have a degree @xmath273 spanning map to @xmath342 .
this shows that @xmath108 can not be hyperelliptic , so this case is also ruled out .
we have now finished showing that it is impossible to have two distinct points @xmath250 .
the same proof works equally well if @xmath232 is infinitesimally near @xmath231 ; this double point defines a tangent direction , and @xmath120 should be chosen as the tangent line in this direction . the main case as before is when @xmath345 , and as before we get three cases : either @xmath346 , @xmath347 , or @xmath348 .
the main principle here is that sections of @xmath301 have to vanish on both @xmath231 and the nearby point @xmath232 , that is to say they have to vanish to order @xmath4 at @xmath231 . with this , the same proofs as above hold , so this shows that if @xmath229 is nonempty , then it is a single reduced point .
this completes the proof of the proposition . for the next discussion
, we assume that there is a point @xmath349 , unique by above .
consider the schematic structure of @xmath261 around this point @xmath350 .
an argument similar to the above , allows us to show that @xmath261 ca nt contain the third infinitesimal neighborhood of @xmath350 ; however , we havent been able to rule out the possibility that it might contain the second neighborhood
. we will formulate this statement precisely in the form of the following lemma , even though we havent really defined the schematic structure of @xmath261 .
recall that @xmath219 .
suppose @xmath351 is a general plane section passing through @xmath350 .
choose @xmath352 , vanishing to order @xmath0 at @xmath350 .
let @xmath353 be the sub - line bundle generated by @xmath58 .
then there exists a section @xmath354 such that the projection of @xmath355 as a section of @xmath356 vanishes to order at most @xmath4 at @xmath350 .
suppose on the contrary that all sections vanish to order @xmath328 in @xmath306 .
as this is true on a general @xmath108 , we may also specialize @xmath108 and it remains true .
in particular , choose a tangent line @xmath120 to @xmath19 at @xmath350 such that the second fundamental form vanishes .
choose a smooth plane section @xmath108 corresponding to a plane containing @xmath120 .
then @xmath357 is a divisor of class @xmath326 , and we can write @xmath358 where , as far as we know for now , @xmath253 and @xmath254 might be the same , and one or both might be equal to @xmath350 . choose a nonzero section @xmath352 which has only a simple zero at @xmath350 . recall that this is possible , by the result that @xmath229 is reduced in the previous proposition .
let @xmath246 be the subline bundle generated by @xmath58 , and let @xmath359 be the quotient .
the contrary hypothesis says that all sections of @xmath307 vanish to order @xmath273 at @xmath350 , when projected into @xmath306 .
this means that the condition of a section vanishing to order @xmath273 at @xmath350 imposes only two additional conditions .
indeed , etale - locally we can choose a basis for @xmath307 compatible with the subbundle @xmath301 , and impose two conditions stating that the first coordinate ( corresponding to @xmath301 ) vanishes to order @xmath273 ( it automatically vanishes to order @xmath0 already ) .
this implies that the section vanishes , since the second coordinate vanishes to order @xmath273 by hypothesis .
now since @xmath360 , we can impose two further conditions and obtain a section @xmath355 vanishing at @xmath253 .
the divisor of vanishing of @xmath355 is therefore @xmath361 where @xmath362 .
if @xmath363 then we would get a morphism @xmath364 nonzero at @xmath350 , but the line bundle @xmath365 has a section nonvanishing at @xmath350 , and this would contradict @xmath349 . therefore we can conclude that @xmath366 .
we now note that @xmath253 and @xmath254 must be distinct from @xmath350 .
for example , if @xmath367 , choose a section @xmath58 vanishing at @xmath253 , and as described above , we can assume vanishing to order @xmath273 at @xmath350 which imposes two additional conditions . if @xmath58 vanishes to order @xmath368 at @xmath350
this would give a section in @xmath262 which ca nt happen , so we can assume that @xmath369 and again this has a section nonvanishing at @xmath350 , contradicting @xmath349 .
so , this case ca nt happen .
similarly if @xmath370 , vanishing to order @xmath273 imposes two conditions , and vanishing to order @xmath134 imposes two more conditions so again there is a section @xmath58 which generates @xmath371 , but this @xmath301 has a section nonvanishing at @xmath350 contradicting @xmath349 . from these arguments
we conclude that @xmath243 are different from @xmath350 .
next , use the fact that a cubic polynomial on @xmath120 vanishing at @xmath134 points in @xmath117 , must also vanish on the fifth point .
suppose first that @xmath372 .
our section @xmath355 viewed as a section of @xmath55 defines a zero - scheme @xmath63 , which contains its zeros on @xmath108 .
in particular , @xmath63 contains the scheme @xmath373 on @xmath120 as well as the scheme @xmath253 .
note on the other hand that @xmath374 otherwise we would get a section in @xmath262 .
we conclude that any element of @xmath257 has to vanish at @xmath254 .
it follows that @xmath375 . by symmetry
, we get also @xmath271 .
now , vanishing of sections at @xmath253 and @xmath254 imposes @xmath4 conditions , and vanishing at @xmath376 imposes @xmath4 conditions as discussed above .
this gives a section in @xmath183 vanishing at all of @xmath357 , hence a nonzero section of @xmath262 .
we get a contradiction in this case .
to finish the proof of the lemma , we have to treat the case where @xmath377 , that is @xmath378 . as described previously , @xmath379 .
basically the same argument as before gives @xmath271 .
indeed , we can consider a section @xmath355 which vanishes at @xmath380 .
it ca nt have a zero of order @xmath4 at @xmath253 otherwise we would get @xmath381 .
let @xmath246 be the subline bundle generated by @xmath355 , and let @xmath306 be the quotient .
write @xmath382 with @xmath366 and @xmath248 disjoint from @xmath383 .
we may also consider @xmath355 as a section defined over @xmath19 , inducing a quotient morphism @xmath384 .
when restricted to @xmath108 this provides a morphism @xmath385 which is the same as the map to @xmath306 generically .
hence it must factor through @xmath386 .
this is more precisely given by @xmath387 .
sections of @xmath55 map to sections of @xmath388 vanishing on the zero locus @xmath63 of @xmath355 , which contains @xmath389 ( a subscheme of length @xmath134 ) .
these sections must vanish on all of @xmath120 , hence they vanish on @xmath313 .
thus the image of any section in @xmath306 has to be a section of @xmath390 .
this means that the sections of @xmath55 , evaluated at @xmath253 , must lie in @xmath391 . in other words , @xmath271 as claimed .
vanishing of a section at @xmath253 therefore imposes a single condition .
so there are two linearly independent sections @xmath392 which vanish at @xmath380 .
no nonzero linear combination of these can have a zero of order @xmath4 at @xmath253 .
it follows that the derivatives of @xmath393 and @xmath394 at @xmath253 are linearly independent .
let @xmath395 and @xmath396 denote the sub - line bundles of @xmath307 generated by the @xmath397 .
we have @xmath398 with @xmath399 and @xmath400 .
but the line bundle @xmath365 has a section nonvanishing at @xmath253 , so @xmath401 has a section nonvanishing at @xmath253 .
but the @xmath402 are generated by the derivatives of @xmath397 , which are linearly independent .
thus the @xmath403 generate @xmath404 .
but as there are sections of @xmath401 nonvanishing at @xmath253 , this contradicts @xmath271 .
this completes the proof of the lemma .
[ lengthtwoatb2 ] suppose @xmath349 .
then for a general section @xmath215 , the scheme of zeros of @xmath58 locally at @xmath350 is either the reduced point @xmath350 , or a length @xmath4 subscheme ( infinitesimal tangent vector ) at @xmath350 . from the proposition before ,
the sections of @xmath55 define @xmath350 as a reduced subscheme .
this means that for any tangent direction , there is at least one section whose derivative in that direction does nt vanish .
so , if @xmath351 is a generic curve through @xmath350 , then the zero scheme @xmath63 of a general section @xmath58 has @xmath405 being a reduced subscheme locally at @xmath350 .
it follows that @xmath63 is curvilinear at @xmath350 .
assume that the general @xmath63 has length @xmath328 locally at @xmath350 .
consider the two sections @xmath406 given by the previous lemma .
their zero sets are therefore curvilinear subschemes of length @xmath328 at @xmath350 . given @xmath58
, we may choose @xmath355 general , then the zero set of @xmath355 is transverse to that of @xmath58 at @xmath350 . for
otherwise this would mean that the tangent directions of the zero sets are always the same , but that would give an infinitesimal tangent vector in @xmath229 contradicting the above proposition .
so these curvilinear subschemes are transversal .
we may choose local coordinates at @xmath350 so that they go along the coordinate axes , up to order @xmath273 at least .
if @xmath407 are these coordinates with @xmath408 , we may write @xmath409 where @xmath278 and @xmath279 are sections of @xmath55 nonvanishing at @xmath350 .
furthermore , @xmath108 is transverse to @xmath410 .
we may assume that the sub - line bundle of @xmath307 generated by @xmath58 is generated by @xmath411 .
notice that if @xmath412 is linearly independent from @xmath413 then @xmath414 would be a section whose zero scheme is the reduced point @xmath350 so we would be done . therefore we may assume , after possibly multiplying by a scalar , that @xmath415 .
the conclusion of the lemma says that @xmath355 is not a section of this subline bundle to order @xmath273 , which means that @xmath416 is not a multiple of @xmath278 to order @xmath4 i.e. modulo quadratic terms .
we may therefore write @xmath417 with one of @xmath418 nonzero modulo @xmath413 .
look at the section @xmath419 for variable @xmath420 ; its leading term is @xmath421 . by our contrary hypothesis
, we suppose that the zero set of this section is curvilinear to order @xmath273 for all @xmath422 , which means that there is a factorization of @xmath419 as a multiple of a single section of @xmath55 , up to terms of order @xmath273 .
the first term has to be @xmath421 , so we can write @xmath423 this expands to @xmath424 or simplifying ( always modulo terms of order @xmath273 ) , @xmath425 now compare terms modulo the section generated by @xmath278 ; we get @xmath426 and from the @xmath427 term we get @xmath428 again modulo @xmath413 . putting these together
gives that @xmath429 modulo @xmath413 , for all @xmath422 .
this is possible only if @xmath430 and @xmath431 are multiples of @xmath413 ; but the conclusion of the previous lemma said that this was nt the case .
this contradiction completes the proof of the corollary .
the above discussion may seem somewhat complicated : let us explain the geometric picture , in terms of a schematic notion of the base locus @xmath261 .
the problem is that @xmath261 could have some `` layers '' surrounding the point @xmath349 .
locally , this would mean that the subsheaf of @xmath55 generated by global sections , looks like a rank @xmath0 subsheaf over @xmath261 , the layers of which would give a certain infinitesimal neighborhood of @xmath350 .
in the lemma , we say that if we cut by a general plane section @xmath108 going through @xmath350 , then the intersection with @xmath261 has length at most @xmath4 .
intuitively this means that while @xmath261 might have a single layer around @xmath350 , it ca nt have two layers .
notice that in some directions @xmath261 might be bigger , but in a general direction it has length @xmath4 .
then , in the corollary , we say that if the general section has a curvilinear zero set of length @xmath328 , that would mean that @xmath261 had to have at least two layers around @xmath350 .
let @xmath432 denote the hilbert scheme of subschemes @xmath65 which satisfy @xmath67 .
let @xmath433 denote the hilbert scheme of subschemes @xmath78 which satisfy @xmath67 .
we call these the _ cb - hilbert schemes_. let @xmath434 and @xmath435 denote the subschemes parametrizing @xmath63 such that @xmath436 and @xmath167 , and ( in the second case ) such that @xmath63 is contained in at least one smooth quintic surface . in that case ,
as we have seen in lemma [ sncase ] , any bundle @xmath21 extending @xmath171 by @xmath172 has seminatural cohomology , so we call them the _ seminatural cb - hilbert schemes_. furthermore , as in corollary [ snunique ] , the isomorphism class of @xmath21 is uniquely determined by @xmath63 . since @xmath21 is stable , it does nt have any nontrivial automorphisms .
[ dimensions ] the seminatural cb - hilbert scheme @xmath434 has pure dimension @xmath437 ; the seminatural cb - hilbert scheme @xmath435 has pure dimension @xmath79 .
denote by @xmath438 $ ] and @xmath439 $ ] the fiber bundles over these , parametrizing pairs @xmath440 where @xmath63 is a seminatural cb hilbert point , and @xmath441 is a @xmath4-dimensional subspace .
these have pure dimensions @xmath442 and @xmath443 respectively .
a point in @xmath434 corresponds to a choice of bundle @xmath21 in @xmath17 plus a section @xmath215 up to scalar .
as the moduli space has dimension @xmath43 and , for the seminatural case , @xmath444 has dimension @xmath134 , the total dimension of @xmath434 is @xmath437 .
the hilbert scheme of pairs @xmath445 with @xmath446 fibers over the @xmath447 dimensional space of quintics @xmath19 ( note that @xmath448 ) with @xmath437-dimensional fibers , so it has dimension @xmath449 . on the other hand , for a fixed @xmath450 ,
the space of quintics @xmath19 containing @xmath63 is @xmath451 .
notice that if @xmath63 is contained in at least one @xmath19 then the discussion of section [ seminatural ] implies that @xmath452 so @xmath453 and the space of quintics containing @xmath63 is an open subset of @xmath454 .
so , the dimension of the hilbert scheme @xmath435 is @xmath455 .
the fiber bundles parametrizing choices of @xmath441 are bundles of grassmanians of dimension @xmath134 , so they have dimensions @xmath442 and @xmath443 respectively .
all irreducible components have the same dimension because the same discussion works for all of them . [ b1dimzero ]
suppose @xmath21 is a general point of its irreducible component .
the subset @xmath261 of points at which @xmath55 is not generated by global sections , has dimension @xmath90 .
equivalently , if @xmath58 is a general section of @xmath55 and @xmath63 its subscheme of zeros , then the base locus in @xmath19 of the linear system of cubics @xmath257 , has dimension @xmath90 ( it remains possible that the base locus in @xmath64 could have dimension @xmath0 , indeed that will be a major case treated in section [ sec - ccc ] below ) .
the proof takes up the rest of this section , using three further lemmas .
note first the equivalence of the two formulations .
the section @xmath58 generates a rank one subsheaf of @xmath55 at all points outside @xmath63 .
thus , if @xmath261 had positive dimension , this would mean that all sections restrict to multiples of @xmath58 on @xmath261 , so all sections of @xmath257 would factor as a function vanishing on @xmath261 times some other function .
so the second statement implies the first . in the other direction ,
suppose all elements of the linear system factored as @xmath87 where @xmath88 is a fixed form , either linear or quadratic .
then the zero set of @xmath88 would provide a positive dimensional component of @xmath261 . to be proven ,
is that the elements of the linear system @xmath257 can not all share a common factor @xmath88 .
suppose to the contrary that they did , and let @xmath93 be the zero - set of @xmath88 .
it is a divisor either in the linear system @xmath29 or @xmath456 , which is to say that it is either a plane section or a conic section of @xmath19 .
let @xmath94 be the residual subscheme of @xmath63 along @xmath95 ( i.e. roughly speaking @xmath457 ) . recall that we are assuming that @xmath63 is not contained in a conic section , so @xmath458 and @xmath94 is nonempty .
the statement that elements of @xmath257 vanish along @xmath95 , means that the map @xmath459 is an isomorphism .
recall also that the right hand side has dimension @xmath134 in our situation , so we get @xmath460 too .
it is now easy to rule out the case where @xmath95 is a conic section .
indeed , in that case we would have @xmath461 , but @xmath462 and the space of sections generates @xmath29 everywhere , so there are at most @xmath273 sections vanishing on a nonempty subscheme @xmath94 giving a contradiction .
therefore , we may now say that @xmath95 is a plane section of @xmath19 . from above , @xmath463 .
next , we claim that @xmath94 satisfies @xmath99 , that is cayley - bacharach for @xmath388 which is the same as @xmath464 . indeed , if @xmath89 is a section of @xmath388 and @xmath88 is the equation of @xmath95 then @xmath87 is a section of @xmath66 .
suppose @xmath465 is a colength @xmath0 subscheme .
then it induces a colength @xmath0 subscheme @xmath68 such that @xmath466 is the residual of @xmath467 , notice that @xmath468 may be viewed as the ideal @xmath469 inside @xmath470 so an ideal of length @xmath0 in @xmath468 gives an ideal of length @xmath0 in @xmath470 .
now if @xmath89 vanishes on @xmath466 then @xmath87 vanishes on @xmath467 , so by @xmath67 for @xmath63 we get that @xmath87 vanishes on @xmath63 which in turn says that @xmath89 vanishes on @xmath94 .
this proves that @xmath94 satisfies @xmath99 as claimed .
it follows that @xmath94 also satisfies @xmath471 .
the next remark is that @xmath94 is not contained in a plane , for if it were then the union of this plane with the one defining @xmath95 would be a conic containing @xmath63 , contrary to our situation .
we have the following lemma , which is a preliminary version of the structural result of proposition [ structureofp ] below .
notice that here we havent yet shown that @xmath261 has dimension @xmath90 , so we use the specific current situation in the proof instead .
[ pdecomp ] in the situation of the proof of the present proposition , consider a general section @xmath472 , and let @xmath65 be its subscheme of zeros .
then @xmath63 decomposes as a disjoint union @xmath473 such that @xmath101 is reduced , and @xmath467 is either empty , consists of a point @xmath350 , or an infinitesimal tangent vector at @xmath350 , in the latter two cases @xmath350 is the unique point of @xmath229 .
choose first any section @xmath215 with zero - scheme @xmath63 , corresponding to a subsheaf @xmath474 .
let @xmath475 be another section linearly independent from @xmath58 , and let @xmath476 be the subsheaf generated by @xmath475 and @xmath58 .
let @xmath477 be the quotient .
let @xmath478 be the image of @xmath475 considered as a section of @xmath169 . under our hypothesis of the proof of the proposition
, the zero scheme @xmath479 decomposes as @xmath480 where @xmath248 is a conic section . for @xmath475 sufficiently general
, @xmath248 does nt contain @xmath95 ( otherwise the conic section @xmath481 would be a common zero of the linear system , and we have ruled that out ) .
thus , @xmath479 is smooth on the complement of a finite set .
now , we can choose @xmath355 so that it is nonvanishing at all isolated points of @xmath261 ( except maybe @xmath349 ) , and at the finite set of singularities .
thus , if the zero scheme @xmath482 of @xmath355 meets @xmath95 , it meets it at a point where @xmath479 is reduced ( which we think of heuristically , as points where @xmath261 is reduced even though we havent given a scheme structure to @xmath261 ) .
furthermore , since @xmath482 moves ( except maybe at @xmath349 ) , its intersection with @xmath95 is reduced . on the other hand , at points located on @xmath95 , @xmath482 has to be locally contained in @xmath95 , otherwise we could add a small multiple of @xmath475 to split off the part of the subscheme sticking out of @xmath95 .
these together imply that the points of @xmath482 contained in @xmath95 are reduced ( except possibly at @xmath229 ) .
the points outside of @xmath95 are reduced because they are located at places where @xmath55 is generated by global sections , again with the possible exception of @xmath349 . from the discussion of the above subsection , the local structure of @xmath482 near the possible single point @xmath349 is at most an infinitesimal tangent vector of length @xmath4 .
this gives the claim of the lemma .
the length of @xmath94 is @xmath483 .
consider first the case where @xmath229 is empty , or @xmath63 is reduced at @xmath349 , or else @xmath484 . in this case , @xmath485 with @xmath486 a reduced subscheme .
for fixed @xmath95 , the dimension of the space of choices of @xmath487 is @xmath488 . on the other hand , @xmath94 is located at the intersection @xmath489 where @xmath490 and @xmath491 are conics , whose intersection has dimension @xmath0 .
furthermore , no component of @xmath492 is contained in @xmath19 , indeed the former has degree @xmath134 while curves in @xmath19 have degre @xmath493 because of the condition @xmath494 .
therefore , @xmath489 is a finite set .
since @xmath94 is reduced except for a possible tangent vector at the unique point @xmath349 , we get that the dimension of the set of choices of @xmath94 for a given @xmath495 , is @xmath227 . on the other hand ,
suppose @xmath496 are three general conics through @xmath94 .
if their intersection is finite , it contains at most @xmath497 points ; but with @xmath498 this ca nt happen and we must have a nontrivial curve in the intersection ; this means that a double intersection @xmath492 has to split into two pieces .
the dimension of the space of such double intersections is the dimension of the grassmanian of @xmath4-planes in @xmath499 , this grassmanian has dimension @xmath77 .
however , as may be seen by a calculation of the possible cases of splitting , the subvariety of the grassmanian corresponding to double intersections which split into at least two components , is @xmath500 . together with the possible one dimensional choice of tangent vector at @xmath350
, we get altogether that the space of choices for @xmath94 together with the two - dimensional subspace spanned by @xmath501 , is @xmath502 . putting in @xmath487
, we get that the dimension of the space of choices of @xmath63 plus a @xmath4-dimensional subspace of @xmath257 , is less than @xmath503 .
the @xmath273 is for the space of choices of plane section @xmath95 . now if @xmath498 then @xmath504 and this dimension is @xmath505 .
the dimension of the corresponding bundle over the seminatural cb - hilbert scheme @xmath438 $ ] is @xmath442 ( proposition [ dimensions ] ) so such a bundle @xmath21 can not be general in its irreducible component .
we are left to treat the case where the unique point @xmath349 lies on @xmath95 , and @xmath63 includes a tangent vector here .
let @xmath506 denote the subscheme of @xmath63 located set - theoretically along @xmath95 , and @xmath507 the complement .
then the dimension of the space of choices of @xmath506 is still @xmath508 , and the same argument as above gives that the dimension of the space of choices of @xmath507 plus a two - dimensional subspace of conics , is @xmath502 .
we get as before @xmath509 . on the other hand
, the tangent vector at @xmath350 might go outside of @xmath95 and contribute to @xmath94 .
if the tangent vector stays inside @xmath95 then @xmath510 and we are done .
if the tangent vector goes outside of @xmath95 , then the estimate from above says only that @xmath511 ; however , we get an additional condition saying that the conics have to vanish at this point @xmath512 , and this condition ( which may be seen , for example , as a condition on the choice of @xmath506 once @xmath507 and the conics are fixed ) gets us back to the estimate @xmath513 .
there is a plane section @xmath514 such that @xmath515 has length @xmath493 .
suppose not , that is to say , suppose that any plane section meets @xmath94 in a subscheme of length @xmath516 . in order to obtain a contradiction
, we show that under this hypothesis , @xmath498 .
choose a plane section meeting @xmath94 in a subscheme of length @xmath328 , call this intersection @xmath517 and let @xmath518 denote the residual subscheme .
then the condition @xmath99 for @xmath94 implies @xmath471 for @xmath518 .
the results of our previous paper @xcite therefore apply : ( a ) @xmath519 ; ( b ) if @xmath520 or @xmath54 then @xmath518 is contained in a line ; ( c ) if @xmath521 or @xmath522 then @xmath518 is contained in a plane
. however , our hypothesis for the proof of the lemma says that no plane contains a subscheme of @xmath94 of length @xmath493 , so the cases @xmath523 , @xmath110 , @xmath522 ca nt happen . if @xmath524 then @xmath518 is contained in a line , and we can choose a plane which meets furthermore a point of @xmath517 , again giving a plane with more than @xmath54 points .
this shows that we must have @xmath525 and since @xmath526 we get @xmath498 as claimed ( under the hypothesis contrary to the lemma ) .
this contradicts the estimate of the previous lemma , which completes the proof of the present one .
now choose a plane section @xmath527 such that @xmath515 has maximal length .
write @xmath528 and let @xmath529 be the residual subscheme with respect to @xmath527
. then @xmath518 satisfies @xmath471 .
if @xmath530 then @xmath531 and by @xcite , @xmath518 must consist of @xmath134 or @xmath54 points on a line . choose a new plane section passing through this line but not meeting @xmath517 ; we conclude that @xmath517 must also consist of @xmath54 points on a line , but then in fact we could choose a plane section meeting @xmath94 in @xmath110 points . thus the case @xmath530 does nt happen . if @xmath532 then @xmath518 would consist of @xmath533 points , but there are no such subschemes satisfying @xmath471 , so this ca nt happen either .
we conclude that @xmath534 , hence @xmath518 must be @xmath134 points on a line .
if @xmath337 is any point of @xmath517 then there is a plane containing @xmath518 and @xmath337 , so the remaining points of @xmath517 are either on this same plane , or else contained in a line .
if the two lines meet at a point , this would give a plane section containing too many points .
hence , we conclude that there are @xmath4 skew lines containing at least @xmath497 of the @xmath535 points in @xmath94 . because of @xmath99 for @xmath94 , in fact all of the points must be on the two skew lines .
count now the dimension of the space of such configurations : there are @xmath497 parameters for the two skew lines .
once this configuration is fixed , the subscheme @xmath94 is specified up to a finite set of choices .
the choice of @xmath95 counts for @xmath273 , and the choice of @xmath535 points in @xmath95 counts for @xmath535 .
the full dimension of this space of choices is therefore @xmath536 . as in the proof of the previous lemma , the case where our double point at @xmath349 lies on @xmath95 but the tangent direction extends out of @xmath95 ,
does nt add an extra dimension because we get a point participating in @xmath94 which constrains the choice of points on @xmath95 .
in view of the fact that @xmath537 , this situation can not happen for a general @xmath21 .
this contradiction completes the proof of the proposition : for a general @xmath21 in its irreducible component , the base locus @xmath261 has dimension @xmath90 .
the previous results were as close as we could get to saying that @xmath55 is generated by global sections , with the techniques we could find . choose a general section @xmath215 and let @xmath63 denote its scheme of zeros .
if @xmath538 ( but not in @xmath229 ) then a general section will not vanish at @xmath337 .
furthermore , if there is a point @xmath350 in @xmath229 then the structure of @xmath63 near @xmath350 is at most an infinitesimal tangent vector .
[ structureofp ] suppose @xmath215 is a general element , and let @xmath63 be the subscheme of zeros
. we can write @xmath539 where @xmath467 consists of the possible point of @xmath63 located at @xmath229 , and @xmath101 is all the rest . with this notation ,
@xmath101 consists of @xmath540 , @xmath541 or @xmath43 isolated points , and @xmath467 is respectively an infinitesimal tangent vector at @xmath349 ; or the isolated point @xmath350 ; or empty . at any point @xmath542 , the map @xmath543 is surjective , meaning that @xmath337 is locally the complete intersection of two general sections of @xmath257 . by proposition [ b1dimzero ] ,
the base locus @xmath261 has dimension zero .
for any point @xmath544 with @xmath545 , a sufficiently general section @xmath215 is nonzero at @xmath193 .
therefore , for @xmath58 general the scheme of zeros @xmath63 does nt meet @xmath261 except possibly at @xmath229 .
divide @xmath63 into two pieces , @xmath467 at @xmath229 and @xmath101 which does nt meet either @xmath261 or @xmath229 . by proposition [ b2onepoint
] , the base locus @xmath229 consists of at most one point which we shall denote by @xmath350 if it exists . therefore , @xmath467 is either empty or has a single point . by corollary [ lengthtwoatb2 ] ,
the zero scheme of a general @xmath58 at @xmath350 has length at most two .
so , if @xmath467 has a point , then it is scheme - theoretically either this reduced point , or an infinitesimal tangent vector there . at a point @xmath542 , since @xmath337 is not in the base locus @xmath261 , it means that @xmath55 is generated by global sections at @xmath337 . from the standard exact sequence for @xmath58
we see that @xmath546 , so the generation of @xmath547 by global sections is exactly the surjectivity of the last claim in the proposition .
the points of @xmath101 are `` interchangeable '' .
this can be phrased using galois theory .
write @xmath548 $ ] .
put @xmath549 and let @xmath550 be the tautological point .
think of @xmath551 .
let @xmath552 be the subscheme of zeros .
the decomposition @xmath539 is canonical , hence defined over @xmath553 . on the other hand ,
@xmath101 consists of @xmath540 to @xmath43 points , but the points are only distinguishable over @xmath554 , which is to say @xmath555 is a set with @xmath540 , @xmath541 or @xmath43 points .
the galois group @xmath556 acts .
[ doublytransitive ] the action of @xmath556 on the set @xmath557 is doubly transitive : it means that any pair of points can be mapped to any other pair . for general @xmath58 , the part @xmath101
is contained in the open subset @xmath558 where @xmath55 is generated by global sections .
suppose @xmath559 and @xmath560 are two pairs of points in @xmath101 . consider a continuous path of pairs @xmath561 contained in @xmath562 defined for @xmath563\subset { { \mathbb r}}$ ] , with @xmath564 and @xmath565 .
vanishing of a section at @xmath566 and @xmath567 imposes @xmath134 conditions on elements of the @xmath54-dimensional space @xmath138 , so we get a family of sections @xmath568 leading to a family of subschemes @xmath569 . for general choice of path ,
the @xmath569 will all be reduced with @xmath540 , @xmath541 or @xmath43 points . at @xmath570 and @xmath571 ,
the section is the same as @xmath58 up to a scalar since it is uniquely determined by the vanishing conditions .
we obtain an element of the fundamental group of an open subset of the parameter space of sections @xmath58 , whose action on the covering determined by the points in @xmath101 , sends @xmath572 to @xmath573 .
this shows that the action is doubly transitive , and it is the same as the galois action after applying the grothendieck correspondence between galois theory and covering spaces .
[ singlecomponent ] suppose @xmath63 is the scheme of zeros of a general section @xmath215 , written @xmath539 as above .
let @xmath574 be the intersection of two cubic hypersurfaces corresponding to general elements of @xmath257 .
this @xmath575 is a complete intersection : @xmath576 .
there is a single irreducible component @xmath577 of @xmath575 such that @xmath101 is contained in the smooth locus of @xmath577 . at points of @xmath101 ,
@xmath577 is transverse to @xmath19 .
the only irreducible components of @xmath575 which can be non - reduced are those , other than @xmath577 , which are fixed as the cubic hypersurfaces vary .
the points of @xmath101 lie in the subset @xmath558 where sections generate @xmath55 . in the standard exact sequence , sections of @xmath55 map to sections of @xmath169 , and the fiber @xmath578 maps
to @xmath579 for @xmath580 . as sections generate the fiber , it implies that sections of @xmath169 generate the ideal @xmath581 ( which is the maximal ideal at @xmath130 ) .
two general sections therefore have linearly independent derivatives at @xmath580 when restricted to @xmath19 , so the same is true of the cubics in @xmath64 which means that @xmath575 is a transverse complete intersection at any @xmath580 .
the doubly transitive action from the previous proposition implies that for general sections and general choice of @xmath575 , the points of @xmath101 must all lie in the same irreducible component @xmath577 of @xmath575 .
note , on the other hand , that any other component of @xmath575 must also have dimension @xmath0 , otherwise we would get a @xmath0-dimensional base locus @xmath261 of sections of @xmath55 on @xmath19 and this possibility has been ruled out in proposition [ b1dimzero ] . note that @xmath577 is reduced since its smooth locus is nonempty .
if @xmath582 is a non - reduced component , then by sard s theorem it has to be a fixed part of the family of complete intersections of the form @xmath575 .
we need to know something about what curves @xmath575 can arise as the complete intersection of two cubics in @xmath64 .
the degree of @xmath575 is @xmath583 .
if @xmath575 is smooth , then @xmath584 is a line bundle of degree @xmath540 , so the genus of @xmath575 is @xmath535 .
the choice of @xmath575 corresponds to a choice of two - dimensional subspace @xmath585 ; furthermore @xmath586 and @xmath587 ( as may be seen from the exact sequences of restriction to one of the cubics @xmath588 and then from @xmath588 to @xmath575 ) .
the dimension of the grassmanian of @xmath4-dimensional subspaces of @xmath589 is @xmath590 .
denote this grassmanian by @xmath591 ; we have a universal family @xmath592 let @xmath593 denote the subset of @xmath594 defining a reduced complete intersection , i.e. such that the fiber @xmath595 of @xmath596 over @xmath594 has dimension @xmath0 and is reduced .
suppose @xmath597 is a decomposition with @xmath598 , @xmath599 , so @xmath600 and we may assume @xmath601 .
we are nt saying necessarily that the pieces @xmath602 and @xmath577 are irreducible , though .
this gives @xmath603 .
[ leq5inconic ] suppose @xmath575 is a complete intersection of two cubic hypersurfaces in @xmath64 .
if @xmath582 is a reduced irreducible component of @xmath575 of degree @xmath197 , then either @xmath582 is contained in a quadric , or the normalization of @xmath582 has genus @xmath604 or @xmath0 and the space of such curves has dimension @xmath605 .
if @xmath606 then it is contained in a quadric , so we may assume the degree is @xmath54 . let @xmath607 be the normalization and let @xmath88 denote the genus of @xmath108 .
projecting from a point on @xmath582 gives a presentation of @xmath108 as the normalization of a plane curve of degree @xmath134 , so it has genus @xmath608 .
the line bundle @xmath109 has degree @xmath535 which is therefore in the range @xmath609 , so @xmath610 . if @xmath611 then this is @xmath612 and the map @xmath613 is not injective , giving a quadric containing @xmath582 .
if @xmath604 , @xmath614 , and the embedding to @xmath64 corresponds to a map @xmath615 .
this yields @xmath437 parameters , minus @xmath0 for scalars , minus @xmath273 for @xmath616 , so there are @xmath43 parameters . if @xmath617 , the moduli space of elliptic curves provided with a line bundle @xmath326 of degree @xmath54 , has dimension @xmath0 ( the line bundles are all equivalent via translations ) . here
@xmath618 so the space of parameters for the embedding has dimension @xmath541 , this gives a @xmath43 dimensional space altogether .
[ 63 ] let @xmath619 denote the locally closed subset of @xmath593 parametrizing complete intersections @xmath595 such that @xmath620 with @xmath577 irreducible of degree @xmath110 .
the degree @xmath273 piece @xmath602 is allowed to have other irreducible components .
then @xmath619 is the union of four irreducible components parametrizing : ( a ) the case where @xmath602 is a rational normal space cubic ; ( b ) the case where @xmath602 is a plane cubic ; ( c ) the case where @xmath602 is a disjoint union of a plane conic and a line ; and ( d ) the case where @xmath602 is a disjoint union of three lines .
these have dimensions @xmath442 , @xmath621 , @xmath622 and @xmath437 respectively . for ( a )
there is an open set on which @xmath623 with @xmath602 and @xmath577 being smooth and meeting transversally in @xmath497 points .
we divide into cases corresponding to the piece @xmath602 obtained by removing the degree @xmath110 irreducible component @xmath577 . in the first case ( a )
, we include all degree @xmath273 curves @xmath602 which are connected chains of rational curves with no loops or self - intersections , which then have to span @xmath64 . in these cases , @xmath624 is always @xmath134-dimensional , and @xmath602 deforms to a smooth rational normal space cubic . for a given @xmath602 ,
the space of choices of @xmath594 is the grassmanian of @xmath4-planes in @xmath625 and @xmath626 has dimension @xmath535 ; one can check ( case by case ) that the restriction map from @xmath627 is surjective , so @xmath628 and the grassmanian has dimension @xmath77 .
the space of choices of rational normal space cubic is irreducible , equal to the space of choices of basis for the @xmath134-dimensional space @xmath629 ( @xmath630 ) , modulo scalars ( @xmath631 ) and the automorphisms of the rational curve @xmath602 ( @xmath632 ) . in the case of a chain
the dimension of the automorphism group goes up so those pieces are of smaller dimension in the closure of the open set where @xmath602 is smooth .
the dimension of this component is therefore @xmath633 a general point corresponds to a smooth @xmath602 with general choice of @xmath594 yielding a smooth curve @xmath577 of degree @xmath110 meeting @xmath602 at @xmath497 points .
the remaining possibilities are ( b ) , ( c ) and ( d ) , which are irreducible and one counts the dimensions as : ( b ) a plane @xmath632 plus a cubic @xmath634 plus a subspace of @xmath635 , @xmath636 for a total of @xmath621 ; ( c ) a plane @xmath632 plus a conic @xmath637 plus a disjoint line @xmath638 plus a subspace of @xmath639 , @xmath640 for a total of @xmath622 ; ( d ) three lines @xmath641 plus a subspace of @xmath642 , @xmath641 for a total of @xmath437 . [ 72 ] let @xmath643 denote the locally closed subset of @xmath593 parametrizing complete intersections @xmath595 such that @xmath620 with @xmath577 irreducible of degree @xmath522 .
the degree @xmath4 piece @xmath602 is allowed to have other irreducible components .
then @xmath643 is the union of two irreducible components parametrizing : ( a ) the case where @xmath602 is a plane conic ; and ( b ) the case where @xmath602 is a disjoint union of two lines .
these have dimensions @xmath621 and @xmath442 respectively . for ( a ) there is an open set on which @xmath623 with @xmath602 and @xmath577 being smooth and meeting transversally in @xmath110 points .
the complementary curve @xmath602 has degree @xmath4 .
if irreducible , it has to be a plane conic .
if reducible , it is the union of two lines .
if the lines meet , this still corresponds to ( a ) , if they are disjoint it is case ( b ) .
both cases have irreducible spaces of parameters .
to count the dimensions , in case ( a ) the choice of plane @xmath644 is @xmath632 , the choice of conic in the plane is @xmath637 , and @xmath645 .
one can check that the restriction map is surjective ( since @xmath602 is reduced there are only two cases , an irreducible conic or two crossing lines ) ; so @xmath646 and the grassmanian of @xmath4-planes in here has dimension @xmath647 .
the total dimension is therefore @xmath648 . in case
( b ) the choice of two disjoint lines is @xmath497 dimensional , and @xmath649 ; the grassmanian of @xmath4-planes has dimension @xmath43 so the total dimension here is @xmath442 .
[ 81 ] the subvariety @xmath650 parametrizing @xmath620 with @xmath577 irreducible of degree @xmath497 , is irreducible of dimension @xmath651 .
it has an open set on which @xmath577 is smooth and meets the line @xmath602 transversally in @xmath134 points .
note that @xmath602 has to be a line .
the space of lines has dimension @xmath134 and @xmath652 .
the grassmanian of @xmath4-planes @xmath594 in @xmath625 has dimension @xmath442 , so the total dimension is @xmath651 .
the general element is contained in a smooth cubic surface , on which the relevant linear system defining @xmath577 has no base points so a general @xmath577 is smooth ; the intersection @xmath653 has @xmath134 points by the adjunction formula .
consider a general bundle @xmath21 in its irreducible component , a general section @xmath215 , and a general two - dimensional subspace @xmath654 .
let @xmath574 be the intersection defined by @xmath594 , which is a complete intersection by corollary [ singlecomponent ] .
write @xmath539 as usual and let @xmath655 be the irreducible component containing @xmath101 .
let @xmath656 be the intersection of the four independent cubics spanning @xmath657 .
it is contained in @xmath575 , indeed it is the intersection of @xmath575 with the other two cubics spanning the complement of @xmath658 .
hence @xmath659 , and also of course @xmath660 .
write @xmath661 where @xmath662 is the union of @xmath0-dimensional pieces of @xmath136 and @xmath663 is the remaining @xmath90-dimensional part .
notice that @xmath664 and @xmath665 correspond to galois invariant pieces in the situation where @xmath58 is a generic geometric point , so by proposition [ doublytransitive ] it follows that if @xmath666 is nonempty then @xmath667 and similarly for @xmath663 .
our situation therefore breaks down into two distinct cases : the _ common curve case _ when the @xmath0-dimensional part @xmath662 contains the big variable part @xmath101 ; or the _ variable curve case _ when @xmath668 . in this section
, we would like to rule out the first possibility ; reasoning by contradiction suppose on the contrary that we are in the common curve case . since @xmath669 , and by corollary [ singlecomponent ] there is a single irreducible component @xmath577 of @xmath575 containing @xmath101
, it follows that @xmath670 .
the common curve case is therefore equivalent to the following hypothesis , which will be in vigour throughout the section until it is ruled out .
[ ccc ] all of the sections in @xmath657 vanish along @xmath577 .
[ containedquartic ] suppose that @xmath671 .
then @xmath577 is contained in a quadric , from which it follows that @xmath63 is contained in a quadric .
if we can show that @xmath577 is contained in a quadric , then it follows that @xmath63 is contained in the same quadric by lemma [ inquadric ] .
suppose @xmath672 .
then by lemma [ leq5inconic ] , either @xmath577 is contained in a quadric or it runs in a space of dimension @xmath605 . in the latter case , for each choice of @xmath577 we have a space of possible choices of @xmath63 of dimension @xmath43 , @xmath673 , or @xmath674 depending on whether @xmath467 is empty , a single point , or an infinitesimal tangent vector . in all cases ,
this results in a space of possible subschemes @xmath63 of dimension @xmath675 , so by proposition [ dimensions ] it ca nt contribute to a general point in the irreducible component .
suppose @xmath676 ( and of course @xmath655 so @xmath677 )
. choose a hyperplane @xmath644 not passing through a point of @xmath101 , and let @xmath678 be the intersection .
it is finite of length @xmath318 . using hypothesis [ ccc ]
, we get a @xmath134-dimensional space @xmath95 of sections of @xmath679 ( one can note that , by the same argument as lemma [ inquadric ] , sections of @xmath680 vanishing on @xmath577 vanish also on @xmath63 so @xmath681 ) . consider the restriction map @xmath682 if @xmath151 lies in the kernel , it means that @xmath151 factors as the linear form defining @xmath119 times a quadric .
then this quadric contains @xmath577 , so it contains @xmath63 by lemma [ inquadric ] .
we now show that @xmath475 is not injective .
[ cubicconditions ] suppose @xmath683 is a subscheme of length @xmath522 .
if @xmath684 does nt contain @xmath54 points on a line ( i.e. the intersection with any line has length @xmath516 ) then @xmath684 imposes @xmath522 independent conditions on @xmath685 .
choose a line @xmath120 with maximal value of the length @xmath91 of @xmath686 . then @xmath687 . on the @xmath535 dimensional space of cubics we can try to impose @xmath273 further conditions and
should prove that this makes sections vanish .
imposing @xmath90,@xmath0 or @xmath4 additional conditions on cubics restricted to @xmath120 makes them vanish .
the residual subscheme @xmath688 of @xmath684 with respect to @xmath120 has length @xmath689 , and we have to show that it imposes this number of conditions on conics . again choosing a line @xmath690 with maximal contact ( of order @xmath691 ) with @xmath688 , imposing @xmath90 or @xmath0 additional conditions we get vanishing of the conics on @xmath690 ; we are left with a further residual subscheme @xmath692 of length @xmath693 , which is between @xmath90 and @xmath273 ; however if @xmath692 consisted of @xmath273 colinear points that would imply @xmath694 so @xmath692 would have length @xmath102 and this is ruled out . hence @xmath692 imposes independent conditions on linear sections .
we conclude that @xmath684 imposed @xmath522 independent conditions on cubics .
to finish the proof of lemma [ containedquartic ] , consider the subscheme @xmath684 from that proof .
it has length @xmath522 , @xmath497 or @xmath583 . since @xmath119 was general , no @xmath54 points of @xmath684 lie on a line .
applying the previous lemma to a subscheme of length @xmath522 , we see that @xmath684 has to impose at least @xmath522 conditions on cubics . but
@xmath695 is a subspace of the @xmath535-dimensional @xmath685 , vanishing on @xmath684 .
thus @xmath696 showing that @xmath475 ca nt be injective .
this completes the proof . to finish this section
, we just have to consider the case when @xmath697 .
consider first the case where a general @xmath575 is reduced , and apply lemma [ 63 ] .
notice that for each choice of @xmath575 a reduced complete intersection @xmath698 , @xmath699 , the space of possible choices of @xmath63 has dimension @xmath605 . from @xmath700
this is clear when @xmath63 is reduced .
the other possibility is that @xmath467 is an infinitesimal tangent vector . in that case
, @xmath101 is to be chosen in the smooth subset of @xmath577 , giving an @xmath540 dimensional space of choices .
when @xmath467 is in the smooth part of @xmath602 it really only corresponds to a @xmath0 dimensional space of choices , giving @xmath541 in all ; when @xmath467 is in the singular set of @xmath575 , the choice of @xmath350 is @xmath90-dimensional and the choice of tangent vector @xmath102-dimensional , so we get a space of choices of dimension @xmath605 in all . from lemma [ 63 ] ,
the space of choices of pairs @xmath701 in case ( a ) has dimension @xmath702 .
this is the same as the dimension of the component of the hilbert scheme we are looking at .
however , a general pair @xmath703 with @xmath698 in the @xmath442 dimensional piece @xmath704 and @xmath700 general , does nt occur .
indeed , the degree @xmath110 piece @xmath577 is a smooth curve of genus @xmath273 so there are @xmath647 sections of @xmath705 on @xmath577 , and imposing up to @xmath43 conditions ca nt make the sections vanish there ; on the other hand , we could start by imposing up to @xmath4 independent conditions from @xmath467 on the degree @xmath273 piece @xmath602 . thus , a general choice of @xmath700 imposes @xmath43 conditions on the @xmath706-dimensional space @xmath707 , leaving only @xmath522 sections to add to @xmath708 giving @xmath709 .
so , for a general choice of @xmath700 we have @xmath710 and @xmath63 ca nt satisfy @xmath67 .
hence the space of @xmath701 such that @xmath575 decomposes with a degree @xmath110 piece @xmath577 , is a proper subspace of our irreducible component so for general bundles @xmath21 this case does nt occur . in case
( b ) of lemma [ 63 ] , consider the plane @xmath119 containing @xmath602 .
the subspace @xmath594 of cubics vanishing on @xmath575 has dimension @xmath4 , whereas @xmath711 has dimension @xmath0 ( the plane cubic @xmath602 determines its equation uniquely up to a scalar ) .
therefore the restriction map from @xmath594 to @xmath712 is not injective ; but an element @xmath713 mapping to zero on @xmath119 must be a product of a quadric and the linear equation of @xmath119 .
this gives a quadric containing @xmath577 and hence @xmath63 .
so , for bundles with @xmath59 , this case does nt occur . in cases ( c ) and ( d ) of lemma
[ 63 ] , the total dimension is @xmath714 which is too small , so these do nt contribute for general bundles @xmath21 .
this completes the analysis of the case where a general @xmath575 is reduced .
if a general @xmath575 is non - reduced , the non - reduced components @xmath582 must be fixed , but different from @xmath577 . as @xmath577 is also fixed ( when we vary the two - dimensional subspace @xmath594 of cubics ) , there must be at least one variable component @xmath715 .
the degree of the complementary piece to @xmath577 is @xmath273 so the only possibility is a fixed line of multiplicity two , and a variable line of multiplicity @xmath0 .
but then we have a @xmath134 dimensional space of cubics passing through the degree @xmath497 curve @xmath716 , so as in the proof of lemma [ containedquartic ] , this would give a quartic containing @xmath717 .
we have finished the proof of the following theorem ruling out the common curve case .
[ variablecurve ] hypothesis [ ccc ] leads to a contradiction .
therefore , for a general seminatural bundle @xmath21 in its irreducible component and a general section @xmath215 defining a scheme of zeros @xmath63 , if the intersection of the four cubics passing through @xmath63 has a @xmath0-dimensional piece @xmath662 , then the big interchangeable collection of points @xmath100 does nt meet @xmath662 .
in other words , we are in the variable curve case .
the common curve case is ruled out by the previous section .
hence we are in the variable curve case , when @xmath718 .
it means that the choice of @xmath95 , which determines @xmath136 , then determines @xmath101 and hence almost @xmath467 ( note however that @xmath467 could still be in a @xmath0-dimensional piece of @xmath662 ) .
let @xmath658 be a general @xmath4-dimensional subspace determining a complete intersection @xmath698 .
in this section , we consider the case when @xmath575 is not irreducible , a possibility which we would like to rule out .
as was argued before , the points of @xmath101 are indistinguishable under the galois group ; the subspace @xmath594 may be chosen defined over the same field as @xmath63 , so @xmath101 must be contained in the smooth points of a single irreducible component @xmath577 of @xmath575 . write @xmath597 where the remaining piece @xmath602 is allowed to be reducible .
applying lemmas [ leq5inconic ] ( as in the first paragraph of the proof of [ containedquartic ] ) and [ inquadric ] as well as the hypothesis @xmath59 so @xmath63 is not contained in a quadric , gives that @xmath719 .
the idea is to use a dimension count .
the dimensions of the cases go all the way up to @xmath720
. however , the subspace @xmath95 determines @xmath101 , and in turn @xmath95 is determined by a smaller subset of points than @xmath63 , so the dimension count can still work .
choose a subscheme @xmath721 of length @xmath77 as follows : start with @xmath722 of length @xmath4 containing @xmath467 .
note that @xmath722 imposes @xmath4 independent conditions on @xmath627 . then for @xmath723 let @xmath724 with @xmath725 chosen in @xmath101 such that it imposes a nontrivial condition on @xmath726 .
this exists because @xmath727 for @xmath728 we get @xmath729 imposing @xmath77 independent conditions , and @xmath730 .
it follows that @xmath731 in particular , @xmath95 is determined by @xmath729 .
however , the remaining four points of @xmath732 are all in @xmath101 , in particular they are reduced points .
because of the `` variable curve case '' theorem [ variablecurve ] , the intersection @xmath136 of the cubics in @xmath95 has dimension @xmath90 at the points of @xmath101 ; therefore , the locations of the remaining four points are determined ( up to a finite choice ) by @xmath95 .
we get that @xmath63 is determined by @xmath729 .
we may now count the dimension of the space of choices of pair @xmath701 where @xmath698 for a general subspace @xmath658 .
the space of choices of @xmath575 containing a degree @xmath110 or degree @xmath522 piece , is @xmath733 .
the dimension of the space of choices of @xmath729 inside @xmath575 is @xmath734 if we assume @xmath575 reduced , or @xmath735 in any case , so the total dimension there is @xmath736 which is too small . for the case of @xmath575 containing a piece @xmath577 of degree @xmath497
, we get a dimension of @xmath737 so this looks possible
. however , the general element @xmath575 of the parameter space corresponds to the union of a smooth degree @xmath497 curve @xmath577 meeting a line @xmath602 in @xmath134 points . in order to get to dimension @xmath443
, we must have @xmath63 general , in particular @xmath101 is a general collection of @xmath540 points in @xmath577 .
now @xmath577 has genus @xmath522 .
the line bundle @xmath738 is a general one of degree @xmath110 , which on a curve of genus @xmath522 will not have any sections .
hence , all cubics containing @xmath63 must vanish on @xmath577 , which would put us back into the `` constant curve case '' .
so , this case does nt occur .
we have finished ruling out the possibility that @xmath575 would be reducible , resulting in the following theorem .
[ zirred ] for a general seminatural bundle @xmath21 in its irreducible component and a general section @xmath215 defining a scheme of zeros @xmath63 , choose a general @xmath4-dimensional subspace @xmath739 defining a complete intersection @xmath595 .
then @xmath595 is irreducible .
in this section we complete the proof that the hilbert scheme @xmath435 is irreducible , by treating the case @xmath740 where @xmath595 is an irreducible complete intersection of degree @xmath583 .
we first indicate how to construct an open set of the hilbert scheme .
consider a smooth complete intersection curve @xmath595 for a general @xmath4-dimensional subspace @xmath741 .
the grassmanian of choices of @xmath594 has dimension @xmath742 and there is a dense open set where @xmath698 is smooth of genus @xmath535 .
now @xmath700 will be a subscheme of length @xmath43 , which is a cartier divisor since @xmath575 is smooth . by varying any collection of @xmath535 points ,
we obtain a family which surjects to the jacobian @xmath743 .
the line bundle @xmath744 has degree @xmath745 .
note that the map @xmath746 is surjective , with kernel of dimension @xmath497 .
hence , in order to obtain @xmath205 one should ask for @xmath747 that is , @xmath748 . as @xmath749
we get @xmath750 .
the condition @xmath748 is therefore equivalent to @xmath751 or by duality , @xmath752 .
now , @xmath753 has degree @xmath540 , so @xmath754 is a line bundle of degree @xmath4 .
asking for it to have a section is equivalent to asking that @xmath755 for a degree @xmath4 effective divisor @xmath756 .
the dimension of choices of @xmath301 is @xmath4 and the space of choices is irreducible . for each choice of @xmath301
, we have @xmath757 , and the space of choices of divisor @xmath63 such that @xmath758 is a projective space of dimension @xmath759 putting these together , we get an irreducible @xmath760 dimensional space of choices of @xmath700 such that @xmath205 . including the variation of @xmath575 in a @xmath742 dimensional space , these fit together to form an irreducible @xmath443 dimensional variety .
if we replace @xmath63 by a subscheme @xmath761 of colength @xmath0 in the above argument , then @xmath301 changes to @xmath762 where @xmath763 .
as this is a general point of @xmath575 , we still have @xmath764 giving the cayley - bacharach condition @xmath67 for @xmath63 .
hence , there is a dense open subset of the @xmath443 dimensional variety parametrizing pairs @xmath703 where @xmath63 satisfies @xmath67 .
this is our irreducible component of @xmath439 $ ] .
abstracting out the choice of @xmath575 gives an irreducible @xmath79-dimensional component of the hilbert scheme @xmath435 . [ only ] the irreducible component constructed above is the only one in @xmath435 .
the argument above shows the basic idea .
however , we need to do some more work to treat the case when @xmath575 is singular and specially the possibility of a point or infinitesimal tangent vector in @xmath467 .
the first step is to rule out this last possibility .
a general @xmath63 in its irreducible component is reduced .
given @xmath63 we can choose a quintic surface @xmath19 containing it , and write @xmath539 .
we have @xmath101 reduced and if @xmath467 is non - reduced , it consists of a single infinitesimal tangent vector .
furthermore we may assume that @xmath63 is at a smooth point of its hilbert scheme .
choose a local smoothing infinitesimal deformation of @xmath467 ; we would like to extend that to a deformation of @xmath63 preserving the @xmath67 condition . as the cayley - bacharach property is open
, it is equivalent to preserving the property @xmath205 .
one can check that the obstruction to finding a deformation of @xmath101 which , when added to the given deformation of @xmath467 , preserves @xmath765 , would be the existence of a section @xmath766 such that @xmath355 vanishes to order @xmath4 at all the points of @xmath101 in @xmath64 . consider a complete intersection of cubics @xmath575 containing @xmath63 , and we may assume that @xmath101 lies on the smooth locus of @xmath575 . from the results of the previous sections , we may assume that @xmath575 is an irreducible curve of degree @xmath583 .
hence @xmath767 is a line bundle of degree @xmath742 .
our section @xmath355 vanishes at @xmath768 , but also at the points of @xmath467 .
together these are at least @xmath769 points , so it follows that @xmath355 vanishes on @xmath575 .
let @xmath770 be one of the cubics defining @xmath575 .
the residual of the scheme @xmath771 of multiplicity @xmath4 at @xmath101 , intersected with @xmath588 , consists of all the points of @xmath101 .
the restriction @xmath772 , divided by the other equation of @xmath575 , corresponds to a linear section vanishing at these points ; but the points of @xmath101 are not all contained in a plane ( indeed they are not even contained in a quadric ) , so @xmath773 .
then @xmath355 divided by the equation of @xmath588 is a linear form again vanishing on @xmath101 , so it is zero .
thus , @xmath570 .
this proves that the obstruction to lifting our smoothing deformation of @xmath467 to a deformation of @xmath63 , vanishes .
therefore , for a general point @xmath63 the piece @xmath467 has to consist of at most a single reduced point .
this proves the lemma .
suppose next that @xmath63 is a cartier divisor on @xmath575
. this will always be the case at points of @xmath101 which are smooth points of @xmath575 , but it remains a possibility that @xmath467 is a non - movable point at a singularity of @xmath575 .
we will deal with this problem below , but for now in the interest of better explaining the argument , assume that @xmath774 is a line bundle which we may think of as being a restriction from a small analytic neighborhood of @xmath575 .
now @xmath575 is a complete intersection , so duality still applies .
this can be seen , for example , by using serre duality on @xmath64 and the equations for @xmath575 which provide resolutions for @xmath775 ; the local @xmath776 sheaves may be tensored with @xmath120 which exists on a neighborhood of @xmath575 .
we get @xmath777 applying this to @xmath778 , we get @xmath779 on the other hand , @xmath780 and as before , @xmath781 , so the condition @xmath205 is equivalent to @xmath782 , which in turn is equivalent to asking that the degree @xmath4 line bundle @xmath783 be effective .
the picard scheme @xmath784 is still a group scheme , hence smooth ; and its tangent space at the origin is @xmath785 .
the exact sequences for @xmath786 ( where @xmath588 is one of the cubics cutting out @xmath575 ) give @xmath787 which is @xmath535-dimensional .
so the group scheme , as well as its torsors @xmath788 are @xmath535 dimensional .
an infinitesimal argument with exact sequences also shows that for @xmath535 general points in @xmath575 , the map from the product of their tangent spaces to the picard scheme is surjective .
as @xmath63 consists of @xmath43 points , and the picard scheme has dimension @xmath535 , at least @xmath535 points can move generally , keeping the same divisor @xmath63 .
the effective divisors form a two dimensional subscheme of @xmath789 . thus , at a general @xmath700 satisfying @xmath205 , the hilbert scheme of such @xmath63 has dimension @xmath760 .
the locus of singular @xmath575 has dimension @xmath790 , so the pairs @xmath703 with @xmath575 singular lie on a subscheme of dimension @xmath736 , and can not therefore correspond to a general bundle @xmath21 in its irreducible component .
this finishes the proof of theorem [ only ] in the case where @xmath63 corresponds to a cartier divisor .
some further argument is needed for the general case .
the reader may calculate directly that the dimension of the space of @xmath703 such that @xmath575 is a nodal curve and @xmath63 contains a point @xmath350 located at the node , is @xmath791 and does nt contribute .
this indicates that we do nt get a new irreducible component in this way . to give a more complete argument , consider @xmath703 with @xmath575 singular ( but still reduced and irreducible ) and @xmath63 including a point @xmath792 located at a singular point of @xmath575 .
consider general hyperplanes @xmath105 passing through @xmath350 , let @xmath793 ( meaning the local piece of @xmath794 at @xmath350 ) and let @xmath795 .
this is now a cartier divisor on @xmath575 so the previous considerations apply .
let @xmath91 denote the length of @xmath553 .
the condition that @xmath575 is not contained in a plane means that the general intersection @xmath794 ca nt be concentrated at a single point , on the other hand @xmath350 is singular in @xmath575 , so @xmath796 .
the exact sequences for complete intersections imply that @xmath553 imposes @xmath91 independent conditions on cubics .
our point @xmath350 is in the base locus @xmath229 for the bundle @xmath21 , meaning that sections in @xmath797 vanish to order @xmath798 at @xmath350 in @xmath19 .
this is true for any general quintic @xmath19 passing through @xmath63 , so sections of @xmath799 vanish to order @xmath798 at @xmath350 in @xmath64 . in particular , @xmath575 contains the multiplicity two fat point at @xmath350 . in turn
, this implies that @xmath553 contains the multiplicity two fat point at @xmath350 in @xmath119 .
we have @xmath800 , which translates , using duality and calculating the euler characteristic , into @xmath801 .
that is to say , @xmath802 is an effective line bundle of degree @xmath803 ( the case @xmath804 would correspond to the case treated previously ) .
the dimension of the space of choices of @xmath805 satisfying this effectivity condition , at general @xmath101 in its linear system , is @xmath806 .
note that since @xmath63 is reduced , @xmath805 determines @xmath63 . for a given @xmath807 ,
the space of choices of @xmath575 passing through @xmath553 is the grassmanian of @xmath4-planes in @xmath808 , so it has dimension @xmath809 .
we consider the space of choices of @xmath810 .
the choices of @xmath811 form a @xmath54 dimensional space .
let @xmath812 denote the dimension of the space of choices of @xmath807 located at a given point @xmath350 .
then altogether , the space of choices of @xmath810 has dimension @xmath813 this should be compared with the dimension of the hilbert scheme , plus the number of choices of @xmath119 ( @xmath4-dimensional ) for each @xmath63 , which is to say @xmath814 .
the dimension count is now taken care of by noting that @xmath807 contains the fat point of multiplicity @xmath4 at @xmath350 and this part is fixed without parameters .
the remaining parameters for the choice of @xmath553 therefore correspond to the length of the remaining subscheme , which is to say @xmath815 .
this gives a count of @xmath816 for the space of @xmath810 corresponding to the singular situation , which is @xmath817 so it does nt contribute to the general points of the hilbert scheme of @xmath703 .
this completes the proof of theorem [ only ] .
to complete the proof of theorem [ main ] , we should go back from the hilbert scheme of @xmath67 subschemes in @xmath64 , to the hilbert scheme of @xmath67 subschemes of a general quintic @xmath19 .
note first of all that we have looked above at the hilbert scheme @xmath439 $ ] of pairs @xmath703 .
however , for a given @xmath63 the space of choices of @xmath575 is just a grassmanian of @xmath4-planes @xmath818 .
so , irreducibility of the @xmath443-dimensional hilbert scheme @xmath819 implies irreducibility of the @xmath79-dimensional hilbert scheme @xmath435 which for brevity we denote just by @xmath820 and so forth .
consider now the incidence variety of pairs @xmath445 such that @xmath19 is a smooth quintic hypersurface containing @xmath63 .
the map @xmath821 is a fibration in projective spaces of dimension @xmath822 , indeed by the seminatural condition @xmath63 imposes @xmath43 conditions on the @xmath823 dimensional space @xmath824 and we should also divide out by scalars .
thus , the incidence variety @xmath825 is irreducible of dimension @xmath449 .
the space of quintics denoted @xmath826 is an open subset of @xmath827 , and the hilbert scheme of choices of @xmath63 for a given general @xmath19 , is the @xmath437-dimensional fiber of the map @xmath828 up to now , we have shown that the source of this map is irreducible .
an additional argument is needed to show that the fibers are irreducible .
we will use the same argument as was used in @xcite , which was pointed out to us by a. hirschowitz .
the idea is to say that there is a specially determined irreducible component of each fiber ; then this component is invariant under the galois action of the galois group of the function field of the base , on the collection of irreducible components of the fiber . on the other hand , irreducibility of the total space
means that the galois group acts transitively on the set of irreducible components of the fiber , and together these imply that the fiber is irreducible . in order to isolate a special irreducible component ,
notice that the singular locus of the moduli space of bundles was identified in @xcite .
it has a some explicit irreducible components corresponding to the choice of @xmath471 subschemes of length @xmath535 in @xmath19 , yielding the case of bundles with @xmath829 ( this is the case we have been explicitly avoiding throughout the bulk of the argument above ) .
we consider the @xmath541-dimensional component of the singular locus whose general point is a bundle @xmath21 fitting into an exact sequence @xmath830 where @xmath831 is a general collection of @xmath535 points on @xmath832 for a quadric @xmath588 . for a general such bundle @xmath21 , there is a unique co - obstruction , which is to say a unique exact sequence as above , and the zariski normal space to the singular locus may naturally be identified with @xmath833 which has dimension @xmath4 .
the second order obstruction map is the same as the quadratic form associated to the symmetric bilinear form obtained from duality @xmath834 .
this quadratic form defines a pair of lines inside @xmath833 .
these are the two actual normal directions of the moduli space of bundles along the singular locus at @xmath21 . in order to show that this component of the singular locus meets a canonically defined irreducible component of the moduli space
, it suffices to show that these two lines are interchanged as @xmath835 moves about in the hilbert scheme of @xmath535-tuples of points in @xmath108 .
the @xmath4-dimensional space @xmath833 together with its quadratic form , depend only on the arrangement @xmath836 of @xmath535 points on a quadric @xmath837 , in a way we now explain .
the homogeneous coordinates of the @xmath535 points give a map @xmath838 .
we get a map @xmath839 , and the equation of the quadric @xmath588 is an element of the kernel ; as @xmath840 has dimension @xmath535 itself , there is an element @xmath841 in the cokernel , unique up to scalars .
the @xmath471 condition , which holds for general @xmath835 , corresponds to asking that @xmath842 for all @xmath843 .
therefore @xmath844 defines a nondegenerate symmetric bilinear form on @xmath845 denoted @xmath846 the condition that @xmath844 vanish on the image of @xmath840 says that @xmath847 is an isotropic subspace . in other words , it is contained in its orthogonal subspace @xmath848 .
the quotient @xmath849 is our two - dimensional space @xmath833 and @xmath850 induces a quadratic form on here .
we are interested in the two isotropic lines .
fix @xmath583 of the points in a general way ; then our two dimensional subspace with quadratic form , depends on a single choice of @xmath851 .
a calculation shows that the discriminant divisor of the quadratic form contains reduced components in @xmath588 .
so if one has a curve of points @xmath852 which intersect this divisor transversally , the two lines are interchanged when we go around the intersection point on the curve .
now , one can choose @xmath19 to pass through the given @xmath853 as well as transversally through a general reduced point on the discriminant divisor .
for such @xmath19 , the tangent directions are interchanged as @xmath835 moves around in @xmath832 , so the same is also true for any general @xmath19 .
this completes the construction of a specified irreducible component of the moduli space of bundles .
notice that for the singular points @xmath21 constructed above , we still have @xmath163 , so a soon as we move off the singular locus to get @xmath48 , this gives a bundle with seminatural cohomology .
thus , our specified irreducible component corresponds to bundles with seminatural cohomology .
now , hirschowitz s argument plus theorem [ only ] saying that the hilbert scheme of choices @xmath820 is irreducible , combine to show that there is only one irreducible component in the moduli space of stable bundles on @xmath19 of degree @xmath0 and @xmath1 having seminatural cohomology .
this completes the proof of theorem [ main ] .
we indicate here how one should be able to treat conjecture [ all ] .
notice that we made the hypothesis that @xmath163 , and this implied seminatural cohomology .
so , in the non - seminatural case we have @xmath854 , and @xmath184 . if @xmath855 with subscheme of zeros @xmath63 then @xmath856 . the first step will be to show that sections of @xmath55 have a base locus consisting of at most one point @xmath350 , and that a general @xmath63 has to be reduced at @xmath467 , with @xmath541 points making up @xmath101 with doubly - transitive galois action .
this should be similar to our arguments of sections [ baseloci ] and [ pstructure ] .
one can also point out , right away , that this allows us to rule out the `` common curve case '' as in section [ sec - ccc ] , indeed even in the case when @xmath662 has degree @xmath110 , the same argument as we used for degrees @xmath522 and @xmath497 works to show that @xmath662 would have to be contained in a quadric .
so , we are in the variable curve case . if @xmath575 is a complete intersection of two cubics passing through @xmath63 , then @xmath597 with @xmath577 irreducible , containing @xmath101 in its smooth locus .
part of the argument consisted of ruling out @xmath857 by a dimension count .
here we ca nt just transpose the arguments , indeed the dimension of the hilbert scheme of possible collections @xmath63 might be strictly smaller than @xmath79 , because each @xmath63 can contribute a positive dimensional space of extension classes .
so we should divide the argument into two cases .
if @xmath858 , i.e. @xmath75 in the notations of section [ notations ] , then the dimension of @xmath859 is @xmath0 and the extension class is unique up to scalars . in this case , the dimension of the hilbert scheme @xmath820 remains @xmath79 ( and including the complete intersection curve @xmath575 gives @xmath819 of dimension @xmath443 ) .
the dimension count may then proceed as we have done and this should allow us to treat this case . in the case when @xmath860 , each choice of @xmath63 corresponds to a positive dimensional space of choices of extension class up to scalars .
however , in this case we can degenerate the extension class to one which no longer satisfies the cayley - bachrach condition meaning that , viewed as a dual element to @xmath861 , it vanishes on one or more points .
the doubly transitive galois action on @xmath63 implies that the images of the points @xmath63 in the projective space of extension classes , can not generically bunch up in groups of more than one .
therefore , it is possible to degenerate the extension class towards one which vanishes at exactly one point of @xmath63 .
this means an extension which corresponds to a torsion - free sheaf @xmath21 with a singularity at a single point .
it therefore corresponds to a point in the boundary of the moduli space , at the boundary component coming from @xmath862 .
this boundary piece has codimension @xmath0 and we should be able to analyze the nearby bundles and conclude that we remain in the principal irreducible component ( indeed it suffices to say that nearby bundles have seminatural cohomology ) .
the technique of localizing our picture on the boundary of the moduli space is obviously a necessary and important one which needs to be further developed in order to treat this type of question .
this will be left for a future work .
another interesting direction will be to look at reider s theory of nonabelian jacobians @xcite @xcite for bundles on a general quintic surface .
the structures we have encountered in an _ ad hoc _ way in the course of our proof , are actually pieces of reider s theory . | in this paper we continue our study of the moduli space of stable bundles of rank two and degree @xmath0 on a very general quintic surface .
the goal in this paper is to understand the irreducible components of the moduli space in the first case in the `` good '' range , which is @xmath1 .
we show that there is a single irreducible component of bundles which have seminatural cohomology , and conjecture that this is the only component for all stable bundles .
this paper is the next in a series , starting with @xcite , in which we study the moduli spaces of rank two bundles of odd degree on a very general quintic hypersurface @xmath2 .
this series is dedicated to professor maruyama , who brought us together in the study of moduli spaces , a subject in which he was one of the first pioneers . in the first paper
, we showed that the moduli space @xmath3 , of stable bundles of rank @xmath4 , degree @xmath0 and given @xmath5 , is empty for @xmath6 , irreducible for @xmath7 , and good ( i.e. generically smooth of the expected dimension ) for @xmath8 . on the other hand
, nijsse has shown that the moduli space is irreducible for @xmath9 @xcite using the techniques of ogrady @xcite @xcite .
this leaves open the question of irreducibility for @xmath10 .
[ all ] the moduli space @xmath11 is irreducible .
we havent yet formulated an opinion about the cases @xmath12 . in the present paper , due to lack of time and for length reasons ,
we treat a special case of the conjecture : the case of bundles with _
seminatural cohomology _ , meaning that only at most one of @xmath13 , @xmath14 or @xmath15 can be nonzero for each @xmath16 .
let @xmath17 denote the open subvariety of the moduli space consisting of bundles with seminatural cohomology . in section [ seminatural ]
we show that the seminatural condition is a consequence of assuming just @xmath18 .
the main result of this paper is : [ main ] the moduli space @xmath17 is irreducible .
recall from @xcite that our inspiration to look at this question came from the recent results of yoshioka , for the case of calabi - yau surfaces originating in @xcite .
yoshioka shows that the moduli spaces are irreducible for all positive values of @xmath5 , when @xmath19 is an abelian or k3 surface @xcite @xcite .
his results apply for example when @xmath19 is a general quartic hypersurface .
we thought it was a natural question to look at the case of a quintic hypersurface , which is one of the first cases where @xmath19 has general type , with @xmath20 being as small as possible .
_ remark on the difficulty of this project : _ we were somewhat surprised by the diversity of techniques needed to treat this question .
much of the difficulty stems from the possibilities of overdetermined intersections which need to be ruled out at various places in the argument .
this question is inherently very delicate , because there is not , to our knowledge , any general principle which would say whether the moduli space is `` supposed to be '' irreducible or not . on the one hand ,
the present case is close to the abelian or k3 case , so it is nt too surprising if the moduli space remains irreducible ; however on the other hand , at some point new irreducible components will be appearing as has been shown by the first author in @xcite .
so , we are led to analyse a number of cases for various aspects of the argument .
if any case is mistakenly ignored , it might hide a new irreducible component which would then be missed .
a natural question to wonder about is whether `` derived algebraic geometry '' could help here , but it would seem that those techniques need to be further developed in order to apply to some basic geometric situations such as we see here .
furthermore , each place in the argument where some case is ruled out , constitutes a possible reason for there to be additional irreducible components in more complicated situations ( such as on a sextic hypersurface ) .
so , in addition to the theorem itself which only goes a little way into the range that remains to be treated , much of the interest lies in the geometric situations which are encountered along the way . |
recently , the opposite deformations between proton and neutron densities in c isotopes were theoretically suggested @xcite by the method of antisymmetrized molecular dynamics ( amd ) : the proton density has an oblate deformation , while the neutron density has a prolate deformation , and the symmetry axis of proton is perpendicular to that of neutron in @xmath4c and @xmath0c . based on this picture , the author also gave a qualitative explanation @xcite for unusually small electric transition strength @xmath5 in @xmath0c , which is derived from life time measurement @xcite . according to ref.@xcite ,
the 2@xmath6 state is a rotational excited state , and the rotational axis is perpendicular to the neutron symmetry axis . in this excitation mechanism ,
the proton transition strength is reduced due to the difference of deformation between the proton and neutron distributions mentioned above , and therefore , the @xmath1 transition is dominated by the neutron excitation . in order to search for the possible difference of proton and neutron contributions to excitation of the @xmath7 state in @xmath0c
, inelastic scattering experiment of @xmath0c on @xmath2pb target was performed @xcite applying the coulomb - nuclear interference method .
the analysis was carried out by using the deformed potential model , and the proton and neutron transition matrix elements , @xmath8 and @xmath9 , were extracted . in ref.@xcite , it is mentioned that the experimental transition probability is inconsistent with theoretical ones ( the amd , extended amd and shell - model calculations are cited ) .
however , the phenomenological analysis done in ref.@xcite contains some assumptions , and hence , it seems inappropriate to compare the @xmath8 and @xmath9 values evaluated in ref.@xcite with those calculated theoretically . to test the @xmath0c internal wave function
, we should directly link the cross section with the wave function by calculating the differential cross sections of the inelastic scattering of @xmath0c on @xmath2pb target with the microscopic coupled - channels ( mcc ) method , and compare the calculated result with the experimental reaction data measured in ref.@xcite .
the mcc method has been applied for studying reactions of stable nuclei , such as the @xmath10li and @xmath11be elastic and inelastic scattering @xcite , the resonance reactions of the @xmath12c+@xmath12c system leading to inelastic @xcite and @xmath13be+@xmath0o @xmath14-transferred channels @xcite , the rainbow scattering of @xmath0o+@xmath0o system @xcite , etc . , adopting the microscopic cluster model wave functions of @xmath10li @xcite , @xmath11be @xcite , @xmath12c @xcite , and @xmath0o @xcite . because the microscopic cluster model wave functions well reproduce the measured charge form factors not only of the elastic electron - scattering but also the inelastic one , the wave functions are reliable and adopted for studying nuclear reaction mechanisms .
the mcc calculations successfully reproduce the experimental reaction data .
the reliability of the method has already been established .
hence , we think it is possible to examine inversely the validity of calculated internal wave functions by comparing the result of mcc with experimental reaction data . in this paper
, we adopt the amd internal wave function of @xmath0c .
the reason why we think it is worthy to test the validity of the amd wave function in the present study is as follows .
( 1 ) because no inert cores and no clusters are assumed , the amd wave function is flexible
. therefore , amd is suited for structure study of general unstable nuclei .
the applicability has been proved in many works @xcite .
( 2 ) deformations of proton and neutron densities are obtained dynamically .
in other words , electromagnetic transition probability can be calculated without introducing effective charge .
( 3 ) in amd , it is easy to carry out the spin - parity projection , which is necessary for microscopic calculation of transition density .
it should be noted that any nuclear structure models are applicable to mcc , if the diagonal density and transition ones to excited states can be calculated from a model wave function that gives no spurious center - of - mass component . in the next section ,
we briefly describe the mcc method .
more detailed description is made in refs.@xcite .
the coupled - channels equation describing the collision of two nuclei for a total angular momentum of the system @xmath15 is written as @xmath16 \chi_{\alpha l}^{(j)}(r ) \hspace{2.5 cm } \nonumber \\ % = - \sum_{(\alpha ' l ' ) \neq ( \alpha l ) } u^{(j)}_{\alpha l , \alpha ' l'}(r ) \
\chi_{\alpha ' l'}^{(j)}(r ) , \label{eq : cc}\end{aligned}\ ] ] where @xmath17 denotes the kinetic - energy operator , and @xmath14 and @xmath18 denote a channel and the orbital angular momentum associated to the relative coordinate @xmath19 . in the present study
, we take into account the elastic and @xmath0c excitation channels , while only the ground state ( @xmath20 ) is considered for the target @xmath2pb nucleus .
thus , the channel @xmath14 is designated by the spin i and the excitation energy @xmath21 of @xmath0c .
@xmath22 represents the center - of - mass energy for the projectile - target relative motion in channel @xmath14 ( @xmath23 ) .
@xmath24 is the relative wave function and is obtained by solving eq.([eq : cc ] ) numerically . in eq.([eq :
cc ] ) , @xmath25 represents the diagonal @xmath26 or the coupling @xmath27 potential , which is composed of the nuclear part @xmath28 and the coulomb part @xmath29 .
the nuclear part is given by the double - folding model and defined as @xmath30 % \left [ \
y_{\lambda}(\hat{\mathbf r}_1 ) \ \otimes y_{\lambda}(\hat{\mathbf r } ) \ \right ] _ { 00 } , \label{eq : fold } \\[0.3 cm ] % & & ( { \mathbf x}={\mathbf r}_1 + { \mathbf r } - { \mathbf r}_2 ) \nonumber\end{aligned}\ ] ] where @xmath31 represents the ordinary racah coefficient and @xmath32 is @xmath33 . @xmath34 and @xmath35 ( @xmath36 and @xmath37 ) are the radial components of proton and neutron transition densities of @xmath0c ( @xmath2pb ) , respectively , which will be mentioned in the next subsection in detail .
@xmath38 represents the spin- and isospin - scalar ( @xmath39 ) component of an effective nucleon - nucleon interaction , while @xmath40 represents the spin - scalar , isospin - vector ( @xmath41 , @xmath42 ) component . for this effective interaction ,
we adopt the ddm3y ( density - dependent michigan three - range yukawa ) @xcite , which is defined by @xmath43 where @xmath44 is the internucleon separation , and @xmath45 is a density dependent factor @xmath46 .\ ] ] here , @xmath47 denotes an incident energy per nucleon in the laboratory system .
the coefficients @xmath48 , @xmath49 , and @xmath50 in the density - dependent factor @xmath45 were determined at each energy by fitting a volume integral of the @xmath51 to the real part of the optical potential felt by a nucleon in the nuclear matter @xcite .
@xmath52 in eq.([eq : ddm3y ] ) is the original m3y interaction @xcite : @xmath53 with @xmath54 and @xmath55 with @xmath56 the units for @xmath47 and @xmath57 are mev / nucleon and fm , respectively . the coulomb part @xmath29 is also given by the double - folding model .
the double - folded coulomb potential is written in the same form as eq.([eq : fold ] ) , by replacing the neutron densities and the nucleon - nucleon interaction as @xmath58 since ddm3y has no imaginary part , we add the imaginary potential @xmath59 to the nuclear part , which is assumed as @xmath60 , where @xmath61 is the only a phenomenological parameter of the present mcc formalism .
the simple assumption for the imaginary part should be valid in the present case , since we only discuss the cross sections at very forward scattering angles , which is not sensitive to the detail shape of the potential in whole radial range . hence , the interaction potential has the form as @xmath62 .
@xmath63 , @xmath64 and @xmath65 values of @xmath0c calculated by amd , in which the strength of the spin - orbit force is set to ( i ) @xmath66=900 mev , ( ii ) @xmath66=1500 mev , and ( iii ) @xmath66=2000 mev . the experimental data of @xmath63 is taken from ref.@xcite .
[ cols="^,^,^,^,^",options="header " , ] the diagonal or transition density of proton at a position @xmath44 with respect to the center - of - mass of the nucleus can be expanded into multipole components : @xmath67 where @xmath68 represents the proton wave function in the nucleus .
@xmath69 represents the radial component of the transition density , which is used in eq.([eq : fold ] ) .
the radial component of the neutron transition density @xmath70 is obtained in the same manner as the proton case in terms of the neutron wave function @xmath71 .
the proton or neutron matrix element of rank @xmath72 is defined as @xmath73 where @xmath74 represents @xmath75 or @xmath76 .
the proton matrix element is related with the electric transition strength @xmath77 as @xmath78 here , we use the amd wave function for @xmath79 and @xmath80 to calculate the transition densities defined in eq.([eq : den ] ) .
we consider the ground state ( 0@xmath6 ) and first excited 2@xmath6 state . in ref.@xcite ,
two versions of @xmath0c internal wave function are obtained in the variation before projection ( vbp ) formalism changing the strength of the spin - orbit force , ( i ) @xmath66=900 mev and ( ii ) @xmath66=1500 mev .
the @xmath63 values as well as the @xmath81 and @xmath82 ones obtained with amd wave functions ( i ) and ( ii ) are summarized in table [ tab : be2 ] with the experimental data of @xmath63 . with the spin - orbit forces ( i ) and
( ii ) , the amd calculation reproduces well the systematic behavior of the @xmath63 value and root - mean - square radius of the c isotopes as shown in ref.@xcite . in particular , the systematic feature that the @xmath63 value of @xmath0c is abnormally small compared with other c isotopes ( @xmath4c , @xmath12c and @xmath83c ) is well reproduced by amd with the spin - orbit ( i ) and ( ii ) , although the @xmath63 value of @xmath0c is slightly overestimated .
in addition to above two , we also use the amd wave function for which the strength of the spin - orbit force is set to ( iii ) @xmath66=2000 mev so as to reduce the @xmath63 value and to be close to the experimental value .
the @xmath63 value as well as the @xmath81 and @xmath82 ones of the case ( iii ) is also shown in table [ tab : be2 ] .
one might think that the case ( iii ) gives the best wave function .
however , increasing the strength of the spin - orbit force to reduce the @xmath63 value of only @xmath0c as done in the case ( iii ) may lead to an unrealistic situation , because the systematic behavior of the other c isotopes is not reproduced with such strong spin - orbit force . hereafter
, we refer the three versions of amd wave function as amd(i ) , amd(ii ) and amd(iii ) . c. ( a ) @xmath72=0 component .
the solid curves represent the proton and neutron diagonal density of the ground state ( 0@xmath6 ) , while the dashed curves represent the those of the 2@xmath6 state .
the proton density of the @xmath7 state is almost identical to that of the @xmath84 state and the difference can not be seen in the figure .
( b ) @xmath72=2 component .
the dotted and solid curves represent the proton and neutron transition densities from 2@xmath6 to 0@xmath6 .
the double - dot - dashed and dashed curves represent the @xmath72=2 components of the proton and neutron diagonal densities for the 2@xmath6 state .
these two curves are displayed with opposite sign .
, width=226 ] figure [ fig : trd1 ] shows the radial components of the diagonal and transition densities obtained with amd(i ) . in the upper panel ( a )
, the @xmath72=0 components of the diagonal density are shown .
the solid curves represent the proton and neutron densities of the ground state ( @xmath84 ) , and the dashed curves represent those of the 2@xmath6 state .
it is found that the shape of the @xmath7 diagonal densities are almost the same as those of @xmath84 , except for the region around the origin .
the proton density of the @xmath7 state is almost identical to that of the @xmath84 state and the difference can not be seen in the figure . in the lower panel ( b ) ,
the @xmath72=2 components of the diagonal and transition density are shown .
the dotted and solid curves represent the proton and neutron transition densities , respectively , for the @xmath85 transition .
the transition is found to be dominated by the neutron component especially in the surface region .
the double - dot - dashed and dashed curves are the @xmath72=2 components of the 2@xmath6 diagonal density for the proton and neutron .
these two curves are displayed with opposite sign in the figure .
the shapes of the double - dot - dashed and dashed curves are similar to the dotted and solid curves , respectively .
it should be noted that the proton part of @xmath72=2 component of the diagonal density is proportional to the electric quadrupole moment of the 2@xmath6 state .
we neglect the @xmath72=4 component of the 2@xmath6 diagonal density , because this component is very small and is expected to have a small contribution to the inelastic scattering . .
the solid , dotted and dashed curves represent the transition densities of amd(i ) , amd(ii ) and amd(iii ) , respectively . the solid curves in ( a ) and ( b ) are the same as the dotted and solid curves in fig.[fig : trd1](b ) , respectively .
, width=226 ] in fig.[fig : trd2](a ) , the proton transition density of amd(i ) ( solid curve ) is compared with those of amd(ii ) and amd(iii ) , which are represented by the dotted and dashed curves , respectively .
the transition densities of amd(ii ) and amd(iii ) are smaller than that of amd(i ) , and the transition density of amd(iii ) is slightly shifted to small @xmath57 side .
the difference of the proton transition density causes the difference of the electric transition strength , as shown in table [ tab : be2 ] . in fig.[fig : trd2](b ) , the neutron transition densities of amd(i ) , ( ii ) and ( iii ) are shown by the solid , dotted and dashed curves , respectively .
although the overall shapes of transition densities of ( i ) and ( ii ) are almost the same , the magnitudes are found to be slightly different .
the magnitude of transition density of amd(iii ) is suppressed when it is compared with those of ( i ) and ( ii ) , especially in inner region .
the proton density distribution of the @xmath2pb nucleus is obtained by unfolding the charge density @xcite , which was obtained by a model - independent analysis of electron - scattering experiment , with the realistic proton charge form factor @xcite .
the neutron density distribution is obtained by assuming that the shape of the neutron density is the same as that of proton one , namely @xmath86 .
this assumption is known to be valid for stable nuclei . in the next section ,
we show the results of the mcc calculation using the amd transition densities described above .
c+@xmath2pb system calculated with amd(i ) are represented by the solid curves .
the dotted and dashed curves are the contributions of the proton and neutron components of @xmath0c , respectively , where both the proton and neutron components of @xmath2pb(0@xmath6 ) are included .
, width=226 ] in fig.[fig : foldpn ] , we show the double - folding model potential using the densities obtained by amd(i ) described in subsection [ ssec : trd ] .
the solid curve represents the nuclear potential eq.([eq : fold ] ) .
the dotted and dashed curves are the contributions of proton and neutron components of @xmath0c , respectively , where both the proton and neutron components of @xmath2pb(0@xmath6 ) are included . in fig.[fig : foldpn](a ) , the diagonal potential of @xmath0c(0@xmath6 ) + @xmath2pb elastic channel is shown . since the diagonal density of the 2@xmath6 state in @xmath0c resembles closely that of 0@xmath6 state as shown in fig.[fig : trd1](a ) , the diagonal potential of @xmath0c(2@xmath6 ) + @xmath2pb is almost the same as that of @xmath0c(0@xmath6 ) + @xmath2pb , and therefore , not shown here . in fig.[fig : foldpn](b ) , the coupling potential of @xmath0c(0@xmath87)+@xmath2pb is shown .
it is found that the neutron component has a dominant contribution to the total potential , especially in vicinity of the strong absorption radius @xmath88 fm . .
, width=226 ] in order to see the effect of the isovector component , we also decompose the diagonal and coupling potentials into the isoscalar and isovector components .
the result is shown in fig.[fig : foldisv ] .
the solid curves are the same as the solid ones in fig.[fig : foldpn ] , and the dotted and dashed curves are the isoscalar ( is ) and isovector ( iv ) components , respectively . compared with the isoscalar component ,
the magnitude of the isovector component is about 5@xmath89 , and the sign is opposite .
the 5@xmath89 reduction of the coupling potential leads to about 10@xmath89 reduction of the inelastic scattering cross sections and is not negligible in the present case .
therefore , we include the isovector component throughout the present calculations .
c(@xmath1 ) inelastic scattering on @xmath2pb target at e / a=52.7 mev .
the crosses are the experimental data and the solid curve is the result of our calculation .
the dotted and dashed curves are the nuclear and coulomb components , respectively .
, width=226 ] we perform a coupled - channels calculation using the diagonal and coupling potentials shown in fig.[fig : foldpn ] . because of the high incident energy ,
the coupled - channels equations are solved numerically with the relativistic kinematics , which has a non - negligible effect at forward - angle cross sections .
the parameter for the imaginary potential @xmath61 is set to 1.2 .
following the procedure of ref.@xcite , the calculated cross sections are smoothed by gaussian functions according to the experimental angular uncertainty of 0.28@xmath90 .
the result is shown in fig.[fig : xsc ] .
the differential cross sections are shown as a function of scattering angle @xmath91 in the laboratory system .
the crosses are the experimental data @xcite and the solid curve represents the result of the coupled - channels calculation .
it is found that the mcc calculation with amd(i ) reproduces the experimental data fairly well , although it slightly underestimates the magnitude of the cross sections at large angles . while the oscillatory shape of angular distribution is formed by interference between the nuclear and coulomb components , which are represented by the dotted and dashed curves , respectively , in fig.[fig : xsc ] , the average strength of the calculated cross section is determined by the nuclear excitation .
particularly , the neutron component dominates the nuclear excitation as understood from fig.[fig : foldpn](b ) .
therefore , present result indicates that amd(i ) slightly underestimates the neutron excitation strength by about 10 % , while it overestimates the proton excitation one as shown in table [ tab : be2 ] .
dependence of the calculated differential cross sections . the dotted , dashed , solid , dot - dashed and double - dot - dashed curves are the results of the mcc calculations with @xmath61=1.0 , 1.1 , 1.2 , 1.3 and 1.4 , respectively .
, width=226 ] the parameter @xmath61 can not be determined theoretically in the present mcc framework . in order to see the @xmath61 dependence of the calculated result
, we perform the same calculation with different @xmath61 values .
the results are shown in fig.[fig : idep ] .
the dotted , dashed , solid , dot - dashed and double - dot - dashed curves are the results of the mcc calculations with @xmath61=1.0 , 1.1 , 1.2 , 1.3 and 1.4 , respectively .
it is seen that @xmath61-dependence is very weak , although the cross sections at very forward angles slightly changes with @xmath61 .
the angular distribution around @xmath92 = 3 - 4 degrees is seen to be independent of the @xmath61 value .
because the calculation with @xmath61=1.2 reproduces the shape of the data a little better than others , we choose @xmath61=1.2 in the present calculation .
next , we perform the mcc calculations using amd(ii ) and ( iii ) . in fig.[fig : hikaku](a ) , the coupling potential with amd(i ) represented by the solid curve is compared with those with ( ii ) and ( iii ) , which are represented by the dotted and dashed curves , respectively .
it is noticed that the strength of the coupling potential is almost proportional to the @xmath93 value ; 16.1 @xmath94 for ( i ) , 14.8 @xmath94 for ( ii ) and 11.1 @xmath94 for ( iii ) .
the results of the mcc calculations with amd(ii ) and ( iii ) are shown in fig.[fig : hikaku](b ) by the dotted and dashed curves , compared with the result of amd(i ) shown by the solid curve which is the same as that in fig.[fig : xsc ] , and the experimental data . as expected from the strength of the coupling potential , the differential cross sections with amd(ii )
are slightly smaller than those calculated with amd(i ) , and those with amd(iii ) severely underestimates the magnitude of the measured cross sections .
the magnitude of differential cross section of inelastic scattering directly reflects the electric and hadronic transition strength of the @xmath0c nucleus . for amd(iii ) , the proton transition seems good because it gives the @xmath63 value being close to the measured one .
however , the nuclear excitation strength of ( iii ) is too small as shown in fig.[fig : hikaku](b ) .
since the nuclear excitation is dominated by the neutron component , this result indicates that amd(iii ) fails to describe the neutron excitation correctly .
this fact can not be known by the experimental data of electromagnetic probe .
therefore , we think it is very important that the internal wave function of a nucleus obtained theoretically is tested by the experimental data of hadronic probe to investigate the behavior of neutron component . for further investigation , it is very interesting to analyze the experimental data of the @xmath0c inelastic scattering on proton measured at riken @xcite , which is more sensitive to the neutron excitation .
( b ) the result of the mcc calculation when the aligned proton density shown in ( a ) is used .
, title="fig:",width=226 ] ( b ) .
( b ) the result of the mcc calculation when the aligned proton density shown in ( a ) is used .
, title="fig:",width=226 ] finally , we see how the angular distribution changes if the symmetry axis of proton is aligned to that of neutron . in the original amd wave function , the symmetry axis of proton is perpendicular to that of neutron .
we artificially make the aligned proton density from the amd wave function ( ii ) by rotating the proton wave function to set its symmetry axis to be parallel to the neutron one .
the aligned proton density is shown in fig.[fig : align](a ) by the dotted curve .
the solid curve is the neutron transition density which is the same as the dotted curve in fig.[fig : trd2](b ) .
the proton transition density has the opposite sign to the neutron one in the surface region .
the result of the mcc calculation using this aligned proton transition density is shown in fig.[fig : align](b ) .
it is found that the calculated angular distribution is out - of - phase compared with the experimental data .
this result indicates that the proton transition density should have the same sign as the neutron transition density in the surface region .
in order to test the @xmath0c internal wave function , we studied the @xmath0c(@xmath1 ) inelastic scattering on @xmath2pb target at @xmath3=52.7 mev @xcite by the microscopic coupled - channels ( mcc ) method using the internal wave function of the @xmath0c nucleus obtained by the antisymmetrized molecular dynamics ( amd ) @xcite . in ref.@xcite ,
two versions of wave function are obtained with the strength of spin - orbit force ( i ) @xmath95= 900 mev and ( ii ) @xmath96=1500 mev .
it was shown in ref.@xcite that these amd calculations reproduced the systematic behavior of @xmath63 value and root - mean - square ( rms ) radius of c isotopes .
the mcc calculations using these wave functions of @xmath0c reproduce well the measured differential cross sections , although they slightly underestimate the magnitude of the cross sections at large angles .
especially , the shape is rather well reproduced around @xmath92= 3 - 4 degrees , where the angular distribution is independent of the strength parameter @xmath61 of the imaginary potential .
while the shape of differential cross section due to the interference between the nuclear and coulomb excitation components is sensitive to the strength of the proton excitation , the magnitude of the cross section is sensitive to the strength of neutron excitation , because the nuclear excitation is dominated by the neutron one in the present case .
therefore , we can conclude that the amd wave function of ( i ) predicts the neutron excitation strength of @xmath0c reasonably well , although the strength may be slightly underestimated by about 10 % .
we also performed a coupled - channels calculation using amd wave function , for which the strength of the spin - orbit force is set to ( iii ) @xmath95=2000 mev .
although this wave function gives @xmath63 value of @xmath0c close to the measured one , the systematic behavior of @xmath63 and rms radius of the other c isotopes are failed to be reproduced due to the unrealistic strength of the spin - orbit force .
the mcc calculation using amd(iii ) severely underestimates the differential cross sections , which indicates that the neutron excitation is not properly described when the spin - orbit force ( iii ) is used .
it can be said that testing the validity of calculated wave function only with the electromagnetic experimental data , such as @xmath63 value , may be insufficient . especially for neutron - rich nuclei
, it is expected that the proton density is largely different from the neutron one .
therefore , we think it is very important that the internal wave functions of a nucleus obtained by any nuclear structure theory should be tested not only by the experimental data of electromagnetic probe but also those of hadronic probe , particularly to investigate the behavior of the neutron component , as done in the present paper .
we showed that the mcc calculation is a useful tool to link the inelastic scattering data with the internal wave functions obtained theoretically .
note that since the diagonal density is also reflected by the behavior of the calculated differential cross section through the diagonal potential as the distorting effect , the overall feature of diagonal and transition densities of nucleus can be tested in the consistent procedure .
nuclear reaction data themselves are available for the nuclear structure study similarly to the rms radius , electromagnetic transition strength , charge form factor , etc .
, by applying the mcc method .
the authors would like to thank prof .
t. motobayashi for valuable comments .
they are also thankful to dr .
z. elekes for providing us the experimental data .
one of the authors ( m.t . ) is grateful for the financial assistance from the special postdoctoral researchers program of riken .
this work is partially performed in the `` research project for study of unstable nuclei from nuclear cluster aspects '' sponsored by riken . | in order to test the @xmath0c internal wave function , we perform microscopic coupled - channels ( mcc ) calculations of the @xmath0c(@xmath1 ) inelastic scattering by @xmath2pb target at @xmath3=52.7 mev using the antisymmetrized molecular dynamics ( amd ) wave functions of @xmath0c , and compare the calculated differential cross sections with the measured ones .
the mcc calculations with amd wave functions reproduce the experimental data fairly well , although they slightly underestimate the magnitude of the cross sections .
the absolute magnitude of calculated differential cross sections is found to be sensitive to the neutron excitation strength .
we prove that the mcc method is a useful tool to connect the inelastic scattering data with the internal wave functions . |
brown dwarfs , objects whose central temperatures never reach the critical threshold for stable thermonuclear burning ( @xcite ; @xcite ) , are the lowest mass products of star formation .
hundreds of examples are now known , enabling the study of brown dwarfs as a population in their own right ( @xcite ) .
the study of brown dwarfs helps to constrain mechanisms for small - object formation , which include turbulent fragmentation ( @xcite ; @xcite ) , magnetic field confinement ( @xcite ) , stellar embryo ejection through dynamical interactions ( @xcite ; @xcite ) , and photo - evaporation of embryos by nearby hot stars ( @xcite ) .
brown dwarfs also represent a `` fossilized '' record of star formation throughout the galaxy s history because their mass is never ejected back into the interstellar medium .
they therefore preserve information on metallicity enrichment over the lifetime of the milky way ( @xcite ) .
solitary brown dwarfs have also proven to be excellent calibrators of the atmospheric models on which our inference of the properties of giant exoplanets depends ( @xcite ; @xcite ; @xcite ) .
their effective temperatures are similar to those of the exoplanets discovered thus far but their spectra lack the complication of irradiation from a host star . despite uncovering hundreds of brown dwarfs ,
previous surveys have allowed us to identify only the warmest examples .
the latest object currently having a measured spectrum is ugps [email protected] , whose effective temperature is estimated to be 520@xmath640 k ( @xcite ; bochanski et al .
, submitted , find @xmath7 = 500 - 600k ) and whose spectrum is used as the near - infrared t9 spectral standard ( cushing et al . , accepted)t8.5 have been reclassified now that the end of the t dwarf sequence and beginning of the y dwarf sequence has been defined ( cushing et al . , accepted ) . ] .
two other objects wd 0806@xmath2661b ( also known as gj 3483b ; @xcite ) and cfbdsir j145829 + 101343b ( @xcite ) are probably even colder and later in type than ugps [email protected] , although both currently lack spectroscopic confirmation .
both of these objects underscore the fact that the coldest brown dwarfs are extremely faint even at near - infrared wavelengths where ground - based spectroscopy has its best chance of characterizing the spectra .
wd 0806@xmath2661b , a common proper motion companion to a white dwarf with a measured distance of [email protected] pc , has yet to be detected in ground - based imaging observations down to j=23.9 mag ( luhman et al . ,
submitted ; see also @xcite ) .
+ 101343b is the secondary in a system with a composite spectral type of t9 and a measured distance of [email protected] pc ( @xcite ) . a combination of its close proximity ( 0.11 arcsec ) to the primary along with a faint magnitude ( [email protected] ) make the acquisition of a spectrum challenging . of the two ,
wd 0806@xmath2661b is less luminous at j band and presumably intrinisically fainter bolometrically ( see also @xcite ) .
finding even closer and brighter examples of cold brown dwarfs will be necessary to maximize our chances of best characterizing them .
canvassing the immediate solar neighborhood for such cold objects is one of the goals of the all - sky mission performed by the wide - field infrared survey explorer ( @xcite ) .
discoveries will directly measure the low - mass cutoff of star formation ( figure 12 of @xcite ) and provide even colder fiducial atmospheres for modeling cold exoplanets and understanding the gas giants of our own solar system .
the discovery of cold objects raises the question of whether a new spectral class , dubbed `` y '' ( @xcite ; @xcite ; see also @xcite ) , will be needed beyond the t class . in this paper
we present an overview of the our first @xmath5100 wise brown dwarf discoveries and show that objects colder than those previously known , including y - class brown dwarfs , are being uncovered .
wise is an earth - orbiting nasa mission that surveyed the entire sky simultaneously at wavelengths of 3.4 , 4.6 , 12 , and 22 @xmath1 m , hereafter referred to as bands w1 , w2 , w3 , and w4 , respectively .
as shown in figures 6 , 7 , and 13 of @xcite as well as figure 2 of @xcite , the w1 and w2 bands were specifically designed to probe the deep , 3.3 @xmath1 m ch@xmath8 absorption band in brown dwarfs and the region relatively free of opacity at @xmath54.6 @xmath1 m .
since the peak of the planck function at low temperatures is in the mid - infrared , a large amount of flux emerges in the 4.6 @xmath1 m window , and this makes the w1-w2 colors of cool brown dwarfs extremely red ( see 2.1 ) .
such red colors , which are almost unique among astronomical sources , make the identification of cool brown dwarfs much easier . wise launched on 2009 dec 14 and , after an in - orbit checkout , began surveying the sky on 2010 jan 14 .
its sun - synchronous polar orbit around the earth meant that each location along the ecliptic was observed a minimum of eight times , with larger numbers of re - visits occurring at locations nearer the ecliptic poles .
wise completed it first full pass of the sky on 2010 jul 17 and its second pass on 2011 jan 09 . during this second pass ,
the outer , secondary tank depleted its cryogen on 2010 aug 05 , rendering the w4 band unusable , and the inner , primary tank depleted its cryogen on 2010 sep 30 , rendering the w3 band unusable .
thus , this second full sky pass is partly missing bands w3 and w4 .
fortunately , the bands most crucial for brown dwarf selection w1 and w2 were little affected by this cryogen exhaustion .
wise continued to collect data on a third , incomplete sky pass in bands w1 and w2 until data acquisition was halted on 2011 jan 31 .
preliminary processing of the data , including single - frame and coadded images and photometrically and astrometrically characterized detections , has been used to search for cold brown dwarf candidates , as described in detail below .
this is the same version of the pipeline software that produced the wise preliminary data release , details of which can be found in the explanatory supplement . for a more detailed description of the wise mission and data products , see @xcite and the nasa / ipac infrared science archive ( irsa ; http://irsa.ipac.caltech.edu ) . because processing of the data continues as of this writing ,
our candidate selection is on - going , and only a fraction of our candidates has been followed up , it is not possible to estimate the sky coverage or volume surveyed for discoveries presented herein .
however , objects discussed here can be added to the growing census of brown dwarfs in the solar neighborhood and can be used to place lower limits , as we do in 5.3 below , to the brown dwarf space density as a function of type or temperature . as such
, this paper should be regarded as a progress report on the continuing wise search for previously missed brown dwarfs in the sun s immediate vicinity . before beginning the hunt for brown dwarfs ,
it is necessary to establish empirically the locus of known brown dwarfs in wise color space and to understand what other kinds of astrophysical objects might fall in the same area .
this will not only inform the search of wise color space itself but also dictate the kinds of photometric follow - up that need to take place before time - intensive spectroscopic characterization begins .
we have performed a positional cross - correlation of nearby stars from dwarf archives against source lists derived from the wise coadded data .
many of these stars are known to have substantial proper motion , so it was necessary to verify each cross - match by visually inspecting the wise and two micron all - sky survey ( 2mass ; @xcite ) images .
the final cross - identifications are given in table [ photometry_knownmlt ] , which lists photometry from 2mass ( when detected ) and wise for 118 previously cataloged t dwarfs , 142 l dwarfs , and 92 m dwarfs .
figure [ w1w2_vs_type ] shows the resulting trend of wise w1-w2 color as a function of spectral type for these objects , ranging from early - m through late - t .
note that there is a slow increase in the w1-w2 color between early - m and early - l , with the color stagnating near 0.3 mag between early- and mid - l .
the w1-w2 color then rapidly increases at types later than mid - l , corresponding to the appearance of the methane fundamental band at 3.3 @xmath1 m ( @xcite ) .
the average w1-w2 color is @xmath50.6 mag at t0 and @xmath51.5 mag at t5 , with the color increasing to above 3.0 mag at late t. the red w1-w2 colors ( @xmath41.7 mag ) of dwarfs of type mid - t and later are almost , but not entirely , unique among astrophysical sources .
dust - obscured galaxies ( dogs ) and asymptotic giant branch stars ( agbs ) are the major sources of contamination at these red w1-w2 colors , as analysis of the spitzer deep wide - field survey ( sdwfs ) results of @xcite has shown .
the three short - wavelength bands of wise which are close in wavelength to the 3.6 , 4.5 , and 8.0 @xmath1 m bands ( hereafter denoted as ch1 , ch2 , and ch4 , respectively ) of the infrared array camera ( irac ) onboard the _ spitzer _ space telescope can help to distinguish between these populations . as figure 1 of @xcite shows ,
most agbs with very red w1-w2 ( or ch1-ch2 ) colors can be distinguished by their very red w2-w3 ( or ch2-ch4 ) colors .
brown dwarfs with similar w1-w2 colors are much _ bluer _ in w2-w3 color than these contaminants .
similarly , dogs should be easily separable from brown dwarfs because , like agbs , their w2-w3 ( or ch2-ch4 ) colors tend to the red for objects with very red w1-w2 colors .
this is further demonstrated in figure 12 of @xcite , where the bulk of the extragalactic menagerie , including red active galactic nuclei , can be distinguished from cold brown dwarfs using a color of [email protected] as the dividing line . as with any set of generic color cuts , however , one should be ever vigilant for exceptions . as figure [ w1w2_vs_type ] and
figure [ w2w3_vs_type ] show , very late - t dwarfs have colors approaching [email protected] mag , near the locus of extragalactic sources , but their w1-w2 colors are extreme ( @xmath43.0 mag ) .
few extragalactic sources have w1-w2 colors this red , so the w2-w3 color criterion can be relaxed for the coldest objects ( see figure [ w1w2_vs_w2w3 ] ) .
indeed , brown dwarfs with @xmath10 are expected to turn to the red in w2-w3 color ( figure 14 of @xcite ) .
it should also be noted that the location of low - gravity or low - metallicity brown dwarfs may not follow the general rule set by normal - gravity , solar - metallicity cases , so it is important to use other data ( proper motion , parallax ) when possible to identify brown dwarfs independent of photometric selections .
nonetheless , a number of possibly low - metallicity t dwarfs have been uncovered using these same photometric selections , as further discussed in the appendix .
we have used two different sets of criteria to search the wise source lists for nearby brown dwarfs : \(1 ) to find the coldest brown dwarfs , we selected high signal - to - noise ( snr@xmath43 at w2 ) detections having w1-w2 colors ( or limits ) greater than 1.5 mag , corresponding roughly to types @xmath0t5 .
( because of the relative depths of the w1 and w2 bands , the w1-w2 requirement imposes a more severe w2 snr limit of its own , generally w2 snr@xmath47 . ) in order to assure that an object is real , we required it to have been detected at least eight times in the individual w2 exposures ; this eliminated spurious sources like cosmic rays and satellite trails that would otherwise not be eliminated during the outlier rejection step in coadd image creation ( see section iv.5.a.v of the explanatory supplement to the wise preliminary data release ) . for our initial candidate selection
, we also required w2-w3 @xmath11 3.0 mag if the object has a detection in w3 .
\(2 ) to find bright , nearby ( i.e. , high proper motion ) l and t dwarfs that other surveys have missed , we searched for objects with w1-w2 colors greater than 0.4 mag ( roughly types @xmath0l5 ) , w2 snr values greater than 30 , and no association with a 2mass source ( implying that the j - w2 color is either very red or the object has moved ) .
it should be noted that the wise source lists report 2mass associations falling within 3 arcseconds of each wise source .
as with the first search , we required the object to have been detected at least eight times in w2 to assure its reliability . to eliminate extragalactic contaminants
, we imposed one additional criterion that w1-w2 @xmath4 0.96(w2-w3 ) @xmath2 0.96 ( see dashed line in figure [ w1w2_vs_w2w3 ] ) fall more than one sigma to the right of this line .
these objects are the m dwarfs cti 064951.4 + 280442 and cti 065950.5 + 280228 and the l dwarfs sdss j082030.12 + 103737.0 , sdss j102947.68 + 483412.2 , and sdss [email protected] .
visual inspection of the wise images for each of these shows that the w3 detections for all five sources are almost certainly spurious , the photometry code having found a low - level w3 `` detection '' at the sky location of the object .
the signal - to - noise values for each of these w3 detections are 3.1 , 2.2 , 3.1 , 2.0 , and 4.4 , respectively .
because the w3 band is sensitive to extended structure within the milky way as well as background variations by bright stars and the moon , low - level w3 detections should be regarded with caution . ] . for both searches , no constraint on galactic latitude
was imposed , although additional constraints were placed on object detections in order to assure that they were real , particularly for our earliest searches of the wise source lists .
first , the reduced @xmath12 value from the psf photometric fitting ( `` rchi2 '' in the wise preliminary release source catalog ) was required to fall between 0.5 and 3.0 to assure that the source was pointlike .
second , the early version of the wise data processing pipeline automatically flagged artifacts bright star halos , diffraction spikes , latent images , ghost reflections from bright stars , etc . only on individual frames and not on the coadded images . for search # 2 above , because those objects all have high - snr w2 detections , we are able to use these individual frame flags to remove objects marked as spurious . for fainter objects found in search # 1 , we created three - color images from the w1 ( blue ) , w2 ( green ) , and w3 ( red ) coaded images , which were then inspected by eye to eliminate artifacts
. third , for all objects passing the above tests , we created finder charts showing the digitized sky survey ( dss ; @xmath13 , @xmath14 , @xmath15 ) , the sloan digital sky survey ( sdss ; @xcite @xmath16 , @xmath17 , @xmath18 , @xmath19 , and @xmath20 , if available ) , 2mass ( @xmath21 , @xmath22 , @xmath23 ) , and wise ( w1 , w2 , w3 , w4 , + 3-color image made from w1+w2+w3 ) images .
a visual inspection of these finder charts allowed us to remove other spurious sources and objects clearly visible in the short wavelength optical bands , while also allowing us to check for proper motion between surveys for objects bright enough to have been detected at shorter wavelengths , such as @xmath21 and @xmath22 for brighter t dwarf candidates .
example images are shown in figure [ finder_chart1 ] through figure [ finder_chart19 ] for our spectroscopically confirmed candidates .
wise photometry for these same sources is given in table [ wise_photometry_discoveries ] .
follow - up imaging observations of wise candidates are important not only for verifying that the source has the characteristics of a brown dwarf at shorter wavelengths but also for determining how bright the object is at wavelengths observable from the ground .
this latter knowledge is necessary for determining which facility to use for spectroscopic confirmation .
ground - based near - infrared observations at @xmath21 and @xmath22 bands are technically the easiest to acquire . before discussing specifics
, it should be noted that the two main near - infrared filter systems being used today
the 2mass system and the mauna kea observatories near - infared ( mko - nir , or just mko ) system will yield somewhat different results for the same objects .
the 2mass bandpasses are illustrated in figure 2 of @xcite , and the mko filter profiles are shown in figure 1 of @xcite .
a comparison of the two filter sets is illustrated in figure 4 of @xcite . for @xmath21 and @xmath22 bands ,
the main difference between the two systems lies in the width of the @xmath21-band filter ; the @xmath22-band filters are very similar .
the 2mass @xmath21-band profile extends to bluer wavelengths than does the mko @xmath21-band profile , and the overall shapes of the two @xmath21 filters are quite different , the 2mass @xmath21 filter lacking the top - hat shape that is characteristic of most filter profiles . as a result ,
the measured @xmath21-band magnitude of a brown dwarf , whose spectral signature is also quite complex at these same wavelengths , can be considerably different between the two systems .
this is dramatically illustrated in figure 3 of @xcite , which shows that although the @xmath22-band magnitudes are ( as expected ) very similar between the two systems for a wide range of brown dwarf types , the @xmath21-band magnitudes ( and hence , @xmath24 colors ) can differ by as much as 0.5 mag for late - type t dwarfs . in the discussion that follows ,
we note the systems on which our photometry was obtained and we mark photometry from the two systems with different colors and symbols on the plots . as figure [ jh_vs_type ] and figure [ jh_vs_w1w2 ] show , mid- to late - t dwarfs ( w1-w2 @xmath4 1.5 mag ) have relatively blue colors , @xmath25 mag , on the 2mass filter system . the upper righthand panel of figure 5 of @xcite shows the trend of @xmath24 color versus spectral type using the mko filter system , and there we find that @xmath26 mag for dwarfs @xmath0t5 .
these blue colors in both systems are a consequence of stronger methane bands and collision - induced h@xmath27 absorption at @xmath22-band compared to @xmath21-band .
this color stands in contrast to the majority of other astrophysical sources , whose @xmath24 colors are much redder than this . for example , figure 19 of @xcite shows the 2mass @xmath24 color distribution of detected sources at high galactic latitude and confirms that most 2mass objects have colors redder than those of mid- to late - t dwarfs
. nevertheless , there are true astrophysical sources
e.g. , main sequence stars earlier than type k0 ( table 2 of @xcite ) ; certain agns , particularly those with @xmath28 ( figure 2 of @xcite ) that are nt brown dwarfs and have @xmath24 colors below 0.4 mag . fortunately , other colors like @xmath21-w2 and @xmath22-w2 can also be used to distinguish populations .
figure [ jw2_vs_type ] and figure [ hw2_vs_type ] show these colors as a function of spectral type and demonstrate that the @xmath21-w2 color of mid- to late - t dwarfs runs from @xmath52.0 mag at t5 to @xmath44.0 mag at late - t ; @xmath22-w2 color runs from @xmath51.5 mag to @xmath55.0 mag for the same range of types .
figure [ jw2_vs_w1w2 ] and figure [ hw2_vs_w1w2 ] show the trend of @xmath21-w2 and @xmath22-w2 with w1-w2 color .
the correlation is very tight for m and l dwarfs ( w1-w2 @xmath11 0.6 mag ) , but at redder w1-w2 colors , corresponding to the l / t transition and beyond , there is a much larger spread of @xmath21-w2 and @xmath22-w2 colors at a given w1-w2 color .
nonetheless , both @xmath21-w2 and @xmath22-w2 color increase dramatically beyond w1-w2 @xmath41.5 mag ( @xmath4t5 ) . with our list of brown dwarf candidates in hand ,
we have obtained @xmath21 and @xmath22 observations and in some cases , @xmath29 and @xmath23 as well using a variety of different facilities in both hemispheres .
details of those observations are given below , and a listing of the resultant photometry can be found in table [ followup_photometry_discoveries ] .
the fan mountain near - infrared camera ( fancam ) at the university of virginia s 31-inch telescope has a 1024@xmath301024-pixel hawaii-1 array ( pixel scale of 0.51 arcsec / pix ) that images an 8.7-arcminute square field of view ( @xcite ) .
observations of eight of our candidates were obtained in 2mass @xmath21 and @xmath22 bands , and for some objects @xmath29-band observations were also acquired .
details on observing strategy and data reduction are described further in @xcite .
persson s auxiliary nasmyth infrared camera ( panic ) at the 6.5 m magellan baade telescope has a 1024@xmath301024 hawaii array with a plate scale of 0@xmath31125 pixel@xmath32 , resulting in a 2@xmath302-arcminute field of view ( @xcite ) .
observations of three of our candidates were obtained in carnegie ( essentially mko ) @xmath21 and @xmath22 bands .
a series of dithered images of short integration times was obtained in each band , as is the standard in near - infrared imaging observations .
the data were reduced using custom interactive data language ( idl ) routines written by one of us ( m.c.c . ) .
each image was first corrected for non - linearity using the relation given in the panic observer s manual .
sky flat fields were constructed by first subtracting a median - averaged dark frame from each twilight flat frame .
the dark - subtracted twilight flats were then scaled to a common flux level and then medianed .
after each frame was sky subtracted and flat fielded , we corrected for optical distortions as described in the panic observer s manual
. frames were then aligned and combined using a median average to produce the final mosaic .
2mass stars in the field of view were used to both astrometrically calibrate the mosaic as well as determine the zero points for the images . the 2mass camera on the 1.5 m kuiper telescope on mt .
bigelow , arizona , has three 256@xmath30256 nicmos3 arrays capable of observing simultaneously in 2mass @xmath21 , @xmath22 , and @xmath23 filters ( @xcite ) .
the plate scale for all three arrays is 1@xmath3165 pixel@xmath32 , resulting in a 7@xmath307 arcmin field of view .
images for twenty - three of our candidates were obtained using a 3@xmath303 box dither pattern , and to reduce the overheads associated with nodding the telescope four images were taken at each of nine dither positions . at the conclusion of each dither sequence , the telescope was offset slightly before the start of the next sequence .
the data were reduced using custom idl routines .
flat fields in each band were constructed using on - source frames .
these images were scaled to a common flux level and combined using an average , the latter process rejecting the lowest 10 and highest 20 values at each pixel location to eliminate imprinting of real objects .
each raw frame was then flat fielded with these derived flats .
sky subtraction was then accomplished in a two - step process .
first , all of the coadded frames were median averaged to produce a first - order sky frame .
this frame was then subtracted from each of the coadded frames so that stars could be easily identified .
a sky frame was then constructed for each coadded image using the five previous frames in the sequence and the next five frames in the sequence .
stars identified in the previous step were masked out before the ten frames were median averaged .
the sky frame was then scaled to the same flux level as the image and subtracted . after all of the coadded frames had been sky subtracted , the frames were co - registered on the sky and averaged .
2mass stars in the field of view were used to both astrometrically calibrate the mosaic as well as determine the zero points for the images . the peters automated infrared imaging telescope ( pairitel ; @xcite ) on mt .
hopkins , arizona , is a 1.3 m telescope equipped with three 256@xmath30256 nicmos3 arrays with 2@xmath33 pixels that cover an [email protected] arcminute field of view simultaneously in 2mass @xmath21 , @xmath22 , and @xmath23 filters ( @xcite ) .
this camera is , in fact , one of the original 2mass cameras ( @xcite ) and the telescope is the original 2mass northern facility , now roboticized for transient follow - up and other projects .
thirty - three of our candidates were observed here . upon acquiring a target field ,
the system acquires a series of 7.8 s exposures until the desired integration time ( 1200 s ) is met . after every third exposure the telescope shifts to a position offset by about one tenth of the field of view in order to account for bad pixels on the arrays . within hours of a night
s observations an automated data pipeline delivers flexible image transport system ( fits ; @xcite ) mosaic images of the combined stack of observations for each object in each of the three wavebands .
the pipeline subtracts a background from each raw image based on the median of several images adjacent in time .
the resulting individual frames are calibrated with a flat field constructed from sky observations spanning many nights .
astrometric and photometric calibrations were accomplished using observations of 2mass - detected stars in the field of view .
the wide - field infrared camera ( wirc ) at the 5 m hale telescope at palomar observatory has a pixel scale of 0@xmath312487 pixel@xmath32 and uses a 2048@xmath302048-pixel hawaii-2 array to image an 8.7-arcminute square field of view ( @xcite ) .
observations of eighteen of our candidates were obtained in mko @xmath21 and @xmath22 bands .
images were reduced using a suite of iraf scripts and fortran programs kindly provided by t. jarrett .
the images were first linearized and dark subtracted .
sky background and flat field images were created from the list of input images , and then these were subtracted from and divided into , respectively , each input image . at this stage ,
wirc images still contained a significant bias not removed by the flat field .
comparison of 2mass and wirc photometric differences across the array showed that this flux bias had a level of @xmath910% and the pattern was roughly the same for all filters . using these 2mass - wirc differences for many fields , we created a flux bias correction image that was then applied to each of the `` reduced '' images .
finally , we astrometrically calibrated the images using 2mass stars in the field .
the images were then mosaicked together .
this final mosaic was photometrically calibrated using 2mass stars and a custom idl script .
magnitudes were calculated using the zero points computed using 2mass stars .
the gemini infrared imaging camera at the 3 m shane telescope at lick observatory uses two 256@xmath30256-pixel arrays ( pixel scale of 0.70 arcsec / pix ) for simultaneous observations of a 3@xmath303-arcminute field of view ( @xcite ) over each array .
the short - wavelength channel uses a nicmos3 hgcdte array , and the long - wavelength channel uses an insb array .
observations were obtained for three of our candidates and used 2mass @xmath21 , @xmath22 , and @xmath23 filters .
observations were acquired in pairs ( @xmath21 + @xmath23 or @xmath22 + @xmath23 ) so that twice as much integration time could be obtained on the insb array at @xmath23 as in either the @xmath21 or @xmath22 filter on the hgcdte array .
observations were obtained in dithered sequences .
data reduction was handled similarly to that described for the mt .
bigelow/2mass data .
the spartan infrared camera ( spartanirc ) at the 4.1 m southern astrophysics research ( soar ) telescope on cerro pachn , chile , uses a mosaic of two 2048@xmath302048-pixel hawaii-2 arrays to cover either a 3-arcminute square ( 0.04 arcsec / pix ) or 5-arcminute square ( 0.07 arcsec / pix ) field of view ( @xcite ) . for seven of our candidates
, we used the 5@xmath305-arcminute mode and acquired observations in the mko @xmath21 and @xmath22 ( and for one object object , @xmath34 ) filters .
observing strategy and data reductions followed the same prescription discussed in @xcite .
the second - generation near - infrared camera ( nirc2 ) at the 10 m w. m. keck observatory atop mauna kea , hawaii , employs a 1024@xmath301024 aladdin-3 array .
used with the keck ii laser guide star adaptive optics system ( @xcite ; @xcite ) , it can provide deep , high - resolution , imaging for our faintest targets . in the wide - field mode ,
the camera has a plate scale of 0.039686 arcsec / pix resulting in a field of view of 40@xmath3040 arcsec .
wisepa j182831.08 + 265037.8 and wisepc j205628.90 + 145953.3 are the only targets in this paper whose photometry comes solely from the nirc2 data because they were undetected in @xmath21- and/or @xmath22-band at other facilities .
the camera employs mko @xmath21 and @xmath22 filters . for wisepa j182831.08 +
265037.8 we used the @xmath14=16 star usno - b 1168 - 0346313 ( @xcite ) , located 41@xmath35 from the target , for the tip - tilt reference star . the tip - tilt reference star for wisepc j205628.90
+ 145953.3 was usno - b 1050 - 0583683 ( @xmath14=13 , separation=13@xmath35 ) .
a three - position dither pattern was used to avoid the noisy lower - left quadrant .
each position of the dither pattern consisted of a 120 s exposure ; the pattern was repeated two or three times with a different offset for each repeat in order to build up deeper exposures .
the images were reduced using a custom set of idl scripts written by one of us ( c.r.g . ) .
the raw images were first sky - subtracted using a sky frame constructed from all of the images .
next , a dome flat was used to correct for pixel - to - pixel sensitivity variations . in order to shift the reduced images to a common astrometric grid for the creation of the mosaic , we used the header keywords aotsx and aotsy , which record the position of the ao tip - tilt sensor stage .
as the telescope is dithered , this sensor must move so that the tip - tilt star stays properly centered .
although this method of computing image offsets is more precise than using the right ascension and declination offsets in the header , it can be prone to small positional errors . to account for this , we computed the minimum of the residuals of the shifted image relative to the reference image ( the first image of the stack ) over a 5@xmath305 pixel grid centered on the aotsx / aotsy - computed offset .
this correction was generally @xmath112 pixels .
the aligned images were then medianed to form the final mosaic .
there are typically very few , if any , 2mass stars in the nirc2 field of view that can be used to determine the photometric zero point .
we therefore bootstrapped a photometric zero point using faint stars in the nirc2 images that are also present in the much wider field - of - view palomar / wirc images that have been calibrated using 2mass stars . for this ensemble
we computed the average difference between the calibrated wirc magnitudes and the nirc2 instrumental magnitudes .
we used the standard deviation of the differences as the error in the photometric calibration and note that it is the dominant source of uncertainty .
the noao extremely wide field infrared imager ( newfirm ) at the 4 m victor m. blanco telescope on cerro tololo , chile , consists of four 2048@xmath302048 insb arrays arranged in a 2@xmath302 grid . with a pixel scale of 0.40 arcsec / pixel
, this grid covers a total field of view of [email protected] arcmin .
( @xcite ) . the only source in this paper with newfirm photometry is wisepa [email protected] , so
the discussion that follows is specific to this set of observations . at @xmath21-band ,
ten sets of images were obtained with integration times of 30 s and two coadds ; at @xmath22-band ten sets with exposure times of 5 s and twelve coadds were obtained .
thus , the total integration time was 600 s in each filter , and these are on the mko system . because of the large field of view and the 35@xmath35 gap between the arrays , we positioned the target near the center of the north - east array , designated sn013 in the proposal preparation guide .
photometry was performed on the mosaics produced by the newfirm pipeline ( @xcite ) .
although the pipeline computed a magnitude zero - point based on the photometry of 2mass stars in the mosaic , the default aperture used was so large ( @xmath9 6 pixel radius ) that wisepa [email protected] overlapped with a close neighbor .
furthermore , wisepa [email protected] was barely visible on the mosaics , meaning that a smaller aperture was warranted to increase the s / n of our measurement .
we therefore re - calibrated the mosaics using an aperture of 2-pixel radius .
some of the wise brown dwarf candidates were detected by 2mass or by the united kingdom infrared deep sky survey ( ukidss ; @xcite ) . for 2mass
, we have searched the all - sky point source catalog and the reject table and found photometry for forty - five of our sources , the other fifty - one being too faint for 2mass detection . for ukidss , we searched data release 6 ( or 5 for those mini - surveys not released in 6 ) and found five of our objects three in the large area survey and two in the galactic clusters survey .
some of the coldest brown dwarf candidates detected by wise have w1-w2 color _ limits _ because these sources are too faint to be detected in the w1 band .
acquiring deeper images at similar bandpasses would therefore be extremely beneficial in further deciding which of the candidates are the most interesting for spectroscopic follow - up .
fortunately , the two shortest wavelength bandpasses of the infrared array camera ( irac ; @xcite ) onboard the _ spitzer _ space telescope ( @xcite ) are operating during the warm _ spitzer _ mission and can be used to provide this missing info .
although these bandpasses at 3.6 and 4.5 @xmath1 m ( aka ch1 and ch2 , respectively ) are similar to the two shortest wavelength bands of wise ( see figure 2 of @xcite ) , they were not designed specifically for brown dwarf detection and therefore yield less extreme colors ( ch1-ch2 vs. w1-w2 ) for objects of the same spectral type .
nonetheless , irac photometry in the warm spitzer era provides a powerful verification tool for wise brown dwarf candidates .
figure [ ch1ch2_vs_type ] shows the trend of ch1-ch2 color versus spectral type for early - m through late - t dwarfs taken from @xcite , @xcite , and @xcite . as noted by these authors
, there is a clear trend of increasing ch1-ch2 color with later spectral type , as expected .
figure [ ch1ch2_vs_w1w2 ] maps the ch1-ch2 color onto w1-w2 color for these same sources .
two additional plots , figure [ jch2_vs_type ] and figure [ hch2_vs_type ] , show composite colors made from ground - based near - infrared and _ spitzer _ follow - up and demonstrate that both @xmath21-ch2 and @xmath22-ch2 colors can serve as proxies for spectral type for objects of type mid - t and later .
plots of @xmath22-ch2 color versus type have been presented in earlier papers ( the upper lefthand panel of figure 6 of @xcite ; figure 5 of @xcite ) .
even earlier , @xcite noted the tight correlation between @xmath22-ch2 color and effective temperature for later t dwarfs and suggested that the color is relatively insensitive to gravity and metallicity , which makes it a reliable indicator of t@xmath36 or type ( see also @xcite ) .
we have therefore obtained irac ch1 and ch2 photometry of many of our wise brown dwarf candidates ( program id=70062 ; j.d .
kirkpatrick , pi ) .
data were collected in both channels using a 5-position , medium - scale `` cycling '' dither script .
single images were 30s per position , for a total on - source exposure time of 150s per channel .
although the @xmath37 science center ( ssc ) suggests performing photometry on custom - built mosaics from the basic calibrated data ( bcd ) rather than the post - bcd mosaics , we have found that the photometric differences between the two mosaics are negligible for our data sets .
therefore , photometry of our sources was measured on the post - bcd mosaics as produced by the ssc irac pipeline , software version s18.18 .
the post - bcd mosaics have a pixel scale of 0.6@xmath35/pixel , which is half of the native pixel scale .
we used a 4-pixel aperture , a sky annulus of 24 - 40 pixels , and applied the aperture corrections listed in table 4.7 of the irac instrument handbook@xmath38 to account for our non - standard aperture size .
resulting irac photometry for those confirmed candidates observed so far is listed in table [ followup_photometry_discoveries ] .
for candidates whose follow - up imaging strengthened their credibility as a bona fide brown dwarf , we obtained spectroscopic confirmation at near - infrared wavelengths . the facilities and instruments we have used , along with the data reductions employed ,
are discussed below .
the low resolution imaging spectrometer ( lris , @xcite ) at the 10 m w. m. keck observatory atop mauna kea , hawaii , uses two channels to simultaneously observe blue and red optical wavelengths .
our observational setup employed only the two 2k@xmath304k ccds in the red channel . when used with the 400 lines / mm
grating blazed at 8500 and a 1@xmath35 slit , the red channel produces 10- - resolution spectra covering the range from 6300 to 10100 .
the og570 order - blocking filter was used to eliminate second - order light .
flat - field exposures of the interior of the telescope dome were used to normalize the response of the detector , and hgnear arc lamps were taken after each program object to obtain the wavelength calibration . because our targets were late - t dwarfs , for which the 9300- band of h@xmath27o is the most important spectral diagnostic , observations of g0 dwarf stars near in airmass and near in time were needed to correct for telluric h@xmath27o at these same wavelengths .
observations were typically acquired with the slit oriented to the parallactic angle to minimize wavelength - dependent slit losses .
once per night , a standard star from the list of @xcite was observed to provide flux calibration .
the data were reduced and calibrated using standard iraf routines as described in @xcite and @xcite .
spex , a medium - resolution spectrograph and imager at nasa s 3 m infrared telescope facility ( irtf ) on mauna kea , hawaii , uses a 1024@xmath301024 insb array for its spectroscopic observations ( @xcite ) .
we used the prism mode with a 0@xmath315 wide slit to achieve a resolving power of @xmath39 over the range 0.8 - 2.5 @xmath1 m .
a series of 120s or 180s exposures were typically obtained at two different nod positions along the 15@xmath35 long slit .
a0 dwarf stars at similar airmass to the target were observed near in time for telluric correction and flux calibration .
observations were typically obtained with the slit oriented to the parallactic angle to minimize slit loses and spectral slope variations due to differential atmospheric refraction .
finally , a set of exposures of internal flat field and argon arc lamps were obtained for flat fielding and wavelength calibration .
the data were reduced using spextool ( @xcite ) the idl - based data reduction package for spex .
the raw images were first corrected for non - linearity , pair subtracted , and then flat fielded .
for some of the fainter sources , multiple pair - subtracted images were averaged in order to facilitate tracing .
the spectra were then optimally extracted ( e.g. , @xcite ) and wavelength calibrated using the argon lamp exposures .
multiple spectra were then averaged and the resulting spectrum was corrected for telluric absorption and flux calibrated using observations of an a0 v star using the technique described in @xcite . the near - infrared spectrometer ( nirspec , @xcite ) at the 10 m w. m. keck observatory on mauna kea , hawaii , uses a 1024@xmath301024 insb array for spectroscopy . in low - resolution mode , use of the @xmath40 slit results in a resolving power of r @xmath41 . our brown dwarf candidates were observed in either or both of the n3 and n5 configurations ( see @xcite ) that cover the portion of the @xmath21-band window from 1.15 to 1.35 @xmath1 m and the portion of the @xmath22-band window from 1.5 to 1.8 @xmath1 m .
data were typically obtained in four or more sets of dithered pairs , with a 300-second exposure obtained at each dither position . to measure telluric absorption and to calibrate the flux levels ,
a0 dwarf stars were observed near in time and airmass to the target .
other calibrations consisted of neon and argon arc lamp spectra , dark frames , and spectra of a flat - field lamp .
we employed standard reductions using the redspec package , as described in @xcite .
the folded - port infrared echellette ( fire ; @xcite , @xcite ) at the 6.5 m magellan baade telescope uses a 2048@xmath302048 hawaii-2rg array . in its high - throughput prism mode
, it covers the wavelength range from 0.8 to 2.5 @xmath1 m at a resolution ranging from r=500 at @xmath21-band to r=300 at @xmath34-band for a slit width of 0@xmath316 .
typically , each observation used the 50@xmath33-long slit aligned to the parallactic angle , and consisted of a series of nod pairs taken with exposure times not exceeding 120s per position .
the spectrograph detector was read out using the 4-amplifier mode at `` high gain '' ( 1.2 counts per e@xmath42 ) with the fowler-8 sampling mode .
we also obtained exposures of a variable voltage quartz lamp ( set at 1.2 v and 2.2 v ) for flat fielding purposes and neon / argon arc lamps were used for wavelength calibration .
a0 dwarf stars were used for telluric and flux calibration .
data were reduced using a modified version of spextool .
in contrast to spex , the spatial axis of the fire slit is not aligned with the columns of the detector so that the wavelength solution is not only a function of the column number but also of the row number .
we therefore derived a two - dimensional wavelength solution in two steps .
first , a one - dimensional wavelength solution applicable to the center of the slit was determined using the night - sky oh emission lines ( @xcite ) .
second , the oh emission lines were used to map the optical distortions in the spatial and dispersion axes within each order .
the 1d wavelength solution and distortion maps were then combined to assign a wavelength and spatial position to each pixel in each order . with the wavelength calibration completed
, the remainder of the reduction steps could proceed .
pairs of images taken at different positions along the slit were subtracted in order to remove the bias and dark current as well as to perform a first - order sky subtraction .
the resulting pair - subtracted image was then flat fielded using a normalized flat constructed from dome flats taken at the start of each night .
the spectral extraction followed the technique described in @xcite .
a set of `` pseudorectangles '' was defined spanning each order to map out positions of constant wavelength on the detector .
these rectangles are themselves composed of pseudopixels with a width and height of @xmath51 detector pixel .
the pseudorectangles were then extracted using a polygon clipping algorithm ( @xcite ) , producing 1d profiles of the slit at each wavelength .
spectral extraction including residual background subtraction could then proceed in the standard way .
the raw spectra were then combined and corrected for telluric absorption and flux calibrated using the observations of an a0 v star and the technique described in @xcite .
the triple spectrograph ( tspec ) at the 5 m hale telescope at palomar observatory uses a 1024@xmath302048 hawaii-2 array to cover simultaneously the range from 1.0 to 2.45 @xmath1 m ( @xcite ) . with a 1@xmath3030-arcsecond slit
, it achieves a resolution of @xmath52700 .
observations were acquired in an abba nod sequence with an exposure time per nod position not exceeding 300s to mitigate problems with changing oh background levels .
observations of a0 dwarf stars were taken near in time and near in airmass to the target objects and were used for telluric correction and flux calibration .
dome flats were taken to calibrate the pixel - to - pixel response .
data reduction is identical to that discussed above for fire because spextool required the same changes for tspec reductions as it did for fire reductions .
the ohio state infrared imager / spectrometer ( osiris ) mounted at the 4.1 m southern astrophysical research telescope ( soar ) located at cerro pachn , chile , uses a 1024@xmath301024 hgcdte array .
the 1@xmath310-wide slit yields a resolving power of @xmath43 across the 1.18 - 2.35 @xmath1 m wavelength range in three spectral orders .
short exposures ( 180s ) were taken at six or seven positions nodded along the 24@xmath35-long slit .
a0 dwarf stars were observed for telluric correction and flux calibration .
wavelength calibration was based on the oh airglow lines .
the data were reduced using a modified version of the spextool data reduction package .
( see [ irtf - section ] for a description of spextool . )
the three spectral orders were then merged into a single spectrum covering the entire wavelength range . the wide field camera 3 ( wfc3 ) onboard the _ hubble _ space telescope employs a 1024@xmath301024 hgcdte detector with a plate scale of 0@xmath3113 pixel@xmath32 to image a field of view of 123@xmath30126 arcseconds .
we used the g141 grism to acquire slitless spectroscopy over the 1.1 - 1.7 @xmath1 m range for faint targets not observable from ground - based facilities ( program 12330 , pi kirkpatrick ) .
the resulting resolving power , @xmath14@xmath9130 , is ideal for the broad characterization of faint sources .
three spectra those of wisepc j145018.40 + 553421.4 , wisepa j173835.53 + 273258.9 , and wisepa j182831.08 + 265037.8 were obtained with this setup . for each target
, we first obtained four direct images through the f140w filter in multiaccum mode using the spars25 sampling sequence .
the telescope was offset slightly between each exposure .
we then obtained four dispersed images with the g141 grism using mutliaccum mode and a spars50 sequence .
these dispersed images were acquired at the same positions / dithers as used in the direct images .
bias levels and dark current were first subtracted from the data images using version 2.3 of the calwfc3 pipeline .
calwfc3 also flat fields the direct images , but the grism images are flat fielded during the extraction process described below .
spectra were then extracted using the axe software ( @xcite ) . because axe requires knowledge of the position and brightness of the targets in the field of view , we combined the four direct images using multidrizzle ( @xcite ) and the latest instrument distortion coefficient table ( idctab ) .
sextractor ( @xcite ) was then used to produce a catalog of objects in the field .
for each cataloged object , two - dimensional subimages centered on the first - order spectra of each object were combined using axedrizzle to produce a final spectral image .
these subimages were used to extract flux - calibrated spectra .
additional details on the data reduction process are discussed in cushing et al .
( accepted ) .
the list of spectroscopically confirmed wise brown dwarfs is given in table [ spectral_types ] .
abbreviated source namesddmm '' .
full designations can be found in table [ wise_photometry_discoveries ] . ]
are shown in column 1 ; optical spectral types are shown in column 2 ; near - infrared types are shown in column 3 ; and the source of the spectrum , integration time , telluric corrector star ( for ground - based observations ) , and observation date are shown in columns 4 through 7 . the optical and near - infrared classifications of these sources are discussed further below .
keck / lris spectra were obtained for seven of our candidates .
reduced spectra from 8000 to 10000 are shown in figure [ optical_spectra ] , all of which have been corrected for telluric absorption over the regions 68677000 ( the fraunhofer b band , caused by o@xmath27 absorption ) , 75947685 ( the fraunhofer a band , again caused by o@xmath27 ) , and 71867273 , 81628282 , and @xmath589509650 ( all caused by h@xmath27o absorption ) .
these spectra show the hallmarks of t dwarf optical spectra : strong h@xmath27o absorption with a bandhead at 9250 , along with absorption at 8521 and 8943 in the earlier objects and ch@xmath8 absorption between 8800 and 9200 in the later objects .
by - eye comparisons to the t dwarf optical spectral standards of @xcite show that these spectra range in type from t@xmath445 for wise 1841 + 7000 to later than t@xmath448 for wise 1741 + 2553 .
( the `` o '' subscript is used to denote spectral types assigned based on optical spectra . )
this latter spectrum is unusual in that it has stronger 8800 - 9200 ch@xmath8 , stronger 9250 - 9400 h@xmath27o bands of the 3(@xmath45,@xmath46 ) transition , and stronger 9450 - 9800 h@xmath27o bands of the 2(@xmath45,@xmath46 ) + 2(@xmath47 ) transition than the latest t dwarf optical standard , the t@xmath448 dwarf 2mass j04151954@xmath20935066 ( @xcite ) .
we therefore propose that wise 1741 + 2553 be the spectral standard for a newly adopted t@xmath449 spectral class .
figure [ optical_spectral_sequence ] illustrates the entire sequence of t dwarf optical standards from @xcite appended with the proposed t@xmath440 standard sdssp [email protected] from @xcite and this newly proposed t@xmath449 standard .
further evidence in support of a new optical standard is shown in figure [ optical_t9s ] . shown here
is a comparison of the spectrum of wise 1741 + 2553 with our lris spectrum of the bright , late - t dwarf ugps [email protected] , which cushing et al .
( accepted ) have proposed as the infrared spectral standard for type t9 . as our comparison shows ,
the ch@xmath8 and h@xmath27o depths are very similar between these two objects and both are distinctly different from the t@xmath448 standard .
thus , identifying wise 1741 + 2553 as the new , t@xmath449 anchor point would help link the optical and near - infrared sequences , especially since wise 1741 + 2553 is also classified as a t9 on the near - infrared scheme .
see further discussion in the appendix .
all of the confirmed brown dwarfs listed in table [ spectral_types ] have near - infrared follow - up spectra .
these spectra were classified using the the near - infrared m and l sequences of @xcite and the near - infrared t0-to - t8 dwarf sequence of @xcite .
cushing et al .
( accepted ) have extended classifications to t9 and y0 and have reclassified previously discovered @xmath4t8 dwarfs from the literature on this system .
this extension of the classification system uses ugps 0722@xmath20540 as the near - infrared t9 standard and wise 1738 + 2732 as the y0 standard .
assignment of spectral types was done by overplotting spectra of these standards onto the candidate spectra and determining by - eye which standard provided the best match . in some cases two adjacent standards , such as t7 and t8 , provided an equally good match ,
so the candidate spectrum was assigned an intermediate type , in this case , of t7.5 . for l dwarfs ,
the comparison was done at @xmath21-band and , following the prescription discussed in @xcite , any anomalies at @xmath22- and @xmath34-bands were noted .
spectra that did not match any of the standards well are marked with a `` pec '' suffix to indicate that they are peculiar . as a further example , an object that best fit the l9 spectral standard at @xmath21-band but failed to provide a good match to the l9 standard at longer wavelengths because it was considerably redder than the standard was assigned a type of `` l9 pec ( red ) . ''
see @xcite for examples of similar classifications . in figure [ other_spectra_lt_t0 ] through figure [ other_spectra_gt_t8 ]
, we show the near - infrared spectra for each of our sources . because of the narrow wavelength ranges covered by the keck / nirspec and soar / osiris spectra ,
those data are plotted separately in figure [ nirspec_spectra ] and figure [ osiris_spectra ] .
near - infrared spectral types ( and optical spectral types , for those with keck / lris spectra ) are listed in table [ spectral_types ] for all wise brown dwarf discoveries . for objects with near - infared spectral types of t0 or later ,
figure [ spec_type_histogram ] shows the number of newly discovered objects per spectral type bin compared to the number of objects previously published . whereas there were 16 objects known previously with types of t8 or later ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite )
, the tally now stands at 58 once our objects are added .
wise has already identified seventeen new objects with types equal to or later than the t9 ugps [email protected] , the previous record holder for latest measured spectral type , and six of these belong to the y dwarf class ( cushing et al . ,
accepted ) .
figure [ w1w2_vs_type ] through figure [ w1w2_vs_w2w3 ] and figure [ jh_vs_type ] through figure [ hch2_vs_type ] , discussed previously , show the locations of these newly discovered wise brown dwarfs ( black symbols ) in color space .
the t9 , t9.5 , and early - y dwarfs continue the trend toward redder w1-w2 , ch1-ch2 , @xmath21-w2 , @xmath22-w2 , @xmath21-ch2 , and @xmath22-ch2 colors , with the reddest object being the @xmath4y0 dwarf wise 1828 + 2650 ( j - w2 = [email protected] mag ; @xmath21 on the mko filter system ) .
the blueward trend in @xmath24 color seen for later t dwarfs , however , begins to reverse near a spectral type of y0 .
in particular , the @xmath24 color of wise 1828 + 1650 is dramatically redder ( @xmath48 mag ; mko filter system ) than any of the late - t or y0 dwarfs , the latter of which show a large scatter in @xmath24 colors themselves . cushing et al .
( accepted ) explore the trend of @xmath24 colors in more detail and show that the synthetic photometry derived from our observed spectra generally agree with photometry measured from direct imaging . given
the large spread in @xmath24 color observed for the six y dwarfs already identified , @xmath49 colors alone can not be used to confirm or deny objects as cold as y dwarfs .
distances to the new wise brown dwarf discoveries can be estimated based on their w2 magnitudes and measured spectral types .
first , however , the relation between absolute w2 magnitude and spectral type needs to be established using objects with measured trigonometric parallaxes and wise w2 detections .
figure [ mw2_vs_type ] shows the trend of absolute w2 magnitude as a function of near - infrared spectral type for previously published objects whose measured parallaxes are at least three times the measurement error ( table [ abs_mags ] ) .
a third - order least squares relation , weighted by the errors on the @xmath50 values , is shown by the black curve in figure [ mw2_vs_type ] . for this fit , objects known to be binary ( red points ) have been omitted .
the resulting relation is @xmath51 where @xmath52 is the near - infrared spectral type on the system where l0=0 , l5=5 , t0=10 , t5=15 , and y0=20 . using this relation ,
we have estimated distances to our wise discoveries .
these are given in column 2 of table [ astrometry ] .
( the distance to the lone m dwarf , wise 0106 + 1518 , was estimated using 2mass magnitudes and the near - infrared absolute magnitudes listed in table 3 of @xcite . )
these distance estimates for the late - t dwarfs and y dwarfs are shown graphically in figure [ distances ] .
also shown in the figure are previously published late - t dwarfs from other surveys .
wise has sufficient sensitivity to detect the latest t dwarfs out to 15 to 20 pc and because of its all - sky coverage can complete the census of the solar neighborhood for these objects . as the figure shows , twelve of our objects have estimated distances placing them within 10 pc of the sun , and two of these have estimates placing them within 5 pc .
it should also be noted that the fitted relation shown in figure [ mw2_vs_type ] may lead to overestimated distances for objects at the latest types because the relation falls above all four of the previously published t8.5 and t9 dwarfs on that plot .
furthermore , the extrapolation of this relation to even later y dwarf types may lead to even more discrepant distance overestimates , as discussed further in the caption to figure [ mw2_vs_type ] .
measuring trigonometric parallaxes for more of these latest t dwarfs and early y dwarfs will be an important , early step in characterizing the physical nature of these objects . because these objects should all lie very close to the sun , their observed parallaxes will be large .
thanks to its survey strategy , wise performed its two passes of the sky with observations always near 90@xmath53 solar elongation , thus capitalizing on the maximum parallactic angle at both epochs .
( for objects observed during the final @xmath52 weeks of wise operations , three epochs of wise data are available . )
objects will , of course , also show displacements due to proper motion , so observations at other epochs and/or from other surveys are necessary to disentangle the two effects .
hence , ancillary astrometry from 2mass and sdss and our own follow - up observations from the ground and from space ( _ spitzer _ and _ hst _ ) are invaluable .
currently available astrometric data points are shown in columns 3 through 8 of table [ astrometry ] .
it should be noted here that positions of objects in the wise preliminary data release may be offset from their true positions by many times the quoted positional uncertainty .
approximately 20% of the sources fainter than [email protected] mag in the preliminary release source catalog suffer from a pipeline coding error that biases the reported position by @xmath50.2 - 1.0 arcsec in the declination direction while an increasingly smaller fraction of the sources suffer this effect to magnitudes as bright as [email protected] mag .
the cautionary notes section of the wise preliminary release explanatory supplement describes the origin and nature of this effect in detail . for this paper
, we have rerun the wise images for our sources through a version of the wise pipeline that eliminates this source of systematic error , and we list those remeasured positions in table [ astrometry ] .
this version of the pipeline is essentially the same one used to process data for the wise final data release .
astrometric fits were made to the multiple observations of each source .
these fits solved for five parameters : initial ( time = @xmath54 ) positional offsets of @xmath55 and @xmath56 in right ascension ( @xmath57 ) and in delination ( @xmath58 ) , the right ascension component of proper motion ( @xmath59 ) , the declination component of proper motion ( @xmath60 ) , and the parallax ( @xmath61 ) . for all but four sources , the data were not sufficient to find an accurate distance , so the distance was forced to equal the spectrophotometric estimate , and the fit only solved for the first four parameters ; the four sources with a preliminary parallax measurement are listed in table [ prelim_parallaxes ] and are discussed individually in the appendix .
the equations used are @xmath62 @xmath63 subscript @xmath19 refers to the individual astrometric measurements , where @xmath54 is the observation time in years , and @xmath64 is the vector position of the observer relative to the sun in celestial coordinates and astronomical units . @xmath65 and @xmath66 are unit vectors pointing north and west from the position of the source . @xmath64 is the position of the earth for 2mass , sdss , wise , and _ hst _ observations ; for _ spitzer _ observations , @xmath64 is the position of the spacecraft .
the observed positional difference on the left hand side is in arcseconds , the parameters @xmath55 and @xmath56 are in arcseconds , the proper motion @xmath67 and @xmath68 are in arcseconds / year , and the parallax @xmath61 is in arcseconds .
these equations are solved using standard weighted least - squares techniques , which also provide the uncertainties in the parameters .
these uncertainties come from propagating the uncertainties in the input .
the @xmath69 and number of degrees of freedom are also given and can be used to assess the quality of the fit .
the resulting proper motions in ra and dec are listed in columns 9 and 10 of table [ astrometry ] . for motions with a significance of @xmath70 ,
the total proper motion is listed in column 11 along with the tangential velocity in column 12 .
the brown dwarf discoveries presented here represent only a fraction of the brown dwarf candidates identified so far from wise data .
wise coadded data are not available across the entire sky , and many of those coadds do not reach the full survey depth . nonetheless , we can use these preliminary results to assess our progress toward completing the tally of cold brown dwarfs in the solar neighborhood and gauging the functional form of the mass function for these objects by using lower limits to their space densities .
our goal is to complete an all - sky census of objects out to a specified maximum distance for each spectral subtype of t6 or later .
table [ space_density_numbers ] divides our discoveries into six spectral type bins ( column 1 ) from t6 through @xmath4y0 .
the approximate range in effective temperature is given for each bin in column 2 .
these temperature bins are assigned as follows .
we took the values of @xmath7 for t dwarfs of type t6 and later as computed by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , bochanski et al .
( submitted ) , and cushing et al .
( accepted ) or compiled by @xcite .
then , when necessary , we re - assigned spectral types to these objects so that they matched the near - infrared spectral classification scheme of @xcite or its extension beyond t8 by cushing et al .
( accepted ) .
we then examined the distribution of temperature within each integral spectral type bin and found that a 150k width for each bin was enough to encompass the @xmath7 values for most of the objects .
the final assignments are given in column 2 of table [ space_density_numbers ] .
as shown in figure [ distances ] , the depth of our current search translates to different distances for each bin . in the spirit of determining the space density using a well defined census of the solar neighborhood , we limit our sample to those objects falling within 20 pc of the sun even if wise can sample that spectral type to larger distances . only in the last three bins the t9-t9.5 bin and
the y dwarf bins is wise incomplete at this distance , so those bins are limited to volumes with smaller radii .
these values , called @xmath71 , are listed in column 3 .
next , the number of objects per spectral type bin lying closer than the value of @xmath71 is tabulated for previously published objects ( column 4 ) , for our new wise discoveries ( column 5 ) , and in total ( column 6 ) .
distances are determined using trigonometric parallaxes , if available , or spectrophotometric estimates if no parallax has been measured .
the resulting space density in each bin is given in column 7 .
this simple calculation of the space densities can be overestimated for the following reasons : \(1 ) spectrophotometric distance estimates have an inherent bias .
the absolute magnitude versus spectral type relation is based on parallaxes , and those parallax measurements lead to a bias in estimated distances because a parallax value of @xmath72 is more likely to represent an object further away ( @xmath73 ) than an object closer ( @xmath74 ) because the volume of space between parallax values of @xmath61 and @xmath73 is larger than that sampled between parallax values of @xmath61 and @xmath74 .
thus , the observed values of @xmath61 are larger than the true values and the measured absolute values will be systematically too large .
a correction can be applied that depends only on the value of @xmath75 ( see table 1 of @xcite ) .
most of the parallaxes in table [ abs_mags ] have @xmath75 values of less than 5% where the correction to the absolute magnitude is @xmath760.02 mag , and those with larger errors have already been downweighted in our fit .
we therefore conclude that the lutz - kelker effect is negligible here .
\(2 ) both the malmquist bias and the eddington bias can be largely accounted for by limiting the sample over which we derive our space densities .
malmquist bias ( @xcite ) , in which more luminous objects can preferentially bias statistics in a magnitude limited sample , can be eliminated by calculating space densities in narrow spectral type bins in which all objects have the same ( or nearly the same ) intrinisic luminosity .
the eddington bias ( @xcite ; @xcite ) , in which random errors will bias magnitude measures to brighter values due to the fact that there are more objects in the more distant ( fainter ) population than in the closer ( brighter ) one , can be reduced by operating at magnitudes where the random errors are still small . by using the brighter and better measured w2 values to estimate distances
, we can reduce the effects of eddington bias on our derived densities .
( for further discussion , see also @xcite . )
\(3 ) unresolved binarity will cause distances to be underestimated .
this may cause a more distant object to appear closer than it really is and falsely inflate the space density .
empirical data presented earlier can be used to estimate this degree of binary contamination .
figure [ mw2_vs_type ] shows a well - defined binary sequence ( red ) overlying the sequence of single objects on the @xmath50 versus spectral type diagram .
if we compute the ratio of known binaries to total objects between l0 and t4 , we find that 12/35 , or 34% , are binary .
( see also section 7.4 of @xcite for an in - depth discussion of instrinic binarity , which varies from @xmath520% at early - l to @xmath542% at the l / t transition . )
while this is a sizable percentage , it does not mean that all of the unresolved binaries fall outside of the sample considered .
some small fraction will still be contained within the distance limit and will have been undercounted by a factor of two .
nonetheless , binarity is likely the largest contributor to inflating density estimates . on the other hand , other biases discussed below lead to an underestimate in the space densities .
these effects are believed to overwhelm the effects detailed above , and hence our simple space density calculations , although preliminary , can be considered as lower limits to the true densities : \(1 ) although wise has taken data covering the entire sky at multiple epochs , the available coadded data cover less than 75% of the entire sky .
also none of the sky has been coadded to its full depth using all available frames and the detection threshold for first - pass processing was set higher , in units of signal - to - noise ratio , than it will be for final processing .
the latter points are particularly important as they will , in the future , enable more robust colors or color limits for potential y dwarf candidates .
moreover , we have followed up less than 50% of the brown dwarf candidates already culled from sections of the sky for which we have access .
thus , we believe that our current space density estimates are gross underestimates .
\(2 ) except for 2mass , other surveys providing data in column 4 of table [ space_density_numbers ] do not have all - sky coverage and can only provide limited help in completing this nearby sample . moreover , none of the current or planned ground - based surveys canvassing the sky for brown dwarfs can reach sizable populations of the coldest objects because y dwarfs are intriniscally dim at ground - observable wavelengths
this is highlighted in figure [ mh_vs_type ] , which shows the absolute @xmath22-band magnitude as a function of spectral type .
note that the y0 dwarf wise 1541@xmath22250 has @xmath77 mag ; its absolute magnitude is @xmath78 mag and it is presumably even fainter than this shortward of @xmath21 band .
wise operates at wavelengths where these objects are their brightest five thousand times brighter at w2 than at @xmath21 and @xmath22 bands , in the case of wise 1828 + 2650 so it is the only survey capable of detecting the coldest brown dwarfs in significant numbers .
\(3 ) as mentioned earlier , the @xmath50 vs. spectral type relation of figure [ mw2_vs_type ] likely overpredicts distances to dwarfs of type @xmath0t9 .
this means that our surveyed volume may be overestimated , leading to an underestimate of the space density .
\(4 ) despite the all - sky covereage of wise , the galactic plane will restrict our ability to probe to the same depths as other parts of the sky due to higher backgrounds and confusion .
this loss of coverage is not currently accounted for in our density estimates .
we have checked the distance distribution of objects in each spectral type bin by performing the @xmath79 test ( column 8 of table [ space_density_numbers ] ) .
this test was first proposed by @xcite to check the uniformity of a distribution of objects in space .
the quantity @xmath80 is the volume of space interior to object @xmath19 at distance @xmath81 , and @xmath82 is the full volume of space contained within the distance limit , @xmath71 , of the sample . for a uniform sample ,
the average value , @xmath83 , should be 0.5 because half of the sample should lie in the nearer half of the volume and the rest should lie in the farther half . if this number is not near 0.5 , then the sample is either non - isotropic or incomplete . for our sample
we find that the t dwarf bins have @xmath83 values considerably less than 0.5 .
this points to incompleteness in the sample our low @xmath83 values are almost certainly a consequence of the fact that the brighter ( closer ) candidates tend to be followed up first .
this gives further credence to the assertion that our number densities are lower limits only . for the y dwarf bins , however , the values of @xmath83 are above 0.5 , which is further suggestive that our assumed distances to these objects are overestimates .
figure [ space_density ] shows these preliminary results ( column 9 of table [ space_density_numbers ] ) relative to measurements made by other surveys and relative to predictions based on different forms of the underlying mass function .
previous results by @xcite , @xcite , and @xcite are shown by the open symbols and , with the possible exception of the @xcite point , support values of @xmath57 of zero or greater , where the functional form is given as @xmath84 .
our incomplete , volume - limited sample so far fails to put tighter constraints on the mass function at warmer temperatures than previously published work , but the preliminary lower limit to the y0 space density already rules out the @xmath85 model and may soon , with additional y0 discoveries , be able to distinguish between the @xmath86 and @xmath87 models .
results for the early - y dwarfs already suggest that the low - mass cutoff of star formation must be below 10 @xmath88 if @xmath89 .
this result is in accordance with findings in young star formation regions ( e.g. , @xcite , @xcite , @xcite , @xcite ) , but the derived masses for those objects should be considered cautiously , as discussed in @xcite , because models with ages of @xmath51 myr or younger are highly sensitive to untested assumptions about initial conditions . using models of older ages is far less sensitive to these assumptions , so field brown dwarfs provide a better check of star formation s low - mass cutoff .
model fits by cushing et al .
( accepted ) suggest that our y dwarf discoveries have masses as high as 30 @xmath88 or as low as 3 @xmath88 or less , which agrees roughly with the mass values inferred from figure [ space_density ] .
this paper represents the culmination of a year s worth of effort following up the first batch of brown dwarf candidates identified by wise .
there are many hundreds more candidates still being scrutinized , and there are still areas of sky not yet searched .
it is therefore clear that these first hundred brown dwarf discoveries are harbingers of a much larger trove of brown dwarfs yet to be uncovered by wise .
not only is the wise data archive uniquely suited to finding even colder objects than the current batch of early - y dwarfs , the all - sky and multi - epoch nature of the mission will enable many other brown dwarf studies the search for the lowest mass objects in nearby moving groups , hunting for low - metallicity objects via their high proper motions , etc .
that are well beyond the scope of the photometric search presented here .
this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration .
we acknowledge fruitful discussions with tim conrow , roc cutri , and frank masci , and acknowledge assistance with magellan / fire observations by emily bowsher .
this publication also makes use of data products from 2mass , sdss , and ukidss .
2mass is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
sdss is funded by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
ukidss uses the wide field camera at the united kingdom infrared telescope atop mauna kea , hawaii .
we are grateful for the efforts of the instrument , calibration , and pipeline teams that have made the ukidss data possible .
we acknowledge use of the dss , which 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 oschin schmidt telescope on palomar mountain and the uk schmidt telescope .
this research has made use of the nasa / ipac infrared science archive ( irsa ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
our research has benefitted from the m , l , and t dwarf compendium housed at dwarfarchives.org , whose server was funded by a nasa small research grant , administered by the american astronomical society .
we are also indebted to the simbad database , operated at cds , strasbourg , france .
this work is based in part on observations made with the _ spitzer _ space telescope , which is operated by the jet propulsion laboratory , california institute of technology , under a contract with nasa .
support for this work was provided by nasa through an award issued to program 70062 by jpl / caltech .
this work is also based in part on observations made with the nasa / esa _ hubble _ space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
these observations are associated with program # 12330 .
support for program # 12330 was provided by nasa through a grant from the space telescope science institute .
some of the spectroscopic data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m . keck foundation . in acknowledgement of our observing time at keck and the irtf
, we further wish to recognize the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
we acknowledge use of pairitel , which is operated by the smithsonian astrophysical observatory ( sao ) and was made possible by a grant from the harvard university milton fund , the camera loaned from the university of virginia , and the continued support of the sao and uc berkeley .
the pairitel project is supported by nasa grant nng06gh50 g .
this paper also includes data gathered with the 6.5 m magellan telescopes located at las campanas observatory , chile .
notes are given below for objects with unusual spectra , spectrophotometric distance estimates placing them within 10 parsecs of the sun , spectral types later than t9 , or possible companionship with a nearby object previously cataloged .
in the sections below , the spectra of objects are assumed to be normal unless peculiarities are specifically mentioned .
the @xmath21-band spectrum of this object , which was earlier cataloged as sips j0031@xmath23840 and identified to be nearby star via its high proper motion by @xcite , is a good match to the spectrum of the l2 standard ( figure [ 0031m3840 ] ) , but the spectrum is much bluer than the standard at @xmath22 and @xmath34 bands .
because this source does not exhibit other telltale signs of low metallity , such as strong hydride bands , we classify it as an `` l2 pec ( blue ) '' . @xcite
give an optical spectral type of l2.5 and do not note any peculiarities that might be attributable to low metallicity , either .
its near - infrared spectral morphology and discrepancy relative to its nearest standard is most similar to the `` l1 pec ( sl .
blue ) '' object 2mass j14403186@xmath21303263 shown in figure 32 of @xcite .
furthermore , this object has a larger w1-w2 color than a typical l2 ( see figure [ w1w2_vs_type ] ) and a bluer @xmath24 color ( see figure [ jh_vs_type ] ) . as discussed in @xcite , the physical interpretation of these `` blue l dwarfs ''
is not fully known and may differ from object to object . some blue l dwarfs are blue , for example , because they are composte l + t dwarf binaries ( e.g. , @xcite ) .
model fitting suggests that others have thin cloud decks and/or large grains in their atmosphere , though neither seems to be directly attributable to gravity or metallicity effects ( @xcite ) .
kinematic analysis by @xcite has shown that the blue l dwarfs have kinematics older than the field l dwarf population , but not nearly as old as that of low - metallicity m dwarfs .
it is possible that some of the blue l dwarfs may be slightly metal poor , and that even a subtle lowering of the metal abundance in these objects may result in the directly measurable effects on the spectral energy distribution seen here .
despite the fact that the spectrum of this object matches very well to the @xmath21-band spectrum of the m8 standard , the @xmath22-band spectrum is more peaked than that seen in a normal m8 , with the h@xmath27o bands on either side of the @xmath22 peak being stronger than in the standard ( figure [ 0106p1518 ] ) .
this object , which we classify as `` m8 pec '' , is similar to the peculiar late - m dwarf 2mass j18284076 + 1229207 shown in figure 38 of @xcite .
the cause of the peculiarity is not known , but appears not to be due to low gravity , as there are no peculiarities in the strength of the feh bands between 0.9 and 1.3 @xmath1 m when compared to the m8 standard .
( see , for example , near - ir spectra of low - gravity late - m dwarfs in figure 14 of @xcite . )
this object also has a larger w1-w2 color than a typical m8 ( see figure [ w1w2_vs_type ] ) and a bluer @xmath24 color ( see figure [ jh_vs_type ] ) .
curiously , this object has a sizable motion @xmath1 = [email protected] arcsec / yr and a tangential velocity of [email protected] km / s , suggesting that it may belong to an old population .
the peculiar spectroscopic features may be caused in part by a slightly subsolar metallicity .
@xcite also identified this object as the high motion source ulas2mass j0106 + 1518 , and their proper motion determination ( @xmath90 arcsec / yr ) agrees with the one we derive here .
the near - infrared spectrum of this object , discussed in cushing et al .
( accepted ) , is distinctly later in type than the t9 near - infrared standard , ugps [email protected] , and is therefore classified as t9.5 .
our spectrophotometric distance places it at 12.1 pc ( table [ astrometry ] ) .
this is the only one of our @xmath0t9.5 discoveries not detected in w3 and along with wise 1738 + 2732 is one of only two @xmath0t9.5 dwarfs detected in w1 ( table [ wise_photometry_discoveries ] ) .
the @xmath21-band spectrum of this object closely matches that of the l9 spectral standard , but the @xmath22- and @xmath34-band portions are much redder than those of the standard l9 ( figure [ 0206p2640 ] ) .
this extremely red color is supported by independent photometry , namely @[email protected] mag , from the 2mass all - sky point source catalog .
this color is somewhat redder than the mean @xmath91 color , @xmath51.78 mag , of very late l s ( figure 14 of @xcite ) . because of its spectral peculiarity
, we classify this object as `` l9 pec ( red ) '' , and add it to the growing list of l dwarfs that appear red for reasons not obviously attributable to low gravity ( see table 6 of @xcite ) . the underlying physical cause for these `` red l dwarfs ''
is not known , although two have been studied in detail by @xcite . using the small sample of red l dwarfs
then known , @xcite found that these objects , unlike young , low - gravity l dwarfs that are also redder than spectral standards of the same type , appear to have older kinematics than that of the field l dwarf population .
this object is a nearby t8 dwarf at a spectrophotometric distance of d=6.9 pc .
our astrometry ( table [ astrometry ] ) over a 10.4yr baseline indicates a high motion of @xmath1 = [email protected] arcsec / yr and a large tangential velocity of [email protected] km / s .
our preliminary trigonometic parallax measurement ( table [ prelim_parallaxes ] ) places this object at @xmath92 pc , in excellent agreement with the spectrophotomeric estimate .
this t8.5 dwarf has a spectrophotometric distance estimate of only 8.1 pc .
the near - infrared spectrum of this object , discussed in cushing et al .
( accepted ) , is classified as y0 .
our distance estimate places it 9.0 pc from the sun , and our measurement of the proper motion indicates that it may also be a high mover @xmath1 = [email protected] arcsec / yr although the error bar is large ( table [ astrometry ] ) . as with most of the other y dwarf discoveries , this object is detected by wise only in bands w2 and w3 and not in w1 or w4 ( table [ wise_photometry_discoveries ] ) .
the depths of the h@xmath27o and ch@xmath8 absorption bands in the @xmath21- and @xmath22-band spectra of this object best fit the t5 standard ; however , the spectrum shows excess flux in the @xmath29 band around 0.95 to 1.10 @xmath1 m and a flattening of the entire @xmath34-band spectrum ( figure [ 0448m1935 ] ) .
we therefore classify this object as a `` t5 pec '' . excess flux at @xmath29 band and a flattening at @xmath34
have also been noted in @xcite for the t6 pec dwarf 2mass j09373487 + 2931409 , which may be slightly metal - poor ( @xmath93 \approx -0.5 $ ] to @xmath94 ) based on fits to model spectra . @xcite and @xcite have noted the same peculiarities in the spectrum of the t7.5 dwarf ulas j141623.94 + 134836.3 ( @xcite ; @xcite ) , which is a common proper motion companion to the nearby , late - l dwarf sdss j141624.08 + 134826.7 ( @xcite ; @xcite ) .
the latter is classified by @xcite as sdl7 and by @xcite as d / sdl7 .
given that the two objects are presumably coeval , it can be assumed that they have the same metallicity and that the peculiar features in the spectrum of ulas j141623.94 + 134836.3 and by extension , wise 0448@xmath21935 are caused by a metal content below solar .
the high motion of wise 0448@xmath21935
@xmath1 = [email protected] arcsec / yr , which translates into a tangential velocity of [email protected] km / s for an estimated distance of 21.4 pc also indicates that this object belongs to an old kinematic population .
this object was discussed in detail by @xcite .
our distance estimate of 7.3 pc assumes a single source , but analysis of laser guide star adaptive optics data from @xcite indicates that the source is a double in which the components have @xmath95 mag .
individual magnitudes are measured as @xmath96 , @xmath97 and @xmath98 , @xmath99 on the mko filter system . @xcite
suggest an actual distance to the system of [email protected] pc and individual spectral types of t8.5 and t9 .
the @xmath21-band spectrum of this object best fits the t6 standard , but there are discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 0525 + 6739 shows excess flux relative to the standard , and at @xmath34 band the spectrum shows less flux relative to the standard ( figure [ 0525p6739 ] ) .
we therefore classify this object as a `` t6 pec '' . the physical cause ,
as discussed above for wise 0448@xmath21935 , may be low metal content .
the @xmath21-band spectrum of this object best fits the t7 standard , but there are discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 0528@xmath23308 shows excess flux relative to the standard , and at @xmath34 band the spectrum shows less flux relative to the standard ( figure [ 0528m3308 ] ) .
we therefore classify this object as a `` t7 pec '' . the physical cause , as discussed above for both wise 0448@xmath21935 and wise 0525 + 6739 , may be low metal content .
the @xmath21-band spectrum of this object best fits the t8 standard , but there are major discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 0836@xmath21859 shows excess flux relative to the standard , and at @xmath34 band the spectrum shows less flux relative to the standard ( figure [ 0836m1859 ] ) .
we therefore classify this object as a `` t8 pec '' .
several other objects discussed in this section wise 0448@xmath21935 , wise 0525 + 6739 , wise 0528@xmath23308 , wise 1436@xmath21814 , wise 2134@xmath27137 , and wise 2325@xmath24105 have similar @xmath29- and @xmath34-band discrepancies , which may result from low metal content , but none are as severe as the discrepancies in this object . unlike in those objects ,
the @xmath22-band flux in wise 0836@xmath21859 is markedly lower than the closest matching standard at @xmath21 , making this object the most peculiar one of the group , and perhaps also the most metal poor . ideally , we would like to study the frequency of these metal - poor t dwarfs to see
if the numbers found are what star formation theory would predict for an old , field population .
unfortunately , we do not have cold models across a large grid of metallicities with which to determine the [ m / h ] values of our spectra .
we suspect that in cool objects such as these , slight changes in the metal content can have profound effects on the emergent spectra . as a result
, we may be able to detect via spectroscopy smaller changes in [ m / h ] for t dwarfs than are possible in hotter stars due to the richness of molecular species and the important role of condensation in determining the absorbing species of colder objects .
in other words , the metal content of these t dwarfs may not be too different from solar .
as discussed above for wise 0448@xmath21935 , the t7.5 dwarf ulas j141623.94 + 134836.3 shows peculiarities in its spectrum that are similar to the ones seen in these unusual wise t dwarfs , yet models with a subsolar abundance of only [ m / h ] = @xmath20.3 provide good fits to the emergent spectrum of that object ( @xcite ) .
this object , a normal t6 dwarf , lies 265 arcseconds away from the nearby m5 v star lhs 302 ( gj 3657 ) . our spectrophotometric distance estimate for wise 1122 + 2550 ( 16.9 pc ; table [ astrometry ] ) is very similar to the distance of 17.2 pc obtained via trigonometric parallax ( [email protected] arcsec ) for lhs 302 ( @xcite ) .
moreover , the right ascension and declination components of the proper motion of wise 1122 + 2550 are measured by us to be @[email protected] and @[email protected] arcsec / yr , respectively , which are only @xmath53@xmath100 different from those measured for lhs 302 ( @[email protected] arcsec / yr and @[email protected] arcsec / yr ; @xcite ) . if these two objects are a common proper motion binary , the projected separation between them is @xmath54500 au .
other stellar + substellar binaries of large separation are known .
examples are the gliese 570 system comprised of k4 v , m1.5 v + m3 v , and t7.5 componenets with the latter having a projected separation of 1500 au from the k star ; the gliese 584 system comprised of g1 v + g3 v and l8 components with a projected separation of 3600 au ( @xcite ) ; the gliese 417 system comprised of g0 v and l4.5 components with a projected separation of 2000 au ( @xcite ) ; the gliese 618.1 system comprised of m0 v and l2.5 components with a projected separation of 1000 au ( @xcite ) ; the hd 89744 system comprised of f7 iv - v and l0 components with a projected separation of 2500 au ( @xcite ) ; and the hd 2057 system comprised of f8 and l4 components with a projected separation of 7000 - 9000 au ( @xcite ) . each of these systems , however , has a more massive primary than the one discussed here . assuming that lhs 302 has a mass of @xmath50.2 @xmath101 in concert with other m5 dwarfs ( see @xcite ) , then our projected separation of 4500 au falls well outside the @xmath102 au stability limit suggested empirically by @xcite
this suggests that wise 1122 + 2550 and lhs 302 are either physically unbound while sharing a common proper motion or are totally unassociated .
our distance estimate places this t8 dwarf 9.6 pc from the sun .
our distance estimate places this t9 dwarf 6.7 pc from the sun .
our measurement of @[email protected] arcsec / yr , based on astrometry covering only 0.7 yr , may also indicate a high proper motion , but the uncertainty in this value is very large .
the near - infrared spectrum of this source is an excellent match to the l5 standard at @xmath21 band . at longer wavelengths , however , the spectrum is considerably bluer , as shown in figure [ 1311p3629 ] .
there is no evidence in the @xmath21-band that this source has a low - metallicity , and therefore that the @xmath22 and @xmath34 bands are being suppressed by the relatively stronger collision - induced absorption by h@xmath27 one would expect in a metal - starved atmosphere .
the notch near 1.62 @xmath1 m in the top of the @xmath22 band peak is very similar to the interesting feature noted by @xcite in the spectrum of sdss j080531.84 + 481233.0 , which those authors claim is an unresolved mid - l + mid - t binary .
the @xmath22-band notch is also seen by @xcite in the spectrum of 2mass j1315@xmath22649 , which those authors have successfully split , via high - resolution imaging and spectroscopy , into an l5 + t7 double .
we classify wise 1311 + 3629 as an `` l5 blue '' and note that its peculiar features may be caused by unresolved binarity as well .
our formal fits to synthetic binaries ( see @xcite for details ) show that the most likely spectral types of the two components are [email protected] and [email protected] with estimated relative magnitudes of @xmath103 , @xmath104 , and @xmath105 mag .
this object would be an excellent target for high - resolution imaging .
we note that this object was also identified as a brown dwarf candidate by @xcite and given the designation sdss j131142.11 + 362923.9 .
this object is tentatively classfied as y0 ( pec ? ) by cushing et al .
( accepted ) , who describe its spectral features and derived physical parameters .
we estimate a distance of 8.6 pc and find a high proper motion of [email protected] arcsec / yr and high tangential velocity of [email protected] km / s . as with most of the other y dwarf discoveries ,
this object is detected by wise only in bands w2 and w3 and not in w1 or w4 ( table [ wise_photometry_discoveries ] ) .
the @xmath21-band spectrum of this object best fits the t8 standard , but there are discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 1436@xmath21814 shows excess flux relative to the standard , and at @xmath34 band the spectrum shows less flux relative to the standard ( figure [ 1436m1814 ] ) .
we therefore classify this object as a `` t8 pec '' . the physical cause ,
as discussed above for several other objects , may be low metal content .
this t6 dwarf is estimated to fall only 4.9 pc from the sun .
its large motion ( @xmath1 = [email protected] arcsec / yr ) placed it nearly in front of a star , now to the southeast , of similar near - infrared brightness during the 2mass survey ( see figure [ finder_chart12 ] ) , and this confusion led to the source having been missed in photometric searches of the 2mass point source catalog .
the near - infrared spectrum of this object , discussed in cushing et al .
( accepted ) , is classified as y0 because of its similarity to the spectrum of the y0 near - infrared standard , wise 1738 + 2732 . as with most of the other y dwarf discoveries ,
this object is detected by wise only in bands w2 and w3 and not in w1 or w4 ( table [ wise_photometry_discoveries ] ) .
in contrast to our crude spectrophotometric distance estimate from table [ astrometry ] of 8.2 pc , we measure a trignometric parallax placing it at @xmath3 pc ( table [ prelim_parallaxes ] ) , along with a proper motion of @xmath106 arcsec / yr .
this parallax result is significant only at the 3@xmath100 level and is measured only over a 1.2yr baseline , so it should be treated as preliminary only .
nonetheless , if confirmed , this distance would place wise 1541@xmath22250 as the seventh closest stellar system to the sun after the @xmath57 centauri system ( d=1.3 pc , @xcite ) , barnard s star ( d=1.8 pc , @xcite ) , wolf 359 ( d=2.4 pc , @xcite ) , lalande 21185 ( d=2.5 pc , @xcite ) , sirius ab ( d=2.6 pc , @xcite ) , and l 726 - 8 ab ( also known as bl ceti and uv ceti ; d=2.7 pc , @xcite )
. the measured distance implies absolute magnitudes of @xmath78 and @xmath107 mag on the mko filter system and @xmath108 mag .
this indicates a rapid dimming at those wavelengths in just a single spectral subclass from t9 to y0 ( see , e.g. , figure [ mw2_vs_type ] and figure [ mh_vs_type ] ) .
the @xmath21-band spectrum of this object best fits the l9 standard , but there is excess flux at @xmath22 and particularly @xmath34 relative to the standard itself ( figure [ 1647p5632 ] ) .
we therefore classify this object as an `` l9 pec ( red ) '' .
this object adds to a growing list of red l dwarfs whose red colors can not obviously be attributed to low gravity .
it becomes the seventh example of this class , which now includes 2mass j21481633 + 4003594 and 2mass j18212815 + 1414010 from @xcite ; 2mass j13313310 + 3407583 , 2mass j23174712@xmath24838501 , and 2mass j23512200 + 3010540 from @xcite ; and wise 0206 + 2640 from above . as mentioned in @xcite ,
the kinematics of the first five examples suggest that these objects derive from an old population , making them distinct from the red l dwarfs that have low - gravity spectral signatures and young kinematics .
we note , however , that this object has a low tangential velocity of only [email protected] km / s .
however , this velocity assumes our spectrophotometric distance estimate of 20.2 pc , and our preliminary astrometric measurements over an 11.8yr baseline indicate a closer distance of @xmath109 pc , along with a motion of @xmath110 arcsec / yr . continued astrometric monitoring of this object is needed to see if this closer distance is confirmed , as this would provide another clue in deciphering the physical nature of this rare class of red ( non low-@xmath17 ) l dwarfs .
cushing et al . ( accepted ) propose this object as the y0 spectroscopic standard .
it is the only one of our y dwarfs detected in all three short - wavelength bands of wise ( w1 , w2 , and w3 ) .
our spectrophotometric distance estimate places it at 10.5 pc ; the available astrometry for this y0 dwarf spans barely six months ( table [ astrometry ] ) , so we are not yet able to derive proper motion or parallax .
this nearby dwarf of near - infrared spectral type t9 is detected in the 2mass and sdss surveys but was overlooked because of its weak detection in both . using a 10.4yr baseline , we find a high proper motion of @xmath1 = [email protected] arcsec / yr and estimate a distance of 4.7 pc .
our preliminary trigonometric parallax measurement places it at @xmath111 pc ( table [ prelim_parallaxes ] ) . this object has identical optical and near - infrared types to another nearby object , ugps [email protected] , whose trigonometric parallax from @xcite places it at a distance of 4.1 pc .
@xcite note that wise 1741 + 2553 appears single in near - infared imaging observations with laser guide star adaptive optics .
the near - infrared spectrum of this object is classified as t9.5 : because of its similiarity to the spectrum of the t9.5 dwarf wise 0148@xmath27202 .
although noisy , the narrowness of the j - band peaks falls intermediate between that of the t9 standard ugps [email protected] and the y0 standard wise 1738 + 2732 .
we estimate that this object falls at a distance of 13.0 pc .
the wise detections for this object are very similar to those seen for y dwarfs ; namely , the object is detected only in bands w2 and w3 and not in w1 or w4 ( table [ wise_photometry_discoveries ] ) .
this object , with a tentative classification of @xmath4y0 from cushing et al .
( accepted ) , is the latest object so far found with wise .
the spectrum is unique among late - t and y dwarfs in that the @xmath21- and @xmath22-band peaks , in units of @xmath112 , are nearly the same height ( figure [ other_spectra_gt_t8 ] ) .
this reddening of the near - infrared colors ( @xmath48 mag ; table [ followup_photometry_discoveries ] ) is predicted by model atmosphere calculations to occur at effective temperatures below 300 - 400k ( @xcite ) .
this effect is due to the fact that the wien tail of the spectral energy distribution becomes the overwhelming effect shaping the spectrum at those wavelengths , and this may be even more dramatically illustrated by the extremely red @xmath113 and @xmath114 colors measured for this object ( figure [ jw2_vs_type ] and figure [ hw2_vs_type ] ) .
our spectrophotometric distance estimate of @xmath119.4 pc implies exceedingly dim absolute magnitudes of @xmath115 and @xmath116 mag , on the mko filter system .
these values agree with expectations that wise 1828 + 2650 should be dimmer at these wavelengths than the presumably warmer wise 1541@xmath22250 , which also has exceedingly dim @xmath21 and @xmath22 magnitudes ( table [ prelim_parallaxes ] ) .
this y0 dwarf , discussed in cushing et al .
( accepted ) , is estimated to lie at a distance of 7.7 pc . as with several other y dwarf discoveries , this object is detected by wise only in bands w2 and w3 and not in w1 or w4 ( table [ wise_photometry_discoveries ] ) . the @xmath21-band spectrum of this object best fits the t9 standard , but there are discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 2134@xmath27137 shows excess flux relative to the standard , and at @xmath34 band the spectrum shows less flux relative to the standard ( figure [ 2134m7137 ] ) .
we therefore classify this object as a `` t9 pec '' . the physical cause ,
as discussed above for several other objects , may be low metal content . as with the previous object ,
the @xmath21-band spectrum best fits the t9 standard , but there are discrepancies at @xmath29 and @xmath34 bands . at @xmath29 band
the spectrum of wise 2325@xmath24105 shows excess flux relative to the t9 standard , and at @xmath34 band the spectrum shows less flux relative to the t9 standard ( figure [ 2325m4105 ] ) .
we therefore classify this object as a `` t9 pec '' . the physical cause ,
as discussed above for other objects , may be low metal content .
this object has the near - infrared spectrum of a normal l9 dwarf , but its wise color of w1-w2 = [email protected] is markedly redder than the other l9 dwarfs on figure [ w1w2_vs_type ] . the _
spitzer_/irac color of ch1-ch2 = [email protected] is redder than all other l and early - t dwarfs in figure [ ch1ch2_vs_type ] .
effects such as low - gravity or low - metallicity would cause the near - infrared spectrum of this object to appear unusually red or unusually blue , respectively , in the near - infrared ( @xcite ) , and this is not seen .
one hypothesis is that this object is an unresolved l+t binary , but not one with such a warm t dwarf that the near - infrared spectrum of the composite shows itself to be peculiar .
we can test this as follows .
the w1-w2 color of wise 2327@xmath22730 is @xmath50.23 mag redder than the mean w1-w2 for other l9 dwarfs . using the absolute w2 versus spectral type plot of figure [ mw2_vs_type ] along with the trend of w1-w2 color with spectral type in figure [ w1w2_vs_type ]
, we estimate that the type of the hypothesized companion would have to be roughly t6.5 .
however , as figure 4 of @xcite shows , an object with a composite type of @xmath5l9 and secondary of @xmath5t6.5 would have a noticeably peculiar near - infrared spectrum which would distinguish it from a normal l9 .
hence , binarity appears not to be the cause of the redder w1-w2 and ch1-ch2 colors .
the reason for this color peculiarity remains unexplained .
albert , l. , artigau , . ,
delorme , p. , reyl , c. , forveille , t. , delfosse , x. , & willott , c. j. 2011 , , 141 , 203 artigau , . ,
radigan , j. , folkes , s. , jayawardhana , r. , kurtev , r. , lafrenire , d. , doyon , r. , & borissova , j. 2010 , , 718 , l38 artigau , . , doyon , r. , lafrenire , d. , nadeau , d. , robert , j. , & albert , l. 2006 , , 651 , l57 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. 2003
, brown dwarfs , 211 , 41 barman , t. s. , hauschildt , p. h. , & allard , f. 2005 , , 632 , 1132 bate , m. r. , & bonnell , i. a. 2005 , , 356 , 1201 becklin , e. e. , & zuckerman , b. 1988 , , 336 , 656 bertin , e. , & arnouts , s. 1996 , , 117 , 393 bessell , m. s. 2005 , , 43 , 293 bessell , m. s. , & brett , j. m. 1988 , , 100 , 1134 bloom , j. s. , starr , d. l. , blake , c. h. , skrutskie , m. f. , & falco , e. e. 2006 , astronomical data analysis software and systems xv , 351 , 751 boss , a. p. 2004
, , 350 , l57 bouvier , j. , kendall , t. , & meeus , g. 2009 , american institute of physics conference series , 1094 , 497 bouy , h. , brandner , w. , martn , e. l. , delfosse , x. , allard , f. , & basri , g. 2003 , , 126 , 1526 bowler , b. p. , liu , m. c. , & dupuy , t. j. 2010 , , 710 , 45 boyd , d. f. a. , & whitworth , a. p. 2005
, , 430 , 1059 burgasser , a. j. , sitarski , b. n. , gelino , c. r. , logsdon , s. e. , & perrin , m. d. 2011 , arxiv:1107.1484 burgasser , a. j. , et al .
2011 , arxiv:1104.2537 burgasser , a. j. , et al .
2010 , , 725 , 1405 burgasser , a. j. , looper , d. , & rayner , j. t. 2010 , , 139 , 2448 burgasser , a. j. 2008 , 14th cambridge workshop on cool stars , stellar systems , and the sun , 384 , 126 burgasser , a. j. , looper , d. l. , kirkpatrick , j. d. , cruz , k. l. , & swift , b. j. 2008 , , 674 , 451 burgasser , a. j. 2007 , , 134 , 1330 burgasser , a. j. , geballe , t. r. , leggett , s. k. , kirkpatrick , j. d. , & golimowski , d. a. 2006 , , 637 , 1067 burgasser , a. j. , burrows , a. , & kirkpatrick , j. d. 2006 , , 639 , 1095 burgasser , a. j. 2004 , , 155 , 191 burgasser , a. j. , mcelwain , m. w. , kirkpatrick , j. d. , cruz , k. l. , tinney , c. g. , & reid , i. n. 2004 , , 127 , 2856 burgasser , a. j. , kirkpatrick , j. d. , liebert , j. , & burrows , a. 2003 , , 594 , 510 burgasser , a. j. , kirkpatrick , j. d. , mcelwain , m. w. , cutri , r. m. , burgasser , a. j. , & skrutskie , m. f. 2003 , , 125 , 850 burgasser , a. j. , mcelwain , m. w. , & kirkpatrick , j. d. 2003 , , 126 , 2487 burgasser , a. j. , et al .
2002 , , 564 , 421 burgasser , a. j. , et al .
2000 , , 120 , 1100 burgasser , a. j. , et al .
2000 , , 531 , l57 burgasser , a. j. , et al . 1999 , , 522 , l65 burningham , b. , et al .
2011 , , 414 , l90 burningham , b. , et al .
2011 , , l259 burningham , b. , et al . 2010 , , 406 , 1885 burningham , b. , et al . 2010 , , 404 , 1952 burningham , b. , et al . 2009 , , 395 , 1237 burningham , b. , et al .
2008 , , 391 , 320 burrows , a. , sudarsky , d. , & lunine , j. i. 2003 , , 596 , 587 chiu , k. , fan , x. , leggett , s. k. , golimowski , d. a. , zheng , w. , geballe , t. r. , schneider , d. p. , & brinkmann , j. 2006 , , 131 , 2722 costa , e. , mndez , r. a. , jao , w .- c . ,
henry , t. j. , subasavage , j. p. , & ianna , p. a. 2006 , , 132 , 1234 cruz , k. l. , et al .
2007 , , 133 , 439 cruz , k. l. , burgasser , a. j. , reid , i. n. , & liebert , j. 2004 , , 604 , l61 cruz , k. l. , reid , i. n. , liebert , j. , kirkpatrick , j. d. , & lowrance , p. j. 2003 , , 126 , 2421 cushing , m. c. , vacca , w. d. , & rayner , j. t. 2004 , , 116 , 362 dahn , c. c. , et al .
2002 , , 124 , 117 dahn , c. c. , et al .
1988 , , 95 , 237 deacon , n. r. , hambly , n. c. , king , r. r. , & mccaughrean , m. j. 2009 , , 394 , 857 deacon , n. r. , hambly , n. c. , & cooke , j. a. 2005 , , 435 , 363 delfosse , x. , tinney , c. g. , forveille , t. , epchtein , n. , borsenberger , j. , fouqu , p. , kimeswenger , s. , & tiphne , d. 1999 , , 135 , 41 delfosse , x. , et al .
1997 , , 327 , l25 delorme , p. , et al .
2010 , , 518 , a39 delorme , p. , et al .
2008 , , 482 , 961 eddington , a. s. , sir 1940 , , 100 , 354 eddington , a. s. 1913 , , 73 , 359 eisenhardt , p. r. m. , et al .
2010 , , 139 , 2455 ellis , s. c. , tinney , c. g. , burgasser , a. j. , kirkpatrick , j. d. , & mcelwain , m. w. 2005 , , 130 , 2347 faherty , j. k. , burgasser , a. j. , cruz , k. l. , shara , m. m. , walter , f. m. , & gelino , c. r. 2009 , , 137 , 1 fan , x. , et al .
2000 , , 119 , 928 fazio , g. g. , et al .
2004 , , 154 , 10 fortney , j. j. , marley , m. s. , lodders , k. , saumon , d. , & freedman , r. 2005 , , 627 , l69 geballe , t. r. , et al .
2002 , , 564 , 466 gelino , c. r. , et al .
2011 , , 142 , 57 giclas , h. l. , burnham , r. , & thomas , n. g. 1971 , flagstaff , arizona : lowell observatory , 1971 , gizis , j. e. 2002 , , 575 , 484 gizis , j. e. , monet , d. g. , reid , i. n. , kirkpatrick , j. d. , liebert , j. , & williams , r. j. 2000 , , 120 , 1085 goldman , b. , marsat , s. , henning , t. , clemens , c. , & greiner , j. 2010 , , 405 , 1140 hall , p. b. 2002 , , 564 , l89 hamuy , m. , suntzeff , n. b. , heathcote , s. r. , walker , a. r. , gigoux , p. , & phillips , m. m. 1994 , , 106 , 566 hanisch , r. j. , farris , a. , greisen , e. w. , pence , w. d. , schlesinger , b. m. , teuben , p. j. , thompson , r. w. , & warnock , a. , iii 2001 , , 376 , 359 hawkins , m. r. s. , & bessell , m. s. 1988 , , 234 , 177 hawley , s. l. , et al .
2002 , , 123 , 3409 hayashi , c. , & nakano , t. 1963 , progress of theoretical physics , 30 , 460 herter , t. l. , et al .
2008 , , 7014 , horne , k. 1986 , , 98 , 609 kanneganti , s. , park , c. , skrutskie , m. f. , wilson , j. c. , nelson , m. j. , smith , a. w. , & lam , c. r. 2009 , , 121 , 885 kendall , t. r. , jones , h. r. a. , pinfield , d. j. , pokorny , r. s. , folkes , s. , weights , d. , jenkins , j. s. , & mauron , n. 2007 , , 374 , 445 kirkpatrick , j. d. , et al .
2010 , , 190 , 100 kirkpatrick , j. d. 2008 , 14th cambridge workshop on cool stars , stellar systems , and the sun , 384 , 85 kirkpatrick , j. d. , et al .
2008 , , 689 , 1295 kirkpatrick , j. d. , barman , t. s. , burgasser , a. j. , mcgovern , m. r. , mclean , i. s. , tinney , c. g. , & lowrance , p. j. 2006 , , 639 , 1120 kirkpatrick , j. d. 2005 , , 43 , 195 kirkpatrick , j. d. , dahn , c. c. , monet , d. g. , reid , i. n. , gizis , j. e. , liebert , j. , & burgasser , a. j. 2001 , , 121 , 3235 kirkpatrick , j. d. 2000 , from giant planets to cool stars , 212 , 20 kirkpatrick , j. d. , et al .
2000 , , 120 , 447 kirkpatrick , j. d. , et al . 1999 , , 519 , 802 kirkpatrick , j. d. , beichman , c. a. , & skrutskie , m. f. 1997 , , 476 , 311 kirkpatrick , j. d. , & mccarthy , d. w. , jr .
1994 , , 107 , 333 kirkpatrick , j. d. , mcgraw , j. t. , hess , t. r. , liebert , j. , & mccarthy , d. w. , jr .
1994 , , 94 , 749 knapp , g. r. , et al .
2004 , , 127 , 3553 koekemoer , a. m. , fruchter , a. s. , hook , r. n. , & hack , w. 2002 , the 2002 hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system , 337 kouzuma , s. , & yamaoka , h. 2010 , , 509 , a64 kumar , s. s. 1963 , , 137 , 1121 kmmel , m. , walsh , j. r. , pirzkal , n. , kuntschner , h. , & pasquali , a. 2009 , , 121 , 59 lawrence , a. , et al .
2007 , , 379 , 1599 leggett , s. k. , et al .
2010 , , 710 , 1627 leggett , s. k. , et al .
2009 , , 695 , 1517 leggett , s. k. , et al .
2000 , , 536 , l35 lpine , s. , & shara , m. m. 2005 , , 129 , 1483 liebert , j. , kirkpatrick , j. d. , cruz , k. l. , reid , i. n. , burgasser , a. , tinney , c. g. , & gizis , j. e. 2003 , , 125 , 343 liu , m. c. , et al . 2011 ,
arxiv:1103.0014 lodieu , n. , scholz , r .- d . , & mccaughrean , m. j. 2002 , , 389 , l20 loh , e. d. , biel , j. d. , chen , j .- j . , davis , m. , laporte , r. , & loh , o. y. 2004 , , 5492 , 1644 looper , d. l. , et al .
2008 , , 686 , 528 looper , d. l. , kirkpatrick , j. d. , & burgasser , a. j. 2007 , , 134 , 1162 lpez - morales , m. 2007 , , 660 , 732 lucas , p. w. , et al .
2010 , , 408 , l56 lucas , p. w. , weights , d. j. , roche , p. f. , & riddick , f. c. 2006 , , 373 , l60 luhman , k. l. , burgasser , a. j. , & bochanski , j. j. 2011 , , 730 , l9 luhman , k. l. , et al .
2007 , , 654 , 570 luhman , k. l. 2007 , , 173 , 104 luhman , k. l. , rieke , g. h. , young , e. t. , cotera , a. s. , chen , h. , rieke , m. j. , schneider , g. , & thompson , r. i. 2000 , , 540 , 1016 lutz , t. e. , & kelker , d. h. 1973 , , 85 , 573 luyten , w. j. 1980 , nltt catalogue .
volume_iii .
0__to -30_. , by luyten , w. j .. published by university of minnesota , mineapolis , minnesota , usa , 283 p. , luyten , w. j. 1979 , minneapolis : university of minnesota , 1979 , 2nd ed . ,
luyten , w. j. 1979 , new luyten catalogue of stars with proper motions larger than two tenths of an arcsecond , 1 , 0 ( 1979 ) , 0 luyten , w. j. 1979 , new luyten catalogue of stars with proper motions larger than two tenths of an arcsecond , 2 , 0 ( 1979 ) , 0 luyten , w. j. , & kowal , c. t. 1975 , proper motion survey with the forty - eight inch schmidt telescope .
one hundred and six faint stars with large proper motions .
, by luyten , w. j. ; kowal , c. t .. separate print univ .
minnesota , minneapolis , minnesota , 2 p. , 43
, 1 luyten , w. j. 1974 , proper motion survey with the forty - eight inch schmidt telescope._xxxvii .
proper motions for 4483 faint stars .
, by luyten , w. j .. separate print univ .
minnesota , minneapolis , minnesota , 44 p. , 37 , 1 luyten , w. j. 1974 , proper motion survey with the forty - eight inch schmidt telescope .
proper motions for 6955 faint stars . , by luyten ,
w. j .. separate print univ .
minnesota , minneapolis , minnesota , 64 p. , 36 , 1 luyten , w. j. 1972 , proper motion survey with the forty - eight inch schmidt telescope .
proper motions for 2520 faint stars .
, by luyten , w. j .. separate print univ .
minnesota , minneapolis , minnesota , 24 p. , 31 , 1 mainzer , a. , et al .
2011 , , 726 , 30 malmquist , k. g. 1920 , medd .
lund astron .
ser . , 2 .
22 marocco , f. , et al .
2010 , , 524 , a38 marois , c. , macintosh , b. , barman , t. , zuckerman , b. , song , i. , patience , j. , lafrenire , d. , & doyon , r. 2008 , science , 322 , 1348 martn , e. l. , et al .
2010 , , 517 , a53 martini , p. , persson , s. e. , murphy , d. c. , birk , c. , shectman , s. a. , gunnels , s. m. , & koch , e. 2004 , , 5492 , 1653 metchev , s. a. , kirkpatrick , j. d. , berriman , g. b. , & looper , d. 2008 , , 676 , 1281 mclean , i. s. , mcgovern , m. r. , burgasser , a. j. , kirkpatrick , j. d. , prato , l. , & kim , s. s. 2003 , , 596 , 561 mclean , i. s. , graham , j. r. , becklin , e. e. , figer , d. f. , larkin , j. e. , levenson , n. a. , & teplitz , h. i. 2000 , , 4008 , 1048 mclean , i. s. , et al .
1998 , , 3354 , 566 mclean , i. s. , et al .
1994 , , 2198 , 457 milligan , s. , cranton , b. w. , & skrutskie , m. f. 1996 , , 2863 , 2 monet , d. g. , et al .
2003 , , 125 , 984 muench , a. a. , lada , e. a. , lada , c. j. , & alves , j. 2002 , , 573 , 366 noll , k. s. , geballe , t. r. , leggett , s. k. , & marley , m. s. 2000 , , 541 , l75 oke , j. b. , et al . 1995 ,
pasp , 107 , 375 padoan , p. , kritsuk , a. , michael , norman , l. , & nordlund , .
2005 , , 76 , 187 patten , b. m. , et al .
2006 , , 651 , 502 perryman , m. a. c. , et al .
1997 , , 323 , l49 phan - bao , n. , et al .
2008 , , 383 , 831 pinfield , d. j. , et al .
2008 , , 390 , 304 rayner , j. t. , toomey , d. w. , onaka , p. m. , denault , a. j. , stahlberger , w. e. , vacca , w. d. , cushing , m. c. , & wang , s. 2003 , , 115 , 362 reid , i. n. , cruz , k. l. , kirkpatrick , j. d. , allen , p. r. , mungall , f. , liebert , j. , lowrance , p. , & sweet , a. 2008 , , 136 , 1290 reid , i. n. , kirkpatrick , j. d. , liebert , j. , gizis , j. e. , dahn , c. c. , & monet , d. g. 2002 , , 124 , 519 reid , i. n. , gizis , j. e. , kirkpatrick , j. d. , & koerner , d. w. 2001 , , 121 , 489 reid , i. n. , kirkpatrick , j. d. , gizis , j. e. , dahn , c. c. , monet , d. g. , williams , r. j. , liebert , j. , & burgasser , a. j. 2000 , , 119 , 369 reipurth , b. , & clarke , c. 2001 , , 122 , 432 reyl , c. , et al .
2010 , , 522 , a112 rodriguez , d. r. , zuckerman , b. , melis , c. , & song , i. 2011 , , 732 , l29 ross , f. e. 1928 , , 38 , 117 ruiz , m. t. , leggett , s. k. , & allard , f. 1997 , , 491 , l107 ruiz , m. t. , takamiya , m. y. , & roth , m. 1991 , , 367 , l59 schilbach , e. , rser , s. , & scholz , r .- d .
2009 , , 493 , l27 schmidt , m. 1968 , , 151 , 393 schmidt , s. j. , west , a. a. , burgasser , a. j. , bochanski , j. j. , & hawley , s. l. 2010 , , 139 , 1045 schneider , d. p. , et al .
2002 , , 123 , 458 scholz , r .- d . 2010 , , 510 , l8 scholz , r .- d .
2010 , , 515 , a92 scholz , r .- d . , mccaughrean , m. j. , lodieu , n. , & kuhlbrodt , b. 2003 , , 398 , l29 sheppard , s. s. , & cushing , m. c. 2009 , , 137 , 304 simcoe , r. a. , et al .
2010 , , 7735 , simcoe , r. a. , et al . 2008 , , 7014 , skrutskie , m. f. , et al .
2006 , , 131 , 1163 smith , j. d. t. , et al .
2007 , , 119 , 1133 stephens , d. c. , & leggett , s. k. 2004 , , 116 , 9 stern , d. , et al .
2007 , , 663 , 677 strauss , m. a. , et al . 1999 , , 522 , l61 sutherland , i. e. & hodgman , g. .w . ,
1974 , communications of the association for computing machinery , 17 swaters , r. a. , valdes , f. , & dickinson , m. e. 2009 , astronomical data analysis software and systems xviii , 411 , 506 teerikorpi , p. 2004
, , 424 , 73 tinney , c. g. , burgasser , a. j. , kirkpatrick , j. d. , & mcelwain , m. w. 2005 , , 130 , 2326 tinney , c. g. , burgasser , a. j. , & kirkpatrick , j. d. 2003 , , 126 , 975 tinney , c. g. 1993 , , 105 , 1169 tody , d. 1986 , , 627 , 733 tokunaga , a. t. , simons , d. a. , & vacca , w. d. 2002 , , 114 , 180 tsvetanov , z. i. , et al .
2000 , , 531 , l61 vacca , w. d. , cushing , m. c. , & rayner , j. t. 2003 , , 115 , 389 van altena , w. f. , lee , j. t. , & hoffleit , e. d. 2001 , vizier online data catalog , 1238 , 0 van altena , w. f. , lee , j. t. , & hoffleit , d. 1995 , vizier online data catalog , 1174 , 0 van dam , m. a. , et al . 2006 , , 118 , 310 van leeuwen , f. 2007 , , 474 , 653 vrba , f. j. , et al .
2004 , , 127 , 2948 warren , s. j. , et al .
2007 , , 381 , 1400 werner , m. w. , et al .
2004 , , 154 , 1 west , a. a. , hawley , s. l. , bochanski , j. j. , covey , k. r. , reid , i. n. , dhital , s. , hilton , e. j. , & masuda , m. 2008 , , 135 , 785 whitworth , a. p. , & zinnecker , h. 2004 , , 427 , 299 wilson , j. c. , et al .
2003 , , 4841 , 451 wilson , j. c. , miller , n. a. , gizis , j. e. , skrutskie , m. f. , houck , j. r. , kirkpatrick , j. d. , burgasser , a. j. , & monet , d. g. 2003 , brown dwarfs , 211 , 197 wilson , j. c. , kirkpatrick , j. d. , gizis , j. e. , skrutskie , m. f. , monet , d. g. , & houck , j. r. 2001 , , 122 , 1989 wizinowich , p. l. , et al .
2006 , , 118 , 297 wright , e. l. , et al .
2010 , , 140 , 1868 wright , e. l. , mainzer , a. , gelino , c. , & kirkpatrick , d. 2010 , arxiv:1104.2569 wroblewski , h. , & torres , c. 1991 , , 91 , 129 york , d. g. , et al .
2000 , , 120 , 1579 zhang , z. h. , et al .
2009 , , 497 , 619 lllcccccccc + wisepc j003402.80 - 005207.4 & ulas j003402.77 - 005206.7 & 1 & [email protected] & [email protected] & @xmath411.801 & @xmath49.224 & [email protected] & [email protected] & [email protected] & t8.5 + wisepc j005021.03 - 332229.2 & 2mass j00501994 - 3322402 & 2 & [email protected] & [email protected] & [email protected] & @xmath48.989 & [email protected] & [email protected] & [email protected] & t7 + wisepc j005911.09 - 011400.6 & cfbds j005910.90 - 011401.3 & 3 & [email protected] & [email protected] & [email protected] & @xmath49.290 & [email protected] & [email protected] & [email protected] & t9 + wisepc j013657.45 + 093347.0 & ipms j013656.57 + 093347.3 & 4 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t2.5 + wisepa j015024.39 + 135924.3 & ulas j015024.37 + 135924.0 & 5 & [email protected] & [email protected] & @xmath412.108 & @xmath48.942 & [email protected] & [email protected] & [email protected] & t7.5 + wisepa j015142.21 + 124429.8 & sdss j015141.69 + 124429.6 & 6 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t1 + wisepc j020742.96 + 000056.9 & sdss j020742.48 + 000056.2 & 6 & [email protected] & [email protected] & @xmath412.618 & @xmath49.231 & [email protected] & @xmath416.396 & @xmath415.412 & t4.5 + wisepc j024313.48 - 245331.5 & 2massi j0243137 - 245329 & 7 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t6 + wisepa j024749.98 - 163111.4 & sdss j024749.90 - 163112.6 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.114 & [email protected] & [email protected] & [email protected] & t2 : + wisepa j032553.11 + 042540.6 & sdss j032553.17 + 042540.1 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.091 & [email protected] & @xmath416.080 & @xmath416.525 & t5.5 + wisepc j034807.34 - 602234.9 & 2mass j03480772 - 6022270 & 9 & [email protected] & [email protected] & [email protected] & @xmath48.875 & [email protected] & [email protected] & [email protected] & t7 + wisepa j035104.53 + 481046.5 & sdss j035104.37 + 481046.8 & 8 & [email protected] & [email protected] & [email protected] & @xmath48.827 & [email protected] & [email protected] & [email protected] & t1 : + wisepa j040708.95 + 151455.4 & 2mass j04070885 + 1514565 & 10 & [email protected] & [email protected] & [email protected] & @xmath48.868 & [email protected] & [email protected] & [email protected] & t5 + wisepa j041521.20 - 093500.6 & 2massi j0415195 - 093506 & 7 & [email protected] & [email protected] & [email protected] & @xmath48.712 & [email protected] & [email protected] & [email protected] & t8 + wisepa j042348.32 - 041402.5 & sdssp j042348.57 - 041403.5 & 6 & [email protected] & [email protected] & [email protected] & @xmath48.462 & [email protected] & [email protected] & [email protected] & t0 + wisepa j043038.81 + 130956.8 & cfht - hy-20 & 11 & [email protected] & [email protected] & @xmath411.837 & @xmath48.722 & & & @xmath416.080 & t2 + wisepa j051035.35 - 420809.2 & 2mass j05103520 - 4208140 & 12 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t5 + wisepa j051609.28 - 044552.4 & 2mass j05160945 - 0445499 & 9 & [email protected] & [email protected] & [email protected] & @xmath49.010 & [email protected] & [email protected] & [email protected] & t5.5 + wisepa j051859.89 - 282840.2 & 2mass j05185995 - 2828372 & 13 & [email protected] & [email protected] & [email protected] & @xmath48.798 & [email protected] & [email protected] & [email protected] & t1p + wisepa j055919.58 - 140452.6 & 2mass j05591914 - 1404488 & 14 & [email protected] & [email protected] & [email protected] & @xmath48.666 & [email protected] & [email protected] & [email protected] & t4.5 + wisepa j060206.60 + 404356.8 & 2mass j06020638 + 4043588 & 12 & [email protected] & [email protected] & @xmath412.678 & @xmath49.123 & [email protected] & [email protected] & [email protected] & t4.5 + wisepa j072227.27 - 054029.9 & ugps j072227.51 - 054031.2 & 15 & [email protected] & [email protected] & [email protected] & @xmath48.637 & [email protected] & [email protected] & & t9 + wisepa j072719.13 + 170952.0 & 2massi j0727182 + 171001 & 7 & [email protected] & [email protected] & [email protected] & @xmath48.200 & [email protected] & [email protected] & [email protected] & t7 + wisepa j072859.49 - 395346.7 & 2mass j07290002 - 3954043 & 12 & [email protected] & [email protected] & [email protected] & @xmath49.022 & [email protected] & [email protected] & @xmath415.290 & t8pec + wisepa j073922.43 + 661502.5 & sdss j073922.26 + 661503.5 & 8 & [email protected] & [email protected] & @xmath412.456 & @xmath49.180 & [email protected] & [email protected] & [email protected] & t1.5 : + wisepa j074148.97 + 235124.0 & sdss j074149.15 + 235127.5 & 16 & [email protected] & [email protected] & @xmath412.002 & @xmath48.450 & [email protected] & [email protected] & @xmath415.847 & t5 + wisepa j074201.00 + 205517.1 & sdss j074201.41 + 205520.5 & 16 & [email protected] & [email protected] & @xmath411.982 & @xmath48.973 & [email protected] & [email protected] & @xmath415.225 & t5 + wisepa j075547.93 + 221214.4 & 2massi j0755480 + 221218 & 7 & [email protected] & [email protected] & [email protected] & @xmath49.107 & [email protected] & [email protected] & [email protected] & t5 + wisepa j075840.16 + 324721.0 & sdss j075840.33 + 324723.4 & 16 & [email protected] & [email protected] & [email protected] & @xmath48.612 & [email protected] & [email protected] & [email protected] & t2 + wisepa j081729.74 - 615504.1 & denis j081730.0 - 615520 & 17 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t6 + wisepa j083048.90 + 012826.9 & sdss j083048.80 + 012831.1 & 16 & [email protected] & [email protected] & [email protected] & @xmath48.602 & [email protected] & [email protected] & @xmath416.358 & t4.5 + wisepa j083717.18 - 000021.2 & sdssp j083717.22 - 000018.3 & 18 & [email protected] & [email protected] & @xmath412.770 & @xmath49.163 & [email protected] & [email protected] & @xmath415.674 & t1 + wisepa j085834.06 + 325628.2 & sdss j085834.42 + 325627.7 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.243 & [email protected] & [email protected] & [email protected] & t1 + wisepa j085910.62 + 101013.8 & ulas j085910.69 + 101017.1 & 19 & [email protected] & [email protected] & @xmath412.442 & @xmath49.372 & [email protected] & [email protected] & [email protected] & t7 + wisepa j090900.49 + 652526.5 & sdss j090900.73 + 652527.2 & 8 & [email protected] & [email protected] & & @xmath4 null & [email protected] & [email protected] & [email protected] & t1.5 + wisepa j092012.04 + 351740.1 & 2massw j0920122 + 351742 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.200 & [email protected] & [email protected] & [email protected] & t0p + wisepa j092615.39 + 584718.8 & sdssp j092615.38 + 584720.9 & 6 & [email protected] & [email protected] & [email protected] & @xmath49.209 & [email protected] & [email protected] & [email protected] & t4.5 + wisepc j092624.75 + 071138.3 & ulas j092624.76 + 071140.7 & 5 & [email protected] & [email protected] & @xmath412.683 & @xmath49.451 & [email protected] & [email protected] & [email protected] & t3.5 + wisepc j093735.61 + 293127.7 & 2massi j0937347 + 293142 & 7 & [email protected] & [email protected] & [email protected] & @xmath48.651 & [email protected] & [email protected] & [email protected] & t6p + wisepc j094908.57 - 154548.5 & 2mass j09490860 - 1545485 & 2 & [email protected] & [email protected] & [email protected] & @xmath49.128 & [email protected] & [email protected] & [email protected] & t2 + wisepc j100732.94 - 455512.9 & 2mass j10073369 - 4555147 & 12 & [email protected] & [email protected] & [email protected] & @xmath49.276 & [email protected] & [email protected] & [email protected] & t5 + wisepc j102109.57 - 030421.5 & sdss j102109.69 - 030420.1 & 18 & [email protected] & [email protected] & [email protected] & @xmath48.850 & [email protected] & [email protected] & [email protected] & t3 + wisepc j103931.34 + 325624.7 & sdss j103931.35 + 325625.5 & 8 & [email protected] & [email protected] & [email protected] & @xmath48.699 & [email protected] & [email protected] & [email protected] & t1 + wisepc j104829.08 + 091939.4 & sdss j104829.21 + 091937.8 & 8 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath416.365 & t2.5 + wisepc j105213.53 + 442254.6 & sdss j105213.51 + 442255.7 & 8 & [email protected] & [email protected] & [email protected] & @xmath48.957 & [email protected] & [email protected] & [email protected] & t0.5 : + wisepc j105951.63 + 304159.0 & 2mass j10595185 + 3042059 & 21 & [email protected] & [email protected] & [email protected] & @xmath49.090 & [email protected] & [email protected] & [email protected] & t4 + wisepc j110611.72 + 275417.5 & 2mass j11061197 + 2754225 & 12 & [email protected] & [email protected] & [email protected] & @xmath48.718 & [email protected] & [email protected] & [email protected] & t2.5 + wisepc j111009.86 + 011608.7 & sdssp j111010.01 + 011613.1 & 6 & [email protected] & [email protected] & [email protected] & @xmath49.296 & [email protected] & [email protected] & @xmath415.129 & t5.5 + wisepc j111448.80 - 261828.2 & 2mass j11145133 - 2618235 & 2 & [email protected] & [email protected] & [email protected] & @xmath49.463 & [email protected] & [email protected] & @xmath416.109 & t7.5 + wisepc j112208.14 - 351239.7 & 2mass j11220826 - 3512363 & 2 & [email protected] & [email protected] & [email protected] & @xmath48.824 & [email protected] & [email protected] & [email protected] & t2 + wisepc j115323.23 - 014828.0 & ulas j115338.74 - 014724.1 & 5 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t6 + wisepc j115700.56 + 061103.2 & sdss j115700.50 + 061105.2 & 16 & [email protected] & [email protected] & @xmath412.312 & @xmath49.229 & [email protected] & [email protected] & @xmath415.251 & t1.5 + wisepc j120257.02 + 090158.1 & ulas j120257.05 + 090158.8 & 5 & [email protected] & [email protected] & @xmath412.041 & @xmath48.744 & [email protected] & [email protected] & [email protected] & t5 + wisepc j120602.54 + 281328.2 & sdss j120602.51 + 281328.7 & 8 & [email protected] & [email protected] & @xmath412.080 & @xmath49.519 & [email protected] & [email protected] & @xmath415.817 & t3 + wisepc j120746.83 + 024426.3 & sdss j120747.17 + 024424.8 & 22 & [email protected] & [email protected] & [email protected] & @xmath48.884 & [email protected] & [email protected] & [email protected] & t0 + wisepc j120956.33 - 100405.8 & 2mass j12095613 - 1004008 & 10 & [email protected] & [email protected] & [email protected] & @xmath49.012 & [email protected] & [email protected] & [email protected] & t3 + wisepc j121441.14 + 631643.6 & sdss j121440.95 + 631643.4 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.636 & [email protected] & [email protected] & [email protected] & t3.5 : + wisepc j121544.12 - 342103.9 & 2mass j12154432 - 3420591 & 12 & [email protected] & [email protected] & @xmath412.327 & @xmath49.062 & [email protected] & [email protected] & @xmath416.317 & t4.5 + wisepc j121710.30 - 031113.2 & 2massi j1217110 - 031113 & 23 & [email protected] & [email protected] & [email protected] & @xmath49.372 & [email protected] & [email protected] & @xmath415.887 & t7.5 + wisepc j122554.67 - 273954.0 & 2mass j12255432 - 2739466 & 23 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t6 + wisepc j123146.73 + 084722.3 & 2mass j12314753 + 0847331 & 10 & [email protected] & [email protected] & [email protected] & @xmath48.781 & [email protected] & [email protected] & [email protected] & t5.5 + wisepc j123737.39 + 652608.2 & 2mass j12373919 + 6526148 & 23 & [email protected] & [email protected] & [email protected] & @xmath49.343 & [email protected] & [email protected] & @xmath416.058 & t6.5 + wisepc j123828.37 + 095352.5 & ulas j123828.51 + 095351.3 & 24 & @xmath417.806 & [email protected] & @xmath412.453 & @xmath48.620 & [email protected] & [email protected] & & t8.5 + wisepc j125011.60 + 392547.8 & sdss j125011.65 + 392553.9 & 8 & [email protected] & [email protected] & @xmath412.404 & @xmath49.303 & [email protected] & [email protected] & [email protected] & t4 + wisepc j125453.57 - 012245.8 & sdssp j125453.90 - 012247.4 & 18 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t2 + wisepc j130041.65 + 122114.6 & ross 485c & 25 & [email protected] & [email protected] & [email protected] & @xmath49.257 & [email protected] & [email protected] & [email protected] & t8 + wisepc j130217.09 + 130850.7 & ulas j130217.21 + 130851.2 & 5 & [email protected] & [email protected] & @xmath411.958 & @xmath48.611 & [email protected] & [email protected] & [email protected] & t8.5 + wisepc j132434.95 + 635827.5 & 2mass j13243559 + 6358284 & 12 & [email protected] & [email protected] & [email protected] & @xmath49.671 & [email protected] & [email protected] & [email protected] & t2:pec + wisepc j133553.41 + 113004.7 & ulas j133553.45 + 113005.2 & 24 & [email protected] & [email protected] & [email protected] & @xmath49.027 & [email protected] & [email protected] & [email protected] & t9 + wisepa j134646.09 - 003151.4 & sdssp
j134646.45 - 003150.4 & 26 & [email protected] & [email protected] & [email protected] & @xmath48.456 & [email protected] & [email protected] & [email protected] & t6.5 + wisepc j135852.69 + 374708.7 & sdss j135852.68 + 374711.9 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.251 & [email protected] & [email protected] & @xmath416.099 & t4.5 : + wisepa j140255.66 + 080053.9 & sdss j140255.66 + 080055.2 & 8 & [email protected] & [email protected] & @xmath412.570 & @xmath48.834 & [email protected] & [email protected] & [email protected] & t1.5 + wisepa j140449.73 - 315933.0 & 2mass j14044941 - 3159329 & 12 & [email protected] & [email protected] & [email protected] & @xmath49.045 & [email protected] & [email protected] & [email protected] & t2.5 + wisepc j141530.10 + 572426.1 & sdss j141530.05 + 572428.7 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.617 & [email protected] & @xmath415.821 & @xmath415.544 & t3 : + wisepc j142950.87 + 333011.5 & irac j142950.8 + 333011 & 27 & [email protected] & [email protected] & @xmath412.039 & @xmath49.256 & @xmath416.880 & & @xmath416.990 & t4.5 + wisepa j143553.29 + 112949.3 & sdss j143553.25 + 112948.6 & 8 & [email protected] & [email protected] & @xmath412.090 & @xmath49.090 & [email protected] & @xmath416.150 & @xmath416.906 & t2 : + wisepa j143945.75 + 304219.2 & sdss j143945.86 + 304220.6 & 8 & [email protected] & [email protected] & @xmath412.466 & @xmath49.152 & [email protected] & @xmath416.280 & @xmath415.881 & t2.5 + wisepa j145715.84 - 212207.5 & gliese 570d & 28 & [email protected] & [email protected] & [email protected] & @xmath48.577 & [email protected] & [email protected] & [email protected] & t7.5 + wisepa j145829.35 + 101341.8 & cfbdsir j145829 + 101343 & 29 & @xmath417.883 & [email protected] & @xmath412.421 & @xmath49.057 & [email protected] & [email protected] & [email protected] & t8 + + wisepa j150319.70 + 252525.4 & 2mass j15031961 + 2525196 & 30 & [email protected] & [email protected] & [email protected] & @xmath49.031 & [email protected] & [email protected] & [email protected] & t5 + wisepa j150411.81 + 102717.4 & sdss j150411.63 + 102718.4 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.073 & [email protected] & @xmath416.909 & @xmath417.023 & t7 + wisepa j151114.49 + 060740.9 & sdss j151114.66 + 060742.9 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.165 & [email protected] & [email protected] & [email protected] & t0 : : + wisepa j151603.01 + 025927.6 & sdss j151603.03 + 025928.9 & 16 & [email protected] & [email protected] & & @xmath4 null & [email protected] & [email protected] & [email protected] & t0 : + wisepa j151642.98 + 305344.3 & sdss j151643.01 + 305344.4 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.119 & [email protected] & [email protected] & [email protected] & t0.5 : + wisepa j152040.02 + 354617.2 & sdss j152039.82 + 354619.8 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.011 & [email protected] & [email protected] & [email protected] & t0 : + wisepa j152103.14 + 013144.1 & sdss j152103.24 + 013142.7 & 16 & [email protected] & [email protected] & [email protected] & @xmath48.884 & [email protected] & [email protected] & [email protected] & t2 : + wisepa j153417.03 + 161547.0 & sdss j153417.05 + 161546.1ab & 8 & [email protected] & [email protected] & @xmath412.517 & @xmath49.542 & [email protected] & [email protected] & @xmath416.411 & t3.5 + wisepa j153449.98 - 295230.8 & 2massi j1534498 - 295227 & 7 & [email protected] & [email protected] & [email protected] & @xmath49.064 & [email protected] & [email protected] & [email protected] & t5.5 + wisepa j154614.83 + 493204.7 & 2mass j15461461 + 4932114 & 31 & [email protected] & [email protected] & [email protected] & @xmath48.915 & [email protected] & [email protected] & @xmath414.900 & t2.5 : + wisepa j154627.30 - 332508.0 & 2massi j1546291 - 332511 & 7 & [email protected] & [email protected] & @xmath412.306 & @xmath48.664 & [email protected] & [email protected] & [email protected] & t5.5 + wisepa j155301.94 + 153238.6 & 2massi j1553022 + 153236 & 7 & [email protected] & [email protected] & [email protected] & @xmath49.204 & [email protected] & [email protected] & [email protected] & t7 + wisepa j161504.36 + 134004.2 & 2mass j16150413 + 1340079 & 12 & [email protected] & [email protected] & [email protected] & @xmath49.302 & [email protected] & [email protected] & @xmath415.859 & t6 + wisepa j162414.09 + 002916.0 & sdssp j162414.37 + 002915.6 & 32 & [email protected] & [email protected] & [email protected] & @xmath49.103 & [email protected] & [email protected] & @xmath415.518 & t6 + wisepa j162838.99 + 230818.1 & sdss j162838.77 + 230821.1 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.163 & [email protected] & [email protected] & [email protected] & t7 + wisepc j163022.92 + 081819.8 & sdss j163022.92 + 081822.0 & 8 & [email protected] & [email protected] & @xmath412.210 & @xmath49.418 & [email protected] & [email protected] & @xmath416.611 & t5.5 + wisepa j163239.30 + 415002.9 & sdss j163239.34 + 415004.3 & 16 & [email protected] & [email protected] & @xmath412.713 & @xmath49.455 & [email protected] & [email protected] & @xmath415.748 & t1 : + wisepa j175023.81 + 422239.0 & sdss j175024.01 + 422237.8 & 16 & [email protected] & [email protected] & [email protected] & @xmath49.424 & [email protected] & [email protected] & [email protected] & t2 + wisepa j175033.07 + 175905.5 & sdssp j175032.96 + 175903.9 & 6 & [email protected] & [email protected] & @xmath412.742 & @xmath48.905 & [email protected] & [email protected] & [email protected] & t3.5 + wisepa j175805.46 + 463317.5 & sdss j175805.46 + 463311.9 & 16 & [email protected] & [email protected] & [email protected] & @xmath49.305 & [email protected] & [email protected] & [email protected] & t6.5 + wisepa j182835.97 - 484903.0 & 2mass j18283572 - 4849046 & 10 & [email protected] & [email protected] & [email protected] & @xmath48.970 & [email protected] & [email protected] & [email protected] & t5.5 + wisepa j190106.18 + 471818.4 & 2mass j19010601 + 4718136 & 10 & [email protected] & [email protected] & [email protected] & @xmath49.618 & [email protected] & [email protected] & [email protected] & t5 + wisepc j205235.45 - 160928.7 & sdss j205235.31 - 160929.8 & 8 & [email protected] & [email protected] & @xmath412.366 & @xmath49.194 & [email protected] & [email protected] & [email protected] & t1 : + wisepc j212414.00 + 010002.6 & sdss j212413.89 + 010000.3 & 16 & [email protected] & [email protected] & @xmath412.493 & @xmath49.238 & [email protected] & [email protected] & @xmath416.144 & t5 + wisepc j213927.09 + 022023.7 & 2mass j21392676 + 0220226 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.018 & [email protected] & [email protected] & [email protected] & t1.5 + wisepc j214428.77 + 144606.2 & hn peg b & 34 & [email protected] & [email protected] & [email protected] & @xmath49.123 & [email protected] & [email protected] & [email protected] & t2.5 + wisepc j215138.86 - 485356.2 & 2mass j21513839 - 4853542 & 35 & [email protected] & [email protected] & [email protected] & @xmath49.432 & [email protected] & [email protected] & [email protected] & t4 + wisepc j215432.94 + 594213.7 & 2mass j21543318 + 5942187 & 12 & [email protected] & [email protected] & @xmath413.013 & @xmath49.678 & [email protected] & [email protected] & @xmath415.338 & t5 + wisepc j220415.66 - 564723.5 & eps indi bab & 36 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t2.5 + wisepc j222829.00 - 431029.4 & 2mass j22282889 - 4310262 & 9 & [email protected] & [email protected] & [email protected] & @xmath48.937 & [email protected] & [email protected] & [email protected] & t6 + wisepc j225418.97 + 312352.2 & 2massi j2254188 + 312349 & 7 & [email protected] & [email protected] & [email protected] & @xmath49.383 & [email protected] & [email protected] & [email protected] & t4 + wisepc j230600.99 + 130225.2 & ulas j230601.02 + 130225.0 & 5 & @xmath417.793 & [email protected] & @xmath412.801 & @xmath48.841 & [email protected] & [email protected] & [email protected] & t6.5 + wisepc j231557.68 + 132255.6 & ulas j231557.61 + 132256.2 & 5 & [email protected] & [email protected] & @xmath412.458 & @xmath49.301 & [email protected] & [email protected] & [email protected] & t6.5 + wisepc j232035.36 + 144830.2 & ulas j232035.28 + 144829.8 & 37 & [email protected] & [email protected] & @xmath412.633 & @xmath49.266 & [email protected] & [email protected] & [email protected] & t5 + wisepc j232123.83 + 135453.5 & ulas j232123.79 + 135454.9 & 37 & [email protected] & [email protected] & @xmath412.394 & @xmath49.435 & [email protected] & [email protected] & [email protected] & t7.5 + wisepc j232802.12 + 134545.2 & ulas j232802.03 + 134544.8 & 5 & [email protected] & [email protected] & @xmath412.617 & @xmath48.758 & [email protected] & [email protected] & [email protected] & t7 + wisepc j233123.87 - 471828.0 & 2mass j23312378 - 4718274 & 10 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & t5 + wisepc j233910.51 + 135218.8 & 2massi j2339101 + 135230 & 7 & [email protected] & [email protected] & @xmath412.215 & @xmath49.248 & [email protected] & [email protected] & [email protected] & t5 + wisepc j235654.40 - 155318.5 & 2massi j2356547 - 155310 & 7 & [email protected] & [email protected] & [email protected] & @xmath49.508 & [email protected] & [email protected] & [email protected] & t5.5 + + + wisepc j001544.82 + 351600.0 & 2massw j0015447 + 351603 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.985 & [email protected] & [email protected] & [email protected] & l2 + wisepc j003043.78 + 313931.8 & 2massw j0030438 + 313932 & 38 & [email protected] & [email protected] & [email protected] & @xmath49.239 & [email protected] & [email protected] & [email protected] & l2 + wisepa j012912.38 + 351757.4 & 2massw j0129122 + 351758 & 38 & [email protected] & [email protected] & [email protected] & @xmath49.283 & [email protected] & [email protected] & [email protected] & l4 + wisepa j013118.73 + 380155.0 & 2mass j01311838 + 3801554 & 39 & [email protected] & [email protected] & [email protected] & @xmath49.171 & [email protected] & [email protected] & [email protected] & l4 : + wisepc j013535.83 + 120517.7 & 2massw j0135358 + 120522 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.987 & [email protected] & [email protected] & [email protected] & l1.5 + wisepc j014435.67 - 071616.3 & 2mass j01443536 - 0716142 & 40 & [email protected] & [email protected] & [email protected] & @xmath48.841 & [email protected] & [email protected] & [email protected] & l5 + wisepa j014733.46 + 345310.5 & 2massw j0147334 + 345311 & 38 & [email protected] & [email protected] & [email protected] & @xmath49.194 & [email protected] & [email protected] & [email protected] & l0.5 + wisepc j020333.45 - 010813.4 & sdss
j020333.26 - 010812.5 & 16 & [email protected] & [email protected] & @xmath412.479 & @xmath48.700 & @xmath416.858 & [email protected] & [email protected] & l9.5 ( ir ) + wisepa j020503.70 + 125142.2 & 2massw j0205034 + 125142 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.764 & [email protected] & [email protected] & [email protected] & l5 + wisepc j020529.69 - 115929.1 & denis - p j0205.4 - 1159 & 41 & [email protected] & [email protected] & [email protected] & @xmath48.651 & [email protected] & [email protected] & [email protected] & l7 + wisepa j020735.76 + 135554.4 & sdss j020735.60 + 135556.3 & 22 & [email protected] & [email protected] & @xmath412.495 & @xmath48.958 & [email protected] & [email protected] & [email protected] & l3 + wisepa j020818.65 + 254253.0 & 2massw j0208183 + 254253 & 20 & [email protected] & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] & l1 + wisepa j020823.82 + 273738.7 & 2massw j0208236 + 273740 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.526 & [email protected] & [email protected] & [email protected] & l5 + wisepa j020854.98 + 250049.0 & 2massw j0208549 + 250048 & 20 & [email protected] & [email protected] & @xmath411.418 & @xmath48.106 & [email protected] & [email protected] & [email protected] & l5 + wisepc j021828.98 - 313324.1 & 2massi j0218291 - 313322 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.474 & [email protected] & [email protected] & [email protected] & l3 + wisepc j021928.20 - 193842.7 & ssspm j0219 - 1939 & 43 & [email protected] & [email protected] & [email protected] & @xmath48.868 & [email protected] & [email protected] & [email protected] & l1 + wisepc j022710.62 - 162450.7 & 2mass j02271036 - 1624479 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.842 & [email protected] & [email protected] & [email protected] & l1 + wisepa j022811.20 + 253737.8 & 2massw j0228110 + 253738 & 44 & [email protected] & [email protected] & [email protected] & @xmath49.240 & [email protected] & [email protected] & [email protected] & l0 : + wisepa j023015.66 + 270405.9 & 2mass j02301551 + 2704061 & 39 & [email protected] & [email protected] & [email protected] & @xmath49.023 & [email protected] & [email protected] & [email protected] & l0 : + wisepa j023547.53 - 084919.7 & sdss j023547.56 - 084919.8 & 22 & [email protected] & [email protected] & @xmath412.048 & @xmath49.076 & [email protected] & [email protected] & [email protected] & l2 + wisepa j023618.01 + 004852.8 & sdssp j023617.93 + 004855.0 & 6 & [email protected] & [email protected] & [email protected] & @xmath49.050 & [email protected] & [email protected] & [email protected] & l6 + wisepa j023942.48 - 173548.3 & 2massi j0239424 - 173547 & 42 & [email protected] & [email protected] & [email protected] & @xmath48.598 & [email protected] & [email protected] & [email protected] & l0 + wisepa j024153.91 - 124107.5 & 2massi j0241536 - 124106 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.295 & [email protected] & [email protected] & [email protected] & l2 : + wisepa j024243.68 + 160737.6 & 2massw j0242435 + 160739 & 38 & [email protected] & [email protected] & @xmath412.324 & @xmath49.075 & [email protected] & [email protected] & [email protected] & l1.5 + wisepa j025601.91 + 011046.3 & sdss j025601.86 + 011047.2 & 22 & [email protected] & [email protected] & @xmath412.464 & @xmath49.080 & [email protected] & [email protected] & [email protected] & l0 + wisepa j030136.57 + 002057.0 & sdssp j030136.53 + 002057.9 & 45 & [email protected] & [email protected] & @xmath412.585 & @xmath49.167 & [email protected] & [email protected] & @xmath415.619 & l1 : + wisepa j030321.24 - 000938.0 & sdssp j030321.24 - 000938.2 & 45 & [email protected] & [email protected] & [email protected] & @xmath48.742 & [email protected] & [email protected] & [email protected] & l0 + wisepa j030909.04 - 194939.0 & 2massw j0309088 - 194938 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.444 & [email protected] & [email protected] & [email protected] & l4.5 + wisepa j031013.90 - 275645.8 & 2mass j03101401 - 2756452 & 39 & [email protected] & [email protected] & [email protected] & @xmath49.008 & [email protected] & [email protected] & [email protected] & l5 + wisepa j031645.21 - 284852.8 & 2massi j0316451 - 284852 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.191 & [email protected] & [email protected] & [email protected] & l0 : + wisepa j031854.37 - 342128.7 & 2mass j03185403 - 3421292 & 46 & [email protected] & [email protected] & [email protected] & @xmath49.214 & [email protected] & [email protected] & [email protected] & l7 + wisepa j033703.73 - 175806.6 & 2massw j0337036 - 175807 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.151 & [email protected] & [email protected] & [email protected] & l4.5 + wisepa j035822.59 - 411605.0 & 2mass j03582255 - 4116060 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.419 & [email protected] & [email protected] & [email protected] & l5 + wisepa j065219.57 - 253449.5 & denis - p j0652197 - 253450 & 47 & [email protected] & [email protected] & [email protected] & @xmath48.784 & [email protected] & [email protected] & [email protected] & l0 + wisepa j065230.58 + 471036.3 & 2massi j0652307 + 471034 & 42 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & l4.5 + wisepa j071716.22 + 570543.6 & 2massw j0717163 + 570543 & 44 & [email protected] & [email protected] & [email protected] & @xmath48.907 & [email protected] & [email protected] & [email protected] & l3 + wisepc j071932.08 - 505141.5 & 2mass j07193188 - 5051410 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.190 & [email protected] & [email protected] & [email protected] & l0 + wisepa j072314.69 + 572705.6 & 2mass j07231462 + 5727081 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.427 & [email protected] & [email protected] & [email protected] & l1 + wisepa j074642.23 + 200031.5 & 2massi j0746425 + 200032 & 48 & [email protected] & [email protected] & [email protected] & @xmath48.985 & [email protected] & [email protected] & [email protected] & l0.5 + wisepa j075259.41 + 413634.6 & sdss j075259.43 + 413634.6 & 22 & [email protected] & [email protected] & @xmath412.600 & @xmath48.969 & [email protected] & [email protected] & @xmath415.191 & l0 + wisepa j075332.07 + 291710.3 & 2massi j0753321 + 291711 & 20 & [email protected] & [email protected] & @xmath411.874 & @xmath48.419 & [email protected] & [email protected] & [email protected] & l2 + wisepa j075515.18 + 293445.0 & sdss j075515.26 + 293445.4 & 16 & [email protected] & [email protected] & @xmath412.325 & @xmath49.073 & [email protected] & [email protected] & [email protected] & l3.5 : ( ir ) + wisepa j075625.28 + 124454.2 & 2massi j0756252 + 124456 & 20 & [email protected] & [email protected] & @xmath412.032 & @xmath48.803 & [email protected] & [email protected] & [email protected] & l6 + wisepa j075656.43 + 231455.4 & sdss j075656.54 + 231458.5 & 16 & [email protected] & [email protected] & @xmath411.893 & @xmath49.174 & [email protected] & [email protected] & [email protected] & l3.5 : ( ir ) + wisepa j080048.13 + 465825.8 & sdss j080048.13 + 465825.5 & 22 & [email protected] & [email protected] & @xmath412.307 & @xmath49.124 & [email protected] & [email protected] & [email protected] & l2 + wisepa j080140.31 + 462845.8 & 2massw j0801405 + 462850 & 20 & [email protected] & [email protected] & @xmath412.846 & @xmath49.310 & [email protected] & [email protected] & [email protected] & l6.5 + wisepa j080322.74 + 123845.2 & sdss j080322.77 + 123845.3 & 49 & [email protected] & [email protected] & @xmath412.339 & @xmath48.947 & [email protected] & [email protected] & [email protected] & l2.5 + wisepa j080531.37 + 481233.7 & sdss j080531.84 + 481233.0 & 22 & [email protected] & [email protected] & [email protected] & @xmath48.829 & [email protected] & [email protected] & [email protected] & l4 + wisepa j080958.86 + 443419.5 & sdss j080959.01 + 443422.2 & 16 & [email protected] & [email protected] & [email protected] & @xmath49.002 & [email protected] & [email protected] & [email protected] & l6 ( ir ) + wisepc j081231.81 - 244443.9 & denis - p j0812316 - 244442 & 47 & [email protected] & [email protected] & [email protected] & @xmath49.100 & [email protected] & [email protected] & [email protected] & l2.5+/-1 + wisepa j082029.82 + 450027.4 & 2massw j0820299 + 450031 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.996 & [email protected] & [email protected] & [email protected] & l5 + wisepa j082030.08 + 103736.8 & sdss j082030.12 + 103737.0 & 8 & [email protected] & [email protected] & [email protected] & @xmath48.911 & [email protected] & [email protected] & [email protected] & l9.5 : ( ir ) + wisepa j082307.65 + 612518.5 & 2mass j08230838 + 6125208 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.989 & [email protected] & [email protected] & [email protected] & l2 : + wisepa j082348.05 + 242858.2 & 2mass j08234818 + 2428577 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.070 & [email protected] & [email protected] & [email protected] & l3 + wisepa j082906.58 + 145619.6 & 2massw j0829066 + 145622 & 20 & [email protected] & [email protected] & @xmath412.156 & @xmath48.997 & [email protected] & [email protected] & [email protected] & l2 + wisepa j082957.00 + 265509.4 & 2massw j0829570 + 265510 & 20 & [email protected] & [email protected] & @xmath412.625 & @xmath49.089 & [email protected] & [email protected] & [email protected] & l6.5 + wisepa j083204.55 - 012835.8 & 2massw j0832045 - 012835 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.026 & [email protected] & [email protected] & [email protected] & l1.5 + wisepa j083506.10 + 195303.6 & sdss j083506.16 + 195304.4 & 8 & [email protected] & [email protected] & @xmath411.939 & @xmath48.618 & [email protected] & [email protected] & [email protected] & l4.5 ( ir ) + wisepa j083558.21 + 054830.5 & 2mass j08355829 + 0548308 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.443 & [email protected] & [email protected] & [email protected] & l2 : + wisepa j084016.40 + 543001.4 & sdss j084016.42 + 543002.1 & 49 & [email protected] & [email protected] & [email protected] & @xmath49.106 & [email protected] & [email protected] & [email protected] & l1 + wisepc j084728.84 - 153239.4 & 2massi j0847287 - 153237 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.052 & [email protected] & [email protected] & [email protected] & l2 + wisepa j085234.85 + 472031.0 & sdss j085234.90 + 472035.0 & 16 & [email protected] & [email protected] & [email protected] & @xmath49.056 & [email protected] & [email protected] & [email protected] & l9.5 : ( ir ) + wisepa j085757.95 + 570847.5 & sdssp j085758.45 + 570851.4 & 6 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & l8 + wisepa j085938.38 + 634130.4 & 2mass j08593854 + 6341355 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.135 & [email protected] & [email protected] & [email protected] & l0 + wisepc j090347.54 + 011445.2 & sdss j090347.55 + 011446.0 & 49 & [email protected] & [email protected] & @xmath412.855 & @xmath49.400 & [email protected] & [email protected] & [email protected] & l2 + wisepa j090546.55 + 562312.9 & 2mass j09054654 + 5623117 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.113 & [email protected] & [email protected] & [email protected] & l5 + wisepa j090837.56 + 503203.6 & 2massi j0908380 + 503208 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.153 & [email protected] & [email protected] & [email protected] & l5 + wisepc j090957.35 - 065818.4 & denis - p j0909 - 0658 & 50 & [email protected] & [email protected] & [email protected] & @xmath48.466 & [email protected] & [email protected] & [email protected] & l0 + wisepa j091112.50 + 740106.6 & 2mass j09111297 + 7401081 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.896 & [email protected] & [email protected] & [email protected] & l0 + wisepc j091303.23 + 184147.9 & 2massw j0913032 + 184150 & 38 & [email protected] & [email protected] & @xmath412.600 & @xmath49.136 & [email protected] & [email protected] & [email protected] & l3 + wisepc j091534.08 + 042205.1 & 2mass j09153413 + 0422045 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.502 & [email protected] & [email protected] & [email protected] & l7 + wisepa j091714.75 + 314824.1 & sdss j091714.76 + 314824.8 & 49 & [email protected] & [email protected] & @xmath411.971 & @xmath48.955 & [email protected] & [email protected] & [email protected] & l2 + wisepc j091838.44 + 213404.1 & 2massw j0918382 + 213406 & 38 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath414.580 & [email protected] & l2.5 + wisepa j092757.41 + 602746.0 & sdss j092757.46 + 602746.3 & 22 & [email protected] & [email protected] & @xmath412.380 & @xmath49.218 & [email protected] & [email protected] & [email protected] & l1 + wisepc j092933.44 + 342951.8 & 2massw j0929336 + 342952 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.158 & [email protected] & [email protected] & [email protected] & l8 + wisepc j094402.83 + 313132.1 & 2massw j0944027 + 313132 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.614 & [email protected] & [email protected] & [email protected] & l2 + wisepc j102214.76 + 411425.6 & hd 89744b & 51 & [email protected] & [email protected] & @xmath412.021 & @xmath48.766 & [email protected] & [email protected] & [email protected] & l0 + wisepc j102552.58 + 321231.5 & sdss j102552.43 + 321234.0 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.395 & @xmath415.913 & [email protected] & [email protected] & l7.5::(ir ) + wisepc j102947.62 + 483411.8 & sdss j102947.68 + 483412.2 & 49 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & l0 + wisepc j103322.01 + 400547.8 & sdss j103321.92 + 400549.5 & 8 & [email protected] & [email protected] & @xmath412.258 & @xmath48.967 & [email protected] & [email protected] & @xmath415.408 & l6 ( ir ) + wisepc j103602.46 + 372447.4 & sdss j103602.44 + 372448.5 & 49 & [email protected] & [email protected] & @xmath412.351 & @xmath49.434 & [email protected] & [email protected] & [email protected] & l0 + wisepc j105118.71 + 561305.5 & 2mass j10511900 + 5613086 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.098 & [email protected] & [email protected] & [email protected] & l2 + wisepc j110009.49 + 495745.4 & 2mass j11000965 + 4957470 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.801 & [email protected] & [email protected] & [email protected] & l3.5 + wisepa j110830.33 + 683014.8 & 2massw j1108307 + 683017 & 52 & [email protected] & [email protected] & [email protected] & @xmath49.464 & [email protected] & [email protected] & [email protected] & l1 + wisepc j113833.15 + 674038.8 & sdss j113833.10 + 674040.3 & 22 & [email protected] & [email protected] & @xmath412.553 & @xmath49.527 & [email protected] & [email protected] & [email protected] & l0 + wisepc j115442.09 - 340038.8 & 2mass j11544223 - 3400390 & 53 & [email protected] & [email protected] & [email protected] & @xmath49.502 & [email protected] & [email protected] & [email protected] & l0 + wisepc j120703.54 - 315130.5 & 2mass j12070374 - 3151298 & 33 & [email protected] & [email protected] & @xmath412.420 & @xmath49.059 & [email protected] & [email protected] & [email protected] & l3 : + wisepc j120737.92 - 390904.6 & 2mass j12073804 - 3909050 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.048 & [email protected] & [email protected] & [email protected] & l2 : + wisepc j122815.35 - 154736.4 & denis - p j1228.2 - 1547 & 41 & [email protected] & [email protected] & [email protected] & @xmath49.162 & [email protected] & [email protected] & [email protected] & l5 + wisepc j125656.74 + 014615.9 & 2mass j12565688 + 0146163 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.018 & [email protected] & [email protected] & [email protected] & l2 : + wisepc j125737.34 - 011336.5 & sdssp j125737.26 - 011336.1 & 6 & [email protected] & [email protected] & @xmath411.856 & @xmath48.969 & [email protected] & [email protected] & [email protected] & l4 + wisepc j130154.60 - 151023.0 & 2mass j13015465 - 1510223 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.196 & [email protected] & [email protected] & [email protected] & l1 + wisepa j130539.94 - 254106.1 & kelu-1aab & 54 & [email protected] & [email protected] & [email protected] & @xmath49.051 & [email protected] & [email protected] & [email protected] & l2 + wisepa j131530.35 - 264954.6 & 2massi j1315309 - 264951 & 55 & [email protected] & [email protected] & [email protected] & @xmath48.957 & [email protected] & [email protected] & [email protected] & l5.5 + wisepa j132619.81 - 272937.2 & 2massw j1326201 - 272937 & 56 & [email protected] & [email protected] & [email protected] & @xmath48.560 & [email protected] & [email protected] & [email protected] & l5 + wisepc j132629.66 - 003832.6 & sdssp
j132629.82 - 003831.5 & 57 & [email protected] & [email protected] & [email protected] & @xmath49.150 & [email protected] & [email protected] & [email protected] & l8 : + wisepc j133406.18 + 194036.4 & 2massw j1334062 + 194034 & 38 & [email protected] & [email protected] & [email protected] & @xmath49.170 & [email protected] & [email protected] & [email protected] & l1.5 + wisepa j134111.63 - 305252.1 & 2mass j13411160 - 3052505 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.039 & [email protected] & [email protected] & [email protected] & l2 : : + wisepc j134203.12 + 134022.4 & sdss j134203.11 + 134022.2 & 8 & [email protected] & [email protected] & @xmath411.353 & @xmath47.451 & [email protected] & [email protected] & [email protected] & l5.5 ( ir ) + wisepa j135955.14 - 403503.3 & 2mass j13595510 - 4034582 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.241 & [email protected] & [email protected] & [email protected] & l1 + wisepa j140231.60 + 014830.1 & sdss j140231.75 + 014830.3 & 22 & [email protected] & [email protected] & [email protected] & @xmath49.046 & [email protected] & [email protected] & [email protected] & l1 + wisepa j140441.70 + 023547.5 & sdss j140441.68 + 023550.1 & 22 & [email protected] & [email protected] & @xmath411.891 & @xmath48.882 & [email protected] & [email protected] & [email protected] & l1 + wisepa j140753.39 + 124111.0 & 2mass j14075361 + 1241099 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.028 & [email protected] & [email protected] & [email protected] & l1 : : + wisepa j140903.19 - 335755.8 & 2mass j14090310 - 3357565 & 46 & [email protected] & [email protected] & [email protected] & @xmath49.098 & [email protected] & [email protected] & [email protected] & l2 + wisepa j141011.08 + 132900.8 & sdss j141011.14 + 132900.8 & 49 & [email protected] & [email protected] & @xmath412.196 & @xmath48.902 & [email protected] & [email protected] & [email protected] & l4 + wisepa j141224.51 + 163310.5 & 2massw j1412244 + 163312 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.828 & [email protected] & [email protected] & [email protected] & l0.5 + wisepa j142227.23 + 221558.3 & sdss j142227.25 + 221557.1 & 8 & [email protected] & [email protected] & @xmath412.830 & @xmath49.104 & [email protected] & [email protected] & [email protected] & l6.5::(ir ) + wisepa j142438.96 + 091709.8 & gd 165b & 58 & [email protected] & [email protected] & @xmath412.466 & @xmath49.095 & [email protected] & [email protected] & [email protected] & l4 + wisepa j143043.42 + 291542.1 & 2massi j1430435 + 291540 & 44 & [email protected] & [email protected] & [email protected] & @xmath48.465 & [email protected] & [email protected] & [email protected] & l2 + wisepa j143055.92 + 001351.8 & sdss j143055.90 + 001352.1 & 22 & [email protected] & [email protected] & @xmath412.506 & @xmath48.435 & [email protected] & [email protected] & [email protected] & l0 + wisepa j143130.64 + 143653.0 & 2mass j14313097 + 1436539 & 21 & [email protected] & [email protected] & [email protected] & @xmath49.140 & [email protected] & [email protected] & [email protected] & l2 + wisepa j143517.22 - 004612.6 & sdss
j143517.20 - 004612.9 & 22 & [email protected] & [email protected] & @xmath412.107 & @xmath48.924 & [email protected] & [email protected] & [email protected] & l0 + wisepa j143535.75 - 004347.4 & sdss
j143535.72 - 004347.0 & 22 & [email protected] & [email protected] & @xmath412.450 & @xmath48.405 & [email protected] & [email protected] & [email protected] & l3 + wisepa j143855.11 - 130910.8 & 2massw j1438549 - 130910 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.250 & [email protected] & [email protected] & [email protected] & l3 : + wisepa j143940.90 + 182637.7 & 2massw j1439409 + 182637 & 38 & [email protected] & [email protected] & @xmath412.224 & [email protected] & [email protected] & [email protected] & [email protected] & l1 + wisepa j144001.75 + 002146.7 & sdssp j144001.82 + 002145.8 & 57 & [email protected] & [email protected] & @xmath412.625 & @xmath49.056 & [email protected] & [email protected] & [email protected] & l1 + wisepa j144016.21 + 002639.5 & sdss j144016.20 + 002638.9 & 22 & [email protected] & [email protected] & @xmath412.115 & @xmath49.024 & [email protected] & [email protected] & [email protected] & l3 + wisepa j144600.73 + 002451.2 & sdssp j144600.60 + 002452.0 & 6 & [email protected] & [email protected] & [email protected] & @xmath48.712 & [email protected] & [email protected] & [email protected] & l6 + wisepc j144938.12 + 235536.3 & 2massw j1449378 + 235537 & 20 & [email protected] & [email protected] & @xmath412.473 & [email protected] & [email protected] & [email protected] & [email protected] & l0 + wisepa j151136.27 + 353511.5 & sdss j151136.24 + 353511.4 & 49 & [email protected] & [email protected] & @xmath412.632 & @xmath49.139 & [email protected] & [email protected] & [email protected] & l1 + wisepa j151506.18 + 443648.1 & sdss j151506.11 + 443648.3 & 8 & [email protected] & [email protected] & [email protected] & @xmath48.848 & [email protected] & [email protected] & [email protected] & l7.5::(ir ) + wisepa j152322.73 + 301454.3 & gl 584c & 59 & [email protected] & [email protected] & [email protected] & @xmath49.187 & [email protected] & [email protected] & [email protected] & l8 + wisepa j152613.85 + 204337.0 & 2massi j1526140 + 204341 & 20 & [email protected] & [email protected] & [email protected] & @xmath48.734 & [email protected] & [email protected] & [email protected] & l7 + wisepa j154009.25 + 374227.7 & sdss j154009.36 + 374230.3 & 8 & [email protected] & [email protected] & [email protected] & @xmath49.053 & [email protected] & [email protected] & [email protected] & l9 : : ( ir ) + wisepa j154508.84 + 355528.5 & sdss j154508.93 + 355527.3 & 8 & [email protected] & [email protected] & @xmath412.566 & @xmath49.099 & [email protected] & [email protected] & [email protected] & l7.5 + wisepa j161542.46 + 495321.3 & 2mass j16154255 + 4953211 & 39 & [email protected] & [email protected] & [email protected] & @xmath48.862 & [email protected] & [email protected] & [email protected] & l4 + wisepa j163229.39 + 190439.9 & 2massw j1632291 + 190441 & 38 & [email protected] & [email protected] & [email protected] & @xmath49.398 & [email protected] & [email protected] & [email protected] & l8 + wisepc j171714.33 + 652620.9 & sdss j171714.10 + 652622.2 & 22 & [email protected] & [email protected] & [email protected] & @xmath49.827 & [email protected] & [email protected] & [email protected] & l4 + wisepa j172006.70 + 615536.5 & sdss j172006.69 + 615537.7 & 49 & [email protected] & [email protected] & @xmath413.442 & @xmath49.372 & & & & l3 + wisepa j180715.97 + 501530.2 & 2massi j1807159 + 501531 & 42 & [email protected] & [email protected] & [email protected] & @xmath48.967 & [email protected] & [email protected] & [email protected] & l1.5 + wisepa j184108.66 + 311728.3 & 2massw j1841086 + 311727 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.017 & [email protected] & [email protected] & [email protected] & l4pec + wisepa j190908.18 - 193749.6 & denis - p j1909081 - 193748 & 47 & [email protected] & [email protected] & & @xmath4 null & [email protected] & [email protected] & [email protected] & l1+/-1 + wisepa j200250.64 - 052153.6 & 2mass j20025073 - 0521524 & 39 & [email protected] & [email protected] & [email protected] & @xmath48.553 & [email protected] & [email protected] & [email protected] & l6 + wisepc j202820.42 + 005226.6 & sdss j202820.32 + 005226.5 & 22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & l3 + wisepc j203437.63 + 082656.0 & 2mass j20343769 + 0827009 & 33 &
[email protected] & [email protected] & [email protected] & @xmath48.225 & [email protected] & [email protected] & [email protected] & l1 + wisepc j203603.08 + 105127.8 & 2mass j20360316 + 1051295 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.995 & [email protected] & [email protected] & [email protected] & l3 + wisepa j204142.87 - 350645.8 & 2mass j20414283 - 3506442 & 39 & [email protected] & [email protected] & [email protected] & @xmath48.779 & [email protected] & [email protected] & [email protected] & l2 : + wisepc j204317.70 - 155104.0 & sdss j204317.69 - 155103.4 & 8 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & l9 ( ir ) + wisepc j205754.09 - 025230.9 & 2massi j2057540 - 025230 & 42 & [email protected] & [email protected] & [email protected] & @xmath49.101 & [email protected] & [email protected] & [email protected] & l1.5 + wisepa j210754.20 - 454406.2 & 2mass j21075409 - 4544064 & 33 & [email protected] & [email protected] & [email protected] & @xmath49.137 & [email protected] & [email protected] & [email protected] & l0 : + wisepc j220645.11 - 421723.1 & 2massw j2206450 - 421721 & 20 & [email protected] & [email protected] & [email protected] & @xmath49.392 & [email protected] & [email protected] & [email protected] & l2 + wisepc j224316.94 - 593223.0 & 2mass j22431696 - 5932206 & 60 & [email protected] & [email protected] & [email protected] & @xmath48.359 & [email protected] & [email protected] & [email protected] & l0 : : + + + wisepc j002127.85 - 635108.6 & f 079 - 18 & 80 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m5.5 v + wisepc j003519.14 + 280715.5 & cti 003439.1 + 280309 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.950 & [email protected] & [email protected] & [email protected] & m4 v + wisepc j004324.51 + 280543.7 & cti 004244.4 + 280140 & 62 & [email protected] & [email protected] & @xmath412.893 & @xmath49.414 & [email protected] & [email protected] & [email protected] & m6 v + wisepc j004446.85 + 280741.8 & cti 004406.7 + 280336 & 62 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m1.5 v + wisepc j004605.87 + 280841.3 & cti 004525.7 + 280437 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.900 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepc j010412.78 + 280634.3 & cti 010332.0 + 280234 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.029 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepc j011749.92 + 280245.3 & cti 011708.6 + 275850 & 62 & [email protected] & [email protected] & @xmath412.704 & @xmath49.258 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepa j011908.43 + 280910.9 & cti 011826.7 + 280514 & 62 & [email protected] & [email protected] & @xmath412.223 & @xmath49.154 & [email protected] & [email protected] & [email protected] & m6 v + wisepa j012559.40 + 280640.4 & cti 012517.7 + 280247 & 62 & [email protected] & [email protected] & @xmath412.079 & @xmath49.103 & [email protected] & [email protected] & [email protected] & m0 v + wisepa j012739.04 + 280552.3 & cti 012657.5 + 280202 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.992 & [email protected] & [email protected] & [email protected] & m8.5 v + wisepc j012821.03 + 021116.5 & lhs 1252 & 70 & [email protected] & [email protected] & [email protected] & @xmath48.673 & [email protected] & [email protected] & [email protected] & m6 v + wisepa j013825.39 + 280943.7 & cti 013743.0 + 280553 : & 62 & [email protected] & [email protected] & @xmath412.595 & @xmath48.570 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j015158.49 + 280838.4 & cti 015115.8 + 280458 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.240 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepa j015209.72 + 340034.1 & 2masp j0152096 + 340037 & 67 & [email protected] & [email protected] & [email protected] & @xmath48.935 & [email protected] & [email protected] & [email protected] & m6.5 v + wisepa j015421.12 + 280414.9 & cti 015338.5 + 280036 & 62 & [email protected] & [email protected] & @xmath412.085 & @xmath48.708 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepa j015748.95 + 280854.5 & cti 015705.9 + 280519 & 62 & [email protected] & [email protected] & @xmath412.348 & @xmath48.557 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j015908.24 + 280457.8 & cti 015825.4 + 280120 & 62 & [email protected] & [email protected] & & @xmath47.048 & [email protected] & [email protected] & [email protected] & m1.5 v + wisepa j020215.77 + 102011.1 & lhs 1326 & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j021816.35 - 274049.1 & lp 885 - 45 & 68 & [email protected] & [email protected] & [email protected] & @xmath49.391 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j022537.02 + 240833.8 & lp 353 - 59 & 72 & [email protected] & [email protected] & [email protected] & @xmath48.217 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j024840.95 - 165124.8 & lp 771 - 21 & 68 & [email protected] & [email protected] & [email protected] & @xmath48.671 & [email protected] & [email protected] & [email protected] & m8 v + wisepa j025226.18 + 005621.5 & tvlm 832 - 10443 & 66 & [email protected] & [email protected] & [email protected] & @xmath48.603 & [email protected] & [email protected] & [email protected] & m8 v + wisepa j031547.91 + 280600.1 & cti 031502.6 + 280315 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.090 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepa j032059.95 + 185420.3 & lp 412 - 31 & 63 & [email protected] & [email protected] & [email protected] & @xmath48.461 & [email protected] & [email protected] & [email protected] & m8 v + wisepa j032527.98 + 280636.0 & cti 032442.5 + 280400 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.965 & [email protected] & [email protected] & [email protected] & m4 v + wisepa j032744.93 + 280146.3 & cti 032659.6 + 275912 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.044 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j033935.48 - 352541.0 & lp 944 - 20 & 69 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m9 v + wisepa j065038.51 + 280347.1 & cti 064951.4 + 280442 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.055 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j065237.89 + 280214.3 & cti 065150.8 + 280311 & 62 & [email protected] & [email protected] & @xmath412.280 & @xmath48.617 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j065757.51 + 621913.9 & lhs 1885 & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j070037.46 + 280122.5 & cti 065950.5 + 280228 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.496 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j070753.25 - 490046.2 & eso 207 - 61 & 74 & [email protected] & [email protected] & [email protected] & @xmath49.515 & [email protected] & [email protected] & [email protected] & m8.5 v + wisepa j071931.60 + 324944.9 & gl 270 & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m0 v + wisepa j073625.27 + 070440.0 & g 89 - 32 & 75 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j075059.51 + 280415.4 & cti 075013.2 + 280613 & 62 & [email protected] & [email protected] & @xmath412.435 & @xmath49.115 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j075353.10 + 085359.9 & lhs 5129 & 70 & [email protected] & [email protected] & [email protected] & @xmath49.134 & [email protected] & [email protected] & [email protected] & m6 v + wisepa j075809.36 + 280329.2 & cti 075723.2 + 280533 & 62 & [email protected] & [email protected] & @xmath412.587 & @xmath49.150 & [email protected] & [email protected] & [email protected] & m4 v + wisepa j083618.33 - 192040.1 & lp 785 - 18 & 68 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m3 v + wisepa j085335.80 - 032934.2 & lhs 2065 & 64 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m9 v + wisepa j090205.32 + 280138.2 & cti 090120.6 + 280439 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.844 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j090206.59 + 003318.5 & lhs 5142 & 70 & [email protected] & [email protected] & [email protected] & @xmath48.858 & [email protected] & [email protected] & [email protected] & m6 v + wisepc j091800.96 + 280220.5 & cti 091716.8 + 280531 & 62 & [email protected] & [email protected] & @xmath412.380 & @xmath49.414 & [email protected] & [email protected] & [email protected] & m6 v + wisepa j092416.05 + 351644.1 & g 117-b15 b & 75 & [email protected] & [email protected] & [email protected] & @xmath48.871 & [email protected] & [email protected] & [email protected] & m3.5 v + wisepc j092507.00 + 275727.8 & cti 092423.0 + 280044 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.344 & [email protected] & [email protected] & [email protected] & m3 v + wisepc j092623.81 + 275700.9 & cti 092539.9 + 280018 & 62 & [email protected] & [email protected] & @xmath411.995 & @xmath49.000 & [email protected] & [email protected] & [email protected] & m6.5 v + wisepc j093714.90 + 275913.2 & cti 093631.3 + 280237 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.289 & [email protected] & [email protected] & [email protected] & m3 v + wisepc j094222.68 + 424533.4 & tvlm 262 - 111511 & 66 & [email protected] & [email protected] & [email protected] & @xmath49.281 & [email protected] & [email protected] & [email protected] & m8.5
v + wisepc j095147.80 + 423348.7 & tvlm 262 - 70502 & 66 & [email protected] & [email protected] & [email protected] & @xmath49.274 & [email protected] & [email protected] & [email protected] & m6.5 v + wisepc j095723.93 + 263250.6 & lp 315 - 11 & 72 & [email protected] & [email protected] & [email protected] & @xmath48.564 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepc j100203.26 + 454456.3 & lp 166 - 56 & 71 & [email protected] & [email protected] & [email protected] & @xmath49.261 & [email protected] & [email protected] & [email protected] & m5 v + wisepc j100743.69 + 113429.3 & lspm j1007 + 1134 & 78 & [email protected] & [email protected] & [email protected] & @xmath49.389 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepc j101634.54 + 275144.1 & lhs 2243 & 65 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m8 v + wisepc j102236.87 + 301657.6 & pss 1022 + 3016 & 80 & [email protected] & [email protected] & [email protected] & @xmath49.192 & [email protected] & [email protected] & [email protected] & m6 v + wisepc j102921.73 + 275937.9 & cti 102840.1 + 280330 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.564 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepc j103042.10 + 280040.4 & cti 103000.3 + 280432 & 62 & [email protected] & [email protected] & @xmath411.871 & @xmath48.676 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j133100.85 - 453507.2 & wt 426 & 79 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m3.5 v + wisepc j135421.74 + 280041.2 & lp 324 - 30 & 72 & [email protected] & [email protected] & [email protected] & @xmath48.710 & [email protected] & [email protected] & [email protected] & m6 v + wisepa j135620.19 - 280350.7 & lhs 2826 & 70 & [email protected] & [email protected] & [email protected] & @xmath48.644 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j135644.07 - 173014.0 & lp 799 - 3 & 68 & [email protected] & [email protected] & @xmath412.042 & @xmath48.553 & [email protected] & [email protected] & [email protected] & m4.5 : : v + wisepa j135910.02 - 195005.4 & lhs 2836 & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m4.5 v + wisepc j141107.82 + 275826.3 & cti 141034.2 + 280158 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.092 & [email protected] & [email protected] & [email protected] & m4.5 v + wisepc j141144.97 + 275808.2 & lp 325 - 1 & 72 & [email protected] & [email protected] & [email protected] & @xmath49.388 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepc j141311.37 + 280226.4 & cti 141237.7 + 280557 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.411 & [email protected] & [email protected] & [email protected] & m3 v + wisepc j141403.20 + 275801.0 & cti 141329.8 + 280131 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.524 & [email protected] & [email protected] & [email protected] & m2.5 v + wisepa j145824.46 + 283953.3 & 2mass j14582453 + 2839580 & 33 & [email protected] & [email protected] & [email protected] & @xmath48.876 & [email protected] & [email protected] & [email protected] & m8.5 v + wisepa j150108.17 + 225001.4 & tvlm 513 - 46546 & 66 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m8.5 v + wisepa j150618.61 + 275824.0 & cti 150546.8 + 280117 & 62 & [email protected] & [email protected] & @xmath412.177 & @xmath49.038 & [email protected] & [email protected] & @xmath414.941 & m5.5 v + wisepa j151016.58 - 024107.6 & tvlm 868 - 110639 & 66 & [email protected] & [email protected] & [email protected] & @xmath49.004 & [email protected] & [email protected] & [email protected] & m9 v + wisepa j151420.46 + 234104.1 & tvlm 513 - 8328 & 66 & [email protected] & [email protected] & [email protected] & @xmath49.224 & [email protected] & [email protected] & [email protected] & m8 v + wisepa j151943.12 + 260936.2 & 2masp j1519431 + 260937 & 80 & [email protected] & [email protected] & [email protected] & @xmath49.146 & [email protected] & [email protected] & [email protected] & m8 v + wisepa j152047.56 + 300208.5 & sdss j152047.63 + 300209.0 & 76 & [email protected] & [email protected] & [email protected] & @xmath48.405 & [email protected] & [email protected] & [email protected] & m8.5 v + wisepa j152424.72 + 292525.2 & 2masp j1524248 + 292535 & 77 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m7.5 v + wisepa j153946.68 + 275951.0 & cti 153915.5 + 280214 & 62 & [email protected] & [email protected] & @xmath412.229 & @xmath49.307 & [email protected] & @xmath417.395 & [email protected] & m4 : v + wisepa j153946.73 + 280222.3 & cti 153915.6 + 280445 & 62 & [email protected] & [email protected] & @xmath412.351 & @xmath49.298 & [email protected] & [email protected] & [email protected] & m6.5 v + wisepa j154016.16 + 280010.2 & cti 153945.1 + 280233 & 62 & [email protected] & [email protected] & [email protected] & @xmath49.146 & [email protected] & [email protected] & [email protected] & m1.5 v + wisepa j154019.25 + 280058.3 & cti 153948.1 + 280322 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.954 & [email protected] & [email protected] & [email protected] & m2 v + wisepa j154302.42 + 280139.5 & cti 154231.3 + 280401 & 62 & [email protected] & [email protected] & @xmath412.092 & @xmath49.047 & [email protected] & [email protected] & [email protected] & m5.5 v + wisepa j183436.70 + 400724.3 & lp 229 - 17 & 71 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath47.104 & m4 v + wisepa j184110.13 + 244715.5 & gj 1230 b & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j185848.03 + 280147.8 & cti 185818.3 + 280045 & 62 & [email protected] & [email protected] & [email protected] & @xmath48.873 & [email protected] & [email protected] & [email protected] & m2.5 v + wisepa j190705.22 + 205313.4 & gl 745 a & 61 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m3 v + wisepa j190712.86 + 205233.9 & gl 745 b & 61 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m2 v + wisepa j191246.79 + 280346.1 & cti 191216.9 + 280228 & 62 & [email protected] & [email protected] & @xmath412.568 & @xmath49.035 & [email protected] & [email protected] & [email protected] & m4 v + wisepa j192033.38 - 073945.4 & gl 754.1 b & 68 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m3 v + wisepa j193238.07 + 003439.3 & gl 761.2 & 72 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m0.5 v + wisepa j194932.81 - 624921.4 & lhs 3487 & 70 & [email protected] & [email protected] & [email protected] & @xmath48.996 & [email protected] & [email protected] & [email protected] & m0 v + wisepa j201916.77 - 562123.2 & lhs 3538 & 70 & [email protected] & [email protected] & [email protected] & @xmath48.456 & [email protected] & [email protected] & [email protected] & m4 v + wisepc j202948.79 + 094121.5 & gl 791.2ab & 70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & m4.5 v + wisepa j211832.15 - 450557.1 & hb 2115 - 4518 & 73 & [email protected] & [email protected] & [email protected] & @xmath48.837 & [email protected] & [email protected] & [email protected] & m8.5 v + wisepc j212310.80 + 641415.3 & lp 75 - 10 & 72 & [email protected] & [email protected] & [email protected] & @xmath49.220 & [email protected] & [email protected] & [email protected] & m2.5 v + wisepc j212726.25 - 421519.7 & hb 2124 - 4228 & 73 & [email protected] & [email protected] & [email protected] & @xmath48.602 & [email protected] & [email protected] & [email protected] & m7.5 v + wisepc j213008.67 - 444632.2 & hb 2126 - 4459 & 73 & [email protected] & [email protected] & @xmath412.137 & @xmath48.797 & [email protected] & [email protected] & [email protected] & m8.5 v + cccrccccccc wisepc [email protected] & 1 & @xmath276.3 & [email protected] & [email protected] & @xmath412.235 & [email protected] & [email protected] & @xmath112.308 & 13 + wisepc [email protected] & 1 & @xmath277.7 & [email protected] & [email protected] & [email protected] & @xmath48.763 & [email protected] & [email protected] & 12 + wisepc j004928.48 + 044100.1 & 1 & @xmath258.1 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 14 + wisepc j010637.07 + 151852.8 & 1 & @xmath247.3 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 13 + wisepa j012333.21 + 414203.9 & 1 & @xmath220.7 & [email protected] & [email protected] & @xmath412.339 & @xmath48.839 & [email protected] & @xmath112.509 & 13 + wisepc [email protected] & 1 & @xmath263.7 & [email protected] & [email protected] & [email protected] & @xmath48.957 & [email protected] & [email protected] & 14 + wisepc [email protected] & 1,5 & @xmath244.4 & [email protected] & [email protected] & @xmath412.579 & @xmath49.521 & [email protected] & @xmath112.005 & 24 + wisepa j015010.86 + 382724.3 & 1 & @xmath222.9 & [email protected] & [email protected] & [email protected] & @xmath48.779 & [email protected] & [email protected] & 10 + wisepa j020625.26 + 264023.6 & 1 & @xmath233.2 & [email protected] & [email protected] & [email protected] & @xmath49.144 & [email protected] & [email protected] & 11 + wisepa j022105.94 + 384202.9 & 1 & @xmath220.9 & [email protected] & [email protected] & @xmath412.087 & @xmath48.685 & [email protected] & @xmath112.534 & 12 + wisepc [email protected] & 1 & @xmath269.6 & [email protected] & [email protected] & [email protected] & @xmath49.544 & [email protected] & [email protected] & 15 + wisepa [email protected] & 1 & @xmath256.2 & [email protected] & [email protected] & @xmath412.285 & @xmath48.998 & [email protected] & @xmath112.258 & 13 + wisepa j025409.45 + 022359.1 & 1 & @xmath248.2 & [email protected] & [email protected] & [email protected] & @xmath49.067 & [email protected] & [email protected] & 11 + wisepa j030533.54 + 395434.4 & 1 & @xmath216.0 & [email protected] & [email protected] & @xmath412.243 & @xmath49.169 & [email protected] & @xmath112.400 & 13 + wisepa j030724.57 + 290447.6 & 1 & @xmath225.0 & [email protected] & [email protected] & @xmath412.339 & @xmath48.977 & [email protected] & @xmath112.543 & 11 + wisepa j031325.96 + 780744.2 & 1 & 17.2 & [email protected] & [email protected] & [email protected] & @xmath48.760 & [email protected] & [email protected] & 16 + wisepc [email protected] & 1 & @xmath247.9 & [email protected] & [email protected] & [email protected] & @xmath49.287 & [email protected] & [email protected] & 17 + wisepc [email protected] & 1 & @xmath247.6 & [email protected] & [email protected] & [email protected] & @xmath49.746 & [email protected] & [email protected] & 20 + wisepa j041022.71 + 150248.5 & 1 & @xmath225.9 & @xmath418.101 & [email protected] & [email protected] & @xmath48.923 & @xmath43.911 & [email protected] & 12 + wisepa j041054.48 + 141131.6 & 1 & @xmath226.3 & [email protected] & [email protected] & @xmath412.001 & @xmath49.069 & [email protected] & @xmath113.020 & 12 + wisepa [email protected] & 1 & @xmath235.4 & [email protected] & [email protected] & [email protected] & @xmath48.950 & [email protected] & [email protected] & 19 + wisepa j045853.89 + 643452.9 & 2 & 13.3 & [email protected] & [email protected] & [email protected] & @xmath49.260 & [email protected] & [email protected] & 15 + wisepa [email protected] & 1 & @xmath230.2 & [email protected] & [email protected] & @xmath412.500 & @xmath48.845 & [email protected] & @xmath111.473 & 16 + wisepa j051317.28 + 060814.7 & 1 & @xmath218.5 & [email protected] & [email protected] & @xmath412.374 & @xmath49.118 & [email protected] & @xmath111.460 & 11 + wisepa j052536.33 + 673952.3 & 1 & 17.3 & [email protected] & [email protected] & @xmath412.727 & @xmath49.128 & [email protected] & @xmath112.172 & 16 + wisepa [email protected] & 1 & @xmath230.8 & [email protected] & [email protected] & [email protected] & @xmath49.312 & [email protected] & [email protected] & 18 + wisepa [email protected] & 1 & @xmath220.6 & [email protected] & [email protected] & @xmath412.743 & @xmath49.202 & [email protected] & @xmath112.021 & 14 + wisepa [email protected] & 1 & @xmath222.4 & [email protected] & [email protected] & @xmath412.071 & @xmath49.068 & [email protected] & @xmath111.836 & 15 + wisepa [email protected] & 1 & @xmath210.7 & [email protected] & [email protected] & [email protected] & @xmath48.818 & [email protected] & [email protected] & 13 + wisepa [email protected] & 1 & @xmath221.2 & [email protected] & [email protected] & @xmath412.468 & @xmath48.954 & [email protected] & @xmath111.576 & 18 + wisepa [email protected] & 1 & @xmath226.4 & [email protected] & [email protected] & [email protected] & @xmath49.720 & [email protected] & [email protected] & 44 + wisepa j061407.49 + 391236.4 & 1 & 10.1 & [email protected] & [email protected] & @xmath411.968 & @xmath48.960 & [email protected] & @xmath111.665 & 13 + wisepa [email protected] & 1 & @xmath28.5 & [email protected] & [email protected] & [email protected] & @xmath48.661 & [email protected] & [email protected] & 16 + wisepa j062542.21 + 564625.5 & 1 & 19.0 & [email protected] & [email protected] & [email protected] & @xmath48.886 & [email protected] & [email protected] & 15 + wisepa [email protected] & 1 & @xmath210.4 & [email protected] & [email protected] & [email protected] & @xmath49.206 & [email protected] & [email protected] & 26 + wisepa j065609.60 + 420531.0 & 1 & 18.6 & [email protected] & [email protected] & [email protected] & @xmath48.846 & [email protected] & [email protected] & 14 + wisepa j074457.15 + 562821.8 & 1 & 29.5 & [email protected] & [email protected] & [email protected] & @xmath49.091 & [email protected] & [email protected] & 15 + wisepa j075003.84 + 272544.8 & 1 & 24.2 & @xmath418.338 & [email protected] & @xmath412.658 & @xmath48.765 & @xmath43.855 & @xmath111.825 & 12 + wisepa [email protected] & 1 & @xmath222.9 & [email protected] & [email protected] & [email protected] & @xmath49.306 & [email protected] & [email protected] & 45 + wisepc [email protected] & 1 & @xmath29.9 & [email protected] & [email protected] & [email protected] & @xmath49.120 & [email protected] & [email protected] & 19 + wisepa [email protected] & 1 & 17.8 & [email protected] & [email protected] & [email protected] & @xmath48.926 & [email protected] & [email protected] & 13 + wisepa j082131.63 + 144319.3 & 1 & 26.4 & [email protected] & [email protected] & @xmath412.572 & @xmath48.893 & [email protected] & @xmath111.711 & 10 + wisepc [email protected] & 1 & 12.9 & [email protected] & [email protected] & @xmath412.755 & @xmath49.156 & [email protected] & @xmath112.269 & 15 + wisepa j085716.25 + 560407.6 & 1 & 39.6 & [email protected] & [email protected] & @xmath412.291 & @xmath49.270 & [email protected] & @xmath111.740 & 15 + wisepa j090649.36 + 473538.6 & 1 & 42.1 & [email protected] & [email protected] & [email protected] & @xmath48.746 & [email protected] & [email protected] & 12 + wisepc j092906.77 + 040957.9 & 1 & 36.6 & [email protected] & [email protected] & [email protected] & @xmath49.140 & [email protected] & [email protected] & 10 + wisepc j095259.29 + 195507.3 & 1 & 48.7 & [email protected] & [email protected] & @xmath412.468 & @xmath49.183 & [email protected] & @xmath111.917 & 11 + wisepc [email protected] & 1 & 26.2 & [email protected] & [email protected] & @xmath412.222 & @xmath48.771 & [email protected] & @xmath111.866 & 14 + wisepa j101905.63 + 652954.2 & 1 & 44.8 & [email protected] & [email protected] & [email protected] & @xmath49.147 & [email protected] & [email protected] & 18 + wisepc [email protected] & 1 & 17.6 & @xmath418.496 & [email protected] & @xmath412.746 & @xmath49.099 & @xmath43.981 & @xmath111.769 & 14 + wisepc j112254.73 + 255021.5 & 1 & 70.1 & [email protected] & [email protected] & @xmath412.317 & @xmath49.429 & [email protected] & @xmath111.648 & 9 + wisepc j115013.88 + 630240.7 & 1 & 52.7 & [email protected] & [email protected] & [email protected] & @xmath48.831 & [email protected] & [email protected] & 16 + wisepc j121756.91 + 162640.2 & 1 & 76.7 & [email protected] & [email protected] & [email protected] & @xmath48.710 & [email protected] & [email protected] & 11 + wisepc j131106.24 + 012252.4 & 1 & 63.8 & [email protected] & [email protected] & @xmath412.110 & @xmath49.287 & [email protected] & @xmath112.623 & 12 + wisepc j131141.91 + 362925.2 & 1 & 79.7 & [email protected] & [email protected] & [email protected] & @xmath48.830 & [email protected] & [email protected] & 15 + wisepc j132004.16 + 603426.2 & 1 & 56.2 & [email protected] & [email protected] & @xmath412.046 & @xmath49.087 & [email protected] & @xmath112.443 & 9 + wisepa [email protected] & 1,4 & 38.6 & [email protected] & [email protected] & @xmath412.017 & [email protected] & [email protected] & @xmath111.859 & 9 + wisepc j134806.99 + 660327.8 & 1 & 50.0 & [email protected] & [email protected] & [email protected] & @xmath49.760 & [email protected] & [email protected] & 18 + wisepc j140518.40 + 553421.4 & 1,5 & 58.5 & @xmath417.989 & [email protected] & [email protected] & @xmath49.115 & @xmath43.904 & [email protected] & 21 + wisepa [email protected] & 1 & 38.0 & [email protected] & [email protected] & [email protected] & @xmath48.799 & [email protected] & [email protected] & 12 + wisepc j145715.03 + 581510.2 & 1 & 51.9 & [email protected] & [email protected] & [email protected] & @xmath49.558 & [email protected] & [email protected] & 18 + wisepc j150649.97 + 702736.0 & 1 & 42.6 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 25 + wisepc j151906.64 + 700931.5 & 1 & 42.0 & [email protected] & [email protected] & [email protected] & @xmath410.043 & [email protected] & [email protected] & 29 + wisepa [email protected] & 1.5 & 25.2 & @xmath417.018 & [email protected] & [email protected] & @xmath49.064 & @xmath43.036 & [email protected] & 11 + wisepa [email protected] & 1 & 12.3 & @xmath416.881 & [email protected] & @xmath412.150 & @xmath48.744 & @xmath42.796 & @xmath111.935 & 13 + wisepa j161441.45 + 173936.7 & 1,4 & 42.3 & [email protected] & [email protected] & @xmath412.318 & @xmath49.151 & [email protected] & @xmath111.911 & 14 + wisepa j161705.75 + 180714.3 & 3 & 42.0 & [email protected] & [email protected] & [email protected] & @xmath49.255 & [email protected] & [email protected] & 16 + wisepa [email protected] & 1 & 26.8 & [email protected] & [email protected] & @xmath411.958 & @xmath48.760 & [email protected] & @xmath112.058 & 11 + wisepa j162725.64 + 325525.5 & 1,4 & 43.3 & [email protected] & [email protected] & [email protected] & @xmath49.197 & [email protected] & [email protected] & 18 + wisepa j164715.59 + 563208.2 & 1 & 39.3 & [email protected] & [email protected] & [email protected] & @xmath49.208 & [email protected] & [email protected] & 41 + wisepa j165311.05 + 444423.9 & 1,4 & 39.2 & [email protected] & [email protected] & [email protected] & @xmath49.517 & [email protected] & [email protected] & 23 + wisepa j171104.60 + 350036.8 & 1 & 34.8 & [email protected] & [email protected] & [email protected] & @xmath49.454 & [email protected] & [email protected] & 21 + wisepa j171717.02 + 612859.3 & 1 & 34.7 & [email protected] & [email protected] & [email protected] & @xmath49.142 & [email protected] & [email protected] & 36 + wisepa j172844.93 + 571643.6 & 1 & 33.6 &
[email protected] & [email protected] & [email protected] & @xmath49.899 & [email protected] & [email protected] & 52 + wisepa j173835.53 + 273258.9 & 1,5 & 27.2 & [email protected] & [email protected] & [email protected] & @xmath49.182 & [email protected] & [email protected] & 18 + wisepa j174124.26 + 255319.5 & 1,4 & 26.1 & [email protected] & [email protected] & [email protected] & @xmath48.580 & [email protected] & [email protected] & 17 + wisepa j180435.40 + 311706.1 & 1 & 23.0 & @xmath418.423 & [email protected] & [email protected] & @xmath49.391 & @xmath43.714 & [email protected] & 19 + wisepa j181210.85 + 272144.3 & 3 & 20.1 &
[email protected] & [email protected] & @xmath412.369 & @xmath49.284 & [email protected] & @xmath111.812 & 15 + wisepa j182831.08 + 265037.8 & 1,5 & 16.5 & @xmath418.452 & [email protected] & [email protected] & [email protected] & @xmath44.176 & [email protected] & 18 + wisepa j183058.57 + 454257.9 & 1 & 22.4 & [email protected] & [email protected] & @xmath412.833 & @xmath49.286 & [email protected] & @xmath111.261 & 42 + wisepa j184124.74 + 700038.0 & 1,4 & 26.2 & [email protected] & [email protected] & [email protected] & @xmath49.424 & [email protected] & [email protected] & 77 + wisepa j185215.78 + 353716.3 & 1 & 15.2 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 21 + wisepa j190624.75 + 450808.2 & 1 & 16.3 & [email protected] & [email protected] & [email protected] & @xmath49.480 & [email protected] & [email protected] & 30 + wisepa j195246.66 + 724000.8 & 1 & 21.4 & [email protected] & [email protected] & [email protected] & @xmath49.141 & [email protected] & [email protected] & 47 + wisepa [email protected] & 1 & @xmath227.9 & [email protected] & [email protected] & @xmath412.287 & @xmath48.789 & [email protected] & @xmath111.531 & 13 + wisepa [email protected] & 3 & @xmath231.8 & [email protected] & [email protected] & [email protected] & @xmath48.977 & [email protected] & [email protected] & 12 + wisepc j205628.90 + 145953.3 & 1,5 & @xmath219.1 & @xmath417.742 & [email protected] & [email protected] & @xmath48.646 & @xmath43.890 & [email protected] & 12 + wisepa [email protected] & 1 & @xmath238.0 & [email protected] & [email protected] & [email protected] & @xmath48.830 & [email protected] & [email protected] & 13 + wisepc j215751.38 + 265931.4 & 1 & @xmath221.6 & [email protected] & [email protected] & @xmath411.927 & @xmath48.441 & [email protected] & @xmath112.626 & 8 + wisepc [email protected] & 1 & @xmath254.1 & [email protected] & [email protected] & @xmath411.861 & @xmath49.124 & [email protected] & @xmath111.925 & 10 + wisepc j221354.69 + 091139.4 & 1 & @xmath237.3 & [email protected] & [email protected] & @xmath412.329 & @xmath49.155 & [email protected] & @xmath112.179 & 14 + wisepc j222623.05 + 044003.9 & 1 & @xmath242.6 & [email protected] & [email protected] & @xmath411.992 & @xmath49.075 & [email protected] & @xmath112.633 & 11 + wisepc [email protected] & 1 & @xmath251.9 & [email protected] & [email protected] & @xmath412.460 & @xmath48.854 & [email protected] & @xmath112.200 & 10 + wisepc j223937.55 + 161716.2 & 1 & @xmath236.0 & [email protected] & [email protected] & [email protected] & @xmath49.173 & [email protected] & [email protected] & 12 + wisepc [email protected] & 1 & @xmath264.4 & [email protected] & [email protected] & @xmath411.975 & @xmath49.165 & [email protected] & @xmath112.105 & 10 + wisepa [email protected] & 3 & @xmath235.5 & [email protected] & [email protected] & [email protected] & @xmath49.312 & [email protected] & [email protected] & 22 + wisepc [email protected] & 1 & @xmath267.3 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11 + wisepc [email protected] & 1 & @xmath267.4 & [email protected] & [email protected] & @xmath412.061 & @xmath49.251 & [email protected] & @xmath112.051 & 12 + wisepc [email protected] & 1 & @xmath271.3 & [email protected] & [email protected] & [email protected] & @xmath49.283 & [email protected] & [email protected] & 12 + wisepc [email protected] & 1 & @xmath264.3 & [email protected] & [email protected] & @xmath411.977 & @xmath49.477 & [email protected] & @xmath111.581 & 11 + wisepa [email protected] & 1 & @xmath242.0 & [email protected] & [email protected] & [email protected] & @xmath49.405 & [email protected] & [email protected] & 19 + wisepc j234446.25 + 103415.8 & 1 & @xmath248.9 & @xmath417.949 & [email protected] & @xmath412.429 & @xmath48.543 & @xmath43.054 & @xmath112.466 & 10 + wisepc [email protected] & 1 & @xmath267.7 & [email protected] & [email protected] & @xmath411.958 & @xmath49.525 & [email protected] & @xmath112.310 & 13 + wisepa [email protected] & 3 & @xmath243.0 & [email protected] & [email protected] & [email protected] & @xmath49.446 & [email protected] & [email protected] & 18 + lcllllllllll wise j0008@xmath21739 & 4 & & [email protected] & [email protected] & @xmath417.0 & & & & [email protected] & [email protected] & [email protected] + wise j0031@xmath23840 & 1 & & [email protected] & [email protected] & [email protected] & & & & extant & & + wise j0049 + 0441 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0106 + 1518 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 10 & [email protected] & & & & [email protected] & [email protected] & [email protected] & & & + wise j0123 + 4142 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0138@xmath20322 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0148@xmath27202 & 8 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0150 + 3827 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 5 & & [email protected] & [email protected] & & & & & & & + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0206 + 2640 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0221 + 3842 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0223@xmath22932 & 4 & & [email protected] & [email protected] & @xmath416.94 & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0226@xmath20211 & 5 & & [email protected] & @xmath419.10 & & & & & [email protected] & [email protected] & [email protected] + wise j0254 + 0223 & 1 & & [email protected] & [email protected] & @xmath416.006 & & & & [email protected] & [email protected] & [email protected] + & 7 & & & & & [email protected] & [email protected] & & & & + wise j0305 + 3954 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0307 + 2904 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 10 & & & & & & & [email protected] & & & + wise j0313 + 7807 & 6 & [email protected] & [email protected] & [email protected] & & & & & [email protected] & [email protected] & [email protected] + wise j0323@xmath26025 & 8 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0333@xmath25856 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0410 + 1502 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0410 + 1411 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 10 & & & & & & & [email protected] & & & + wise j0448@xmath21935 & 2 & & [email protected] & [email protected] & @xmath415.858 & & & & & & + & 2 & & [email protected] & [email protected] & @xmath416.556 & & & & & & + wise j0458 + 6434 & 6 & [email protected] & [email protected] & [email protected] & & & & & [email protected] & [email protected] & [email protected] + wise j0500@xmath21223 & 9 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0513 + 0608 & 1 & & [email protected] & @xmath416.890 & @xmath415.936 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0525 + 6739 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j0528@xmath23308 & 9 & & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + wise j0539@xmath21034 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0542@xmath21628 & 1 & & [email protected] & @xmath415.887 & @xmath415.934 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0611@xmath20410 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0612@xmath23036 & 2 & & [email protected] & @xmath415.808 & @xmath415.781 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0612@xmath24920 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j0614 + 3912 & 2 & & [email protected] & [email protected] & @xmath416.283 & & & & [email protected] & [email protected] & [email protected] + wise j0623@xmath20456 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0625 + 5646 & 1 & & [email protected] & [email protected] & @xmath415.771 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0627@xmath21114 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0656 + 4205 & 1 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0744 + 5628 & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0750 + 2725 & 4 & & @xmath418.57 & @xmath417.92 & @xmath416.91 & & & & [email protected] & [email protected] & [email protected] + & 3 & & 19.02 @xmath60.21 & @xmath419.49 & @xmath418.63 & & & & & & + & 7 & & & & & [email protected] & [email protected] & & & & + & 10 & & & & & [email protected] & & & & & + wise j0751@xmath27634 & 9 & & & & & [email protected] & @xmath419.02 & & [email protected] & [email protected] & [email protected] + wise j0759@xmath24904 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j0819@xmath20335 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0821 + 1443 & 1 & & [email protected] & [email protected] & @xmath417.089 & & & & & & + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0836@xmath21859 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j0857 + 5604 & 3 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j0906 + 4735 & 4 & & [email protected] & @xmath418.33 & @xmath417.36 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j0929 + 0409 & 4 & & [email protected] & @xmath418.26 & @xmath417.37 & & & & [email protected] & [email protected] & [email protected] + & 10 & & & & & & [email protected] & [email protected] & & & + wise j0952 + 1955 & 2 & & [email protected] & @xmath417.166 & @xmath416.447 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & @xmath417.337 & & & & & & + wise j1018@xmath22445 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j1019 + 6529 & 1 & & [email protected] & @xmath416.328 & @xmath417.016 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1042@xmath23842 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j1122 + 2550 & 1 & & [email protected] & @xmath416.062 & @xmath416.824 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1150 + 6302 & 4 & & [email protected] & @xmath418.01 & @xmath416.65 & & & & [email protected] & [email protected] & [email protected] + wise j1217 + 1626 & 4 & & @xmath418.52 & @xmath417.50 & @xmath416.64 & & & & [email protected] & [email protected] & [email protected] + & 7 & & & & & [email protected] & [email protected] & & & & + wise j1311 + 0122 & 7 & & & & & [email protected] & & & [email protected] & [email protected] & [email protected] + wise j1311 + 3629 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & @[email protected] + wise j1320 + 6034 & 2 & & [email protected] & @xmath416.806 & @xmath415.622 & & & & [email protected] & [email protected] & [email protected] + & 2 & & [email protected] & @xmath415.759 & @xmath416.921 & & & & & & + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1322@xmath22340 & 4 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j1348 + 6603 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1405 + 5534 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j1436@xmath21814 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j1457 + 5815 & 2 & & [email protected] & @xmath417.636 & @xmath417.320 & & & & [email protected] & [email protected] & [email protected] + & 2 & & [email protected] & @xmath416.725 & @xmath415.464 & & & & & & + & 2 & & [email protected] & [email protected] & @xmath416.704 & & & & & & + & 4 & & @xmath417.48 & @xmath416.42 & @xmath415.03 & & & & & & + wise j1506 + 7027 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1519 + 7009 & 4 & & @xmath417.58 & @xmath416.82 & & & & & [email protected] & [email protected] & [email protected] + & 7 & & & & & [email protected] & [email protected] & & & & + wise j1541@xmath22250 & 12 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j1612@xmath23420 & - & & & & & & & & [email protected] & [email protected] & [email protected] + wise j1614 + 1739 & 6 & [email protected] & & & & & & & [email protected] & [email protected] & [email protected] + & 9 & & & & & [email protected] & [email protected] & & & & + wise j1617 + 1807 & 6 & [email protected] & & & & & & & [email protected] & [email protected] & [email protected] + & 9 & & & & & [email protected] & [email protected] & & & & + wise j1622@xmath20959 & 1 & & [email protected] & @xmath416.615 & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1627 + 3255 & 1 & & [email protected] & [email protected] & @xmath417.362 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & & & & & & & + & 6 & [email protected] & [email protected] & & & & & & & & + wise j1647 + 5632 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1653 + 4444 & 6 & [email protected] & [email protected] & [email protected] & & & & & [email protected] & [email protected] & [email protected] + wise j1711 + 3500 & 4 & & [email protected] & @xmath418.12 & & & & & [email protected] & [email protected] & [email protected] + wise j1717 + 6129 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + & 4 & & @xmath418.93 & @xmath418.17 & @xmath416.75 & & & & & & + wise j1728 + 5716 & 6 & [email protected] & [email protected] & [email protected] & & & & & [email protected] & [email protected] & [email protected] + wise j1738 + 2732 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j1741 + 2553 & 1 & & [email protected] & [email protected] & @xmath416.785 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & [email protected] & & & & & & + & 6 & [email protected] & [email protected] & [email protected] & & & & & & & + wise j1804 + 3117 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + & 4 & & @xmath418.88 & @xmath418.24 & @xmath416.86 & & & & & & + wise j1812 + 2721 & 7 & & & & & [email protected] & [email protected] & & & & + & 4 & & @xmath418.58 & @xmath417.77 & @xmath417.03 & & & & & & + wise j1828 + 2650 & 11 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j1830 + 4542 & 2 & & @xmath418.752 & [email protected] & [email protected] & & & & & & + wise j1841 + 7000 & 2 & & [email protected] & [email protected] & @xmath415.626 & & & & [email protected] & [email protected] & [email protected] + & 7 & & & & & [email protected] & [email protected] & & & & + & 4 & & [email protected] & [email protected] & & & & & & & + wise j1852 + 3537 & 1 & & [email protected] & @xmath417.407 & @xmath417.076 & & & & [email protected] & [email protected] & [email protected] + & 4 & & @xmath416.70 & @xmath415.88 & & & & & & & + wise j1906 + 4508 & 1 & & [email protected] & [email protected] & @xmath416.107 & & & & & & + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j1952 + 7240 & 4 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j1959@xmath23338 & 1 & & [email protected] & @xmath416.077 & @xmath416.227 & & & & & & + & 9 & & & & & [email protected] & [email protected] & & & & + wise j2018@xmath27423 & - & & & & & & & & & & + wise j2056 + 1459 & 7 & & & & & [email protected] & @xmath419.5 & & [email protected] & [email protected] & [email protected] + & 4 & & @xmath417.6 & @xmath417.1 & & & & & & & + & 11 & & & & - & [email protected] & [email protected] & & & & + wise j2134@xmath27131 & 8 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j2157 + 2659 & 5 & & [email protected] & [email protected] & & & & & [email protected] & [email protected] & [email protected] + wise j2209@xmath22734 & 1 & & [email protected] & @xmath416.408 & @xmath416.903 & & & & [email protected] & [email protected] & [email protected] + wise j2213 + 0911 & 2 & & [email protected] & @xmath416.046 & @xmath416.092 & & & & [email protected] & [email protected] & [email protected] + & 2 & & [email protected] & @xmath416.071 & @xmath417.175 & & & & & & + & 4 & & [email protected] & [email protected] & @xmath416.2 & & & & & & + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j2226 + 0440 & 4 & & [email protected] & @xmath417.3 & & & & & [email protected] & [email protected] & [email protected] + wise j2237@xmath20614 & 2 & & [email protected] & @xmath416.009 & @xmath417.054 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & @xmath416.8 & & & & & & + wise j2239 + 1617 & 1 & & [email protected] & [email protected] & @xmath414.890 & & & & [email protected] & [email protected] & [email protected] + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j2255@xmath23118 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & @xmath417.7 & @xmath416.4 & & & & & & + wise j2313@xmath28037 & 2 & & [email protected] & @xmath416.192 & @xmath416.358 & & & & & & + wise j2319@xmath21844 & 2 & & [email protected] & @xmath417.225 & @xmath416.009 & & & & [email protected] & [email protected] & [email protected] + & 7 & & & & & [email protected] & [email protected] & & & & + wise j2325@xmath24105 & 9 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j2327@xmath22730 & 1 & & [email protected] & [email protected] & [email protected] & & & & [email protected] & [email protected] & [email protected] + wise j2340@xmath20745 & 1 & & [email protected] & [email protected] & @xmath416.271 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & @xmath416.3 & & & & & & + wise j2343@xmath27418 & 1 & & [email protected] & @xmath416.142 & [email protected] & & & & & & + wise j2344 + 1034 & 7 & & & & & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] + wise j2348@xmath21028 & 1 & & [email protected] & @xmath416.441 & @xmath415.977 & & & & [email protected] & [email protected] & [email protected] + & 4 & & [email protected] & [email protected] & @xmath416.443 & & & & & & + & 3 & & [email protected] & [email protected] & [email protected] & & & & & & + wise j2359@xmath27335 & - & & [email protected] & [email protected] & @xmath415.159 & & & & [email protected] & [email protected] & [email protected] + lcccccc wise j0008@xmath21739 & & t6 & keck / nirspec - j & 1200 & hd 13936 & 2010 dec 18 + wise j0031@xmath23840 & & l2 pec ( blue ) & irtf / spex & 720 & hd 4065 & 2010 nov 17 + wise j0049 + 0441 & & l9 & irtf / spex & 1200 & hd 7215 & 2010 sep 14 + wise j0106 + 1518 & & m8 pec & irtf / spex & 960 & hd 7215 & 2010 sep 14 + wise j0123 + 4142 & & t7 & keck / nirspec - j & 600 & hd 13936 & 2010 dec 19 + wise j0138@xmath20322 & & t3 & irtf / spex & 960 & hd 13936 & 2010 aug 17 + wise j0148@xmath27202 & & t9.5 & magellan / fire & 960 & hd 1881 & 2010 sep 18 + wise j0150 + 3827 & & t0 & irtf / spex & 1920 & hd 19600 & 2010 sep 12 + wise j0206 + 2640 & & l9 pec ( red ) & irtf / spex & 1680 & hd 19600 & 2010 sep 12 + wise j0221 + 3842 & & t6.5 & palomar / tspec & 1200 & hd 19600 & 2011 jan 22 + wise j0223@xmath22932 & & t7 & keck / nirspec - h & 1200 & hd 25792 & 2010 jul 18 + & & t7.5 & irtf / spex & 1560 & hd 20423 & 2010 aug 17 + wise j0226@xmath20211 & & t7 & keck / nirspec - h & 1200 & hd 21875 & 2010 jul 19 + wise j0254 + 0223 & & t8 & irtf / spex & 960 & hd 21379 & 2010 jul 14 + wise j0305 + 3954 & & t6 & palomar / tspec & 1200 & hd 19600 & 2011 jan 22 + wise j0307 + 2904 & & t6.5 & irtf / spex & 720 & hd 19600 & 2010 dec 17 + wise j0313 + 7807 & & t8.5 & palomar / tspec & 1800 & hd 210501 & 2010 nov 16 + wise j0323@xmath26025 & & t8.5 & magellan / fire & 1268 & hd 62762 & 2010 dec 24 + wise j0333@xmath25856 & & t3 & soar / osiris & 1080 & hd 16636 & 2010 dec 28 + wise j0410 + 1502 & & y0 & magellan / fire & 600 & hd 18620 & 2010 nov 18 + wise j0410 + 1411 & & t6 & irtf / spex & 600 & hd 25175 & 2010 dec 18 + wise j0448@xmath21935 & & t5 pec & irtf / spex & 240 & hd 29433 & 2010 aug 17 + wise j0458 + 6434 & & t8.5 & & & & + wise j0500@xmath21223 & & t8 & irtf / spex & 960 & hd 31743 & 2010 oct 29 + wise j0513 + 0608 & & t6.5 & palomar / tspec & 1200 & hd 40686 & 2010 nov 15 + wise j0525 + 6739 & & t6 pec & irtf / spex & 3240 & hd 38831 & 2010 nov 17 + wise j0528@xmath23308 & & t7 pec & palomar / tspec & 2400 & hd 38056 & 2010 nov 16 + wise j0539@xmath21034 & & t5.5 & palomar / tspec & 2400 & hd 38386 & 2011 jan 22 + wise j0542@xmath21628 & & t6.5 & palomar / tspec & 2400 & hd 21127 & 2010 nov 16 + wise j0611@xmath20410 & & t0 & irtf / spex & 360 & hd 56525 & 2010 dec 18 + wise j0612@xmath23036 & & t6 & palomar / tspec & 1800 & hd 32855 & 2010 nov 16 + wise j0612@xmath24920 & & t3.5 & soar / osiris & 1080 & hd 48169 & 2010 dec 28 + wise j0614 + 3912 & & t6 & palomar / tspec & 1800 & hd 56385 & 2010 nov 15 + wise j0623@xmath20456 & & t8 & irtf / spex & 900 & hd 45137 & 2010 dec 17 + wise j0625 + 5646 & & t6 & palomar / tspec & 1200 & hd 38831 & 2011 jan 22 + wise j0627@xmath21114 & & t6 & irtf / spex & 720 & hd 56525 & 2010 dec 18 + wise j0656 + 4205 & & t3 & irtf / spex & 240 & hd 56525 & 2010 dec 18 + wise j0744 + 5628 & & t8 & palomar / tspec & 2400 & hd 45105 & 2011 jan 22 + wise j0750 + 2725 & & t8.5 & keck / nirspec - j & 1800 & hd 71906 & 2010 dec 24 + & & t8.5 & keck / nirspec - h & 3000 & hd 71906 & 2010 dec 24/25 + wise j0751@xmath27634 & & t9 & magellan / fire & 760 & hd 98671 & 2010 apr 03 + wise j0759@xmath24904 & & t8 & magellan / fire & 760 & cpd-44 2691 & 2011 mar 25 + wise j0819@xmath20335 & & t4 & irtf / spex & 360 & hd 65241 & 2010 dec 18 + wise j0821 + 1443 & & t5.5 & irtf / spex & 720 & hd 65241 & 2010 dec 18 + wise j0836@xmath21859 & & t8 pec & magellan / fire & 1522 & hd 75159 & 2011 mar 27 + wise j0857 + 5604 & & t8 & palomar / tspec & 2400 & hd 45105 & 2011 jan 22 + wise j0906 + 4735 & & t8 & palomar / tspec & 3600 & hd 82191 & 2010 jun 04 + wise j0929 + 0409 & & t6.5 & palomar / tspec & 1200 & hd 121409 & 2011 jan 22 + wise j0952 + 1955 & & t6 & irtf / spex & 1800 & hd 89239 & 2011 mar 09 + wise j1018@xmath22445 & & t8 & magellan / fire & 507 & hd 90738 & 2011 mar 27 + wise j1019 + 6529 & & t6 & irtf / spex & 1680 & hd 143187 & 2010 may 27 + & & t6 & palomar / tspec & 2400 & sao 15429 & 2010 may 30 + & t@xmath447 & & keck / lris & 3000 & hd 151506 & 2010 jun 18 + wise j1042@xmath23842 & & t8.5 & magellan / fire & 1014 & hd 90738 & 2011 mar 27 + wise j1122 + 2550 & & t6 & irtf / spex & 1440 & hd 99966 & 2010 jul 14 + wise j1150 + 6302 & & t8 & palomar / tspec & 2400 & hd 121409 & 2011 jan 22 + wise j1217 + 1626 & & t9 & palomar / tspec & 2400 & hd 19600 & 2011 jan 22 + wise j1311 + 0122 & & t9 : & keck / nirspec - h & 900 & hd 71906 & 2010 dec 25 + wise j1311 + 3629 & & l5 pec ( blue ) & irtf / spex & 1200 & hd 109615 & 2011 jan 26 + wise j1320 + 6034 & & t6.5 & irtf / spex & 720 & hd 118214 & 2010 jul 02 + wise j1322@xmath22340 & & t8 & irtf / spex & 1920 & hd 114345 & 2010 may 24 + wise j1348 + 6603 & & l9 & irtf / spex & 1200 & hd 71906 & 2011 jan 26 + wise j1405 + 5534 & & y0 ( pec ? ) & hst / wfc3 & 2212 & & 2011 mar 14 + wise j1436@xmath21814 & & t8 pec & magellan / fire & 507 & hd 130755 & 2011 mar 27 + wise j1457 + 5815 & & t7 & irtf / spex & 960 & hd 143187 & 2010 jul 14 + & t@xmath448 & & keck / lris & 3600 & hd 238493 & 2010 jul 17 + wise j1506 + 7027 & & t6 & keck / nirspec - h & 960 & hd 25792 & 2010 oct 20 + & & t6 & palomar / tspec & 960 & hd 145454 & 2011 jan 22 + wise j1519 + 7009 & & t8 & palomar / tspec & 2400 & hd 145454 & 2010 jun 04 + wise j1541@xmath22250 & & y0 & magellan / fire & 1522 & hd 130755 & 2011 mar 27 + wise j1612@xmath23420 & & t6.5 & keck / nirspec - h & 1200 & hd 152384 & 2010 jul 19 + wise j1614 + 1739 & & t9 & magellan / fire & 600 & hd 98671 & 2010 apr 03 + wise j1617 + 1807 & & t8 & & & & + & t@xmath448 & & keck / lris & 3600 & bd+18 3241 & 2010 jul 17 + wise j1622@xmath20959 & & t6 & irtf / spex & 1200 & hd 148968 & 2010 apr 23 + wise j1627 + 3255 & & t6 & keck / nirspec - h & 2400 & hd 145647 & 2010 feb 24 + & & t6 & irtf / spex & 4080 & hd 145647 & 2010 feb 28 + wise j1647 + 5632 & & l9 pec ( red ) & irtf / spex & 960 & hd 179933 & 2010 aug 17 + wise j1653 + 4444 & & t8 & irtf / spex & 1440 & hd 143187 & 2010 apr 21 + & t@xmath448 & & keck / lris & 2400 & hd 159518 & 2010 jul 17 + wise j1711 + 3500 & & t8 & irtf / spex & 1440 & hd 165029 & 2010 jul 13 + wise j1717 + 6129 & & t8 & keck / nirspec - j & 1200 & hd 179933 & 2010 jul 19 + & & t8 & keck / nirspec - h & 1200 & hd 179933 & 2010 jul 19 + wise j1728 + 5716 & & t6 & irtf / spex & 3120 & hd 143187 & 2010 apr 21 + wise j1738 + 2732 & & y0 & hst / wfc3 & 2012 & & 2011 may 12 + wise j1741 + 2553 & & t9 & magellan / fire & 1800 & hd 98671 & 2010 apr 03 + & t@xmath449 & & keck / lris & 3900 & hd 335701 & 2010 jun 18 + wise j1804 + 3117 & & t9.5 : & keck / nirspec - h & 1200 & hd 171623 & 2010 jul 19 + wise j1812 + 2721 & & t8.5 : & & & & + wise j1828 + 2650 & & @xmath4y0 & hst / wfc3 & 2012 & & 2011 may 09 + wise j1830 + 4542 & & l9 & irtf / spex & 1920 & hd 178207 & 2010 sep 12 + wise j1841 + 7000 & & t5 & palomar / tspec & 2400 & hd 179933 & 2010 jun 04 + & t@xmath445 & & keck / lris & 3600 & bd+70 1059 & 2010 jul 17 + wise j1852 + 3537 & & t7 & irtf / spex & 1680 & hd 174567 & 2010 may 25 + wise j1906 + 4508 & & t6 & irtf / spex & 1200 & hd 174567 & 2010 nov 17 + wise j1952 + 7240 & & t4 & palomar / tspec & 1200 & hd 18187 & 2010 nov 16 + wise j1959@xmath23338 & & t8 & palomar / tspec & 3600 & hd 194272 & 2010 jun 04 + wise j2018@xmath27423 & & t7 & & + wise j2056 + 1459 & & y0 & keck / nirspec - j & 2400 & hd 198070 & 2010 oct 21 + & & y0 & keck / nirspec - h & 1800 & hd 198069 & 2010 nov 22 + wise j2134@xmath27137 & & t9 pec & magellan / fire & 600 & hd 223296 & 2010 sep 18 + wise j2157 + 2659 & & t7 & palomar / tspec & 1200 & hd 208108 & 2010 sep 29 + wise j2209@xmath22734 & & t7 & keck / nirspec - h & 600 & hd 212643 & 2010 jul 18 + wise j2213 + 0911 & & t7 & irtf / spex & 2400 & hd 210501 & 2010 aug 04 + wise j2226 + 0440 & & t8.5 & keck / nirspec - j & 1200 & hd 190807 & 2010 jul 18 + & & t8 & keck / nirspec - h & 1200 & hd 190807 & 2010 jul 18 + & & t8 & irtf / spex & 1920 & hd 210501 & 2010 aug 04 + wise j2237@xmath20614 & & t5 & irtf / spex & 1200 & hd 219833 & 2010 jul 14 + wise j2239 + 1617 & & t3 & irtf / spex & 1200 & hd 210501 & 2010 aug 17 + wise j2255@xmath23118 & & t8 & keck / nirspec - h & 1200 & hd 202941 & 2010 jul 19 + & & t8 & irtf / spex & 2400 & hd 202025 & 2010 sep 10 + wise j2313@xmath28037 & & t8 & & & & + wise j2319@xmath21844 & & t7.5 & irtf / spex & 2880 & hd 222332 & 2010 aug 17 + wise j2325@xmath24105 & & t9 pec & magellan / fire & 960 & hd 221805 & 2010 sep 18 + wise j2327@xmath22730 & & l9 & irtf / spex & 1200 & hd 219290 & 2010 sep 12 + wise j2340@xmath20745 & & t7 & irtf / spex & 960 & hd 219833 & 2010 jul 14 + & t@xmath447 & & keck / lris & 2400 & hd 1920 & 2010 jul 17 + wise j2343@xmath27418 & & t6 & magellan / fire & 634 & hd 223296 & 2010 dec 24 + wise j2344 + 1034 & & t9 & keck / nirspec - j & 1200 & hd 198070 & 2010 oct 21 + wise j2348@xmath21028 & & t7 & irtf / spex & 960 & hd 219833 & 2010 jul 14 + wise j2359@xmath27335 & & t5.5 & & & & + lccccc a ) objects not known to be binary + ugps [email protected] & 246@xmath633 & 2 & t9 & [email protected] & [email protected] + denis j081730.0@xmath2615520 & 203@xmath613 & 3 & t6 & [email protected] & [email protected] + denis - p j0255@xmath24700 & [email protected] & 4 & l8 & [email protected] & [email protected] + 2massi j0415195@xmath2093506 & [email protected] & 5 & t8 & [email protected] & [email protected] + gliese 570d & [email protected] & 6 & t7.5 & [email protected] & [email protected] + 2massi j0937347 + 293142 & [email protected] & 7 & t6 & [email protected] & [email protected] + 2massw j1507476@xmath2162738 & [email protected] & 8 & l5 & [email protected] & [email protected] + 2massw j0036159 + 182110 & [email protected] & 8 & l3.5 & [email protected] & [email protected] + 2massi j0727182 + 171001 & [email protected] & 5 & t7 & [email protected] & [email protected] + cfbds [email protected] & [email protected] & 1 & t8.5 & [email protected] & [email protected] + 2mass j05591914@xmath21404488 & [email protected] & 8 & t4.5 & [email protected] & [email protected] + ulas j133553.45 + 113005.2 & [email protected] & 1 & t8.5 & [email protected] & [email protected] + 2mass j12373919 + 6526148 & [email protected] & 5 & t6.5 & [email protected] & [email protected] + 2massi j0825196 + 211552 & [email protected] & 8 & l7.5 & [email protected] & [email protected] + 2massi j0243137@xmath2245329 & [email protected] & 5 & t6 & [email protected] & [email protected] + sdssp j162414.37 + 002915.6 & [email protected] & 9 & t6 & [email protected] & [email protected] + 2massi j1217110@xmath2031113 & [email protected] & 9 & t7.5 & [email protected] & [email protected] + 2massi j1546291@xmath2332511 & [email protected] & 9 & t5.5 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 8 & t2 & [email protected] & [email protected] + ulas [email protected] & [email protected] & 1 & t8.5 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 5 & l5 & [email protected] & [email protected] + 2massw j1439284 + 192915 & [email protected] & 8 & l1 & [email protected] & [email protected] + 2massi j2356547@xmath2155310 & [email protected] & 5 & t5.5 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 9 & t6.5 & [email protected] & [email protected] + 2massw j1632291 + 190441 & [email protected] & 8 & l8 & [email protected] & [email protected] + denis - p j1058.7@xmath21548 & [email protected] & 8 & l3 & [email protected] & [email protected] + 2massw j1658037 + 702701 & [email protected] & 8 & l1 & [email protected] & [email protected] + gliese 584c & [email protected] & 6 & l8 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 5 & l8 & [email protected] & [email protected] + sdss j015141.69 + 124429.6 & [email protected] & 5 & t1 & [email protected] & [email protected] + sdssp j144600.60 + 002452.0 & [email protected] & 5 & l6 & [email protected] & [email protected] + sdssp j175032.96 + 175903.9 & [email protected] & 5 & t3.5 & [email protected] & [email protected] + 2massw j0030300@xmath2145033 & [email protected] & 5 & l7 & [email protected] & [email protected] + 2massi j1711457 + 223204 & [email protected] & 5 & l6.5 & [email protected] & [email protected] +
gd 165b & [email protected] & 10 & l4 & [email protected] & [email protected] + 2massw j1328550 + 211449 & [email protected] & 8 & l5 & [email protected] & [email protected] + 2massw j0326137 + 295015 & [email protected] & 8 & l3.5 & [email protected] & [email protected] + sdssp j003259.36 + 141036.6 & [email protected] & 5 & l8 & [email protected] & [email protected] + sdss j020742.48 + 000056.2 & [email protected] & 1 & t4.5 & [email protected] & @xmath416.396 + ulas [email protected] & [email protected] & 1 & t5.5 & [email protected] & [email protected] + hd 89744b & [email protected] & 6 & l0 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 5 & l0 & [email protected] & [email protected] + + b ) known binaries / multiples : + @xmath118 ind bab & [email protected] & 6 & t1 & [email protected] & [email protected] + 2massi j0746425 + 200032 & [email protected] & 8 & l0.5 & [email protected] & [email protected] + 2mass j12255432@xmath22739466 & [email protected] & 9 & t6 & [email protected] & [email protected] + 2massi j1534498@xmath2295227 & [email protected] & 9 & t5.5 & [email protected] & [email protected] + sdssp [email protected] & [email protected] & 5 & t0 & [email protected] & [email protected] + hn peg b & [email protected] & 6 & t2.5 & [email protected] & [email protected] + kelu-1ab & [email protected] & 8 & l2 & [email protected] & [email protected] + denis - p j0205.4@xmath21159 & [email protected] & 8 & l7 & [email protected] & [email protected] + denis - p j1228.2@xmath21547 & [email protected] & 8 & l5 & [email protected] & [email protected] + gliese 417bc & [email protected] & 6 & l4.5 & [email protected] & [email protected] + 2massw j1728114 + 394859 & [email protected] & 5 & l7 & [email protected] & [email protected] + g 124 - 62b & [email protected] & 4 & l0.5 & [email protected] & [email protected] + sdss [email protected] & [email protected] & 9 & t3 & [email protected] & [email protected] + 2massw j2101154 + 175658 & [email protected] & 5 & l7.5 & [email protected] & [email protected] + ccllccclrrrr wise 0008 - 1739 & 22.0 & 2.2073557 & -17.6563363 & 0.179 & 0.185 & wise epoch 1 & 55363.39 & [email protected] & [email protected] + & & 2.2073061 & -17.6565639 & 0.279 & 0.293 & wise epoch 2 & 55542.36 + & & 2.2073379 & -17.6564659 & 0.485 & 0.309 & spitzer & 55571 + + wise 0031 - 3840 & 22.0 & 7.830238 & -38.676575 & 0.06 & 0.07 & 2mass psc & 51391.3054 & [email protected] & [email protected] & [email protected] & [email protected] + & & 7.8323524 & -38.6766992 & 0.045 & 0.044 & wise epoch 1 & 55357.83 + & & 7.8324281 & -38.6767498 & 0.046 & 0.044 & wise epoch 2 & 55535.75 + + wise 0049 + 0441 & 19.2 & 12.367782 & 4.682658 & 0.07 & 0.12 & 2mass psc & 51768.4283 & [email protected] & [email protected] & [email protected] & [email protected] + & & 12.3686648 & 4.6833142 & 0.053 & 0.053 & wise epoch 1 & 55382.38 + & & 12.3686625 & 4.6832837 & 0.066 & 0.062 & wise epoch 2 & 55560.23 + & & 12.3686292 & 4.6833140 & 0.206 & 0.260 & spitzer & 55591 + + wise 0106 + 1518 & 43.8 & 16.655734 & 15.315453 & 0.06 & 0.06 & 2mass psc & 50696.4341 & [email protected] & [email protected] & [email protected] & [email protected] + & & 16.65554497 & 15.31529756 & 0.058 & 0.045 & sdss dr7 & 51464.3 + & & 16.6544791 & 15.3146837 & 0.048 & 0.047 & wise epoch 1 & 55298.38 + & & 16.6543939 & 15.3146032 & 0.053 & 0.052 & wise epoch 2 & 55567.30 + & & 16.6543859 & 15.3146712 & 0.157 & 0.351 & spitzer & 55633 + + wise 0123 + 4142 & 21.9 & 20.8883990 & 41.7010723 & 0.301 & 0.309 & wise epoch 1 & 55217.30 & [email protected] & [email protected] & [email protected] & [email protected] + & & 20.8885524 & 41.7010216 & 0.230 & 0.237 & wise epoch 2 & 55408.13 + & & 20.8886559 & 41.7012568 & 0.282 & 0.292 & wise epoch 3 & 55583.18 + & & 20.8887898 & 41.7012899 & 0.222 & 0.182 & spitzer & 55646 + + wise 0138 - 0322 & 17.7 & 24.652009 & -3.371703 & 0.18 & 0.17 & 2mass psc & 51079.2142 & [email protected] & [email protected] & [email protected] & [email protected] + & & 24.6524238 & -3.3725727 & 0.065 & 0.066 & wise epoch 1 & 55298.71 + & & 24.6524451 & -3.3726575 & 0.080 & 0.081 & wise epoch 2 & 55570.43 + & & 24.6523636 & -3.3726830 & 0.441 & 0.318 & spitzer & 55624 + + wise 0148 - 7202 & 12.1 & 27.0301973 & -72.0496466 & 0.179 & 0.166 & wise epoch 1 & 55324.30 & [email protected] & [email protected] + & & 27.0306266 & -72.0497359 & 0.158 & 0.149 & wise epoch 2 & 55504.92 + & & 27.0306647 & -72.0496657 & 0.207 & 0.212 & spitzer epoch 1 & 55458 + & & 27.0308488 & -72.0496856 & 0.188 & 0.177 & spitzer epoch 2 & 55580 + + wise 0150 + 3827 & 18.8 & 27.541553 & 38.457203 & 0.08 & 0.09 & 2mass psc & 50771.1947 & [email protected] & [email protected] & [email protected] & [email protected] + & & 27.5452547 & 38.4567195 & 0.060 & 0.061 & wise epoch 1 & 55221.20 + & & 27.5454455 & 38.4567394 & 0.055 & 0.056 & wise epoch 2 & 55410.43 + & & 27.5455516 & 38.4566765 & 0.064 & 0.063 & wise epoch 3 & 55586.55 + & & 27.5457112 & 38.4567635 & 0.258 & 0.172 & spitzer & 55646 + + wise 0206 + 2640 & 17.9 & 31.603883 & 26.673265 & 0.09 & 0.08 & 2mass psc & 51506.2524 & [email protected] & [email protected] & [email protected] & [email protected] + & & 31.6052461 & 26.6732349 & 0.056 & 0.055 & wise epoch 1 & 55220.34 + & & 31.6053520 & 26.6732556 & 0.056 & 0.056 & wise epoch 2 & 55408.11 + & & 31.6053749 & 26.6732325 & 0.056 & 0.053 & wise epoch 3 & 55585.36 + & & 31.6053859 & 26.6732165 & 0.240 & 0.237 & spitzer & 55646 + + wise 0221 + 3842 & 21.4 & 35.2747550 & 38.7009027 & 0.221 & 0.236 & wise epoch 1 & 55227.70 & [email protected] & [email protected] + & & 35.2750666 & 38.7008127 & 0.225 & 0.244 & wise epoch 2 & 55417.64 + & & 35.2749576 & 38.7007578 & 0.279 & 0.308 & wise epoch 3 & 55592.18 + & & 35.2747902 & 38.7010427 & 0.239 & 0.184 & spitzer & 55651
+ + wise 0223 - 2932 & 13.3 & 35.8432105 & -29.5492102 & 0.114 & 0.120 & wise epoch 1 & 55298.25 & [email protected] & [email protected] + & & 35.8432970 & -29.5493427 & 0.155 & 0.161 & wise epoch 2 & 55570.15 + & & 35.8433178 & -29.5492604 & 0.420 & 0.158 & spitzer epoch 1 & 55457 + & & 35.8434107 & -29.5492880 & 0.543 & 0.153 & spitzer epoch 2 & 55586 + + wise 0226 - 0211 & 19.3 & 36.5998759 & -2.1952331 & 0.214 & 0.228 & wise epoch 1 & 55217.31 & [email protected] & [email protected] + & & 36.5999766 & -2.1951802 & 0.253 & 0.274 & wise epoch 2 & 55406.74 + & & 36.5999836 & -2.1951423 & 0.240 & 0.252 & wise epoch 3 & 55580.86 + & & 36.5998761 & -2.1955429 & 0.216 & 0.297 & spitzer epoch 1 & 55457 + & & 36.5997878 & -2.1950815 & 0.609 & 0.498 & spitzer epoch 2 & 55620 + + wise 0254 + 0223 & 6.9 & 43.532860 & 2.398985 & 0.28 & 0.39 & 2mass psc & 51808.4049 & [email protected] & [email protected] & [email protected] & [email protected] + & & 43.5393853 & 2.3995999 & 0.078 & 0.079 & wise epoch 1 & 55222.06 + & & 43.5398386 & 2.3996419 & 0.077 & 0.076 & wise epoch 2 & 55412.34 + & & 43.5400634 & 2.3996809 & 0.089 & 0.084 & wise epoch 3 & 55588.27 + & & 43.5399385 & 2.3996560 & 0.270 & 0.275 & spitzer epoch 1 & 55457 + & & 43.5401933 & 2.3996251 & 0.138 & 0.218 & spitzer epoch 2 & 55623 + + wise 0305 + 3954 & 23.1 & 46.3897630 & 39.9095717 & 0.214 & 0.231 & wise epoch 1 & 55236.15 & [email protected] & [email protected] + & & 46.3898024 & 39.9096596 & 0.238 & 0.252 & wise epoch 2 & 55426.37 + & & 46.3898707 & 39.9096158 & 0.235 & 0.224 & spitzer & 55666 + + wise 0307 + 2904 & 24.0 & 46.8524209 & 29.0798600 & 0.288 & 0.313 & wise epoch 1 & 55233.50 & [email protected] & [email protected] + & & 46.8523979 & 29.0798509 & 0.339 & 0.363 & wise epoch 2 & 55423.59 + & & 46.8524465 & 29.0798329 & 0.229 & 0.213 & spitzer & 55666 + + wise 0313 + 7807 & 8.1 & 48.3581817 & 78.1289629 & 0.078 & 0.081 & wise epoch 1 & 55256.85 & [email protected] & [email protected]
+ & & 48.3584055 & 78.1289989 & 0.071 & 0.071 & wise epoch 2 & 55448.07 + & & 48.3587833 & 78.1290043 & 0.182 & 0.131 & spitzer epoch 1 & 55551 + & & 48.3585620 & 78.1289581 & 0.233 & 0.249 & spitzer epoch 2 & 55675 + + wise 0323 - 6025 & 15.5 & 50.9063954 & -60.4318079 & 0.171 & 0.163 & wise epoch 1 & 55369.41 & [email protected] & [email protected] + & & 50.9065228 & -60.4318044 & 0.158 & 0.157 & wise epoch 2 & 55549.31 + & & 50.9066237 & -60.4318259 & 0.288 & 0.170 & spitzer epoch 1 & 55457 + & & 50.9065579 & -60.4319125 & 0.214 & 0.238 & spitzer epoch 2 & 55580 + + wise 0333 - 5856 & 16.8 & 53.456318 & -58.936623 & 0.08 & 0.07 & 2mass psc & 51484.2013 & [email protected] & [email protected] & [email protected] & [email protected] + & & 53.4555971 & -58.9383889 & 0.053 & 0.052 & wise epoch 1 & 55377.88 + & & 53.4555377 & -58.9385002 & 0.054 & 0.053 & wise epoch 2 & 55554.87 + & & 53.4556255 & -58.9385088 & 0.290 & 0.121 & spitzer & 55550 + + wise 0410 + 1502 & 9.0 & 62.5946668 & 15.0468130 & 0.187 & 0.202 & wise epoch 1 & 55246.29 & [email protected] & [email protected] & [email protected] & [email protected] + & & 62.5949435 & 15.0465208 & 0.164 & 0.178 & wise epoch 2 & 55434.04 & & & & + & & 62.5949913 & 15.0464124 & 0.215 & 0.126 & spitzer epoch 1 & 55490 + & & 62.5950070 & 15.0461156 & 0.201 & 0.189 & spitzer epoch 2 & 55666 + + wise 0410 + 1411 & 27.5 & 62.7269959 & 14.1920334 & 0.285 & 0.319 & wise epoch 1 & 55246.20 & [email protected] & [email protected] + & & 62.7269105 & 14.1920574 & 0.316 & 0.358 & wise epoch 2 & 55434.04 + & & 62.7269833 & 14.1919793 & 0.209 & 0.214 & spitzer & 55666 + + wise 0448 - 1935 & 21.4 & 72.219501 & -19.598743 & 0.24 & 0.22 & 2mass reject & 51838.3186 & [email protected] & [email protected] & [email protected] & [email protected] + & & 72.219098 & -19.599138 & 0.34 & 0.32 & 2mass reject & 51146.2679 + & & 72.2220593 & -19.5968744 & 0.127 & 0.136 & wise epoch 1 & 55248.32 + & & 72.2221445 & -19.5966820 & 0.139 & 0.147 & wise epoch 2 & 55439.14 + + wise 0458 + 6434 & 7.3 & 74.7245596 & 64.5812869 & 0.084 & 0.085 & wise epoch 1 & 55262.15 & [email protected] & [email protected]
+ & & 74.7247231 & 64.5813115 & 0.082 & 0.085 & wise epoch 2 & 55453.42 + & & 74.7248445 & 64.5812701 & 0.140 & 0.164 & spitzer epoch 1 & 55499 + & & 74.7247249 & 64.5813482 & 0.153 & 0.167 & spitzer epoch 2 & 55670 + + wise 0500 - 1223 & 12.4 & 75.0126995 & -12.3953585 & 0.139 & 0.149 & wise epoch 1 & 55251.81 & [email protected] & [email protected] + & & 75.0126275 & -12.3952592 & 0.120 & 0.130 & wise epoch 2 & 55444.82 + & & 75.0126456 & -12.3952697 & 0.162 & 0.195 & spitzer & 55487 + + wise 0513 + 0608 & 14.8 & 78.321983 & 6.138618 & 0.13 & 0.14 & 2mass psc & 51539.2078 & [email protected] & [email protected] & [email protected] & [email protected] + & & 78.3219766 & 6.1374197 & 0.114 & 0.122 & wise epoch 1 & 55259.05 + & & 78.3220117 & 6.1373744 & 0.110 & 0.118 & wise epoch 2 & 55450.18 + & & 78.3220245 & 6.1372354 & 0.175 & 0.187 & spitzer & 55666 + + wise 0525 + 6739 & 25.6 & 81.4012734 & 67.6644117 & 0.211 & 0.236 & wise epoch 1 & 55266.05 & [email protected] & [email protected]
+ & & 81.4011703 & 67.6638300 & 0.169 & 0.189 & wise epoch 2 & 55457.33 + & & 81.4014639 & 67.6644615 & 0.165 & 0.142 & spitzer epoch 1 & 55499 + & & 81.4012915 & 67.6644678 & 0.168 & 0.173 & spitzer epoch 2 & 55670 + + wise 0528 - 3308 & 19.0 & 82.1854678 & -33.1399839 & 0.171 & 0.186 & wise epoch 1 & 55259.30 & [email protected] & [email protected] + & & 82.1855415 & -33.1399867 & 0.149 & 0.161 & wise epoch 2 & 55450.45 + & & 82.1854770 & -33.1400365 & 0.218 & 0.276 & spitzer & 55505 + + wise 0539 - 1034 & 26.0 & 84.9875161 & -10.5768926 & 0.235 & 0.261 & wise epoch 1 & 55264.73 & [email protected] & [email protected] + & & 84.9874342 & -10.5768627 & 0.241 & 0.268 & wise epoch 2 & 55456.07 + & & 84.9875164 & -10.5771147 & 0.159 & 0.149 & spitzer & 55669 + + wise 0542 - 1628 & 15.3 & 85.631043 & -16.475651 & 0.16 & 0.15 & 2mass psc & 51196.1551 & [email protected] & [email protected] & [email protected] & [email protected] + & & 85.6302683 & -16.4747751 & 0.127 & 0.134 & wise epoch 1 & 55265.19 + & & 85.6303505 & -16.4747098 & 0.128 & 0.136 & wise epoch 2 & 55456.47 + & & 85.6303011 & -16.4747332 & 0.184 & 0.204 & spitzer epoch 1 & 55548 + & & 85.6302423 & -16.4746955 & 0.183 & 0.203 & spitzer epoch 2 & 55671 + + wise 0611 - 0410 & 17.7 & 92.896153 & -4.172446 & 0.06 & 0.07 & 2mass psc & 51095.3441 & [email protected] & [email protected] & [email protected] & [email protected] + & & 92.8963777 & -4.1733529 & 0.055 & 0.054 & wise epoch 1 & 55275.79 + & & 92.8964286 & -4.1733818 & 0.052 & 0.053 & wise epoch 2 & 55467.18 + & & 92.8964138 & -4.1734137 & 0.165 & 0.167 & spitzer & 55669 + + wise 0612 - 3036 & 17.0 & 93.058322 & -30.602640 & 0.23 & 0.22 & 2mass reject & 51470.3267 & [email protected] & [email protected] & [email protected] & [email protected] + & & 93.0578204 & -30.6033715 & 0.113 & 0.119 & wise epoch 1 & 55276.30 + & & 93.0578414 & -30.6033901 & 0.121 & 0.133 & wise epoch 2 & 55467.73 + & & 93.0578281 & -30.6034545 & 0.217 & 0.180 & spitzer epoch 1 & 55553 + & & 93.0577910 & -30.6034781 & 0.243 & 0.167 & spitzer epoch 2 & 55675 + + wise 0612 - 4920 & 29.2 & 93.0362290 & -49.3400502 & 0.065 & 0.069 & wise epoch 1 & 55277.56 & [email protected] & [email protected] + & & 93.0362551 & -49.3399726 & 0.063 & 0.066 & wise epoch 2 & 55469.23 + & & 93.0362060 & -49.3400468 & 0.212 & 0.223 & spitzer & 55592 + + wise 0614 + 3912 & 14.5 & 93.531962 & 39.211395 & 0.24 & 0.23 & 2mass reject & 51519.4327 & [email protected] & [email protected] & [email protected] & [email protected] + & & 93.5312108 & 39.2099957 & 0.112 & 0.119 & wise epoch 1 & 55275.58 + & & 93.5312727 & 39.2099085 & 0.111 & 0.118 & wise epoch 2 & 55466.94 + & & 93.5312418 & 39.2098538 & 0.150 & 0.183 & spitzer epoch 1 & 55548 + & & 93.5311786 & 39.2098092 & 0.161 & 0.208 & spitzer epoch 2 & 55670 + + wise 0623 - 0456 & 11.3 & 95.7913950 & -4.9401879 & 0.120 & 0.130 & wise epoch 1 & 55277.49 & [email protected] & [email protected] & [email protected] & [email protected] + & & 95.7912970 & -4.9400740 & 0.123 & 0.131 & wise epoch 2 & 55468.97 + & & 95.7911133 & -4.9400708 & 0.183 & 0.142 & spitzer & 55675 + + wise 0625 + 5646 & 19.9 & 96.426281 & 56.774120 & 0.29 & 0.24 & 2mass psc & 51448.4972 & [email protected] & [email protected] & [email protected] & [email protected] + & & 96.4258596 & 56.7737642 & 0.144 & 0.159 & wise epoch 1 & 55276.24 + & & 96.4261210 & 56.7737633 & 0.166 & 0.182 & wise epoch 2 & 55467.53 + & & 96.4260484 & 56.7737822 & 0.184 & 0.209 & spitzer epoch 1 & 55551 + & & 96.4260262 & 56.7737448 & 0.182 & 0.144 & spitzer epoch 2 & 55675 + + wise 0627 - 1114 & 12.0 & 96.833665 & -11.240032 & 0.10 & 0.08 & 2mass psc & 51182.2328 & [email protected] & [email protected] & [email protected] & [email protected] + & & 96.8336050 & -11.2411311 & 0.068 & 0.069 & wise epoch 1 & 55278.95 + & & 96.8336478 & -11.2411066 & 0.064 & 0.066 & wise epoch 2 & 55470.56 + & & 96.8336096 & -11.2411756 & 0.170 & 0.179 & spitzer & 55671 + + wise 0656 + 4205 & 16.6 & 104.038623 & 42.091709 & 0.09 & 0.07 & 2mass psc & 51122.4585 & [email protected] & [email protected] & [email protected] & [email protected] + & & 104.0399904 & 42.0922078 & 0.063 & 0.067 & wise epoch 1 & 55280.98 + & & 104.0401314 & 42.0922872 & 0.062 & 0.066 & wise epoch 2 & 55474.39 + + wise 0744 + 5628 & 15.7 & 116.2384910 & 56.4725580 & 0.169 & 0.182 & wise epoch 1 & 55288.54 & [email protected] & -0.122@xmath6 0.528 + & & 116.2386617 & 56.4725324 & 0.193 & 0.209 & wise epoch 2 & 55480.15 + + wise 0750 + 2725 & 14.1 & 117.5157736 & 27.4293219 & 0.221 & 0.245 & wise epoch 1 & 55295.29 & [email protected] & [email protected] + & & 117.5157144 & 27.4290782 & 0.220 & 0.250 & wise epoch 2 & 55486.76 + & & 117.5156237 & 27.4291158 & 0.190 & 0.164 & spitzer & 55540 + + wise 0751 - 7634 & 13.0 & 117.7865794 & -76.5804610 & 0.120 & 0.116 & wise epoch 1 & 55247.30 & [email protected] & [email protected] + & & 117.7864634 & -76.5804277 & 0.117 & 0.111 & wise epoch 2 & 55427.39 + & & 117.7868452 & -76.5804157 & 0.187 & 0.178 & spitzer epoch 1 & 55465 + & & 117.7870092 & -76.5804892 & 0.206 & 0.192 & spitzer epoch 2 & 55665 + + wise 0759 - 4904 & 11.5 & 119.9457963 & -49.0816846 & 0.117 & 0.123 & wise epoch 1 & 55327.64 & [email protected] & [email protected] + & & 119.9458493 & -49.0817034 & 0.102 & 0.104 & wise epoch 2 & 55518.12 + & & 119.9457564 & -49.0816408 & 0.205 & 0.207 & spitzer & 55563 + + wise 0819 - 0335 & 14.1 & 124.992527 & -3.590735 & 0.07 & 0.07 & 2mass psc & 51158.3227 & [email protected] & [email protected] & [email protected] & [email protected] + & & 124.9918916 & -3.5912481 & 0.065 & 0.068 & wise epoch 1 & 55309.51 + & & 124.9919050 & -3.5912565 & 0.061 & 0.062 & wise epoch 2 & 55500.87 + & & 124.9918753 & -3.5912920 & 0.177 & 0.134 & spitzer & 55582 + + wise 0821 + 1443 & 20.9 & 125.382219 & 14.723092 & 0.27 & 0.27 & 2mass psc & 50757.4660 & [email protected] & [email protected] & [email protected] & [email protected] + & & 125.3817907 & 14.7220453 & 0.182 & 0.202 & wise epoch 1 & 55304.68 + & & 125.3819009 & 14.7220247 & 0.189 & 0.212 & wise epoch 2 & 55495.78 + + wise 0836 - 1859 & 20.1 & 129.1713270 & -18.9964194 & 0.300 & 0.335 & wise epoch 1 & 55319.57 & [email protected] & [email protected] + & & 129.1715540 & -18.9964381 & 0.327 & 0.366 & wise epoch 2 & 55510.58 + & & 129.1715373 & -18.9963358 & 0.188 & 0.132 & spitzer & 55563 + + wise 0857 + 5604 & 13.0 & 134.3176838 & 56.0687846 & 0.139 & 0.154 & wise epoch 1 & 55299.85 & [email protected] & [email protected] + & & 134.3175161 & 56.0687720 & 0.132 & 0.143 & wise epoch 2 & 55494.98 + & & 134.3173712 & 56.0687644 & 0.231 & 0.202 & spitzer & 55548 + + wise 0906 + 4735 & 16.3 & 136.7056493 & 47.5940957 & 0.240 & 0.262 & wise epoch 1 & 55306.35 & [email protected] & [email protected] + & & 136.7055665 & 47.5939434 & 0.233 & 0.247 & wise epoch 2 & 55497.55 + & & 136.7055675 & 47.5940058 & 0.349 & 0.337 & spitzer & 55554 + + wise 0929 + 0409 & 16.8 & 142.2782394 & 4.1660918 & 0.158 & 0.171 & wise epoch 1 & 55324.92 & [email protected] & [email protected] + & & 142.2782409 & 4.1659982 & 0.196 & 0.213 & wise epoch 2 & 55515.74 + & & 142.2783300 & 4.1659286 & 0.234 & 0.188 & spitzer & 55563 + + wise 0952 + 1955 & 20.5 & 148.247322 & 19.919182 & 0.21 & 0.21 & 2mass reject & 50836.4072 & [email protected] & [email protected] + & & 148.2470846 & 19.9189556 & 0.210 & 0.227 & wise epoch 1 & 55325.26 + & & 148.2471023 & 19.9190241 & 0.204 & 0.221 & wise epoch 2 & 55515.94 + & & 148.2470632 & 19.9189825 & 0.293 & 0.249 & spitzer & 55563 + + wise 1018 - 2445 & 13.1 & 154.5335346 & -24.7660780 & 0.148 & 0.158 & wise epoch 1 & 55349.86 & [email protected] & [email protected] + & & 154.5335261 & -24.7661840 & 0.168 & 0.175 & wise epoch 2 & 55539.49 + & & 154.5335257 & -24.7661780 & 0.241 & 0.204 & spitzer & 55588 + + wise 1019 + 6529 & 16.6 & 154.773976 & 65.497971 & 0.20 & 0.19 & 2mass psc & 51496.4619 & [email protected] & [email protected] & [email protected] & [email protected] + & & 154.7734412 & 65.4984032 & 0.115 & 0.119 & wise epoch 1 & 55307.00 + & & 154.7734862 & 65.4983792 & 0.109 & 0.112 & wise epoch 2 & 55498.47 + & & 154.7735020 & 65.4983754 & 0.186 & 0.222 & spitzer & 55540 + + wise 1042 - 3842 & 14.5 & 160.6884450 & -38.7106301 & 0.200 & 0.213 & wise epoch 1 & 55364.71 & [email protected] & [email protected] + & & 160.6887008 & -38.7105165 & 0.229 & 0.236 & wise epoch 2 & 55554.90 + & & 160.6885579 & -38.7107245 & 0.169 & 0.216 & spitzer & 55649 + + wise 1122 + 2550 & 16.9 & 170.731266 & 25.840298 & 0.17 & 0.20 & 2mass psc & 51324.1702 & [email protected] & [email protected] & [email protected] & [email protected] + & & 170.72963579 & 25.83988882 & 0.118 & 0.098 & sdss dr7 & 53387.5 + & & 170.7280297 & 25.8394998 & 0.128 & 0.137 & wise epoch 1 & 55344.11 + & & 170.7279454 & 25.8394037 & 0.156 & 0.165 & wise epoch 2 & 55534.13 + & & 170.7277844 & 25.8393187 & 0.231 & 0.311 & spitzer & 55581 + + wise 1150 + 6302 & 9.6 & 177.5577364 & 63.0448752 & 0.092 & 0.093 & wise epoch 1 & 55322.30 & [email protected] & [email protected] + & & 177.5579005 & 63.0447128 & 0.106 & 0.104 & wise epoch 2 & 55513.36 + & & 177.5578860 & 63.0447494 & 0.336 & 0.284 & spitzer & 55554 + + wise 1217 + 1626 & 6.7 & 184.4871110 & 16.4446764 & 0.104 & 0.108 & wise epoch 1 & 55362.98 & [email protected] & [email protected] & [email protected] & [email protected] + & & 184.4873517 & 16.4444386 & 0.117 & 0.121 & wise epoch 2 & 55549.01 + & & 184.4873182 & 16.4443118 & 0.274 & 0.287 & spitzer & 55633 + + wise 1311 + 0122 & 14.4 & 197.7760031 & 1.3817110 & 0.218 & 0.240 & wise epoch 1 & 55293.42 & [email protected] & [email protected] + & & 197.7758861 & 1.3815585 & 0.315 & 0.322 & wise epoch 2 & 55568.13 + & & 197.7760406 & 1.3814727 & 0.249 & 0.215 & spitzer & 55649 + + wise 1311 + 3629 & 26.4 & 197.926132 & 36.489887 & 0.08 & 0.06 & 2mass psc & 50920.2820 & [email protected] & [email protected] & [email protected] & [email protected] + & & 197.9246218 & 36.4901989 & 0.051 & 0.052 & wise epoch 1 & 55363.90 + & & 197.9246197 & 36.4902302 & 0.056 & 0.058 & wise epoch 2 & 55554.11 + & & 197.9245666 & 36.4901695 & 0.168 & 0.169 & spitzer & 55623 + + wise 1320 + 6034 & 20.0 & 200.020494 & 60.573997 & 0.29 & 0.26 & 2mass reject & 51685.2443 & [email protected] & [email protected] & [email protected] & [email protected] + & & 200.021008 & 60.573971 & 0.19 & 0.24 & 2mass reject & 51237.5119 + & & 200.0173925 & 60.5739840 & 0.168 & 0.171 & wise epoch 1 & 55339.40 + & & 200.0174182 & 60.5738862 & 0.142 & 0.141 & wise epoch 2 & 55529.70 + & & 200.0173022 & 60.5738479 & 0.151 & 0.101 & spitzer & 55567 + + wise 1322 - 2340 & 11.9 & 200.6401733 & -23.6713770 & 0.128 & 0.134 & wise epoch 1 & 55303.81 & [email protected] & [email protected] + & & 200.6401337 & -23.6713074 & 0.172 & 0.182 & wise epoch 2 & 55579.11 + & & 200.6401028 & -23.6713742 & 0.184 & 0.220 & spitzer & 55664 + + wise 1348 + 6603 & 27.9 & 207.028236 & 66.058762 & 0.30 & 0.29 & 2mass psc & 51274.83561 & [email protected] & [email protected] & [email protected] & [email protected] + & & 207.0291320 & 66.0579122 & 0.060 & 0.061 & wise epoch 1 & 55334.60 + & & 207.0293434 & 66.0578476 & 0.068 & 0.070 & wise epoch 2 & 55525.60 + & & 207.0291262 & 66.0579238 & 0.281 & 0.269 & spitzer & 55562 + + wise 1405 + 5534 & 8.6 & 211.3266880 & 55.5726255 & 0.132 & 0.135 & wise epoch 1 & 55355.75 & [email protected] & [email protected] & [email protected] & [email protected] + & & 211.3261242 & 55.5725614 & 0.149 & 0.150 & wise epoch 2 & 55545.11 + & & 211.3259057 & 55.5725984 & 0.147 & 0.108 & spitzer & 55583 + + wise 1436 - 1814 & 16.8 & 219.0091232 & -18.2394159 & 0.254 & 0.282 & wise epoch 1 & 55230.10 & [email protected] & [email protected] + & & 219.0092676 & -18.2394081 & 0.242 & 0.264 & wise epoch 2 & 55408.01 + & & 219.0091654 & -18.2394558 & 0.187 & 0.257 & spitzer & 55671 + + wise 1457 + 5815 & 17.7 & 224.315332 & 58.253193 & 0.29 & 0.34 & 2mass reject & 51716.2127 & [email protected] & [email protected] & [email protected] & [email protected] + & & 224.315344 & 58.253124 & 0.33 & 0.35 & 2mass reject & 51699.2167 + & & 224.315402 & 58.252907 & 0.31 & 0.36 & 2mass reject & 51592.4008 + & & 224.3127027 & 58.2528351 & 0.131 & 0.132 & wise epoch 1 & 55363.89 + & & 224.3124989 & 58.2527587 & 0.150 & 0.147 & wise epoch 2 & 55554.11 + & & 224.3124228 & 58.2527638 & 0.380 & 0.177 & spitzer & 55586 + + wise 1506 + 7027 & 4.9 & 226.7082158 & 70.4600383 & 0.038 & 0.042 & wise epoch 1 & 55327.90 & [email protected] & [email protected] & [email protected] & [email protected] + & & 226.7078583 & 70.4600650 & 0.041 & 0.043 & wise epoch 2 & 55518.52 + & & 226.7077094 & 70.4600814 & 0.320 & 0.286 & spitzer epoch 1 & 55553 + & & 226.7075505 & 70.4602557 & 0.069 & 0.216 & spitzer epoch 2 & 55675 + + wise 1519 + 7009 & 12.9 & 229.7776851 & 70.1587724 & 0.100 & 0.097 & wise epoch 1 & 55332.47 & [email protected] & [email protected]
+ & & 229.7778701 & 70.1585885 & 0.111 & 0.104 & wise epoch 2 & 55523.36 + & & 229.7775452 & 70.1586150 & 0.474 & 0.267 & spitzer epoch 1 & 55540 + & & 229.7778838 & 70.1585951 & 0.327 & 0.145 & spitzer epoch 2 & 55666 + + wise 1541 - 2250 & 8.2 & 235.4651859 & -22.8404607 & 0.189 & 0.211 & wise epoch 1 & 55244.67 & [email protected] & [email protected] + & & 235.4648903 & -22.8402853 & 0.167 & 0.186 & wise epoch 2 & 55424.68 + & & 235.4649058 & -22.8405027 & 0.193 & 0.198 & spitzer & 55665 + + wise 1612 - 3420 & 16.2 & 243.0664600 & -34.3412500 & 0.149 & 0.165 & wise epoch 1 & 55252.92 & [email protected] & [email protected] + & & 243.0663368 & -34.3412819 & 0.148 & 0.164 & wise epoch 2 & 55434.47 + & & 243.0663573 & -34.3412643 & 0.183 & 0.184 & spitzer epoch 1 & 55458 + & & 243.0663181 & -34.3414016 & 0.190 & 0.191 & spitzer epoch 2 & 55675 + + wise 1614 + 1739 & 11.3 & 243.6727451 & 17.6598663 & 0.170 & 0.186 & wise epoch 1 & 55243.86 & [email protected] & [email protected]
+ & & 243.6728541 & 17.6598448 & 0.142 & 0.154 & wise epoch 2 & 55423.69 + & & 243.6728699 & 17.6598646 & 0.190 & 0.195 & spitzer epoch 1 & 55459 + & & 243.6730063 & 17.6598073 & 0.179 & 0.185 & spitzer epoch 2 & 55669 + + wise 1617 + 1807 & 12.7 & 244.2739640 & 18.1206173 & 0.134 & 0.148 & wise epoch 1 & 55244.39 & [email protected] & [email protected]
+ & & 244.2739204 & 18.1206136 & 0.115 & 0.120 & wise epoch 2 & 55424.22 + & & 244.2739060 & 18.1205622 & 0.197 & 0.142 & spitzer epoch 1 & 55459 + & & 244.2739485 & 18.1205905 & 0.264 & 0.140 & spitzer epoch 2 & 55669 + + wise 1622 - 0959 & 17.3 & 245.536974 & -9.993136 & 0.26 & 0.29 & 2mass psc & 51276.4110 & [email protected] & [email protected] + & & 245.5372651 & -9.9929576 & 0.151 & 0.161 & wise epoch 1 & 55251.08 + & & 245.5372808 & -9.9929933 & 0.154 & 0.166 & wise epoch 2 & 55432.49 + & & 245.5372360 & -9.9929079 & 0.200 & 0.206 & spitzer epoch 1 & 55458 + & & 245.5372316 & -9.9929665 & 0.131 & 0.141 & spitzer epoch 2 & 55669 + + wise 1627 + 3255 & 14.3 & 246.857210 & 32.924686 & 0.30 & 0.27 & 2mass psc & 50908.4198 & [email protected] & [email protected] & [email protected] & [email protected] + & & 246.85704879 & 32.92431447 & 0.126 & 0.126 & sdss dr7 & 52402.2 + & & 246.85701573 & 32.92417129 & 0.114 & 0.114 & sdss dr7 & 52788.2 + & & 246.8568583 & 32.9235349 & 0.096 & 0.104 & wise epoch 1 & 55242.60 + & & 246.8568878 & 32.9234726 & 0.080 & 0.086 & wise epoch 2 & 55422.86 + & & 246.8568177 & 32.9234703 & 0.241 & 0.225 & spitzer epoch 1 & 55459 + & & 246.8568639 & 32.9234009 & 0.352 & 0.183 & spitzer epoch 2 & 55668 + + wise 1647 + 5632 & 20.2 & 251.815856 & 56.534924 & 0.10 & 0.08 & 2mass psc & 51308.4127 & [email protected] & [email protected] & [email protected] & [email protected] + & & 251.8149734 & 56.5356318 & 0.048 & 0.048 & wise epoch 1 & 55226.51 + & & 251.8148281 & 56.5356748 & 0.045 & 0.046 & wise epoch 2 & 55406.42 + & & 251.8149394 & 56.5356781 & 0.054 & 0.055 & wise epoch 3 & 55592.29 + & & 251.8148690 & 56.5357524 & 0.112 & 0.173 & spitzer & 55621 + + wise 1653 + 4444 & 11.2 & 253.2960324 & 44.7396889 & 0.098 & 0.104 & wise epoch 1 & 55245.44 & [email protected] & [email protected]
+ & & 253.2959638 & 44.7396403 & 0.076 & 0.079 & wise epoch 2 & 55425.41 + & & 253.2958575 & 44.7396392 & 0.256 & 0.209 & spitzer epoch 1 & 55459 + & & 253.2959160 & 44.7387647 & 0.340 & 1.476 & spitzer epoch 2 & 55667 + + wise 1711 + 3500 & 17.0 & 257.7691694 & 35.0102353 & 0.168 & 0.189 & wise epoch 1 & 55259.07 & [email protected] & [email protected]
+ & & 257.7691388 & 35.0102031 & 0.162 & 0.177 & wise epoch 2 & 55437.93 + & & 257.7691150 & 35.0102595 & 0.209 & 0.194 & spitzer epoch 1 & 55458 + & & 257.7691399 & 35.0101974 & 0.232 & 0.115 & spitzer epoch 2 & 55669 + + wise 1717 + 6129 & 19.6 & 259.3209221 & 61.4831687 & 0.219 & 0.226 & wise epoch 1 & 55228.46 & [email protected] & [email protected]
+ & & 259.3209456 & 61.4831246 & 0.177 & 0.185 & wise epoch 2 & 55407.56 + & & 259.3213856 & 61.4830062 & 0.273 & 0.291 & wise epoch 3 & 55591.70 + & & 259.3206293 & 61.4832354 & 0.140 & 0.252 & spitzer epoch 1 & 55458 + & & 259.3207534 & 61.4831919 & 0.204 & 0.291 & spitzer epoch 2 & 55586 + + wise 1728 + 5716 & 25.1 & 262.1872683 & 57.2785228 & 0.152 & 0.165 & wise epoch 1 & 55250.60 & [email protected] & [email protected]
+ & & 262.1873665 & 57.2785841 & 0.181 & 0.199 & wise epoch 2 & 55431.66 + & & 262.1872315 & 57.2785926 & 0.214 & 0.237 & spitzer epoch 1 & 55458 + & & 262.1872521 & 57.2785553 & 0.225 & 0.287 & spitzer epoch 2 & 55621 + + wise 1738 + 2732 & 10.5 & 264.6480391 & 27.5497726 & 0.172 & 0.193 & wise epoch 1 & 55269.06 & [email protected] & [email protected] + & & 264.6480755 & 27.5496620 & 0.165 & 0.182 & wise epoch 2 & 55451.22 + & & 264.6480850 & 27.5496349 & 0.140 & 0.179 & spitzer & 55458 + + wise 1741 + 2553 & 4.7 & 265.352586 & 25.892878 & 0.26 & 0.29 & 2mass psc & 51645.3888 & [email protected] & [email protected] & [email protected] & [email protected] + & & 265.3517935 & 25.89105715 & 0.109 & 0.098 & sdss dr7 & 53264.1 + & & 265.3510834 & 25.8887959 & 0.060 & 0.060 & wise epoch 1 & 55270.05 + & & 265.3509014 & 25.8886009 & 0.057 & 0.056 & wise epoch 2 & 55451.80 + & & 265.3508910 & 25.8886468 & 0.189 & 0.190 & spitzer & 55458 + + wise 1804 + 3117 & 13.0 & 271.1475485 & 31.2849339 & 0.165 & 0.185 & wise epoch 1 & 55277.66 & [email protected] & [email protected] + & & 271.1474413 & 31.2849933 & 0.176 & 0.191 & wise epoch 2 & 55459.94 + & & 271.1473240 & 31.2851309 & 0.189 & 0.253 & spitzer & 55465 + + wise 1812 + 2721 & 12.2 & 273.0452354 & 27.3623147 & 0.154 & 0.167 & wise epoch 1 & 55280.04 & [email protected] & [email protected] + & & 273.0451342 & 27.3622949 & 0.132 & 0.143 & wise epoch 2 & 55462.27 + + wise 1828 + 2650&@xmath119.4 & 277.1295227 & 26.8438370 & 0.173 & 0.192 & wise epoch 1 & 55288.38 & [email protected] & [email protected]
+ & & 277.1295880 & 26.8438296 & 0.167 & 0.182 & wise epoch 2 & 55467.61 + & & 277.1296054 & 26.8438464 & 0.195 & 0.164 & spitzer epoch 1 & 55387 + & & 277.1296928 & 26.8438604 & 0.151 & 0.236 & spitzer epoch 2 & 55534 + + wise 1830 + 4542 & 32.8 & 277.743778 & 45.715626 & 0.26 & 0.26 & 2mass reject & 50969.2963 & [email protected] & [email protected] & [email protected] & [email protected] + & & 277.7440428 & 45.7159649 & 0.062 & 0.064 & wise epoch 1 & 55291.22 + & & 277.7440349 & 45.7159869 & 0.064 & 0.065 & wise epoch 2 & 55472.33 + + wise 1841 + 7000 & 22.4 & 280.354101 & 70.009018 & 0.28 & 0.31 & 2mass reject & 51325.3783 & [email protected] & [email protected] & [email protected] & [email protected] + & & 280.3530928 & 70.0105747 & 0.086 & 0.085 & wise epoch 1 & 55226.23 + & & 280.3531427 & 70.0107065 & 0.122 & 0.119 & wise epoch 2 & 55590.80 + & & 280.3531553 & 70.0106363 & 0.263 & 0.201 & spitzer epoch 1 & 55459 + & & 280.3530727 & 70.0107399 & 0.233 & 0.239 & spitzer epoch 2 & 55586 + + wise 1852 + 3537 & 16.2 & 283.064682 & 35.622112 & 0.14 & 0.16 & 2mass psc & 50925.4682 & [email protected] & [email protected] & [email protected] & [email protected] + & & 283.0659051 & 35.6212634 & 0.105 & 0.111 & wise epoch 1 & 55297.44 + & & 283.0659721 & 35.6211283 & 0.096 & 0.101 & wise epoch 2 & 55480.31 + & & 283.0656975 & 35.6211998 & 0.214 & 0.220 & spitzer & 55465 + + wise 1906 + 4508 & 15.8 & 286.603188 & 45.136463 & 0.30 & 0.26 & 2mass psc & 50972.2922 & [email protected] & [email protected] & [email protected] & [email protected] + & & 286.6031075 & 45.1353261 & 0.085 & 0.086 & wise epoch 1 & 55307.23 + & & 286.6030488 & 45.1352444 & 0.087 & 0.090 & wise epoch 2 & 55489.38 + + wise 1952 + 7240 & 13.6 & 298.1944081 & 72.6669049 & 0.048 & 0.049 & wise epoch 1 & 55224.36 & [email protected] & [email protected] & [email protected] & [email protected] + & & 298.1942123 & 72.6668797 & 0.050 & 0.051 & wise epoch 2 & 55416.38 + & & 298.1942250 & 72.6668371 & 0.053 & 0.055 & wise epoch 3 & 55590.99 + & & 298.1940345 & 72.6664128 & 0.115 & 0.197 & spitzer & 55562 + + wise 1959 - 3338 & 11.6 & 299.773538 & -33.642067 & 0.17 & 0.17 & 2mass psc & 51348.2416 & [email protected] & [email protected] & [email protected] & [email protected] + & & 299.7736161 & -33.6426513 & 0.123 & 0.132 & wise epoch 1 & 55301.80 + & & 299.7735368 & -33.6427860 & 0.123 & 0.132 & wise epoch 2 & 55482.83 + + wise 2018 - 7423 & 13.3 & 304.6040683 & -74.3909858 & 0.118 & 0.122 & wise epoch 1 & 55290.76 & [email protected] & [email protected] + & & 304.6041586 & -74.3911074 & 0.100 & 0.104 & wise epoch 2 & 55471.93 + + wise 2056 + 1459 & 7.7 & 314.1204202 & 14.9981500 & 0.167 & 0.175 & wise epoch 1 & 55329.26 & [email protected] & [email protected] + & & 314.1203454 & 14.9982440 & 0.171 & 0.176 & wise epoch 2 & 55511.01 + & & 314.1205276 & 14.9982240 & 0.194 & 0.145 & spitzer & 55540 + + wise 2134 - 7137 & 10.1 & 323.7363420 & -71.6290364 & 0.159 & 0.162 & wise epoch 1 & 55301.54 & [email protected] & [email protected] + & & 323.7367230 & -71.6290448 & 0.120 & 0.123 & wise epoch 2 & 55482.63 + & & 323.7370297 & -71.6291393 & 0.227 & 0.187 & spitzer & 55516 + + wise 2157 + 2659 & 19.1 & 329.4640457 & 26.9920831 & 0.203 & 0.215 & wise epoch 1 & 55351.02 & [email protected] & [email protected] + & & 329.4639163 & 26.9920130 & 0.187 & 0.203 & wise epoch 2 & 55530.32 + & & 329.4639525 & 26.9920269 & 0.165 & 0.224 & spitzer & 55554 + + wise 2209 - 2734 & 13.4 & 332.344856 & -27.576269 & 0.17 & 0.18 & 2mass psc & 51119.0341 & [email protected] & [email protected] & [email protected] & [email protected] + & & 332.3420794 & -27.5776833 & 0.161 & 0.165 & wise epoch 1 & 55331.90 + & & 332.3418675 & -27.5777599 & 0.137 & 0.145 & wise epoch 2 & 55512.35 + & & 332.3419420 & -27.5778063 & 0.240 & 0.300 & spitzer & 55540 + + wise 2213 + 0911 & 18.8 & 333.478168 & 9.194306 & 0.26 & 0.25 & 2mass reject & 51738.3571 & [email protected] & [email protected] & [email protected] & [email protected] + & & 333.478251 & 9.194472 & 0.28 & 0.23 & 2mass reject & 51738.3621 + & & 333.4778840 & 9.1942733 & 0.169 & 0.180 & wise epoch 1 & 55348.46 + & & 333.4779537 & 9.1942974 & 0.212 & 0.222 & wise epoch 2 & 55526.69 + & & 333.4777956 & 9.1942456 & 0.224 & 0.160 & spitzer & 55562 + + wise 2226 + 0440 & 16.7 & 336.5960623 & 4.6677288 & 0.279 & 0.301 & wise epoch 1 & 55346.85 & [email protected] & [email protected] + & & 336.5960856 & 4.6675165 & 0.210 & 0.220 & wise epoch 2 & 55527.75 + & & 336.5959869 & 4.6676643 & 0.229 & 0.190 & spitzer & 55562 + + wise 2237 - 0614 & 16.4 & 339.372445 & -6.243156 & 0.23 & 0.24 & 2mass reject & 51087.1236 & [email protected] & [email protected] & [email protected] & [email protected] + & & 339.3730006 & -6.2428667 & 0.281 & 0.298 & wise epoch 1 & 55345.46 + & & 339.3728426 & -6.2428775 & 0.304 & 0.323 & wise epoch 2 & 55526.42 + & & 339.3729068 & -6.2428666 & 0.218 & 0.158 & spitzer & 55562 + + wise 2239 + 1617 & 18.3 & 339.904942 & 16.286875 & 0.12 & 0.12 & 2mass psc & 50711.2743 & [email protected] & [email protected] & [email protected] & [email protected] + & & 339.9064502 & 16.2878008 & 0.071 & 0.072 & wise epoch 1 & 55356.91 + & & 339.9065291 & 16.2877786 & 0.080 & 0.080 & wise epoch 2 & 55534.89 + & & 339.9064381 & 16.2877837 & 0.216 & 0.230 & spitzer & 55563 + + wise 2255 - 3118 & 13.0 & 343.9198262 & -31.3116495 & 0.165 & 0.173 & wise epoch 1 & 55339.91 & [email protected] & [email protected] + & & 343.9197639 & -31.3116969 & 0.154 & 0.159 & wise epoch 2 & 55521.06 + & & 343.9197363 & -31.3117099 & 0.288 & 0.178 & spitzer & 55548 + + wise 2313 - 8037 & 11.2 & 348.397355 & -80.615623 & 0.25 & 0.29 & 2mass reject & 51519.0917 & [email protected] & [email protected] & [email protected] & [email protected] + & & 348.4015688 & -80.6167123 & 0.093 & 0.089 & wise epoch 1 & 55297.70 + & & 348.4018567 & -80.6168599 & 0.086 & 0.084 & wise epoch 2 & 55480.40 + + wise 2319 - 1844 & 12.1 & 349.912834 & -18.735308 & 0.26 & 0.28 & 2mass reject & 51002.3174 & [email protected] & [email protected] & [email protected] & [email protected] + & & 349.9130580 & -18.7345555 & 0.144 & 0.152 & wise epoch 1 & 55351.15 + & & 349.9130362 & -18.7344856 & 0.159 & 0.165 & wise epoch 2 & 55530.59 + & & 349.9130638 & -18.7344989 & 0.141 & 0.302 & spitzer & 55562 + + wise 2325 - 4105 & 10.9 & 351.3314055 & -41.0930503 & 0.165 & 0.171 & wise epoch 1 & 55340.97 & [email protected] & [email protected] + & & 351.3314102 & -41.0930321 & 0.173 & 0.175 & wise epoch 2 & 55522.12 + & & 351.3314778 & -41.0930998 & 0.394 & 0.393 & spitzer & 55548 + + wise 2327 - 2730 & 21.7 & 351.868790 & -27.515934 & 0.09 & 0.08 & 2mass psc & 51341.3803 & [email protected] & [email protected] & [email protected] & [email protected] + & & 351.8697702 & -27.5157049 & 0.061 & 0.063 & wise epoch 1 & 55349.70 + & & 351.8697700 & -27.5157391 & 0.063 & 0.066 & wise epoch 2 & 55528.80 + & & 351.8698347 & -27.5157046 & 0.268 & 0.215 & spitzer & 55562 + + wise 2340 - 0745 & 12.1 & 355.110349 & -7.751374 & 0.22 & 0.19 & 2mass psc & 51098.1200 & [email protected] & [email protected] & [email protected] & [email protected] + & & 355.1109091 & -7.7522765 & 0.120 & 0.122 & wise epoch 1 & 55361.14 + & & 355.1108439 & -7.7523496 & 0.114 & 0.124 & wise epoch 2 & 55541.19 + & & 355.1109954 & -7.7523579 & 0.267 & 0.306 & spitzer & 55571 + + wise 2343 - 7418 & 14.9 & 355.960273 & -74.313484 & 0.36 & 0.37 & 2mass psc & 51774.2181 & [email protected] & [email protected] & [email protected] & [email protected] + & & 355.9633757 & -74.3130319 & 0.090 & 0.085 & wise epoch 1 & 55310.21 + & & 355.9637530 & -74.3130163 & 0.089 & 0.086 & wise epoch 2 & 55492.35 + + wise 2344 + 1034 & 15.6 & 356.1926713 & 10.5710647 & 0.334 & 0.347 & wise epoch 1 & 55369.15 & [email protected] & [email protected] + & & 356.1926410 & 10.5711379 & 0.381 & 0.385 & wise epoch 2 & 55548.91 + & & 356.1927398 & 10.5710142 & 0.196 & 0.169 & spitzer & 55580 + + wise 2348 - 1028 & 16.8 & 357.169203 & -10.479364 & 0.15 & 0.13 & 2mass psc & 51105.0337 & [email protected] & [email protected] & [email protected] & [email protected] + & & 357.1712843 & -10.4788976 & 0.172 & 0.182 & wise epoch 1 & 55361.94 + & & 357.1713877 & -10.4789185 & 0.189 & 0.194 & wise epoch 2 & 55541.59 + & & 357.1713313 & -10.4788145 & 0.256 & 0.204 & spitzer & 55571 + + wise 2359 - 7335 & 13.1 & 359.918122 & -73.584862 & 0.19 & 0.23 & 2mass psc & 51528.1013 & [email protected] & [email protected] & [email protected] & [email protected] + & & 359.9211060 & -73.5846766 & 0.073 & 0.070 & wise epoch 1 & 55316.75 + & & 359.9211513 & -73.5847461 & 0.071 & 0.069 & wise epoch 2 & 55494.74 + & & 359.9212621 & -73.5847480 & 0.260 & 0.222 & spitzer & 55552 + cccccccc wise 0254 + 0223 & t8 & [email protected] & 4.7 - 8.4 & [email protected] & [email protected] & [email protected] + wise 1541 - 2250 & y0 & [email protected] & 2.2 - 4.1 & [email protected] & [email protected] & [email protected] + wise 1647 + 5632 & l9 pec ( red ) & [email protected] & 6.9 - 11.5 & [email protected] & [email protected] & [email protected] + wise 1741 + 2732 & t9 & [email protected] & 4.5 - 6.9 & [email protected] & [email protected] & [email protected] + ccccccccc t6-t6.5 & 900 - 1050 & 20 & 14 & 16 & 30 & @xmath49.0e-4 & 0.36 & @xmath42.5 + t7-t7.5 & 750 - 900 & 20 & 13 & 12 & 25 & @xmath47.5e-4 & 0.34 & @xmath42.1 + t8-t8.5 & 600 - 750 & 20 & 9 & 26 & 35 & @xmath41.0e-3 & 0.30 & @xmath42.9 + t9-t9.5 & 450 - 600 & 15 & 1 & 9 & 10 & @xmath47.1e-4 & 0.40 & @xmath42.0 + y0 & 300 - 450 & 10 & 0 & 4 & 4 & @xmath49.5e-4 & 0.59 & @xmath42.7 + @xmath4y0 & @xmath11300 & 10 & 0 & 1 & 1 & @xmath42.4e-4 & 0.83 & @xmath40.7 + | we present ground - based spectroscopic verification of six y dwarfs ( see also cushing et al . ) , eighty - nine t dwarfs , eight l dwarfs , and one m dwarf identified by the wide - field infrared survey explorer ( wise ) .
eighty of these are cold brown dwarfs with spectral types @xmath0t6 , six of which have been announced earlier in mainzer et al . and burgasser et al .
we present color - color and color - type diagrams showing the locus of m , l , t , and y dwarfs in wise color space .
near - infrared and , in a few cases , optical spectra are presented for these discoveries .
near - infrared classifications as late as early y are presented and objects with peculiar spectra are discussed . using these new discoveries
, we are also able to extend the optical t dwarf classification scheme from t8 to t9 . after deriving an absolute wise 4.6 @xmath1 m ( w2 ) magnitude vs. spectral type relation
, we estimate spectrophotometric distances to our discoveries .
we also use available astrometric measurements to provide preliminary trigonometric parallaxes to four our discoveries , which have types of l9 pec ( red ) , t8 , t9 , and y0 ; all of these lie within 10 pc of the sun .
the y0 dwarf , wise 1541@xmath22250 , is the closest at @xmath3 pc ; if this 2.8 pc value persists after continued monitoring , wise 1541@xmath22250 will become the seventh closest stellar system to the sun .
another ten objects , with types between t6 and @xmath4y0 , have spectrophotometric distance estimates also placing them within 10 pc .
the closest of these , the t6 dwarf wise 1506 + 7027 , is believed to fall at a distance of @xmath54.9 pc .
wise multi - epoch positions supplemented with positional info primarily from _ spitzer_/irac allow us to calculate proper motions and tangential velocities for roughly one half of the new discoveries .
this work represents the first step by wise to complete a full - sky , volume - limited census of late - t and y dwarfs . using early results from this census , we present preliminary , lower limits to the space density of these objects and discuss constraints on both the functional form of the mass function and the low - mass limit of star formation |
theoretical study of resonant formation of the muonic molecule @xmath0 in condensed deuterium targets is the main subject of this paper .
the resonant @xmath0 formation , first observed by dzhelepov and co - workers @xcite , is a key process of muon catalyzed fusion ( @xmath7cf ) in deuterium ( see e.g. reviews @xcite ) . a muonic deuterium atom @xmath1 is created when a negative muon @xmath8 is captured into an atomic orbital in a deuterium target .
after @xmath1 deexcitation to the @xmath9 state and slowing down , the @xmath0 molecule can be formed in @xmath1 atom collision with one of the d@xmath2 target molecules .
the resonant formation is possible due to presence of a loosely bound state of @xmath0 , characterized by the rotational number @xmath10 and vibrational number @xmath11 , with binding energy @xmath12 1.97 ev .
this energy , according to the vesman mechanism @xcite , is completely transferred to excited rovibrational states of the molecular complex @xmath13 $ ] .
the scheme of calculation of @xmath0 formation rate in gaseous deuterium has been developed for many years @xcite , and has lead to a good agreement with the experiments performed in gaseous targets @xcite .
on the other hand , this theory , when directly applied to solid deuterium targets , leads to strong disagreement with the experimental results @xcite .
therefore , it is necessary to calculate the @xmath0 formation rate with solid state effects taken into account , which is the main purpose of this paper .
our calculations are based on the theoretical results ( transition matrix elements , resonance energies ) obtained in the case of @xmath0 formation in a single d@xmath2 molecule . in sec .
[ sec : freemol ] the main formulas used for this case are briefly reported . a general formula for the energy - dependent @xmath0 formation rate in a d@xmath2 condensed target is derived in sec .
[ sec : formalism ] , using the van hove formalism of the single - particle response function @xcite .
this formula is then applied ( sec .
[ sec : formsol ] ) for harmonic solid targets , in particular for a cubic bravais lattice .
a phonon expansion of the response function is used to study phonon contributions to the resonant formation .
numerical results for 3 k zero pressure frozen deuterium targets ( triumf experimental conditions @xcite ) , with the fcc polycrystalline structure , are shown in sec .
[ sec : deut3k ] .
the formation rates have been calculated assuming the isotropic debye model of the solid and the values of debye temperature and lattice constant observed in neutron scattering experiments .
the calculated rates of resonant @xmath0 formation and back decay have been used for monte carlo simulations of @xmath3 fusion neutron and proton time spectra .
since the initial distributions of @xmath9 muonic atom energy contain contributions from hot @xmath1 s ( @xmath14 1 ev ) @xcite , influence of slow deceleration of @xmath1 atoms below 10 mev @xcite on these time spectra is investigated in sec .
[ sec : monte_carlo_form ] .
the simulations take into account the processes of incoherent and coherent @xmath1 atom scattering in solid deuterium .
in particular , the bragg scattering , phonon scattering , and rovibrational transitions in d@xmath2 molecules are included .
we consider a dependence of the resonant formation rate and time spectra on broadening of the rovibrational d@xmath2 energy levels , due to the binding of the molecules in the lattice @xcite . since it has been predicted in refs .
@xcite that strong @xmath0 formation takes place only in solid para - d@xmath2 , study of this process in pure ortho - d@xmath2 and para - d@xmath2 targets is another aim of this work . the neutron spectra calculated for these two
solids are discussed in sec .
[ sec : monte_carlo_form ] .
first we consider resonant formation of the @xmath0 molecule in the following reaction @xmath15_{\nu_fk_f } \,,\ ] ] where d@xmath2 is a free deuterium molecule in the initial rovibrational state @xmath16 and the total nuclear spin @xmath17 .
the muonic atom @xmath1 has total spin @xmath18 and cms kinetic energy @xmath19 .
the complex @xmath13 $ ] is created in the rovibrational state @xmath20 and the molecular ion @xmath0 , which plays the role of a heavy nucleus of the complex , has total spin @xmath21 .
the rate @xmath22 of the process above depends on the elastic width @xmath23 of @xmath13 $ ] complex decay @xcite in reactions @xmath24_{\nu_fk_f } & \\ & \xrightarrow[\tilde{\lambda}_f ] \quad \text{stabilization~processes , } \end{split}\ ] ] where @xmath25 is the total rate of the stabilization processes , i.e. deexcitation and nuclear fusion in @xmath0 @xmath26 when fusion takes place , the muon is generally released and can again begin the @xmath7cf cycle .
however , sometimes the muon is captured into an atomic orbital of helium ( sticking ) , which stops further reactions .
the value of @xmath23 is given in atomic units ( @xmath27 1 ) by the formula @xmath28 where @xmath29 is a transition matrix element and @xmath30 is a resonance energy defined in ref .
the factor @xmath31 is due to averaging over initial and summing over final projections of spins and angular momenta of the system .
vector @xmath32 is the momentum of relative @xmath1 and d@xmath2 motion @xmath33 and @xmath34 is the reduced mass of the system .
integration of eq .
( [ eq : el_width ] ) over @xmath32 leads to @xmath35 since @xmath23 and @xmath25 are much lower ( @xmath36 mev ) than @xmath19 , vesman s model can be applied and the energy - dependent resonant formation rate has the dirac delta function profile @xmath37 where @xmath38 is the density of deuterium nuclei in the target . according to ref .
@xcite the coefficients @xmath31 and @xmath39 in the above equations are equal to @xmath40 where @xmath41 and the curly brackets stand for the wigner @xmath42 symbol . in formula ( [ eq : ab_def ] ) the usual boltzmann factor describing the population of rotational states in a gas target is omitted because we calculate the formation rate separately for each initial rotational state . if the muonic atoms in a gas have a steady kinetic energy distribution @xmath43 at target temperature @xmath44 , eq .
( [ eq : resrate ] ) can be averaged over the atom motion leading to a mean resonant rate @xmath45 .
since a muonic deuterium atom can be approximately treated as a small neutron - like particle , methods used for description of neutron scattering and absorption in condensed matter are applicable in the case @xmath0 formation in dense deuterium targets .
below we adapt the method developed by lamb @xcite , and then generalized by singwi and sjlander @xcite using the van hove formalism of the single - particle response function @xmath46 @xcite , for calculation of the resonant @xmath0 formation rates .
a hamiltonian @xmath47 of a system , consisting of a @xmath1 atom in the 1@xmath48 state and a heavy condensed d@xmath2 target , can be written down as follows @xmath49 where @xmath50 is the @xmath1 mass and @xmath51 denotes the position of @xmath1 center of mass in the coordinate frame connected with the target ( see fig . [ fig : sys_cmplx ] ) .
operator @xmath52 is the hamiltonian of a free @xmath1 atom , @xmath53 is @xmath1 internal vector ; @xmath54 denotes the internal hamiltonian of a free d@xmath2 molecule .
it is assumed that @xmath0 formation takes place in collision with the l - th d@xmath2 target molecule .
the position of its mass center in the target frame is denoted by @xmath55 ; @xmath56 is a vector connecting deuterons inside this molecule .
function @xmath57 stands for the potential of the @xmath1d@xmath2 interaction @xcite , leading to @xmath0 resonant formation .
vector @xmath58 connects the @xmath1 and d@xmath2 centers of mass .
we neglect contributions to the potential @xmath57 from the molecules other than the l - th molecule because we assume here that distances between different molecules in the target are much greater than the d@xmath2 size .
the kinetic energy @xmath19 of the @xmath1 atom and its momentum @xmath32 in the target frame are connected by the relation @xmath59 the hamiltonian @xmath60 of a pure d@xmath2 target , corresponding to the initial target energy @xmath61 , has the form @xmath62 where @xmath63 is the position of @xmath64-th molecule center of mass in the target frame ( fig .
[ fig : sys_lattice ] ) , @xmath65 denotes interaction between the @xmath64-th and @xmath66-th molecule , and @xmath67 is the mass of a single target molecule .
the coordinate part @xmath68 of the initial wave function of the system can be written as a product @xmath69 where @xmath70 stands for the initial wave function of the condensed d@xmath2 target , corresponding the total energy @xmath61 .
eigenfunctions of the operators @xmath52 and @xmath54 are denoted by @xmath71 and @xmath72 , respectively . using the relation @xmath73
, the wave function @xmath68 takes the form @xmath74 which is similar to that used in the case of @xmath0 formation on a single d@xmath2 , except the factor @xmath75 .
this factor depends only on positions of mass centers of the target molecules .
after formation of @xmath13 $ ] complex , the total hamiltonian of the system is well approximated by the operator @xmath76 @xmath77 where @xmath78 is an internal hamiltonian of @xmath0 molecular ion , vectors @xmath79 and @xmath80 are its jacobi coordinates .
relative motion of @xmath0 and @xmath81 in the complex is described by a hamiltonian @xmath82 which depends on the respective internal vector @xmath83 .
the final hamiltonian @xmath84 of the target , with the eigenfunction @xmath85 and energy eigenvalue @xmath86 , is expressed by the formula @xmath87 where @xmath88 is the mass of the complex .
the respective coordinate part @xmath89 of the total final wave function of the system is @xmath90 where @xmath91 and @xmath92 denote eigenfunctions of the hamiltonians @xmath78 and @xmath82 , respectively .
the energy - dependent resonant @xmath0 formation rate @xmath93 in the condensed target , for the initial @xmath70 and final @xmath85 target states and a fixed @xmath1 total spin @xmath94 , is calculated using the formula @xmath95 with the resonance condition @xmath96 taking into account the initial and final energy of the target .
the resonant energy for a free d@xmath2 is denoted by @xmath30 and the transition matrix element is given by @xmath97 using eqs .
( [ eq : wf_tot2 ] ) and ( [ eq : wf_tot ] ) the matrix element ( [ eq : matrixel1 ] ) can be written as a product @xmath98 where @xmath29 is the transition matrix element calculated for a single d@xmath2 molecule @xcite .
the rate ( [ eq : resratsol0 ] ) can be additionally averaged over a distribution @xmath99 of the initial target states at a given temperature @xmath44 and summed over the final target states , which leads to @xmath100 factor @xmath39 , defined by eqs .
( [ eq : ab_def ] ) , is due to the averaging over the initial projections and summation over the final projections of spin and rovibrational quantum numbers .
this factor takes also into account a symmetrization of the total wave function of @xmath1+d@xmath2 system over three deuterium nuclei .
now we introduce a time variable @xmath101 to eliminate the @xmath102 function in the equation above and then we involve time - dependent operators , which is familiar in scattering theory ( see , e.g. , refs @xcite ) .
using the fourier expansion of the @xmath102 function @xmath103 one has @xmath104 assuming that the perturbation operator @xmath105 is well - approximated by its mean value @xmath106 which is valid when the target relaxation time is much smaller than the @xmath0 lifetime of the order of 10@xmath107 s , the matrix element in eq .
( [ eq : resratsol2 ] ) can be expressed as @xmath108 where @xmath109 denotes the heisenberg operator and @xmath110 in formula ( [ eq : res_shift ] ) is the mean kinetic energy of the target molecule at temperature @xmath44 . using the identity @xmath111 in eq .
( [ eq : resratsol2 ] ) we obtain @xmath112 where @xmath113 denotes both the quantum mechanical and the statistical averaging at temperature @xmath44 , and @xmath114 being the resonance energy @xmath115 shifted by @xmath116 .
note that such a resonant energy shift was neglected in papers @xcite , where absorption of neutrons and @xmath117-rays by heavy nuclei were considered .
an estimation of the shift in the case of @xmath117 emission from a nucleus bound in a solid , similar to eq .
( [ eq : res_shift ] ) was given in ref .
@xcite .
a self pair correlation function @xmath118 is defined by the following equation @xcite @xmath119 and the single - particle response function @xmath120 is given by the formula @xmath121 thus , by virtue of eqs .
( [ eq : resratsol3 ] ) and ( [ eq : resp_def ] ) , the resonant formation rate in a condensed target can by expressed in terms of the response function @xmath122 where the momentum transfer @xmath123 and energy transfer @xmath124 to the target are defined as follows @xmath125 the advantage of the van hove method is that all properties of the target , for given momentum and energy transfers , are contained in the factor @xmath126 .
it is possible to rigorously calculate @xmath46 in the case of a perfect gas and in the case of a harmonic solid . however , a liquid target or a dense gas target is a difficult problem to solve .
proceeding as above one can obtain a similar formula for @xmath127 in a condensed target ( in general , @xmath1 spin @xmath128 after back decay can be different from @xmath1 spin @xmath94 before the formation ) @xmath129 @xmath130 is the response function calculated for the state @xmath131 and @xmath132 where @xmath133 denotes the mean kinetic energy of the complex bound in the target .
it has been shown by van hove @xcite that the self correlation function in the case of a gas or a solid with cubic symmetry takes the general form @xmath134 for a cubic bravais lattice , in which each atom is at a center of inversion symmetry , @xmath135 is given by the formula @xmath136 where @xmath137 is the normalized vibrational density of states such that @xmath138 @xmath139 is the bose factor @xmath140^{-1}\ , , \quad \beta = ( k_{\text{b } } t)^{-1 } \,.\ ] ] and the boltzmann constant is denoted by @xmath141 . the response function ( [ eq : resp_def ] ) , after substitution of eqs .
( [ eq : corr_gen ] ) , ( [ eq : gamma_def ] ) and integration over @xmath79 , can be written as follows @xmath142 \biggrb ) \
, , \end{split}\ ] ] @xmath143 denotes the limit of @xmath135 at @xmath144 .
this formula can be expanded in a power series of the momentum transfer @xmath145 , which leads to @xmath146 \,,\ ] ] where @xmath147 is the debye - waller factor , familiar in the theory of neutron scattering , @xmath148 and the functions @xmath149 are given by @xmath150 \
, , \\ & g_n(w , t ) = \int_{-\infty}^{\infty } dw ' \ , g_1(w - w',t)\ , g_{n-1}(w',t ) \ , ,
\\ & \int_{-\infty}^{\infty } dw \ , g_n(w ) = 1 \ , .
\end{split}\ ] ] in the case of a cubic crystal structure @xmath147 can also be expressed as @xmath151 where @xmath152 is the displacement of a molecule from its lattice site .
substitution of eq .
( [ eq : resp2 ] ) to eq .
( [ eq : resratsol4 ] ) leads to the following formation rate @xmath153 \,,\ ] ] the first term in expansion ( [ eq : resratsol5 ] ) represents a sharp peak describing the @xmath102 profile recoilless formation .
the next terms give broad distributions corresponding to subsequent multi - phonon processes .
in particular , the term with @xmath154 describes formation connected with creation or annihilation of one phonon . if @xmath155 we deal with so - called strong binding @xcite where only the few lowest terms in the above expansion are important . on the other hand , in the limit @xmath156 ( weak - binding )
many multi - phonon terms give comparable contributions to ( [ eq : resratsol5 ] ) .
therefore , for sufficiently large @xmath157 it is convenient to use the impulse approximation in which @xmath135 is replaced by its value near @xmath158 @xmath159 this leads to the asymptotic formula for @xmath46 @xmath160 where @xmath161 the mean kinetic energy @xmath110 of a molecule in the solid , which also determines the resonance energy shift ( [ eq : res_shift ] ) , is equal to @xmath162 \,.\ ] ] the energy @xmath110 contains a contribution from the zero - point vibrations and it approaches @xmath163 only at high temperatures @xmath164 .
function ( [ eq : resp_asym ] ) is a gaussian with response centered at the recoil energy @xmath165 .
therefore in the weak binding region the resonant formation rate takes the doppler form obtained by bethe and placzek ) is the limit of the bethe formula in the case of a very narrow natural resonance width @xmath166 . ] for resonant absorption of neutrons in gas targets @xcite .
however , the resonance width ( [ eq : dopp_width ] ) in the solid at temperature @xmath44 is different from the doppler width in a maxwellian gas @xmath167 unless the temperature is sufficiently high . this phenomenon was pointed out by lamb in his paper @xcite concerning resonant neutron absorption in solid crystals . by virtue of the equations above one
can introduce for the solid an effective temperature @xmath168 @xmath169
the following considerations concern the solid deuterium crystals used in the triumf experiments @xcite , though the results presented below can be applied to targets obtained in similar conditions @xcite . at triumf
thin solid deuterium layers have been formed by rapid freezing of gaseous d@xmath2 on gold foils at @xmath170 k and zero pressure . according to ref .
@xcite such deuterium layers have the face - centered cubic ( fcc ) polycrystalline structure .
since the distance between the neighboring molecules is a few times greater than the diameter of a d@xmath2 molecule and the van der waals force that binds the solid is weak , one can neglect perturbations of the resonant formation potential @xmath57 due to these neighbors .
the deuterium crystals at zero pressure are quantum molecular crystals .
the amplitude of zero - point vibration at 3 k equals 15% of the nearest neighbor distance .
a single - particle potential in this case is not harmonic and the standard lattice dynamics leads to imaginary phonon frequencies .
however , the standard dynamics can be applied after a renormalization of the interaction potential , taking into account the short - range pair correlations between movement of the neighbors @xcite . in result ,
the theoretical calculations @xcite of the phonon dispersion relations give a good agreement with the neutron scattering experiments @xcite and the debye model for solid deuterium can be used as a good approximation of the phonon energy distribution @xmath171 with the debye energy @xmath172 and debye temperature @xmath173 taken from the neutron experiments . for @xmath170
k we use the debye model of an isotropic solid with @xmath174 k corresponding to the maximal phonon energy @xmath175 mev .
thus , we are dealing with the limit @xmath176 where @xmath177 are very good approximations of eqs .
( [ eq:2w ] ) , ( [ eq : meankin ] ) and ( [ eq : t_eff ] ) .
the debye - waller factor and mean kinetic energy @xmath110 at lowest temperatures are determined by contributions from the zero - point d@xmath2 vibration in the lattice , and therefore these quantities do not tend to zero at @xmath178 .
the zero - point energy is not accessible energy , but its effects are always present .
the values of the resonance energies depend on initial and final rovibrational quantum numbers of the system . in solid hydrogens at low pressures
these quantum numbers remain good quantum numbers , but excited energy levels broaden into energy bands ( rotons and vibrons ) due to coupling between neighboring molecules @xcite .
the calculations presented in the literature concern pure solid h@xmath2 , hd and d@xmath2 targets and only lowest quantum numbers . the problem of a heavier impurity , such as @xmath179 complex in d@xmath2 , has not been considered yet .
however , knowing that the width of the rotational bands can reach about 1 mev @xcite , a possible influence of this effect on the calculated formation rates and fusion neutron time spectra is discussed in the next section . at low temperatures all d@xmath2 molecules are in the ground vibrational state @xmath180 and @xmath0
is formed via the excitation of the complex to the state @xmath181 . unless a catalyst is applied , rapidly frozen deuterium is a mixture of ortho - d@xmath2 ( @xmath182 ) and para - d@xmath2 ( @xmath183 ) . in
the triumf experiments gaseous deuterium was pumped through a hot palladium filter before freezing .
therefore the solid target was a statistical mixture ( 2:1 ) of the ortho- and para - states ( @xmath184 ) .
since the para - ortho relaxation without a catalyst is very slow ( 0.06%/h ) in solid deuterium @xcite , the population of these states is not changed during experiments of a few days .
the lowest resonance energies @xmath30 and @xmath185 , for fixed @xmath186 , @xmath187 and different values of @xmath94 , @xmath188 , @xmath48 and @xmath189 are shown in table [ tab : eres ] @xcite .
a few of them have negative values , which means that to satisfy the resonance condition @xmath190 an energy excess in the @xmath1+d@xmath2 system should be transferred to external degrees of freedom .
this is possible in dense targets , where energy of neighboring molecules can be increased . such an effect ,
due to triple collisions in gas targets , has been firstly discussed in ref .
@xcite . in a solid
, the energy excess is lost through incoherent phonon creation . according to ( [ eq : res_shift ] ) , ( [ eq : eres_sol ] ) , and ( [ eq:2w_asym ] ) , in the considered 3 k solid deuterium all resonant energies @xmath185
are shifted by @xmath191 mev .
one can see that all resonances for @xmath192 are placed at higher energies , which is caused by @xmath1 hyperfine splitting @xmath193 48.5 mev .
all resonance energies @xmath194 10 mev are connected with formation from the upper spin state @xmath195 of @xmath1 .
however , only resonances corresponding to the dipole transitions @xmath196 and @xmath197 can give a significant contribution to the formation rate at lowest energies .
other transition matrix elements described in ref .
@xcite tend to zero when @xmath198 ( see figs . [
fig : v0f ] and [ fig : v1f ] obtained for @xmath182 and @xmath183 ) . the low energy rates ( @xmath199 )
are calculated using formula ( [ eq : resratsol5 ] ) with a few most significant terms of the response function expansion ( [ eq : resp2 ] ) taken into account .
[ fig : resp ] shows the function @xmath200 corresponding to the two dipole transitions in para - d@xmath2 .
the subthreshold resonance , with @xmath201 mev , gives contributions to the formation rate only through the phonon creation processes . for @xmath202 mev ,
the non - phonon process is possible and it is represented by a vertical line .
different peaks in this figure describe processes connected with different numbers of created phonons .
in particular , one - phonon processes , which are proportional to @xmath137 with the characteristic debye cutoff , can be clearly distinguished . since
the @xmath203-phonon term in ( [ eq : resp2 ] ) is proportional to @xmath204 , the @xmath0 formation rate tends to zero at @xmath198 .
note that the phonon annihilation gives negligible contribution to the rate at very low target temperatures @xmath205 . in order to compare the calculated formation rates with experiments the summed rates @xmath206 are introduced @xmath207 in fig .
[ fig : rate3_abs ] the formation rates @xmath206 in the solid ortho - d@xmath2 and para - d@xmath2 are shown for @xmath195 . in the case of resonances satisfying the condition
@xmath208 we have @xmath209 and the expansion ( [ eq : resratsol5 ] ) is used .
the two strong peaks represent the recoilless formation process , without phonon excitations .
the delta function profile of every peak is shown as a rectangle with a height equal to the formation rate strength divided by the total decay width ( @xmath210 mev ) .
the strength defined as the value of the factor standing before @xmath211 in the expansion ( [ eq : resratsol5 ] ) , is equal to 0.1061 ev@xmath212s@xmath213 for the resonance @xmath214 in solid ortho - d@xmath2 .
the transition @xmath215 in para - d@xmath2 gives 0.07544 ev@xmath212s@xmath213 as the resonance strength .
higher resonance energies involve many multi - phonon terms and therefore we use the asymptotic form ( [ eq : resp_asym ] ) of @xmath216 for @xmath217 .
all formation rates presented in the figures are normalized to the liquid hydrogen density @xmath218 atoms/@xmath219 .
though in monte carlo simulations , involving energy - dependent rates of different processes , the `` absolute '' formation rates @xmath206 should be used , it is convenient to introduce an effective formation rate @xmath220 which leads to the nuclear @xmath3 fusion in @xmath13 $ ] complex
. back decay of the complex to the @xmath221d@xmath2 system , characterized by the quantum numbers @xmath222 and @xmath128 , strongly influences the fusion process because the back - decay rates are comparable with the effective fusion rate @xmath223 374 @xmath7s@xmath213 @xcite . since in a solid target rotational deexcitation of the asymmetric
complex is much faster than back decay and fusion , it is assumed that back decay takes place only from the state @xmath224 .
the effective formation rate is then defined by the following formula @xmath225 where the fusion fraction @xmath226 is given by @xmath227 since the frequency of lattice vibrations ( @xmath228 @xmath7s@xmath213 ) is many orders of magnitude greater than the back - decay and fusion rates , energetic phonons created during the @xmath0 formation process are dissipated . at 3 k the number of phonons with energies @xmath229 0.26 mev is strongly suppressed by the bose factor @xmath139 . therefore back decay with phonon annihilation at @xmath205 is negligible . in particular , the phonon channel of decay of @xmath0 , formed from @xmath1 state @xmath195 due to the subthreshold resonances , is closed because this would require an annihilation of a phonon with energy of a few mev . in this case back decay is connected with the spin - flip transition to @xmath230 . since the corresponding energy release of a few tens of mev is much greater than the debye energy ( @xmath231 ) , the @xmath0 decay rate is dominated by contributions from simultaneous phonon creation processes .
after integration of formula ( [ eq : elwidthsol1 ] ) over direction of vector @xmath32 one obtains @xmath232 and then substitution of expansion ( [ eq : resp2 ] ) and integration of the recoilless term lead to @xmath233 \ , , \end{split}\ ] ] where @xmath234 it is assumed in the formula above that the phonon energy spectrum of solid deuterium containing @xmath13 $ ] is similar to that of a pure deuterium lattice .
the problem of lattice dynamics of a quantum solid deuterium crystal containing a small admixture of a heavier isotope has not been considered yet in literature , at least to the knowledge of the authors .
however , this approximation is reasonable since the debye temperatures of solid hydrogen and deuterium at 3 k are very similar @xcite , independently of the mass difference of these isotopes
. therefore it is assumed that during the @xmath0 lifetime the mean kinetic energy @xmath133 of the complex reaches the energy @xmath110 characterizing a pure deuterium solid .
thus the resonance energy shift ( [ eq : eres_cplx ] ) is approximated by @xmath235 which gives @xmath236 mev .
the effective formation rates in 3 k solid deuterium for @xmath195 are shown in fig .
[ fig : eddm3 ] .
the phonon part of the rates below a few mev is about two orders of magnitude lower than the average rate of 2.7 @xmath7s@xmath213 derived from the experiment @xcite .
this means that at @xmath237 the phonon contribution to the total resonant formation rate is even smaller than the non resonant @xmath0 formation rate @xmath238 0.44 @xmath7s@xmath213 @xcite , and that the estimation of the phonon contribution given in ref .
@xcite is strongly overestimated .
therefore , the experimental results can only be explained by resonant @xmath0 formation at energies @xmath239 mev , where the rate exceeds significantly the value of 1 @xmath7s@xmath213 . a cusp at 0.3 mev in para - d@xmath2 is due to the formation with simultaneous one - phonon creation , connected with the subthreshold resonance @xmath240 .
this implies a significant difference between the resonant formation in ortho - d@xmath2 and para - d@xmath2 below 1 mev .
however , this difference is difficult to measure because of a broad distribution of @xmath1 energy .
note that a similar subthreshold phonon effect in the case of resonant @xmath241 formation in solid deuterium has been discussed in ref @xcite . in the solid target
the fusion fraction @xmath242 0.3 and the total resonance width @xmath243 mev for both @xmath244 and @xmath245 .
the back - decay rate @xmath246 from @xmath244 to @xmath230 equals about 843 @xmath7s@xmath213 .
decay @xmath247 is impossible . in the case of
@xmath245 we have obtained @xmath248 281 @xmath7s@xmath213 for @xmath230 and @xmath248 610 @xmath7s@xmath213 for @xmath249 .
phonon creation processes give dominant contributions to the back - decay rates , e.g. , the non - phonon part of @xmath246 , given by the first term of expansion ( [ eq : width_expan ] ) , equals 169 @xmath7s@xmath213
. therefore the @xmath1 energy spectrum , after back decay in the solid , is not discrete . in fig .
[ fig : eddm1 ] the effective rates in solid deuterium for @xmath192 are presented . for the sake of comparison the formation rate for 3 k ortho - d@xmath2 gas
is also plotted .
the `` gas '' curve has been calculated using the asymptotic formula ( [ eq : resp_asym ] ) for @xmath46 with @xmath250 k.this figure shows that in a real solid deuterium target the rates are smeared much more than in a gas target with the same temperature , because of the zero - point vibrations . therefore even at relatively high @xmath1 energies of some 0.1 ev one should not neglect the solid effects and use the formation rates calculated for a 3 k maxwellian gas .
the calculated energy - dependent @xmath0 formation rates have been applied in our monte carlo simulations of @xmath7cf in 3 k solid deuterium targets .
the final @xmath1 energy distribution after back decay , including simultaneous phonon creation processes , has been determined through a numerical integration of eq .
( [ eq : width_expan ] ) . the calculated distribution is shown in fig .
[ fig : ebck ] for @xmath244 , @xmath224 and @xmath230 .
the rotational transitions to @xmath251 with no phonon creation are seen as the delta peaks .
the continuous energy spectrum describes phonon creation contribution to @xmath1 energy .
note that , opposite to @xmath0 formation rates , this phonon contribution ( for a given rotational transition peak ) extends towards lower energies .
the average @xmath1 energy after @xmath0 back decay equals about 30 mev , for the presented spectrum .
the @xmath3 fusion neutron and proton spectra depend on the time evolution of @xmath1 energy .
this energy is determined by differential cross sections of different scattering processes of @xmath1 atoms in a given solid target , including elastic scattering , rovibrational transitions , spin - flip reactions and phonon processes .
the scattering cross sections in a solid are calculated using the van hove method .
some results of such calculations for @xmath1 atoms in fcc solid deuterium have been presented and discussed in ref .
the incoherent processes , such as spin - flip or rovibrational transitions , are described by the self pair correlation function @xmath118 defined by eq .
( [ eq : selfcor ] ) .
the bragg scattering and coherent phonon scattering are connected with a pair correlation function @xmath252 @xcite . in fig .
[ fig : xfddd22 ] is shown the total cross section for @xmath253 scattering in the statistical mixture of 3 k solid ortho - d@xmath2 and para - d@xmath2 .
bragg scattering , with the bragg cutoff at @xmath254 1.1 mev , and incoherent elastic scattering do not change @xmath1 energy because of the very large mass of the considered solid target .
below 1.7 mev the @xmath1 atom is effectively accelerated , mainly due to the rotational deexcitation of para - d@xmath2 molecules @xcite .
this transition is enabled by muon exchange between deuterons in @xmath221d@xmath2 scattering .
the curve `` @xmath255 '' in fig .
[ fig : xfddd22 ] , describing the rotational deexcitation , includes contributions from simultaneous incoherent phonon processes .
this cross section at @xmath256 2.5 mev equals 0.22@xmath257 @xmath258 , which is about three times less ( taking into account the statistical factor of 1/3 for @xmath259 states ) than the estimation given in paper @xcite .
phonon annihilation is a much weaker @xmath1 acceleration mechanism than the rotational deexcitation .
since the coherent amplitude for @xmath1 elastic scattering on a single d@xmath2 molecule is greater by two orders of magnitude than the incoherent amplitude , the coherent processes involving conservation of momentum dominate low energy @xmath1 scattering in solid deuterium .
it is especially important below a few mev , where the shapes of coherent and incoherent cross sections differ strongly .
the small phonon creation cross section below 1.1 mev , leading to @xmath1 energy loss , is due to the incoherent amplitude .
coherent phonon creation is impossible below @xmath260 .
this limit is obtained in the case of coherent one - phonon creation process , for the total momentum conservation involving the smallest ( non - zero ) inverse lattice vector @xmath261 , which also fixes the position of the first peak of the bragg scattering at @xmath254 1.1 mev . for @xmath262 one - phonon creation
is possible only if the @xmath1 velocity is not lower than the sound velocity in the crystal , which is well - known in neutron physics .
according to ref .
@xcite the mean sound velocity in solid deuterium equals about 1.2@xmath263 cm / s and this corresponds to @xmath1 energy of 15 mev .
therefore , neglecting the inverse lattice contribution to the one - phonon creation cross section in ref .
@xcite leads to the severe underestimation of @xmath1 slowing down at lowest energies and subsequent overestimation of @xmath1 kinetic energy . above 1.7 mev phonon creation
already prevails over all acceleration processes .
however , the effective deceleration rate below @xmath264 is strongly suppressed by the dominating bragg elastic scattering . at energies above some 10 mev
subsequent rotational and then vibrational excitations of d@xmath2 molecules become important and they provide a very fast mechanism of @xmath1 deceleration at higher energies .
the total cross section for @xmath253 scattering in a pure 3 k ortho - d@xmath2 target ( see fig . [
fig : xfddd22k0 ] ) is quite similar to that shown in fig .
[ fig : xfddd22 ] .
a significant difference is the lack of rotational deexcitation .
therefore phonon annihilation is the only , and weak , acceleration mechanism .
it dominates the inelastic cross section below 1.4 mev .
[ fig : avendm3 ] presents the time evolution of average @xmath253 atom energy @xmath265 , obtained from our monte carlo calculations .
it is assumed that the target is infinite and that @xmath1 atoms have initially a maxwellian energy distribution with a mean energy of 1 ev .
a statistical initial population of @xmath1 total spin is used and the theoretical non - resonant part of the total spin - flip rate @xmath266 is multiplied by a single scaling factor of 0.4 , in order to keep agreement with the experimental values @xcite of the spin - flip rate .
the calculations have been performed for ortho - d@xmath2 , para - d@xmath2 and their statistical mixture ( stat ) .
one can see that @xmath1 mean energy of 10 mev is reached already after 5 ns . then , below the debye energy , deceleration become very slow .
the lowest value of @xmath265 is determined by the intersection point of the cross sections of the acceleration processes and phonon creation process . in the case of a statistical mixture @xmath267 1.7 mev , for @xmath268
we have @xmath267 1.4 mev . finally , for pure para - d@xmath2 , with a contribution to the total cross section from the rotational transition @xmath269 three times greater than that shown in fig .
[ fig : xfddd22 ] , @xmath267 2.2 mev .
thus , @xmath1 atoms are never thermalized and their energy is significantly greater than 1 mev .
for para - d@xmath2 the mean energy is always greater than the energy of the lowest resonance peak @xmath270 1.6 mev .
however , even if @xmath265 is smaller than @xmath185 , a significant part of @xmath1 atoms has energy @xmath271 because of a large admixture of hot @xmath1 atoms at @xmath158 @xcite and slow deceleration below 10 mev . since at energies of a few mev
the lowest delta peaks are dominant in the resonant formation , their contributions to the mean effective formation rate are shown in fig .
[ fig : vesav ] for gas and solid deuterium ( stat ) targets , assuming steady maxwell distributions of @xmath253 energy , with different @xmath265 .
the maximum average rate of about 6 @xmath7s@xmath213 in the solid is due to the resonance energy shift of @xmath2721.8 mev .
the experimental result of 3 @xmath7s@xmath213 can be explained because @xmath265 is greater than 1 mev .
however , in order to obtain large fusion neutron and proton yields through resonant @xmath0 formation , the width @xmath273 of the resonance peaks in solid can not be too narrow .
the peak resonant rates of a few 10@xmath274s@xmath213 have been obtained assuming the discrete values of the rovibrational d@xmath2 energies in solid deuterium and @xmath275 mev .
these resonant rates are many orders of magnitude greater than the inelastic scattering rate @xmath14 10 @xmath7s@xmath213 .
in such a case @xmath1 atoms are very quickly ( compared to @xmath253 lifetime ) removed from the regions of resonance peaks and the contribution of the recoilless resonances to the neutron yield is negligible .
the monte carlo simulations have shown that the neutron yield from the phonon part of the resonant rates gives only some 10% of the yield observed in the experiments . in result ,
the calculated time spectra , obtained for the small @xmath273 , are dominated by weak non - resonant @xmath0 formation , which disagrees with the experimental data .
therefore , we have investigated influence of a broadening of the non - phonon resonant peaks , due to the presence of molecular rovibrational bands in solid , discussed in ref . @xcite . since in the literature
there is no information concerning the profile of such bands , we have assumed a rectangular shape of the resonance peaks .
the resonance strengths have been fixed and their widths have been changed in the limits 0.0011 mev .
it turns out that good monte carlo results are obtained for @xmath276 0.5 mev , which is consistent with the rotational bandwidths of about 1 mev reported in ref .
this gives the resonant formation rate of 294 @xmath7s@xmath213 for the recoilless peak in ortho - d@xmath2 , and respectively 214 @xmath7s@xmath213 in para - d@xmath2 . in fig .
[ fig : eddmdistat3 ] one sees the resonant formation rate at lowest energies for @xmath277 0.5 mev and for the statistical mixture of ortho- and para - states .
also shown is the monte carlo distribution of @xmath253 energy , calculated for times @xmath278 10 ns and @xmath278 30 ns . the maxwell distribution of initial @xmath1 energy , with @xmath279 1 ev , has been assumed .
two minima in the @xmath1 energy distribution appear quickly at the positions of the resonance peaks since the respective @xmath0 formation rates are comparable with the total inelastic scattering rate of about 30 @xmath7s@xmath213 . the @xmath3 fusion neutron spectrum , calculated assuming the same initial @xmath1 energy and resonance profiles , is shown in fig .
[ fig : npfita ] . a 3.2@xmath280 concentration of nitrogen is included in order to fit the triumf target conditions .
the solid line plotted in this figure has been calculated using the steady - state kinetics model with the effective formation rate @xmath281 3 @xmath7s@xmath213 and total spin - flip rate @xmath282 36 @xmath7s@xmath213 taken from the fits to the experimental data @xcite .
the slope of the spectrum at @xmath283 80 ns is determined by the rates @xmath284 , @xmath266 , and @xmath1 scattering rate which also changes the population of @xmath253 atoms in the vicinity of the resonant peaks .
the steady - state kinetics model does not include the process of @xmath1 deceleration .
therefore , fits using this model could entangle the deceleration rate with the formation and spin - flip rates .
the mean formation rate , calculated directly in the monte carlo runs , is a function of time , and it stays at the level of 13 @xmath7s@xmath213 . the spectrum slope at large times @xmath285 100 ns , when @xmath253 atoms practically disappear , are due to the nonresonant @xmath0 formation from @xmath192 and to the muon transfer to nitrogen contamination .
the shape of the time spectra practically does not change when the mean energy @xmath265 of the initial single maxwell distribution varies in the limits 0.011 ev . on the other hand , the spectra change strongly if a significant part of @xmath1 atoms at @xmath158 has energy smaller than the energy of the lowest resonant peak , which can be observed using a more complicated ( e.g. two - maxwell distribution ) . assuming that @xmath273 is greater than 0.5 mev we obtain results which begin to differ significantly from the analytical curve calculated with the experimental parameters .
in particular , the ratios of neutron yields from the short and large times begin to disagree .
fits of the calculated spectra to the experimental data would enable a better determining of @xmath273 and a shape of the initial @xmath1 energy .
however , this is not the purpose of this work .
a qualitative comparison of monte carlo spectra with the experimental data has already been performed in article @xcite . in this case
good fits were not obtained since at that time the resonant @xmath0 formation rates in solid d@xmath2 and @xmath1 scattering rates including coherent effects in the solid were not yet available .
our calculations show that strong resonant @xmath0 formation takes place both in ortho - d@xmath2 and para - d@xmath2 .
there are certain differences between the neutron time spectra from these targets ( see fig .
[ fig : neutronl ] ) , caused by the different positions and strengths of the lowest resonance peaks .
also @xmath1 slowing down process differs slightly in the two cases .
the neutron yield at larger times is smaller for ortho - d@xmath2 since in this case the resonance peak is placed at higher energy of 2.3 mev .
therefore , @xmath1 atoms are removed faster from the peak compared to the situation in para - d@xmath2 , where the resonance is observed at 1.6 mev .
a greater mean @xmath1 energy in para - d@xmath2 ( cf . fig . [ fig : avendm3 ] ) leads also to a stronger overlap of the resonance peak and @xmath1 energy distribution at @xmath286 ns .
however , the differences between the spectra can be clearly seen only in high - statistics experiments .
the methods used for description of resonant neutron and @xmath117-ray absorption in condensed matter have been directly applied for calculation of resonant @xmath0 formation and back - decay rates in condensed deuterium targets .
these rates are expressed in terms of the van hove single - particle function , which depends on properties of a given target . in particular , we have derived the analytical formulas for the rate in the case of resonant @xmath0 formation in a harmonic solid deuterium .
the calculations show great differences between resonant @xmath0 formation in 3 k solid deuterium and in 3 k d@xmath2 gas . in solid ,
the formation at a few mev , which determines the experimental results , is dominated by presence of the strong recoilless resonant peaks . on the other hand ,
the formation with simultaneous phonon creation is important above the debye energy .
the resonance profiles in the solid at higher energies are similar to that in d@xmath2 gas , but with the effective temperature equal to 40 k. this temperature is determined by the energy of zero - point vibration of d@xmath2 molecule in the lattice .
phonon creation is always important in the case of @xmath0 back decay because it is connected with energy release of a few tens mev , which is much greater than the debye energy .
a condition @xmath287 is fulfilled for any solid deuterium target at low pressure .
therefore , the parameters determining solid state effects ( debye - waller factor , mean energy of d@xmath2 vibration in solid ) weakly depend on target temperature @xmath44 .
they are expressed in terms of the debye energy @xmath264 which does not significantly change with the varying solid temperature @xmath44 . in result ,
the resonant @xmath0 formation rates in solid deuterium for different @xmath44 are very similar and one may expect that the average formation rates , derived from measurements performed at different temperatures , will also be very close .
this is confirmed by the results of experiments carried out at triumf and at jinr .
the structure of a solid deuterium target depends on its temperature and history .
targets maintained at @xmath288 4 k have the hcp structure @xcite .
though our calculations have been performed for fcc crystals , the obtained results are also good approximations of the resonant rates in hcp polycrystals since the debye temperature and nearest neighbor distance are similar for these two lattices . in general
, the formulas derived in this paper can be used in a wide range of target temperature and density , with appropriate experimental values of the debye temperature and lattice constant taken into account .
the monte carlo calculations show that @xmath1 deceleration below the debye energy is very slow and that mean energy of @xmath253 atom is always significantly greater than 1 mev .
the energy distribution of @xmath1 s during their lifetime is very broad ( at least a few mev ) , therefore a strong overlap of this distribution and lowest resonance peaks takes place , leading to a large mean @xmath0 formation rate in solid deuterium .
however , explanation of the experiments is possible only if the broadening of rovibrational molecular levels in solid is taken into account .
we obtained reasonable results assuming that the strengths of the recoilless resonant peaks are constant and that the rotational bands increase the resonance peak width to 0.5 mev .
note that , according to ref .
@xcite , high pressures lead to a greater broadening and even to a mixing of rotational states .
this could complicate a comparison of theory and high - pressure experiments .
the phonon part of the resonant rate give only about 10% contribution to the calculated neutron time spectra .
the @xmath3 fusion neutron spectra calculated for ortho - d@xmath2 and para - d@xmath2 solid targets are quite similar .
small differences between the spectra are due to the different energies and strengths of the lowest resonant peaks , and to a slightly higher mean @xmath1 energy in para - d@xmath2 . these differences can be clearly seen only in high - statistics experiments .
our calculations do not confirm a lack of strong resonant @xmath0 formation in solid ortho - d@xmath2 , predicted in the papers @xcite . in order to verify the theory it is necessary to perform measurements in pure ortho - d@xmath2 and para - d@xmath2 solid targets under the same conditions .
.the lowest resonance energies of @xmath0 formation in @xmath1 scattering from single d@xmath2 molecule ( @xmath30 ) and from 3 k solid deuterium target ( @xmath185 ) .
these energies are given in the respective cms systems . [ cols=">,<,>,<,^,^,^,^ " , ] | the rate of @xmath0 muonic molecule resonant formation in @xmath1 atom collision with a condensed deuterium target is expressed in terms of a single - particle response function . in particular , @xmath0 formation in solid deuterium at low pressures is considered .
numerical calculations of the rate in the case of fcc polycrystalline deuterium at 3 k have been performed using the isotropic debye model of solid .
it is shown that the energy - dependent @xmath0 formation rates in the solid differ strongly from those obtained for d@xmath2 gaseous targets , even at high @xmath1 kinetic energies .
monte carlo neutron spectra from @xmath3 fusion in @xmath0 molecules have been obtained for solid targets with different concentrations of ortho- and para - deuterium .
the recent experimental results performed in low pressure solid targets ( statistical mixture of ortho - d@xmath2 and para - d@xmath2 ) are explained by the presence of strong recoilless resonance peaks in the vicinity of 2 mev and very slow deceleration of @xmath1 atoms below 10 mev .
a good agreement between the calculated and experimental spectra is achieved when a broadening of d@xmath2 rovibrational levels in solid deuterium is taken into account .
it has been shown that resonant @xmath0 formation with simultaneous phonon creation in solid gives only about 10% contribution to the fusion neutron yield .
the neutron time spectra calculated for pure ortho - d@xmath2 and para - d@xmath2 targets are very similar .
a practically constant value of the mean @xmath0 formation rate , observed for different experimental conditions , is ascribed to the fact that all the recent measurements have been performed at temperatures @xmath4 19 k , much lower than the target debye temperature @xmath5 110 k. in result , the formation rate , obtained in the limit @xmath6 1 , depends weakly on the temperature . |
the internal motions of galaxies are largely set by dark matter , which outweighs baryonic matter by at least five to one .
a key goal of galaxy astrophysics is to relate the visible and dark mass in any given system , which to first order means determining the correlations between the structural parameters of galaxies ( e.g. @xmath2 , @xmath3 , @xmath4 and hubble type @xmath5 ) and those of dark matter haloes ( e.g. @xmath6 , @xmath7 , @xmath8 ) .
this programme requires extensive observation of the rotation and velocity dispersion profiles of galaxies and their baryonic mass distributions , combined with detailed dynamical modelling under a variety of assumptions about the galaxy halo connection .
traditional summaries of the relation between the baryonic mass distributions and internal motions of galaxies relate 1-point statistics of these functions , reducing the former to a total galaxy mass and size , and the latter to a single measure of velocity .
these are the tully
fisher , mass size and faber jackson relations , along with the fundamental plane .
more information , however , can be found in the full _ radial _ dependence of mass and velocity , and analysis of this dependence may be expected to afford not only a more stringent test of specific galaxy formation models , but also a richer foundation for a bottom - up determination of the galaxy halo connection . the local correlation of dark and visible matter is usefully described by the ratio of enclosed dynamical mass ( determined kinematically ) to enclosed baryonic mass ( determined photometrically ) . a proxy for this quantity is @xmath9 , to which it is equal in the case of spherical mass distribution .
@xmath9 is known as the `` mass discrepancy '' @xcite , and will be denoted @xmath10 hereafter . in general , @xmath10 may be any function of position @xmath11 within a galaxy , with parametric dependence on global galaxy properties @xmath12 ( e.g. @xmath2 , @xmath3 , @xmath4 , @xmath5 ) .
the full functional form of the mass discrepancy , @xmath13 , is a kinematic parameterization of the relation between dark matter and baryons , and _ a fortiori _ of the galaxy halo connection . while in principle @xmath10 may have an arbitrary dependence on position , its utility is significantly enhanced by the fact that it is known to correlate strongly with acceleration @xmath14 in all galaxies in which it has been measured in detail @xcite
this allows us to focus our attention on @xmath15 , which is known as the `` mass discrepancy acceleration relation '' or mdar .
the tightness of this relation and its inclusion of information across a galaxy s profile may be expected to make it among the most useful scaling relations for constraining models of the galaxy halo connection .
the value of the mdar relative to 1-point summaries , regardless of tightness , may be approximately established by a simple counting argument : while the latter are limited to one data point per galaxy , the former contains as many data points as one can measure across the entirety of a galaxy s rotation curve .
the aim of this paper is to construct a framework for using the mdar to test galaxy formation models , and hence deduce the dynamically - relevant correlations of the galaxy halo connection .
although the mdar has been known observationally for decades , few studies have sought to systematically extract the information contained within it .
the tightness of the relation is used by some to argue against all @xmath0cdm - based models of galaxy formation @xcite , where stochasticity in the galaxy halo connection may be expected to introduce a significant scatter into the relation between @xmath10 and @xmath16 . in addition
, it is argued that the presence of a `` characteristic acceleration '' @xmath17 at which @xmath10 consistently becomes @xmath18 ( indicating a dynamically - insignificant quantity of dark matter ) is not compatible with standard galaxy formation . a generic galaxy
halo connection would predict a spread in @xmath10 at high @xmath16 , and no clear transition in acceleration space between the dark matter and baryon - dominated regimes .
other authors , however , claim that the salient features of the mdar arise naturally in @xmath0cdm models which have been tuned to match the tully
fisher @xcite or @xmath19 , @xmath20 and @xmath21 relations @xcite , and consensus concerning the relation s significance does not seem at hand .
furthermore , no study to date has systematically investigated the dependence of @xmath10 on global galaxy properties ( @xmath12 ) at fixed @xmath16 , or quantified the dependence of mdar residuals on @xmath11 .
our specific task is twofold .
first , we create a set of statistics to quantify four significant features of the mdar : its shape , its scatter , the presence of a `` characteristic acceleration , '' and the correlation of its residuals with other variables ( @xmath2 , @xmath3 , @xmath4 , @xmath5 and @xmath11 ) .
this is motivated in part by the prevalence of purely qualitiative claims in the literature concerning the compatibility of the mdar with various models , from which it is difficult to determine the exact degree or nature of the agreement .
our statistics enable the conversion of verbal assertions into precise statistical comparisons , and we hope that they will sharpen future discussions of the mdar regardless of theoretical perspective .
the salience of these statistical features , however , is best appreciated in the context of specific model expectations .
our second task , therefore , is to develop a semi - empirical framework in @xmath0cdm to generate predictions for the mdar .
we adopt a fully bottom - up methodology , beginning with the simplest and best motivated correlations between galaxy and halo variables , and introducing more when required by the data .
an empirical approach is preferable to the use of a specific model in this case as it affords transparency concerning the correlations of galaxy and halo properties responsible for particular aspects of the predictions , provides model - independent information about the galaxy halo connection , and ensures tight enough control over the model systematics for statistical tests to be meaningful .
by comparing the predicted and observed mdars we will deduce the extent to which semi - empirical models are able to account for the significant features of the relation , and the concrete extensions to basic models that are required to match the relation s more detailed features .
we intend in this way to lay the groundwork for a phenomenological determination of the galaxy halo connection from information - rich relations such as the mdar , as well as formulating precise tests for specific models .
the starting point of our framework is the technique of halo abundance matching ( am ) , which imposes a nearly monotonic relationship between galaxy stellar mass and halo mass or velocity at a particular epoch @xcite .
mocks based on am have been shown to agree well with clustering and satellite fraction measurements @xcite , and , at least to first order , with 1-point galaxy scaling relations @xcite . from a phenomenological perspective , am specifies the relation between stellar mass and halo mass and concentration required to fit observations , but in its basic form neglects gas mass as well as galaxy size and type .
we therefore augment the model by allowing correlations of these variables with @xmath6 and @xmath7 at fixed @xmath2 .
for a given set of correlations constituting the galaxy halo connection , we generate a large number of mock data sets from our theoretical population with baryonic properties identical to the real data and halo properties specified by the model .
we then calculate the mdar statistics of these mock data sets , and evaluate the models by asking whether the observed values of the statistics could plausibly have been drawn from the mock data distributions .
the mdar is considered by some an important piece of evidence in favour of modified newtonian dynamics ( mond ) as an alternative to @xmath0cdm for solving the missing mass problem , and is a central relation in mond phenomenology ( see @xcite and references therein ) .
indeed , the founding papers of mond were the first to predict that @xmath10 would be more tightly correlated with acceleration than velocity or galactocentric radius @xcite , a hypothesis not verified empirically for many years . in mond , the mdar is a direct manifestation of the breakdown of newtonian gravity or mechanics at low acceleration @xmath22 , with the result that it is predicted to have no intrinsic scatter , residuals systematically uncorrelated with any other variable , and a clear acceleration scale @xmath23 . while the mond mdar must go to 1 at @xmath24 ( the newtonian regime ) and has a shape fixed by the theory at @xmath25 ( the deep - mond regime ) , the behaviour at intermediate @xmath16
is specified by an _
ad hoc _ interpolating function .
our statistical analysis will shed light on the compatibility of the observations with the mond hypothesis , and our comparison with @xmath0cdm models will bring into sharper focus the relative evidence accorded to mond and @xmath0cdm by the relation .
the structure of this paper is as follows . in section [ sec : data ]
we describe our observational mdar sample and the n - body simulations on which we build our theoretical framework . in section
[ sec : method ] we lay out our procedure for constructing the galaxy halo connection and document the mdar statistics with which we evaluate our models . in section [ sec : results ] we present our comparison of theory and data , first for a fiducial model best motivated by prior analyses , and then allowing variations in our model assumptions to maximise agreement with the mdar .
section [ sec : discussion ] discusses previous mdar studies in the context of our results , locates our parameter constraints and model requirements relative to previous findings , and elaborates on the broader implications for our understanding of galaxy astrophysics .
section [ sec : conclusion ] concludes .
we take the observational mdar from the sparc sample @xcite , a recent compilation of @xmath26 high - quality resolved rotation curves augmented with spitzer @xmath27 photometry .
this sample spans a wide range in all dynamically - relevant galaxy properties ( @xmath28 , @xmath29 , @xmath30 , and hubble type s0 to im / bcd ) , making it ideal for our analysis . following the suggestion of @xcite
, we remove galaxies with quality flag @xmath31 ( indicating major asymmetries , non - circular motions , and/or offsets between the stellar and hi distributions ) and those with inclination @xmath32 degrees , which may be subject to large systematic uncertainties in their rotation curves .
we take @xmath33 to account for molecular hydrogen , and adopt a stellar mass - to - light ratio of 0.5 @xcite .
the mdar of the final sample of 153 galaxies is shown in figure [ fig : fig1 ] .
our theoretical framework is based on dark matter haloes from a simulation in the darksky simulation suite @xcite .
the darksky suite assumes a flat @xmath0cdm cosmology with @xmath34 , @xmath35 , @xmath36 , and @xmath37 .
here we use the darksky-400 simulation ( ` ds14_i_4096 ` ) , a @xmath38 box with @xmath39 particles and mass resolution @xmath40 , run with the 2hot code @xcite .
this simulation was previously used in the studies of @xcite , @xcite , and @xcite .
haloes were identified using the rockstar halo finder @xcite , and a merger tree was constructed using the consistent trees code @xcite .
we construct a series of empirical models of increasing complexity for the effective galaxy halo connection of the sparc sample .
we begin with the standard am ansatz that the only galaxy property that correlates with halo mass and concentration is stellar mass .
we then consider the possibility that the sparc galaxies inhabit a systematically biased subset of the total halo population , the potential for correlation between halo concentration and galaxy size at fixed stellar mass , and finally the impact of disk formation on the halo density profile . to test the models we use them to generate a large number of mock mdar data sets with galaxy properties identical to the real data , but halo properties specified by the model .
for each mock data set , we calculate a series of statistics based on important features of the mdar , which we then compare to the values derived from the real data .
we describe our theoretical framework in section [ sec : model ] and our statistical analysis of the mdar in section [ sec : comparison ] . for quick reference ,
the free parameters of the framework are listed in table [ tab : params1 ] , and the mdar statistics in table [ tab : params2 ] . [ cols= "
< , < " , ] 1 . @xmath41 + where @xmath42 is the number of data points in the range @xmath43 .
this is the average @xmath10 in the first @xmath16 bin , .
the first and last bins therefore occupy the ranges @xmath44 and @xmath45 respectively . ] and describes the relative amount of dark and baryonic matter in low acceleration regions .
2 . @xmath46 + the corresponding statistic for the last @xmath16 bin .
this statistic and the above give an indication of the shape of the mdar .
3 . @xmath47 + this measures the spread in @xmath10 in the first @xmath16 bin and hence quantifies the effect of variations between galaxies at low acceleration .
4 . @xmath48 + the corresponding statistic for the last @xmath16 bin .
5 . @xmath49
+ where @xmath50 is the bin index and @xmath51 is the lower limit of bin @xmath50 .
this is the average variance in @xmath10 over all the @xmath16 bins , weighted by the number of points in each bin .
@xmath52 measures the overall scatter of the mdar .
6 . @xmath53 + where @xmath54 is the @xmath16 value at which galaxy @xmath55 crosses @xmath56 , in the data ( see below ) , and hence maximises the number of galaxies for which @xmath57 is defined .
raising the threshold increases @xmath58 and decreases @xmath59 , and vice versa .
moderate perturbations to the threshold value ( @xmath60 ) do not significantly affect the results . to calculate @xmath57 , we linearly interpolate between the two measured accelerations on either side of @xmath56 . ] and @xmath61 is the number of galaxies within this data set for which this happens ( see below ) .
@xmath54 is a proxy for the acceleration at which dark matter becomes dynamically negligible in galaxy @xmath55 , and hence @xmath62 indicates the `` characteristic acceleration scale '' of this data set .
note that in a given data set not all galaxies have some measured points at @xmath63 and some at @xmath64 ; for galaxies that do not @xmath57 is undefined .
the number of such galaxies is measured by @xmath59 and @xmath58 , below . 7 .
@xmath65 + the spread in @xmath57 values among galaxies measures the similarity of the `` acceleration scales '' of different galaxies within a data set .
the smaller @xmath66 , the more unique this scale and hence the more it may be said to be `` characteristic .
@xmath59 , the number of galaxies with @xmath63 across their rotation curve .
these galaxies do not possess a value for @xmath57 and are therefore excluded in the calculation of @xmath62 and @xmath66 above .
@xmath58 , the number of galaxies with @xmath64 across their rotation curve . as above . and
@xmath58 between mock and real data sets requires that the rotation curves be sampled at identical radii for each .
analysis at this level of detail could not therefore be performed using the method of @xcite or @xcite , for example . ]
10 . @xmath67 + where + @xmath68 + @xmath69 is defined in eq . 5
, @xmath70 labels points at low @xmath12 , and @xmath71 at high @xmath12 . for @xmath72
, @xmath70 corresponds to galaxies with @xmath12 in the bottom third of the sparc sample , and @xmath71 to galaxies in the upper third .
for @xmath73 , @xmath70 corresponds to the third of measurement points taken at lowest galactocentric radius ( separately for each galaxy ) , and @xmath71 to the corresponding upper third . for @xmath74
, @xmath70 denotes galaxies with hubble type s0sb ( 27 galaxies ) , and @xmath71 those with type sd
sdm ( 26 galaxies ) .
we weight by @xmath75 in order to prioritise the bins with the lowest shot noise , which contain many points from both subsamples .
+ these statistics measure the average vertical offset between the mdars at low and high @xmath12 .
an offset significantly different from 0 would imply that the mdar is dependent on @xmath12 , and _ a fortiori _ that @xmath10 is not a function of @xmath16 alone .
the values of @xmath12 we consider here are the dynamically - relevant galaxy properties that may be expected to influence the mdar in a generic theory of galaxy formation . in particular , by comparing the predicted and observed @xmath76 we can hope to get a handle on the following : the dynamical validity of the am stellar mass halo mass relation ( @xmath77 ) , the correlation of galaxy size with halo properties at fixed stellar mass ( @xmath78 ) , the validity of the assumption that the halo proxy in am is uncorrelated with cold gas mass at fixed stellar mass ( @xmath79 ) , the relation between galaxy morphology and halo properties ( @xmath74 ) , and finally the radius dependence of the halo density distribution ( @xmath73 ) . associated with each statistic for a given model is a two - tailed @xmath80-value equal to the fraction of mock data sets with a more extreme value for that statistic than the data .
we will say that a model fails to account for statistic @xmath55 ( and hence the associated feature of the mdar ) if a value of @xmath55 at least as extreme as the real data s has a @xmath81 probability of being randomly drawn from the mock data distribution .
we begin with a fiducial model in which @xmath82 , am scatter and @xmath83 are chosen in accordance with the results of previous am studies ( 0.6 , 0.16 and -0.5 respectively ; @xcite ) , and @xmath84 and @xmath85 are set to 0 .
this is the maximum - likelihood prior expectation for the description of the galaxy halo connection by am , assuming no dependence of galaxy size or type on halo properties .
the results are shown in tables [ tab : table1][tab : table3 ] ( second row ) and figs .
[ fig : fig2][fig : fig7 ] ( in red where models are compared ) .
the columns in tables [ tab : table1][tab : table3 ] corresponds to the statistics described in section [ sec : comparison ] ; those pertaining to mdar shape and scatter are shown in table [ tab : table1 ] , characteristic acceleration in table [ tab : table2 ] , and the correlation of mdar residuals in tables [ tab : table3 ] .
the first row lists the values in the real data , while subsequent rows contain the results for various models . in these cases , we give the modal average value of the statistic over all the monte carlo mock data sets , as well as the upper and lower limits enclosing 95 per cent of the results . if the value in the data lies outside these limits ( and assuming negligible systematic error ) , then the null hypothesis that the data was drawn from the model may be rejected at the 95 per cent confidence level , according to that statistic . fig .
[ fig : fig2 ] compares the real data in blue to a stack of 2000 monte carlo realisations of the theoretical galaxy population in magenta and a randomly - chosen example mock data set in green .
[ fig : fig3a ] shows the average mass discrepancy in each bin of acceleration , @xmath86 , averaged over all mock data sets ( thick line ) with the associated @xmath87 scatter between mock data sets ( thin lines ) ; fig .
[ fig : fig3b ] is the analogue for the standard deviations @xmath88 .
fig [ fig : fig4 ] compares the weighted average variance in @xmath10 , @xmath89 , for the mock and real data .
[ fig : fig5 ] is a normalised histogram of @xmath57 for the real galaxies ( red ) , and the mock galaxies stacked over all monte carlo realisations ( blue ) .
the insets show the distributions of @xmath59 and @xmath58 over the mock data sets , compared to the corresponding values for the real data . figs .
[ fig : fig6a ] and [ fig : fig6b ] plot the distributions of @xmath62 and @xmath66 respectively . finally , fig .
[ fig : fig7 ] shows the average offset @xmath76 in @xmath10 between subsets of the data split by @xmath2 , @xmath90 , @xmath91 , @xmath5 , and @xmath11 .
the results are as follows . 1 . a visual inspection of fig .
[ fig : fig2 ] reveals that the shape of the observed mdar roughly traces the model expectation .
@xmath10 is low at high @xmath16 ( where baryons dominate ) , and dark matter becomes increasingly important towards lower @xmath16 .
this result is known from the work of @xcite and @xcite .
however , it is also evident from fig .
[ fig : fig2 ] that both @xmath92 and @xmath93 are too high at low @xmath16 ; that is , there exists an excess of dark matter in low acceleration regions , in addition to excessive scatter between galaxies . in a given realisation of the model ,
several measured points would be expected at @xmath94 , while none are observed .
these results are confirmed in fig .
[ fig : fig3 ] , and the @xmath95 and @xmath96 columns of table [ tab : table1 ] , where it is shown that the discrepancy in the first @xmath16 bin is @xmath97 in @xmath92 and @xmath98 in @xmath93 .
the discrepancy in the weighted average variance , @xmath89 , is @xmath99 .
3 . a crossover occurs at @xmath100 , beyond which the simulated measurements lie below the observations in both @xmath92 and @xmath93 .
this shows that standard am ( with moderate halo expansion ) readily accounts for the baryon - domination of high - acceleration regions , and furthermore that variations in the dark matter content among galaxies at high @xmath16 are naturally small .
in fact , it is seen from the @xmath101 and @xmath102 columns of table [ tab : table1 ] that the predicted @xmath92 and @xmath93 are both @xmath103 too _ low _ in the @xmath104 bin , which implies at face value that high - acceleration regions ought to possess more dark matter than allowed for by our fiducial model .
although the quoted measurement uncertainties in the sparc data ( @xmath105 per cent in @xmath106 ) are too small to reconcile the measured values with the theoretical expectations , there exist additional uncertainties due to disc inclination , mass - to - light ratio , distance , and three dimensional baryon structure ( personal communication , f. lelli ) . taken together , these could account for the discrepancy in @xmath107 and @xmath108 , but would exacerbate the overprediction of @xmath109 and @xmath89 .
when the line traced by a galaxy in the @xmath110 plane crosses @xmath56 , it does so at an acceleration @xmath57 in the range @xmath111 .
the distribution of @xmath57 values over all galaxies in a mock data set is comparable to that of the real data , although typically centred on a slightly higher value ( figs .
[ fig : fig5a ] and [ fig : fig6a ] , and column 3 of table [ tab : table2 ] ) . remarkably , the spread in @xmath57 , @xmath66 , is actually _ lower _ in the model than in the data ( fig .
[ fig : fig6b ] and column 4 of table [ tab : table2 ] ) , indicating a _ greater _ degree of regularity in the acceleration beyond which dark matter becomes dynamically irrelevant .
these results imply that if the real galaxies are said to exhibit a `` characteristic acceleration , '' so too should the model galaxies .
however , only a fraction of the galaxies in each data set have points on either side of @xmath56 .
some have @xmath63 at all observed points along their rotation curve ( typically low surface brightness galaxies that are dark matter - dominated even near the centre ) and hence contribute to @xmath59 , while others have @xmath64 everywhere ( typically those with insufficient gas to probe the outer low-@xmath16 regions ) and hence contribute to @xmath58 .
we see from the inset of fig .
[ fig : fig5a ] , as well as columns 5 and 6 of table [ tab : table2 ] , that many more mock galaxies have @xmath63 than observed , but fewer have @xmath63 .
this is a consequence of the mismatch of the shape of the predicted and observed mdars , which is manifest also in the fact that @xmath112 is too high and @xmath107 too low .
the observed values of @xmath113 , @xmath114 and @xmath115 are compatible with the corresponding theoretical distributions ( figs .
[ fig : fig7a ] , [ fig : fig7c ] and [ fig : fig7d ] and @xmath95 , @xmath96 and @xmath102 columns of table [ tab : table3 ] ) .
this implies that the dependence of halo mass on stellar mass generated by am is adequate , and that the mdar does not favour a modification to stellar mass - based am to include gas .
that @xmath115 is consistent between theory and data suggests that at fixed stellar mass , s0 and sd galaxies do not live in haloes with significantly different mass or concentration .
is too large ) .
although statistically insignificant , this offset is in the same direction as the indications from weak lensing and satellite kinematics @xcite .
] we note , however , that the large spread of @xmath76 values in the mock data ( due to the excessive scatter in @xmath10 ) makes it difficult to draw definitive conclusions .. this statistic depends on the contingent details of the baryonic mass distributions and sampling radii of the sparc galaxies . the only way to acquire information concerning these aspects of the galaxy halo connection , therefore , is to compare mock and real data sets in which these distributions and radii are the same . ]
we caution also that these results are sensitive to the free parameters of the model , an issue to which we will return in section [ sec : varying_params ] .
@xmath116 is typically larger in the mock data than the real data ( fig .
[ fig : fig7b ] and @xmath101 column of table [ tab : table3 ] ) .
this implies that smaller galaxies at fixed stellar mass ( i.e. those of higher surface brightness ) live in more massive or more concentrated haloes than they have been assigned , and vice versa for larger galaxies . in section
[ sec : varying_params ] we will use a correlation between galaxy size and halo concentration to explore this issue further . 8 .
@xmath117 is smaller in the mock data than the real data by around @xmath118 ( fig .
[ fig : fig7e ] and @xmath119 column of table [ tab : table3 ] ) .
this discrepancy a relative of the well - known cusp / core problem for dwarf galaxies indicates that too much dark matter resides in the inner regions of galaxies in the model . in section
[ sec : varying_params ] we will use this result as further evidence for the hypothesis that the sparc galaxies must reside in a low - concentration subset of the total halo population if an am - based model is to be able to account for the mdar in detail . 9 .
the magnitudes of @xmath76 in the data are all small ( @xmath120 ) .
the average measurement uncertainty of 7 per cent in @xmath106 corresponds to a 14 per cent uncertainty in @xmath10 , which equates to @xmath121 when averaged over the entire mdar .
hence the systematic offset in the mdar as each @xmath12 is varied is small enough to be accounted for by measurement error alone , and @xmath10 is consistent with being a function only of @xmath16 . ) in the mdar for various model assumptions , compared to the value in the data .
the red , blue and green histograms correspond to the @xmath122 , @xmath119 and @xmath123 rows in tables [ tab : table1][tab : table3 ] respectively . only a strong selection on halo concentration
can reduce the predicted @xmath89 to acceptable levels.,scaledwidth=50.0% ] we now investigate the effect of variations in the model parameters , with a particular eye to mitigating the discrepancies noted above .
we first consider the possibility of reducing the predicted mdar scatter by decreasing the am scatter : the results of a model with am scatter = 0 are shown in the @xmath95 row of tables [ tab : table1][tab : table3 ] .
this has a negligible effect on the statistics characterising the shape and scatter of the relation , indicating that @xmath93 originates almost entirely from the spread in @xmath10 values generated by haloes of fixed proxy ( due to scatter in @xmath6 and @xmath7 at fixed @xmath124 ) , rather than the spread in proxy values at fixed @xmath2 .
that @xmath109 and @xmath89 are significantly overpredicted even without any scatter in the galaxy halo connection demonstrates the extraordinary tightness of the mdar .
since am scatter = 0 is in any case ruled out by clustering @xcite , however , we will not consider this model further . as in @xcite
, we find also that the effect of varying @xmath82 within the limits set by clustering measurements is small and does not significantly impact any of our statistics .
we therefore conclude that varying the am parameters can not substantially improve the mdar prediction .
we explore next the effect of modifications to the halo density profile due to disk formation , as parameterized by @xmath83 .
the excessive value of @xmath112 in the fiducial model , in addition to @xmath117 being too low , suggests that the degree of halo expansion corresponding to @xmath125 is insufficient .
we therefore show in fig . [
fig : fig5b ] and row 4 of tables [ tab : table1][tab : table3 ] the results of changing @xmath83 to @xmath126 , roughly the lowest value deemed plausible by @xcite .
this corresponds to more dark matter being expelled from the baryon - dominated inner regions of galaxies .
although @xmath112 , @xmath62 , @xmath59 and @xmath117 all move closer to their observed values , they remain discrepant with the data .
this suggests at least within the parameter space spanned by our implementation that feedback is unlikely to resolve the low-@xmath16 discrepancies . for reference
, we show in the @xmath96 row of tables [ tab : table1][tab : table3 ] the results of a model with standard adiabatic contraction ( @xmath127 ) . in accordance with many studies in the literature
, this clearly generates an excess of dynamical mass within the regions probed by kinematic measurements .
the scatter in the theoretical mdar may be more efficiently reduced by imposing a correlation between galaxy morphology and halo concentration , effectively situating the late - type sparc galaxies in haloes less concentrated than average . as described in section [ sec : model ] , this may be achieved by removing at each stellar mass some fraction @xmath84 of the highest concentration haloes before assigning haloes to the mock galaxies .
the blue lines in fig .
[ fig : fig3 ] show the mean mdar and its dispersion when the 80 per cent of highest concentration haloes at each stellar mass are removed from the theoretical parent population ( i.e. @xmath128 ; see also the @xmath102 row of tables [ tab : table1]-[tab : table3 ] ) .
this is found to be the approximate value required for the predicted @xmath112 and @xmath109 to be consistent with that observed . by reducing mass discrepancies across the mdar
, this modification to the effective host halo population also reduces @xmath62 and @xmath59 .
in addition , both @xmath113 and @xmath114 are increased .
this occurs because lower concentration galaxies have relatively more dark matter mass at large radius , while the rotation curves of galaxies with higher gas or stellar mass tend to be probed out to larger radii in the data .
hence the average @xmath10 of high @xmath2 and @xmath3 galaxies are increased relative to those with low values .
this effect corresponds also to a large increase in @xmath117 , bringing it into agreement with the observed value .
we discuss this model further in section [ sec : problems ] . as an alternative selection model
, we report in row 7 of tables [ tab : table1]-[tab : table3 ] the case in which each sparc galaxy is required to occupy a distinct halo that is not located inside the virial radius of a larger halo .
we find a moderate reduction in both @xmath92 and @xmath93 across the acceleration range , showing that subhaloes have larger @xmath10 on average .
changes to @xmath76 are small , and there is a negligible difference in statistics pertaining to the characteristic acceleration scale . in terms of its effect on the mdar ,
this is a middle ground between the fiducial model of section [ sec : fiducial ] and the @xmath128 model described above .
none of the model alterations examined so far have had a significant impact on @xmath116 ( table [ tab : table3 ] , column 4 ) , which has remained in all cases at least @xmath129 too high .
this indicates that larger galaxies at fixed stellar mass ought to reside in less massive haloes than have been assigned by the model , i.e. that @xmath4 as well as @xmath2 ought to correlate with halo @xmath7 and @xmath6 .
we anticipated this possibility with the size concentration correlation model described in section [ sec : model ] .
we now consider the case @xmath130 , and tune @xmath85 until the discrepancy in @xmath116 is removed .
we find @xmath131 for the observed value of @xmath116 to lie within the 95 per cent confidence interval of the theoretical prediction , and display the best - fitting case @xmath132 in figs .
[ fig : fig3][fig : fig7 ] and row 8 of tables [ tab : table1]-[tab : table3 ] .
this result suggests that in principle there is sufficient information in the mdar to determine the correlation of galaxy size with halo properties at fixed stellar mass .
this correlation has two further consequences .
first , since a galaxy of given size can no longer occupy a halo with any concentration , the spread in @xmath10 is lowered , leading to a reduction in both @xmath109 and @xmath89 .
these remain in disagreement with the observed values , however , even when the correlation between @xmath7 and @xmath90 is made very tight .
this suggests that the spread in @xmath6 alone at fixed @xmath2 , @xmath4 and @xmath7 overpredicts the mdar scatter .
a second consequence of setting @xmath133 is that @xmath114 is reduced ( fig .
[ fig : fig7c ] ) .
this is because @xmath4 and @xmath3 are correlated at fixed @xmath2 : larger galaxies have lower surface density and hence a lower star formation rate and greater ratio of @xmath3 to @xmath2 . reducing @xmath10 for larger galaxies
therefore reduces it also for more gas - rich galaxies , so that @xmath116 and @xmath114 shift together . in particular ,
when the predicted @xmath116 agrees with the observations , @xmath114 is @xmath134 too low .
although this discrepancy is not statistically significant , taken at face value it suggests that more gas - rich galaxies ought to be assigned more massive haloes than gas - poor galaxies at fixed @xmath2 and @xmath4 .
this could be achieved by abundance matching on total baryonic mass rather than stellar mass .
future analyses along these lines will determine whether this modification is necessary .
we discuss the @xmath135 correlation further in section [ sec : constraints ] .
there have been few theoretical studies of the mdar , and none to the level of detail presented here .
those that do discuss the relation either use it to argue that @xmath0cdm - based models must not be able to account for the detailed dynamical properties of galaxies @xcite , or , in reaction to such claims , that the relation does emerge from basic assumptions concerning galaxy formation @xcite . by means of detailed analysis of the statistical properties of the relation and its quantitative agreement with specific assumptions for the galaxy halo connection , we intend our work to break this impasse and achieve a non - binary appreciation of the relation s significance . in this section
, we discuss the mdar literature in light of our results
. a recurring theme will be the assertion that a given model `` can fit the mdar '' or `` can not in principle fit the mdar , '' where in fact the associated analysis warrants only the conclusion that it can or can not fit one particular mdar statistic and even that not established quantitatively .
a series of papers @xcite argue that the mdar poses a serious fine - tuning problem for all @xmath0cdm - based galaxy formation models .
in particular , it is claimed that neither the tightness of the observed mdar nor the bottoming out of @xmath10 at high @xmath16 can be accounted for .
we confirm that the small scatter of the mdar is a serious problem , although in our analysis the discrepancy is not as severe as suggested for example by fig . 3 of @xcite .
this is because @xcite considers as _ a priori _ plausible correlations between galaxy and halo variables that violate the assumptions of abundance matching ; we on the other hand contend that the success of am in matching to first order both clustering and kinematic statistics implies that the constraint it imposes on the galaxy halo connection must obtain in any plausible galaxy formation scenario .
that models inconsistent with the am stellar mass halo mass relation lie off the mdar need not argue for a fine tuning problem with @xmath0cdm .
they may instead be interpreted as a success of am .
further , we do not find low values of @xmath10 at high @xmath16 to be problematic ; indeed , given an amount of halo expansion compatible with the tully fisher and faber jackson relations , our models predict a _
@xmath92 and @xmath93 at @xmath136 than is found in the data .
we find in addition that the @xmath16 values at which @xmath10 becomes small ( in our case , @xmath64 ) are no less uniform in mock data than real data , and are clustered around a comparable central value .
this success derives from the stellar - to - halo mass fractions of am , combined with halo expansion and realistic galaxy sizes and baryonic mass profiles .
if @xmath62 and @xmath66 adequately capture the meaning of the terms `` acceleration scale '' or `` characteristic acceleration , '' we conclude that if present in the data these are present also in the mocks . that the acceleration scale of dark matter is around @xmath137 is perhaps not so surprising if we consider that the average dark matter density is an @xmath138 multiple of @xmath139 , which is proportional to @xmath140 . nevertheless , many of the claims of @xcite , @xcite and @xcite are borne out by our investigation .
not only are are @xmath92 , @xmath109 and @xmath89 naturally overpredicted , they remain so under highly conservative assumptions about galaxy formation , including no scatter in the galaxy halo connection and a significant reduction in the spread of halo concentrations associated with a galaxy of fixed @xmath2 and @xmath4 .
furthermore , the details of the transition region from baryon to dark matter domination ( as quantified by @xmath59 and @xmath58 ) are not understood , and @xmath10 is indeed compatible with being a function of @xmath16 alone ( @xmath141 ) .
our analysis therefore serves to refine and focus questions concerning the compatibility of the mdar with concordance cosmology .
@xcite are concerned in part to show that galaxy formation scenarios with correct phenomenology in other respects must fit the mdar .
they present a model with feedback parameters tuned to match the tully fisher relation , among others , and use it to generate 40 mock galaxies which they sample at regular intervals to produce a theoretical mdar .
they find that @xmath10 is a tighter function of @xmath16 than @xmath11 or @xmath142 , and that it tends to 1 at high @xmath16 .
they claim on the basis of these results that the model successfully exhibits a `` characteristic acceleration scale . ''
although our study qualitatively agrees with several results of @xcite , we believe their claim of success to be too strong such a conclusion could only be supported by a statistical analysis in which a mock data set with all the important features of the observed mdar is not extraordinary given the parent galaxy halo population .
in particular , claiming success for _ any _ reasonable model precludes using the relation to constrain models , which we believe we have shown to be both possible and fruitful .
@xcite examine the mdar of galaxies produced in several cdm and wdm hydrodynamical simulations , and conclude that the predicted and observed relations are incompatible .
this conclusion is supported by an application of the wilcoxon signed rank test , which quantifies the agreement of the ranks of @xmath10 in the observed and simulated data at given @xmath16 .
although this test clearly demonstrates that the expected mdar is on the whole normalised too high ( a conclusion corroborated here ) , we caution that it need not be considered as decisive as the authors claim : since @xmath10 is reduced to its rank , a simulated mdar consistently lying only slightly above the observations will appear as incompatible with them as one lying much higher , despite the fact that uncertainties in either data set may suffice to make them fully compatible according to any goodness - of - fit test that uses @xmath10 directly .
furthermore , the failure of individual simulations does not imply impossibility within the simulational framework .
we suspect in addition that @xcite use mond mass - to - light ratios , which have no justification in @xmath0cdm .
nevertheless , we concur that the scatter in the expected mdar is on the whole too high , and for the reasons given : there is a large spread in halo properties for given baryonic properties .
we have placed this discrepancy on firm numerical footing , and furthermore brought to light the conditions required for a quantitative solution . whether or not such conditions may be reasonably achieved by hydrodynamical simulations in @xmath0cdm remains to be seen .
@xcite construct a semi - empirical model for the mdar using average galaxy scaling relations from observations and halo scaling relations from n - body simulations .
although similar in spirit , our methodology improves upon that of @xcite in several ways : we use an am model that matches clustering measurements , we consider selection effects and correlations of halo variables with galaxy properties besides stellar mass , we give mock galaxies @xmath3 ,
@xmath4 , @xmath143 and @xmath144 values identical on a galaxy - by - galaxy basis with the data and sample their rotation curves at the same radii , we take the correlations between halo variables directly from a state - of - the - art n - body simulation rather than approximating them by power - laws with fixed log - normal scatter , and finally we make quantitative comparisons . while a confirmation of the qualitative results of @xcite with updated measurements is valuable , @xcite do not investigate the relation in sufficient detail to claim that all its pertinent aspects arise in @xmath0cdm . finally , @xcite compare the observed mdar to that of 22 galaxies generated by the magicc and clues hydrodynamical simulations .
although the shape and scatter of the relation can not be quantified with such a small sample , both seem to be in approximate agreement with the observations . with reference to analyses of this type ,
we argue that an empirical framework which offers complete transparency concerning its constituent elements provides a better handle on the aspects of the galaxy halo connection responsible for important features of the predicted mdar .
nevertheless , hydrodynamical simulations will produce correlations between galaxy and halo variables that lie outside our parameterizations , making a direct comparison with the mdar valuable and complementary to our approach .
we hope that future comparisons will utilise the statistics we have developed .
the mdar imposes several interesting constraints on am - based galaxy formation models in @xmath0cdm , and provides evidence concerning a number of important correlations of the galaxy halo connection .
we discuss these issues here .
first , in agreement with @xcite , @xcite , @xcite and @xcite , the observed @xmath10 values fit more comfortably with expanded than contracted haloes : a high value of @xmath83 ( such as that corresponding to standard adiabatic contraction ) degrades the fit for the majority of our statistics .
we find little sensitivity to the am parameters within the ranges allowed by clustering measurements , although more generally the tightness of the mdar argues for a small am scatter .
we find no evidence from @xmath113 for deviations from the am stellar mass halo mass relation , and at best weak evidence from @xmath114 that am based on total baryonic mass would perform better than the conventional stellar mass - based scheme . in addition , the agreement of @xmath115 with our fiducial model suggests that galaxies of different hubble type at fixed @xmath2 do not reside in haloes with markedly different dynamical properties .
we note however that the sparc sample does not include any true early - type galaxies , and hence we can not use this statistic to rule out the scenario in which the haloes of galaxies earlier than s0 differ substantially from those later .
the small offset between the mdars of galaxies with the same stellar mass but different size requires at the @xmath1 level that lower surface brightness galaxies be placed in less massive or less concentrated haloes than is allowed for by standard am .
we have constructed a simple toy model for such a relation , and found that the observed value of @xmath116 implies an average correlation @xmath145 , with @xmath131 ( 95 per cent confidence ) and best - fit value @xmath132 .
the weakness of this correlation lends justification to the assumption used in many semi - analytic and empirical models that galaxy size is uncorrelated with halo properies at fixed @xmath2 ( e.g. @xcite ) , and indeed models that make this assumption yield approximate agreement with observed @xmath146 correlations ( e.g. the tilt of the fundamental plane in the case of @xcite and @xcite ) while those that impose strong @xmath135 or @xmath147 correlations _ a priori _ do not ( e.g. the correlation of velocity and size residuals in @xcite ) .
we stress however that the correlation between galaxy size and halo properties has important dynamical consequences and no strong prior ( indeed , it is predicted by the celebrated @xcite model ) , and should therefore be explicitly considered in any galaxy formation model that seeks to explain galaxies @xmath148 scaling relations . in an upcoming paper ( desmond et al .
2016 , in prep ) , we will show that the eagle hydrodynamical simulation @xcite produces a @xmath135 correlation in approximate agreement with both mdar and tully fisher statistics , although the velocity and size residuals remain somewhat too strongly anticorrelated . remarkably , the `` characteristic acceleration '' at which @xmath10 becomes small is predicted to be no less uniform than that observed for all the models we consider , and hence the presence of this feature can not be used to constrain the theory .
we could not find any plausible choice for the galaxy halo connection that significantly breaks this agreement ( except perhaps the case of strong halo contraction ) . within the context of our framework ,
@xmath112 , @xmath109 , @xmath89 and @xmath117 may argue for the sparc galaxies occupying a subset of the total halo population highly biased in concentration or mass .
we discuss the plausibility of such a model further in the following section .
we have identified several aspects of the mdar that can not be matched or can be matched only with extreme difficulty in our framework .
foremost among these is the scatter , which we find to be too high in the theory both at low @xmath16 ( @xmath109 ) and in an averaged sense ( @xmath89 ) , regardless of the details of the am scheme , the halo response to disk formation , and the correlation between galaxy size and halo concentration at fixed stellar mass .
this is a particularly pressing problem because am imposes a tight galaxy
halo connection by construction , and there is no guarantee that a full galaxy formation theory would naturally produce a correlation between @xmath2 and @xmath124 with only @xmath149 dex scatter .
indeed , the mdar scatter is overpredicted even in the case of a perfectly monotonic galaxy halo connection , when am scatter is switched off .
in addition , we have neglected observational uncertainties from disc inclination , mass - to - light ratio , distance , and three dimensional baryon structure , which would increase the intrinsic tightness of the relation still further .
finally , we have been careful to ensure that there are no differences between the local baryonic properties of mock and real galaxies or the radii at which their rotation curves are sampled , both of which could artificially inflate the predicted scatter .
our conclusions concerning @xmath93 ought therefore to be highly conservative .
the unexpected regularity of galaxy kinematics has been noted previously in the context of the baryonic tully fisher relation , most recently for the sparc sample in @xcite .
a possible solution to this problem is to suppose that the sparc galaxies occupy a substantially biased subset of the full halo population .
we find that excluding the @xmath150 per cent of haloes with highest concentration at each stellar mass reduces @xmath109 and @xmath89 to acceptable levels , in addition to mitigating the offset of the @xmath10 values of low and high @xmath11 points , as measured by @xmath117 . in line with @xcite , it is tempting to interpret this as a correlation between morphology and host halo structure . that such a model may be fleshed out in detail , however , is doubtful , for four reasons : 1 .
the sparc galaxies span the range from s0 to im / bcd , a morphological subset that accounts for @xmath151 per cent of the total galaxy population ( e.g. @xcite ) .
2 . even supposing one could justify discarding 80 per cent of haloes
, it is implausible that these should be the exact haloes with highest concentration .
a non - monotonic selection function would require a further increase in @xmath84 to be effective .
when the fraction of haloes removed is constant over stellar mass , @xmath10 is also reduced for @xmath152 where it already lies below the observed values in the fiducial model . since the high - acceleration regime is mostly populated by high - mass galaxies
, this would suggest that the fraction of haloes removed ought to be lower at higher @xmath2 , yet the fraction of early - types rises with @xmath2 .
4 . as discussed in section [ sec : results ] , the agreement between the observed and predicted @xmath115 reveals that the average difference in the masses or mass distributions of haloes of galaxies with different morphology within the sparc sample can not be large . forcing galaxies of earlier type to inhabit more concentrated haloes would reduce @xmath115 , probably out of consistency with the data .
the second most significant challenge is the excess @xmath92 predicted at low @xmath16 .
this continues a long line of results indicating that the amount of dark matter in galaxies predicted by galaxy formation models applied to n - body simulations is on the whole too high , a problem which manifests itself in many and varied forms ( e.g. @xcite ) .
we note that while this discrepancy can be remedied in the case of galaxy 1-point scaling relations
the tully fisher and faber jackson relations by means of a global halo expansion parameterized in the same way as adiabatic contraction @xcite , the radius - dependent information in the mdar reveals that this method is inadequate in detail .
no value of @xmath83 in the range that correctly predicts the tully fisher normalisation , for example , can generate a value of @xmath112 as low as that observed .
low acceleration points come from low surface brightness galaxies ( lsbs ) and the outer regions of high surface brightness galaxies ( hsbs ) .
two unique features of the sparc sample are its high proportion of lsbs and its wide range of stellar mass ( down to @xmath153 ) .
tully fisher and faber jackson data sets , on the other hand , tend to be dominated by hsbs and extend no lower than @xmath154 .
the excess mass discrepancy at low @xmath16 is largely due to low mass lsbs : these galaxies are placed in haloes that are too massive .
there are hints that this would impact the baryonic tully fisher relation were it extended to low mass @xcite , and the discrepancy receives full expression in the too big to fail @xcite and cusp - core @xcite problems at the dwarf scale .
potential solutions to those problems apply here also , most prominently the possibility of strong feedback .
indeed , the low value of @xmath117 in the fiducial model ( indicating an excess of dark matter at low over high radii ) may also be evidence of feedback : a successful scheme would not only need to move dark matter outwards , but do so preferentially in low mass , low surface brightness galaxies .
if feedback is powered by star formation , it might be thought that such galaxies ought to have a _ smaller _ impact on their haloes , although some simulations ( e.g. @xcite ) do predict such an effect .
we stress however that simply reducing @xmath10 by moving dark matter outwards is not enough : the spread in @xmath10 must also be reduced , for example by homogenising the haloes .
it is unclear whether this is expected .
the discrepancy in @xmath112 may also be alleviated by modifying the galaxies stellar masses .
we have assumed that the stellar masses assigned by am are identical to those of the sparc sample , although in practice they have been derived using different assumptions for galaxy photometry , mass - to - light ratios and imf . since the am stellar - to - halo mass ratio peaks around @xmath155 ,
if the galaxies at the low - mass end of the sparc sample in fact correspond to galaxies of higher stellar mass in our model then the predicted mass discrepancies would be reduced at low @xmath16 .
this effect is degenerate with a modification to am which places lower mass galaxies in less massive haloes .
it is also possible that extrapolating the stellar and halo mass function below their resolution limits ( @xmath154 ) could introduce systematic errors at the low-@xmath16 end of the mdar .
if none of these modifications prove effective , non - standard dark matter interactions may be required ( e.g. @xcite ) .
four more mdar statistics can not be matched at the 95 per cent level within our framework : the high - acceleration average mass discrepancy and scatter ( @xmath107 and @xmath108 ) and the number of galaxies within a given data set with points on either side of @xmath156 ( @xmath59 and @xmath58 ) .
we doubt however that these discrepancies provide additional information on the galaxy halo connection : @xmath107 and @xmath108 likely point to additional measurement uncertainty , and correct prediction of @xmath59 and @xmath58 may well follow from success in matching @xmath92 at low @xmath16 .
since the mond prediction for the mdar has less freedom than that of @xmath0cdm , the relation offers less potential to constrain it ; on the flip - side , this makes the mond prediction easier to rule out .
we note three ways in which our analysis has relevance for mond . *
_ prima facie _ ,
@xmath157 at @xmath158 implies the existence of missing mass in the newtonian regime , violating a basic principle of mond .
since we have shown that this is equally unexpected for am , however , it is likely due to additional measurement error .
this may derive from uncertainties in disc inclination , mass - to - light ratio , distance , or the three dimensional distributions of the baryons . *
absent systematic uncertainty , @xmath10 depends _ only _ on @xmath16 in mond , and hence mdar residuals may not correlate with any other variable . in our analysis , this implies @xmath159 for each of @xmath160 .
as noted in section [ sec : fiducial ] , the magnitudes of each of these variables in the data are @xmath161 and hence compatible with 0 within the measurement uncertainty , supporting the hypothesis that @xmath10 is a function of @xmath16 alone .
this finding is made somewhat more significant by the fact that our model predictions for @xmath162 significantly exceed 1 in some cases .
* in a bayesian model comparison sense , an increase in the evidence achievable by @xmath0cdm models disfavours mond , and vice versa .
we have shown that the presence of a `` characteristic acceleration '' in the mdar can not discriminate against am mocks .
on the other hand , we have considerably strengthened the argument that the scatter in the relation is lower than predicted by standard galaxy formation , especially at low @xmath16 .
the mdar provides a map between the distribution of a galaxy s baryonic and dark matter , and therefore contains crucial information about the galaxy halo connection . in this paper
we have laid the groundwork for extracting this information for use in evaluating models of galaxy formation .
we have analysed the mdar using a set of 14 statistics that quantify its four most important features : its shape , its scatter , the presence of a `` characteristic acceleration '' beyond which mass discrepancy consistently goes to @xmath18 , and the correlation of its residuals with other galaxy properties .
in addition to using these statistics to focus discussion of the observed relation itself , we have engaged them to construct a data - driven framework for the galaxy halo connection .
building upwards from the simplest case of stellar mass - based abundance matching in @xmath0cdm , we have successively incorporated selection effects , a correlation between galaxy size and halo concentration , and a mass - dependent prescription for the impact of disk formation on halo density profiles . comparing to data from the sparc sample ,
our most significant findings are the following . * a basic am model readily accounts for several features of the mdar , including its approximate overall shape , its small scatter at high acceleration , and the independence of its residuals on stellar mass . in addition , we find no grounds for the claim that the acceleration at which the mass discrepancy becomes low is more uniform in the data than is expected in theory ( a `` characteristic acceleration scale '' ) . *
nevertheless , the predicted mdar has significantly too high a normalisation and scatter at low @xmath16 .
this remains true under highly conservative assumptions for galaxy formation , including no scatter in the galaxy halo connection and a significant reduction in the spread of halo concentrations associated with a galaxy of fixed @xmath2 and @xmath4 .
this indicates too much dark matter predicted in the outer regions of high surface brightness galaxies and ( especially ) in low surface brightness galaxies , and too large a variation among halo properties .
in addition , dark matter is more concentrated towards the centres of galaxies than the mdar suggests .
these discrepancies argue for halo expansion in response to disk formation , and a quantitative resolution requires also the exclusion of a large fraction ( @xmath163 per cent ) of the haloes with highest concentration at each stellar mass .
* the mdar may be used to to detect correlations of halo properties with ( at least ) three galaxy properties at fixed stellar mass : gas mass , disc size , and hubble type .
our analysis suggests at the @xmath164 level that galaxy size is anticorrelated with halo concentration at fixed stellar mass .
there exists in addition weak evidence in favour of basing am on total baryonic mass rather than stellar mass , but negligible direct evidence that the haloes of early - type galaxies have systematically different properties than those of later type .
we hope that this work will stimulate interest in the mdar as a source of information about the galaxy halo connection .
looking forward , we identify three ways in which further progress could be made .
first , additional models and assumptions need to be tested against the relation to the level of rigour achieved here .
it is unclear to what extent the outputs of many hydrodynamical simulations or semi - analytic models , for example , are consistent with mdar statistics .
specific implementations of the correlations of halo properties with galaxy variables besides @xmath2 ( e.g. am using total baryonic mass , or the angular momentum partition model of @xcite ) should be tested individually .
second , a firmer theoretical basis needs to be given for the empirical galaxy halo correlations argued for by the mdar ; this may be possible by mapping hydrodynamical simulations onto phenomenological frameworks such as ours .
finally , an increase in the size and precision of mdar data sets may be expected to yield a considerable gain in the constraining power of analyses of this type , pinning down the values of the statistics in the real data and reducing the widths of their distributions in the mock data .
the full power of the mdar likely remains to be harnessed .
i thank xavier hernandez , federico lelli , stacy mcgaugh and risa wechsler for fruitful discussions during the course of this work , and simon foreman for comments on the manuscript .
i am particularly grateful to federico lelli for access to and guidance with the sparc data .
this work made use of one of the dark sky simulations , which were produced using an incite 2014 allocation on the oak ridge leadership computing facility at oak ridge national laboratory .
i thank the dark sky collaboration for creating and providing access to this simulation , and sam skillman and yao - yuan mao for running rockstar and consistent trees on it , respectively .
additional computation was performed at slac national accelerator laboratory . | we use the mass discrepancy acceleration relation ( the correlation between the ratio of dark - to - visible mass and acceleration in galaxies ; mdar ) to test the galaxy halo connection .
we analyse the mdar using a set of 14 statistics which quantify its four most important features : its shape , its scatter , the presence of a `` characteristic acceleration scale , '' and the correlation of its residuals with other galaxy properties .
we construct an empirical framework for the galaxy halo connection in @xmath0cdm to generate predictions for these statistics , starting with conventional correlations ( halo abundance matching ; am ) and introducing more where required . comparing to the sparc data @xcite , we find : 1 ) the approximate shape of the mdar is readily reproduced by am , and there is no evidence that the acceleration at which dark matter becomes negligible has less spread in the data than in am mocks ; 2 ) even under conservative assumptions , am significantly overpredicts the scatter in the relation and its normalisation at low acceleration , and furthermore positions dark matter too close to galaxies centres on average ; 3 ) the mdar affords @xmath1 evidence for a correlation of surface brightness with halo mass or concentration .
our analysis lays the groundwork for a bottom - up determination of the galaxy halo connection from relations such as the mdar , provides concrete statistical tests for specific galaxy formation models , and brings into sharper focus the relative evidence accorded by galaxy kinematics to @xmath0cdm and modified gravity alternatives .
= 1 galaxies : formation - galaxies : fundamental parameters - galaxies : haloes - galaxies : kinematics and dynamics - dark matter . |
* high resolution x - ray diffraction * the mean value of _ _ d__@xmath4 could also be determined by high resolution x - ray diffraction ( hrxrd ) analysis . *
[ fig : figs1 ] * shows the @xmath17 - 2@xmath18-scan of the gan(0002)-reflex of samples a ( blue ) and sample b ( black ) .
both diffraction patterns exhibit clearly resolved satellite peaks up to the third order indicating a good layer periodicity . in order to assign the satellite peaks , a simulation using a dynamical diffraction theory
is performed and also shown in * fig .
[ fig : figs1 ] * for sample a. values of _ _ d__@xmath4 obtained by simulation of the diffractograms were 7.7 nm ( sample a ) and 2.6 nm ( sample b ) which is in excellent agreement with the values obtained by stem analysis and confirms a high degree of out - of - plane orientation as well as a good lateral homogeneity of the respective thicknesses . * numerical simulations * numerical simulations of the carrier confinement
were performed to estimate the influence of the free carrier concentration _
n _ on the internal electric fields in aln / gan ndsls .
we have used the software package nextnano@xmath23 to obtain self - consistent solutions of the 1d schrdinger - poisson equation within an effective mass model@xcite using material parameters from ref .
the simulated structure consists a 10-fold ndsl with _
_ d__@xmath4=4 nm and 4 nm thick aln barriers . *
[ fig : figs2 ] * shows the dependence of the calculated transition energies on the free carrier concentration ( bottom x - axis ) as extracted from the confined one - particle electron and hole ground - states .
black squares represent simulations assuming ideal pseudomorphic growth , i.e. including contributions from both spontaneous and piezoelectric polarization .
results represented by red circles neglect piezoelectric polarization ( piezoelectric constant set to zero ) , i.e. assume full strain relaxation in the nds to account for the formation of misfit dislocations that has been observed for nd with a thickness exceeding 3 nm . in both cases the simulated emission energies exhibit a sigmoidal shape and a significant blue shift of the emission energy
is found above _ _
n__=1 x 10@xmath14@xmath1 . at very high carrier concentrations of _ _
n__@xmath2810@xmath0@xmath1 the transition energies converge against the value of a field - free nd in agreement with results on 2d qws of ref .
@xcite . experimental transition energies ( green diamonds )
are plotted as a function of bep@xmath11 ( upper x - axis ) and fall between these two cases due to partial strain relaxation at the aln / gan interfaces @xcite . as the number of activated donors is not known , the correlation between bep@xmath11 and @xmath24 is adjusted to obtain good agreement between simulation and experiment .
linear interpolation of the corresponding ge - concentration according to ref.@xcite yields a ratio of ionized donors of 0.88 , 0.8 , 0.74 and 0.66 with increasing bep@xmath11 .
a background doping concentration of 10@xmath29@xmath1 has been assumed for the not intentionally doped ( n.i.d . ) sample . at high @xmath24
the simulated emission energies overestimate the experimental values , which is attributed to many body effects that have been neglected in the applied numerical model .
44ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1143/jjap.36.l459 [ * * , ( ) ] link:\doibase 10.1103/physrevb.62.16826 [ * * , ( ) ] link:\doibase 10.1143/jjap.43.l1524 [ * * , ( ) ] http://stacks.iop.org/0957-4484/21/i=19/a=195202 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/nl2006802 [ * * , ( ) ] link:\doibase 10.1063/1.4823550 [ * * , ( ) ] http://stacks.iop.org/0957-4484/22/i=27/a=275505 [ * * , ( ) ] link:\doibase 10.1021/nl303021v [ * * , ( ) ] http://www.sciencedirect.com/science/article/b6txf-42yw41x-3/2/db77913808abbc4facb4c9ca3ebb3a3b [ * * , ( ) ] link:\doibase 10.1063/1.2204836 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.3694674 [ * * , ( ) ] link:\doibase 10.1039/c3tc30556b [ * * , ( ) ] link:\doibase 10.1103/physrevb.68.125305 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.94.146102 [ * * , ( ) ] link:\doibase 10.1103/physrevb.75.045316 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.121305 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.205303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.58.r13371 [ * * , ( ) ] link:\doibase 10.1063/1.123851 [ * * , ( ) ] link:\doibase 10.1063/1.126831 [ * * , ( ) ] \doibase 10.1002/1521 - 396x(200112)188:2<851::aid - pssa851>3.0.co;2-b [ * * , ( ) ] \doibase 10.1002/1521 - 396x(200203)190:1<87::aid - pssa87>3.0.co;2-l [ * * , ( ) ] link:\doibase 10.1002/pssc.200776584 [ * * , ( ) ] link:\doibase 10.1063/1.1753085 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.2335400 [ * * , ( ) ] link:\doibase 10.1063/1.4761815 [ * * , ( ) ] link:\doibase 10.1063/1.1763976 [ * * , ( ) ] link:\doibase 10.1016/j.jcrysgro.2004.03.024 [ * * , ( ) ] link:\doibase 10.1063/1.3695172 [ * * , ( ) ] link:\doibase 10.1063/1.4848555 [ * * , ( ) ] link:\doibase 10.1063/1.4820264 [ * * , ( ) ] link:\doibase 10.1021/nl300840q [ * * , ( ) ] http://dx.doi.org/10.1039/c2nr32173d [ * * , ( ) ] link:\doibase 10.1002/pssc.200778479 [ * * , ( ) ] see section _ supplemental material _ for further information on hrxrd and numerical simulations .
link:\doibase 10.1103/physrevb.66.233311 [ * * , ( ) ] link:\doibase 10.1063/1.2980341 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.155320 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/24/11/115704 [ * * , ( ) ] the emission energy for _ _
d__@xmath4 = 1.6 nm was extracted from a spectrum recorded upon excitation with a pulsed nd : yag laser ( 266 nm ) .
see nextnano website [ http://www.wsi.tum.de/nextnano3 ] for tutorial input files and detailed documentation .
link:\doibase 10.1007/s11671 - 009 - 9347 - 1 [ * * , ( ) ] `` , '' ( , ) chap .
a convolution of two exponential decays with the same time - scale but opposing time - axes and an offset of 6.4ns was used for fitting .
this takes into account the so called back - sweep for decay - times longer than half of the repetition rate of the excitation pulses .
it has to be noted that 24ns is the upper detection limit of the used setup .
hence , this value should be interpreted as a lower limit and the actual decay times might be even higher .
link:\doibase 10.1063/1.1600519 [ * * , ( ) ] | we report on electrostatic screening of polarization - induced internal electric fields in aln / gan nanowire heterostructures with germanium - doped gan nanodiscs embedded between aln barriers .
the incorporation of germanium at concentrations above 10@xmath0@xmath1 shifts the photoluminescence emission energy of gan nanodiscs to higher energies accompanied by a decrease of the photoluminescence decay time . at the same time , the thickness - dependent shift in emission energy is significantly reduced . in spite of the high donor concentration a degradation of the photoluminescence properties is not observed .
reports on the self - assembled growth of gan nanowires ( nws ) by plasma - assisted molecular beam epitaxy ( pambe ) more than a decade ago @xcite kindled the interest in this promising technology platform for the realization of light emitters with improved efficiency @xcite .
more recently , applications such as photocatalytic water splitting @xcite or optochemical sensing have been proposed @xcite .
iii - n nws exhibit a much lower density of structural defects compared to planar thin films / structures due to their three - dimensional geometry and the corresponding efficient strain - relaxation @xcite . hence
, these structures are also considered to be a model system for the analysis and relation of basic structural and optical properties @xcite in addition to their huge application potential .
the realization of al@xmath2ga@xmath3n / gan nw heterostructures ( nwhs ) @xcite triggered the analysis of carrier confinement mechanisms in nwhs with embedded gan nanodiscs ( nds ) . besides the influence of the specific nw geometry via the so - called `` strain - confinement '' @xcite ,
the large polarization - induced internal electrical fields for al@xmath2ga@xmath3n barriers with high al - content exert strong influence on the optical properties of nwhs via the quantum - confined stark effect ( qcse ) . as a result
, the photoluminescence ( pl ) emission energy of nds embedded in al@xmath2ga@xmath3n barrier material has been shown to be shifted below the gan band gap for a sufficiently large al - content and disc thickness _ _
d__@xmath4 @xcite , as it has been reported before for planar al@xmath2ga@xmath3n / gan quantum well ( qw ) structures @xcite .
the possibility of screening the polarization - induced internal electric fields in planar al@xmath2ga@xmath3n / gan qws by doping has been theoretically investigated in ref.@xcite
. experimentally , screening by photoexcited carriers after excitation with different excitation powers
was observed in al@xmath2ga@xmath3n / gan qws @xcite as well as in aln / gan quantum dots ( qds ) @xcite .
si - doping of the active qd - region has been reported not to affect the internal electric fields in aln / gan qd layers @xcite most likely due to the limited doping efficiency of si and its deteriorating influence on the crystal quality when incorporated in high concentrations @xcite . in ref.@xcite
even a red shift of the pl emission of al@xmath5ga@xmath6n
/ gan qws due to si - doping of the qws was observed .
recently , efficient n - type doping of gan layers @xcite and gan nws @xcite using germanium as a donor was demonstrated and ge - concentrations above 10@xmath0@xmath1 could be achieved without structural degradation of the material , opening the possibility for screening of the qcse by free carriers .
+ in this work , we report on the electrostatic screening of internal electric fields in aln / gan nd superlattices ( ndsls ) by free carriers due to ge - doping of the gan nds .
an increasingly suppressed qcse with increasing ge - concentration is demonstrated by a combination of continuous - wave ( cw- ) and time - resolved ( tr-)pl analysis .
growth of n - polar iii - n nwhs was carried out by pambe using nitrogen - rich growth conditions on highly conductive
n - si(111 ) substrates @xcite . after growth of a 600 nm
nominally undoped gan nw base a ndsl structure containing 40 gan nds ( nd thickness _ _ d__@xmath4=1.67.7 nm ) embedded in aln barriers ( nominally 4 nm ) was grown and covered with a 20 nm gan cap .
the ga beam equivalent pressure ( bep@xmath7 ) was kept constant at 3x10@xmath8 mbar leading to a growth rate of 6.5nm / min .
the aln barriers were grown with a bep@xmath9 of 8x10@xmath10 mbar resulting in a growth rate of 2.8nm / min .
germanium was used as n - type dopant of the gan nds and the bep@xmath11 was varied between 5x10@xmath12 and 3x10@xmath13 mbar .
linear extrapolation of the data reported for gan nws in ref.@xcite implies ge - concentrations between 9x10@xmath14 and 6.2x10@xmath0@xmath1 in the nds .
however , as an identical doping efficiency of nws and nds for small values of _ _ d__@xmath4 can not be guaranteed we refer to the bep@xmath11 to describe the amount of incorporated ge in the following sections .
structural characterization was carried out by high - resolution transmission electron microscopy ( hrtem ) and aberration corrected high angle annular dark field ( haadf ) scanning transmission electron microscopy ( stem ) in a fei tecnai f20 field emission gun microscope operated at 200kv and a titan 60300 low base microscope operated at 300kv , respectively .
a panalytical xpert pro mrd setup was used for high resolution x - ray diffraction ( hrxrd ) experiments .
low temperature ( 4k ) cw - ensemble pl measurements were performed using a hecd laser operating at 325 nm , a jobin yvon thr 1000 spectrometer and a hamamatsu r375 photomultiplier .
the tr - pl analysis was carried out at 10k in a standard streak - camera setup @xcite with excitation by a frequency - tripled ti : sapphire laser at 276 nm ( 4.49ev ) , emitting 100fs pulses at a repetition rate of 78mhz .
the streak camera provides a time resolution of 10ps with an overall time window of 1.5ns .
the spectral resolution of the setup is about 5 nm . in * fig .
[ fig : fig1 ] * results of the stem analyses of two nwhs with ge - doped ( bep@xmath11=8.7x10@xmath12 mbar ) gan nds with different _ _
d__@xmath4 are displayed , revealing a mean thickness of the aln barrier of ( [email protected])nm in both samples and mean values of _ _ d__@xmath4=([email protected])nm ( ( [email protected])nm ) for sample a ( b ) , measured in the nwh center .
images of the entire nwhs are shown in * fig .
[ fig : fig1]a * and * d*. coalescence of adjacent nwhs and growth continuation as one nwh with an enlarged diameter is observed in numerous cases .
atomic resolution images of the nw top and middle sections are depicted in * figs .
[ fig : fig1]b , c , e , f*. the aln barriers as well as the gan : ge nds exhibit a truncated pyramidal shape introducing semipolar \{1@xmath1602 } facets .
no interdiffusion of ga and al was found in electron energy loss spectroscopy analyses , confirming the presence of sharp aln / gan(:ge ) interfaces ( not shown ) .
lateral growth during barrier deposition leads to the formation of a lateral aln shell and an increasing nd diameter in growth direction @xcite .
the shell thickness is less than 1 nm at the top and up to 10 nm at the gan nw base ( the nw outer facets are indicated by dashed lines in * figs .
[ fig : fig1]b , c , e , f * ) . similar
to the results reported in ref.@xcite the formation of misfit dislocations was observed when the nd height exceeded 3 nm , facilitating partial strain relaxation .
hrxrd analysis of the nwh ensembles revealed a high degree of out - of - plane orientation as well as a good lateral homogeneity of the ndsls . here ,
the @xmath17 - 2@xmath18-scan of the gan ( 0002)-reflex of samples a and b exhibits clearly resolved satellite peaks up to the third order . by simulation of the diffraction patterns values for _ _ d__@xmath4 of 7.7nm@xcite ( sample a ) and 2.6 nm ( sample b ) were obtained , in excellent agreement with the atomic resolution haadf stem results .
the results of the low - temperature cw - pl analysis are shown in * fig .
[ fig : fig2]a*. all spectra exhibit an emission line from the donor bound a exciton ( d@xmath19x@xmath20 ) emission originating from the gan base that is slightly blue shifted by up to 4mev with respect to the emission of unstrained gan ( 3.472ev @xcite ) due to compressive strain induced by the aln shell @xcite .
the energy of the emission band from the ndsl is shifted below the gan band gap as a consequence of the qcse due to polarization - induced internal electric fields @xcite . as displayed in the inset of * fig .
[ fig : fig2]a * the aln / gan : ge ndsl luminescence also red shifts with increasing _ _
d__@xmath4 for a constant bep@xmath11 of 8.7x10@xmath12 mbar .
+ the main panel of * fig .
[ fig : fig2]a * displays low temperature pl spectra of samples with nominally _ _
d__@xmath4=4 nm grown at different bep@xmath11 and a nwh sample with not intentionally doped ( n.i.d . ) gan nds for comparison .
an increase of the bep@xmath11 results in a blue shift of the ndsl emission by up to 0.6ev for bep@xmath11=1x10@xmath13 mbar ( cf .
inset of * fig .
[ fig : fig2]b * ) , strongly indicating a reduction of the qcse due to screening of the internal electric fields by free carriers from ionized ge donors @xcite . the spectral shape of the ndsl emission band remains almost unchanged besides an increase in its full width at half maximum ( fwhm ) from 250mev ( n.i.d . ) to 400mev ( ge - doped ) .
+ the influence of the ge - doping on the dependence of the ndsl emission energy on _
_ d__@xmath4 is summarized in * fig .
[ fig : fig2]b*. the n.i.d . samples ( black squares ) show a pronounced reduction of the ndsl - related emission energy from 3.7ev to 2.5ev for an increase of _ _ d__@xmath4 from 1.6 nm to 6.1nm@xcite which is in good agreement with the results obtained for nine - fold aln
/ gan nd structures from ref.@xcite ( green diamonds ) .
the red circles represent data obtained from aln / gan : ge nds grown with a bep@xmath11 of 8.7x10@xmath12 mbar . in that case
the red shift of the emission energy due to the qcse is notably attenuated , again indicating a strong reduction of the internal electric fields .
+ germanium has been shown to be an efficient donor in gan and the bep@xmath11 used for the growth of gan nds here correspond to ge - concentrations above 10@xmath0@xmath1 in gan nws @xcite .
for doping concentrations of 6x10@xmath21@xmath1 screening of internal electric fields by free carriers was not observed in si - doped gan qws with a thickness of _ _ d__@xmath22=3.4 nm embedded in undoped al@xmath5g@xmath6n barriers@xcite .
therefore , it can be concluded that noticeable screening of the internal electric fields requires a free carrier concentration above 10@xmath14@xmath1 . in order to estimate the onset of free carrier screening effects we performed 1d simulations of the carrier confinement in gan nds using the schrdinger - poisson - solver nextnano@xmath23@xcite .
the results indicate a significant modification of the triangular nd potential , i.e. the formation of an edge potential in the nd , and a resulting blue shift of the emission above a free carrier density of @xmath24 @xmath25 1 x 10@xmath14@xmath1 in agreement with the experimental data@xcite .
these findings also agree to the theoretical work on qws by riyopoulos et al .
@xcite stating that the impact of doping on the internal electric fields mainly depends on the ratio of _ _ d__@xmath22 and the debye length @xmath26 . for increasing carrier density the formation of an edge potential in the qw
was reported while the band profile is similar to the field - free case in the middle of the qw . at very high doping concentrations of @xmath24 @xmath25
10@xmath0@xmath1 the transition energies approach the value of a field - free qw above the gan band gap energy as the confined state exits the edge potential .
interestingly , the blue shift of the ndsl emission observed here saturates at a value of 180 mev below the gan band gap for bep@xmath11 @xmath25 1x10@xmath13 mbar ( cf .
inset in * fig .
[ fig : fig2]b * ) .
further increase of the ge - concentration does not result in a further blue shift of the emission energy .
this trend was also confirmed by pl analysis for different excitation powers . in * fig .
[ fig : fig3]a * the shift of the ndsl pl emission relative to its value obtained for 0.25@xmath27w excitation is displayed for different excitation intensities as a function of the bep@xmath11 . whereas the n.i.d .
nwhs show a strong blue shift of the ndsl emission energy with increasing excitation power due to screening of the internal electric fields by photo - excited carriers , no significant increase of the emission energy was observed for samples grown with a bep@xmath11 of 1x10@xmath13 mbar or higher .
we assign this saturation behavior and the discrepancy between the numerical simulations and the experimental values to many body effects@xcite that have been neglected in ref.@xcite and in our simulations .
+ in addition , the modifications of the nd confinement potential due to electrostatic screening by free carriers should result in an increasing oscillator strength and hence a decreasing pl lifetime . in order to analyze the influence of the ge - concentration on the pl decay time tr - pl experiments at different excitation powers
have been performed for samples grown with different bep@xmath11 , the results are depicted in * * fig .
[ fig : fig3]b**@xcite . a strong decrease of the pl decay time with increasing bep@xmath11 was found .
decay times of 24ns for the n.i.d . sample and of 6.3ns for the sample with a bep@xmath11 of 1.5x10@xmath13 mbar
were determined at a low excitation power , which is in reasonable agreement with ref.@xcite , reporting a pl decay time of 10ns for n.i.d .
nine - fold aln / gan ndsls ( _ _ d__@xmath4=3.5 nm ) and similar excitation conditions@xcite .
+ due to the additional screening by photo - generated carriers this effect becomes less pronounced with increasing excitation power , resulting in decay times of 7.9ns ( 2.6ns ) for the n.i.d . sample ( the sample with bep@xmath11=1.5x10@xmath13 mbar ) at an excitation power of 500@xmath27w .
this further confirms the assignment of the decreasing pl lifetime to electrostatic screening of the internal electric fields . at the same time the evolution of the decay times also excludes a degradation of the crystal quality and the formation of non - radiative recombination centers due to the high concentration of incorporated germanium .
in conclusion we have shown that polarization - induced internal electric fields in gan nds embedded in aln / gan nwhs can be efficiently screened by ge - doping of the nds .
the cw - pl and tr - pl analysis unambiguously reveal that for high ge - concentrations above 10@xmath0@xmath1 the oscillator strength in the nds is significantly increased compared to n.i.d .
structures . for the highest doped samples we found a saturation behavior of the pl emission energy below the gan band gap , indicating that due to the high density of free carriers many - body effects impact the optical transition energies .
in contrast to the case of silicon , doping with high concentrations of germaniun does not result in an enhanced accumulation of strain or the formation of structural defects .
consequently , high concentration of free carriers can be obtained in optically active areas .
this opens new possibilities for the design of complex nanostructures for optoelectronic applications such as intersubband devices .
authors from the justus - liebig - universitt acknowledge financial support within the loewe program of excellence of the federal state of hessen ( project initiative store - e ) . the research leading to these results has received funding from the european union seventh framework programme under grant agreement 312483 - esteem2 ( integrated infrastructure initiative
i3 ) .
the marburg group gratefully acknowledges funding through the research training group 1782 - functionalization of semiconductors by the german science foundation ( dfg ) . |
terminology in this note will follow @xcite . a matroid is _
internally @xmath1-connected _ if it is @xmath4-connected and , for every @xmath4-separation @xmath5 of @xmath0 , either @xmath6 or @xmath7 is a triangle or a triad of @xmath0 .
the purpose of this note is to prove the following technical result .
[ bigun ] let @xmath0 be a binary internally @xmath1-connected matroid in which every element is in exactly three triangles .
then @xmath0 has at least four elements @xmath2 such that @xmath8 is internally @xmath1-connected .
morever , if @xmath0 has fewer than six such elements , then these elements are in a @xmath1-element cocircuit .
this section introduces some basic material relating to matroid connectivity . for a matroid @xmath0 ,
let @xmath9 be the ground set of @xmath0 and @xmath10 be its rank function .
the _ connectivity function _
@xmath11 of @xmath0 is defined on all subsets @xmath6 of @xmath9 by @xmath12 .
equivalently , @xmath13 we will sometimes abbreviate @xmath11 as @xmath14 . for a positive integer @xmath15 , a subset @xmath6 or a partition @xmath16 of @xmath9
is _ @xmath15-separating _ if @xmath17 .
a @xmath15-separating partition @xmath16 of @xmath9 is a _ @xmath15-separation _ if @xmath18 .
if @xmath19 is an integer exceeding one , a matroid is _
@xmath19-connected _ if it has no @xmath15-separations for all @xmath20 .
let @xmath5 be a @xmath4-separation in a matroid @xmath0 . if @xmath21 , then we call @xmath22 , or @xmath5 a
_ @xmath23-violator _ since it certifies that @xmath0 is not internally @xmath1-connected .
for example , if @xmath6 is a _
@xmath1-fan _ , that is , a @xmath1-element set containing a triangle and a triad , then @xmath6 is a @xmath23-violator provided @xmath24 . in a matroid @xmath0 ,
a set @xmath25 is _ fully closed _ if it is closed in both @xmath0 and @xmath26 .
the _ full closure _ @xmath27 of a set @xmath28 in @xmath0 is the intersection of all fully closed sets containing @xmath28 .
the full closure of @xmath28 may be obtained by alternating between taking the closure and the coclosure until both operations leave the set unchanged .
let @xmath5 be a partition of @xmath29 .
if @xmath5 is @xmath15-separating in @xmath0 for some positive integer @xmath15 , and @xmath30 is an element of @xmath7 that is also in @xmath31 or @xmath32 , then it is well known and easily checked that @xmath33 is @xmath15-separating , and we say that we have _ moved _
@xmath30 into @xmath6 .
more generally , @xmath34 is @xmath15-separating in @xmath0 .
the following elementary result will be used repeatedly .
[ 3conn ] if @xmath0 is an internally @xmath1-connected binary matroid and @xmath35 , then @xmath36 is @xmath4-connected .
the result is easily checked if @xmath37 , so we may assume that @xmath38 .
since @xmath0 is @xmath4-connected and binary , @xmath39 and both @xmath40 and @xmath36 are @xmath41-connected . if @xmath42 , then @xmath0 is isomorphic to @xmath43 , or @xmath44 and again the result is easily checked .
thus we may assume that @xmath45 .
now let @xmath46 and suppose that @xmath47 has a @xmath41-separation @xmath5 .
we may assume that @xmath48 .
suppose @xmath49 .
then @xmath7 is a @xmath41-cocircuit @xmath50 of @xmath47 . as @xmath50 is not
a @xmath41-cocircuit of @xmath40 and @xmath0 is binary , we see that , in @xmath40 , either one or both of @xmath51 and @xmath52 is in a @xmath41-element parallel class .
thus we may assume that @xmath40 has @xmath53 as a circuit and @xmath54 as a cocircuit , or @xmath40 has @xmath53 and @xmath55 as circuits and has @xmath56 as a cocircuit .
hence @xmath0 has @xmath57 as a @xmath1-fan or has @xmath56 as both a circuit and a cocircuit . since @xmath45 , each possibility contradicts the fact that @xmath0 is internally @xmath1-connected .
we conclude that @xmath58 .
let @xmath59 be obtained from @xmath5 by adjoining each element of @xmath60 to the side of @xmath5 that contains an element parallel to it .
then @xmath61 and @xmath62 , so @xmath59 is a @xmath41-separation of @xmath40 . hence @xmath63 and @xmath64 are @xmath4-separations of @xmath0 . as @xmath65 and @xmath66 ,
this gives a contradiction .
let @xmath19 be an integer exceeding one . if @xmath0 is @xmath19-connected , an @xmath19-separation @xmath67 of @xmath0 is _ sequential _ if @xmath68 or @xmath69 is @xmath29 . in particular ,
when @xmath70 , there is an ordering @xmath71 of the elements of @xmath72 such that @xmath73 is @xmath19-separating for all @xmath74 in @xmath75 . when this occurs , the set @xmath72 is called _
we begin this section by noting two useful results .
[ dracula ] let @xmath0 be a matroid in which every element is in exactly three triangles .
then @xmath0 has exactly @xmath76 triangles .
consider the set of ordered pairs @xmath77 where @xmath78 and @xmath79 is a triangle of @xmath0 containing @xmath2 .
the number of such pairs is @xmath80 since each element is in exactly three triangles . as each triangle contains exactly three elements , this number is also three times the number of triangles of @xmath0 .
[ noodd ] let @xmath0 be an internally @xmath1-connected binary matroid in which every element is in exactly three triangles .
then @xmath0 has no cocircuits of odd size . for a cocircuit @xmath81 of @xmath0
, we construct an auxiliary graph @xmath82 as follows .
let @xmath81 be the vertex set of @xmath82 , and let @xmath83 be an edge exactly when @xmath84 and @xmath85 are members of @xmath81 that are contained in a triangle of @xmath0 .
since every element in is three triangles of @xmath0 , every vertex in @xmath82 has degree three by orthogonality and the fact that @xmath0 is binary .
hence @xmath86 , which equals the number of vertices of @xmath82 with odd degree , is even . to prove the next lemma
, we shall use the following theorem of qin and zhou @xcite .
[ qzthm ] let @xmath0 be an internally @xmath1-connected binary matroid with no minor isomorphic to any of @xmath87 , @xmath88 , @xmath89 , or @xmath90 .
then either @xmath0 is isomorphic to the cycle matroid of a planar graph , or @xmath0 is isomorphic to @xmath91 or @xmath92 .
[ smalltime ] let @xmath0 be an internally @xmath1-connected binary matroid in which every element is in exactly three triangles and @xmath93 .
then @xmath0 is isomorphic to @xmath91 or @xmath89 .
hence @xmath36 is internally @xmath1-connected for all elements @xmath2 of @xmath0 .
assume that @xmath0 is not isomorphic to @xmath91 or @xmath89 .
suppose first that @xmath0 has none of @xmath87 , @xmath88 , @xmath89 , or @xmath90 as a minor .
as @xmath92 has no triangles , it follows that @xmath0 is isomorphic to the cycle matroid of a planar graph @xmath82 .
as every edge of @xmath82 is in exactly three triangles , but @xmath94 is internally @xmath1-connected , every vertex has degree at least four .
hence @xmath95 .
moreover , by lemma [ noodd ] , every vertex of @xmath82 has even degree .
clearly @xmath96 .
moreover , @xmath97 , otherwise @xmath98 ; a contradiction . as @xmath99
, it follows that @xmath100 and @xmath101
. then @xmath82 is obtained from @xmath102 by deleting the edges of a perfect matching . but
no edge of this graph is in exactly three triangles .
we may now assume that @xmath0 has an @xmath103-minor for some @xmath103 in @xmath104 . by the splitter theorem for @xmath4-connected matroids
, there is a sequence @xmath105 of @xmath4-connected matroids such that @xmath106 and @xmath107 , while @xmath108 for all @xmath74 in @xmath109 . since @xmath110 and @xmath93
, it follows that @xmath111 .
suppose that some @xmath112 is obtained from its successor by contracting an element @xmath2 .
then @xmath40 has an @xmath103-minor .
but @xmath36 has at most nine elements .
thus @xmath113 and @xmath103 is @xmath87 or @xmath88 .
since @xmath36 must contain triangles , @xmath103 is @xmath88 .
now , by lemma [ noodd ] , every cocircuit of @xmath40 is even .
moreover , @xmath40 has exactly three 2-circuits .
the union of these three 2-circuits can not have rank two in @xmath40 otherwise @xmath0 has @xmath91 as a restriction but the remaining six elements of @xmath0 can not all be in exactly three triangles of @xmath0 .
let @xmath114 and @xmath115 be the three elements of @xmath88 that are in @xmath41-circuits in @xmath40 .
then one easily checks that there are two intersecting triangles of @xmath88 whose union contains exactly two elements of @xmath116 .
the cocircuit of @xmath40 whose complement is the union of the closure of these two triangles is odd ; a contradiction .
we now know that @xmath0 is an extension of @xmath103 by at most four elements .
let @xmath117 .
then @xmath118 so @xmath119 .
moreover , @xmath103 has at least @xmath120 triangles .
it is straightforward to check that the last number is positive , so @xmath103 can not be @xmath87 or @xmath90 .
thus @xmath103 is @xmath88 or @xmath89 .
each element of @xmath89 is in three triangles , so @xmath121 since each element of @xmath122 must be in a triangle with some element of @xmath89 ; a contradiction .
we deduce that @xmath123 .
now @xmath88 has exactly six triangles with each element being in precisely two triangles .
thus , in @xmath0 , there are six triangles each containing a single element of @xmath88 and two elements of @xmath122 . as @xmath124 , there are at most six triangles containing exactly two elements of @xmath122 .
we deduce that @xmath113 so @xmath0 can be obtained from @xmath125 by deleting exactly two elements . as @xmath125 has exactly seven triangles containing each element , deleting two elements
leaves each element in at least five triangles ; a contradiction .
in this section , we move towards proving the main result by dealing with 4-cocircuits and certain special 6-cocircuits in @xmath0 . throughout the section
, we will assume that @xmath0 is an internally @xmath1-connected binary matroid in which every element is in exactly three triangles , and @xmath126 .
[ no5 ] if @xmath81 is a @xmath1-element cocircuit of @xmath0 , then , for all @xmath2 in @xmath81 , the matroid @xmath36 is internally @xmath1-connected having no triads .
suppose that @xmath127 and @xmath36 is not internally @xmath1-connected . as @xmath0 is internally @xmath1-connected , @xmath128 .
as @xmath2 is in three triangles of @xmath0 , there are elements @xmath129 such that @xmath130 is a triangle for all @xmath74 . as @xmath131 is in three triangles for all @xmath74 , by orthogonality and the fact that @xmath0 is binary , there are elements @xmath132 such that @xmath133 , and @xmath134 are triangles .
this forces @xmath135 , and @xmath136 to be triangles , so @xmath137 is in no other triangle of @xmath0 for all @xmath74 .
let @xmath138 .
lemma [ 3conn ] implies that @xmath47 is @xmath4-connected .
the set @xmath139 forms an @xmath140-restriction in @xmath47 .
suppose @xmath47 has a non - sequential @xmath4-separation .
then we may assume that @xmath139 is contained in one side of the @xmath4-separation .
since @xmath141 is a circuit in @xmath40 , we may add @xmath142 , and @xmath143 to the side containing the @xmath140-restriction , and then add @xmath2 to get a @xmath23-violator of @xmath0 ; a contradiction .
we deduce that a @xmath23-violator of @xmath36 is a sequential @xmath4-separation .
we show next that [ notriads ] @xmath144 has no triads .
suppose @xmath144 has a triad @xmath145 .
then @xmath146 has @xmath145 as a cocircuit . by lemma [ noodd ]
, we may assume that @xmath147 or @xmath148 is a cocircuit of @xmath0 .
by orthogonality , in the first case , @xmath149 while , in the second case , @xmath150 . in the first case ,
let @xmath151 . then @xmath152 while @xmath153 , so @xmath154 , a contradiction as @xmath126 . in the second case
, @xmath0 has a @xmath1-cocircuit @xmath155 such that @xmath156 and @xmath157 .
apart from @xmath158 , the other triangles containing @xmath159 must meet @xmath160 in distinct elements and must meet @xmath161 in distinct elements .
thus @xmath162 and @xmath163 , so @xmath164 ; a contradiction since @xmath126 .
thus [ notriads ] holds . by [ notriads ]
, @xmath144 has no @xmath1-fans and so has no sequential @xmath4-separation that is a @xmath23-violator .
this contradiction completes the proof .
[ no6 ] take @xmath35 and the three triangles @xmath165 and @xmath166 containing @xmath2 . if @xmath167 is a cocircuit @xmath81 , then @xmath168 is internally @xmath1-connected for every element @xmath169 of @xmath81 .
let @xmath170 for each @xmath171 .
note that @xmath172 , and @xmath166 are not coplanar , otherwise their union forms an @xmath91-restriction , and @xmath81 contains a triangle ; a contradiction to the fact that @xmath0 is binary .
suppose the lemma fails .
then we may assume that @xmath173 is not internally @xmath1-connected .
as @xmath159 is in two triangles other than @xmath174 , orthogonality and the fact that @xmath0 is binary imply that each of these triangles contains an element of @xmath175 .
if @xmath176 and @xmath177 are each contained in a triangle , then the plane containing @xmath174 and @xmath178 is an @xmath91-restriction , so @xmath2 is in a fourth triangle ; a contradiction .
hence @xmath159 is in a single triangle with an element of @xmath179 and a single triangle with an element of @xmath180 . without loss of generality , @xmath181 and @xmath182 are triangles . by taking the symmetric difference of these triangles with the circuits @xmath183 and @xmath184 , respectively , we see that @xmath185 and @xmath186 are also triangles .
we have now identified all three of the triangles containing each element in @xmath187 .
but , for each element in @xmath175 , one of the triangles containing the element remains undetermined .
either @xmath188 and @xmath189 are triangles , or @xmath190 and @xmath191 are triangles . in each of these cases , we will obtain the contradiction that @xmath173 is internally @xmath1-connected . by lemma [ 3conn ] ,
@xmath192 is @xmath4-connected .
take @xmath67 to be a @xmath23-violator in @xmath47 .
let @xmath193 . clearly
the restriction of @xmath194 to @xmath6 is isomorphic to @xmath140 .
we may assume that @xmath195 where @xmath7 is @xmath196 or @xmath197 depending on whether @xmath198 or @xmath199 is a triangle of @xmath0 .
without loss of generality , we may also assume that @xmath25 spans @xmath6 in @xmath47 .
then @xmath200 is @xmath4-separating in @xmath47 and it follows that @xmath201 is 3-separating in @xmath0 . since @xmath0 has no @xmath23-violator , we deduce that @xmath72 is a sequential @xmath4-separating set in @xmath47 .
thus @xmath47 has a triad @xmath202 . by lemma [ noodd ]
, @xmath0 has a cocircuit @xmath155 where @xmath155 is @xmath203 or @xmath204 for some @xmath30 in @xmath7 . in the first case , by orthogonality , @xmath205 .
the last inclusion also follows by orthogonality in the second case since @xmath202 must meet @xmath6 and @xmath206 .
hence @xmath207 contains at least two cocircuits . since @xmath208
, it follows that @xmath209 ; a contradiction as @xmath126 .
[ seqsep ] let @xmath5 be an exact @xmath1-separation in @xmath0 with @xmath210 . if @xmath0 has no @xmath1-cocircuits , then @xmath6 is coindependent , @xmath211 , and @xmath212 .
if @xmath213 , then @xmath7 contains a basis of @xmath0 , and @xmath6 is coindependent . as @xmath214 , the rank of @xmath6 is at most three , and the result holds . if @xmath215 , then @xmath6 is independent , so @xmath216 . as @xmath217
, it follows that @xmath6 is a @xmath1-cocircuit ; a contradiction . beginning with @xmath7 ,
look at @xmath218 until the first time we get @xmath29 .
consider the set @xmath219 that occurs before @xmath29 in this sequence , let @xmath220 , and let @xmath2 be the last element that was added in taking the closure or coclosure that equals @xmath219 .
then either @xmath219 is a hyperplane and @xmath221 is a cocircuit , or @xmath219 is a cohyperplane and @xmath221 is a circuit .
suppose @xmath221 is a circuit .
as @xmath222 , we see that @xmath223 .
thus , as @xmath221 does not contain a @xmath1-cocircuit , it is coindependent , so it has size at most four .
we may assume that @xmath224 , otherwise the lemma holds .
suppose @xmath225 .
then both @xmath226 and @xmath59 are exact @xmath1-separations .
thus @xmath227 or @xmath228 .
the latter holds otherwise @xmath0 has a @xmath1-cocircuit ; a contradiction . but
@xmath219 is coclosed , so @xmath2 was added by coclosure ; that is , @xmath229 and we have a contradiction to orthogonality since @xmath230 .
it remains to consider the case when @xmath231 .
then @xmath232 .
the lemma holds if @xmath233 , so there is an element @xmath234 of @xmath235 that was added immediately before @xmath2 in the construction of @xmath219 .
now if @xmath234 is added via closure , then we can also add @xmath2 and @xmath221 via closure , so we violate our choice of @xmath219 . thus @xmath234 is added via coclosure so @xmath236 . hence @xmath0 has a @xmath1-cocircuit ; a contradiction .
we may now assume that @xmath221 is a cocircuit .
then @xmath221 has at least six elements .
as @xmath221 is @xmath1-separating , @xmath237 .
hence @xmath238 , so @xmath239 is a restriction of @xmath125 . as @xmath0 is binary , @xmath221 contains no triangle and no @xmath240-circuits , so @xmath239 is a restriction of @xmath241 .
as @xmath221 has six or eight elements , it follows that @xmath221 is a union of @xmath1-circuits so @xmath242 can not contain @xmath221 ; a contradiction .
[ seqok ] assume @xmath0 has no @xmath1-cocircuits .
if every exact @xmath1-separation in @xmath0 is sequential , then , for every element @xmath35 , the matroid @xmath36 is internally @xmath1-connected with no triads .
let @xmath130 be a triangle for all @xmath171 .
the matroid @xmath138 is @xmath4-connected by lemma [ 3conn ] .
let @xmath67 be a @xmath23-violator in @xmath47 . then @xmath243 .
add @xmath131 to the side of the @xmath4-separation containing @xmath137 for all @xmath171 to obtain @xmath244 , a @xmath4-separation in @xmath40 .
neither @xmath245 nor @xmath246 is a @xmath4-separation in @xmath0 . hence both are @xmath1-separations in @xmath0
thus , by hypothesis , each is a sequential @xmath1-separation in @xmath0 .
lemma [ seqsep ] implies that , without loss of generality , either @xmath247 is coindependent and has rank at most three in @xmath0 ; or both @xmath248 and @xmath249 have rank at most three and are contained in @xmath250 and @xmath251 , respectively . in the first case , as @xmath247
is contained in a plane , @xmath25 is contained in a triangle in @xmath3 ; a contradiction . in the second case , @xmath252 , so @xmath248 and @xmath249 span planes in @xmath125 .
these planes meet in a line , so @xmath253 .
hence @xmath254 ; a contradiction .
suppose @xmath144 has a triad @xmath116 .
then , by lemma [ no5 ] , @xmath0 has @xmath255 as a cocircuit , so we may assume that @xmath256 .
now @xmath0 has a triangle containing @xmath159 and exactly one of @xmath257 , or @xmath258 .
it follows that @xmath36 has a triangle meeting @xmath129 , so @xmath36 has a @xmath1-fan ; a contradiction . .
] the next three lemmas deal with a plane and a line in @xmath0 .
[ baddie ] suppose @xmath0 contains a plane @xmath259 and a line @xmath260 that are skew and are labelled as in figure [ pl ] where not every element in the figure must be in @xmath0 .
if @xmath261 , and @xmath262 are in @xmath0 , and @xmath263 and @xmath264 are cocircuits in @xmath0 , then @xmath265 is internally @xmath1-connected for all @xmath266 in @xmath267 . by symmetric difference , @xmath268 is a cocircuit . as @xmath262 is in three triangles of @xmath0
, orthogonality implies that @xmath262 is in a triangle with @xmath115 , say @xmath269 , and a triangle with @xmath234 , say @xmath270 .
likewise , @xmath169 is in triangles @xmath271 and @xmath272 , while @xmath30 is in triangles @xmath273 and @xmath274 , for some elements @xmath275 .
as @xmath259 and @xmath260 are skew , all of @xmath276 are distinct and none is in @xmath259 or @xmath260 . by symmetry , it suffices to show that @xmath277 is internally @xmath1-connected .
let @xmath278 .
let @xmath279 .
the restriction of @xmath47 to @xmath28 is isomorphic to @xmath88 .
suppose @xmath67 is a @xmath23-violator of @xmath47 .
without loss of generality , @xmath25 spans @xmath28 in @xmath47 .
thus @xmath25 spans @xmath280 .
hence @xmath281 is @xmath4-separating in @xmath282 , so @xmath283 is @xmath4-separating in @xmath0 .
thus @xmath72 is a sequential @xmath4-separating set in @xmath47 , so @xmath72 contains a triad @xmath145 .
thus either @xmath284 or @xmath285 is a cocircuit of @xmath0 for some @xmath286 in @xmath287 .
the first possibility gives a contradiction to orthogonality with @xmath273 .
thus @xmath288 , @xmath289 , or @xmath290 is a cocircuit .
suppose @xmath288 or @xmath289 is a cocircuit .
then orthogonality implies that @xmath145 contains @xmath291 or @xmath292 and so we get a contradiction to orthogonality with at least one of @xmath293 and @xmath274 .
thus @xmath290 is a cocircuit .
this cocircuit also contains @xmath169 so either contains @xmath30 and elements from each of @xmath294 and @xmath295 , or contains @xmath262 and elements from each of @xmath296 and @xmath297 .
each case gives a contradiction to orthogonality .
we conclude that @xmath277 is internally @xmath1-connected , so the lemma holds .
[ notfig ] assume @xmath0 has no @xmath1-cocircuits .
let @xmath67 be a non - sequential @xmath1-separation of @xmath0 where @xmath25 is closed and @xmath72 is contained in the union of a plane @xmath259 and a line @xmath260 of @xmath0 .
then either @xmath72 is @xmath298-cocircuit , or @xmath299 and @xmath300 .
moreover , @xmath301 is internally @xmath1-connected for at least six elements @xmath302 of @xmath72 . by lemma [ noodd ] , each cocircuit contained in @xmath72 has exactly six elements otherwise it contains a triangle .
suppose @xmath303 .
as @xmath304 , we know that @xmath72 is coindependent . hence it is contained in @xmath305 ; a contradiction . evidently @xmath306 .
we use figure [ pl ] as a guide for the points that may exist in @xmath72 .
we consider which positions are filled , keeping in mind that @xmath72 is the union of circuits and the union of cocircuits .
suppose @xmath72 has rank four and view @xmath72 as a restriction of @xmath307 .
then @xmath308 is a point of @xmath309 , so we may suppose @xmath310 .
furthermore , as @xmath304 , we know that @xmath72 contains , and therefore is , a cocircuit .
thus @xmath311 . as @xmath72 contains no triangles , @xmath312 , and @xmath313 .
thus @xmath314 .
without loss of generality , the points in @xmath72 are @xmath315 , and @xmath30 , and the result follows by lemma [ no6 ] provided @xmath35 .
we assume therefore that @xmath316 . we know
that @xmath317 . by orthogonality , without loss of generality , the three triangles of @xmath0 containing @xmath169 are @xmath318 , and @xmath319 .
thus @xmath0 has as triangles each of @xmath320 , and @xmath321 .
hence @xmath0 has no other triangles containing @xmath169 or @xmath30 .
thus the remaining triangles containing @xmath322 must be in @xmath259 , and so contain @xmath115 and @xmath323 .
but then @xmath116 and @xmath324 are triangles of @xmath0 , so @xmath325 is in four triangles ; a contradiction .
suppose that @xmath326 . then @xmath259 and @xmath260 are skew , and @xmath72 is the union of two @xmath298-cocircuits , @xmath81 and @xmath155 .
by orthogonality , each of @xmath81 and @xmath155 contains at most four elements of @xmath259 .
thus , by orthogonality , @xmath327 so @xmath328 . hence @xmath329 and @xmath299
. then , without loss of generality , each of @xmath81 and @xmath155 meets @xmath259 in four elements and @xmath260 in two elements .
the result now follows by lemma [ baddie ] .
[ nospike ] if @xmath0 has a @xmath298-element cocircuit @xmath330 where @xmath331 and @xmath332 are circuits , then @xmath168 is internally @xmath1-connected for all @xmath169 in @xmath81 . by symmetric difference , @xmath333 is also a circuit .
thus @xmath81 is the union of three disjoint pairs , @xmath334 , @xmath335 , and @xmath336 such that the union of any two of these pairs is a circuit . if one of these pairs is in a triangle with some element @xmath169 , then each of the pairs is in a triangle with @xmath169 and the lemma follows by lemma [ no6 ] .
thus we may assume that each of @xmath337 and @xmath338 is in a triangle .
hence so are @xmath339 and @xmath340 .
thus each of @xmath341 and @xmath323 is in exactly one triangle with an element of @xmath336 .
hence @xmath2 and @xmath234 can not both be in exactly three triangles ; a contradiction .
[ sixy ] let @xmath342 be an exact @xmath1-separation of @xmath0 such that @xmath343 is closed .
if @xmath344 , then @xmath345 is a @xmath298-cocircuit and @xmath346 is internally @xmath1-connected for all @xmath15 in @xmath345 .
we have @xmath347 and @xmath348 .
if @xmath349 , then @xmath345 is a cocircuit ; a contradiction .
thus @xmath350 .
since @xmath345 is a union of cocircuits each of which has even cardinality , it follows that @xmath351 .
hence @xmath345 is a @xmath298-cocircuit .
thus @xmath352 so @xmath345 contains two circuits such that they and their symmetric difference have even cardinality .
hence @xmath345 is the union of two 4-circuits that meet in exactly two elements and the result follows by lemma [ notfig ] .
the next lemma essentially completes the proof of theorem [ bigun ] .
[ no4 ] let @xmath0 be an internally @xmath1-connected binary matroid in which every element is in exactly three triangles .
suppose @xmath0 has no @xmath1-cocircuits .
then @xmath0 has at least six elements @xmath2 such that @xmath36 is internally @xmath1-connected . by lemma [ smalltime ] , we know that @xmath126 .
assume that the lemma fails .
by lemma [ seqok ] , @xmath0 has a non - sequential @xmath1-separation @xmath5 where @xmath6 is minimal
. then @xmath7 is fully closed . by lemma
[ sixy ] , @xmath353 and @xmath6 contains an element @xmath354 such that @xmath355 is not internally @xmath1-connected .
let @xmath356 be a triangle for all @xmath171 .
now @xmath357 is not internally @xmath1-connected . by lemma [ 3conn ] ,
it is @xmath4-connected .
take a @xmath23-violator @xmath244 in @xmath47
. then @xmath358 .
hence @xmath359 and @xmath360 exceed two .
add @xmath131 to the side containing @xmath137 for all @xmath171 to obtain @xmath361 .
then both @xmath362 and @xmath363 are exact @xmath1-separations of @xmath0 .
since @xmath364 and @xmath365 , we deduce that @xmath366 and @xmath367 .
moreover , by lemma [ seqsep ] , both @xmath362 and @xmath363 are non - sequential . without loss of generality
, we may assume that @xmath368 and , when equality holds , @xmath369 .
. then [ bigger ] @xmath371 , and , when equality holds , @xmath372 .
we show next that [ nonempty ] @xmath373 , and @xmath374 are all non - empty . as @xmath375 ,
the first set is not empty .
if the second is empty , then , as @xmath376 is in the closure of @xmath377 , we can move @xmath376 to @xmath7 to get @xmath378 as a non - sequential @xmath1-separation of @xmath0 ; a contradiction to our choice of @xmath5 .
if the third is empty , then @xmath379 , and @xmath380 contradicts our choice of @xmath5 .
likewise , if the fourth set is empty , then @xmath381 , and @xmath382 violates our choice of @xmath5 .
this completes our proof of [ nonempty ] . by submodularity of the connectivity function , @xmath383
we now break the rest of the argument into the following two cases , which we shall then consecutively eliminate . 1 .
@xmath384 and @xmath385 ; or 2 .
@xmath386 .
[ nope ] ( a ) does not hold .
suppose that ( a ) holds .
as @xmath0 is internally @xmath1-connected , @xmath374 is a triangle , or a triad , or contains at most two elements . clearly , this set is not a triad .
suppose @xmath387 .
then , by submodularity again , @xmath388 , so @xmath389
. then @xmath390 , so @xmath7 contains and so is a cocircuit . as this cocircuit can not contain a triangle , it follows that @xmath391 , so @xmath392 ; a contradiction . thus @xmath393 .
if @xmath394 , then @xmath395 is contained in a triangle , so @xmath72 is contained in the union of two lines ; a contradiction since @xmath72 contains a cocircuit that must have six elements and so contain a triangle .
we deduce that @xmath396 .
hence @xmath397 .
lemma [ seqsep ] implies that @xmath395 has rank at most three .
thus @xmath72 is contained in the union of a line @xmath260 and a plane @xmath259 .
it now follows by lemma [ sixy ] that [ nope ] holds .
next we show that [ nopeb ] ( b ) does not hold .
assume that ( b ) holds . since @xmath386 and @xmath398 is properly contained in @xmath6 , either @xmath399 , or @xmath400 .
it follows using lemma [ seqsep ] that @xmath401 .
thus , by [ bigger ] , @xmath402 .
if @xmath403 , then @xmath6 is contained in the union of a plane and a line .
then , arguing as in ( a ) , it follows that @xmath404 or @xmath405 and @xmath168 is internally @xmath1-connected for all @xmath169 in @xmath6 .
each alternative gives a contradiction .
thus , by [ bigger ] , @xmath406 and @xmath407 .
hence @xmath408 .
now view @xmath0 as a restriction of @xmath409 , where @xmath410 .
as @xmath5 is an exact @xmath1-separation , @xmath411 is a plane @xmath259 of @xmath309 .
because @xmath7 is fully closed , no element of @xmath6 is in @xmath259 .
it follows by orthogonality , since @xmath6 is a union of cocircuits of @xmath0 , that each triangle that meets an element of @xmath6 is either contained in @xmath6 or contains exactly two elements of @xmath6 with the third element being in @xmath259 .
we show that [ 5or6 ] @xmath412 .
suppose not .
then @xmath413 and @xmath414 .
so @xmath6 is contained in an @xmath241-restriction of @xmath0 . as @xmath6 is a cocircuit , @xmath415 or @xmath416 .
since @xmath417 and each is at least three , it follows that @xmath416 . to have a triangle meeting @xmath6
, there must be an element @xmath30 of @xmath7 in @xmath259 .
but @xmath30 is the tip of a binary spike in @xmath418 so it is in at least four triangles .
this contradiction proves [ 5or6 ] .
we show next that [ only5 ] @xmath419 .
. then @xmath420 . as @xmath421
, we deduce that @xmath422 , where we recall that @xmath409 and @xmath423 .
suppose @xmath424 meets @xmath259 .
as @xmath425 , we know that @xmath426 . then @xmath427 is a flat with rank at most @xmath428 .
hence its complement , which is contained in @xmath398 , contains a cocircuit . but
this cocircuit contains at least six elements by lemma [ no5 ] , so it contains a triangle in @xmath398 .
we deduce that @xmath424 avoids @xmath259 . by symmetry , so does @xmath429 .
it follows that each triangle that meets @xmath6 is either contained in @xmath398 or @xmath395 , or contains an element of each of @xmath430 , and @xmath259 .
if @xmath431 , then @xmath432-restriction , so each element in @xmath398 is in three triangles contained in @xmath398 .
then each element in @xmath395 is contained in three triangles in @xmath395 , so @xmath433 , and @xmath434 ; a contradiction to [ bigger ] .
thus @xmath435 and [ bigger ] implies that @xmath436 .
thus @xmath395 contains an element @xmath302 that is in at most one triangle in @xmath395 .
hence @xmath302 is in triangles @xmath437 and @xmath438 for some @xmath439 and @xmath440 in @xmath398 , and @xmath441 and @xmath442 in @xmath259 .
take @xmath443 in @xmath398 such that @xmath444 is a basis for @xmath398 .
then @xmath445 is a flat of rank at most @xmath428 whose complement , which is contained in @xmath395 , contains a cocircuit .
this cocircuit has at most five elements ; a contradiction to lemma [ no5 ] .
hence [ only5 ] holds .
we now know that @xmath419 .
it follows , since @xmath421 , that @xmath446 is a point @xmath447 of @xmath309 .
moreover , @xmath448 , so @xmath449 since @xmath450 , otherwise @xmath395 contains a cocircuit of @xmath0 that either has fewer than six elements or contains a triangle .
similarly , @xmath451 .
the following is an immediate consequence of the fact that @xmath25 is closed .
[ wheresp ] if @xmath452 , then @xmath453 .
let @xmath454 and @xmath455 .
neither @xmath456 nor @xmath286 is in @xmath6 , so @xmath457 hence @xmath458 .
recall that @xmath459 .
as @xmath460 , it follows that @xmath461 .
hence [ cap56 ] @xmath462 .
call a triangle of @xmath0 _ special _ if it contains an element of @xmath398 , an element of @xmath395 , and an element of @xmath259 . construct a bipartite graph @xmath463 with vertex classes @xmath398 and @xmath395 with @xmath464 being an edge , where @xmath465 and @xmath466 , precisely when @xmath467 is contained in a special triangle .
clearly @xmath468 next we show the following . [ star2 ]
every vertex @xmath169 of @xmath469 has its degree in @xmath470 .
let @xmath471 and take @xmath472 such that @xmath473 .
let @xmath474 be the element of @xmath475 .
thus @xmath476 .
clearly @xmath477 .
assume @xmath478 .
then @xmath479 contains @xmath480 , at least three distinct elements of @xmath481 , and @xmath474 .
thus @xmath479 contains @xmath481 .
hence @xmath482 contains at most five elements of @xmath0 ; a contradiction to the fact that every cocircuit of @xmath0 has at least six elements .
thus @xmath483 .
next suppose that @xmath484 .
then all three triangles containing @xmath169 are contained in @xmath485 . thus @xmath486 .
hence , for @xmath487 , the three triangles containing @xmath262 are contained in @xmath488 . thus @xmath489 .
hence @xmath490 contains a point of @xmath0 that is in six triangles ; a contradiction .
thus [ star2 ] holds .
now either 1 . @xmath491 ; or 2 .
@xmath492 , and @xmath447 are distinct .
suppose that ( i ) holds .
assume first that @xmath493 . by [ star2 ] , for @xmath494 , every element of @xmath495 is in a triangle contained in @xmath496 .
thus either @xmath497 and @xmath498 ; or @xmath499 and @xmath500 . since @xmath501 , we obtain a contradiction using ( [ eq ] ) .
thus @xmath502 .
as @xmath503 by [ cap56 ] , we see that @xmath504 , otherwise @xmath505 , and @xmath484 for every @xmath506 ; a contradiction to [ star2 ] .
we deduce that @xmath507 , and @xmath508 .
now @xmath447 is in two triangles in @xmath509 .
thus , of the three triangles in @xmath510 containing @xmath447 , at most one contains two elements of @xmath511 .
hence , using [ star2 ] , we see that @xmath512 comprises two triangles with a single element , not @xmath447 , in common .
thus @xmath513 ; a contradiction to equation [ eq ] .
we conclude that ( i ) does not hold .
we now know that @xmath492 , and @xmath447 are distinct .
we show next that [ wheresp2 ] @xmath452 .
suppose @xmath514 .
then @xmath504 so @xmath515 .
thus @xmath516 is five when @xmath517 and nine otherwise .
as @xmath518 for each @xmath466 by [ star2 ] , it follows that @xmath519 .
then @xmath520 is eight or seven depending on whether @xmath521 is @xmath522 or @xmath523 .
thus , by ( [ eq ] ) , we have a contradiction .
hence [ wheresp2 ] holds .
suppose @xmath524 .
then @xmath525 , otherwise there is an element of @xmath526 with degree zero in @xmath463 ; a contradiction to [ star2 ] . then @xmath527 .
suppose @xmath519 .
if the line through @xmath528 contains a third point of @xmath0 , say @xmath529 , then each of the other two lines through @xmath447 in @xmath424 contains at most one point of @xmath0 .
thus @xmath530 and , as @xmath531 , we see that @xmath532 is the unique triangle in @xmath512 containing @xmath529 . as
this triangle is special , it follows that @xmath533 ; a contradiction to [ star2 ] .
evidently the line through @xmath528 does not contain a third point of @xmath0 .
we deduce that @xmath512 comprises two triangles that have one element , not @xmath447 or @xmath286 , in common . then @xmath534 ; a contradiction .
we deduce that @xmath535 .
then exactly one of the lines in @xmath536 through @xmath447 contains exactly three points .
since no point of @xmath395 has degree three in @xmath463 , it follows that @xmath512 comprises two triangles with a point , not @xmath447 , in common . as @xmath537
, it follows that @xmath513 ; a contradiction .
we conclude that @xmath538 .
it remains to consider the case that @xmath504 and @xmath515 .
then @xmath516 is five or nine depending on whether or not @xmath456 is in @xmath7 . from [ wheresp2 ] , @xmath539 .
thus @xmath540 $ ] consists of two three - point lines meeting in a point @xmath262 . if @xmath541
, then @xmath520 is four or eight , depending on whether or not @xmath286 is in @xmath7 ; a contradiction .
hence @xmath542 .
thus the third element on the line containing @xmath528 is in @xmath6 . again
@xmath520 is seven , if @xmath535 , or four , if @xmath519 ; a contradiction to ( [ eq ] ) .
we conclude that [ nopeb ] holds and the lemma follows .
it is now straightforward to complete the proof of our main result .
if @xmath0 has a @xmath1-cocircuit , then the result follows by lemma [ no5 ] .
if @xmath0 has no @xmath1-cocircuits , then the theorem follows by lemma [ no4 ] .
it is natural to ask whether , for an internally @xmath1-connected binary matroid @xmath0 with every element in exactly three triangles , @xmath36 is internally @xmath1-connected for every element @xmath2 .
we now describe an example where this is not the case .
begin with @xmath543 having vertex classes @xmath544 and @xmath545 .
form the graph @xmath82 by adjoining three new vertices @xmath546 , @xmath302 , and @xmath266 , each adjacent to all of @xmath547 , and @xmath548 but not to each other . the vertex - edge incidence matrix of @xmath82 is the matrix @xmath549 shown below .
@xmath550 then @xmath94 is an internally @xmath1-connected matroid in which every element is in exactly three triangles .
now adjoin the matrix @xmath551 to @xmath549 where @xmath551 is shown below .
@xmath552 the matroid @xmath103 that is represented by @xmath553 $ ] has each element in @xmath94 in exactly three triangles , and each element of @xmath267 is in exactly two triangles . to see this , observe that @xmath554 .
moreover , no element of @xmath94 lies on a line with two elements of @xmath267 and it is straightforward to check that no element of @xmath267 is in a triangle with two elements of @xmath94 .
now take the generalized parallel connection of @xmath89 and @xmath103 across @xmath267 to get an internally @xmath1-connected binary matroid @xmath0 in which every element is in exactly three triangles .
evidently @xmath555 is not internally @xmath1-connected for all @xmath262 in @xmath267 . | let @xmath0 be an internally @xmath1-connected binary matroid with every element in three triangles .
then @xmath0 has at least four elements @xmath2 such that @xmath3 is internally @xmath1-connected . |
we drop the magnetic braking term in the integral on right - hand side of eqn .
( [ tpxb ] ) , as explained in the text , and obtain : @xmath116 where , @xmath117 defining @xmath118 and substituting @xmath119 in the above , we get @xmath120_{b_f}^{b_i } , \label{int2}\ ] ] where the indefinite integral @xmath57 is given by , @xmath121 standard expressions for integrals of type @xmath122 are given in gradshteyn and ryzhik ( 1980 ) , from which we get , @xmath123 @xmath124 \label{int4}\ ] ] @xmath125\ ] ] from equations ( [ tpxb1 ] ) , ( [ int1 ] ) , ( [ int2 ] ) , ( [ int4 ] ) , @xmath126_{x = b_f}^{x = b_i } \label{tpxb2}\ ] ] [ lastpage ] | we consider essential mechanisms for orbit - shrinkage or `` hardening '' of compact binaries in globular clusters to the point of roche - lobe contact and x - ray emission phase , focussing on the process of collisional hardening due to encounters between binaries and single stars in the cluster core .
the interplay between this kind of hardening and that due to emission of gravitational radiation produces a characteristic scaling of the orbit - shrinkage time with the single - star binary encounter rate @xmath0 in the cluster which we introduce , clarify , and explore .
we investigate possible effects of this scaling on populations of x - ray binaries in globular clusters within the framework of a simple `` toy '' scheme for describing the evolution of pre - x - ray binaries in globular clusters .
we find the expected qualitative trends sufficiently supported by data on x - ray binaries in galactic globular clusters to encourage us toward a more quantitative study .
[ firstpage ] globular clusters : general scattering stellar dynamics
binaries : close stars : low - mass stars : horizontal branch
x - rays : binaries introduction .
it is well - known that globular clusters contain far more than their fair share of compact x - ray binaries per unit stellar mass , compared to their host galaxies ( @xcite , @xcite ) .
the enhancement factor is @xmath1 in the milky way and m31 ( @xcite , @xcite ) , and possibly much higher in elliptical galaxies , as recent _ chandra _ observations have suggested ( @xcite , @xcite ) .
the origin of this overabundance of close binaries has been realized for some thirty years now to be the dynamical formation of such binaries through tidal capture and/or exchange interactions which can proceed at a very significant rate in dense cores of globular clusters ( henceforth gcs ) because of the high stellar - encounter rates there ( @xcite , @xcite , @xcite ) , but whose rate is negligible over the rest of the galaxy , where the stellar density is low by comparison .
the gc x - ray binaries that we shall be mainly concerned with in this work are those which are powered by accretion onto compact stars
. these can be either ( a ) low - mass x - ray binaries ( henceforth lmxbs ) , containing neutron stars accreting from low - mass companions , or , ( b ) cataclysmic variables ( henceforth cvs ) , containing white dwarfs accreting from low - mass companions .
accordingly , we shall not explicitly consider here binaries which contain either ( a ) two `` normal '' solar - mass stars , one or both of which are coronally active , or , ( b ) recycled neutron stars operating as rotation - powered millisecond pulsars , with a white - dwarf or a low - mass normal companion , although such binaries can be low - luminosity x - ray sources .
however , general considerations on the dynamical formation of close binaries do apply to these as well ; indeed , the latter binaries are now widely accepted as evolutionary products of lmxbs ( @xcite ) . in tidal - capture formation of a compact - star binary , the compact star ( neutron star / white dwarf ) passing close to a normal star dissipates its kinetic energy significantly by creating tidal deformation in the latter star , and so becomes bound to it . in the exchange process of formation , a compact star replaces one of the stars of an existing binary system of two normal stars during a dynamical encounter ( @xcite , @xcite , @xcite ) .
after formation in such an encounter , the compact - star binary continues to undergo stellar encounters in the dense cores of gcs , and it is on one particular effect of the continuing encounters that we focus in this paper . in the mid-1970s , it was realized that a major effect of the binary - single star encounters would be to extract energy from a given binary , making it more tightly bound or _
harder _ , and giving this energy to the motion of the single stars in the gc , thus `` heating '' the cluster ( @xcite , @xcite , @xcite ) .
we can call this effect _ collisional hardening _ of the compact - star binary , which makes the binary s orbit shrink at a rate higher than that which would obtain if it were not subject to the above stellar encounters , _
i.e. _ , if it were not in a gc .
the latter rate is believed to be determined by a combination of two processes , _ viz .
_ , ( a ) emission of gravitational radiation and ( b ) magnetic braking . in the former process ,
a compact binary emits gravitational radiation and so loses energy and angular momentum , which makes its orbit shrink @xcite .
the latter mechanism is envisaged as follows .
the low - mass companion to the compact star has a significant magnetic field , and also has its rotation tidally coupled or `` locked '' to that of orbital revolution .
the companion drives a wind , which carries away angular momentum at an enhanced rate because the magnetic field enforces co - rotation of the wind out to a radius considerably larger than that of the star , and this angular momentum ultimately comes from the orbit because of the above tidal locking , thus making the orbit shrink @xcite .
we discuss in this paper the relative roles of the above mechanisms for binary hardening , particularly the role of collisional hardening _ vis - - vis _ that due to gravitational radiation , indicate and clarify a scaling that naturally emerges from this interplay , and briefly suggest possible observational signatures of this scaling . in sec . 2
, we discuss the hardening of compact binaries by the three mechanisms discussed above , bringing out the particular role of collisional hardening .
we show that a characteristic scaling of the orbit - shrinkage time of such binaries with an essential gc parameter emerges because of the interplay between collisional hardening and that due to gravitational radiation . in sec . 3 , we explore possible observational signatures of this scaling .
we sketch a very simple `` toy '' scheme for describing the evolution of compact x - ray binaries in gcs , and indicate a possible signature of the above scaling within the bounds of this toy model .
we show that current data on x - ray binaries in gcs are consistent with this signature .
we discuss our results in sec . 4 , exploring possible lines of future inquiry .
hardening of compact binaries .
we consider the shrinking of the orbital radius of a compact binary by the three mechanisms introduced above .
consider gravitational radiation first .
the rate at which the radius @xmath2 of a binary decreases due to this process is given by ( see , _
e.g. _ , dinverno 1992 and references therein ) : @xmath3 in this equation , @xmath4 and @xmath5 are respectively the masses of the compact star and its companion in solar masses , @xmath6 is the total mass in the same units , and the orbital radius @xmath2 is expressed in units of solar radius .
we shall use these units throughout this paper . now consider magnetic braking .
the orbit shrinkage rate due to this process is given in the original verbunt - zwaan prescription @xcite as : @xmath7 where @xmath8 is the radius of the companion .
we give below further discussion on this mechanism .
finally consider the rate of orbit shrinkage due to collisional hardening , which is given by @xcite , @xmath9 here , @xmath10 is the mass of the single stars in the gc which are undergoing encounters with the binary . in this introductory work , we assume @xmath10 to be a constant , representing a suitable average value for a gc core , which we take to be @xmath11 .
the parameter @xmath0 is a measure of the encounter rate between a given binary and the background of single stars in the core of the gc : it is a crucial property of the gc for our purposes , so that we shall use it constantly in this work .
it scales as @xmath12 with the ( average ) core density @xmath13 of the gc , and the velocity dispersion @xmath14 of the stars in the core . following the convention often employed in the gc literature ( @xcite , @xcite ) ,
we can , in fact , _ define _ this parameter as : @xmath15 then the unit of @xmath0 is @xmath16 sec , corresponding to the units of @xmath13 and @xmath14 commonly used in the gc literature , namely , @xmath17 and km sec@xmath18 respectively . in these units ,
values of @xmath0 generally run in the range @xmath19 ( see below ) . in this work ,
we take the mass of the compact star to be @xmath20 and that of the companion @xmath21 , the latter corresponding to a typical average mass of stars in a gc ( see above ) .
according to the mass - radius relation for low - mass main - sequence stars , the radius of a main - sequence companion will then be @xmath22 .
furthermore , we consider only circular orbits in this work , returning to this point in sec .
4 .
the total rate of orbit shrinkage due to the combination of the above mechanisms is given by : @xmath23 we emphasize that the first two terms in eq .
( [ tot ] ) are always operational , irrespective of whether the binary is in a gc or not , and it is the relative effect of the third term , which represents the effects of the encounters in a gc core , that we wish to study here .
the interplay between the first and the third term was investigated in a pioneering study by @xcite , before the magnetic braking mechanism was postulated @xcite .
due to gravitational radiation , magnetic braking and collisional hardening , shown as functions of the binary separation @xmath2 .
also shown is the total shrinkage rate .
value of @xmath0 as indicated.__,width=340 ] note first that the three terms have different regions of dominance , as shown in fig .
[ dominance ] .
collisional hardening dominates at large values of the orbital separation @xmath2 , _ i.e. _ , for wide binaries , while hardening by gravitational radiation and magnetic braking dominates at small @xmath2 , _
i.e. _ , for narrow binaries . between the latter two ,
magnetic braking dominates at the smallest orbital separations , if we adopt the original verbunt - zwaan ( henceforth vz ) scaling for it ( see below ) .
the relative orbit shrinkage rate @xmath24 thus scales as @xmath2 at large orbit separations , passes through a minimum at a critical separation @xmath25 where the gravitational radiation shrinkage rate , scaling as @xmath26 , takes over from collisional hardening , and finally rises at very small separations as @xmath27 due to vz magnetic braking .
the change - over from gravitational radiation shrinkage to that due to magnetic braking occurs at a radius @xmath28 .
these two critical radii are easily obtained from eqs .
( [ gw ] ) , ( [ mb ] ) , and ( [ coll ] ) , and are given by @xmath29 note the scaling @xmath30 , which is crucial for much of our discussion here , as we shall see below .
the critical orbital separation @xmath25 varies in the range @xmath31 for the canonical range of values of the above gc parameter @xcite @xmath32 in the above units @xcite .
the relevance of this to close compact - star binaries in gcs is as follows .
when such a binary is formed , its orbital separation in most cases is such that the low - mass companion is not in roche lobe contact , since the roche - lobe radius has to be @xmath33 or less for this to happen for a typical low - mass main sequence or subgiant companion of mass @xmath34 ( see above ) .
mass transfer does not occur under such circumstances , so that such binaries are pre - lmxbs or pre - cvs , and we can call them by the general name _
pre - x - ray binaries _ , or pxbs for short .
it is the above orbit - shrinkage or hardening process that brings the companion into roche - lobe contact , so that mass transfer begins , and the pxb turns on as an x - ray binary ( lmxb or cv ) , or xb for short .
depending on the initial separation @xmath35 of the binary , some or all of the above processes can thus play significant roles in shrinking it to the point where mass transfer begins .
recent numerical simulations suggest that tidal - capture binaries are born with orbital radii ( or semi - major axes ) in the range @xmath36 for main - sequence ( henceforth ms ) or early subgiant companions , and in the range @xmath37 for horizontal - branch companions @xcite .
for binaries formed by exchange encounters , the orbital radii are generally expected to be somewhat larger than those for corresponding tidal binaries with identical members .
thus , both collisional hardening and gravitational radiation are expected to play major roles in the orbital shrinkage to roche - lobe contact for most of the pxbs in gcs , whether dynamically formed or primordial .
although we have included magnetic braking as above for completeness , its role in hardening of pxbs into xbs appears to be rather insignificant , at least for the vz scaling adopted above .
this is evident from the fact that there is little change in any of the results described here whether magnetic braking is included or not . for the vz
scaling , this is easy to understand . with a steep increase at small @xmath2 ,
this effect is significant only at very small orbital separation , when the pxb has already come into roche - lobe contact and become an xb .
thus , this process may well be significant in the further orbital evolution of the xb as mass - transfer proceeds ( @xcite , @xcite ) , but not in the pxb - hardening process under study here .
actually , further study of magnetic braking since the original vz formulation has revealed many interesting points .
the nature and strength of this effect very likely depends on the mass and evolutionary state of the companion star .
for example , magnetic braking may become totally ineffective for very low - mass companions with @xmath38 or less , as these stars are fully convective .
whereas a significant convective envelope is necessary for strong magnetic braking , `` anchoring '' of the magnetic field in a radiative core is also believed to be essential for it , and it is argued that the effect would basically vanish when the star becomes fully convective ( @xcite , @xcite ) . indeed , this forms the basis for the standard explanation for the _ period gap _ in cvs ( @xcite and references therein ) .
further , studies of the rotation periods of stars in open clusters have suggested that magnetic braking may be less effective than that given by the vz prescription : this has been modelled in recent literature by either ( a ) the vz scaling as above , but a smaller numerical constant than that given above , or , ( b ) a `` saturation '' effect below a critical value of @xmath2 , wherein the scaling changes from the vz @xmath39 scaling of eq . ( [ mb ] ) to a much slower @xmath40 scaling below this critical @xmath2-value ( @xcite ) .
while these modifications are of relevance to xb evolution , it does not appear that they can alter the pxb - hardening results described here in any significant way .
accordingly , we shall not discuss magnetic braking any further here , and keep this term in the complete equations only to remind ourselves that it is operational , in principle , for companions with @xmath41 . an interesting scaling interplay between collisional hardening and gravitational - radiation hardening near the above critical orbital separation @xmath25 produces a characteristic scaling , which we now describe .
consider the shrinkage time @xmath42 of a pxb from an initial orbital separation @xmath35 to the final separation @xmath43 corresponding to roche - lobe contact and the onset of mass transfer , given by : @xmath44 for given values of stellar masses , @xmath42 scales with the gc parameter @xmath0 introduced above as @xmath45 the scaling is almost exact at high values of @xmath0 , _
i.e. _ , @xmath46 , say , there being a slight fall - off from this scaling at low @xmath0 s . in eq .
( [ tpxb ] ) shown as function of orbital separation @xmath2 , with values of @xmath0 as indicated.__,width=340 ] how does this scaling arise ? to see this , consider first the qualitative features of the integrand on the right - hand side of eq .
( [ tpxb ] ) , _ i.e. _ , the reciprocal of the total shrinkage rate at an orbital separation @xmath2 , which we denote by @xmath47 , and which is displayed in fig .
[ peak ] .
it is sharply peaked at @xmath48 : indeed , the peak would be _ exactly _ at the above critical separation @xmath25 but for the effects of magnetic braking , as can be readily verified .
since the latter effects are not important in the range of @xmath2-values relevant for this problem , as explained above , we can get a good estimate of the actual result by considering only gravitational radiation and collisional hardening .
because of this dominant , sharp peak in @xmath47 , most of the contribution to the integral , _
i.e. _ , to @xmath42 , comes from there , provided that the integration limits ( @xmath49 ) are such that _ all or most of the peak is included_. we assume for the moment that this is so , and return to a discussion in the next subsection of what happens when this condition fails . under the above circumstances
, we can immediately give a rough estimate of @xmath42 , which is the area under the curve in fig .
[ peak ] , as @xmath50(maximum value of the above integrand ) .
this maximum value is simply @xmath51 if we neglect magnetic braking , since the gravitational radiation term equals the collisional hardening term there , as explained above .
this gives @xmath52 , which , with the aid of eq .
( [ acrit ] ) , yields @xmath53 .
this is the basic reason for the scaling given by eq .
( [ scaling ] ) .
an exact evaluation of the integral in eq .
( [ tpxb ] ) , with the magnetic braking term neglected , confirms this , as expected and as detailed in appendix a. the exact result is : @xmath54 .
\label{exact}\ ] ] here , @xmath55 is a dimensionless orbital separation defined by @xmath56 , and the integral @xmath57 is given in appendix a. as @xmath57 has only a logarithmic dependence on @xmath58 under these circumstances , the basic scaling is @xmath59 , as above .
it is this basic scaling that leads to the essential behavior of the shrinkage time @xmath42 discussed in this paper .
breakdown of scaling ? when would the above scaling break down , and why ?
a simple answer is clear from fig .
[ peak ] : this would happen when the integration limits ( @xmath49 ) are such that all or most of the above peak in @xmath47 is _ not _ included .
for the present problem , this basically reduces to an upper bound on @xmath43 , since @xmath35 is normally large enough to ensure that the region of integration in fig .
[ peak ] extends well into considerably larger values of @xmath2 beyond the peak .
when @xmath43 becomes so large as to exceed @xmath25 , the region of integration is severely curtailed from the left in fig .
[ peak ] , so that most of the peak s contribution is missed , and the above scaling breaks down .
we might think that such a situation would arise when the low - mass companion in the pxb is an evolved , horizontal - branch star , which has a much larger radius than a ms / subgiant companion of the same mass , and so would be expected to come into roche - lobe contact at a much larger value of @xmath60 , say @xmath61 , and correspondingly larger values of @xmath43 .
but such binaries are not relevant to our discussion here , since the lifetimes ( @xmath62 y ) of such horizontal - branch stars are too short to be of significance to the long binary - hardening timescales under consideration here .
thus , this possibility is not of practical importance here .
however , there _ is _ a situation in which this scaling is not relevant , not because it breaks down , but , rather because we move into a region of @xmath0-values where @xmath42 computed in the above way exceeds the expected main - sequence lifetime @xmath63 of the low - mass ms / early - subgiant companion . under these circumstances ,
the companion starts evolving into a giant and rapidly fills its roche lobe , for essentially any value that @xmath2 is likely to have at that stage .
this is formally equivalent to saying that @xmath42 saturates at a value @xmath63 in this range of @xmath0 .
we return to this point below .
shrinkage time we now calculate the exact variation of the shrinkage time @xmath42 with the encounter - rate parameter @xmath0 introduced earlier , keeping all terms in eq .
( [ tpxb ] ) . for this
, we need to specify the initial and final values , @xmath35 and @xmath43 , of the orbital separation .
we adopt @xmath64 for ms / subgiant companions corresponding to roche - lobe contact , when the radius of the roche - lobe @xmath60 of the companion becomes equal to the radius of the companion itself .
this translates into the above value of the orbital separation @xmath43 by the well - known paczyski relation : @xmath65 corresponding to @xmath60 being equal to companion radius @xmath66 for a companion mass @xmath67 . in general
, @xmath35 will have a value which is within a possible range @xmath68 , which is indicated in table [ tab1 ] .
this range depends on the formation - mode of the binary , and also on the evolutionary status of the companion .
the former has two possibilities , namely , ( a ) the binary is primordial , _
i.e. _ , it was already a binary when the globular cluster formed , or , ( b ) it formed by tidal capture or exchange interactions in the dense core of the globular cluster .
the latter has also two basic possibilities , namely that the companion is either ( a ) a ms / subgiant , or , ( b ) a horizontal - branch star , as explained earlier . as explained above , however , the short lifetimes of horizontal - branch ( @xcite , @xcite ) stars compared to the timescales of hardening processes under study here make it clear that they are of little importance in this problem , and we shall not consider them any further in this work .
the ranges of @xmath35 adopted in various cases are detailed in table [ tab1 ] , and are taken from current literature ( @xcite ) .
._distribution functions @xmath69 and range of initial orbital separations @xmath35 of compact - star binaries in globular clusters _ [ cols="<,^ ,
< " , ] [ tab2 ] finally , at low values of @xmath0 ( about @xmath70 - @xmath71 ) , the following aspect of the low - mass companion s evolutionary characteristics enters the picture .
the value of @xmath72 calculated in the above manner then exceeds the main - sequence lifetime @xmath63 of the companion , a simple , widely - used estimate for which is @xmath73 for a typical low - mass companion with @xmath74 , therefore , @xmath75 gyr .
when @xmath72 calculated as above exceeds this value of @xmath63 , what happens is that the companion evolves into a giant , and so comes into roche - lobe contact at a time @xmath76 for essentially all plausible values of @xmath2 at this point , irrespective of the calculated value of @xmath72 .
this is formally equivalent to the statement that @xmath72 reaches a saturation value of @xmath63 at low values of @xmath0 , the change - over occurring at @xmath77 such that @xmath78 .
thus , the computed values can be analytically approximated by the prescription that @xmath72 is given by eq .
( [ tapprox ] ) for @xmath79 , and by @xmath80 for @xmath81 .
this is shown in fig .
[ tau ] , where we normalize all the curves to the above , common `` saturation value '' @xmath82 .
note that the lifetimes of gcs are typically @xmath83 gyrs , so that , in a given gc , only those pxbs which reach roche - lobe contact within its lifetime would be relevant for our purposes .
what we have shown in fig .
[ tau ] is the _ formal _ behavior of the distribution - averaged @xmath72 for plausible distributions of @xmath35 . for a given gc
, only that range of values of @xmath35 which corresponds to roche - lobe contact within its lifetime will go into the specific calculation for it .
evolution of compact - star binaries in globular clusters how can we test the above scaling ?
since @xmath84 is not directly observable , are there possible signatures that its scaling with @xmath0 might leave in the observed behavior of the populations of compact x - ray binaries in globular clusters ?
we briefly consider this question now and suggest possible answers . as remarked earlier , pxbs are formed in gcs primarily by tidal capture and exchange interactions in the gc core .
the rate of the former process is proportional to the encounter rate between single stars in the gc core . the latter rate is commonly denoted by @xmath85 in the literature , and it scales as @xmath86 with the average core density @xmath13 , the velocity dispersion @xmath14 of the stars in the core , and the core radius @xmath87 .
we can describe this as a rate of increase of the number @xmath88 of pxbs in the gc which is @xmath89 , where @xmath90 is a constant . in an exchange interaction , one of the members of a binary consisting of two normal stars
is replaced by a heavier compact star .
the rate of this process is proportional to the encounter rate between the above two populations . assuming the population of compact stars in a gc to scale with the entire stellar population in the gc , and also the population of normal - star binaries in a gc to scale with its total population , both of which are normally done
, the rate of the exchange process also scales with the square of the stellar density , and therefore with the above @xmath85 parameter , and we can express it in a similar vein as @xmath91 , where @xmath92 is another constant .
thus , we can write the entire formation rate phenomenologically as @xmath93 , where @xmath94 . after formation
, two main processes affect the fate of the pxbs .
the first is the process of the hardening of the pxb to the point of roche - lobe overflow and conversion into an xb , as discussed in this paper .
this proceeds on a timescale @xmath42 , which means that the pxb population decreases on a timescale @xmath42 , which we can describe phenomenologically by a rate of decrease of @xmath88 which is @xmath95 .
this process may be slightly modified by second - order ones , which are normally ignored .
for example , during the above hardening , an exchange encounter of the pxb with a single normal gc star heavier than the companion in the pxb can replace the latter .
the second process is the destruction of a pxb by its encounter with single stars in the gc core .
this can happen in the two following ways .
first , a fraction of the star - pxb encounters leads to a disruption or _ ionization _ of the pxb .
this leads to a reduction in @xmath88 , whose rate is proportional to @xmath0 , the star - binary encounter rate introduced earlier , and also to @xmath88 .
we can thus express this rate of decrease in @xmath88 phenomenologically as @xmath96 , where @xmath97 is a constant .
secondly , in a smaller fraction of such encounters , a compact star can replace the normal low - mass companion in an exchange encounter , resulting in the formation of a double compact - star binary .
this is equivalent to destroying the pxb , since such a binary will not evolve into an xb .
the rate of this process scales with both @xmath88 and the rate of encounter between a given pxb and compact stars .
if we again argue that the population of compact stars in a gc scales with the entire stellar population in it , this rate is @xmath98 , and the rate of reduction of @xmath88 can be written phenomenologically as @xmath99 , where @xmath100 is a constant .
the total pxb destruction rate can thus be written as @xmath101 , with @xmath102 . a simple `` toy '' evolutionary scheme we now combine the above points into a simple `` toy '' description of pxb and xb evolution in gcs , which we can use in an attempt to extract possible signatures of the scaling described in this work . in this `` toy '' scheme , which is similar in spirit to that of @xcite and @xcite for following the evolution of x - ray binary populations of galaxies outside gcs
, the evolution of the pxb population is given by : @xmath103 wherein the above rates of increase and decrease of @xmath88 have simply been collected together . the evolution of the xb population @xmath104 resulting from the above pxbs is described in a similar manner : @xmath105 here , @xmath106 is the evolutionary timescale for xbs .
the idea here is that xbs are created from pxbs at the rate @xmath107 , and conclude their mass - transfer phase , and so their lifetime as xbs , on a timescale @xmath106 . in the spirit of the `` toy '' model ,
all timescales in equations ( [ pxbevol ] ) and ( [ xbevol ] ) can be considered constants , as can @xmath108 and @xmath109 , while in reality they depend on orbital parameters and stellar properties , as also on other parameters .
these equations can then be solved readily , and of interest to us here is the asymptotic behavior , obtained by setting the time - derivatives to zero in these , which yields an xb population : @xmath110 the effect of collisional hardening is immediately seen on the right - hand side of eqn .
( [ xbasym ] ) , in the second term in the denominator . since collisional hardening always decreases @xmath84 , it increases @xmath104 , other things being equal .
this enhancement in @xmath104 is as expected , as collisional hardening makes a larger number pxbs reach roche - lobe contact .
thus , in a gc with given properties , the number of x - ray sources is expected to be enhanced by collisional hardening compared to what it would be if this effect were negligible .
signature of collisional hardening ? can we look for observational evidence of the above enhancement in xb populations of gcs expected from collisional hardening ?
we discuss briefly an attempt to use _ chandra _ observations of gcs to this end , with the cautionary remark that our evolutionary model , as given in the previous subsection , is still too simple - minded to apply quantitatively to actual gc data .
what we are looking for , therefore , is a possible _ qualitative _ trend that is consistent with the ideas of collisional hardening discussed in this paper , which will encourage us to perform a more detailed study . the trend given by eqn .
( [ xbevol ] ) readily translates into one of the form @xmath111 where @xmath112 and @xmath113 are constants , independent of the cluster parameters .
we can compare this with data obtained from _
_ observations of gcs , as given in @xcite .
this is shown in fig .
[ obs ] . vs. @xmath0 for galactic globular clusters .
observational points with error bars from @xcite .
trend suggested by eq .
( [ xbasym1 ] ) shown for various distributions as indicated . also shown is the trend expected in absence of collisional hardening.__,width=340 ] it is clear that the trend suggested by eq .
( [ xbasym1 ] ) is consistent with the data , while that expected when collisional hardening is entirely neglected is not .
the flattening of the trend on inclusion of collisional hardening is precisely due to the scaling discussed in this paper .
however , we stress again that ours is only a `` toy '' model at this stage , relevant only for exploring feasibility . to study the effect of @xmath35 distributions
, we have normalized the constants @xmath114 and @xmath115 for each distribution by having the curve pass through two chosen points at the lowest and highest values of @xmath0 for which data is available .
the results show clearly that varying the distribution has almost no effect on the trend .
discussions the purpose of this paper is to point out an essential effect in the hardening of pxbs in gcs , _ viz .
_ , that collisional hardening
_ increases _ with increasing @xmath2 and orbital period , while that due to gravitational radiation has the opposite trend , so that their interplay leads to a characteristic scaling of the total hardening rate with gc parameters . in our introductory treatment of this effect here , we have given very simple formulations of many physical processes .
first , collisional hardening is an inherently stochastic process , wherein individual events of varying sizes accumulate to yield a final state , and the shull ( 1979 ) rate we have used is a continuous approximation to it .
secondly , the essential two- and three - body interactions that determine the evolution and fate of a pxb in a gc are also stochastic by nature .
for example , approximating an ionization event in which a binary is disrupted by a continuous term is necessarily a great oversimplification .
thus , an improved treatment must include a proper formulation of these stochastic processes , which we are developing currently .
thirdly , we have confined ourselves to circular orbits here , while binaries created by tidal capture and/or exchange interactions often have quite eccentric orbits , in which tidal circularization must play a dominant role during initial phases of hardening .
fourthly , mass segregation is an essential effect in gcs , which reflects itself in the accumulation of the heaviest objects in the core of a gc , and so in a change in the effective mass - function in the core .
finally , the evolution of the gc must be taken into account in a realistic calculation : this would make the gc parameters time - dependent , while we have treated them as constants here , and may indeed have a large effect if core collapse and bounce are modelled .
the above processes will need to be adequately modelled in a realistic calculation of the evolution of pxbs and xbs in gcs .
such calculations are under way , and the results will be reported elsewhere . *
acknowledgments : * it is a pleasure to thank the referee for thoughtful comments , which improved the paper considerably .
99 angelini l. , loewenstein m. and mushotzky r.f . , 2001 , apj , 557 , l35 .
ashman k.m . and zepf s.e . , 1998 , globular cluster systems , cambridge univ .
press .
clayton d. , 1968 , principles of stellar evolution and nucleosynthesis , mcgraw - hill , new york .
dinverno r. , 1992 , `` introducing einstein s relativity '' , oxford univ . press .
ghosh p. and white n.e .
, 2001 , apj , 559 , l97 .
goodman j. and hut p. , 1993 ,
apj , 403 , 271 .
gradshteyn i.s . and ryzhik i. m. , 1980 , table of integral , series , and products , academic press , san diego .
heggie d. , 1975 , mnras , 173 , 729 .
hut p. , 1985 ,
iaus , 113 , 231h .
hut p. and verbunt f. , 1983 , nature , 301 , 587 .
hut p. , mcmillan s. and romani r.w .
, 1992 , apj , 389 , 527 .
hut p. et .
al . , 1992 , pasp , 104 , 981h .
kippenhan r. and weigert a. , 1990 , stellar structure and evolution , springer - verlag , berlin .
kraicheva z.t . ,
popova e.i . , tutukov a.v . and yungelson l.r . , 1978 , in `` nonstationary evolution of close binaries '' , ed .
zytkow a. , polish academy of sciences , p.25 pooley d. et.al . , 2003 ,
apj , 591 , l131 .
podsiadlowski ph . ,
rappaport , s. and pfahl , e. d. , 2002 , apj , 565 , 1107 .
portegies zwart s.f .
, hut p. , mcmillan s.l.w . and verbunt f. , 1997 , a&a , 328 , 143 .
portegies zwart s.f . and
meinen a.t . , 1993 , a&a , 280 , 174 .
sarazin c.l .
et.al . , 2003 ,
apj , 595 , 743 .
shull j.m . , 1979 ,
apj , 231 , 534 .
spitzer l. jr . , 1987 , dynamical evolution of globular clusters , princeton univ .
press .
spruit , h.c . and ritter , h. , 1983 , a&a , 124 , 267 .
van den heuvel e.p.j . , 1991 ,
in `` neutron stars : theory and observation '' , eds . ventura j. and pines d. , kluwer , dordrecht , p.171 .
van den heuvel e.p.j . , 1992 , in `` x - ray binaries and recycled pulsars '' , eds
. van den heuvel e.p.j . and
rappaport s.a . ,
kluwer , dordrecht , p.233 .
van der sluys m. v. , verbunt f. and pols , o.r .
2005 , a&a , 440 , 973 .
verbunt f. , 2002 , new horizons in globular cluster astronomy , asp conf .
series .
verbunt f. and zwaan c. , 1981 , a&a , 100 , l7 .
verbunt f. and hut p. , 1987
, iaus . , 125 , 187v .
verbunt f. and lewin w.h.g .
, 2004 , in `` compact stellar x - ray sources '' , eds .
lewin w.h.g . and van der klis m. , cambridge univ.press .
white n.e . and ghosh p. , 1998 ,
apj , 504 , l31 .
white r.e . , sarazin c.l . and
kulkarni s.r . , 2002 , apj , 571 , l23 . |
the cosmological evolution survey ( cosmos ) 2 sq .
field has been extensively targeted by a wide array of observations in order to probe the cosmic evolution of galaxies and the large - scale structure in which they are immersed ( scoville et al .
2007a ) . with a wealth of multi - wavelength data spanning from x - rays to radio - wavelengths , and a core deep uv - optical - ir survey with the highest resolution and sensitivity offered by space facilities ( _ hst _ , _ spitzer _ , _ galex _ )
, it provides a unique opportunity to study the relationships and interactions among galaxy populations selected at different wavelengths and across a wide array of environments in cosmic time . a key contribution towards this knowledge comes from the far - ir to millimeter wavelength regime , which has been shown to uncover ultraluminous violently star - forming galaxies at high redshifts ( @xmath11 ) that would have gone undetected at traditional optical - near ir survey wavelengths due to their intrinsic high obscuration ( smail , ivison & blain 1997 , barger et al .
1998 , hughes et al .
1998 ) . named the ( sub-)millimeter galaxy population ( smg for sort ) , this population has been linked to the formation of massive elliptical galaxies , with large luminosities @xmath12 , large star formation rates @xmath13 yr@xmath2 , large reservoirs of gas @xmath14 , and large dynamical @xmath15 and stellar masses @xmath15(e.g .
greve et al .
2005 , tacconi et al .
2008 , dye et al .
2008 ) .
previous mm - wavelength surveys in cosmos have covered areas which were significantly smaller than the full 2 sq .
design survey and the conclusions derived from the limited number of detected galaxies suffered from large field - to - field variations .
cosbo ( bertoldi et al . 2007 ) , a 1.2 mm survey carried out with the max - planck millimeter bolometer array ( mambo ) mounted on the 30 m institute for radioastronomy at millimeter wavelengths ( iram ) telescope , detected 37 candidate sources at a s / n@xmath3 in the central 0.09 sq . deg .
of cosmos , which was mapped at a @xmath16 noise level of @xmath17 mjy beam@xmath2 .
an adjacent 0.15 sq .
field was imaged at 1.1 mm to a 1@xmath18 level of 1.2 to 1.4 mjy beam@xmath2 with the aztec instrument ( wilson et al .
2008a ) mounted on the 15 m james clerk maxwell telescope ( jcmt ) , uncovering 50 candidate sources at a s / n@xmath3 ( scott et al .
the combined area sampled by these two surveys amounts to only @xmath19 per cent of the total @xmath20 cosmos area .
field - to - field variations in the derived overall properties of smgs , such as number counts , redshift distributions , or clustering , had already been reported in these kinds of smaller ( @xmath21 sq .
deg . ) fields and attributed both to variance due to the intrinsically small volume sampled by the surveys ( e.g. coppin et al .
2006 , aretxaga et al .
2007 , wei@xmath22 et al .
2009 , austermann et al . 2010 ) and to the chance amplification by a foreground population of galaxies ( almaini et al .
2005 , austermann et al .
indeed , the environments traced by optical - ir galaxies in the cosmos sub - fields were quite different : while the aztec survey focused on an area with an overdensity of optical - ir galaxies , cosbo sampled lower galaxy - density environments ( see figure 1 ) .
a more representative survey of the cosmos field was thus necessary in order to investigate the culprits for these large variations and to characterize the global blank - field population at 1.1 mm .
aztec alone has surveyed to date @xmath23 sq .
deg . of the blank - field extragalactic sky at 1.1 mm , to @xmath24 to 1.4 mjy beam@xmath2 , coupled to 15 and 10 m telescopes with resolutions 18 and 30 arcsec , respectively ( scott et al . 2008 , 2010 , perera et al .
2008 , austermann et al .
2009 , 2010 , hatsukade et al .
2011 , ikarashi et al . 2011 ) . the cosmos survey of 0.72 sq .
deg . presented in this paper is the largest single - field extragalactic area mapped at 1.1 mm at these depths , and provides important constraints especially when compared to the scuba half degree extragalactic survey ( shades ) fields mapped at 1.1 mm , of comparable extension and depth ( austermann et al .
2010 ) .
section 2 presents the aztec / aste observations in the cosmos field ; section 3 details the data reduction processes employed to produce our 1.1 mm map ; section 4 characterizes the systematic properties of the map and derives the catalog of source candidates from our observations ; section 5 compares the aztec source catalog to other mm and radio - wavelength catalogs in cosmos ; section 6 presents the number counts over the 0.72 sq .
deg . area , which shows an excess over those of the 0.5 sq deg .
shades field ; section 7 explores the origins of the excess in terms of cross - correlations with the optical - ir galaxy population ; and section 8 discusses the results and summarizes our conclusions .
we imaged a 2800 sq .
arcmin field centered at right ascension ra(j2000.0)=@xmath25 and declination dec(j2000.0)= @xmath26 with aztec mounted on the 10 m aste ( ezawa et al . 2004 , 2008 ) , located at 4800 m in the atacama desert of chile . the survey was carried out from october 20 to november 30 , 2008 , during excellent observing conditions , with mean zenith opacity , as reported by the aste monitor , @xmath27 and values of @xmath28 about 76 per cent of the time .
aste was operated remotely with the n - cosmos3 network system ( kamazaki et al .
2005 ) by observers deployed in san pedro de atacama ( chile ) , mitaka and nobeyama ( japan ) , amherst ( usa ) and tonantzintla ( mexico ) .
a total of 112.6 hrs of on - bolometer time was devoted to this field , excluding calibration and pointing observations .
the area was sampled in raster mode at 208 arcsec s@xmath2 along 52 arcmin stripes oriented in azimuth , spaced by 1 arcmin steps in elevation . since the array orientation is fixed in azimuth and elevation , the scan angle in ra - dec continuously changes due to sky rotation , providing excellent cross - linking in the final combined image .
a total of 203 raster - scan maps were acquired in cosmos during the 2008 observing season , each taking 33 minutes of observation in a single passage through the scanning area . out of the 144 nominal bolometers of aztec ,
117 were operative during this season .
pointing observations were acquired every half an hour to an hour , sandwiching every one to two cosmos raster - maps , depending on observing conditions .
the bright qso 1055@xmath29018 , @xmath30 jy , was used to measure shifts from the standard aste pointing model .
we acquired @xmath31 arcmin@xmath32 maps of our pointing target with a continuous lissajous pattern ( scott et al .
2010 ) , and fitted a bi - dimensional gaussian to the resulting map .
the calculated offset - corrections were not implemented in real time .
instead , they were fed into a time series , and a linearly interpolated offset was applied to the telescope - pointing time series of each cosmos raster - map during the reduction process .
a total of 270 pointing observations were obtained for the cosmos field during the campaign .
aztec maps are calibrated using planets as primary calibrators .
each night uranus or neptune was imaged in order to derive the flux conversion factor for each detector . in a single observation of a field the typical statistical calibration error
is found to be @xmath33 per cent ( wilson et al . 2008a ) .
this estimate is supported by the observations of pks 0537@xmath34411 , where we report a 6 per cent scatter in 31 single observations ( wilson et al .
the cosmos data presented in this paper is the weighted sum of 203 observations taken over 27 days .
if the nightly calibration uncertainty is as high as 13 per cent , in the absence of systematic errors , combining the 27 days of observations leaves us with a statistical calibration uncertainty of 2.5 per cent . adding this in quadrature with the 5 per cent uncertainty in the brightness temperature of uranus at 1.1 mm ( griffin & orton 1993 ) gives a 5.6 per cent overall calibration uncertainty .
we reduced the aztec data in a manner similar to that described in detail by scott et al .
( 2008 ) , but with an added set of steps to account for nonlinearities in our atmospheric cleaning technique . for each of our 203 observations of the cosmos field ,
the raw time - stream data from the instrument , which includes both bolometer and pointing data , are despiked and then `` cleaned '' of atmospheric contamination in a row - by - row manner using our standard principal component analysis technique ( see scott et al .
2008 for a description ) .
an astrometric correction is made to all pointing signals in the time stream based on a linear interpolation of the pointing offsets measured by the bracketing pointing observations of the qso 1055@xmath29018 . with this correction in place ,
the bolometer signals are flux - calibrated and binned into @xmath35 arcsec@xmath32 pixels . performing this process for the 203 observations of the field results in 203 nearly independent maps which are then co - added to make a preliminary image of the sky . as in previous aztec analyses
, we also produce 100 noise - only realizations of the cosmos field by jack - knifing the time - stream data on a row - by - row basis as described in scott et al .
these noise maps are used extensively in the characterization of the map properties . as a deviation from previous aztec analyses , we revise the technique presented in scott et al .
( 2008 ) , used to estimate the transfer function of the correlated noise removal algorithm on point sources in the data .
the previous technique estimated the transfer function by creating simulated data that contain only a point source at the center of the field and subtracting the eigenmodes identified for removal in the raw time - stream data .
the revised technique differs in that the transfer function is estimated by executing the cleaning algorithm on data that combines the raw time - stream data with several simulated point sources distributed throughout the field . in the latter case ,
the eigenmodes identified for removal are recalculated from the combination of all sources of signal , noise and simulated point sources . in detail , several gaussian point sources are added to the time - stream data , all map - making steps save optimal filtering are performed , and the resulting maps are differenced from the original true maps of the sky .
the regions near the simulated sources are normalized , stacked and rotationally averaged to produce the point source ` kernel . ' as demonstrated in downes et al .
( 2011 ) , this technique provides a better estimation of the effects of the non - linear correlated noise removal on a typical point source in the map . for the cosmos field
, this has the effect of increasing the measured flux and noise values by @xmath36 per cent relative to the values obtained from the prescription in scott et al .
we use this kernel to optimally filter the maps in an identical manner to that in previous aztec analyses .
the final step in the analysis is to use the mean power spectral density of the noise maps and the newly estimated kernel map to construct an optimal filter for point source detection .
this final set of wiener filtered maps is composed of the filtered signal map , the filtered weight map , and the corresponding signal to noise map .
the 100 noise realizations are filtered with the same wiener filter and will collectively be referred to as `` noise maps '' .
the inner 0.72 sq . deg .
map of cosmos acquired by aztec at aste is shown in figure [ fig : map08 ] .
this area corresponds to a minimum coverage of 50 per cent of the maximum coverage attained in the map , and has been shown to provide excellent noise properties for source extraction and overall population analysis ( e.g. scott et al .
the raster scan strategy for cosmos translates into very uniform noise properties along this section of the map , ranging from 1.23 to 1.27 mjy beam@xmath2 , while at the extreme edges of the map ( not represented ) , the noise increases quickly as one moves away from the center .
the overall noise of the map , characterized by the combined jack - knifed noise simulations of the individual raster - maps , is well represented by a gaussian distribution of rms 1.26 mjy beam@xmath2 ( see fig . [
fig : fluxhisto ] ) .
this value is within the noise range of our previous 0.15 sq .
deg . cosmos map ( scott et al . 2008 ) .
the astrometric accuracy of the aztec map is verified by stacking at the positions of 1471 radio sources in the field that are _ not _ located within 30 arcseconds of @xmath37 or greater peaks in the aztec map .
the radio source locations are taken from the very large array ( vla ) 1.4ghz deep mosaic of cosmos ( schinnerer et al .
2010 ) which has an rms noise level of @xmath38jy beam@xmath2 in the central @xmath39 region , and positional accuracy better than 1 arcsec .
the stacked aztec map has a bright , psf - shaped peak with s / n@xmath40 and an offset from the center of the stacked image of @xmath41 arcsec in ra and @xmath42 arcsec in dec .
both of these mean offset values are small compared to the 3 arcsecond pixel size of the map and so we consider them to be too small to warrant any correction .
the source extraction algorithm employed to derive the candidate source catalog is identical to that used in scott et al .
we identify point sources in the 1.1 mm s / n map by searching for local maxima within 15 arcsec of pixels with s / n@xmath3 inside the 50 per cent coverage region .
the 189 source candidates are marked with circles in figure [ fig : map08 ] and listed in table [ tab : stacks ] , together with their measured s / n , flux densities , @xmath16 photometric errors , and deboosted flux densities ( see [ sec : num_counts ] ) .
all sources in the cosmos catalog appear to be unresolved . .aztec
/ aste source catalog and ancillary sub - mm to radio photometry .
the columns give ( 1 ) aztec identification number , ( 2 ) source name , ( 3 ) s / n of the detection in the aztec map , ( 4 ) measured 1.1 mm flux density , ( 5 ) deboosted flux density and 68 per cent confidence interval , ( 6 ) flux density at @xmath43 m from sma observations ( younger et al . 2007 , 2009 ) , ( 7 ) flux density at 1.2 mm from mambo ( bertoldi et al .
2007 ) , ( 8) flux density at 1.4ghz from vla ( schinnerer et al . 2010 ) , ( 9 ) probability of chance association as per @xmath44-statistics , ( 10 ) angular distance to radio counterpart , and ( 11 ) notes on individual objects , including names for the source in other ( sub-)mm catalogs , where aztec_j refers to the aztec / jcmt catalog ( scott et al . 2008 ) , aztec denotes interferometric sma detections at 890@xmath45 m
( younger et al . 2007 , 2009 )
, mmj denotes detections at 1.2 mm by the cosbo survey performed with mambo ( bertoldi et al .
2007 ) , and b indicates detection at 1.1 mm by the bolocam survey carried out at the caltech submillimeter observatory ( cso ; j. aguirre , private communication ; for a description of the survey see aguirre et al .
2006 ) . also included in these notes are claims for additional radio associations ( ra ) and spectroscopic redshifts , where references are ( a ) younger et al .
2007 ; ( b ) younger et al . 2009 ; ( c ) capak et al .
2011 , riechers et al . 2010 ; ( d ) smolci et al . 2011 ; and ( e ) capak et al . in preparation
[ cols="<,^,>,^,^,^,^,^,^,^ , < " , ]
with the detection of ultra - bright smgs by aztec ( wilson et al . 2008b , ikarashi et al . 2011 ) , mambo ( lestrade et al . 2010 ) , the south pole telescope ( vieira et al . 2010 ) and the _ herschel _ satellite ( negrello et al . 2010 ) , there has been new attention given to the role that lensing plays in our view of the smg population .
our aztec survey of the cosmos field offers a prime opportunity to inspect the relationship between the background smgs and the foreground large scale structure mapped out at optical - ir wavelengths over a large redshift span . indeed , with the 0.15 sq .
aztec / jcmt map of cosmos austermann et al .
( 2009 ) already found a significant spatial correlation between the projected foreground ( @xmath46 ) galaxy population and the 50 smg candidates . in this section
we will take advantage of a much wider range of foreground large scale structures covered within the 0.72 sq .
aztec / aste map of cosmos in order to test if the foreground structure significantly impacts our view of the blank - field smg population and whether this might be the likely origin of the difference in number counts between cosmos and shades .
figure [ fig : galden1 ] shows the projected density of optical - ir galaxies in the aztec field with photometric redshift @xmath47 along with the location of the 129 aztec sources with s / n @xmath48 , @xmath5 per cent of which we expect to be spurious detections .
while there is no apparent correlation between the positions of both populations , we can quantify this impression by comparing the distribution of projected densities of optical - ir galaxies within 30 arcsec radii circles centered on the aztec positions with those of 30 arcsec radii circles centered on random locations in the map ( as in austermann et al .
a kolmogorov - smirnov ( ks ) @xmath49-test can then be used to ask with what probability we can rule out the null hypothesis that the two samples are drawn from the same parent distribution .
as expected , the result , @xmath50 per cent , is consistent with no strong correlation between the optical - ir population at @xmath51 and the @xmath52 aztec catalog .
performing the same test with a flux - cut on the aztec catalog results in a more significant correlation detection . considering only the 41 s / n @xmath53 sources we find @xmath54 per cent , which is tentative evidence ( at @xmath55 level ) that the null hypothesis of no correlation is rejected
indeed , for the 20 aztec sources with deboosted flux - densities @xmath56 mjy we can strongly reject the null hypothesis of identity between the distributions of galaxy - densities around random positions in the aztec coverage area of the cosmos field and galaxy - densities around bright aztec sources : @xmath57 per cent ( @xmath58 ) .
[ fig : twotone ] the latest release of photometric redshifts for optical - ir selected galaxies in the cosmos field has achieved accuracies @xmath59 to @xmath60 at @xmath61 ( ilbert et al .
this statistical precision allows for the identification of large - scale structure pertaining to different redshift slices ( see for example scoville et al .
2007b ) . with this information in hand
, we can search in redshift space for the structures that are more likely associated with the bright aztec sources .
we will again compare the distributions of galaxy - densities around aztec sources and around random positions in the aztec covered cosmos area at the redshift of interest . as a reference ,
if we take into account the aztec catalog of 129 s / n@xmath4 sources , there is no significant signal of statistical differences between the distributions at any redshift ( see fig .
[ fig : z_contrib ] ) .
however , if we restrict the analysis to the 20 sources with deboosted flux densities @xmath56 mjy , we identify the redshift ranges @xmath62 and @xmath63 as the ones in which most significant ( @xmath64 ) differences are found between the median density of galaxies around aztec sources and the median density of galaxies around random positions ( fig .
[ fig : z_contrib ] ) .
the same redshift bins are highlighted if the analysis focuses on the 42 sources with deboosted flux densities @xmath65 mjy , although the probability of rejecting the null hypothesis for the former redshift interval is smaller than for the later .
the optical - ir galaxy density in these redshifts slices and the positioning of the bright aztec sources can be seen in figures .
[ fig : galden_slice0p65 ] and [ fig : galden_slice0p24 ] . in the interval @xmath62
we find that @xmath56 mjy aztec sources have a significantly denser galaxy environment than that found at random positions in the map : the null hypothesis of identity between the medians of the distributions can be rejected at a @xmath66 per cent confidence level , while for @xmath65 mjy aztec sources that level gets reduced to @xmath67 per cent . in the interval @xmath63 the reverse happens , finding that the @xmath68 mjy aztec sources have a denser environment with a significance @xmath69 per cent . the significance gets increased to @xmath70 per cent if @xmath65 mjy sources are considered .
the effect seems to be carried by four @xmath71 mjy sources that coincide with large density peaks , rather than by the general population of intermediate / brightness sources , though .
if one excludes the @xmath56 mjy sources from the analysis , the probability drops to @xmath72 per cent .
these two redshift intervals are the same ones we identified as having the largest correlation between @xmath8 optical - ir galaxies and aztec sources for the smaller 0.15 sq .
deg . field observed by aztec in the jcmt ( austermann et al .
it is thus important to know if the correlations arise mainly due to the very rich clusters located in the area of the sky previously sampled , or if this is a trend observed over the larger 0.72 sq .
deg . field , which might be more representative of a generic blank - field . in austermann
et al . ( 2009 ) we showed that the association of bright aztec sources around areas of large galaxy densities was not exclusively due to the presence of massive rich clusters .
if we mask out the areas marked as # 1 and # 24 in figure [ fig : galden_slice0p65 ] , which represent the fwhm of the two massive clusters at @xmath62 identified in the large - scale - structure analysis of scoville et al .
( 2007b ) , and exclude the sources that fall within those areas , the probability of having a median of galaxy - densities around aztec sources larger than that of random positions by chance is only @xmath73 per cent .
the distributions are shown in figure [ fig : dennocluster ] . likewise ,
if we exclude from the correlation analysis the overlapping area common to the aztec / jcmt and aztec / aste maps ( encircled in dashed blue lines in figures [ fig : galden_slice0p65 ] and [ fig : galden_slice0p24 ] ) , the remaining 12 @xmath56 mjy sources still show a larger tendency of falling within the large galaxy - density regions mapped at optical - ir wavelengths at @xmath8 . in that case
, the null hypothesis of identity between the medians of the distributions can be rejected at a @xmath74 per cent confidence level .
with evidence that our detected smgs are spatially correlated with foreground ( @xmath75 ) large scale structure in the cosmos field , we now ask what effect , if any , this structure might have on our estimation of the smg number counts .
we address this by first dividing the catalog into two sub - samples : smgs that lie in regions of the map with foreground galaxy densities larger than the average density of the field and smgs that lie in regions with foreground galaxy densities smaller than the average galaxy density of the field .
figure [ fig : twotone ] shows the two regions along with the respective locations of the corresponding smgs .
figure [ fig : compare_hl ] shows the number counts resulting from splitting sources in high and low foreground galaxy density environments . while none of the @xmath76 bins individually show a clear , statistically significant deviation from the mean counts over the field , the data clearly suggests that , consistent with the results of section 7 , at high fluxes ( @xmath77mjy ) the counts are systematically higher for the high foreground galaxy density sample .
this result can also be made evident by representing the distribution of flux densities of the smgs that fall within the low and high galaxy - density areas of the map ( see fig . [
fig : dendencompa ] ) . the null hypothesis that the 1.1 mm flux - density distribution of sources that fall within low and high density areas are similar can be rejected , as differences as large as the one measured can only be produced by chance in 0.37 per cent of situations .
the number counts for the low density galaxy sample is closer to the number counts derived for shades , if still systematically above them ( by @xmath78 per cent in flux density ) , while the high density galaxy sample aligns with the results derived for cosmos / jcmt and goods - n , ( about @xmath79 per cent offset in flux density at @xmath80mjy , compare with fig .
it is thus apparent that an accurate description of the overall population will fall somewhere in between these solutions , and that despite having sampled 0.72 sq .
deg . in cosmos and 0.5 sq .
deg in shades at comparable noise levels , this is not yet a large enough area to avoid variance due to intrinsic clustering of the smg population and amplification by foreground structures .
cc the optical - ir galaxies that compose the cosmos galaxy density map have accurate photometric redshifts that place them at @xmath81 , while most of the redshifts of the smgs uncovered by aztec are still unknown .
furthermore , the largest amplitude correlation between the positions of optical - ir galaxies and smgs occurs at @xmath82 ( 7 ) , where only a @xmath83 per cent of 850@xmath45 m smgs with radio associations are statistically located ( chapman et al . 2005 ) .
photometric arguments ( fig . 9 , 5.3 ) ,
place the majority of aztec smgs at @xmath84 ( see also younger et al . 2007 , 2009 ) and the 4 bright targets with refined interferometric positions by sma that have been spectroscopically targeted and have robust redshifts are indeed at @xmath85 ( table 1 , capak et al . 2011 and in prep . , riechers et al .
2010 , smolci et al .
2011 ) . while there is no proof that all bright @xmath86 mjy sources are at high redshifts
, it is likely that most of these sources are at @xmath87 and that their association with the optical - ir galaxy large scale structure is through amplification , an explanation already posed to account for the discrepancies between the number counts measured in the smaller cosmos aztec / jcmt field and shades ( austermann et al .
2009 ) . our new results , thus , confirm this interpretation for a more representative area of the sky that is not dominated by the presence of a rich cluster environment .
lensing of galaxies by foreground galaxies and by foreground groups of galaxies has been shown to be the reason for the presence of a very bright ( sub-)mm selected galaxy population ( vieira et al .
2010 , negrello et al .
2010 , lestrade et al .
2010 , ikarashi et al . 2011 ) .
these galaxies are extreme cases of the phenomena presented here , where moderate @xmath88 per cent amplification factors would be enough to account for the flux density offset between the number counts of shades and cosmos .
light propagation experiments in cosmological simulations using multiple lens - plane algorithms show that at @xmath87 any population is subject to a large chance of amplification by foreground structures ( e.g. martel & premadi 2008 ) .
correlations between bright smgs and foreground optically selected galaxies at @xmath89 were first detected using a small sample of @xmath90 mjy sources in the uk 8-mjy and hubble deep field ( almaini et al . 2005 ) .
the analysis of a statistically robust sample of 2477 350@xmath45m - selected smgs in the lockman - swire field , that includes the smaller uk 8-mjy survey area studied previously , has confirmed correlations between the smg population and @xmath91 and @xmath92 optical and ir - selected galaxy samples ( wang et al .
our first 0.15 sq .
survey in the cosmos field scanned all foreground structures at @xmath93 , yielding significant correlations between the bright 1.1mm - selected smgs detected in the field and optical - ir galaxies at redshifts @xmath94 and @xmath95 ( austermann et al . 2009 ) . whether these correlations were dominated by the optical - ir overdensity of galaxies where the aztec survey had been acquired or inherent of amplifications to the general blank - field population was open to debate .
our new analysis in cosmos allows for a better estimation of the structures that contribute to the possible boosting of flux densities of smgs by foreground structures , identifying the @xmath96 and , most significantly , the @xmath97 redshift bins as those with the largest probability of association with bright @xmath98 mjy sources .
the first redshift interval is common to the @xmath99m - selected smg correlations result . the second redshift interval , however , is still unexplored by other experiments .
other populations of high - z galaxies , also should show similar amplification trends .
the positions of bright lyman break galaxies selected to be at @xmath100 in the canada - france - hawaii - telescope legacy survey are , indeed , positively correlated with optically - selected galaxies at @xmath101 , and this effect has been shown to be consistent with weak lensing by the foreground structures in the line of sight ( hildebrandt , waerbeke & erben 2009 ) .
strong lensing is also predicted to be a dominating effect in the bright number counts of smgs ( blain 1996 , negrello et al .
2007 , lima , jain & devlin 2010 ) , however , neither strong nor weak lensing are often included in the predictions of observables offered by galaxy formation models . figure 10 shows the number counts of cosmos compared with several semi - analytical galaxy formation models that have successfully reproduced overall properties of the smg and smg population ( granato et al .
2004 , baugh et al .
2005 , rowan - robinson 2009 ) .
the 1.1 mm number counts of the shades field were compared with these very same models ( austermann et al .
2010 ) , and it was found that they all overpredicted the number counts in the @xmath102 mjy regime .
the discrepancies with the new 0.72 sq .
aztec / aste cosmos field are not that severe . while the granato et al .
( 2004 ) model for the joint formation of qsos and smgs could be made compatible with the cosmos number counts using a small shift in flux density that would mimic the amplification claimed in this paper , the model over - predicts the @xmath103 mjy number counts , which , although they are poorly constrained by the cosmos data alone , have more robust estimations from smaller deeper fields ( scott et al .
2010 , hatsukade et al . 2011 ) .
the baugh et al .
( 2005 ) model offers a good description of the cosmos number counts at @xmath104 mjy , but it over - predicts the number counts at lower flux densities , while the rowan - robinson ( 2009 ) models have a different functional form to that found for cosmos over the whole flux density range sampled by our study .
the number counts of the cosmos 0.72 sq . deg .
aztec / aste field show an overdensity of sources with respect to the population of smgs found in previous large blank - field surveys such as shades ( austermann et al , 2010 ) .
the number counts derived for the cosmos field display a systematic positive offset over those of shades , but are nevertheless consistent with those derived from smaller fields that were considered too small to characterize the overall blank - field population .
we identify departures to occur more significantly in the @xmath7 mjy regime , and demonstrate that these differences are related to the areas where galaxies at redshifts @xmath8 are more densely clustered .
the positions of optical - ir galaxies in the redshift interval @xmath105 are the most strongly correlated with the positions of the 1.1 mm bright population ( @xmath106 mjy ) , a result which does not depend exclusively on the presence of rich clusters within the survey sampled area . the most likely cause of these departures in number counts at 1.1 mm is lensing by either foreground galaxies or foreground groups of galaxies at moderate amplification levels , that increases in amplitude as one samples larger and larger flux densities .
our results and the comparison with the previously published shades number counts illustrate the fact that even @xmath107 sq .
surveys are still subject to variance due to the small volume sampled by the mapped areas in conjunction to the chance amplification by foreground structures .
this work has been supported in part by conacyt ( mexico ) grants 39953-f and 39548-f , nsf ( usa ) grants ast-0907952 and ast-0838222 , and by the mext grant - in - aid for specially promoted research ( no .
20001003 ) .
observations with aste carried out remotely from japan used nnt s gemnet2 and its partner r&e networks , which are based on the accessnova collaboration of the university of chile , ntt laboratories , and the noaj .
we would like to thank everyone who supported the aztec / aste observations of the cosmos field , including e. akiyama , r. cybulski , k. fukue , s. harasawa , s. ikarashi , h. inoue , m. kawamura , a. kuboi , j. rand , m. tashiro , t. tosaki , t. tsukagoshi , y. shimajiri and c. williams .
the aste project is driven by the nobeyama radio observatory ( nro ) , a branch of the national astronomical observatory of japan ( naoj ) , in collaboration with the university of chile and japanese institutions including the university of tokyo , nagoya university , osaka prefecture university , ibaraki university , and hokkaido university . | we present a @xmath0 sq . deg .
contiguous 1.1 mm survey in the central area of the cosmos field carried out to a @xmath1 mjy beam@xmath2 depth with the aztec camera mounted on the 10 m atacama submillimeter telescope experiment ( aste ) .
we have uncovered 189 candidate sources at a signal - to - noise ratio s / n@xmath3 , out of which 129 , with s / n@xmath4 , can be considered to have little chance of being spurious ( @xmath5 per cent ) .
we present the number counts derived with this survey , which show a significant excess of sources when compared to the number counts derived from the @xmath60.5 sq .
deg . area sampled at similar depths in the scuba half degree extragalactic survey ( shades , austermann et al .
2010 ) .
they are , however , consistent with those derived from fields that were considered too small to characterize the overall blank - field population .
we identify differences to be more significant in the @xmath7 mjy regime , and demonstrate that these excesses in number counts are related to the areas where galaxies at redshifts @xmath8 are more densely clustered .
the positions of optical - ir galaxies in the redshift interval @xmath9 are the ones that show the strongest correlation with the positions of the 1.1 mm bright population ( @xmath10 mjy ) , a result which does not depend exclusively on the presence of rich clusters within the survey sampled area .
the most likely explanation for the observed excess in number counts at 1.1 mm is galaxy - galaxy and galaxy - group lensing at moderate amplification levels , that increases in amplitude as one samples larger and larger flux densities .
this effect should also be detectable in other high redshift populations .
[ firstpage ] surveys galaxies : evolution cosmology : miscellaneous infrared : galaxies submillimetre |
study of the topological structure of flow lines foliations has a long history and leads back to h. poincar .
the question when a partition into curves is a foliation was considered by h. whitney @xcite , @xcite . in two - dimensional case
one - dimensional foliations appeared as level - sets of pseudo - harmonic functions in w. kaplan @xcite , @xcite .
let @xmath6 be a one - dimensional foliation on @xmath7 , and @xmath8 be the space of leaves endowed with the quotient topology .
notice that @xmath8 is usually non - hausdorff .
w. kaplan @xcite showed that 1 .
[ enum : kaplan : p_is_loc_triv ] the quotient map @xmath9 is a locally trivial fibration with fiber @xmath10 ; 2 .
[ enum : kaplan : split_into_strips ] there exists at most countably many leaves @xmath11 of @xmath6 such that the complement @xmath12 is a disjoint union @xmath13 , where each @xmath14 is homeomorphic with @xmath15 so that the lines @xmath16 , @xmath17 , correspond to the leaves of @xmath6 ; 3 .
[ enum : kaplan : pseudoharm_func ] there exists a pseudoharmonic function ( without singularities ) @xmath18 whose foliation by connected components of level - sets coincides with @xmath6 .
see also w. boothby @xcite , @xcite , m. morse and j. jenkins @xcite , @xcite , @xcite , @xcite and m. morse @xcite , @xcite for extensions of kaplan s results to foliations with singularities .
a. haefliger and g. reeb @xcite studied general one - dimensional _ non - hausdorff _ manifolds and showed , in particular , that the above result [ enum : kaplan : pseudoharm_func ] of w. kaplan can be deduced from poincar - bendixon theorem , see also @xcite , @xcite .
later c. godbillon and g. reeb @xcite classified locally trivial fibrations over a .
though they considered a very special case their methods clarify the general situation .
the question when for an arbitrary @xmath19-dimensional foliation @xmath6 on @xmath20 the quotient map @xmath21 has homotopy lifting properties was considered in c. godbillon @xcite , see also g. meigniez @xcite and @xcite for the criterion when @xmath22 is a serre fibration or a locally trivial fibration but mostly in smooth category .
j. harrison @xcite studied similar problem concerning geodesic flows without compact orbits .
also foliations by flow lines on @xmath23-manifolds are classified by s. matsumoto @xcite . in recent years
a progress in the theory of hamiltonial dynamical systems of small degrees of freedom increased an interest to the structure of level - sets functions on surfaces , see e.g. a. fomenko and a. bolsinov @xcite , a. oshemkov @xcite , v. sharko @xcite , @xcite , v. sharko and yu .
soroka @xcite , e. polulyakh and i. yurchuk @xcite , e. polulyakh @xcite .
homotopy properties of foliations on surfaces glued from strips similarly to [ enum : kaplan : split_into_strips ] are studied in s. maksymenko and e. polulyakh @xcite and @xcite and yu .
soroka @xcite .
in @xcite the authors extended kaplan s result [ enum : kaplan : split_into_strips ] to foliations on arbitrary non - compact surfaces @xmath20 .
namely , under certain assumptions including [ enum : kaplan : p_is_loc_triv ] , i.e. that @xmath21 is a locally trivial fibration , the topological structure of the closures @xmath24 of strips @xmath14 was described . in the present paper
we consider an arbitrary one - dimensional foliation @xmath6 with all non - compact leaves on a topological manifold @xmath20 .
our main result gives necessary and sufficient conditions for the quotient map @xmath21 to be a locally trivial fibration , see theorem [ th : charact_loc_triv ] . as mentioned above such types of questions
were extensively studied . however , the essentially new features of theorem [ th : charact_loc_triv ] in comparison e.g. with @xcite , @xcite and others , is that we work in @xmath25 category only and give a characterization in terms of the topology of the quotient space @xmath26 .
let @xmath27 be the closed half - space in @xmath28 .
[ defn : foliated_chart ] let @xmath20 be an @xmath1-dimensional topological manifold , @xmath29 .
a on @xmath0 of codimension @xmath30 is a pair @xmath31 , where @xmath32 is open and @xmath33 is a homeomorphism with @xmath34 being an open subset of @xmath35 and @xmath36 .
the set @xmath37 , @xmath38 , is called a of this foliated chart .
[ defn:1-foliation ] let @xmath39 be a partition of @xmath0 into pathconnected subsets @xmath40 of @xmath0 .
suppose that @xmath0 admits an atlas @xmath41 of foliated charts of codimension @xmath30 such that , for each @xmath42 and each @xmath43 , every pathcomponent of a set @xmath44 is a plaque .
then @xmath6 is said to be a ( and ) and @xmath41 is called a .
each @xmath40 is called a of the foliation and the pair @xmath45 is called a . in @xcite one - dimensional foliations on the plane were also called . in what follows
we will assume that @xmath20 is endowed with some @xmath46-dimensional foliation @xmath6 .
we will also consider only foliated charts included into some ( maximal ) foliated atlas associated to @xmath6 .
let @xmath47 be a leaf of @xmath6 . if @xmath47 is compact , then it is homeomorphic with the circle .
otherwise , there exists a continuous bijection @xmath48 .
moreover , if @xmath49 is another continuous bijection , then @xmath50 is a homeomorphism , c.f .
* proposition 6 ) .
recall that a continuous map @xmath51 is called whenever for each compact @xmath52 its inverse image @xmath53 is compact .
the following lemma is easy and we leave it for the reader .
[ lm : proper_leaves ] consider the following conditions on @xmath54 : * there exists an embedding @xmath55 with @xmath56 ; * there exists a proper injective continuous map @xmath55 with @xmath56 ; * any injective continuous map @xmath55 with @xmath56 is proper ; * @xmath4 is a closed subset of @xmath20 .
then the following equivalences hold true : @xmath57 a leaf @xmath47 satisfying condition ( p ) of lemma [ lm : proper_leaves ] will be said to be .
the union of all leaves of @xmath6 intersecting a subset @xmath58 is called the of @xmath59 and denoted by @xmath60 .
the following lemma is easy to prove .
[ lm : saturation_is_open ] ( * ? ? ?
* proposition 1.5 ) , ( * ? ? ?
* theorem 4.10 ) if @xmath32 is open , then @xmath60 is open as well .
if @xmath61 is a manifold , then a @xmath46-dimensional foliation @xmath6 on the product @xmath62 is a partition of @xmath62 into the lines @xmath63 , @xmath64 .
let @xmath65 be a @xmath46-foliation on @xmath66 , @xmath67 .
then an embedding @xmath68 will be called whenever @xmath69 is contained in some leaf of @xmath70 for each leaf @xmath71 .
in particular , if @xmath33 is a foliated chart as in definition [ defn : foliated_chart ] , then its inverse @xmath72 is an open foliated embedding . in this case
the set @xmath73 is a plaque for each @xmath74 .
* space of leaves .
* let @xmath75 be the space of leaves and @xmath76 be the corresponding quotient map .
endow @xmath61 with the _ quotient topology _ with respect to @xmath22 .
thus a subset @xmath77 is open if and only if its inverse @xmath78 is open in @xmath20 .
notice that for a subset @xmath32 its saturation is @xmath79 .
in particular , lemma [ lm : saturation_is_open ] means that @xmath22 is an open map .
evidently , @xmath61 is a @xmath80-space if and only if each leaf of @xmath6 is a closed subset of @xmath20 .
however , in general , @xmath61 is not a hausdorff space .
* special points .
* let @xmath81 be a point and @xmath82 be a base of neighborhoods of @xmath83 .
then the following set @xmath84 will be called the of @xmath83 .
a point @xmath83 will be called if @xmath85 .
notice that @xmath86 if and only if any two neighborhoods of @xmath83 and @xmath87 intersect .
the latter statement is symmetric with respect to @xmath83 and @xmath87 , and so it is equivalent to the assumption @xmath88 .
however , one easily checks that the property `` belong to hausdorff closure '' is not transitive
. evidently , @xmath61 is hausdorff if and only if @xmath89 for all @xmath81 , that is when @xmath61 has no special points .
the set of all special points of @xmath61 will be denoted by @xmath90 .
we will say that a leaf @xmath47 of @xmath6 is if @xmath91 is a special point of @xmath61 . in particular , @xmath92 is the set of all special leaves of @xmath6 .
the following lemma gives a characterization of special leaves and extends ( * ? ? ?
* proposition 4 ) .
[ lm : charact_spec_leaves ] let @xmath93 be a leaf and @xmath94 be the corresponding point in @xmath61 .
then the following conditions equivalent : 1 .
@xmath83 is a special point of @xmath61 , and so @xmath47 is a special leaf of @xmath6 ; 2 .
there exists a point @xmath88 distinct from @xmath83 and a sequence @xmath95 converging to both points @xmath83 and @xmath87 ; 3 .
there exist two sequences @xmath96 and @xmath97 in @xmath20 such that @xmath98 and @xmath99 belong to the same leaf for all @xmath100 , that is @xmath101 , and @xmath102 equivalence ( 1)@xmath103(2 ) is well known and easy .
( 3)@xmath104(2 ) .
denote @xmath105 , @xmath100 and @xmath106 .
then , by continuity of @xmath22 , the sequence @xmath95 converges to distinct points @xmath83 and @xmath87 .
in particular , @xmath88 .
( 2)@xmath104(3 ) .
choose any points @xmath107 and @xmath108 , and let @xmath109 and @xmath110 be countable bases of topology on @xmath20 at @xmath111 and @xmath112 respectively . since @xmath22 is open , @xmath113 and @xmath114 are open neighborhoods of @xmath83 and @xmath87 respectively .
but these points are special and @xmath86 , whence @xmath115 for all @xmath100 . hence there exist @xmath116 and @xmath117 such that @xmath118 . then @xmath96 and @xmath97 converge to @xmath111 and @xmath112 respectively .
[ def : cross_section ]
let @xmath119 be an open subset of @xmath120 , where @xmath121 is the codimension of the foliation @xmath6 .
a continuous map @xmath122 is called a of @xmath6 if * @xmath123 ; * the composition @xmath124 is an injective map , that is @xmath125 implies that the images @xmath126 and @xmath127 of these points belong to distinct leaves of @xmath6 .
if @xmath128 and @xmath47 is a leaf of @xmath6 containing @xmath111 , then we will also say that @xmath129 as well as .
the aim of this paper is to present necessary and sufficient conditions for the map @xmath22 to be a locally trivial fibration under assumption that all leaves of @xmath6 are non - compact .
[ th : charact_loc_triv ] let @xmath20 be an @xmath1-dimensional manifold and @xmath6 be a one - dimensional foliation on @xmath20 .
suppose that all leaves of @xmath6 are non - compact and the family of all special leaves of @xmath6 is locally finite .
then the following conditions are equivalent . 1 . [ lm : loc_triv_relations : enum : prj_is_ltfibr ] the quotient map @xmath21 is a locally trivial fibration with fiber @xmath10 and @xmath61 is locally homeomorphic with @xmath120 ( though it is not necessary a hausdorff space ) .
[ lm : loc_triv_relations : enum : partit_lt__prj_open ] for each leaf @xmath47 there exists an open saturated neighborhood foliated homeomorphic with @xmath130 , where @xmath131 is an open subset of @xmath120 .
[ lm : loc_triv_relations : leaf_proper_cross_sections ] for each leaf @xmath47 of @xmath6 there exists a cross section passing through @xmath47 .
it is proved in ( * ? ? ?
* chapter iii , propostion 4 and corollary ) that for arbitrary @xmath132-dimensional foliation @xmath6 then the quotient map @xmath21 is a serre fibration whenever it satisfies a certain variant of homotopy extension property and either has a local section at each point or the quotient @xmath26 is a ( possibly non - hausdorff manifold ) .
see also @xcite and @xcite for extensions . our theorem [ th : charact_loc_triv ] claims that for one - dimensional foliations @xmath6 with locally finite family of special leaves existence of cross sections with open subsets of @xmath133 implies that @xmath22 is even a locally trivial fibration and @xmath26 is a possibly non - hausdorff manifold .
equivalence between [ lm : loc_triv_relations : enum : prj_is_ltfibr ] and [ lm : loc_triv_relations : enum : partit_lt__prj_open ] for @xmath134 is proved in @xcite without assumption that @xmath61 is locally homeomorphic with @xmath10 . also in ( * ? ? ? * , proposition 1 ) it is show that @xmath26 is a @xmath46-manifold for one - dimensional foliation on @xmath7 .
r. h. bing @xcite , @xcite constructed a non - manifold @xmath135 such that @xmath136 is homeomorphic with @xmath137 .
in other words , @xmath137 admits a trivial partition into open arcs ( being not a foliation ) such that the quotient space @xmath138 is not a @xmath23-manifold . that example was improved by many authors ,
see e.g r. rosen @xcite , j. kim @xcite , j. bailey @xcite , l. rubin @xcite .
e. dyer and m. hamstrom @xcite studied so called mappings @xmath76 between metric spaces such that the inverse images of all points are in a certain sense `` uniformly homeomorphic '' , and get sufficient conditions when such a map is equivalent to a trivial fibration , see ( * ? ? ?
* theorem 7 ) , and also @xcite , @xcite for generalizations .
we consider here a similar problem , but now the space @xmath61 is not even hausdorff , and we gave conditions when @xmath22 is a locally trivial fibration .
the following statement is proved in ( * ? ? ?
* theorem 1 ) for continuous functions @xmath18 , and in ( * ? ? ?
* item 3 at the end of page 3778 ) for smooth case .
let @xmath139 and @xmath140 be two manifolds such that @xmath141 and @xmath142 be a surjective continuous map such that * @xmath143 and @xmath144 ; * the partition @xmath145 of @xmath139 constitutes a one - dimensional foliation with all non - compact leaves .
then @xmath146 is a locally trivial fibration with fiber @xmath10 . in particular ,
if @xmath140 is contractible , then @xmath146 is a trivial fibration .
we claim that @xmath6 contains no special leaves and each leaf admits a cross section . then it will follow from theorem [ th : charact_loc_triv ] that @xmath146 is a locally trivial fibration with fiber @xmath10 .
_ absence of special leaves .
_ let @xmath147 be the space of leaves endowed with the corresponding factor topology .
then @xmath146 can be written as a composition of the following maps @xmath148 where @xmath149 is the induced continuous bijection . since @xmath140 is hausdorff
, it follows that so is @xmath150 , and therefore @xmath150 contains no special points .
hence @xmath6 contains no special leaves . _ existence of cross sections .
_ let @xmath151 and @xmath152 be a foliated chart at @xmath111 as in definition [ defn:1-foliation ] such that @xmath153 , where @xmath154 .
then the map @xmath155 defined by @xmath156 is a cross section of @xmath6 .
in fact , theorem [ th : charact_loc_triv ] is an easy consequence of the following statements : * lemma [ lemma : proper_leaf1 ] .
* _ let @xmath157 be a leaf of @xmath6 .
suppose that for each leaf @xmath47 of @xmath6 contained in @xmath158 there exists a cross section @xmath129 passing through @xmath47
. then @xmath157 is properly embedded . _
[ th : trivialization_of_cross_section ] let @xmath159 be a cross section intersecting only leaves being simultaneously non - compact , properly embedded , and non - special . then the saturation @xmath160 is open and foliated homeomorphic with @xmath161 . the proof of theorem [ th : charact_loc_triv ] will be given in
[ sect : proof_of_theorem : th : charact_loc_triv ] . in [ sect : prelim ] we will prove some general preliminary results concerning one - dimensional @xmath25 foliations being well known for smooth case . in particular we will prove lemma [ lemma : proper_leaf1 ] .
[ sect : proof_of_theorem : th : trivialization_of_cross_section ] is devoted to the proof of theorem [ th : trivialization_of_cross_section ] using e. michael s theorems about selections of multivalued maps .
[ lm : loc_triv_relations : enum : prj_is_ltfibr]@xmath104[lm : loc_triv_relations : enum : partit_lt__prj_open ] , [ lm : loc_triv_relations : leaf_proper_cross_sections ] .
suppose the quotient map @xmath76 is a locally trivial fibration with fiber @xmath10 and @xmath61 is locally homeomorphic with @xmath120 .
this means that for each @xmath93 there exist * an open neighborhood @xmath162 of its image @xmath163 homeomorphic with an open subset of @xmath120 and * a foliated homeomorphism @xmath164 .
then @xmath78 is an open and saturated neighborhood of @xmath47 and @xmath165 is a foliated homeomorphism required by [ lm : loc_triv_relations : enum : partit_lt__prj_open ] .
moreover , the map @xmath166 defined by @xmath167 is a cross section passing through @xmath47 .
this proves [ lm : loc_triv_relations : leaf_proper_cross_sections ] .
[ lm : loc_triv_relations : leaf_proper_cross_sections]@xmath104[lm : loc_triv_relations : enum : partit_lt__prj_open ] .
suppose each leaf of @xmath6 admits a local cross section .
then it follows from lemma [ lemma : proper_leaf1 ] that all leaves of @xmath6 are properly embedded .
let @xmath168 be a family of all special leaves and @xmath169 be a special leaf .
since each leaf is closed and @xmath168 is a locally finite family , it follows that @xmath170 is a closed set , whence @xmath171 is open and saturated and contains no special leaves .
moreover , since each leaf in @xmath172 admits a local cross section , it follows from theorem [ th : trivialization_of_cross_section ] that each leaf @xmath173 has an open saturated neighborhood @xmath174 foliated homeomorphic with @xmath175 , where @xmath131 is an open subset of @xmath176 .
then @xmath174 is also open in @xmath20 .
this proves [ lm : loc_triv_relations : enum : partit_lt__prj_open ] .
[ lm : loc_triv_relations : enum : partit_lt__prj_open]@xmath104[lm : loc_triv_relations : enum : prj_is_ltfibr ] .
let @xmath81 and @xmath177 be the corresponding leaf of @xmath6 .
suppose there exist an open @xmath178 and a foliated homeomorphism @xmath179 onto some open and saturated neighborhood @xmath180 of @xmath47 .
since @xmath22 is an open map , so is the composition @xmath181 .
hence @xmath182 is an open neighborhood of @xmath83 in @xmath61 .
moreover , the restriction @xmath183 is a continuous and open bijection , and so it is a homeomorphism . thus @xmath61 is locally homeomorphic with @xmath176 and
the map @xmath184 is a trivialization of @xmath22 over @xmath185 , so @xmath22 is a locally trivial fibration with fiber @xmath10 .
theorem [ th : charact_loc_triv ] is completed .
in this section we will assume that @xmath20 is an @xmath1-dimensional manifold with @xmath186 and @xmath6 is a one - dimensional foliation on @xmath20 .
some statements in this section are well known for @xmath187 foliations e.g. @xcite , and some of them are proved for @xmath25 case but for the foliations on @xmath7 , see e.g. w. kaplan @xcite .
however we did not find good exposition in the literature for general @xmath25 foliations needed in our case and therefore short proofs will be presented .
this will also make the paper self - contained
. it will be convenient to regard the graph of a function @xmath188 as the following subset @xmath189 of @xmath190 .
thus we switch the coordinates .
[ lm : make_constrant_graph ] let @xmath191 be a topological space , @xmath192 be continuous functions such that @xmath193 for all @xmath194 and @xmath195 , and @xmath196\ \times { z}\ ] ] be the graph of @xmath197 .
let also @xmath198 be any increasing @xmath19-tuple of numbers .
then there exists a self - homeomorphism @xmath199 of @xmath200 \times { z}$ ] such that 1 .
@xmath199 is fixed on @xmath201 and @xmath202 ; 2 .
[ lm : make_constrant_graph : enum : preserve_lines ] @xmath203\times z ) = [ a , b ] \times z$ ] for all @xmath195 ; 3 .
@xmath204 ; 4 .
[ lm : make_constrant_graph : enum : fixed_on_0 ] if @xmath205 for some @xmath206 and all @xmath207 , then @xmath199 is fixed on @xmath200\times z$ ] .
the proof follows from ( * ? ? ?
* lemma 6.1.1 ) , see also ( * ? ? ?
* lemma 5.2.1 ) .
let us just mention that the situation can be reduced to the case @xmath200 = [ 0,1]$ ] , and that for @xmath208 the desired self - homeomorphism @xmath199 of @xmath209 \times { z}$ ] can be defined e.g. by @xmath210 we leave the details for the reader , see figure [ fig : const_graphs ] .
[ lm : adopt_quasi_tube ] let @xmath211 be an open neighborhood of @xmath212 in @xmath213 and @xmath214 be a cross section of the trivial one - dimensional foliation such that @xmath215 .
then for each @xmath216 there exists an open embedding @xmath217 such that 1 .
[ lm : adopt_quasi_tube : enum : pres_lines ] @xmath218 for all @xmath219 ; 2 .
[ lm : adopt_quasi_tube : enum : crs_is_constant ] @xmath220 for all @xmath219 , i.e. @xmath221 ; 3 . [ lm : adopt_quasi_tube : enum : fixed_on_0 ] if @xmath222 , then @xmath223 for all @xmath224 .
let @xmath225 be the standard projection .
then the assumption that @xmath129 is a cross section means that the composition @xmath226 is an injective map between open subsets of @xmath213 .
hence by brouwer s theorem on domain invariance , e.g. @xcite , @xmath227 is an open neighborhood of @xmath212 in @xmath213 .
therefore we get an open embedding @xmath228 satisfying [ lm : adopt_quasi_tube : enum : pres_lines ] and [ lm : adopt_quasi_tube : enum : fixed_on_0 ] . then @xmath229 can be regarded as a graph of certain continuous function @xmath230 .
hence we get from lemma [ lm : make_constrant_graph ] that @xmath165 can be composed with a foliated homeomorphism of @xmath231 to satisfy [ lm : adopt_quasi_tube : enum : crs_is_constant ] , that is to make @xmath232 being the graph of the constant function @xmath233 .
moreover , statements [ lm : make_constrant_graph : enum : preserve_lines ] and [ lm : make_constrant_graph : enum : fixed_on_0 ] of lemma [ lm : make_constrant_graph ] allow to preserve properties [ lm : adopt_quasi_tube : enum : pres_lines ] and [ lm : adopt_quasi_tube : enum : fixed_on_0 ] respectively .
[ lm : extanding_foliated_chart ] let @xmath47 be a leaf of @xmath6 , @xmath234 be two compact segments such that @xmath235 is a point , @xmath131 be an open @xmath30-disk , @xmath236 , and @xmath237 be two open foliated embeddings such that @xmath238\times 0\bigr ) & = j_1 , & { \psi}_2\bigl ( [ b , c]\times 0\bigr ) & = j_2 , & { \psi}_1(b,0 ) = { \psi}_2(b,0 ) = j_1 \cap j_2,\end{aligned}\ ] ] and the union of the images of @xmath239 and @xmath240 does not contain compact leaves of @xmath6 .
then there exists an open neighborhood @xmath174 of @xmath212 in @xmath131 and an open foliated embedding @xmath241 such that @xmath242\times 0\bigr ) = j_1\cup j_2 $ ] , see figure [ fig : join_fol_charts ] .
notice that the assumption that the union of the images of @xmath239 and @xmath240 does not contain compact leaves of @xmath6 implies that for any @xmath243 the union of the arcs @xmath244 does not contain a non - trivial loop , so the intersection of these arcs is connected ( though possibly empty ) .
since @xmath239 and @xmath240 are open embeddings , there exist @xmath245 and a small open neighborhood @xmath174 of @xmath212 in @xmath131 such that @xmath246 then we have an embedding @xmath247 defined by @xmath248 , @xmath249 .
hence by lemma [ lm : adopt_quasi_tube ] one can find an open foliated embedding @xmath250 such that * @xmath251 for all @xmath252 ; * @xmath253 ; * @xmath254 for all @xmath255 ; * @xmath256 for all @xmath249 .
now define the map @xmath257 by @xmath258 , \\ { \psi}_2({t},{u } ) , & { t}\in[b , c+{\varepsilon } ) .
\end{cases}\ ] ] one easily checks that @xmath165 is an open foliated embedding which coincides with @xmath239 on @xmath259\times 0 $ ] and with @xmath240 on @xmath260
. in particular , @xmath242\times 0\bigr ) = j_1\cup j_2 $ ] .
[ cor : extanding_foliated_chart]c.f .
* lemma 22 ) let @xmath34 be an open @xmath30-disk , @xmath47 be a leaf of @xmath6 , and @xmath261 be a compact segment .
then there exists an open foliated embedding @xmath262 such that @xmath263 \times 0\bigr ) = j$ ] .
let us show that .
indeed , since @xmath47 is either non - compact or is an embedded circle , it follows that @xmath264 .
fix a point @xmath265 . as @xmath0 is a regular space ,
there exist a pair of disjoint open neighborhoods @xmath266 and @xmath267 .
denote @xmath268 .
then by lemma [ lm : saturation_is_open ] @xmath211 is open .
moreover , @xmath269 , so @xmath270 . finally , since @xmath271 , we see that @xmath272 for each @xmath273 , that is @xmath274 . in other words
, @xmath211 does not contain any leaf of @xmath6 .
in particular , @xmath211 can not contain compact leaves .
notice that @xmath275 can be covered by finitely many foliated charts contained in @xmath174 .
lemma [ lm : extanding_foliated_chart ] allows to replace two consecutive foliated charts with one .
hence the proof follows from that lemma by induction on the number of foliated charts covering @xmath275 .
* cross sections . *
the following two lemmas describe general properties of cross sections .
[ lm : good_foliated_chart ] let @xmath276 be an open foliated embedding .
let also @xmath277 , @xmath278 , @xmath74 , be a plaque of @xmath165 , and @xmath122 be a cross section .
then the following statements hold true . 1 .
[ item : full_plaques ] suppose @xmath279 for each @xmath74 .
then then for each @xmath280 the restriction map @xmath281 is a cross section of @xmath6 .
[ item : transverse_is_cross_section ] suppose @xmath282 for some @xmath74 and @xmath283 .
then there exists an open neighborhood @xmath284 of @xmath87 in @xmath119 and an open neighborhood @xmath285 of @xmath87 in @xmath34 such that * @xmath286 ; * @xmath287 for each @xmath288 .
+ in particular , the restriction @xmath289 is a cross sections of @xmath6 .
[ item : cross_sections_at_each_point ] for every @xmath290 there exist an open subset @xmath174 of @xmath213 and a cross - section @xmath291 such that @xmath292 .
[ item : full_plaques ] suppose @xmath279 for all @xmath74 .
since @xmath293 intersects each leaf of @xmath6 in at most one point , it follows that distinct plaques @xmath294 and @xmath295 for @xmath296 belong to distinct leaves of @xmath6 . as @xmath34 is an open subset of @xmath213 ,
the map is a cross section for each @xmath297 .
[ item : transverse_is_cross_section ] consider the following map : @xmath298 where @xmath299 is the standard projection to the second coordinate .
then the assumption @xmath282 for some @xmath74 and @xmath283 implies that @xmath300 and @xmath301 .
since the images of distinct points of @xmath119 under @xmath129 are contained in distinct leaves of @xmath6 , they also belong to distinct plaques of @xmath165 , whence @xmath302 is an injective continuous map between open subsets of @xmath213 .
hence , by brouwer theorem on domain invariance @xmath302 is an open map , @xcite . in particular , @xmath302 yields a homeomorphism of some open neighborhood @xmath284 of @xmath87 onto some open neighborhood @xmath303 of @xmath83 in @xmath34 .
this implies that @xmath304 and @xmath287 for each @xmath305 .
[ item : cross_sections_at_each_point ] let @xmath47 be the leaf containing @xmath111 and @xmath306 . if @xmath307 , then one can put @xmath308 and @xmath309 .
therefore suppose @xmath310 .
let @xmath311 be a closed segment with ends @xmath111 and @xmath112 .
then by corollary [ cor : extanding_foliated_chart ] there exists an open foliated embedding @xmath262 such that @xmath263 \times 0\bigr ) = j$ ] , @xmath312 and @xmath313 .
thus @xmath314 , and so by [ item : transverse_is_cross_section ] there exists a neighborhood @xmath174 of @xmath212 in @xmath34 such that the map @xmath315 defined by @xmath316 is a cross section with @xmath317 .
it remains to note that @xmath318 .
[ lemma : proper_leaf1 ] let @xmath157 be a leaf of @xmath6 .
suppose that for each leaf @xmath47 of @xmath6 contained in @xmath158 there exists a cross section @xmath129 passing through @xmath47
. then @xmath157 is properly embedded , i.e. it satisfyies conditions ( m ) and ( c ) of lemma [ lm : proper_leaves ] .
if @xmath157 is compact , then it is necessarily properly embedded .
therefore assume that @xmath157 is non - compact .
\(m ) let @xmath319 be a leaf of @xmath6 .
by [ item : cross_sections_at_each_point ] of lemma [ lm : good_foliated_chart ] for each @xmath320 there exists an open foliated embedding @xmath321 such that @xmath322 and different plaques of @xmath165 are contained in different leaves of @xmath6 .
in particular , @xmath165 homeomorphically maps @xmath323 onto an open neighbourhood of @xmath111 in @xmath47 .
this implies that @xmath47 is an embedded 1-submanifold of @xmath20 .
\(c ) let @xmath324 . then decreasing @xmath34
one can assume that the image of @xmath165 does not intersect @xmath157 , whence @xmath325 .
from arbitrariness of @xmath326 we conclude that @xmath157 is closed in @xmath20 . * parallel cross sections .
* let @xmath122 be a cross section and @xmath327 be an open subset . then a cross section @xmath328 with @xmath129 , whenever for each @xmath329 the points @xmath330 and @xmath126 belong to the same leaf .
also @xmath331 is if it parametrically agrees with @xmath129 and @xmath332 .
let @xmath333 be two parallel cross sections intersecting only non - compact leaves .
for each @xmath334 let @xmath335 be the leaf containing @xmath336 and @xmath337 , @xmath338 be the compact segment with ends @xmath336 and @xmath337 , and @xmath339 be the interior of @xmath340 .
in this situation we will put : @xmath341
[ lm : parallel_crossections ] there exists a homeomorphism @xmath342\times{{v}}\to { k(\gamma_0 , \gamma_1)}$ ] such that @xmath343\times { u}\bigr ) & = { i_{{u } } } , & { \psi}(0,{u } ) & = { \gamma}_0({u } ) , & { \psi}(1,{u } ) & = { \gamma}_1({u})\end{aligned}\ ] ] for every @xmath334 , see figure [ fig : space_between_crs ] .
in particular , @xmath344 . moreover , @xmath345 is open in @xmath20 .
fix some @xmath236 and denote @xmath346 .
then it follows from corollary [ cor : extanding_foliated_chart ] and lemma [ lm : adopt_quasi_tube ] that for each @xmath334 there exists a neighborhood @xmath303 in @xmath119 and an open foliated embedding @xmath347 having the following properties : 1 .
@xmath348\times{u } ) = { i_{{u}}}$ ] , @xmath349 , and @xmath350 ; 2 .
@xmath351 , @xmath352 , and therefore @xmath353 , are contained in the same leaf of @xmath6 for each @xmath354 ; 3 .
@xmath355 and @xmath356 .
in particular , this implies that the set @xmath357 is open in @xmath20 .
as @xmath119 is paracompact , there is a locally finite cover @xmath358 of @xmath119 and for each @xmath359 an open foliated embedding @xmath360 such that @xmath361\times{u } ) = { i_{{u}}}$ ] for all @xmath362 .
denote @xmath363 and @xmath364 .
then @xmath59 is an open neighborhood of @xmath365 and @xmath366 is a locally finite cover of @xmath59 .
let @xmath367\}_{i\in\lambda}$ ] be a partition of unity subordinated to the cover @xmath358 .
thus @xmath368 and @xmath369 .
let also @xmath370 and @xmath371 be the standard projections , and @xmath372.\ ] ] then @xmath373 , whence @xmath374 extends by zero to a continuous function on all of @xmath59 .
let @xmath375 be the function defined by the following rule : @xmath376 since for each @xmath362 the function @xmath377 $ ] is homeomorphism which maps @xmath336 and @xmath337 to @xmath212 and @xmath46 respectively , and @xmath378 , we see that the restriction @xmath379 is a convex linear combination of orientation preserving homeomorphisms . therefore @xmath380 $ ] is a homeomorphism as well .
let also @xmath381 be the map defined by @xmath382 whenever @xmath383 .
due to ( b ) this definition does not depend on a particular @xmath384 containing @xmath111 .
hence @xmath385 is a well - defined continuous map . then the mapping @xmath386\times{{v}}\ ] ]
is a continuous bijection being also a local homeomorphism , and so it is a homeomorphism .
moreover , @xmath387\times { u}$ ] for all @xmath334 .
therefore @xmath388 is the required homeomorphism .
[ lm : tube_along_leaf ] let @xmath389 , @xmath390 , be a family of pairwise parallel cross sections intersecting only non - compact leaves and @xmath391 be the common saturation of their images .
suppose also that the following two conditions hold : 1 .
@xmath392 for @xmath393 ; 2 .
@xmath394 .
then @xmath59 is open in @xmath20 and foliated homeomorphic with @xmath161
. by lemma [ lm : parallel_crossections ] for each @xmath390 there exists a homeomorphism @xmath395 \times { { v}}\to { k({\gamma}_{i } , { \gamma}_{i+1})}\ ] ] such that for each @xmath334 1 .
@xmath396\times { u})$ ] is a segment of the leaf of @xmath6 between the points @xmath397 and @xmath398 ; 2 .
@xmath399 .
therefore we have a homeomorphism @xmath400 defined by @xmath401 whenever @xmath402 $ ] and @xmath334 .
moreover , @xmath403 is open in @xmath20 .
let @xmath159 be a cross section intersecting only leaves being simultaneously non - compact , properly embedded , and non - special .
we have to prove that its saturation @xmath160 is open and foliated homeomorphic with @xmath161 .
first we will assume that @xmath404 .
the proof of the case @xmath405 will follow from the case @xmath186 by passing to the double @xmath406 of @xmath20 and considering the one - dimensional foliation on @xmath406 induced by @xmath6 .
it will be given at the end op this section .
our proof is based on the following statement which will be proved below .
[ pr : global_crs_outside_compact_set ] let @xmath407 be a compact subset . then one can find two parallel cross sections @xmath408 parametrically agreeing with @xmath129 and satisfying @xmath409 moreover , if @xmath410 are two parallel cross sections parametrically agreeing with @xmath129 , then one can assume that @xmath411 before proving theorem [ th : trivialization_of_cross_section ] let us deduce it from proposition [ pr : global_crs_outside_compact_set ] .
fix any increasing sequence @xmath412 of compact subsets of @xmath20 such that @xmath413 . using proposition [ pr : global_crs_outside_compact_set ]
one constructs a family of parallel cross sections @xmath414 , @xmath100 , parametrically agreeing with @xmath129 and such that 1 .
@xmath415 ; 2 .
@xmath416 for all @xmath417 .
hence @xmath418 exchanging @xmath419 and @xmath420 if necessary and re - denoting them as follows : @xmath421 , and @xmath422 for @xmath100 , one can assume that the sequence of cross sections @xmath423 satisfies assumptions of lemma [ lm : tube_along_leaf ] , see figure [ fig : inf_cross_sect_sequence ] .
hence @xmath160 is open and foliated homeomorphic with @xmath161 .
this proves theorem [ th : trivialization_of_cross_section ] modulo proposition [ pr : global_crs_outside_compact_set ] .
the following lemma guarantees existence of local cross sections in proposition [ pr : global_crs_outside_compact_set ] .
[ lm : local_crs_outside_compact_set ] let @xmath407 be a compact subset .
then for each @xmath334 one can find an open neighborhood @xmath424 in @xmath119 and two parallel cross sections @xmath425 parametrically agreeing with @xmath129 and such that @xmath426 suppose that lemma fails , so there exists @xmath334 belonging to some leaf @xmath47 such that * for any decreasing sequence @xmath427 of neighborhoods of @xmath83 in @xmath119 with @xmath428 * and any family of pairs of parallel cross sections @xmath429 , @xmath100 , parametrically agreeing with @xmath129 the set @xmath430 contains some point @xmath431 .
denote @xmath432 then one can assume , in addition , that the following properties hold : 1 .
the sequence @xmath96 converges to some point @xmath433 ; 2 .
@xmath434 ; 3 .
@xmath435 for all @xmath100 , whence @xmath436 as well , and so @xmath437 . indeed , ( a ) follows from compactness of @xmath438 . to prove ( b ) fix any continuous bijection @xmath48 . by assumption @xmath439
is proper , so one can find @xmath440 such that @xmath441 .
choose @xmath419 and @xmath420 so that @xmath442 , @xmath443 @xmath444 , and @xmath445 . then we will have that @xmath434 . finally , to satisfy ( c ) choose @xmath446 so small that @xmath447 for all @xmath100 .
now let @xmath448 be the leaf of @xmath6 containing @xmath98 , and @xmath449 .
then the sequence @xmath97 converges to @xmath450 .
hence @xmath451 , while @xmath452 for all @xmath100 .
therefore by lemma [ lm : charact_spec_leaves ] @xmath47 is a special leaf which contradicts to the assumption .
the rest of the proof of theorem [ th : trivialization_of_cross_section ] is based on e. michael s result about selections , @xcite .
let @xmath453 be the set of all subsets of @xmath20 and @xmath454 be the set of all closed subsets of @xmath20 .
let also @xmath455 be a subset and @xmath456 be a multivalued map , i.e. a map @xmath457 . then a for the restriction @xmath458 is a continuous map @xmath459 such that @xmath460 for all @xmath461 .
a multivalued map @xmath456 is called whenever for each open @xmath32 the set @xmath462 is open in @xmath119 .
a family @xmath463 is called , @xmath464 , if for every @xmath465 , @xmath466 , and a neighborhood @xmath467 of @xmath111 in @xmath20 , there exists a neighborhood @xmath468 of @xmath111 in @xmath20 such that for every @xmath469 every continuous map @xmath470 of an @xmath471-sphere ( @xmath472 ) is homotopic to a constant map in @xmath473 . a topological space @xmath191 is called @xmath474 , or , @xmath464 , if every continuous map @xmath475 of an @xmath471-sphere ( @xmath472 ) is homotopic to a constant map .
[ th : michaeltheorm ] ( * ? ? ?
* theorem 1.2 ) let @xmath119 be a separable metric space , @xmath476 be a closed subset with @xmath477 , @xmath20 a complete metric space , @xmath478 be equi-@xmath479 and @xmath480 be a lower semi - continuous map .
then every selection for @xmath458 can be extended to a selection for @xmath481 for some open @xmath482 .
if also every @xmath483 is @xmath474 , then one can take @xmath484 .
we will use the following particular case of theorem [ th : michaeltheorm ] .
[ cor : michaeltheorm ] let @xmath119 be a separable metric space , @xmath485 , @xmath20 be a complete metric space , and @xmath478 be an equi-@xmath486 family such that each @xmath487 is contractible .
then every lower semi - continuous multivalued map @xmath480 has a continuous selection .
* proof of proposition [ pr : global_crs_outside_compact_set ] * since @xmath119 is paracompact , it follows from lemma [ lm : local_crs_outside_compact_set ] that there exist * a locally finite open cover @xmath488 of @xmath119 with compact closures @xmath489 , and * a family of pairs of parallel cross sections @xmath490 , @xmath100 , parametrically agreeing with @xmath129 such that @xmath491 where @xmath492 whenever the cross sections @xmath410 are given , and @xmath493 otherwise .
then it follows from lemmas [ lm : parallel_crossections ] and [ lm : make_constrant_graph ] that for each @xmath100 one can find an embedding @xmath494 \times { { w}}_i \to { x}$ ] such that for each @xmath334 1
. @xmath495 \times { u}\bigr)$ ] is contained in the leaf of @xmath6 ; 2 .
@xmath496 , @xmath497 , @xmath498 ; 3 .
@xmath499 ; 4 .
@xmath500 are contained in the same path component of @xmath501 for all @xmath100 .
let @xmath334 , @xmath335 be the leaf of @xmath6 containing @xmath126 , and @xmath502 be any bijection satisfying @xmath503 , and @xmath504 .
therefore @xmath505 for all @xmath506 such that @xmath507
. then there are two numbers @xmath508 such that @xmath509 consists of two half closed intervals @xmath510 $ ] and @xmath511 .
since @xmath335 is a properly embedded leaf , it follows that @xmath512 and @xmath513 are closed in @xmath20 .
moreover , by ( 3 ) they do not intersect @xmath438 .
define the following two maps @xmath514 , i.e. multivalued mappings @xmath515 with closed images , by @xmath516 for @xmath334 .
[ lm : ab_prop ] ( i ) the maps @xmath517 and @xmath518 are lower semi - continuous .
( ii ) the families @xmath519 and @xmath520 are equi-@xmath479 for all @xmath464 .
it suffices to check ( i ) and ( ii ) for @xmath517 only .
\(i ) we should check that for each open @xmath521 the set @xmath522 is open as well .
let @xmath334 be such that @xmath523 , and @xmath524 .
since @xmath525 is open , one can assume that @xmath111 is not the end of @xmath512 , that is @xmath526 . by assumption @xmath527 for some @xmath100 .
then by corollary [ cor : extanding_foliated_chart ] for the closed interval on @xmath335 between @xmath111 and @xmath528 there exists an open neighborhood @xmath529 of @xmath83 in @xmath530 and an open foliated embedding @xmath531 such that * @xmath532 ; * @xmath533 ; * @xmath534 ; * @xmath535 .
it follows from ( a ) and ( d ) that @xmath536 \times { v } ) \subset a_{{v } } \cap { u}$ ] , whence @xmath537 .
thus @xmath538 is open , and so @xmath517 is a lower semi - continuous multivalued map .
\(ii ) notice that for each @xmath539 there exists an open neighborhood @xmath540 such that the intersection of @xmath540 with each leaf @xmath47 is either empty or homeomorphic to an open interval .
therefore intersection of @xmath540 with each set @xmath512 is either empty or homeomorphic to @xmath541 or to @xmath542 $ ] . in the latter two cases @xmath543 is contractible .
hence every continuous map @xmath544 is null homotopic and one can put @xmath545 .
this means that @xmath546 is equi-@xmath479 for all @xmath464 .
since for each @xmath334 the sets @xmath512 and @xmath513 are contractible , it follows from lemma [ lm : ab_prop ] that @xmath517 and @xmath518 satisfy assumptions of corollary [ cor : michaeltheorm ] .
hence they admit continuous selections @xmath408 and these selections are the required cross sections .
this completes proposition [ pr : global_crs_outside_compact_set ] .
* proof of theorem [ th : trivialization_of_cross_section ] .
case @xmath547 .
* we need the following simple lemma whose proof we leave for the reader .
[ lm : double_of_open_set ] let @xmath548 be the involution defined by @xmath549 then for each subset @xmath550 open in the induced topology of @xmath120 , its @xmath551 is open in @xmath213 .
now let @xmath552 be the double of @xmath20 , i.e. the union of two copies @xmath553 and @xmath554 of @xmath20 glued along their boundaries by the identity map .
let also @xmath555 be the involution interchanging @xmath553 and @xmath554 by the identity map .
then the foliation @xmath6 on each of the copies of @xmath20 gives a one - dimensional foliation @xmath556 on @xmath557 .
moreover , let @xmath558 be the double of @xmath119 as in lemma [ lm : double_of_open_set ] .
then @xmath558 is open in @xmath213 and the cross section @xmath129 naturally extends to the cross section @xmath559 of @xmath556 such that @xmath560 and @xmath561 .
since @xmath562 , it follows from the boundary - less case of theorem [ th : trivialization_of_cross_section ] that the saturation @xmath563 is open in @xmath557 and foliated homeomorphic with @xmath564 . that homeomorphism induces a homeomorphism of the open subset @xmath565 of @xmath553 onto @xmath161 .
theorem [ th : trivialization_of_cross_section ] is completed .
the authors are sincerely grateful to olena karlova for pointing out to katetov - tong theorem about characterization of normal spaces which leads us to the proof of theorem [ th : trivialization_of_cross_section ] using e. michael selection theorem .
andr haefliger , _ sur les feuilletages des varits de dimension @xmath30 par des feuilles fermes de dimension @xmath570 _ , colloque de topologie de strasbourg , 19541955 , institut de mathmatique , universit de strasbourg .
, no year given , p. 8 .
louis f. mcauley , _ completely regular mappings , fiber spaces , the weak bundle properties , and the generalized slicing structure properties _ , topology seminar ( wisconsin , 1965 ) , ann . of math .
studies , no .
60 , princeton univ . press , princeton , n.j .
, 1966 , pp .
219227 .
georges reeb , _ les espaces localement numriques non spars et leurs applications un problme classique _ ,
colloque de topologie de strasbourg , 19541955 , institut de mathmatique , universit de strasbourg .
, no year given , p. 12 . | let @xmath0 be an @xmath1-dimensional manifold , @xmath2 be a one - dimensional foliation on @xmath0 , and @xmath3 be a quotient map .
we will say that a leaf @xmath4 of @xmath2 is _ special _ whenever the space of leaves @xmath5 is not hausdorff at @xmath4 .
we present necessary and sufficient conditions for the map @xmath3 to be a locally trivial fibration under assumptions that all leaves of @xmath2 are non - compact and the family of all special leaves of @xmath2 is locally finite . |
a network , or graph , consists of a set of nodes , from which selected pairs are connected by edges .
such mathematical constructions provide a useful description for systems of interacting objects .
more specifically , network concepts are used in the analysis of chemical and metabolic systems as well as food webs and social networks . in recent years , there has been much progress in the analysis of the topology of these networks .
the network topology can be characterized by features such as the number of nodes , @xmath0 , and the average degree @xmath1 $ ] , namely the average number of edges that are connected to a node . a more detailed description of the network topology is given by the degree distribution , @xmath2 , which is the probability that a randomly selected node has exactly @xmath3 edges .
another important topological feature measures the tendency of a network to support the formation of cliques .
a clique is a fully connected set of nodes , namely each pair of nodes in such a set is connected by an edge .
the tendency of a network to form cliques can be characterized by the clustering coefficient @xcite . roughly speaking ,
when a network has a high clustering coefficient it is considered to be highly connected .
a low clustering coefficient implies that the network is only loosely connected .
networks exhibit a unique metric , in which the distance , @xmath4 , between any two nodes is given by the minimal number of edges one has to cross in order to pass from one node to the other . in some cases
, the distance can be used as a measure for the connection between a pair of nodes .
this is based on the assumption that two directly reacting nodes ( @xmath5 ) strongly affect each other , whereas distant nodes weakly affect one another .
the average path length in a network , @xmath6 $ ] , is obtained by averaging over the distance between all pairs of nodes in the network .
the parameters defined above were evaluated for random graphs and their dependence on @xmath0 and @xmath7 was found @xcite .
however , the analysis of realistic networks shows that they are very different from random graphs @xcite . in realistic networks
it is common to find surprisingly low average path lengths , and relatively high clustering coefficients . in many cases
the degree distribution follows a power law form , rather than the poisson distribution which is the signature of random networks .
these features were found to appear in social networks @xcite , the world wide web @xcite , ecological networks @xcite , and metabolic networks @xcite . while the topological properties of realistic networks have been elucidated , the implications on the functionality of these networks are not fully understood . the small average path length and the high clustering of many realistic networks ,
render them as small world networks . at first glance ,
the small world characteristics imply that realistic networks function as highly connected systems .
indeed , one expects that if the distance between two nodes is small , the correlation between them will be strong .
for instance , in the case of a metabolic network , the concentrations of interacting proteins will strongly depend on each other .
a perturbation in the concentration of one protein is likely to affect the concentration of the other .
this might lead to the conclusion that small world networks are highly susceptible to local perturbations , as almost all the nodes are just a short distance away .
the problem with this topological analysis , is that it does not relate to the specific function of a given network or to the strength of the interactions between its nodes @xcite .
consider , for instance , a metabolic network and an ecological network sharing the same topology . in what sense
can these two networks be regarded as similar networks ? even if the two have the same topological structure , the nature of their functional behavior is fundamentally different .
the process of predation may lead to different behavior than the process of chemical reaction between proteins .
even two metabolic networks may function differently if the interaction strengths in one network are higher than in the other . in this paper
, we present a method for obtaining the correlation matrix of a given network .
the elements of this matrix provide the magnitudes of the correlations between pairs of nodes in the network . in certain cases
the matrix can be used to characterize some of the global features of the network s functionality .
for instance , it can be used to identify domains of high correlations versus domains of low correlations .
another use of the correlation matrix is in quantifying the connectivity of a network in a way that accounts both for its topology and for the specific processes taking place between its nodes .
this method , referred to as the network correlation function ( ncf ) method , enables us to determine whether a topological small world ( tsw ) network will also be a functional small world ( fsw ) network .
a network will be regarded as an fsw network if the correlations between its nodes are typically high , and thus the state of one node is highly dependent on that of the others . here
we apply the method to metabolic networks with various topologies and different interaction strengths . in these networks ,
each node represents a reactant , and is assigned a dynamical variable that accounts for the concentration of this reactant .
the time dependence of these concentrations is described by a set of rate equations .
the equations include terms that describe the interaction processes in the given network .
they account both for the topology and for the functionality of the network . from the solution of the rate equations under steady state conditions
one can extract the correlation between each pair of nodes . in certain cases ,
networks are found to have a typical correlation length .
if the distance between two nodes is much higher than this length , the correlation between them is negligible . to quantify the connectivity of the network ,
one compares the correlation length with the average path length . in case that the average path length is smaller than the typical correlation length
, the network will be considered as an fsw network . in this case
, local perturbations will have a global effect on the network .
the fsw network will thus be regarded as strongly connected . on the other hand ,
if the average path length is larger than the typical correlation length , the network will be considered as weakly connected .
the paper is organized as follows . in sec .
ii we present the methodology , and demonstrate its applicability to metabolic networks . in sec .
iii we analyze some simple , analytically soluble networks , and in sec .
iv we present a computational analysis of a set of more complex networks , culminating in an example of a scale free network .
the results are summarized and discussed in sec .
below we present the ncf method for evaluating the connectivity of interaction networks .
for concreteness , we focus on the specific case of metabolic networks .
it is straightforward to generalize the method to other types of networks .
consider a metabolic network consisting of @xmath0 different molecular species , @xmath8 , @xmath9 .
the generation rate of the @xmath8 molecules is @xmath10 ( s@xmath11 ) .
once a molecule is formed it may undergo degradation at a rate @xmath12 ( s@xmath11 ) .
certain pairs of molecules , @xmath8 and @xmath13 , may react to form a more complex molecule @xmath14 ( @xmath15 ) .
in general , the product molecules @xmath14 may be reactive and represented by another node in the network . for simplicity , in the analysis below
, we assume that the @xmath14 molecules are not reactive and thus do not play a further role in the network .
we also limit the discussion to the case in which a molecular species does not react with itself , namely reactions of the form @xmath16 are excluded .
the reaction rate between the @xmath8 and @xmath13 molecules is given by the _ reaction rate matrix _ @xmath17 .
its matrix elements are @xmath18 ( s@xmath11 ) , where @xmath19 .
note that for non - interacting pairs of molecules @xmath20 .
the _ network topology matrix _
, @xmath21 , is also a @xmath22 dimensional matrix , which is defined as follows : @xmath23ij = 1 $ ] if @xmath8 and @xmath13 react with each other , and @xmath23ij = 0 $ ] otherwise .
let @xmath24ij$ ] be the distance between the species @xmath8 and @xmath13 in the metric of the network .
the average path length is thus @xmath25 = \frac{1}{j(j-1 ) } \sum_{i , j=1}^{j}{\mat[d]ij}. \label{eq : average_distance}\ ] ] the parameter @xmath6 $ ] provides some information as to the connectivity of the network , but only in the topological sense . in order to account for the functionality of the network we consider the rate equations , which take the form @xmath26 where @xmath27 is the time dependent concentration of the molecule @xmath8 .
the first term on the right hand side of eq . ( [ eq : metabolic_rate ] ) accounts for the generation of @xmath8 molecules .
the second term accounts for the process of degradation , and the third term accounts for reactions between molecules .
the steady state ( ss ) solution of the rate equations , @xmath28 , can be obtained by setting the left hand side of eq .
( [ eq : metabolic_rate ] ) to zero .
one obtains @xmath29 where @xmath30 is the effective degradation rate .
our goal is to characterize the correlations between the different species around the steady state condition . roughly speaking
, we are asking the following question : while at steady state , to what extent does a small perturbation in the concentration of the species @xmath13 affect the concentration of the species @xmath8 ? to this end
we define the _ first order correlation matrix _ as @xmath31ij = \left .
{ \frac{\partial n_i}{\partial n_j } } \right| _ { \rm ss } , \label{eq:2ni_2nj}\ ] ] which , using eq .
( [ eq : metabolic_ss ] ) takes the form @xmath31ij = - { \frac { a_{ij } g_i}{({w_i^{\rm eff}})^2}}. \label{eq : cij}\ ] ] note that the elements of the first order correlation matrix are non - zero only if the species @xmath8 and @xmath13 directly interact with each other .
topologically , this means that the matrix element @xmath32ij$ ] vanishes unless @xmath24ij = 1 $ ] .
indirect correlations between species that are connected via a third species are not accounted for ( hence the term first order correlation matrix ) . to account for indirect correlations
, one has to compute the complete correlation matrix @xmath33ij = \left .
{ \frac{dn_i}{dn_j}}\right|_{\rm ss}. \label{eq : dni_dnj}\ ] ] clearly , the diagonal terms of this matrix must satisfy @xmath34 for @xmath9 . for the off - diagonal terms , @xmath35 ,
one can write @xmath36 in matrix form , these equations become @xmath37ii = 1 & & \\
\mat[g]ij = \sum_{k=1}^{j}{\mat[c]ik \mat[g]kj } & \,\ , & ( i \ne j ) . \end{array } \right . \label{eq : gij}\end{aligned}\ ] ] eq .
( [ eq : gij ] ) is a set of @xmath22 coupled linear equations .
their solution provides the complete correlation matrix , @xmath38ij$ ] .
typically , one expects the correlation between two species to decay as a function of the distance , @xmath24ij$ ] , between them .
the rate of this decay provides the correlation length . to obtain the correlation function we identify all pairs of species @xmath39 and @xmath40 that are separated by a distance @xmath4 from each other .
we then average the magnitude of the correlations , @xmath41ij|$ ] , over all these pairs .
the correlation function vs. distance takes the form @xmath42ij \right| \delta_{d,\mat[d]ij } } } { \sum_{i , j=1}^{j } \delta_{d,\mat[d]ij } } , \label{eq : g_of_d}\ ] ] where @xmath4 is an integer . the function @xmath43 if @xmath44 and zero otherwise .
note that in the definition of @xmath45 the absolute value of the matrix terms @xmath38ij$ ] was used .
this is because certain pairs of species @xmath8 and @xmath13 may be positively correlated , and others may be negatively correlated . in any case , the focus here is merely on the strength of their mutual correlations and not on the sign of these correlations . to obtain the correlation length
, one may fit the function @xmath45 to an exponent of the form @xmath46 .
the distance @xmath47 is the correlation length .
it approximates the distance within which strong correlations between different species are maintained .
this distance is determined by the dynamical processes and by the characteristic rate constants of a specific network .
it thus accounts not only for the topology of the system , but also for its functionality .
finally , we define the _ connectivity _ of a network as @xmath48}. \label{eq : eta}\ ] ] in the limit where @xmath47 is much greater than the average path length , most of the nodes are within the correlation length from one another , and the components of the network are highly correlated .
the concentrations of different species are strongly dependent on each other , and the network is an fsw network . correspondingly , one obtains that @xmath49 . in case
that @xmath47 is much smaller than the average path length , the effect of a perturbation in the concentration of one species decays on average before it reaches most of the other species .
perturbations are thus local , and the connectivity of the network is said to be low .
while topologically , such a network might be considered a small world network , functionally it is a loosely connected network .
to demonstrate the ncf method we now refer to a set of simple examples , which are analytically soluble . consider a linear metabolic network of @xmath0 species ( @xmath50 )
the species @xmath8 , @xmath51 , reacts with its nearest neighbors , namely @xmath52 and @xmath53 .
this network is shown in fig .
[ fig1_linear_network ] . for simplicity
, we take all the reacting species to have identical parameters , namely @xmath54 and @xmath55 for @xmath51 . also , @xmath56 in case that @xmath57 , and @xmath20 otherwise . taking the limit
in which the number of species @xmath0 is very large , we can avoid the complexities related to the boundaries of the network . under these conditions , the steady state solution for all the species is the same , enabling us to omit the index @xmath39 from the steady state concentrations @xmath28 .
the reaction rate matrix for this network is @xmath58 for a linear network , the average distance between pairs is @xmath6 = ( j+1)/3 $ ] , which for @xmath50 , can be approximated by @xmath25 = \frac{j}{3 } , \label{eq : av_d_linear}\ ] ] namely , @xmath6 $ ] scales linearly with @xmath0 .
the clustering coefficient for this network is zero .
thus , from the topological point of view , the linear network can not be considered a small world .
the rate equation for the linear metabolic network is @xmath59 leading to the steady state solution @xmath60 the first order correlation matrix takes the form @xmath61 where @xmath62 [ eq . ( [ eq : cij ] ) ] . using eq .
( [ eq : ss_linear ] ) , one obtains @xmath63 since the parameters @xmath64 , @xmath65 and @xmath66 are positive , it is easy to see that @xmath67 takes values only in the range @xmath68 .
this fact will be used in the analysis below . to obtain the complete correlation matrix
, one has to solve eq .
( [ eq : gij ] ) . in the case of a linear metabolic network
it takes the form @xmath37ii = 1 & & \\
\mat[g]ij = q(\mat[g]{i+1}{,j } + \mat[g]{i-1}{,j } ) & \,\ , & ( i \ne j ) . \end{array } \right .
\label{eq : gij_linear}\end{aligned}\ ] ] based on the symmetry of the problem , it is clear that for a given choice of the parameters , the correlation between the species @xmath8 and @xmath13 depends only on the distance @xmath69 between them . using this indexation , eq .
( [ eq : gij_linear ] ) becomes @xmath70 & \,\ , & ( d \ge 1 ) , \end{array } \right .
\label{eq : gd_linear}\end{aligned}\ ] ] where @xmath71 is the correlation matrix term for pairs of species @xmath8 and @xmath13 where @xmath72 . since the correlation is expected to decay exponentially as a function of the distance between the nodes , we search for a solution of the form @xmath73 . inserting this expression into eq .
( [ eq : gd_linear ] ) we obtain two possible solutions of the form @xmath74 where @xmath75 . since the parameter @xmath67 is limited to the range @xmath68 , the parameter @xmath76 can take values only in the range @xmath77 .
the physically relevant solution must satisfy the condition that the correlation between very distant species will vanish .
this constraint requires that @xmath78 . to satisfy this condition for @xmath68
, one has to choose the solution where the square root is subtracted .
the result is @xmath79 where @xmath80 .
the correlation between species as a function of the distance @xmath4 between them is thus @xmath81.\ ] ] the pre - factor of the exponent accounts for the fact that since @xmath82 , the correlations between directly interacting species are negative .
thus , pairs of species which are next - nearest neighbors in the network tend to have positive correlations between them .
the correlation function [ eq . ( [ eq : g_of_d ] ) ] is the absolute value of @xmath71 , which comes to be @xmath83 where @xmath84 is the correlation length of the network .
it is interesting to examine the limit in which @xmath85 . in this limit
the correlations are weak and the typical correlation length converges to @xmath86 .
the correlation function approaches @xmath87 . in this limit ,
the correlation between a pair of species is dominated by the shortest path between them . for each step along that path ,
the correlation is multiplied by a factor of @xmath67 .
thus , the magnitude of the correlation between a pair of species at distance @xmath4 from each other is approximated by @xmath88 .
one can identify two limits . in the limit where @xmath89 the correlations are strong , @xmath90 and @xmath91 . in this limit , the reaction process is dominant and long range correlations are observed . in the limit where @xmath92 , the correlations are weak , and @xmath85 . in this limit
the degradation process is dominant and the correlation length is small . in fig .
[ fig2_do_metabolic_graph ] we present the correlation length , @xmath47 , as a function of the parameters @xmath64 , @xmath66 and @xmath65 for a linear metabolic network .
the correlation length increases with @xmath64 and @xmath65 ( as the reaction process becomes dominant ) , and decreases with @xmath66 ( as the process of degradation becomes dominant ) . using eqs .
( [ eq : eta ] ) and ( [ eq : av_d_linear ] ) , the connectivity @xmath93 can be expressed by @xmath94 .
the linear network clearly demonstrates the difference between the concepts of tsw networks and fsw networks . in the topological sense it is as far as a network
can be from a small world network , as the distance scales linearly with the network size , and the clustering coefficient is zero .
however , in the functional sense the linear network can be a small world network , when the reaction terms are sufficiently dominant , enabling @xmath47 to become larger than @xmath0 . in order to examine the theoretical predictions of the method
, we conducted a simulation of the long linear metabolic network described above .
in this simulation we constructed a linear network of @xmath95 reacting species with periodic boundary conditions , namely , @xmath96 reacts with @xmath97 . at time @xmath98 we assigned to each reacting species its steady state concentration @xmath28 .
then we forced the concentration @xmath99 to be slightly above its steady state value , namely @xmath100 , where @xmath101 .
we then let the network relax to its new steady state .
we denote the resulting change in the steady state concentration of the species @xmath8 by @xmath102 . in fig .
[ fig3_perturbation ] we show the absolute value of @xmath103 as a function of @xmath4 , the distance of the node @xmath8 from the perturbed node , @xmath96 .
these results , obtained from direct integration of the rate equations , are shown for different values of the reaction rate @xmath64 ( symbols ) .
when @xmath64 increases the typical correlation length becomes higher , and the effect of the local perturbation of @xmath96 extends to more distant species .
the results are in good agreement with the theoretically derived correlation function , @xmath45 [ eq . ( [ eq : fcor_linear ] ) ] ( solid lines ) .
slight deviations appear for distant species .
this is because in numerical simulations one must choose @xmath104 to be a finite perturbation .
the resulting deviation in the rest of the species is thus affected by higher order terms in the taylor expansion which are not accounted for by our method . here
the generation rates and the degradation rates of all the species are @xmath105 and @xmath106 respectively .
the network becomes an fsw network once @xmath107 , which is approximately the average path length for this network .
this condition is satisfied for @xmath108 .
hierarchical structures are common in realistic networks . for instance , ecological networks have in many cases distinct trophic levels .
social organizations are also constructed in a tree - like framework .
here we relate to a hierarchical metabolic network .
consider a metabolic network of @xmath0 nodes where each node is assigned a level @xmath109 ( @xmath110 ) .
the highest level @xmath111 consists of a single node , referred to as the _
root_. each node at level @xmath109 is then connected to exactly one node at level @xmath112 ( the _ parent _ ) and @xmath113 nodes at level @xmath114 ( the _ siblings _ ) .
the parameter @xmath113 is defined as the order of the tree .
the degree of all the nodes ( except those at the levels zero and @xmath115 ) is thus @xmath116 ( fig .
[ fig4_tree_network ] ) .
since this network is hierarchical , the up and down directions are well defined . stepping from a node at level @xmath109 to a node at level
@xmath112 will be considered going up the network , while stepping from level @xmath109 to level @xmath114 is going down the network .
note that in a tree - like network it is not possible to go sideways , as there is no edge connecting two nodes at the same level .
consider a species @xmath8 , which is at a distance @xmath4 from some other species @xmath13 .
the path between them consists of @xmath117 steps up the network and @xmath118 steps down the network .
the total distance satisfies @xmath119 , and the path between them can be noted by @xmath120 .
for example , the path between the two shaded nodes in fig .
[ fig4_tree_network ] is @xmath121 and the distance is @xmath122 .
two species are said to be located in the same branch if in the path between them either @xmath123 or @xmath124 .
the reaction rate matrix and the first order correlation matrix have non - zero values only for directly interacting species , namely , for pairs of species where either @xmath125 and @xmath124 , or @xmath126 and @xmath123 . in order to avoid the complexities related to the boundaries of the network , we consider the case in which @xmath127 . for simplicity
, we take the generation and the degradation rates to be @xmath128 and @xmath129 for @xmath130 .
the reaction rate is @xmath56 for each pair of nodes @xmath39 and @xmath40 that react with each other . under these conditions , the network is symmetrical and the rate equations are identical for all nodes : @xmath131 the steady state solution is thus @xmath132 and the non - zero elements in the first order correlation matrix [ eqs .
( [ eq:2ni_2nj ] ) and ( [ eq : cij ] ) ] are @xmath133 two limits are observed . in the limit of strong interactions , where @xmath134 , the matrix elements approach @xmath135 . in the limit of weak interactions , where @xmath136 one obtains @xmath137 . in any case
the values that @xmath67 can take are limited to @xmath138 .
for an infinite perfect tree with uniform rate constants , the correlation between all pairs of species with that same values of @xmath117 and @xmath118 are the same .
we denote this correlation by @xmath139 . in each line @xmath39 of the first order correlation matrix there are exactly @xmath140 non - zero terms .
one term for @xmath8 s parent and @xmath113 terms corresponding to @xmath8 s siblings .
the correlation @xmath139 between two species @xmath8 and @xmath13 is thus carried via the the parent of the species @xmath8 , for which the correlation with @xmath13 is @xmath141 , and via the @xmath113 siblings of the species @xmath8 , for which the correlation with @xmath13 is @xmath142 .
( [ eq : gij ] ) thus takes the form @xmath143 & & { \rm for } & v>0 \\ g_{u ,
v } = q(g_{u - 1,v } + m g_{u + 1,v } ) & & { \rm for } & u > 0 \end{array } \right .
\label{eq : gij_tree}\end{aligned}\ ] ] the first equation states that the correlation of every species with itself is unity .
the second equation accounts for the correlations between species at the same branch , measuring the effect of variation in the higher level node on a node at a lower level .
the third equation accounts for all the correlations that are not included in the first two equations .
more specifically , it includes the correlations between species from different branches .
it also includes the correlations between pairs on the same branch , measuring the effect of variation in the lower level node on a node at a higher level .
we seek a solution of the form @xmath144 , where @xmath145 satisfies the condition that correlations vanish between distant species . from the third equation one
obtains @xmath146 } , \label{eq : k1_tree}\ ] ] while from the second equation one obtains @xmath147 \right\ } } , \label{eq : k2_tree}\ ] ] where @xmath148 . in order to satisfy the conditions that @xmath139 does not diverge for @xmath149 while @xmath138 , one has to choose the solution with the plus sign for @xmath150 in eq .
( [ eq : k1_tree ] ) . the same condition for @xmath151 requires one to choose the solution with the plus sign for @xmath152 in eq .
( [ eq : k2_tree ] ) . after some algebraic manipulations it can be shown that @xmath153 .
the correlation between any pair of species is thus @xmath154 where @xmath119 is the distance between the two species , and @xmath155 is the correlation length of the tree - like network .
the correlation function is @xmath156 .
note that for @xmath157 ( @xmath158 ) this solution coincides with the solution obtained for the linear network [ eq .
( [ eq : k_linear2 ] ) ] . in the limit of weak interactions , where @xmath136 and @xmath85 the correlation function approaches @xmath159 . in this limit , due to the weak interactions ,
the correlation between a pair of species is dominated by the shortest path between them . in the limit of strong interactions , where @xmath134 and @xmath160 , the correlation length satisfies @xmath161 .
for @xmath162 the correlation length is always finite . since the average path length of a perfect tree - like network must scale in some form with the number of levels in the tree , one obtains that for a large enough tree network the connectivity will always be less than unity .
thus a perfect tree - like network of order @xmath163 or more will never be an fsw . in fig .
[ fig5_d0_vs_a ] we show the correlation length @xmath47 as obtained for a metabolic network with a perfect tree topology vs. the reaction rate @xmath64 ( symbols ) .
the results are shown for trees of different orders .
here @xmath164 , and @xmath64 is varied .
to demonstrate the applicability of the ncf method , we now refer to the analysis of a set of more complex networks . here
analytical solutions are not available , and the correlation matrix must be obtained numerically .
we analyze three different topologies following the structural classification proposed by estrada @xcite .
the first example represents a class of networks which are organized into highly connected modules with few connections between them .
the second example will be of a network with a highly connected central core surrounded by a sparser periphery , and the last example will be of a scale - free network .
consider a network constructed of three fully connected modules ( communities ) , with a single connection between each pair of communities .
this network is displayed in fig .
[ fig6_three_joint_clusters](a ) . here , each community consists of @xmath165 nodes , adding up to a total of @xmath166 nodes .
to obtain , @xmath28 , @xmath167 , the steady state solution for the concentrations of the different reacting species we solve eq .
( [ eq : metabolic_rate ] ) using a standard runge - kutta stepper .
the parameters we use are @xmath128 and @xmath129 .
the reaction rate @xmath64 between pairs of reacting species is also set to unity .
we then construct the first order correlation matrix , @xmath32ij$ ] , as appears in eq .
( [ eq : cij ] ) . the complete correlation matrix , @xmath38ij$ ] , is obtained from eq .
( [ eq : gij ] ) .
it consists of a set of @xmath168 linear algebraic equations . solving these equations
, one obtains the complete correlation matrix of the network . for this network , the main insight on the global functions of the network can be deduced from the complete correlation matrix , which is displayed in fig . [ fig6_three_joint_clusters](b ) .
the diagonal terms , which are all unity , are omitted from the figure . as expected , strong correlations appear between species within the same community ( sub - matrices along the diagonal ) , and vanishingly small correlations appear between species from different communities .
in fact , the correlation matrix is close to be a partitioned block matrix , except for a few coupling terms between the blocks . in this case
the correlation matrix reflects the topological structure of the network , which is almost fully partitioned into three isolated communities .
we now consider a network , which features a highly connected central core surrounded by a sparser periphery .
this network consists of @xmath169 nodes .
the nodes @xmath170 , @xmath171 , are a fully connected cluster ( the core ) , while the @xmath172 additional nodes are connected to all the nodes in the core , but not to each other ( the periphery ) .
this network is shown in fig . [
fig7_core_and_periphary](a ) . following the same procedure described above one
obtains the correlation matrix for this network [ fig .
[ fig7_core_and_periphary](b ) ] .
the central square ( domain i ) shows the correlations between the nodes in the central core .
domains ii show the correlations between peripheral nodes and central ones .
the value of these correlations is high , expressing the strong dependence of the peripheral nodes on the nodes in the central core . on the other hand for the correlations between the central nodes and the peripheral ones ( domains iii )
one obtains very low correlations .
this is an expected result , as deviations in the population of a node from the periphery should have almost no effect on a node from the core .
an interesting result appears in domains iv .
these domains show the correlations between pairs of nodes that are both from the periphery .
it turns out that the effect of these nodes on each other is stronger than the effect they have on their adjacent nodes from the core .
this is even though the topological distance between peripheral nodes is @xmath173 , while the distance between them and the central nodes is @xmath5 . a small perturbation in a peripheral node results in a very minor effect on all the central nodes .
however this minor change in the core results in a more dramatic effect on all the rest of the peripheral nodes .
this non - trivial result exemplifies the importance of the functional methodology as a complimentary analysis to the common topological approach . in the two examples shown above , we focused on the insights provided by the complete correlation matrix .
below we show an additional numerical example , where we continue the analysis to obtain the correlation length , @xmath47 , and the connectivity @xmath93 .
one of the common characteristics of many realistic networks is their degree distribution that follows a power law , namely @xmath174 , where @xmath175 and @xmath176 are positive constants @xcite .
ecological networks , social networks and metabolic networks are characterized by power - law degree distributions , and are referred to as scale - free networks .
such networks include some nodes , called hubs , with a degree which is orders of magnitude higher than the average degree in the network .
scale free networks are considered as highly connected , because due to these hubs the average path length between nodes is small .
in fact , in metabolic networks the average path length was found to be as small as @xmath6 \simeq 3 $ ] @xcite .
below we examine a scale free network which is a tsw network , and determine whether it is also an fsw network . to construct a scale free network we use the preferential attachment algorithm @xcite . in this algorithm
a single new node is added at each iteration and @xmath113 edges are drawn from it to the set of existing nodes .
the probability of linking the new node to some existing node @xmath8 is proportional to the current degree of the node @xmath8 . this way , nodes which already have a higher degree than others have a high probability of obtaining more links and becoming hubs .
here we constructed a scale free network consisting of @xmath177 nodes .
the number of edges added in each iteration was @xmath178 .
the result is the graph appearing in fig .
[ fig8_scale_free_network ] .
the diameter of this network is @xmath179 and its average path length is @xmath6 = 2.43 $ ] . solving eq .
( [ eq : metabolic_rate ] ) we obtain the steady state solution for the concentrations of all the reactive species .
the parameters are @xmath128 and @xmath129 for @xmath180 .
the reaction rate @xmath64 between pairs of reacting species is varied . in this case , obtaining the complete correlation matrix , @xmath38ij$ ] , requires the solution of @xmath181 linear algebraic equations [ eq .
( [ eq : gij ] ) ] .
we solve these equations and then average over the correlations between equidistant species to obtain the correlation function @xmath182 [ eq .
( [ eq : g_of_d ] ) ] . in fig .
[ fig9_g_of_d_scale_free ] we show the resulting correlation function @xmath182 vs. @xmath4 for three different values of the reaction rate @xmath64 ( symbols ) . when the interaction is suppressed ( @xmath136 ) the correlations decay rapidly .
when the interaction is dominant ( @xmath134 ) , correlations are maintained over long distances . by fitting the correlation functions to exponential functions ( solid lines )
one obtains the typical correlation length , @xmath47 , and the connectivity , @xmath93 , of each of the networks .
the results for @xmath93 vs. the reaction rate @xmath64 are shown in fig .
[ fig10_eta_vs_a_scale_free ] .
it is found that the connectivity increases logarithmically with @xmath64 .
note that for a very wide range of values of the parameter @xmath64 , the connectivity remains lower than unity .
this means that although the examined scale free network is a tsw , for a very wide range of parameters it is not an fsw . only in the extreme cases of very strong interactions
fsw behavior might emerge .
we have presented the ncf method for the analysis and evaluation of the connectivity of interaction networks .
the method complements the topological analysis of networks , taking into account the functional nature of the interactions and their strengths .
the method enables to obtain the correlation matrix , which provides the correlations between pairs of directly and indirectly interacting nodes . in certain cases ,
one may gain insights on the network s functionality by writing down the complete correlation matrix .
for instance , one can identify domains of high and low correlations .
in other cases it is more insightful to extract the macroscopic characteristics of the network from the matrix .
in particular , we have shown how to calculate the typical correlation length of the network .
this correlation length , which has to do with the functionality of the network , can be compared to topological characteristics such as the average minimum path length of the network .
the ratio between these two lengths provides the characteristic connectivity of the network .
it was shown that the topological analysis alone is not sufficient in order to characterize the functionality of a network .
for instance , networks with small world topology may display low connectivity , while networks that do not exhibit small world topology may display high connectivity .
this is because in terms of the functionality of the network , when the correlation length is large , even distant species may be highly correlated .
we demonstrated the method for metabolic networks with different topological structures , and identified the regimes of low connectivity and of high connectivity . as expected , these regimes depend on topological features , such as the number of species or the average minimum path length between pairs of species .
however , they also depend on functional features such as the type of interactions in the network and the rate constants of the different processes .
the ncf method was demonstrated for metabolic networks , but its applicability is much wider .
in fact , the method could be applied to any reaction network that can be modeled by rate equations .
such networks include metabolic networks @xcite , chemical networks @xcite , gene expression networks @xcite and ecological networks @xcite .
it is common to use rate equations for the modeling of these types of networks . in certain models of social networks ,
the flow of information as well as the spreading of viruses can also be described by rate equations .
the method is not suitable for obtaining the correlations in ising type models , where the nodes are assigned discrete variables , which can not be modeled using continuous equations .
the number of elements in the correlation matrix is equal to the number of pairs of nodes in the system .
when applying the ncf method , one writes a single linear equation for each matrix element .
thus , from a computational point of view , the scaling of the ncf method is quadratic in the number of reactive species .
this enables the application of the method to networks which include even thousands of nodes .
it is straightforward to extend the application of the method to the other types of interaction networks mentioned above .
a few examples are addressed below .
consider , for example , gene expression networks .
these networks consist of genes and proteins that interact with each other .
in addition to protein - protein interactions , already analyzed in the context of metabolic networks , genetic networks include transcriptional regulation processes , where some genes regulate the expression of other genes . in recent years
, much information has been acquired about the topology of these networks , for certain organisms such as _ escherichia coli _ @xcite .
the problem is that these networks are very elaborate , and may consist of thousands of nodes .
this limits our ability to simulate their functionality , and thus , currently most of the theoretical and computational analysis of these networks is focused on small modules @xcite . in this analysis ,
one performs simulations of small subnetworks consisting of only a few nodes .
these subnetworks are expected to play specific roles in the functionality of the network as a whole .
such approach is valid if an isolated module maintains its function when incorporated in a large network in which it interacts with many other genes .
we expect the analysis presented here to provide some insight on this matter . by obtaining the complete correlation matrix
, one can characterize the dependence of different proteins and genes on one another .
the network may then be divided into subnetworks , grouping together nodes that are highly correlated , and excluding ones that are not .
it is expected that these modules will not function significantly differently when analyzed in the context of the surrounding network nodes .
in addition , the typical correlation length will provide us with an approximate radius beyond which correlations may be neglected . to simulate a module properly
, one needs to include all the nodes which are within that radius from the module .
other possible applications regard social networks .
for instance , the process of viral spreading could be analyzed @xcite .
many social networks are known to be small world networks .
however , this does not mean that any contagious disease spreads rapidly .
this is , possibly , because for certain diseases the correlation length is small .
using the method presented here , one can obtain this correlation length , taking into account the specific rate constants of the viral flow .
the recent applications of graph theory to many natural macroscopic systems was enabled by focusing on their topology .
this approach has been very fruitful , as it uncovered the mutual structure of networks from many different fields . in particular , the ubiquity of the scale free degree distribution , and the small world topology was found .
however , it still is not completely clear what functional meaning can be given to these topological properties in different contexts .
a recently proposed approach derives the key aspects of the network functionality from its topological structure @xcite .
other approaches use the ising hamiltonian to describe the interaction pattern between nodes on scale free and small world networks @xcite . functional characteristics such as phase transitions , and critical exponents
are then observed .
the ncf method presented in this paper complements these approaches .
it can be applied to a variety of different interaction processes , such as metabolic , ecological or social interactions , all of which can be described by rate equations .
we believe that the approach presented here will lead to new insights on the behavior of networks and their functionality . , of the linear metabolic network versus the generation rate , @xmath65 ( a ) ; the degradation rate , @xmath66 ( b ) ; and the reaction rate , @xmath64 ( c ) .
high connectivity is reached when the primary process is the reaction process ( proportional to @xmath65 and @xmath64 ) .
the correlation length @xmath47 decreases with increasing @xmath66 ( as the degradation becomes dominant ) .
, width=453 ] for the linear metabolic network as obtained from a numerical simulation for different values of the interaction rate @xmath64 ( symbols ) . to conduct the numerical test we integrate the equations for the linear network and bring them to the steady state condition
. then we force a small perturbation @xmath104 on the concentration @xmath99 of the species @xmath96 .
we evaluate the correlation function using @xmath183 . the correlations decay exponentially with the distance between species .
the typical correlation length increases as the reaction rate is increased .
the results are in agreement with the theoretical results of eq .
( [ eq : fcor_linear ] ) ( solid lines ) .
slight deviations appear due to the fact that in numerical simulations @xmath104 must be finite .
, width=453 ] levels .
each node is linked with exactly one node at the level above it ( father ) , and @xmath113 nodes at the level below ( siblings ) .
the top node ( here at level @xmath184 ) is the root node .
the order of the tree is @xmath185 , and the degree of the nodes is @xmath186 .
the path between a pair of nodes is characterized by the number of upward steps followed by the number of downward steps to get from one node to the other . for the path between the two shaded nodes @xmath121 .
, width=453 ] versus the reaction rate @xmath64 for a metabolic network with a perfect tree structure .
the results are shown for trees of different order , @xmath113 ( symbols ) .
for @xmath158 the results coincide with those obtained for the linear network . for higher orders
the correlation length is bound from above by @xmath187 ( gray horizontal lines ) .
thus , for a sufficiently large tree - like network , where the average path length is larger than @xmath188 , the connectivity is always less than unity .
tree like networks are thus not expected to display fsw behavior .
, width=453 ] for different values of the parameter @xmath64 ( symbols ) .
the correlations decay rapidly for low values of @xmath64 , and more gradually for large values of @xmath64 . by fitting the correlation function to an exponential ( solid lines ) , the correlation length @xmath47 can be obtained .
, width=453 ] vs. @xmath64 as obtained for the scale free network shown in fig .
[ fig8_scale_free_network ] .
the connectivity increases logarithmically as a function of @xmath64 .
note that @xmath189 for a very broad range of values of the parameter @xmath64 .
this implies that although scale - free networks are commonly tsw networks , in the functional sense they may not be fsw networks .
, width=453 ] | networks are useful for describing systems of interacting objects , where the nodes represent the objects and the edges represent the interactions between them . the applications include chemical and metabolic systems , food webs as well as social networks .
lately , it was found that many of these networks display some common topological features , such as high clustering , small average path length ( small world networks ) and a power - law degree distribution ( scale free networks ) .
the topological features of a network are commonly related to the network s functionality .
however , the topology alone does not account for the nature of the interactions in the network and their strength . here
we introduce a method for evaluating the correlations between pairs of nodes in the network .
these correlations depend both on the topology and on the functionality of the network .
a network with high connectivity displays strong correlations between its interacting nodes and thus features small - world functionality .
we quantify the correlations between all pairs of nodes in the network , and express them as matrix elements in the correlation matrix . from this information
one can plot the correlation function for the network and to extract the correlation length .
the connectivity of a network is then defined as the ratio between this correlation length and the average path length of the network . using this method we distinguish between a topological small world and a functional small world , where the latter is characterized by long range correlations and high connectivity .
clearly , networks which share the same topology , may have different connectivities , based on the nature and strength of their interactions .
the method is demonstrated on metabolic networks , but can be readily generalized to other types of networks . |
how galaxies build up their stellar mass is a central question in galaxy formation . within the broadly successful cold dark matter scenario
, gas is believed to accrete gravitationally into growing dark matter halos , and then radiative processes enable the gas to decouple from non - baryonic matter and eventually settle into a star - forming disk ( e.g. * ? ? ?
however , many complications arise when testing this scenario against observations , as the stellar assembly of galaxies involves a host of other processes including star formation , kinetic and thermal feedback from various sources , and merger - induced activity , all of which are poorly understood in comparison to halo assembly .
a key insight into gas accretion processes is the recent recognition of the importance of cold mode " accretion @xcite , where gas infalling from the intergalactic medium ( igm ) does not pass through an accretion shock on its way to forming stars .
simulations and analytic models show that cold mode dominates global accretion at @xmath14 @xcite , and dominates accretion in all halos with masses @xmath15m@xmath16 @xcite .
the central features of cold mode accretion are that it is ( 1 ) _ rapid _ ( 2 ) _ smooth _ , and ( 3 ) _ steady_. it is rapid because it is limited by the free - fall time and not the cooling time ; it is smooth because most of the accretion occurs in small lumps and not through major mergers @xcite ; and it is steady because it is governed by the gravitational potential of the slowly growing halo .
a consequence of cold mode accretion is that the star formation rate is a fairly steady function of time ( e.g. * ? ? ?
* ; * ? ? ?
* ) , which results in a tight relationship between the stellar mass @xmath2 and the star formation rate sfr .
hence simulations of star - forming galaxies generically predict a tight relationship with @xmath17sfr that evolves slowly with redshift . in detail ,
a slope slightly below unity occurs owing to the growth of hot halos around higher - mass galaxies that retards accretion .
the stellar mass star formation rate ( @xmath0sfr ) relation for star forming galaxies has now been observed out to @xmath18 , thanks to improving multiwavelength surveys .
as expected from models , the relationship is seen to be fairly tight from @xmath19 , with a slope just below unity and a scatter of @xmath20 dex @xcite .
it also evolves downwards in amplitude to lower redshift in lock step fashion , suggesting a quiescent global quenching mechanism such as a lowering of the ambient cosmic density , again as expected in models .
the range of data used to quantify this evolution is impressive : @xcite used the aegis multi - wavelength survey in the extended groth strip to quantify the @xmath2-sfr relation from @xmath21 ; @xcite used goods at @xmath22 and sdss spectra at @xmath23 ; and @xcite used star - forming bzk - selected galaxies with _ spitzer _ , x - ray , and radio follow - up to study the relation at @xmath24 . in each case , a careful accounting was done of both direct uv and re - radiated infrared photons to measure the total galaxy sfr , paying particular attention to contamination from active galactic nuclei ( agn ) emission ( discussed in [ sec : syst ] ) .
multiwavelength data covering the rest - optical were employed to accurately estimate @xmath2 . this broad agreement in @xmath0sfr slope , scatter , and qualitative evolution between model predictions and these data lends support to the idea that cold mode accretion dominates in these galaxies .
yet all is not well for theory .
closer inspection reveals that the _ amplitude _ of the @xmath0sfr relation evolves with time in a way that is inconsistent with model expectations .
this disagreement is fairly generic , as shown in
[ sec : comparison ] , arising in both hydrodynamic simulations and semi - analytic models , and is fairly insensitive to feedback implementation .
the sense of the disagreement is that going to higher redshifts , the observed sfrs are higher , and/or stellar masses lower , than predicted in current models .
the purpose of this paper is to understand the implications of the observed @xmath0sfr relation for our theoretical view of how galaxies accumulate stellar mass . in
[ sec : tact ] it is argued that @xmath0sfr amplitude evolution implies a typical galaxy star formation history that is difficult to reconcile with not only model expectations , but also other observations of high-@xmath9 galaxies .
possible systematic effects that may bias the estimation of sfr and @xmath2 are considered in [ sec : syst ] , and it is argued that none of them are obvious candidates to explain the discrepancy .
finally a speculative avenue for reconciliation is considered , namely that the stellar initial mass function ( imf ) has a characteristic mass that evolves with redshift ( [ sec : imf ] ) .
this model is constrained based on the @xmath2-sfr relation in
[ sec : imfmodel ] .
the implications for such an evolving imf are discussed in [ sec : mchar ] .
results are summarized in [ sec : summary ] . a @xmath25cdm cosmology with @xmath26 and @xmath27
km / s / mpc is assumed throughout for computing cosmic timescales .
to begin , the observed @xmath0sfr relation is compared with model predictions , in order to understand the origin of the relation .
the primary simulations used are cosmological hydrodynamic runs with gadget-2 @xcite incorporating momentum - driven outflows , which uniquely match a wide variety of igm and galaxy observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the new simulations include stellar recycling assuming a chabrier imf , by returning gas ( and metals ) into the ambient reservoir , in addition to the usual heating , cooling , and star formation , and feedback recipes @xcite .
outflow parameters are computed based on properties of galaxies identified with an on - the - fly group finder .
these and other improvements that go towards making simulations more realistic down to @xmath13 are described in @xcite .
two runs to @xmath13 are considered here , each with 256@xmath28 dark matter and an equal number of gas particle , in cubic volumes of 32 and 64@xmath29 ( comoving ) on a side .
the assumed cosmology is @xmath25cdm with @xmath30 , @xmath31 , @xmath32 km / s / mpc , @xmath33 , and @xmath34 .
the resolved galaxy population , defined here as those with stellar masses greater than 64 gas particle masses , is limited by stellar masses of @xmath35 and @xmath36 in the two volumes ( this is somewhat more conservative than in * ? ? ?
* in order to ensure accurate star formation histories )
. to explore the impact of feedback parameters , simulations with two other outflow models are considered , namely one with no outflows , and one with the @xcite constant wind " outflow model , taken from the suite of outflow runs described in @xcite .
these runs have 32@xmath29 box sizes with @xmath37 particles , and they were evolved only down to @xmath38 .
their cosmology is slightly different , having @xmath39 from first - year wmap parameters , but this is not a large effect at the epochs considered here . in hydro simulations , star formation rates and stellar masses
are obtained by summing over particles in galaxies identified using the group finder skid ( spline kernel interpolative denmax ; http://www-hpcc.astro.washington.edu/tools/skid.html ) ; see @xcite for details .
gas particles that have high enough density to form stars have instantaneous star formation rates computed in gadget-2 ; it is this sfr that is summed to obtain the instantaneous sfr of a galaxy . of course , observed galaxy sfr s and @xmath2 s are measured quite differently than in simulations , a point that will figure prominently in [ sec : syst ] .
note that these hydro simulations do not include any feedback from active galactic nuclei ( agn ) , or any process that explicitly truncates star formation in massive galaxies , hence they do not reproduce the evolution of passive galaxies . here
the focus is on star - forming galaxies only , with the assumption that galaxies are quickly squelched to become passive and so do not appear in star - forming selected samples ( e.g. * ? ? ?
since all galaxies in our simulations would be observationally classified as star - forming , no cuts are made to the simulated galaxy population . to complement the hydro simulations , results for @xmath0sfr
are taken from the millenium simulation plus semi - analytic model ( millenium sam " ) lightcones of @xcite , as fit by @xcite and @xcite .
besides being a completely different modeling approach , the millenium sam also includes agn feedback that broadly reproduces the observed passive galaxy population and its evolution .
additionally , a different semi - analytic model based on the millenium simulation by @xcite was examined , to see if variations in semi - analytic prescriptions can result in significant differences in @xmath0sfr .
it was found that the @xcite model produced a somewhat shallower slope than the @xcite model , which e.g. at @xmath8 resulted in a smaller offset from observations at @xmath40 , but a larger one at @xmath41 . since the results are qualitatively similar , only the @xcite models are discussed further . while there are many more semi - analytic models in the literature , and perhaps even some that are in better agreement with the @xmath0sfr relation , it is beyond the scope of this work to consider them all .
-0.5 in -5.6 in -1.3 in figure [ fig : sfrmstar ] ( first three panels ) shows the @xmath0sfr relation at @xmath42 in simulations , compared to parameterizations from observations ( blue lines ) by @xcite ( @xmath43 ) and @xcite ( @xmath44 ) .
the observed scatter of 0.3 dex @xcite is indicated by the dashed blue lines .
it is worth noting that other @xmath8 observations suggest a larger scatter @xcite , so the observational picture on this is not entirely settled .
finally , the green lines show the @xcite millenium sam results , which generally track the hydro simulation results , particularly in amplitude which will be the key aspect . at @xmath13 , the relations are generally in agreement , though the observed slope of @xmath17sfr@xmath45 is somewhat shallower than that predicted by the hydro simulations ( @xmath46 ) .
the millenium sam produces excellent agreement at @xmath13 , having broadly been tuned to do so . by @xmath47 ,
the predicted hydro simulation slopes are similar to that observed by @xcite though steeper than seen by @xcite , but more importantly the amplitude is low by @xmath48 . at @xmath38 ,
this trend continues , with a @xmath49 amplitude offset .
the millenium sam has a slope that remains near the @xmath13 value , hence it is somewhat shallower than observed at @xmath8 , but the amplitude evolves slowly much like in the hydro simulations .
the rapid evolution in observed @xmath0sfr amplitude has been noted by several authors @xcite , along with the fact that it is more rapid than in the millenium sam @xcite .
the new aspect demonstrated here is that the disagreement persists for current hydrodynamic simulations , and as will be shown in the next section this is true regardless of the main free parameter in such models , namely feedback implementation .
the slope and scatter are a direct reflection of the similarity in the shape of galaxy star formation histories ( sfh ) among various models and at various masses . for a galaxy observed at time @xmath50 , neglecting mass loss from stellar evolution ,
@xmath51 the integral depends purely on the _ form _ of the sfh up to @xmath50 .
so long as the sfh form does not vary significantly with @xmath2 , the slope of @xmath0sfr will be near unity , and so long as the current sfr is similar to the recent past - averaged one , the scatter will be small .
as pointed out by @xcite , the observed tight scatter of @xmath52 dex implies that the bulk of star - forming galaxies can not be undergoing large bursts ; this will be discussed further in [ sec : tact ] . this is consistent with a fairly steady mode of galaxy assembly as generically predicted by cold mode accretion .
note however that the reverse conclusion does not necessarily hold , in the sense that a slope near unity is not uniquely indicative of a steady mode of galaxy formation ; other scenarios ( such as the staged galaxy formation model of * ? ? ?
* ) can be constructed to reproduce a slope near unity , as discussed in
[ sec : tact ] .
example galaxy sfhs from the @xmath53 momentum - driven outflow hydro simulation are shown in the lower right panel of figure [ fig : sfrmstar ] , for a range of @xmath13 stellar masses from @xmath54 , in 100 myr bins .
they are remarkably self - similar , and generally quite smooth .
the increased variance in smaller galaxies owes to discreteness effects in spawning star particles ( these sfhs are computed from the history of galaxy star particles rather than the instantaneous sfr at each time ) ; the true scatter in instantaneous sfr increases only mildly at small masses , from @xmath55 dex at @xmath56 to @xmath57 dex at @xmath58 , as shown in figure [ fig : sfrmstar ] .
the early simulated sfhs are characterized by a rapid rise in star formation to @xmath59 , corresponding to rapid early halo growth @xcite .
next , there is a period of roughly constant star formation from @xmath60 ; this will be the key epoch for the present work .
finally , at @xmath61 the star formation rates decline modestly , with more massive systems showing greater decline .
this pattern reflects the changing a balance between the growth of potential wells being able to attract more matter , and cosmic expansion which lowers the ambient igm density and infall rates .
it is qualitatively similar to the sfh inferred for the milky way . in the models ,
the fact that the sfh depends weakly on @xmath2 gives rise to @xmath17sfr , and the fact that merger - driven starbursts are not a significant growth path for galaxy stellar mass @xcite gives rise to the small scatter .
in detail a slope slightly below unity occurs because more massive galaxies tend to have their accretion curtailed by hot halos ( e.g. * ? ? ?
* ) , resulting in natural downsizing " @xcite .
note that the sfrs of individual galaxies in simulations do not drop by a large amount over a hubble time . even for the most massive star forming galaxies today
, the sfr in simulations drops only be a factor of few since @xmath38 , and smaller galaxies show essentially no drop .
meanwhile , the amplitude of @xmath2-sfr drops by nearly an order of magnitude in the models , and even more in the data .
so the amplitude evolution owes predominantly to the growth of @xmath2 of individual galaxies . in other words ,
models indicate that the @xmath0sfr relation does nt evolve downwards in sfr so much as upwards in @xmath2 .
overall , while the slope and scatter of the @xmath0sfr relation are in broad agreement , there is a stark and generic discrepancy in the amplitude evolution between the models and data .
the agreement in slope and scatter is an encouraging sign that our understanding of gas accretion processes is reasonably sound .
examining the discrepancy in amplitude evolution is the focus of the remainder of this paper .
-0.5 in -0.4 in the amplitude evolution of @xmath0sfr can be recast in terms of the timescale for star formation activity . models generally predict that galaxies have @xmath62sfr@xmath63 , where @xmath64 is the hubble time , perhaps reduced by 1 gyr because few stars were forming at @xmath65 , as seen e.g. in the lower right panel of figure [ fig : sfrmstar ] .
a convenient parameterization is provided by the _ star formation activity parameter _ defined here as the dimensionless quantity @xmath66sfr@xmath67 .
physically , this may be regarded as the fraction of the hubble time ( minus a gyr ) that a galaxy needs to have formed stars at its current rate in order to produce its current stellar mass .
note that @xmath6 is in general a function of @xmath2 .
figure [ fig : tact ] shows @xmath68 for simulations and observations , top panel showing values at @xmath69 , bottom showing @xmath70 .
the values for models have been obtained by fitting power laws to simulated galaxies @xmath0sfr .
the momentum - driven outflow hydro simulation results are shown as the red circles , while cyan and magenta circles show results for no winds and the wind model of @xcite at @xmath71 , respectively .
the outflow model has only a mild effect on @xmath6 , because the _ form _ of the sfhs are broadly similar regardless of outflows , even though the total amount of stars formed is significantly different @xcite .
the most dramatic outlier among models is the high - mass bin of the no - wind simulation , where the rapid gas consumption in large galaxies results in lower current star formation rates relative to the past - averaged one ( i.e. natural downsizing ) ; this actually makes @xmath6 larger and the discrepancy with observations worse .
millenium sam results are quite similar to the hydro runs , again showing little evolution in @xmath6 , although the shallower slope results in a somewhat smaller @xmath6 for the lower mass bin .
overall , it is remarkable that regardless of methodology or feedback processes , all models predict @xmath6 to be around unity with little evolution from @xmath72 , regardless of stellar mass .
this suggests that @xmath4 is a generic consequence of smooth and steady cold accretion dominating the growth of these galaxies .
in contrast , the observed @xmath0sfr relation yields a rapidly rising @xmath73 from @xmath74 at @xmath75
. values at @xmath76 are slightly lower but show similar rapid evolution .
the dramatic difference between observed and predicted amplitude evolution either indicates a significant misunderstanding in our basic picture of galaxy assembly , or else that observations have been misinterpreted in some way . given the currently uncertain state of galaxy formation theory
, one might be inclined towards the former , and call into question the entire picture of cold mode - dominated accretion .
however , it turns out that the rapidly declining @xmath6 also results in uncomfortable conflicts with other _ observations _ of high - redshift galaxies . to see this ,
let us discard the idea of cold mode accretion , or any preconceived theoretical notion , and ask : what possible star formation histories for these galaxies could yield @xmath6 rising with time as observed ?
one scenario is to postulate that observed high-@xmath9 galaxies are progressively more dominated by bursts of star formation .
indeed , some models have suggested that bursts may comprise a significant portion of the high-@xmath9 galaxy population ( e.g. * ? ? ?
however , as @xcite points out , this seems unlikely given the tightness of the @xmath2-sfr relation .
they argue that the observed scatter of 0.3 dex at @xmath77 implies that most star forming galaxies have current star formation rates @xmath78 of their quiescent value , and that the mean of the relation represents the quiescent main sequence " of galaxy assembly .
few galaxies between this main sequence and the passive population are seen , arguing for rapid movement between the two populations . at @xmath8 ,
the constraints are even more severe , as @xcite measure a @xmath79 scatter of 0.29 dex ( 0.16 dex interquartile ) , while the observed @xmath80 implies they are bursting at @xmath81 their quiescent rate . since these observations are able to probe sfrs down to @xmath82/yr , moderate stellar mass ( @xmath83 ) galaxies with @xmath84 would have been seen if present , but are not .
the only way the burst scenario would work is if all observed galaxies somehow conspire to be bursting at approximately the same level , which seems highly contrived .
hence bursts do not seem to be a viable way to produce low @xmath6 .
another possible scenario is that models simply need to begin forming stars at a later epoch .
this would effectively lower @xmath6 as defined here , while still maintaining a fairly constant sfr as required by the observed scatter .
theoretically , such a scenario is easier said than done , as the epoch where galaxy formation begins is not a free parameter even in semi - analytic models , but rather is governed by halo formation times that are well - constrained by current cosmological parameters .
such a scenario would require overwhelmingly strong feedback at early epochs , far stronger than even the strong feedback present in e.g. hydro simulations that match observed @xmath85 galaxy star formation rates @xcite , in order for copious dense gas to be present in early halos yet not form stars .
still , in the spirit of discarding any theoretical preconceptions , delayed galaxy formation " appears to be a viable solution .
however , quantitative consideration of delayed galaxy formation leads to significant problems when confronted by other observations . at @xmath38
, @xmath80 implies that ( constant ) star formation began roughly 0.5 gyr prior which corresponds to @xmath86 ! of course it can not be true that _ all _ galaxies began forming stars at @xmath87 , as many star - forming galaxies are seen at significantly higher redshifts ( e.g. * ? ? ?
* ) . indeed , both the cosmic star formation rate ( e.g. * ? ?
* ) , and the rest - uv luminosity function @xcite are observed to be fairly constant between @xmath88 .
so it appears galaxies were forming stars quite vigorously at @xmath89 .
even if galaxies smoothly ramped up their star formation to values observed at @xmath8 , they would still need to start at @xmath90 .
if most galaxies can not have such a low @xmath6 , perhaps the observed systems represent only a small subset of galaxies at @xmath8 , and the rest are simply unseen .
since current surveys identify both uv - luminous and red , dusty star forming galaxies , it must be that the unseen galaxies are passive , or else so extremely extincted that they only show up in e.g. sub - millimeter surveys .
for the latter case , @xcite found that the cosmic sfr has a relatively minor contribution from such severely extincted systems that escape bzk / lbg selection at @xmath8 , so they are not likely to be the dominant growth mode for galaxies .
moreover , this actually exacerbates the @xmath0sfr problem , because it provides a mode to form many more stars beyond what is formed quiescently , which would make the expected stellar masses even higher than just from the quiescent mode .
hence the unseen galaxies must be passive , or forming stars at some undetectably low rate . in the currently popular downsizing scenario
@xcite , these systems would have formed their stars earlier in a rapid mode , and have now attenuated their star formation through some feedback process so as not to be selected in star - forming samples . perhaps by @xmath8 , downsizing is already well underway , and the observed galaxies only represent the small fraction of systems that have just recently begun forming stars and will soon be quenched .
while qualitatively sensible , the unseen passive galaxies " model is once again quantitatively inconsistent with other observations of high-@xmath9 galaxies .
@xcite used a _
spitzer_/irac - selected sample to determine that the star - forming bzk ( sbzk ) selection technique , i.e. the one employed by @xcite to measure @xmath2-sfr at @xmath8 , encompasses 88% of their galaxies ( 2691/3044 ) from @xmath91 .
this is remarkably efficient , significantly more so than e.g. the lyman break or distant red galaxy selection techniques in obtaining a stellar mass - limited sample .
their stellar mass completeness is estimated to be @xmath92% above @xmath93 for passively evolving galaxies . putting these together
, this means that conservatively at least two - thirds of all galaxies above this mass threshold are selected as sbzk galaxies , and hence less than one - third are unseen passive galaxies . in contrast , the observed @xmath0sfr relation at @xmath8 implies @xmath94 , meaning that @xmath95 of galaxies should be unseen ( passive ) , or else the duty cycle of such objects should be @xmath96 .
this assumes galaxies began significant activity @xmath97 gyr after the big bang , and that @xmath98 at @xmath99 ; the latter is again conservative , as an extrapolation would indicate that @xmath6 should be even lower at @xmath99 .
hence it does not appear that sbzk selection misses nearly enough systems to make the unseen passive galaxies scenario viable to explain @xmath100 .
another way to accomodate the low scatter , unity slope , and rapid evolution in @xmath0sfr is to propose that galaxies evolve _ along _ the @xmath0sfr relation , as proposed by @xcite . at @xmath8
, @xmath98 implies that galaxies must be growing very quickly .
rapid growth means that for any given galaxy , the amplitude of @xmath0sfr is essentially unchanging . now since sfr@xmath101 , an unchanging amplitude implies a solution of @xmath102 in other words , in this scenario , galaxies have exponentially growing star formation rates and stellar masses
! this growth would presumably continue until star formation is truncated by some feedback mechanism , probably when @xmath103 . while at first this exponentially growing galaxies " idea may seem incredible , of all the scenarios
it actually comes closest to reconciling all the high-@xmath9 galaxy data .
one way to test this scenario is to consider the evolution of the number densities @xmath104 of galaxies .
the @xmath105-folding time of stellar mass growth is given by @xmath106sfr . assuming ( conservatively )
that @xmath98 remains constant at @xmath107 , then @xmath108 gyr changes with time means that the assumption of a constant @xmath2-sfr amplitude in equation [ eqn : tau ] is invalid ; however , a simple evolutionary model accounting for this yields virtually identical results . ] .
a simple calculation shows that there are e.g. five @xmath105-folding times ( or 2.2 decades of growth ) from @xmath109 . for a given stellar mass @xmath2 , @xmath110 at @xmath38
should then be similar to @xmath111 at @xmath112 ( neglecting mergers ) .
@xcite measured the number density of galaxies with @xmath113 at @xmath8 to be @xmath114mpc@xmath115 , and those with @xmath116 at @xmath112 have @xmath117mpc@xmath115 ( based on an extrapolation of their sty fits ) , and stellar mass functions from @xcite yield similar values .
hence at face value current data suggests that the stellar mass function does not evolve quite as rapidly as expected in this scenario .
still there are substantial uncertainties in the observed stellar mass function at such low masses and high redshifts , so a factor of a few discrepancy is not compelling enough to rule out the exponentially growing galaxies scenario .
another scenario that has been shown to reproduce @xmath2-sfr evolution is the staged galaxy formation model of @xcite . here
, a delay is introduced in the onset of galaxy formation that is inversely related to the galaxy mass . hence large - mass objects form early and quickly , while small ones start forming later over longer timescales .
this is similar to the exponentially - growing galaxies scenario in the sense that the latter implicitly assumes a sort of delay , because galaxies must accumulate lots of gas in their halo without forming stars in order to have a sufficient gas reservoir to then consume exponentially .
hence it may be possible to constrain staged galaxy formation with similar measurements of the stellar mass function to higher-@xmath9 ; this is being investigated by k. noeske ( priv .
if either exponentially - growing or staged galaxy formation is correct , this would indicate that star formation at early epochs does not scale with gas density as seen locally @xcite , or else that cold gas is somehow prevented from accumulating in halos as expected from growth of structure . in summary , the small @xmath6 at @xmath8 implied by the @xmath0sfr amplitude
is difficult to reconcile with the observed scatter in @xmath0sfr , the observed evolution of star - forming galaxies to higher redshifts , and the observed fraction of passive galaxies . taken together
, the latter data all point to the idea that observed star - forming galaxies represent the majority of galaxies at intermediate masses ( @xmath118 ) at @xmath8 , and have been quiescently forming stars over much of a hubble time , i.e. @xmath4 . while designer models of stellar assembly can not be excluded , they would be highly unexpected , and would require substantial revision to our understanding of how gas accumulates and forms stars in hierarchical structure formation . the general difficulty in reconciling the data together with the models preference for @xmath4 suggests an uncomfortable situation that bears investigation into alternative solutions .
determining @xmath2 and sfr from broad - band observations is fraught with a notoriously large number of systematic uncertainties that could alter the values of @xmath6 inferred from observations .
here it is considered whether such systematics could be responsible for the low value of @xmath6 at high-@xmath9 , particularly focusing on the @xcite analysis at @xmath8 since that is where the discrepancies from model expectations and the range of possible systematics are greatest .
star formation rates there are generally estimated by summing rest - uv flux ( uncorrected for extinction ) with 24@xmath119 flux . for objects that are anomalously high or low in 24@xmath119 to dust - corrected uv sfr estimations
, they use the dust - corrected uv or 24@xmath119 estimate , respectively ; however , only one - third of their sample are anomalous in that sense . meanwhile , stellar masses are estimated from sed fitting : using multi - band photometry or spectra , one fits model stellar population templates ( in their case , * ? ? ?
* ) , varying parameters until the best fit is obtained . in detail , @xcite used a calibration of bzk color to stellar mass based on fitting k20 survey spectra @xcite .
the key assumptions in sed fitting are population synthesis models , dust extinction models , and the form of the star formation history , all of which introduce systematic uncertainties .
current uncertainties in obtaining accurate stellar masses from rest near - ir data revolve around the poorly known contribution of thermally - pulsating asymptotic giant branch ( tp - agb ) stars to the @xmath120-band light @xcite .
this is found to lower the stellar mass estimates for high-@xmath9 galaxies by a factor of two or more @xcite .
however , the sense would be to _ exacerbate _ the problem by lowering @xmath2 and hence further reducing @xmath6 , compared with the @xcite models utilized in @xcite . hence while uncertainties in population synthesis models are not negligible , they seem unlikely to make a @xmath121 difference in the proper direction .
it is possible to hide more stellar mass in objects if one postulates a very old underlying stellar population , in other words by modifying the assumed sfh form .
@xcite found that for uv - selected galaxies , a mass increase of @xmath122 was possible without violating rest - optical constraints .
however , @xmath120-selected samples like that of @xcite are generally redder with more contribution from old stellar light , so the amount that one can hide is more limited ( c. papovich , priv . comm . ) .
more problematic is the fact that doing so requires assuming that the bulk of the galaxy s stars formed very early in the universe , and hence the current sfr is much lower than at that early epoch .
it turns out that this is not a self - consistent solution for reducing @xmath6 .
this is because if such a sfh history were true , one would expect that the current @xmath2 would be significantly _ more _ than obtained by assuming a constant sfh over a hubble time .
this would imply @xmath123 for those galaxies . hence even hiding @xmath124
as much stellar mass as currently inferred would be insufficient , because it would only make @xmath4 , whereas such a sfh implies @xmath123 . in order for there to be a large old stellar population while still having a high current sfr
, it must be that such galaxies formed many stars early , then sat around not forming stars until a recent flare - up ; however , this scenario suffers the same difficulties as the delayed galaxy formation model .
hence one can not self - consistently alter the sfhs to hide a large amount of very old stars in these galaxies . instead
, it may be possible to appeal to some systematic reduction in the inferred star formation rate .
why might the true sfr be lower than estimated ?
one possibility is that the extinction corrections being used for uv - estimated sfrs are too excessive .
now recall that for the majority of galaxies , @xcite estimated the sfr from the uncorrected uv plus 24@xmath119 flux , so for those cases the extinction correction is not relevant ; it is only relevant for the @xmath125 that are mid - ir excess " systems . for those , there may be room to alter the extinction law such that the extinction inferred from the uv spectral slope is overestimated .
@xcite , like most studies of actively star forming galaxies , use the @xcite law . using instead a milky way or lmc law would indeed yield less extinction for a given uv spectral slope , but the typical reduction is less than a factor of two .
so it seems unlikely that extinction alone is causing the low @xmath6 even in the minority mid - ir excess objects , unless the extinction law were dramatically different than anything seen in the local universe .
another possibility is that there is an additional contribution to the rest - uv flux besides star formation .
the obvious candidate are agn , which are common in @xmath8 star forming galaxies . while @xcite found that only @xmath126 of @xmath127 galaxies at @xmath18 have x - ray detectable agn , @xcite identified obscured agn through their mid - ir excess , and found that they appear in @xmath128 of galaxies at @xmath129 , and dominate the overall agn population at @xmath8 .
importantly , they also found ( in agreement with previous studies , e.g. * ? ? ?
* ) that smaller mass galaxies have a lower agn fraction , as expected theoretically @xcite .
recall that @xcite used uv fluxes to estimate sfr in such mid - ir excess objects , and by comparison with x - ray data they estimated that such an inferred sfr is unlikely to be wrong by a significant factor .
moreover , since agn are increasingly rare in smaller mass systems , it would be an odd coincidence if they mimicked the relation @xmath17sfr@xmath130 so tightly . like the burst model , this represents a fine - tuning problem , where agn have to contribute a very specific amount to the inferred sfr at each mass scale for no obvious physical reason .
this can not be ruled out , but seems a priori unlikely .
it is also possible that agn could be present in galaxies that are not mid - ir excess systems .
the strong dichotomy in the x - ray properties suggest there is not a continuum of agn , and that agns are limited to mid - ir excess systems ( e.g. * ? ? ?
* ) . however , there exist agn templates where the contibution to rest - uv and rest-8@xmath119 flux are broadly comparable to that in star - forming galaxies @xcite . these could be mistaken for normal " star - forming systems , whereas in fact the dominant flux is from agn .
still , once again , to mimic the tight @xmath0sfr relation would require agn to be contaminating the sfr estimate at a similar level at all stellar masses .
moreover , if they are dominating the uv flux then they should appear as central point sources , but _
hubble_/acs imaging of sbzk galaxies generally do not show such bright central sources ( m. dickinson , priv .
comm . ) . hence while agn are certainly present , it seems unlikely that they are a major contaminant for uv - derived sfrs .
a perhaps even more troublesome issue is the use of rest-8@xmath119 ( i.e. 24@xmath119 at @xmath8 ) as a star formation rate indicator .
this is quite uncertain , because the dominant flux at those wavelengths is from polycyclic aromatic hydrocarbon ( pah ) emission features , whose nature is not well understood . as @xcite points out
, there is more than a factor of two scatter in star formation rate indicators that assume a fixed ( local ) pah template .
moreover , owing to pah features moving in and out of bands , standard photometric redshift uncertainties of @xmath131 can result in up to an order of magnitude error in inferred sfrs . on the other hand ,
the tight scatter in the @xmath8 @xmath0sfr relation would seem to limit the amount of scatter owing to pah feature variations , or else be an indication that uv light dominates over pah emission ( though the latter is not what is seen locally ; * ? ? ?
@xcite notes that @xmath132 emission has a substantial metallicity dependence , but it is in the sense that lower metallicities have lower emission , so their sfr would be underestimated by using solar metallicity calibrations .
so while it is conceivable that some unknown reason causes pah emission per unit star formation to be systematically higher in high-@xmath9 galaxies compared to local ones , thereby causing the sfr to be overestimated , but there are no obvious indications that this would mimic a tight @xmath2-sfr relation . finally , the observed sfr and @xmath2 s in @xcite are inferred assuming a salpeter imf .
assuming instead a @xcite or @xcite imf would result in shifting _ both _ @xmath2 and sfr down by @xmath133 dex @xcite .
hence to first order it makes no difference as long as the @xmath0sfr slope is around unity ( e. daddi , priv .
fundamentally , this is because at high-@xmath9 both stellar mass and star formation rate indicators are driven by light from stars @xmath134 , and favored present - day galactic imfs generally agree on the shape above this mass . hence more substantial imf variations would be required . in summary ,
there are no obvious paths to systematically alter @xmath2 and sfr determination to make up up the required @xmath135 difference in @xmath6 at @xmath8 .
it is true that many of these systematics can not be ruled out entirely , and it may be possible that a combination of such effects could explain part or even all of the difference .
it is also possible that locally calibrated relations to estimate @xmath2 and sfr may be substantially different at high-@xmath9 for unknown and unanticipated reasons .
the reader is left to assess the plausibility of such scenarios .
the view taken here is that the difficulty in obtaining a straightforward solution motivates the consideration of more exotic possibilities .
one possible way to alter @xmath2-sfr evolution is to invoke an imf that is in some direct or indirect way redshift - dependent .
the key point is that most direct measures of the star formation rate actually trace high - mass star formation , while stellar mass is dominated by lower - mass stars .
hence by modifying the ratio of high to low mass stars formed , i.e. the shape of the imf , it is possible to alter the @xmath0sfr relation .
the possibility of a varying imf has been broached many times in various contexts ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , though typically as a last resort scenario .
it is highly speculative as there is no clear - cut evidence at present that supports a time- or space - varying imf , but it is worth investigating in light of growing observational and theoretical arguments in its favor .
as @xcite points out , a universal imf is not to be expected theoretically , though no variations have been unequivocally detected in local studies of star - forming regions .
the average galactic imf appears to be well - represented by several power laws , steepest at @xmath136 , then significantly shallower at @xmath137 , with a turnover at @xmath138 ( see review by * ? ? ?
forms that follow this include the currently - favored @xcite and @xcite imfs .
but deviations _ are _ seen , at least at face value .
the nearby starburst galaxy m82 appears to have an imf deficient in low - mass stars @xcite .
a similarly bottom - light imf is inferred in the highly active arches cluster near the galactic center , suggesting that vigorous star formation makes the imf top - heavy ( e.g. * ? ? ?
also , the young lmc cluster r136 shows a flattening in the imf below @xmath139 @xcite .
there are hints that presently active star clusters form more low - mass stars compared to the disk - averaged 5 gyr - old imf @xcite , suggesting that the imf was weighted towards heavier stars at lower metallicities and/or earlier times .
even if the imf shape is universal , larger starforming regions may produce more high - mass stars simply because they have enough material to aggregate into larger stars @xcite .
hence there are hints of imf variations , albeit controversial , and interestingly they consistently go in the direction of having a higher ratio of massive to low - mass stars in conditions similar to those in high-@xmath9 galaxies . compared to present - day star - forming systems , galaxies at high-@xmath9 have far higher star formation surface densities than the galactic disk @xcite , have higher gas content and hence presumably more massive starforming clouds @xcite , and have somewhat lower metallicities @xcite . @xcite and @xcite list other pieces of circumstantial evidence that favor early top - heavy imfs .
so while there are no direct observational evidences or strong theoretical arguments for it , it is not inconceivable that the imf in high-@xmath9 galaxies may be more top - heavy or bottom - light than the present - day galactic imf .
such an imf would have three main effects on the @xmath0sfr relation : it would increase the output of uv flux per unit stellar mass formed ; it would cause more stellar mass loss from stellar evolution ; and the mass recycling would provide a larger gas reservoir with which to form stars .
all three effects go towards reconciling theoretical expectations with observations of the @xmath2-sfr relation .
the sfr inferred by assuming a standard imf would be an overestimate , since many fewer low - mass stars are formed per high - mass star . increased recycling losses would lower the theoretical expectations for the amount of stellar mass remaining in these galaxies , bringing them more into line with observed stellar masses .
recycled gas would increase the available reservoir for new stars , effectively delaying star formation in galaxies and pushing models towards the delayed galaxy formation " scenario .
hence even modest evolution in the imf could in principle reconcile the observed and theoretically - expected @xmath6 evolution .
how is the imf expected to evolve with time ? this is difficult to predict since no ab initio theory for the imf exists today . but a simple conjecture by @xcite may have some empirical value .
he notes that there is a mass scale of @xmath140 below which the galactic imf becomes flatter @xcite , and at which point the mass contribution per logarithmic bin ( @xmath141 ) is maximized .
larson uses simple thermal physics arguments to show that this _ characteristic imf mass scale _ ( @xmath142 ) is related to the minimum temperature @xmath143 of molecular clouds , measured today to be around 8 k. in that case , if the temperature in the interstellar media ( ism ) of high-@xmath9 galaxies is higher , this characteristic mass will shift upwards , resulting in an imf that is more bottom - light . to generate a crude imf evolution model ,
let us assume that the @xmath144 , where @xmath145 is some unknown parameter whose value is set by the various effects above .
@xcite predicted analytically that @xmath146 , while recent numerical experiments by @xcite found a less steep scaling of @xmath147 .
so for example an evolution that scaled purely with cmb temperature would yield @xmath148 .
however , other factors may cause an increased ism temperature , e.g. , elevated disk sfr rates causing more supernova heat input ; lower cooling rates owing to lower metallicities ; and/or photoionization from a stronger metagalactic uv flux ( e.g. * ? ? ? * ) .
the approach adopted here is to treat @xmath145 as a free parameter , to be constrained using the observed evolution of @xmath6 .
the assumption is that @xmath149 reflects the present - day galactic imf , while @xmath150 is inferred to be lower purely because of the assumption of a non - evolving imf , whereas in reality @xmath68 ( at any given mass ) is constant as expected in models .
an evolving kroupa " imf is defined in which @xmath151 above @xmath142 , the imf has a form @xmath152 , while below it scales as @xmath153 .
note that the current galactic imfs also have a turn - down at masses below @xmath154 ; the present analysis provides no constraints in this regime , as the stellar mass formed there is a small fraction of the total .
the exact form of the imf is not critical here ; @xcite is chosen for its convenient explicit parameterization in terms of @xmath142 , but the analysis below could equally well have been done with the similar chabrier imf . the goal of the evolving imf is to increase the ratio of the star formation rate of _ high - mass _ stars ( sfr@xmath155 ) to total stellar mass formed , such that the inferred @xmath68 is non - evolving .
note that different probes such as uv luminosity , h@xmath156 emission , and mid - ir emission trace slightly different regimes of high - mass star formation , but so long as the high - mass imf slope remains fixed , the relative ratios of such emission should be constant .
so for example , in order to produce no evolution in @xmath6 from @xmath157 at @xmath76 , the factors by which @xmath158sfr@xmath159 must be raised are 1.9 at @xmath47 and 3.3 at @xmath38 , relative to a standard kroupa imf ( cf .
figure [ fig : tact ] ) .
hence @xmath145 must be chosen to match these factors .
what value of @xmath145 would produce such evolution ? to determine this , the pegase.2 @xcite population synthesis model is employed , taking advantage of its feature allowing user - definable imfs .
pegase outputs a variety of quantities as a function of time from the onset of star formation , given an input star formation history .
the assumed sfh is taken to be exponentially - declining with a 10 gyr decay time and starting 1 gyr after the big bang , in order to approximate a typical model sfh .
an example of such a sfh is shown as the magenta line in the lower right panel figure [ fig : sfrmstar ] , normalized to a star formation rate of 2 @xmath160yr@xmath161 today ; as is evident , it is quite similar to typical sfhs from the simulations , at least for moderate stellar masses .
a constant sfh was also tried , and the results were not significantly different . besides the input sfh , all other parameters are taken as the pegase default values .
no dust extinction is assumed .
the pegase output quantities specifically used here are the stellar mass @xmath2 formed , and the un - extincted 1650 luminosity ( @xmath162 ) .
the latter is used as a proxy for sfr@xmath155 . using these
, it is possible to compute @xmath163 for any given assumed imf at any given time .
note that pegase accounts for both stellar mass accumulation and mass loss owing to stellar evolution , but does not account for the impact of recycled gas . by varying @xmath142 , it is possible to estimate @xmath163 for any assumed @xmath145 , and thereby determine what value of @xmath145 would yield no @xmath6 evolution . in practice , determining @xmath158sfr@xmath159 at any epoch requires integrating the mass growth up to that epoch for an evolving imf . to do so
, first pegase models are computed for a wide range of @xmath164 values .
then the following procedure is used : + ( 1 ) a value of @xmath145 is selected ; + ( 2 ) the following integral is computed : @xmath165 where @xmath166 is the stellar mass growth rate at time @xmath167 taken from pegase models interpolated to the appropriate @xmath164 as given in equation [ eqn : zkroupa ] ; + ( 3 ) @xmath162 is taken at time @xmath167 from pegase , and @xmath168 is computed ( the subscript ev " stands for evolving " ) ; + ( 4 ) similarly , @xmath169 is computed for a standard kroupa imf , where @xmath170 ; + ( 5 ) the ratio @xmath171 is computed , and compared to the factors needed to produce no @xmath6 evolution .
-0.4 in -3.2 in the results of this procedure are shown in figure [ fig : zratio ] , namely @xmath172 as a function of redshift for several values of @xmath145 .
the values by which the @xmath0sfr relation needs to be adjusted to produce no @xmath6 evolution at @xmath173 are indicated by the symbols : squares are for @xmath174 , triangles for @xmath76 , and circles for @xmath175 .
these are computed from published fits to @xmath0sfr @xcite . because the @xmath0sfr relation becomes shallower with time , the required adjustment factors are generally smaller at lower masses .
figure [ fig : zratio ] shows that @xmath176 provides a reasonable fit to the required imf evolution for typical star - forming galaxies at @xmath70 .
the implied values of @xmath142 are @xmath177 and @xmath178 at @xmath179 , respectively .
note that there is no a priori reason why a single value of @xmath145 should fit at all redshifts considered , so the good fit supports the form of @xmath164 evolution assumed in equation [ eqn : zkroupa ] . as a check , an independent method for determining @xmath142 evolution
was explored that is less comprehensive , but conceptually more straightforward . here
, the high - mass star formation rate is assumed to be unchanged , while observations of stellar mass are assumed to reflect stars near the main sequence turnoff mass ( @xmath180 ) , since red optical light is dominated by giant stars . for an evolving kroupa imf , to obtain a reduction of a factor of @xmath181 in the amount of stars produced at @xmath182 , it is easy to show that @xmath183 . @xmath180 may be estimated by noting that stellar lifetimes scale as @xmath184 , and that the sun has a lifetime of 10 gyr ; this yields @xmath185 , ( @xmath64 in gyr ) , which is @xmath186 and @xmath187 at @xmath179 , respectively . inserting values of @xmath188 at @xmath179 yields @xmath189 and @xmath190 .
these values are quite similar to those obtained from pegase modeling , showing that the results are not critically dependent on details of pegase .
the same procedure can be applied to the scenario proposed by @xcite , where the cmb temperature is solely responsible for setting a floor to the ism temperature . in that case ,
@xmath191 $ ] , and @xmath192 .
@xcite predicted @xmath193 , while @xcite found @xmath194 .
the results of this scenario for these two values of @xmath195 are shown as the long and short dot - dashed lines , respectively , in figure [ fig : zratio ] .
as is evident , this imf evolution is not nearly sufficient to reconcile the observed and predicted @xmath6 evolution out to @xmath8 , mainly because @xmath196 only exceeds 8 k at @xmath197 , and hence @xmath142 does not change from @xmath19 . note that since cmb heating of the ism is subdominant at all redshifts , the assumed form of imf evolution ( eq .
[ eqn : zkroupa ] , scaling as @xmath198 ) is not physically well motivated .
while the evolution out to @xmath8 is well - fit by such a form , it could be that the form changes at higher @xmath9 , or that it should be parameterized as some other function of @xmath9 .
for instance , if it is the vigorousness of star formation activity that determines @xmath142 , perhaps imf evolution actually reverses at very high redshifts . for lack of better constraints , the form for imf evolution in equation [ eqn : zkroupa ]
will be assumed at all @xmath9 , but this should be taken as an illustrative example rather than a well - motivated model .
this analysis also makes predictions for @xmath6 at @xmath99 that would be inferred assuming a standard ( non - evolving ) imf .
for instance , figure [ fig : zratio ] implies that at @xmath76 , @xmath199 and @xmath200 .
notably , the evolution slows significantly at @xmath201 , and actually reverses at @xmath202 , though the simple sfh assumed in the pegase modeling may be insufficiently realistic to yield valid results at @xmath202 .
any such extrapolation of this imf evolution should be made with caution , as the @xmath2-sfr relation only constrains it out to @xmath8 . however , in the next section this will be compared to observations that suggest that such imf evolution may be reasonable out to @xmath203 .
-0.4 in -3.2 in figure [ fig : mchar ] illustrates the evolving imf , in terms of stellar mass formed per unit logarithmic mass bin .
a salpeter imf is shown for reference , and evolving kroupa imfs are shown at @xmath42 with characteristic masses @xmath170 , @xmath204 , and @xmath178 , respectively . as expected , the evolution results in more high mass stars compared to low mass ones at higher redshifts . while such a dramatic increase in @xmath142 may seem surprising ,
recall that studies of local highly active star - forming regions suggest a characteristic mass scale of a few @xmath160 @xcite .
note that an imf with a steeper high - mass slope ( e.g. * ? ? ?
* ) would result in even stronger imf evolution being required than in the @xcite case . in summary , an evolving kroupa imf
whose characteristic mass increases with redshift as @xmath205 is able to alter the observed @xmath6 evolution from @xmath206 into one with approximately no evolution , as predicted by models or inferred from complementary observations .
the precise imf shape is not well constrained ; it is certainly possible that an imf with different behavior , for instance an evolving high - mass slope ( e.g. * ? ? ?
* ) , could produce the same @xmath6 evolution .
the only requirements are that it produces the correct ratio of high - mass stars relative to low - mass stars as a function of redshift , and that ( unlike e.g. the merger - induced top - heavy imf suggested by * ? ? ?
* ) it applies to the bulk of star forming galaxies at any epoch .
what are the observational implications of such an evolving imf at early epochs ? surprisingly few , as it turns out , so long as the massive end of the imf remains unaffected .
this is because , as mentioned before , virtually all measures of high-@xmath9 star formation , be they uv , ir or radio , trace light predominantly from high - mass stars . feedback energetics and metallicities likewise reflect predominantly massive star output .
hence so long as an imf preserves the same high mass star formation rate , it is expected to broadly preserve the successes of understanding the relationship between the rest - uv properties of galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) , cosmic metal pollution ( e.g. * ? ? ?
* ; * ? ? ?
* ) , feedback strength , and type ii supernova rates .
the main difference caused by such imf evolution is in stellar mass accumulation rates .
high-@xmath9 measures of stellar mass are therefore critical for testing this type of scenario . @xcite determined the stellar mass function evolution from a k - band survey out to @xmath207 , but at even moderately high redshifts the rest - frame light is actually quite blue , so their stellar mass estimates implicitly involves a significant imf assumption .
the high-@xmath9 stellar mass metallicity relation @xcite has the potential to test the imf ; if the evolving imf is correct , the good agreement found versus recent models @xcite would perhaps not stand , and in particular many more metals would have to be driven out of galaxies of a given @xmath2 via outflows .
type ia supernova rates at high-@xmath9 would be lowered ; the apparent turn - down in type ia rates at @xmath208 argued by @xcite as being indicative of a long type ia delay time of 3 gyr , may instead be accomodated via a more canonical short delay time coupled with an evolving imf .
these are examples of the types of observations that could potentially constrain imf evolution .
-0.3 in -3.2 in the star formation rates inferred for high - redshift galaxies by assuming present - day imfs would have to be revised downwards for an evolving imf . a quantitative estimate of the reduction in star formation rates from the standard to evolving imf case is provided by the ratio @xmath209 as a function of redshift from figure [ fig : zratio ] . specifically , the true " cosmic star formation rate ( assuming the evolving imf is correct ) is the one currently inferred assuming a non - evolving imf , divided by the curve for a particular value of @xmath145 , say @xmath176 .
figure [ fig : madau ] shows the cosmic star formation history with data points compiled by @xcite , and the fit by @xcite .
it also schematically illustrates the effect of an evolving imf by dividing the @xcite fit by the @xmath172 curve for @xmath176 .
note that @xcite used a salpeter imf to infer star formation rates , so these correction factors ( derived for a kroupa imf ) are not exact , but should be reasonably close .
figure [ fig : madau ] shows that the cosmic sfr peak shifts to later epochs in the evolving imf case . the total stellar mass formed in the universe
is reduced by 55% by @xmath13 and 75% by @xmath38 , compared to a standard imf .
interestingly , some current observations seem to _ prefer _ an imf at early epochs that is more top - heavy or bottom - light , and perhaps a later epoch of cosmic star formation than generally believed .
@xcite noted that the integral of the cosmic star formation history exceeds the stellar mass currently detected by a factor of two , assuming a salpeter imf .
@xcite also noted such tension , but suggested that switching from salpeter to a chabrier or kroupa imf might go significantly towards reconciling the difference ; little evidence exists for imf evolution to @xmath210 @xcite .
but to higher redshifts , observations by @xcite of stellar mass density evolution to @xmath211 suggest that mass buildup is peaked towards significantly later epochs than in current models that broadly match the observed cosmic star formation history .
@xcite , updating an analysis done by @xcite , found that the cosmic sfh and present - day cosmic luminosity density are best reconciled using an imf slope that is slightly more top - heavy than salpeter ( @xmath212 vs. @xmath213 ) , and significantly more top - heavy than e.g. @xcite .
@xcite suggests from an archaeological analysis of sdss spectra that the cosmic star formation rate peaked at @xmath214 rather than the more canonical @xmath8 .
@xcite find that it is easier to understand early - type fundamental plane evolution at @xmath210 with a slightly top - heavy imf ( as also argued by * ? ? ?
intracluster medium metallicities indicate that stars formed in clusters at early epochs either had higher yields than today , or else formed with a top - heavy imf @xcite . @xcite and @xcite suggest that the abundance of carbon - enhanced metal - poor stars is indicative of a more top - heavy imf at early times in our galaxy .
@xcite finds that an imf slope more top - heavy than salpeter is required to produce enough early photons to reionize the universe . while it should be pointed out that a large number of observations are consistent with a universal imf , whenever a discrepancy exists , it is invariably in the sense of favoring an imf with more massive stars at early times .
@xcite did perhaps the most careful work on trying to reconcile fossil light measures with redshift - integrated measures of stellar mass . extending the analyses of @xcite and @xcite by including extragalactic background light ( ebl ) constraints
, they found that _ no _ standard imf ( kroupa , chabrier , salpeter , etc . ) , is able to reconcile at the @xmath215 level the cosmic star formation history , the present - day @xmath120-band luminosity density , and the ebl .
they examined how the imf might be altered to make these observations self - consistent , and found that what was required was an imf that had an excess at intermediate masses ( @xmath216 ) over standard imfs .
this imf need not be universal , but instead could be a cumulative imf of all stars that have formed in the universe , since it was obtained from fossil light considerations .
an example of a concordant imf is their paunchy " imf , shown as the blue dot - dashed line in figure [ fig : mchar ] . using the evolving kroupa imf
, it is possible to compute a cumulative imf of all star formation in the universe .
this is done by integrating the evolving imf over cosmic time weighted by the cosmic star formation history , which is taken to be the evolving imf one ( blue line ) from figure [ fig : madau ] .
the result is shown as the dotted cyan line in figure [ fig : mchar ] .
it is similar to the paunchy imf of @xcite , showing elevated formation rates of intermediate mass stars , though it is slightly more top - heavy .
this comparison is preliminary because the ebl and current @xmath120-band light trace stellar output differently than the cosmic star formation history .
still , it is likely that this evolving imf will go significantly towards reconciling the various observations considered in @xcite .
@xcite recently argued for a more top - heavy imf at early times from the evolution of colors and mass - to - light ratios of early - type galaxies in clusters from @xmath217 . parameterizing the evolution in terms of the characteristic mass of a chabrier imf
, he favored a characteristic mass that evolves by @xmath218 from @xmath219 until today ; this is consistent with @xmath220 as proposed here .
note that his analysis only applies to stars formed in early - type cluster galaxies , so is not necessarily applicable to all galaxies . making the assumption that it is widely applicable , @xcite determined a cosmic stellar mass growth rate that is broadly consistent with the evolving imf case shown in figure [ fig : madau ] ( though in detail it is more peaked towards lower @xmath9 ;
see his fig .
13 ) . while many uncertainties are present in the his analysis , it is reassuring that similar imf evolution is inferred from a completely independent line of argument .
the ultimate test of an evolving imf is to directly measure the stellar mass buildup in the universe .
recently , @xcite and @xcite did just that using deep _
spitzer_/irac observations that probe near - ir light all the way out to @xmath203 .
their data of cosmic stellar mass mass growth rates as a function of redshift are shown as the magenta and red data points in figure [ fig : madau ] , respectively ( assuming a salpeter imf ) .
these data points were obtained by differentiating the stellar mass densities as a function of redshift published in those papers .
although those two samples are independent , both show that the stellar mass growth rate is significantly different than that inferred from the observed cosmic star formation history , particularly at @xmath221 .
@xcite notes that this may favor an imf weighted more towards massive stars at high redshifts .
they mention that utilizing a chabrier or kroupa imf will lessen the discrepancy , but such a reduction is essentially by a fixed factor at all redshifts ( modulo minor stellar evolution differences ) , so even if the normalization is adjusted to agree with the cosmic sfh at @xmath13 , it would still yield significantly lower sfrs at @xmath221 .
a major uncertainty in this comparison is that @xcite , in order to integrate up the total stellar mass , assumed a @xcite faint - end slope that is fixed at @xmath222 , even though they can not directly constrain @xmath156 at @xmath221 .
they justify this assumption by noting that there is no evidence for a change in @xmath156 from @xmath223 .
but if @xmath156 becomes progressively steeper with redshift , as suggested by @xcite , it could also reconcile these observations .
note that it would have to steepen quickly , e.g. to @xmath224 at @xmath211 to reconcile the @xmath7 difference , significantly faster than @xmath225 observed by @xcite out to @xmath226 .
@xcite also found that @xmath156 does increase with redshifts slightly in one form of their analysis , though again not by enough to explain the discrepancy .
still , the difficulty of constraining the faint - end slope at high-@xmath9 , along with inherent uncertainties in sed fitting , mean that these observations can not be considered definitive proof of a varying imf .
but at least at face value the agreement with the evolving imf case is quite good .
@xcite independently determined the cosmic stellar mass growth rate by compiling observations of stellar mass densities from the literature for @xmath227 ( note that they do not include the perez - gonzalez et al . or elsner et al .
carefully accounting for various different systematics , they fit a @xcite form to the implied cosmic star formation history ( including effects of stellar mass loss ) , and determine @xmath228^{2.2}$ ] , in @xmath160yr@xmath161mpc@xmath115 .
this relation is shown as the cyan dashed line in figure [ fig : madau ] ; it has been adjusted by @xmath229 dex to change from the imf they assumed ( similar to chabrier ) to a salpeter imf .
it agrees well with other observations , and lies significantly below the star formation rate inferred from high - mass sfr indicators at high-@xmath9 .
wilkins et al . suggest that an evolving imf may be the cause , and propose their own form of imf evolution to explain it .
as seen in figure [ fig : madau ] , the agreement with the evolving imf proposed here is quite good . in short , an evolving kroupa imf does not obviously introduce any large inconsistencies with high-@xmath9 galaxy data , and moreover may be preferred based on low - redshift fossil light considerations and observed cosmic stellar mass growth rates . hence the amplitude evolution of the @xmath0sfr relation may be yet another example of an observation of stellar mass buildup that suggests that the imf is weighted towards more massive stars at higher redshifts .
implications of the observed stellar mass star formation rate correlation are investigated in the context of current theories for stellar mass assembly .
the key point , found here and pointed out in previous studies @xcite , is that the amplitude of the @xmath0sfr relation evolves much more rapidly since @xmath8 in observations that in current galaxy formation models .
it is shown here that this is true of both hydrodynamic simulations and semi - analytic models , and is broadly independent of feedback parameters .
in contrast , the slope and scatter of the observed @xmath0sfr relation are in good agreement with models .
the tight @xmath0sfr relation with a slope near unity predicted in models is a generic result owing to the dominance of cold mode accretion , particularly in early galaxies , which produces rapid , smooth and relatively steady infall .
the slow amplitude evolution arises because star formation starts at @xmath65 for moderately massive star - forming galaxies and continues at a fairly constant or mildly declining level to low-@xmath9 . the large discrepancy in amplitude evolution when compared to observations from @xmath206 hints at some underlying problem either with the models or with the interpretation of data . a convenient parameterization of the problem is through the star formation activity parameter , @xmath230 , where @xmath64 is the hubble time
. a low value corresponds to a starbursting system , a high value to a passive system , and a value near unity to system forming stars constantly for nearly a hubble time . in models , @xmath6 remains constant around unity from @xmath231 , whereas in observations it rises steadily from @xmath232 at @xmath8 to close to unity at @xmath23 .
several ad hoc modifications to the theoretical picture of stellar mass assembly are considered in order to match @xmath6 evolution , but each one is found to be in conflict with other observations of high - redshift galaxies .
bursts seem unlikely given the low scatter in @xmath0sfr .
delaying galaxy formation to match @xmath6 results in such a low redshift for the onset of star formation , it is in conflict with observations of star forming galaxies at earlier epochs . hiding a large population of galaxies as passive runs into difficulty when compared to direct observations of the passive galaxy fraction in mass - selected samples at @xmath8 .
having an exponentially growing phase of star formation or staged " galaxy formation are perhaps the most plausible solutions from an observational viewpoint , but it is difficult to understand theoretically how such scenarios can arise within hierarchical structure formation .
hence if observed star - forming galaxies are quiescently forming stars and represent the majority of galaxies at @xmath8 , as other observations seem to suggest , then it is not easy to accomodate the low @xmath6 inferred from the @xmath0sfr amplitude
. systematic uncertainties in @xmath2 or sfr determinations could bias results progressively more at high-@xmath9 in order to mimic evolution in @xmath6 .
various currently debated sources are considered , such as the contribution to near - ir light from tp - agb stars , extinction corrections , assumptions about star formation history , agn contamination , and pah emission calibration .
it may be possible to concoct scenarios whereby several of these effects combine to mimic @xmath6 evolution , but there are no suggestions from local observations that such scenarios are to be expected
. hence if systematic effects are to explain the low @xmath6 at high-@xmath9 , it would imply significant and unexpected changes in tracers of star formation and stellar mass between now and high redshifts .
a solution is proposed that the stellar initial mass function becomes increasingly bottom - light to higher redshifts .
several lines of arguments are presented that vaguely or circumstantially favor such evolution , though no smoking gun signatures are currently known .
a simple model of imf evolution is constructed , based on the ansatz by @xcite that the minimum temperature of molecular clouds is reflected in the characteristic mass of star formation ( @xmath142 ) where the mass contribution per logarithmic mass bin is maximized .
the minimum temperature may increase with redshift owing to a hotter cosmic microwave background , more vigorous star formation activity , or lower metallicities within early galactic isms . in order to reconcile the observed @xmath6 evolution with theoretical expectations of no evolution ,
an evolving imf of the form @xmath233 is proposed .
the exponent of @xmath142 evolution is constrained by requiring no @xmath6 evolution , through careful modeling with the pegase.2 population synthesis code . while the exact form of imf evolution is not well constrained by present observations ,
what is required is that the imf has progressively more high - mass stars compared to low - mass at earlier epochs , and that this imf applies to the majority of star forming galaxies at any epoch .
it is worth noting that this evolving imf is only constrained out to @xmath8 from the @xmath0sfr relation , though it yields predictions that are consistent with other observations out to @xmath226 .
extrapolating such evolution to higher redshifts is dangerous , since no observational constraints exist and the precise cause of imf evolution is not understood .
implications of such an evolving imf are investigated . by leaving the high - mass end of the imf unchanged , recent successes in understanding the connections between high - mass star formation , feedback , and metal enrichment
are broadly preserved .
the cosmic stellar mass accumulation rate would be altered compared to what is inferred from cosmic star formation history measurements using a standard imf .
it is shown that an evolving imf is at face value in better agreement with direct measures of cosmic stellar mass assembly @xcite .
furthermore , the evolving imf goes towards relieving the generic tension between present - day fossil light measures versus observations of the cosmic star formation history .
in particular , the paunchy imf favored by @xcite in order to reconcile the observed cosmic star formation history , present - day @xmath120-band luminosity density , and extragalactic background light constraints , is qualitatively similar to the cumulative imf of all stars formed by today in the evolving imf case .
individually , each argument has sufficient uncertainties to cast doubt on whether a radical solution such as an evolving imf is necessary . but
taken together , the @xmath0sfr relation adds to a growing body of circumstantial evidence that the ratio of high - mass to low - mass stars formed is higher at earlier epochs .
it is by no means clear that imf variations are the only viable solution to the @xmath0sfr dilemma .
the claim here is only that an evolving imf is an equally ( un)likely solution as invoking unknown systematic effects or carefully crafted star formation histories in order to explain @xmath0sfr evolution .
it is hoped that this work will spur further efforts , both observational and theoretical , to investigate this important issue .
future plans include performing a more careful analysis of the evolving imf in terms of observable quantities , in order to accurately quantify the impact of such imf evolution on the interpretation of uv , near - ir , and mid - ir light .
also , it is feasible to incorporate such an evolving imf directly into simulation runs to properly account for gas recycling in such a scenario . finally , including some feedback mechanism to truncate star formation in massive systems may impact the @xmath0sfr relation in some way , so this will be incorporated into the hydro simulations .
observationally , pushing sfr and @xmath2 determinations to higher redshifts is key ; a continued drop in @xmath6 to @xmath211 would rapidly solidify the discrepancies with current models , and would also generate stronger conflicts with other observations of high-@xmath9 galaxies . assessing the agn contribution and extinction uncertainties from high-@xmath9 systems is critical for accurately quantifying the light from high - mass star formation , for instance through the use of more direct star formation indicators such as paschen-@xmath156 . pushing observations further into the mid - ir such as to 70@xmath119 ,
past the pah bands at @xmath8 , would mitigate pah calibration uncertainties in sfr estimates . obtaining a large sample of spectra for typical high-@xmath9 star - forming systems ( like cb58 ; * ? ? ?
* ) would more accurately constrain the sed than broad - band data .
all of these programs push current technological capabilities to their limits and perhaps beyond , but are being planned as facilities continue their rapid improvement . in summary
, owing to the robust form of star formation histories in current galaxy formation models , the @xmath2-sfr relation represents a key test of our understanding of stellar mass assembly .
current models reproduce the observed slope and scatter remarkably well , but broadly fail this test in terms of amplitude evolution .
whether this reflects some fundamental lack of physical insight , or else some missing ingredient such as an evolving imf , is an issue whose resolution will have a significant impact on our understanding of galaxy formation .
the author is grateful for valuable discussions with emanuele daddi , arjun dey , mark dickinson , david elbaz , mark fardal , kristian finlator , karl glazebrook , andrew hopkins , duan kere , janice lee , heather morrison , kai noeske , casey papovich , naveen reddy , greg rudnick , j. d. smith , pieter van dokkum , david weinberg , and steve wilkins .
the author appreciates stimulus from neal katz , who been espousing a top - heavy imf at high redshifts for some time .
the author thanks h .- w .
rix and the max - planck institt fr astronomie for their gracious hospitality while some of this work was being done .
the simulations used here were run by ben d. oppenheimer on grendel , our department s beowulf system at the university of arizona , and the xeon linux cluster at the national center for supercomputing applications .
support for this work , part of the spitzer space telescope theoretical research program , was provided by nasa through a contract issued by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was also provided by nasa through grant number hst - ar-10946 from the space telescope science institute , which is operated by aura , inc . under nasa contract nas5 - 26555 .
bell , e. f. , zheng , x. z. , papovich , c. , borch , a. , wolf , c. , meisenheimer , k. 2007 , apj , 663 , 834 baldry , i. k. & glazebrook , k. 2003 , apj , 593 , 258 birnboim , y. & dekel , a. 2003 , mnras , 345 , 349 borch , a. , meisenheimer , k. , bell , e. f. , rix , h .- w . , wolf , c. , dye , s. , kleinheinrich , m. , kovacs , z. , wisotzki , l. 2006 , a&a , 453 , 869 bouwens , r. j. , illingworth , g. d. , franz , m. , ford , h. c. 2007 , apj , in press , arxiv:0707.2080 brooks , a. m. , governato , f. , booth , c. m. , willman , b. , gardner , j. p. , wadsley , j. , stinson , g. , quinn , t. 2007 , apjl , 655 , l17 bruzual , g. a. 2007 , in proc .
iau symposium no .
241 stellar populations as building blocks of galaxies " , eds .
a. vazdekis and r. peletier , cambridge : cambridge university press , astro - ph/0703052 bruzual , g. a. & charlot , s. 2003 , mnras , 344 , 1000 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. , koornneef , j. , storchi - bergmann , t. 2000 , apj , 533 , 682 calzetti , d. , et al .
2007 , apj , 666 , 870 chabrier , g. 2003 , pasp , 115 , 763 chary , r .-
2007 , apj , submitted , arxiv:0712.1498 cole , s. et al .
2001 , mnras , 326 , 255 cowie , l. l. , songaila , a. , hu , e. m. , cohen , j. g. 1996 , aj , 112 , 839 daddi , e. , dickinson , m. , morrison , g. , chary , r. , cimatti , a. , elbaz , d. , frayer , d. , renzini , a. , pope , a. , alexander , d. m. , bauer , f. e. , giavalisco , m. , huynh , m. , kurk , j. , mignoli , m. 2007 , apj , submitted , arxiv:0705.2831 daddi , e. , alexander , d. m. , dickinson , m. , gilli , r. , renzini , a. , elbaz , d. , cimatti , a. , chary , r. , frayer , d. , bauer , f. e. , brandt , w. n. , giavalisco , m. , grogin , n. a. , huynh , m. , kurk , j. , mignoli , m. , morrison , g. , pope , a. , ravindranath , s. 2007 , apj , accepted , arxiv:0705.2832 dale , d. a. et al .
2006 , apj , 646 , 161 dav , r. , finlator , k. , & oppenheimer , b. d. 2006 , mnras , 370 , 273 dav , r. & oppenheimer , b. d. 2007 , mnras , 374 , 427 de lucia , g. & blaizot , j. 2007 , mnras , 375 , 2 drory , n. , salvato , m. , gabasch , a. , bender , r. , hopp , u. , feulner , g. , pannella , m. 2005 , apjl , 619 , l131 elbaz , d. , daddi , e. , le borgne , d. , dickinson , m. , alexander , d. m. , chary , r .- r .
, starck , j .- l . , brandt , w. n. , kitzbichler , m. , macdonald , e. , nonino , m. , popesso , p. , stern , d. , vanzella , e. 2007 , a&a , 468 , 33 elsner , f. , feulner , g. , hopp , u. 2007 , a&a , in press , arxiv:0711.0384 erb , d. k. , shapley , a. e. , pettini , m. , steidel , c. c. , reddy , n. a , & adelberger , k. l. 2006a , apj , 644 , 813 erb , d. k. , steidel , c. c. , shapley , a. e. , pettini , m. , reddy , n. a. , adelberger , k. l. 2006b , apj , 646 , 107 erb , d. k. , steidel , c. c. , shapley , a. e. , pettini , m. , reddy , n. a. , adelberger , k. l. 2006c , apj , 647 , 128 fardal , m. a. , katz , n. , weinberg , d. h. , dav , r. , katz , n. 2007 , mnras , 379 , 985 ferguson , h. c. , dickinson , m. , papovich , c. 2002 , apjl , 569 , l65 figer , d. f. 2005 , in the formation and evolution of massive young star clusters " , asp conf .
series , vol .
lamers , l.j .
smith , a. nota , san francisco : asp , p.49 finlator , k. , dav , r. , papovich , c. , & hernquist , l. 2006 , apj , 639 , 672 finlator , k. , dav , oppenheimer , b. d. 2007 , mnras , 376 , 1861 finlator , k. & dav 2007 , mnras , submitted , arxiv:0704.3100 fioc , m. , rocca - volmerange , b. 1997 , a&a , 326 , 950 fontana , a. , pozzetti , l. , donnarumma , i. , renzini , a. , cimatti , a. , zamorani , g. , menci , n. , daddi , e. , giallongo , e. , mignoli , m. , perna , c. , salimbeni , s. , saracco , p. , broadhurst , t. , cristiani , s. , dodorico , s. , gilmozzi , r. 2004 , a&a , 424 , 23 fontana , a. et al .
2006 , a&a , 459 , 745 guo , q. & white , s. d. m. 2007 , arxiv:0708.1814 heavens , a. , panter , b. , jimenez , r. , dunlop , j. 2004 , nature , 428 , 625 hopkins , a. m. 2004 , apj , 615 , 209 hopkins , a. m. & beacom , j. f. 2006 , apj , 651 , 142 hopkins , p. f. , cox , t. j. , kere , d. , hernquist , l. 2007 , apj , submitted , arxiv:0706.1246 jappsen , a .- k . , klessen , r. s. , larson , r. b. , li , y. , mac low , m .- m .
2005 , a&a , 435 , 611 katz , n. , kere , d. , dav , r. , weinberg , d. h. 2003 , in igm / galaxy connection- the distribution of baryons at @xmath13 " , assl conf .
series vol .
281 , eds . j. l. rosenberg & m. e. putman , kluwer , dodrecht , p.185 keres , d. , katz , n. , weinberg , d. h. , & dav , r. 2005 , mnras , 363 , 2 kennicutt , r. c. 1998 , apj , 498 , 541 kitzbichler , m. g. , white , s. d. m. 2007 , mnras , 376 , 2 kobayashi , c. , springel , v. , white , s. d. m. 2007 , mnras , 376 , 1465 kolatt , t. s. , bullock , j. s. , somerville , r. s. , sigad , y. , jonsson , p. , kravtsov , a. v. , klypin , a. a. , primack , j. r. , faber , s. m. , dekel , a. 1999 , 523 , l109 kroupa , p. 2001 , mnras , 322 , 231 kroupa , p. 2007 , in proc .
resolved stellar populations " , eds .
d. valls - gabaud and m. chavez , asp conf .
( in press ) , astro - ph/0703124 kroupa , p. , tout , c. a. , gilmore , g. 1993 , mnras , 262 , 545 lacey , c. g. , baugh , c. m. , frenk , c. s. , silva , l. , granato , g. l. , bressan , a. 2007 , mnras , submitted , arxiv:0704.1562 larson , r. b. 1985 , mnras , 214 , 379 larson , r. b. 1998 , mnras , 301 , 569 larson , r. b. 2005 , mnras , 359 , 211 li , y. , mo , h. j. , van den bosch , f. c. , lin , w. p. 2007 ,
mnras , 379 , 689 lucatello , s. , gratton , r. g. , beers , t. c. , carretta , e. 2005 , apj , 625 , 833 madau , p. , pozzetti , l. , dickinson , m. 1998 , apj , 498 , 106 maraston , c. , daddi , e. , renzini , a. , cimatti , a. , dickinson , m. , papovich , c. , pasquali , a. , pirzkal , n. 2006 , apj , 652 , 85 miller , g. e. & scalo , j. m. 1979 , apjs , 41 , 513 mo , h. j. , mao , s. , white , s. d. m. 1998 , mnras , 295 , 319 murali , c. , katz , n. , hernquist , l. , weinberg , d. h. , dav , r. 2002 , apj , 571 , 1 neistein , e. , van den bosch , f. c. , dekel , a. 2006 , mnras , 372 , 933 noeske , k. g. , et al .
2007a , apjl , 660 , l43 noeske , k. g. , et al .
2007b , apjl , 660 , l47 oppenheimer , b. d. & dav , r. 2006 , mnras , 373 , 1265 oppenheimer , b. d. & dav , r. 2007 , in preparation papovich , c. , dickinson , m. , ferguson , h. c. 2001 , apj , 559 , 620 papovich , c. et al .
2006 , apj , 640 , 92 perez - gonzalez , p. g. , rieke , g. h. , villar , v. , barro , g. , blaylock , m. , egami , e. , gallego , j. , gil de paz , a. , pascual , s. , zamorano , j. , donley , j. l. 2007 , apj , accepted , arxiv:0709.1354 pettini , m. , steidel , c. c. , adelberger , k. l. , dickinson , m. , giavalisco , m. 2000 , apj , 528 , 96 portinari , l. , moretti , a. , chiosi , c. , sommer - larsen , j. 2004 , apj , 604 , 579 reddy , n. a. , erb , d. k. , steidel , c. c. , shapley , a. e. , adelberger , k. l. , pettini , m. 2005 , apj , 633 , 748 reddy , n. a. , steidel , c. c. , pettini , m. , adelberger , k. l. , shapley , a. e. , erb , d. k. , dickinson , m. 2007 , apj , in press renzini , a. 2005 , in the initial mass function 50 years later , " eds .
e. corbelli and f. palle , assl , spinger conf .
v.327 , p.221 rieke , g. h. , lebofsky , m. j. , thompson , r. i. , low , f. j. , tokunaga , a. t. 1980 , apj , 238 , 24 rieke , g. h. , loken , k. , rieke , m. j. , tamblyn , p. 1993
, apj , 412 , 99 rudnick , g. et al .
2006 , apj , 650 , 624 salim , s. et al .
2007 , apj , in press , arxiv:0704.3611 savaglio , s. , et al .
2005 , apj , 635 , 260 schechter , p. 1976
, apj , 203 , 297 scott , j. e. , bechtold , j. , morita , m. , dobrzycki , a. , kulkarni , v. 2002 , apj , 571 , 665 scoville , n. et al .
2007 , apjs , 172 , 150 shapley , a. e. , steidel , c. c. , erb , d. k. , reddy , n. a. , adelberger , k. l. , pettini , m. , barmby , p. , huang , j. 2005 , apj , 626 , 698 sirianni , m. , nota , a. , leitherer , c. , de marchi , g. , clampin , m. 2000 , apj , 533 , 203 smith , j .- d . t. , et al .
2007 , apj , 656 , 770 springel , v. & hernquist , l. 2003 , mnras , 339 , 312 springel , v. 2005 , mnras , 364 , 1105 tumlinson , j. 2007 , apj , 664 , l63 van dokkum , p. g. & van der marel , r. p. 2007 ,
apj , 655 , 30 van dokkum , p. g. 2007 , apj , accepted , arxiv:0710.0875 weidner , c. & kroupa , p. 2006 , mnras , 365 , 1333 wilkins , s. m. , trentham , n. , hopkins , a. m. 2007 , mnras , accepted , arxiv:0801.1594 | the evolution of the galaxy stellar mass star formation rate relationship ( @xmath0sfr ) provides key constraints on the stellar mass assembly histories of galaxies . for star - forming galaxies ,
@xmath0sfr is observed to be fairly tight with a slope close to unity from @xmath1 , and it evolves downwards roughly independently of @xmath2 .
simulations of galaxy formation reproduce these trends , broadly independent of modeling details , owing to the generic dominance of smooth and steady cold accretion in these systems .
in contrast , the observed amplitude of the @xmath0sfr relation evolves markedly differently than in models , indicating either that stellar mass assembly is poorly understood or that observations have been misinterpreted .
stated in terms of a star formation activity parameter @xmath3 , models predict a constant @xmath4 out to redshifts @xmath5 , while the observed @xmath0sfr relation indicates that @xmath6 increases by @xmath7 from @xmath8 until today . the low @xmath6 ( i.e. rapid star formation ) at high-@xmath9 not only conflicts with models , but is also difficult to reconcile with other observations of high-@xmath9 galaxies , such as the small scatter in @xmath0sfr , the slow evolution of star forming galaxies at @xmath10 , and the modest passive fractions in mass - selected samples .
systematic biases could significantly affect measurements of @xmath2 and sfr , but detailed considerations suggest that none are obvious candidates to reconcile the discrepancy .
a speculative solution is considered in which the stellar initial mass function ( imf ) evolves towards more high - mass star formation at earlier epochs . following larson ( 1998 )
, a model is investigated in which the characteristic mass @xmath11 where the imf turns over increases with redshift .
population synthesis models are used to show that the observed and predicted @xmath0sfr evolution may be brought into general agreement if @xmath12 out to @xmath8 .
such imf evolution matches recent observations of cosmic stellar mass growth , and the resulting @xmath13 cumulative imf is similar to the paunchy " imf favored by @xcite to reconcile the observed cosmic star formation history with present - day fossil light measures .
[ firstpage ] galaxies : evolution , galaxies : formation , galaxies : high - redshift , cosmology : theory , stars : mass function |
heavy quark production in hadronic collisions has received much attention in studying aspects of quantum chromodynamics ( qcd ) because its large mass sets the scale for perturbative analyses @xcite . in recent years , productions of quarkonia and open heavy flavor mesons in proton - nucleus ( p@xmath1 )
collisions are investigated from the viewpoint of the gluon saturation in hadrons @xcite by taking into account the nonlinear evolution effects in the color - glass - condensate ( cgc ) formalism @xcite ( see also @xcite ) .
high - energy p@xmath1 collisions can be regarded as a good laboratory for gluon saturation physics : first , the saturation momentum @xmath7 , which separates the nonlinear regime ( @xmath8 ) from the linear regime ( @xmath9 ) of a qcd process with an external scale @xmath10 , is enhanced in the heavy nucleus and at high energies .
this is because the saturation scale is expected to be proportional basically to the gluon density per unit transverse area of the target nucleus and scales as @xmath11 with bjorken s @xmath12 being a longitudinal momentum fraction carried by a gluon and with an empirical parameter @xmath13 @xcite .
second , it is conventionally assumed that no hot and dense quark - gluon medium is formed in p@xmath1 collisions , and thus one can study the initial state effects more directly there collisions @xcite . ] . at forward
rapidities at the lhc , the relevant @xmath12 values become very small ( @xmath14 ) even for the heavy quark production , and the corresponding saturation scale will be @xmath15 . therefore , besides light hadron productions , heavy quark production can be used as a quantitative probe for the gluon saturation in the target nucleus at the lhc . at the same time
, we note that p@xmath1 collisions provide an important baseline for quantifying the nuclear modification of hadron productions by a quark - gluon medium created in @xmath16 collisions .
open heavy flavor meson productions ( @xmath4 and @xmath17 ) in pp and p@xmath1 collisions were previously evaluated in the cgc framework at mid rapidity at the lhc @xcite . in the rhic and lhc experiments , the decay leptons the electrons @xmath18 at mid and the muons @xmath19 at forward rapidity are also detected for the study of heavy quark production .
importantly , the muon detection is the only observable relevant to the heavy flavor production at forward rapidity in the rhic and lhc experiment setup .
then we need to evaluate the leptonic decays of open heavy flavor mesons ( @xmath20 channel ) in order to study the saturation effects in the observed forward muon spectra . as was done in refs .
@xcite , we can compute the lepton production from the semileptonic decays of the heavy flavor mesons by convoluting the meson production cross - sections with the decay functions .
the main purpose of this study is to quantify in the cgc framework the sensitivity of the single lepton spectra to the small-@xmath12 gluon distribution in the target nucleus . in the cgc framework ,
the heavy quark pair production cross - section is available analytically at the leading order in the strong coupling constant @xmath21 and at the order @xmath22 for a dilute - dense system , where @xmath23 ( @xmath24 ) is the color charge density in the proton ( nucleus ) . to this order ,
the physical picture of the pair production is simple : one is the process in which an incoming gluon from the proton splits into a quark pair and the pair is multiply scattered in the target nucleus , and the other is the process in which an incoming gluon is multiply scattered in the target nucleus and then splits into a quark pair . the multiple scattering of the quark pair in the target in the eikonal approximation is represented by multipoint functions , which obey the so - called jimwlk evolution equation in the rapidity @xmath25 @xcite . in the large-@xmath26 approximation with @xmath26 being the number of colors ,
these general multipoint functions are written in terms of the two - point function ( dipole amplitude ) and the four - point function . especially , the quark - pair production cross - section inclusive in the color degrees is written with the dipole amplitude only .
the rapidity evolution of the dipole amplitude is controlled by the balitsky - kovchegov ( bk ) equation @xcite which sums up the small-@xmath12 quantum corrections of orders @xmath27 in the dilute regime and non - linear effects in the dense regime . at the present day , the bk equation with running coupling kernel ( rcbk ) @xcite is regarded as a useful tool for phenomenologies .
the rcbk equation gives the appropriate evolution speed of the saturation scale @xmath28 .
the initial condition for the dipole amplitude in the rcbk equation is constrained by a global analysis of hera dis data .
this paper is organized as follows : in sec . 2
, we first give a brief review of heavy quark production in the cgc framework with the rcbk equation which is utilized in this paper .
then , we introduce a decay distribution function for describing the semileptonic decay of heavy mesons .
after that , in sec . 3 , we show the results of single lepton production spectra and its nuclear modification factor .
in the cgc framework , the heavy quark pair production cross - section at the leading order in @xmath21 with the transverse momentum @xmath29 and rapidity @xmath30 of the ( anti- ) quark in minimum bias events of a dilute - dense system ( e.g. , p@xmath1 ) is given in the large-@xmath26 limit as @xcite @xmath31 where @xmath32 , @xmath33 , and @xmath34 and @xmath35 are the unintegrated gluon distribution function of the proton and the three - point function of the nucleus , respectively ( see fig . [ lo - diagrams ] ) .
the values of bjorken s @xmath12 on the proton and nucleus sides , @xmath36 and @xmath37 , respectively , are determined in the @xmath38 kinematics as @xmath39 with the transverse mass @xmath40 .. ] the relevant hard matrix element @xmath41 can be found in ref .
the integrand depends on two transverse momenta , @xmath42 and @xmath43 , on the nucleus side , by which the @xmath44-factorization is broken in a strict sense @xcite . with this caution in mind , we call this formula @xmath44-factorized formula in a generalized version
. in the forward rapidity region , where @xmath36 becomes large while @xmath37 is much smaller , we can use the hybrid formula @xcite : @xmath45 where @xmath46 is the collinear gluon distribution function with @xmath19 being factorization scale .
since we work at the leading order in @xmath21 , @xmath19 here should be regarded as a model parameter .
the explicit expression for the hard matrix element @xmath47 after the collinear approximation done on the proton side can be found in ref .
. the single heavy quark production cross - section is obtained by integrating the pair production cross - section over the phase space of the anti - quark : @xmath48 the unintegrated gluon distribution function @xmath49 of the proton is expressed in terms of the fourier transform of the dipole amplitude @xmath50 in the adjoint representation , @xmath51 where @xmath52 is the evolution rapidity measured from @xmath53 , and the transverse geometry of the proton is replaced with the effective transverse area @xmath54 . in the large-@xmath26 limit , the adjoint dipole amplitude reduces to the square of the fundamental dipole amplitude in the coordinate space , and therefore @xmath55 with @xmath56 the collinear gluon distribution function of the proton @xmath57 is obtained by integrating @xmath58 over the transverse momentum @xmath44 up to the factorization scale @xmath19 : @xmath59 we note here that @xmath58 is different from the unintegrated gluon distribution function in a strict sense because the former includes higher twist effects .
when we compare the gluon distribution obtained by eq .
( [ eq : xg - cgc ] ) with the collinear gluon distribution , e.g. , cteq6l pdf @xcite , we easily find a difference by a factor of two or so . similarly the three - point function of the nucleus can be expressed in the large-@xmath26 limit with the fundamental dipole amplitude as @xmath60 where @xmath61 is the effective transverse area of the target nucleus .
we ignore the transverse profile of the target nucleus here for simplicity . a recent attempt to include the centrality dependence in p@xmath1 collisions
is found in ref .
@xcite . by construction
the three - point function is related to the two - point function via @xmath62 , and this relation is respected in the large-@xmath26 approximation .
rapidity dependence of the dipole amplitude @xmath63 in the fundamental representation is described in the large-@xmath26 limit by the rcbk equation @xcite . using this solution
, we express the three - point gluon function @xmath64 in the nucleus .
this equation resums quantum corrections of orders @xmath65 in the linear regime , while by unitarity it tames the growth of the transition amplitude in the nonlinear regime .
phenomenological analyses with the rcbk equation have been successful in describing hera - dis data as well as lhc hadron production data @xcite .
the rcbk equation is written in the coordinate space as @xmath66 \ ; , \end{aligned}\ ] ] where the evolution kernel in balitsky s prescription @xcite is @xmath67 \label{eq : rcbk - kernel}\end{aligned}\ ] ] with @xmath68 being the transverse position of the dipole . here
the running coupling constant in the coordinate space appears , for which we assume the following form motivated by the one - loop expression in the momentum space : @xmath69^{-1 } \ , .\end{aligned}\ ] ] note that we have introduced a smooth infrared cutoff @xmath70 so as to satisfy @xmath71 , as was used in ref .
@xcite .
.parameter values of the initial dipole amplitude and the running coupling constant in the coordinate space .
[ tab : par ] [ cols="^,^,^,^,^,^",options="header " , ] of mv@xmath72 ( solid line ) and mv@xmath73 ( dotted line).,height=298 ] by setting the initial condition for the rcbk equation at @xmath74 in a modified form from the mclerran - venugopalan model @xcite : @xmath75 \ , , \end{aligned}\ ] ] the dipole amplitude is constrained by global fit of hera dis data . as we have shown in ref .
@xcite , the original mv model ( @xmath76 ) gives rise to a too - hard @xmath2 spectrum of heavy mesons at high @xmath2 . in this study
we apply two parameter sets , mv@xmath77 and mv@xmath73 , as listed in table [ tab : par ] @xcite with @xmath78 gev . for a heavy nucleus
we replace the initial saturation scale by @xmath79 with @xmath80 in this study , as suggested for minimum bias events from nuclear dis analysis in ref .
@xcite . in fig .
[ phi_gg ] we show the unintegrated gluon distributions of the proton and the nucleus for the sets , mv@xmath72 and mv@xmath73 .
we find that the difference between mv@xmath72 and mv@xmath73 is very small in low @xmath81 region , and that the difference seen in higher @xmath44 region diminishes with the rapidity evolution . at larger @xmath82
, we apply the following extrapolation ansatz @xcite : @xmath83 .
production of heavy flavor meson @xmath84 ( @xmath85 , @xmath86 , etc . ) from the quark @xmath87 is described with the heavy - quark fragmentation function @xmath88 as @xmath89 where the rapidity is set to @xmath90 and the momentum fraction @xmath91 is defined by @xmath92 . in this paper we use the fragmentation function of kartvelishvili s form @xcite @xmath93 with the parameter @xmath94 @xmath95 for @xmath4 @xmath96 meson . we expect that the discussion below will be almost unchanged when we use another form for @xmath97 .
the branching ratio @xmath98 for the quark ( @xmath87 ) to make a transition to the heavy flavor meson @xmath84 satisfies @xmath99 .
for example , @xmath100 @xcite .
now we proceed to lepton production from semileptonic decays of heavy flavor mesons @xmath101 , which can be computed by convoluting the meson production cross - section with the lepton decay function @xmath102 .
then the cross - section is given with the use of eq .
( [ eq : open - cross - section - ln ] ) as : @xmath103 here @xmath104 is the probability for the lepton with momentum @xmath105 to be produced in the decay of the heavy meson with momentum @xmath106 in the laboratory frame , and is expressed as an integral over the phase space @xcite : @xmath107 where @xmath108 is the azimuthal angle between @xmath109 and @xmath110 .
@xmath111 is the distribution of the lepton with energy @xmath112 in the heavy - meson rest frame and is parametrized as @xmath113 the normalization factor @xmath114 $ ] with @xmath115 .
the mass @xmath116 of the produced particle @xmath117 is set to @xmath118 gev in the @xmath4 decay ( @xmath119 gev ) and @xmath120 gev in the @xmath17 decay ( @xmath121 gev ) .
we neglect the lepton masses here because @xmath122 .
here we show the numerical results of the particle spectra computed at rhic and lhc energies .
we adopt cteq6l pdf @xcite with @xmath123 for the collinear gluon distribution function in the hybrid model . here
we neglect the contribution of @xmath124 channel because we expect that this channel gives only a small contribution compared to @xmath125 channel at low @xmath2 and @xmath126 at high @xmath2 . through this paper
, we set @xmath127 fm for the proton s transverse radius and fix @xmath128 in front of the hard matrix elements and use the following branching ratios for the decays of the quark to the lepton @xcite : @xmath129 for @xmath17 decay and @xmath130 for @xmath4 decay . the quark - pair production formula eq .
( [ eq : cross - section - ln ] ) is derived for asymmetric collision systems , and therefore is not best suitable for pp collisions at mid rapidity .
nonetheless it appears as a reference in calculation of the nuclear modification factor at mid rapidity .
let us examine the lepton production in pp collisions here .
[ results - rhic - pp ] shows the @xmath2 distribution of leptons from charmed - hadron and bottomed - hadron semileptonic decays in pp collisions at rhic . in the left panel
, we see that at mid rapidity the cgc result for the charm - decay electrons ( red solid line ) gives a harder @xmath2 slope compared with the data .
the cross - section of the electrons from bottomed - hadron decay is much smaller than that from charmed - hadron decay at lower @xmath2 .
thus we neglect the bottomed - hadron decay contribution as long as we focus on the lepton production less than @xmath131 gev .
we remind here that , at mid rapidity at rhic energy @xmath132 gev , both @xmath36 and @xmath37 are larger than @xmath74 in heavy quark pair production , and that the cgc framework may be only marginally applicable in this kinematical region .
especially , the lepton yields solely depend on the initial condition for the gluon distributions and their extrapolation to larger @xmath133 . in the left panel of fig .
[ results - rhic - pp ] we find little difference between the results with mv@xmath72 ( solid ) and mv@xmath73 ( dotted ) at @xmath134gev . at forward rapidities ,
@xmath37 ( @xmath36 ) becomes smaller ( larger ) than @xmath53 and there appears some room for the effects of the quantum evolution la rcbk to show up in the @xmath2 spectra .
in the right panel of fig .
[ results - rhic - pp ] , we find that the lepton spectrum is still harder than the data at @xmath135 .
the difference between the results with the parametrizations mv@xmath72 and mv@xmath73 is very weak . in this asymmetric case
, we apply the hybrid formula . the resultant @xmath2 spectrum of the muon from charm decay comes closer to the data . at the lhc energy @xmath136 tev ,
the @xmath12 values become much smaller and the use of the cgc framework can be more justified .
the lepton spectra at mid ( electron ) and forward ( muon ) rapidities are shown in fig .
[ results - lhc - pp ] .
although the difference between the numerical results and the data becomes slightly smaller , the spectra from the @xmath44 factorization formula are still harder than the data .
the hybrid formula gives rise to a reasonable description for the muons at forward rapidity both with mv@xmath72 and mv@xmath73 .
the leptons from the b decays ( shown in blue lines ) are much suppressed as compared to those from the c decays in this momentum region @xmath137 gev . at @xmath138 and @xmath135 , respectively , in pp collisions at @xmath132 gev .
red ( blue ) thick solid lines show the lepton yields from the charm ( bottom ) decays obtained by eq .
( [ eq : cross - section - ln ] ) with mv@xmath72 .
the uncertainty band represents the change of quark mass scale : @xmath139 gev for charm and @xmath140 gev for bottom .
dotted lines denote the results with mv@xmath73 .
the results obtained by eq .
( [ eq : cross - section - ln - coll ] ) are depicted with thin lines .
rhic data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite .
, title="fig:",height=298 ] at @xmath138 and @xmath135 , respectively , in pp collisions at @xmath132 gev .
red ( blue ) thick solid lines show the lepton yields from the charm ( bottom ) decays obtained by eq .
( [ eq : cross - section - ln ] ) with mv@xmath72 .
the uncertainty band represents the change of quark mass scale : @xmath139 gev for charm and @xmath140 gev for bottom .
dotted lines denote the results with mv@xmath73 .
the results obtained by eq .
( [ eq : cross - section - ln - coll ] ) are depicted with thin lines .
rhic data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite .
, title="fig:",height=298 ] at @xmath141 and @xmath142 , respectively , in pp collisions at @xmath136 tev .
notations are the same as in fig .
[ results - rhic - pp ] .
lhc data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite .
, title="fig:",height=298 ] at @xmath141 and @xmath142 , respectively , in pp collisions at @xmath136 tev .
notations are the same as in fig .
[ results - rhic - pp ] .
lhc data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite .
, title="fig:",height=298 ] meson as a function of transverse momentum @xmath2 at @xmath143 in ppb collisions at @xmath144 tev .
right : double differential cross - sections of electron from charmed - hadron decay and bottomed - hadron decay as a function of @xmath2 at @xmath145 in ppb collisions at @xmath146 tev .
notations are the same as in fig .
[ results - rhic - pp ] .
lhc data for @xmath4 meson production is taken from ref .
@xcite and for electron from ref .
, title="fig:",height=298 ] meson as a function of transverse momentum @xmath2 at @xmath143 in ppb collisions at @xmath144 tev .
right : double differential cross - sections of electron from charmed - hadron decay and bottomed - hadron decay as a function of @xmath2 at @xmath145 in ppb collisions at @xmath146 tev .
notations are the same as in fig .
[ results - rhic - pp ] .
lhc data for @xmath4 meson production is taken from ref .
@xcite and for electron from ref .
, title="fig:",height=298 ] next , let us turn to heavy - quark decay leptons in p@xmath1 collisions .
we have now two parameters to be fixed for the heavy nucleus , the initial saturation scale for the rcbk evolution and the transverse size of the nucleus .
regarding the initial saturation scale , we choose @xmath147 for the heavy nuclei , au and pb . , which we employed in @xcite . ref .
@xcite used @xmath148 for heavy quark pair production .
] as to the transverse radius of the nucleus , we determine it by imposing the nuclear modification factor @xmath149 at high @xmath2 .
the nuclear modification factor is defined by @xmath150 and it is expected to scale as @xmath151 for large @xmath2 in the cgc formula . in order for @xmath149 at high @xmath2 , we choose here the nuclear radius to be @xmath152 as is done in ref .
this condition with parametrization mv@xmath72 ( mv@xmath73 ) gives @xmath153 ( 6.66 ) fm for pb and 6.08 ( 6.48 ) fm for au .
our calculation can predict now only the @xmath2 and @xmath154 dependence of the nuclear modification factor @xmath155 , which is constrained to be @xmath156 at high @xmath2 .
gev at mid and forward rapidity .
solid ( dotted ) lines are obtained by using eq .
( [ eq : cross - section - ln ] ) ( eq .
( [ eq : cross - section - ln - coll ] ) ) .
the quark mass scale dependence is shown by the different line . filled and
shaded bands are the uncertainty which is reflecting the difference of the initial condition between mv@xmath72 and mv@xmath73 for eq .
( [ eq : cross - section - ln ] ) and ( [ eq : cross - section - ln - coll ] ) , respectively .
rhic data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite .
, title="fig:",height=298 ] gev at mid and forward rapidity .
solid ( dotted ) lines are obtained by using eq .
( [ eq : cross - section - ln ] ) ( eq .
( [ eq : cross - section - ln - coll ] ) ) .
the quark mass scale dependence is shown by the different line . filled and
shaded bands are the uncertainty which is reflecting the difference of the initial condition between mv@xmath72 and mv@xmath73 for eq .
( [ eq : cross - section - ln ] ) and ( [ eq : cross - section - ln - coll ] ) , respectively .
rhic data at mid rapidity is taken from ref .
@xcite and forward data is taken from @xcite . ,
title="fig:",height=298 ] tev .
notations are the same as in fig . [ rpa - rhic ] .
lhc data at mid rapidity is taken from ref .
@xcite , title="fig:",height=298 ] tev .
notations are the same as in fig .
[ rpa - rhic ] .
lhc data at mid rapidity is taken from ref .
@xcite , title="fig:",height=298 ] we use eq .
( [ eq : cross - section - ln ] ) to evaluate the @xmath2 spectra of the @xmath85 and the electron at mid rapidities in p@xmath1 collisions at the lhc energy @xmath146 tev , which is presented in fig .
[ results - lhc - pa ] .
we find that the electron @xmath2 spectrum for charmed - hadron decay of our numerical result is slightly harder than the data although the deviation becomes smaller than in the pp case .
note that the normalization is fixed by the condition for @xmath157 to become unity at large @xmath2 .
now we can compute @xmath157 of the leptons from charmed - hadron decays .
we expect that the systematic uncertainties of the calculation partially cancel out in the ratio . in fig .
[ rpa - rhic ] we show the results of @xmath157 of the electrons at mid rapidities @xmath138 and the muons at forward rapidities @xmath135 at @xmath132 gev , together with the rhic data . at mid rapidity , we find that @xmath157 is almost flat with weak suppression at @xmath158 gev and subtle enhancement at @xmath159 gev .
this structure stems from the multiple scattering effects encoded in the dipole amplitude , while the evolution at @xmath160 is inoperative in the charm production here .
the experimental data of @xmath157 seems systematically larger than unity with some uncertainties . at forward rapidity , where the charm production gets sensitivity to the gluon distribution at @xmath160 ,
the @xmath2 dependence of @xmath157 for the muons becomes slightly more noticeable due to the quantum evolution .
the data is now almost consistent with unity .
we also note that the @xmath44 factorized formula ( [ eq : cross - section - ln ] ) and the hybrid formula ( [ eq : cross - section - ln - coll ] ) give the same result . in fig .
[ rpa - lhc ] we show the electron @xmath157 computed with eq .
( [ eq : cross - section - ln ] ) at mid rapidity @xmath161 at @xmath146 tev .
it is suppressed at low @xmath2 and is recovering to unity monotonically with increasing @xmath2.this behavior reflects the @xmath12-evolution of the gluon distribution , but the smearing through the fragmentation and decay processes seems to make the effect relatively less prominent ( see fig . [ d - rpa - lhc ] below for @xmath4 meson case ) .
we also show in fig .
[ rpa - lhc ] a model prediction of @xmath157 at forward rapidity at @xmath146 tev .
as is found in @xmath4 meson and @xmath162 productions @xcite , we also expect a stronger suppression of the lepton @xmath157 at forward rapidity compared to that of at mid rapidity due to the stronger saturation effect .
spectrum of muon from charmed - hadron decay in pp and p@xmath1 collisions at forward rapidity ( @xmath135 ) at @xmath132 gev : green solid line denotes the full result , red dotted line the contribution from the saturation region ( @xmath163 ) , and blue dashed line the contribution from the extended scaling region ( @xmath164 ) .
the uncertainty band indicates the change of quark mass scale .
all the results are computed with eq .
( [ eq : cross - section - ln - coll ] ) . , title="fig:",height=298 ] spectrum of muon from charmed - hadron decay in pp and p@xmath1 collisions at forward rapidity ( @xmath135 ) at @xmath132 gev :
green solid line denotes the full result , red dotted line the contribution from the saturation region ( @xmath163 ) , and blue dashed line the contribution from the extended scaling region ( @xmath164 ) .
the uncertainty band indicates the change of quark mass scale .
all the results are computed with eq .
( [ eq : cross - section - ln - coll ] ) . , title="fig:",height=298 ] spectrum of the muons from charmed - hadron decay in pp and p@xmath1 collisions at forward rapidity ( @xmath165 ) at @xmath146 tev .
notations are the same as in fig .
[ cgc - extended - rhic ] . , title="fig:",height=298 ] spectrum of the muons from charmed - hadron decay in pp and p@xmath1 collisions at forward rapidity ( @xmath165 ) at @xmath146 tev .
notations are the same as in fig .
[ cgc - extended - rhic ] . ,
title="fig:",height=298 ] in order to quantify the saturation effects more precisely , we investigate fractions of the small-@xmath12 gluon contributions to the total spectrum in pp and p@xmath1 collisions . here , we introduce two cutoff scales in the integration variable @xmath166 of the multipoint function @xmath108 in the integral : one is the saturation scale @xmath167 , which we determined by the condition @xmath168 with the solution of the rcbk equation , the other is the scale @xmath169 which corresponds to the extended geometrical scaling line @xcite
. we will call the momentum regions of @xmath163 and @xmath170 the saturation region and the extended scaling region , respectively .
we use these representative scales in this work , and study the fractional contributions of these kinematical regions to the cross - section of the forward muon production , using the hybrid formula ( [ eq : cross - section - ln - coll ] ) with mv@xmath72 parametrization .
[ cgc - extended - rhic ] shows the fractions of the saturation and extended scaling regions to the differential cross - section of the muons from charmed - hadron decays at the rhic energy . in pp collisions the saturation region and the extended scaling region
give only fractional contributions to the cross - section at low @xmath171 gev .
this is because the saturation scale of proton is small at rhic .
the muons at @xmath172 gev are produced from the gluons outside the saturation regime . in p@xmath1 collisions , the saturation scale of nucleus
is enhanced .
then , the extended scaling region is pushed toward larger @xmath2 region although the saturation regime gives a fractional contribute to the total .
we show in fig .
[ cgc - extended - lhc ] the results at the lhc energy .
one sees that both the saturation and extended scaling regions expand toward higher @xmath2 in pp and p@xmath1 collisions , compared to the results at rhic .
remarkably , the cross - section in p@xmath1 collisions at @xmath173 gev is largely covered by the extended scaling region .
finally , we re - evaluate @xmath157 of @xmath162 and @xmath4 meson productions , since we use @xmath79 with @xmath80 in the rcbk initial condition , which is a different value from what we used in our previous studies @xcite . fig .
[ d - rpa - lhc ] shows the nuclear modification factor @xmath157 of @xmath4 meson as a function of @xmath2 at mid rapidity at the collision energy @xmath146 tev .
the result is in good agreement with the lhc data .
we find a slightly stronger suppression at low @xmath2 ( @xmath174 gev ) compared to the results of the electrons .
this is because the electrons at low @xmath2 come from the decays of the @xmath4 mesons with higher @xmath2 . to the quarkonium production we adopt color - evaporation - model ( cem ) with the same parameters in ref .
[ jpsi - rpa - lhc ] shows @xmath157 of @xmath162 at mid and forward rapidities at the collision energy @xmath146 tev . compared to our previous result of @xmath162 s @xmath157 at forward rapidity @xcite ,
the new estimate comes closer to the experimental data as expected .
this may imply that the use of @xmath79 with @xmath80 in the initial condition of the rcbk equation for heavy nuclei is consistent for nuclear dis and heavy quark production in minimum bias events in p@xmath1 collisions .
we note that the value of @xmath157 for @xmath162 is quite similar to the recent calculation with taking into account the dis constraint and the transverse profile of the nucleus by duclou et al .
@xcite . finally , by inspecting the whole results of the nuclear modification factors @xmath157 of @xmath162 , @xmath4 meson and the electron , our calculation shows a hierarchy of @xmath175 at lower @xmath2 .
meson production at @xmath144 tev .
notations are the same as in fig .
[ rpa - rhic ] .
lhc data is taken from ref .
, height=298 ] production at @xmath144 tev .
notations are the same as in fig .
[ rpa - rhic ] .
lhc data at mid and forward rapidities are taken from ref .
@xcite , title="fig:",height=298 ] production at @xmath144 tev .
notations are the same as in fig .
[ rpa - rhic ] .
lhc data at mid and forward rapidities are taken from ref .
@xcite , title="fig:",height=298 ]
in this paper , we have evaluated the lepton yields from charm - quark semileptonic decays in pp and p@xmath1 collisions at the rhic and lhc energies within the cgc framework .
our result of the @xmath2 spectra of the heavy - quark decay leptons with the @xmath44 factorized formula ( [ eq : cross - section - ln ] ) is harder than the experimental data , but the hybrid formula ( [ eq : cross - section - ln - coll ] ) gives rise to a consistent spectrum of the muons with the data in forward region in pp collisions .
for the heavy nucleus , we adopted the initial saturation scale determined by @xmath176 with the coefficient @xmath80 , which is taken from the nuclear dis analysis @xcite .
we normalized the cross - sections in pa collisions so as to give the nuclear modification factor @xmath149 at large transverse momenta .
the @xmath2 dependence of @xmath4 meson production at mid rapidity in p@xmath1 collisions appears to be consistent with the data up to @xmath177 gev , but that of the electron is slightly harder than the data at @xmath137 gev
. we have shown that @xmath157 has weak @xmath2 dependence in mid and forward rapidity regions at the rhic energy , which can be understood as multiple scattering effects in the heavy nucleus . at the lhc energy
, @xmath157 is below unity at low @xmath2 and recovering to unity at higher @xmath2 , due to the small-@xmath12 evolution of the gluon distribution in the target nucleus .
our results are in agreement with the data , albeit there are large uncertainties .
the factor @xmath157 of the muons is predicted to be more suppressed at forward rapidity at the lhc energy .
we have examined the fractional contributions of the gluons in the nucleus in the saturation and extended scaling regions to the lepton yields . at the rhic energy ,
the contributions from these regions are very limited . on the other hand , at the lhc energy , the lepton yields at @xmath171 gev at forward regions in pa collisions are dominated by the contributions from the extended - scaling region ( @xmath178 ) , and should reflect the saturation effects .
we have re - evaluated the nuclear modification factors @xmath157 of j@xmath179 and @xmath4 meson with the initial saturation scale @xmath176 with @xmath80 and obtained the results closer to the experimental data and consistent with @xcite . from these results ,
we have found that the nuclear modification factors of j@xmath179 , @xmath4 meson and the decay leptons show suppression at forward rapidity at the lhc energy , reflecting the small-@xmath12 evolution of the gluon saturation effect . in this work
, we only considered the minimum bias events in p@xmath1 collisions , but the study of the centrality dependence of the particle production will provide more information on the gluon saturation in the nucleus .
possible ways to treat the centrality dependence of collisions are the glauber model with smooth nuclear thickness function @xcite and the monte carlo implementation @xcite .
we leave this extension for future study
. the azimuthal angle correlation of the decay leptons may may provide further information of the saturation .
in fact , we found in ref .
@xcite that @xmath4-@xmath180 correlation at away side ( @xmath181 ) can probe the low @xmath44 gluon structure where the nonlinear effect is large .
at the same time , one expects that the correlation of the two leptons from semileptonic decays in p@xmath1 collisions will be more smeared .
in fact , back - to - back correlation of the pair at low @xmath2 can be sensitive not only to the saturation effect but also to the so - called sudakov factor @xcite already in the quark production process .
finally we note that the rcbk equation includes only a subset of nlo corrections while the hard matrix elements at lo are adopted here in this phenomenological analysis of pa collisions at the lhc energy .
apparently a consistent nlo formulation , with nlo bk equation @xcite and also the sudakov resummation , is desired for more systematic analysis of the data .
for single quark production , the sudakov effect will not be small . we leave it for future study .
kw wishes to thank komaba particle and nuclear theory group at the university of tokyo for kind hospitality during his visit when this work was finalized .
the authors would like to thank bo - wen xiao for useful comments .
the work of hf was partially supported by jsps kakenhi ( # 24540255 ) .
h. fujii and k. watanabe , nucl .
a * 920 * , 78 ( 2013 ) [ arxiv:1308.1258 [ hep - ph ] ] .
y. q. ma , r. venugopalan and h. f. zhang , phys .
d * 92 * , 071901 ( 2015 ) [ arxiv:1503.07772 [ hep - ph ] ] .
b. duclou , t. lappi and h. mntysaari , phys .
d * 91 * , no .
11 , 114005 ( 2015 ) [ arxiv:1503.02789 [ hep - ph ] ] .
d. kharzeev and k. tuchin , nucl . phys .
a * 735 * , 248 ( 2004 ) [ hep - ph/0310358 ] .
d. kharzeev and k. tuchin , nucl .
a * 770 * , 40 ( 2006 ) [ hep - ph/0510358 ] .
y. v. kovchegov and k. tuchin , phys .
d * 74 * , 054014 ( 2006 ) [ hep - ph/0603055 ] .
d. kharzeev , e. levin , m. nardi and k. tuchin , phys .
lett . * 102 * , 152301 ( 2009 ) [ arxiv:0808.2954 [ hep - ph ] ] , nucl .
a * 826 * , 230 ( 2009 ) [ arxiv:0809.2933 [ hep - ph ] ] .
k. j. golec - biernat and m. wusthoff , phys .
d * 59 * , 014017 ( 1998 ) [ hep - ph/9807513 ] .
b. abelev _ et al . _
[ alice collaboration ] , phys .
b * 719 * , 29 ( 2013 ) [ arxiv:1212.2001 [ nucl - ex ] ] .
b. abelev _ et al . _
[ alice collaboration ] , phys .
c * 90 * , no . 5 , 054901 ( 2014 ) [ arxiv:1406.2474 [ nucl - ex ] ] . g. aad _ et al .
_ [ atlas collaboration ] , phys .
lett . * 110 * , no .
18 , 182302 ( 2013 ) [ arxiv:1212.5198 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 725 * , 60 ( 2013 ) [ arxiv:1303.2084 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , phys .
c * 90 * , no . 4 , 044906 ( 2014 ) [ arxiv:1409.1792 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 718 * , 795 ( 2013 ) [ arxiv:1210.5482 [ nucl - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 724 * , 213 ( 2013 ) [ arxiv:1305.0609 [ nucl - ex ] ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
* 111 * , no .
21 , 212301 ( 2013 ) [ arxiv:1303.1794 [ nucl - ex ] ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
* 114 * , no .
19 , 192301 ( 2015 ) [ arxiv:1404.7461 [ nucl - ex ] ] .
f. gelis , e. iancu , j. jalilian - marian and r. venugopalan , ann .
nucl . part .
sci . * 60 * , 463 ( 2010 ) [ arxiv:1002.0333 [ hep - ph ] ] .
y. v. kovchegov and e. levin , `` quantum chromodynamics at high energy , '' ( 2012 ) , cambridge university press .
i. balitsky , nucl .
b * 463 * , 99 ( 1996 ) [ hep - ph/9509348 ] .
y. v. kovchegov , phys .
d * 54 * , 5463 ( 1996 ) [ hep - ph/9605446 ] .
i. balitsky , phys .
d * 75 * , 014001 ( 2007 ) [ hep - ph/0609105 ] .
j. l. albacete and y. v. kovchegov , phys .
d * 75 * , 125021 ( 2007 ) [ arxiv:0704.0612 [ hep - ph ] ] .
l. d. mclerran and r. venugopalan , phys .
d * 49 * , 2233 ( 1994 ) [ hep - ph/9309289 ] , phys .
d * 49 * , 3352 ( 1994 ) [ hep - ph/9311205 ] , phys .
d * 50 * , 2225 ( 1994 ) [ hep - ph/9402335 ] .
h. fujii , k. itakura , y. kitadono and y. nara , j. phys .
g * 38 * , 124125 ( 2011 ) [ arxiv:1107.1333 [ hep - ph ] ] . j. l. albacete , a. dumitru , h. fujii and y. nara , nucl .
a * 897 * , 1 ( 2013 ) [ arxiv:1209.2001 [ hep - ph ] ] .
j. l. albacete and c. marquet , prog . part .
* 76 * , 1 ( 2014 ) [ arxiv:1401.4866 [ hep - ph ] ] .
k. dusling , f. gelis , t. lappi and r. venugopalan , nucl .
phys . a * 836 * , 159 ( 2010 ) [ arxiv:0911.2720 [ hep - ph ] ] .
m. luszczak , r. maciula and a. szczurek , phys .
d * 79 * , 034009 ( 2009 ) [ arxiv:0807.5044 [ hep - ph ] ] .
m. gronau , c. h. llewellyn smith , t. f. walsh , t. c. yang and s. wolfram , nucl .
b * 123 * , 47 ( 1977 ) .
a. ali , z. phys .
c * 1 * , 25 ( 1979 ) .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) [ hep - ph/0201195 ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
c * 84 * , 044905 ( 2011 ) [ arxiv:1005.1627 [ nucl - ex ] ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
lett . * 109 * , no .
24 , 242301 ( 2012 ) [ arxiv:1208.1293 [ nucl - ex ] ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys . rev .
lett . *
112 * , no .
25 , 252301 ( 2014 ) [ arxiv:1310.1005 [ nucl - ex ] ] .
b. abelev _ et al . _
[ alice collaboration ] , phys .
d * 86 * , 112007 ( 2012 ) [ arxiv:1205.5423 [ hep - ex ] ] .
b. abelev _ et al . _
[ alice collaboration ] , phys .
b * 708 * , 265 ( 2012 ) [ arxiv:1201.3791 [ hep - ex ] ] . t. lappi and h. mntysaari , phys .
d * 91 * , no . 7 , 074016 ( 2015 ) [ arxiv:1502.02400 [ hep - ph ] ] .
e. iancu , j. d. madrigal , a. h. mueller , g. soyez and d. n. triantafyllopoulos , phys .
b * 744 * , 293 ( 2015 ) [ arxiv:1502.05642 [ hep - ph ] ] .
e. iancu , j. d. madrigal , a. h. mueller , g. soyez and d. n. triantafyllopoulos , arxiv:1507.03651 [ hep - ph ] . | we study single lepton production from semileptonic decays of heavy flavor hadrons ( @xmath0 ) in pp and p@xmath1 collisions at rhic and the lhc within the saturation / color - glass - condensate ( cgc ) framework . using the gluon distribution function obtained with the dipole amplitude ,
whose energy dependence is described by the balitsky - kovchegov equation with running coupling effect , we compute the transverse - momentum ( @xmath2 ) spectra of the lepton yields at mid and forward rapidities .
we find that a large fraction of leptons at low @xmath2 stems from the saturation regime of the incoming gluons in the target , especially in p@xmath1 collisions at the lhc .
the resultant @xmath2 spectra is slightly harder than the data , but the nuclear modification factor seems consistent with the data within some uncertainty .
we also update the nuclear modification factors for j/@xmath3 and @xmath4 meson at the lhc energy .
@xmath5institute of physics , university of tokyo , komaba , tokyo 153 - 8902 , japan @xmath6key laboratory of quark and lepton physics ( moe ) and institute of particle physics , central china normal university , wuhan 430079 , china |
investigation of the reaction fronts formed in the reaction - diffusion processes of the type a @xmath0 b @xmath1 0 with initially separated species a and b has attracted a lot of recent interest .
this is due to the fact that not only the kinetics of such systems exhibits many surprising features @xcite , but they are also amenable to experimental studies @xcite . a standard way to treat the initially separated problem analytically is to solve the partial differential equations @xcite @xmath8 { \frac{\textstyle\partial{{\rho_{\mathrm{b}}}}}{\textstyle\partial t } } & = & { d_{\mathrm{b } } } { \frac{\textstyle\partial^2{{\rho_{\mathrm{b}}}}}{\textstyle\partial x^2 } } - r \ , , \end{array}\ ] ] with the initial state given by @xmath9 { \rho_{\mathrm{b}}}(x , t=0 ) & = & b_0 h(x ) \ , , \end{array}\ ] ] where @xmath10 and @xmath11 are the mean local concentrations of a s and b s , @xmath5 is the macroscopic reaction rate , @xmath12 denotes the heavyside step function , and @xmath3 , @xmath4 , @xmath2 and @xmath13 are some constants related to the initial concentrations of species a and b and their diffusion coefficients , respectively . equations ( [ gr ] ) present the macroscopic approach to the problem .
however , it has not been established yet how to relate the macroscopic reaction rate @xmath5 to the microscopic picture of the initially separated reaction - diffusion systems below or at the critical dimension @xmath14 .
dimensional @xcite and renormalisation group analyses @xcite lead to the conclusion that above @xmath15 one can adopt the mean - field approximation @xmath16 , with @xmath17 being a constant . for 2d systems
one expects logarithmic corrections to the mean - field picture @xcite .
one dimensional systems are usually studied by examining microscopic models in which , upon contact , the members of a pair a b react with some probability @xmath18 @xcite .
the analytical form of @xmath19 was derived for @xmath20 @xcite , and for @xmath21 and @xmath22 @xcite .
there are , however , several techniques which enables one to derive a lot of useful information from ( [ gr ] ) even for @xmath23 , i.e. when the explicit form of @xmath5 remains unknown .
they are concentrated on the asymptotic , long - time behaviour of the reaction - diffusion systems , and include the renormalisation group analysis @xcite , the scaling ansatz @xcite , the quasistationary approximation @xcite , and our approach of ref . @xcite . according to the scaling ansatz @xcite , the long - time behaviour of the reaction - diffusion system inside the reaction layer can be described with a help of some scaling functions @xmath24 , @xmath25 and @xmath26 through @xmath27 \label{sb } { \rho_{\mathrm{b}}}(x , t ) & = & { \eta_{\mathrm{b}}}t^{-{\gamma_{\mathrm{b}}}}{s_{\mathrm{b}}}{\!\left ( x - { x_{\mathrm{f}}}(t ) \over w(t ) \right)}\,,\\[1ex ] \label{sr } r(x , t ) & = & { \eta_{\mathrm{r}}}t^{-\beta } { s_{\mathrm{r}}}{\!\left ( x - { x_{\mathrm{f}}}(t ) \over w(t ) \right)}\,,\end{aligned}\ ] ] where @xmath28 denotes the point at which the reaction rate @xmath5 attains its maximal value , @xmath29 is the width of the reaction zone , @xmath30 , @xmath31 and @xmath32 are some parameters independent of @xmath33 and @xmath34 , and exponents @xmath35 , @xmath36 , @xmath37 and @xmath38 are some positive constants given , for @xmath39 and nonzero diffusion constants , by @xmath40 , @xmath41 and @xmath42 . the quasistationary approximation @xcite consists in the assumption that at sufficiently long times the kinetics of the front is governed by two characteristic time scales .
one of them , @xmath43 , controls the rate of change in the diffusive current @xmath44 of particles arriving at the reaction layer .
the other one , @xmath45 , is the equilibration time of the reaction front . if @xmath46 then @xmath47 as @xmath48 .
therefore , as @xmath49 , in the vicinity of @xmath50 the left sides of ( [ gr ] ) become negligibly small compared to other terms .
consequently , if @xmath2 and @xmath13 are both nonzero , the asymptotic form of @xmath51 and @xmath52 inside the reaction layer is governed by much simpler equations @xmath53 { d_{\mathrm{b } } } { \frac{\textstyle\partial^2{{\rho_{\mathrm{b}}}}}{\textstyle\partial x^2 } } & = & r \ , , \end{array}\ ] ] which are to be solved with the boundary conditions @xmath54 \partial{\rho_{\mathrm{b}}}/\partial x \to { j_{\mathrm{b}}}(t),\;{\rho_{\mathrm{a}}}\to 0 & \;\mbox{as}\ ; & x \to + \infty\ , . \end{array}\ ] ] the most important feature of the quasistationary equations ( [ stac ] ) is that they depend only on @xmath33 , with time @xmath34 being a parameter entering their solutions @xmath10 and @xmath11 only through the time dependent boundary currents @xmath55 whose dependence on @xmath34 , @xmath2 , @xmath13 , @xmath3 and @xmath4 has recently been derived analytically @xcite . in a recent paper
@xcite we employed the quasistatic approximation to develop a new method of investigating the asymptotic kinetics of the initially separated reaction - diffusion systems .
a peculiar feature of that approach is that it is concentrated mainly on the properties of the system outside the reaction zone . in this way , without imposing any special restrictions on the form of @xmath5 , it relates _ exactly _ many quantities of physical interest to the values of external parameters @xmath3 , @xmath4 , @xmath2 and @xmath13 , which will enable us to investigate the limit @xmath56 analytically . in particular it was shown that there exist two limits , @xmath57 and @xmath58 .
given @xmath3 , @xmath4 , @xmath2 and @xmath13 , the value of @xmath59 can by computed by solving @xmath60 where @xmath61}\exp(x^{2})\,,\ ] ] and @xmath62 is the error function @xcite
. then @xmath63 can be calculated by solving @xmath64 } = 2\sqrt{{d_{\mathrm{b}}}}{\:\!\mbox{erf$^{-1}$}\left[({c_{\mathrm{b}}}-b_0)/{c_{\mathrm{b}}}\right]}\ ] ] and @xmath65 where @xmath66 and @xmath67 are some constants controlling the form of @xmath51 and @xmath52 outside the reaction zone ; specifically , for @xmath68 we have @xmath69}\,,\ ] ] and for @xmath70 there is @xmath71}\,.\ ] ] it was confirmed by several methods , including the renormalisation group analysis @xcite , numerical simulations @xcite and heuristic arguments @xcite , that the values of the exponents @xmath35 , @xmath36 , @xmath37 and @xmath38 as well as the form of the scaling functions @xmath24 , @xmath25 and @xmath26 do not depend on @xmath2 , @xmath13 , @xmath3 and @xmath4 if the values of these parameters are nonzero .
however , when one of the components is immobile ( or ` static ' ) , the asymptotic kinetics of the reaction front can change dramatically .
for example , in the mean - field approximation the width of the front converges to a stationary value , and the reaction rate at @xmath50 decreases as @xmath72 , which corresponds to @xmath73 and @xmath74 @xcite .
we have therefore two asymptotic universality classes : one characteristic for the ` dynamic ' systems in which both components diffuse , and the ` quasistatic ' one observed if one of the diffusion constants is zero .
henceforth we shall assume @xmath75 , so the two asymptotic universality classes will be distinguished by determining whether @xmath76 or @xmath6 .
although the peculiar kinetics of the systems with @xmath20 was noticed quite early , so far nearly all of the research has been concentrated on the systems in which both of the diffusion constants @xmath2 and @xmath13 are nonzero .
closer inspection of the powerful methods developed in the last decade to investigate the reaction - diffusion systems reveals that only the scaling ansatz , which is a purely mathematical concept , can be trivially employed to investigate both kinds of the systems .
however , since even this relatively simple approach has so far been carried out only for the ` dynamic ' problem @xcite , the theories of the two asymptotic universality classes remain practically disconnected .
several factors have lead to this situation .
theoretical results @xcite derived for the case @xmath20 utilized the basic property of such systems , immobility of particles b , in a way that can not be extended for systems with @xmath21 . on the other hand ,
most of the fundamental techniques developed for the case @xmath6 has been based on the quasistationary approximation which requires that the ratio @xmath77 of two opposite currents of particles a and b entering the reaction zone should asymptotically go to 1 . this condition can not be met by the systems with @xmath76 , as in this case @xmath78 .
the aim of our paper is to unify our understanding of these two kinds of the reaction - diffusion systems .
the procedure we are using consists in detailed examination of the case @xmath76 and comparison of the results with those already derived for @xmath6 .
we study the systems with @xmath76 by means of various methods and for arbitrary ( positive ) values of @xmath3 , @xmath4 , and @xmath2 .
in particular we show the counterpart of the quasistationary approximation ( [ stac ] ) which should be used if @xmath20 .
we argue that the form of these equations , as well as their boundary conditions , determine the properties of the asymptotic universality classes . in subsequent sections we study these properties for the systems with @xmath76 using the heuristic theory of ref .
@xcite , the scaling ansatz , the mean - field approximation , and numerical analysis .
although some aspects of the systems with @xmath76 turn out to be the same as of those with @xmath6 , our analysis implies that these two cases should always be considered separately .
consider equations ( [ cfeq ] ) ( [ rb ] ) with the values of @xmath2 , @xmath3 and @xmath4 fixed at some positive values , and @xmath13 going to 0 .
for physical reasons we expect that as @xmath13 goes to 0 , @xmath59 converges to some nonzero limiting value , and so the argument of @xmath79 on the right hand side of ( [ cfeq ] ) diverges to infinity .
we can therefore use an asymptotic property of the error function , @xmath80 as @xmath81 @xcite , and reduce ( [ cfeq ] ) to @xmath82 since @xmath83 diminishes monotonically from @xmath84 to @xmath85 as @xmath33 grows from @xmath86 to @xmath84 , the above equation has a unique , positive solution .
consequently , ( [ cf ] ) and ( [ cj ] ) imply that @xmath63 and @xmath66 also converge to some positive values , but @xmath67 rapidly diverges to infinity
. one can now use ( [ cf ] ) , ( [ cj ] ) and ( [ cfeq0 ] ) to arrive , after some algebra , at an important relation valid only if @xmath20 @xmath87 this equation has a very natural physical interpretation .
on the one hand the total number @xmath88 of reactions occurring by time @xmath34 is asymptotically equal @xmath89 on the other hand , however , for @xmath70 we have @xmath90 , and for @xmath68 we expect @xmath91 . neglecting terms of order @xmath92 we thus conclude that @xmath88 can as well be estimated by @xmath93 { \d } x \,\sim\ , b_0 { x_{\mathrm{f}}}(t ) \,\sim\ , b_0 { c_{\mathrm{f}}}\sqrt{t } \,,\ ] ] which leads to ( [ relcfcj ] ) .
our theory is consistent with the numerical simulations of larralde _ et al _ @xcite , who considered the one dimensional system ( i.e. for @xmath94 ) with @xmath76 , @xmath95 and @xmath96 .
they found that , respectively , for @xmath97 , 1000 and 5000 the value of @xmath98 was approximately equal to @xmath99 , @xmath100 and @xmath101 , so that @xmath102 @xmath103 , @xmath104 and @xmath105 , respectively .
this is in excellent agreement with our equation ( [ cfeq0 ] ) , whose numerical solution reads @xmath106 .
below , in section [ nums ] , we will also verify our theory using the mean - field approximation ( i.e. for @xmath107 ) .
let @xmath76 and @xmath2 , @xmath3 and @xmath4 take on arbitrary ( positive ) values .
assume that the asymptotic solutions of the glfi and rcz problem ( [ gr ] ) in the long - time limit take on the scaling form ( [ sa ] ) ( [ sr ] ) with @xmath46 . inserting ( [ sa ] ) and ( [ sb ] ) into ( [ gr ] ) and taking the limit @xmath49 we find that at any @xmath33 such that @xmath108 there is @xmath109 hence , because @xmath46 , in the limit @xmath49 the term @xmath110 becomes negligibly small compared to @xmath111 .
this implies that for @xmath112 the form of the scaling functions can be determined by solving @xmath113 - { \frac{\textstyle\partial{{\rho_{\mathrm{b}}}}}{\textstyle\partial t } } & = & r \ , .
\end{array}\ ] ] the appropriate boundary conditions for these equations read @xmath114 { \rho_{\mathrm{a}}}\to 0,\ ; { \rho_{\mathrm{b}}}\to b_0 & \;\mbox{as}\ ; & x \to + \infty\ , , \end{array}\ ] ] where @xmath115 @xcite .
equations ( [ qstac ] ) are the general counterparts of the quasistatic approximation ( [ stac ] ) if @xmath75 and @xmath76 . the boundary conditions ( [ bcro ] ) determine the form of the boundary conditions for @xmath24 and @xmath25 except for a constant multiplier .
we will take advantage of the fact that we are at liberty to multiply @xmath24 and @xmath25 by arbitrary constants ( which can be compensated for by appropriate changes in @xmath30 and @xmath31 ) and assume that the boundary conditions for @xmath24 and @xmath25 read @xmath116 { s_{\mathrm{a}}}(z)\to 0,\ ; { s_{\mathrm{b}}}(z ) \to 1 & \;\mbox{as}\ ; & z
\to + \infty\ , , \end{array}\ ] ] where the prime denotes the derivative with respect to @xmath117 . equations ( [ bcs ] ) immediately imply that @xmath118 \label{eb } { \eta_{\mathrm{b}}}&= & b_0\,.\end{aligned}\ ] ] the diffusive current @xmath119 of particles a for @xmath120 is asymptotically expected to be equal to @xmath121 @xcite . on the other hand , however , we can calculate it by inserting ( [ sa ] ) into @xmath122 , which leads to @xmath123 \label{ea } { \eta_{\mathrm{a}}}&= & { c_{\mathrm{w}}}{c_{\mathrm{j}}}/ { d_{\mathrm{a}}}\,,\end{aligned}\ ] ] where we denoted @xmath124 . upon inserting the scaling ansatz into the first of equations ( [ qstac ] ) we come to @xmath125 . combining it with ( [ rex2 ] ) we arrive at
@xmath126 we thus see that the scaling ansatz imposes on the values of the scaling exponents three relations ( [ rex1 ] ) , ( [ rex2 ] ) and ( [ rex3 ] ) .
only the first of them takes on a different form if @xmath6 , as in that case , by symmetry , @xmath127 @xcite .
equations ( [ qstac ] ) imply that inside the reaction zone @xmath128 inserting into it the scaling ansatz ( [ sa ] ) and ( [ sb ] ) , and carrying out our ` standard ' limiting procedure ( @xmath49 at any @xmath33 such that @xmath129 ) we conclude , after some algebra involving ( [ relcfcj ] ) , that @xmath24 and @xmath25 are related by a simple formula @xmath130 . upon integrating this equation and using the boundary conditions ( [ bcs ] ) to determine the integration
constant we finally come to @xmath131 note that for @xmath6 the scaling functions @xmath24 and @xmath25 satisfy an entirely different relation @xmath132 , which comes from the asymptotic symmetry of the system .
note also that relations ( [ qstac2 ] ) and ( [ sasb ] ) are independent of @xmath5 and , consequently , of @xmath133 . as a quick application of ( [ sasb ] )
consider a one dimensional system with @xmath76 .
it is known @xcite that in this case @xmath134 .
so ( [ sasb ] ) and ( [ bcs ] ) lead to @xmath135 \,.\ ] ]
consider a system governed by ( [ gr ] ) with @xmath136 , @xmath20 and @xmath137 .
this particular form of @xmath5 immediately implies @xmath138 . using now ( [ rex1 ] ) , ( [ rex2 ] ) and ( [ rex3 ] ) we conclude that @xmath139 these values are consistent with numerical simulations @xcite and heuristic arguments @xcite .
it is easy to see that if length and time are measured in units of @xmath140 and @xmath141 , respectively , then the solutions of ( [ qstac ] ) reduce to those obtained for the particular case @xmath142 .
thus , in investigating the mean - field reaction - diffusion systems , it is sufficient to examine in detail only a system with some convenient values of the material parameters @xmath2 , @xmath3 , @xmath4 and @xmath17 .
the solutions for arbitrary values of these parameters can be then easily found by an appropriate choice of the units .
this property guarantees that any asymptotic length @xmath143 satisfies @xmath144 . in particular , the asymptotic width of the reaction zone is given by @xmath145 where @xmath146 denotes the asymptotic width of the reaction zone in the system with @xmath142 .
numerical estimation of this parameter yields @xmath147 ( see the next section for more details ) . essentially the same line of reasoning was used in ref .
@xcite to show that in the mean - field system with @xmath6 the asymptotic width of the reaction zone is given by @xmath148 .
however , in that paper it was incorrectly assumed that @xmath149 .
we estimated the correct value of @xmath150 numerically , obtaining @xmath151 .
inserting now the scaling ansatz into ( [ qstac ] ) and using @xmath152 we conclude that @xmath24 and @xmath25 satisfy @xmath153 \label{esb } { s_{\mathrm{b } } } ' & = & { \tilde{w}}^2 { s_{\mathrm{a}}}{s_{\mathrm{b}}}\,,\end{aligned}\ ] ] upon inserting ( [ sasb ] ) into ( [ esa ] ) we arrive at the nonlinear differential equation for the mean - field scaling function @xmath24 @xmath154 we can use it to estimate the behaviour of @xmath155 and @xmath156 for @xmath157 . in this region
we expect @xmath158 and @xmath159 , so ( [ saeq ] ) reduces to @xmath160 , which implies @xmath161 we can also investigate the tail of particles b which forms for @xmath162 .
in this region we can assume @xmath163 , so that ( [ esb ] ) reduces to @xmath164 , which leads to @xmath165 thus , if @xmath20 , the mean - field form of the scaling function @xmath156 is asymmetric , whereas for @xmath21 this function is always symmetric , which is most easily seen in the symmetric case @xmath166 and @xmath167 .
we will now investigate some properties of the limit @xmath56 which could not be analysed within the framework of the general theory presented in section [ limit ] . as we already showed in our previous paper @xcite , if @xmath6 then the mean - field density of particles b at @xmath50 is asymptotically proportional to @xmath168 .
as this quantity has to be less than @xmath4 , we conclude that @xmath169 .
therefore the time @xmath170 at which the mean - field system enters the long - time regime must satisfy @xmath171 only for times satisfying this relation can one use the quasistatic approximation ( [ stac ] ) .
however , as @xmath56 , the right hand side of ( [ time ] ) diverges to infinity .
consequently , as @xmath13 goes to 0 , @xmath170 diverges to infinity and in the limiting case @xmath20 the kinetics of the system can never be described with the quasistatic approximation equations ( [ stac ] ) .
although we derived these conclusions only within the mean - field approximation , it is reasonable to expect that @xmath172 as @xmath56 in any initially separated reaction - diffusion system . to summarize the differences between the two asymptotic universality classes , in table [ tab1 ] we list their main properties in the mean - field approximation .
the data for the case @xmath21 come from references @xcite , @xcite and @xcite .
.comparison of the two asymptotic universality classes in the mean - field approximation . [ cols="^,^,^",options="header " , ]
to check the theory presented in the previous sections for the case @xmath20 we solved numerically , using the finite - difference ftcs ( forward time centred space ) method , partial differential equations ( [ gr ] ) with the mean - field reaction rate @xmath16 .
we present the data obtained for @xmath173 , @xmath174 , @xmath175 , and @xmath176 .
other values of these parameters yielded similar results .
first of all we verified the theory presented in section [ limit ] . in fig . 1 we show the plot of @xmath51 and @xmath52 in the vicinity of @xmath50 at @xmath177 .
the dotted line was computed from ( [ ra ] ) .
it perfectly matches the numerical solutions up to @xmath178 , i.e. outside the reaction layer .
actually , in the region @xmath179 , the relative error is less than @xmath180 . also , the value of @xmath59 computed from ( [ cfeq0 ] ) is 0.2263 , whereas its numerical estimation @xmath181 for @xmath182 is 0.2253 .
next we investigated the scaling properties of the considered system .
as some aspects of this problem were already considered @xcite , we present briefly only those results which are relevant to our paper . in fig .
2 we show the log - log plot of @xmath183 .
we can see that it initially grows as @xmath184 , which is a typical short time limit behaviour @xcite , but beginning from @xmath185 it quickly converges to a constant value .
this enabled us to estimate @xmath186 . finally , in fig .
3 we present the scaling plot of @xmath187 as a function of @xmath188 , where we used ( [ w ] ) with @xmath189 .
the plots for @xmath190 and @xmath191 are practically indistinguishable .
note also two facts .
first , for @xmath192 the semilog plot of @xmath193 is linear in @xmath117 , in accordance with ( [ taila ] ) .
second , @xmath156 is discontinuous at @xmath194 , which reflects the fact that @xmath11 is discontinuous at @xmath195 . 3.2 in 3.6 in
we have investigated the long - time behaviour of the concentrations @xmath51 and @xmath52 of species a and b in the initially separated diffusion limited systems with @xmath20 .
all of our analysis was carried out for arbitrary values of @xmath2 , @xmath3 and @xmath4 .
first we derived the formulae ( [ cfeq0 ] ) and ( [ relcfcj ] ) which together with ( [ ra ] ) describe the behaviour of @xmath10 outside the reaction zone .
an interesting feature of these equations is that we expect them to be valid for any reaction - diffusion system of the type a @xmath0 b(static ) @xmath1 0 that exhibits the scaling behaviour with @xmath196 .
thus we conclude that many important properties of the reaction - diffusion systems with @xmath76 depend only on the values of @xmath3 , @xmath4 and @xmath2 , but not on the explicit form of @xmath5 .
this includes the location of the reaction zone centre ( controlled by @xmath59 ) , the total reaction rate ( controlled by @xmath63 ) , and the concentration profile of particles a outside the reaction zone ( controlled by @xmath66 and @xmath59 ) .
a similar situation is observed in the systems with @xmath197 @xcite .
next we investigated general consequences of the scaling ansatz .
we concluded that it determines the value of @xmath198 and imposes relations ( [ rex2 ] ) and ( [ rex3 ] ) on the values of @xmath35 , @xmath36 and @xmath37 .
these relations are also valid for @xmath6 .
we proved that the scaling functions @xmath24 and @xmath25 are related by a simple formula ( [ sasb ] ) .
we also concluded that for @xmath20 the asymptotic forms of @xmath51 and @xmath52 inside the reaction front can be derived from equations ( [ qstac ] ) which are to be solved with time depending boundary conditions ( [ bcro ] ) .
thus these equations determine the asymptotic properties of the reaction front .
we also examined in detail the properties of the reaction zone in the mean - field approximation .
in particular we determined the functional forms of @xmath24 , @xmath25 and @xmath26 far from @xmath50 and the dependence of the width of the reaction zone on @xmath17 , @xmath3 , @xmath4 and @xmath2 .
our analysis showed that the main differences in the behaviour of the initially separated reaction - diffusion systems with @xmath20 and @xmath6 arise from the fact that the term @xmath199 can be neglected in ( [ gr ] ) only if the corresponding diffusion constant @xmath13 is nonzero .
therefore , depending on whether @xmath13 is zero or not , the long - time behaviour of the reaction front is governed by entirely different partial differential equations .
if @xmath21 , then we use the usual quasistationary equations ( [ stac ] ) , otherwise we must employ ( [ qstac ] ) .
the different forms of these equations and their boundary conditions imply the different forms of their solutions and , consequently , different asymptotic properties of the two universality classes . therefore the cases @xmath76 and @xmath6 should always be considered separately .
there is , however , evidence that in some special cases the two asymptotic universality classes may be very much alike .
such surprising conclusion follows from the extensive numerical simulations of the one dimensional system carried recently out by cornell @xcite .
he considered a system with @xmath200 and concluded that @xmath201 , @xmath202 , and that asymptotically @xmath5 is a gaussian centred at @xmath50 with its width growing as @xmath203 .
the same results were derived analytically for the one dimensional system with @xmath76 by larralde _
et al _ @xcite .
note also that this form of @xmath5 , or more generally any @xmath5 that depends on @xmath33 and @xmath34 explicitly rather than through @xmath10 and @xmath11 , uniquely determines the form of @xmath10 .
this comes from the fact that whether or not @xmath20 , asymptotical forms of @xmath51 and @xmath5 are related by the same differential equation @xmath204 with the same boundary conditions .
therefore it is quite possible that the only difference between the one dimensional systems with @xmath20 and @xmath197 lies in the form of @xmath25 and the value of @xmath38
. it should be stressed , however , that the asymptotic kinetics of one dimensional systems with @xmath6 has recently become the subject of controversial discussions @xcite , and further exploration of this topic is still required before the final conclusions can be made . | we examine the long - time behaviour of a @xmath0 b(static ) @xmath1 0 reaction - diffusion systems with initially separated species a and b. all of our analysis is carried out for arbitrary ( positive ) values of the diffusion constant @xmath2 of particles a and initial concentrations @xmath3 and @xmath4 of a s and b s .
we derive general formulae for the location of the reaction zone centre , the total reaction rate , and the concentration profile of species a outside the reaction zone .
the general properties of the reaction zone are studied with a help of the scaling ansatz . using the mean - field approximation we find the functional forms of ` tails ' of the reaction rate @xmath5 and the dependence of the width of the reaction zone on the external parameters of the system .
we also study the change in the kinetics of the system with @xmath6 in the limit @xmath7 .
our results are supported by numerical solutions of the mean - field reaction - diffusion equation .
reaction kinetics ; diffusion ; segregation .
05.40.@xmath0j , 82.20.-w |
frequency division multiplexing ( ofdm ) systems with guard - time interval or cyclic prefix can prevent inter - symbol interference ( isi ) .
ofdm has overlapping spectra and rectangular impulse responses .
consequently , each ofdm sub - channel exhibits a sinc - shape frequency response .
therefore , the time variations of the channel during one ofdm symbol duration destroy the orthogonality of different subcarriers , and result in power leakage among subcarriers , known as inter - carrier interference ( ici ) , which causes degradation in system performance . in order to overcome the above drawbacks of ofdm system , several pulse - shaping ofdm systems were proposed @xcite .
it is shown that signal transmission through a rectangular lattice is suboptimal for doubly dispersive ( dd ) channel @xcite . by using results from sphere covering theory , the authors have demonstrated that hexagonal multi - carrier transmission ( hmct ) system obtains lower energy perturbation than ofdm system , hence providing better performance against isi / ici @xcite .
there are mainly two types of hmct systems , ( 1 ) : hmct system with orthogonal prototype pulse@xcite , named as lattice ofdm ( lofdm ) system , which confines the transmission pulses to a set of orthogonal ones .
( 2 ) : hmct system with nonorthogonal prototype pulse@xcite , named as hexagonal multicarrier transmission ( hmt ) system , which abandons the orthogonality condition of the modulated pulses and obtains the high robustness performance on combating the dispersion of dd propagation channel in both time and frequency domain . to optimally combat the impact of the dd propagation channels , the lattice parameters in the time - frequency ( t - f ) plane and the pulse shape of modulation waveform of hmct system are jointly optimized to adapt to the channel scattering function .
it is shown in @xcite that the hmct system outperforms ofdm system from the robustness against channel dispersion point of view .
the basic mathematical operation of the received signal performed by the demodulator is a projection onto an identically structured function set generated by the prototype pulse function @xcite , i.e. an optimal match filter . in hmct system , there is no cyclic prefix and data symbols of hmct signal are transmitted on hexagonal lattice points in the t - f plane .
traditional timing and frequency synchronization schemes can not be applied to hmct system directly . in our previous work @xcite ,
signal to interference and noise ratio ( sinr ) of hmct system over dd channel is analyzed and the max - sinr based timing synchronization scheme is proposed . in this paper ,
a novel preamble structure and two carrier frequency offset estimation algorithms , named as phase differential algorithm and least squares algorithm , for hmct system are proposed .
after detailed derivation , the close form performance lower bound of the proposed carrier frequency offset estimation algorithm is given .
theoretical analyses and simulation results show that the proposed carrier frequency estimation approach for hmct system obtains an approximation to the lower bound mean square error ( mse ) performance over dd channel .
in hmct systems , the transmitted baseband signal can be expressed as @xcite @xmath0 where @xmath1 and @xmath2 are the lattice parameters , which can be viewed as the symbol period and the subcarrier separation , respectively ; @xmath3 denotes the transmitted data symbol , which is assumed to be taken from a specific signal constellation and independent and identically distributed ( i.i.d . ) with zero mean and average power @xmath4 ; @xmath5 and @xmath6 are the position indices in the t - f plane ; @xmath7 and @xmath8 denote the sets from which @xmath9 can be taken , with cardinalities @xmath10 and @xmath11 , respectively . @xmath12 and @xmath13 are the transmitted pulses generated by the prototype pulse @xmath14 . the ambiguity function of the prototype pulse is defined as @xmath15
the baseband dd channel can be modeled as a random linear operator @xmath16 @xcite @xmath17=\int_{0}^{\tau_{\textrm{max}}}\int^{f_{d}}_{-f_{d}}h(\tau,\upsilon)x(t-\tau)e^{j2\pi\upsilon t}d\tau d\upsilon\ ] ] where @xmath18 and @xmath19 are the maximum multipath delay spread and the maximum doppler frequency , respectively@xcite .
@xmath20 is called the delay - doppler spread function , which is the fourier transform of the time - varying impulse response of the channel @xmath21 with respect to @xmath22@xcite . in wide - sense stationary uncorrelated scattering ( wssus )
assumption , the dd channel is characterized by the second - order statistics @xmath23=s_{h}(\tau,\upsilon)\delta(\tau-\tau_{1})\delta(\upsilon-\upsilon_{1})\ ] ] where @xmath24 $ ] denotes the expectation and @xmath25 is called the scattering function , which characterizes the statistics of the wssus channel . without loss of generality ,
@xmath20 is assumed to have zero mean and unit variance , i.e.,@xmath26 .
the received baseband signal can be expressed as @xmath27+w(t)\ ] ] where @xmath28 is the awgn with variance @xmath29 .
the basic mathematical operation of the received signal performed by the demodulator is a projection onto an identically structured function set generated by the prototype pulse function , i.e. an optimal match filter @xcite . to recover the transmitted data symbol @xmath30 , the match filter receiver projects the received signal @xmath31 on prototype pulse function @xmath32 , i.e. , @xmath33 , where @xmath34 denotes the inner product and @xmath35 denotes the complex conjugate .
hence , frequency synchronization plays a critical role in ensuring reliable demodulation .
firstly , we will analyze the effects of carrier frequency offset on the recovered data symbol @xmath36 .
under the assumption that there is a carrier frequency offset @xmath37 between the received signal @xmath31 and the transmitted signal @xmath38 .
after ignoring the impact of additive noise , the received signal @xmath31 can be written as @xmath39 and the recovered data symbol @xmath30 can be expressed as @xmath40 where @xmath41 in ( 7 ) can be expressed as ( 8) at the top of the next page . @xmath42 ' '' '' let @xmath43 and @xmath44 , @xmath45 can be expressed @xmath46 and @xmath47 in ( 7 ) can be expressed as ( 10 ) at the top of the next page .
@xmath48 ' '' '' let @xmath43 and @xmath44 , @xmath49 can be expressed as @xmath50 hence , the data symbol @xmath36 in ( 7 ) can be rewritten as @xmath51 the first term in equation ( 12 ) represents the desired symbol and the last three terms denote isi / ici .
concerning the useful portion , the transmitted symbols @xmath52 are attenuated by @xmath53 which is caused by the carrier frequency offset @xmath37 and doubly dispersive channel .
meanwhile , the transmitted symbols rotated by a phasor @xmath54 .
the proposed preamble is composed of two training sequences @xmath55 and @xmath56 in the frequency domain , as depicted in fig . 1 . @xmath57 $ ] , @xmath58 and @xmath59 denotes the length of training sequence .
we assume that the received signal @xmath65 experiences a carrier frequency offset @xmath37 and the timing offset is completely compensated .
the demodulated training sequences can be expressed as @xmath66 and @xmath67 @xmath68 and @xmath69 denote the awgn in frequency domain .
thanks to the central limit theorem@xcite , the last four terms in ( 14 ) and ( 15 ) can be denoted as @xmath70 and @xmath71 , respectively .
we can rewrite equation ( 14 ) and ( 15 ) as @xmath72 and @xmath73 we assume that @xmath74 and define @xmath75 as ( 18 ) at the top of next page .
@xmath76 denotes the phase rotation caused by frequency offset and @xmath77 is a constant .
let @xmath78 denotes the normalized frequency offset , and @xmath79 denotes the sample period .
@xmath80 is the additive noise with zero mean and average power @xmath81 .
there are two fine frequency offset estimation schemes : phase differential estimation scheme and least squares estimation scheme .
@xmath82 ' '' '' the phase differential carrier frequency offset estimation algorithm can be expressed as @xmath83 the phase differential estimation scheme is valid for @xmath84=[-\rho/2,\rho/2]$ ] , and @xmath85 denotes system spectral efficiency .
we can model the phase of @xmath75 as @xmath86 and @xmath87 where @xmath88 $ ] and @xmath89^t$ ] , @xmath90 $ ] and @xmath91 $ ] .
@xmath92 in ( 21 ) can be expressed as @xmath93[c]{,c / c , } 1&0\\ 1&1\\ \vdots&\vdots\\ 1&n/2 - 1 \end{ieeeeqnarraybox*}\right]\ ] ] the least squares carrier frequency offset estimation algorithm can be expressed as @xmath94 and @xmath95 in ( 23 ) can be obtained by @xmath96[c]{,c / c , } \overline{m^2}&-\overline{m}\\ -\overline{m}&1 \end{ieeeeqnarraybox*}\right]\ ] ] where @xmath97 , @xmath98 and @xmath99 .
hence , the estimation of @xmath100 can be rewritten as @xmath101^{t}=\frac{1}{\delta m^2}\left[\begin{ieeeeqnarraybox*}[][c]{,c / c , } \overline{\gamma}\overline{m^2}&\overline{m}(\overline{\gamma m})\\ \overline{\gamma m}&\overline{m}\cdot\overline{\gamma } \end{ieeeeqnarraybox*}\right ] \end{split}\ ] ] where @xmath102 and @xmath103 .
the phase rotation @xmath104 caused by frequency offset and the constant @xmath77 can be obtained by @xmath105 and @xmath106 the normalized frequency offset @xmath107 can be expressed as @xmath108 the least squares frequency offset estimation scheme is valid for @xmath84=[-\rho/2,\rho/2]$ ] .
let @xmath109 $ ] denotes the prototype pulse with length @xmath110 in discrete hmct system .
in order to recover the transmitted frequency domain training sequence @xmath111 , we need @xmath112 inner product operations with @xmath110 points , which introduce high computational complexity .
let @xmath113 $ ] denote the received signal vector with length @xmath114 , and @xmath115 , @xmath116 , denotes the objective projection subspace .
@xmath117 $ ] and satisfies @xmath118 the time frequency subspace projection of the received signal @xmath119 on the objective projection subspace @xmath120 is equivalent to @xmath121 , @xmath122 .
@xmath123 and @xmath124 denotes the ceiling function . @xmath125 and @xmath126 $ ] .
@xmath127 denotes the hadamard product and @xmath128 denotes the fast fourier transform . the output symbol @xmath129 of the time frequency subspace projector after projecting the received signal @xmath119 to the @xmath130th subspace @xmath131 can be expressed as @xmath132 after performing the ifft on the signal @xmath129 , the transformed signal can be expressed as @xmath133 where @xmath124 denotes the ceiling function .
@xmath134 , @xmath135 , which can be denoted as @xmath125 .
we can conclude from @xmath136 that the transmitted frequency domain training sequence @xmath111 can be obtained by the following two steps .
the first step is to calculate the hadamard product of the received signal @xmath119 and the prototype pulse @xmath137 , that is @xmath138 .
the second step is to superimpose vector @xmath138 with period @xmath139 and perform @xmath139-point fft on the superimposed vector @xmath140 $ ] .
hence , two hadamard product operations , two superimpose operators with period @xmath139 and two @xmath139-point fft are needed to recover the transmitted frequency domain training sequence @xmath111 , which is a low complexity approach compared to the traditional @xmath112 projectors approach
. we can conclude from ( 19 ) and ( 28 ) that the proposed phase differential and least squares carrier frequency offset estimators are functions of @xmath141 .
@xmath141 in awgn channel can be expressed as @xmath142 the conditional probability density function @xmath143 can be expressed as @xmath144 differentiating the log likelihood function @xmath145 with respect to @xmath146 , we have @xmath147 and @xmath148 hence , the cramer - rao lower bound of the proposed carrier frequency offset can be expressed as @xmath149^{-1}\\ & = \frac{n}{2\pi^2m^2\textmd{snr } } \end{split}\ ] ]
in this section , we test the proposed synchronization approach for mcm system with hexagonal t - f lattice via computer simulations based on the discrete signal model . in the following simulations , the number of subcarriers for hmct system is chosen as @xmath150=40 , and the length of prototype pulse @xmath114=600 .
the center carrier frequency is @xmath151=5ghz and the sampling interval is set to @xmath152=@xmath153s .
the system parameters of hmct system are @xmath2=25khz , @xmath1=@xmath154s and signaling efficiency is set to @xmath85=0.8 .
wssus channel is chosen as dd channel with exponential power delay profile and u - shape doppler spectrum . the mse performance of the proposed carrier frequency estimation algorithm over awgn channel is given in fig .
2 . we can see from fig .
2 that the proposed least squares carrier frequency estimation algorithm outperforms phase differential scheme at low snr , and both the proposed schemes can obtain an approximation to the cramer - rao lower bound mse performance . the mse performance of the proposed carrier frequency estimation algorithm over dd channel is given in fig .
3 . as shown in fig .
3 that the proposed least squares carrier frequency estimation algorithm outperforms phase differential scheme at low snr , but the phase differential scheme outperforms least squares algorithm at high snr . the dd propagation channel introduces energy perturbation among the transmitted symbols , hence there is a gap between the mse performance of the proposed two carrier frequency offset estimation algorithms and that of the cramer - rao lower bound .
in this paper , the effects of carrier frequency offset on hmct system is analyzed . meanwhile , a novel carrier frequency offset estimation approach , including preamble structure , carrier frequency offset estimation algorithm , is proposed for hmct system .
after detailed derivation , the close form performance lower bound of the proposed carrier frequency offset estimation alglrithm is given .
theoretical analyses and simulation results show that the proposed carrier frequency offset estimation approach for hmct system obtains an approximation to the cramer - rao lower bound mse performance over the dd channel with exponential power delay profile and u - shape doppler spectrum .
this work was supported by the jiangsu province national science foundation for young scholars under grant ( no .
bk2012055 ) and the young scientists pre - research fund of plaust under grant ( no .
kytyzlxy1211 ) , the national natural science foundation of china ( no . 60972050 ) and the jiangsu province national science foundation under grant ( no .
bk2011002 ) . 00 kumbasar v. , kucur o. ici reduction in ofdm systems by using improved sinc power pulse . _ digital signal processing _ , vol.17 , no.6 , pp.997 - 1006 , nov .
das s. , schniter p. max - sinr isi / ici - shaping multicarrier communication over the doubly dispersive channel .
_ ieee transactions on signal processing _ , vol.55 , no.12 , pp.5782 - 5795 , dec . 2007 .
abbas h.k .
, waleed a. m. , nihad s. , the performance of multiwavelets based ofdm system under different channel conditions _ digital signal processing _ , vol.20 , no.2 , pp.472 - 482 , mar .
gao x. , wang w. , xia x.g . , et al . cyclic prefixed oqam - ofdm and its application to single - carrier fdma .
_ ieee transactions on communications _ , vol.59 , no.5 , pp.1467 - 1480 , may 2011 .
p. jung , g. wunder , the wssus pulse design problem in multicarrier transmission , _ ieee transactions on communications _ ,
vol.55 , no .
10 , pp.1918 - 1928 , oct . 2007 .
g. lin , l. lundheim , n. holte , optimal pulses robust to carrier frequency offset for ofdm / qam systems , _ ieee communications letters _ , vol .
161 - 163 , mar .
p. siohan , c. siclet , n. lacaille , analysis and design of ofdm / oqam systems based on filterbank theory , _ ieee transactions on signal processing _ ,
5 , pp . 1170 - 1183 , may 2002 . m. ma , b. jiao , c. y. lee , a dual - window technique for enhancing robustness of ofdm against frequency offset , _
ieee communications letters _ , vol .
17 - 19 , jan . 2008 .
strohmer t , beaver s. optimal ofdm design for time - frequency dispersive channels .
_ ieee trans .
_ vol.51 , no.7 , pp.1111 - 1122 , jul . 2003 .
yuan z.g . , shen y.h . , a novel lofdm signal and its optimization over doubly - dispersion channels . in _ proc .
3rd ieee conference on industrial electronics and applications , 2008 .
iciea 2008 _ , pp.853 - 856 , jun .
han f.m . , zhang x.d .
hexagonal multicarrier modulation : a robust transmission scheme for time - frequency dispersive channels .
_ ieee transactions on signal processing _ , vol.55 , no.5 , pp.1955 - 1961 , may 2007
han f.m . , zhang x.d .
, mlsd for hexagonal multicarrier transmission with time - frequency localized pulses , _ ieee transactions on vehicular technology _ , vol.58 , no.3 , pp.1598 - 1604 , mar
han f.m . , zhang x.d .
, asymptotic equivalence of two multicarrier transmission schemes in terms of robustness against time - frequency dispersive channels .
_ ieee transactions on vehicular technology_,vol.59 , no.2 , pp.1598 - 1604 , feb . 2010 .
xu k. , shen y. h. , effects of carrier frequency offset , timing offset , and channel spread factor on the performance of hexagonal multicarrier modulation systems .
_ eurasip journal on wireless communications and networking _ ,
vol.2009 , pp.1 - 8 , jan .
xu k. , xu y. , zhang d. , sinr analysis of hexagonal multicarrier transmission systems in the presence of insufficient synchronization for doubly dispersive channel .
_ frequenz _ , vol.65 , no.5 , pp.149 - 157 , aug . 2011 .
xu k. , lv z. , xu y. , zhang d. , max - sinr based timing synchronization scheme in hexagonal multicarrier transmission .
_ wireless personal communications _ ,
doi : 10.1007/s11277 - 012 - 0550 - 5 , 2012 .
xu k. , xu y. , xia x. , zhang d. , on max - sinr receiver for hexagonal multicarrier transmission over doubly dispersive channel . to appear in _ proc .
ieee globecom 2012_. bello p. a.,characterization of randomly time - variant linear channels , _ ieee transactions on communication system _ ,
vol.11 , no.4 , pp.360 - 393 , dec
l. cohen , _ time - frequency analysis_. englewood cliffs , nj : prentice - hall , 1995 .
p. matthias , _ mobile fading channels_. west sussex , england : john wiley & sons , ltd , 2002 .
j. h. conway and n. j. a. sloane , _ sphere packings , lattices and groups _ , 3rd ed .
new york : springer - verlag , 1998 . j. proakis and m. salehi , _ digital communications _ , 5th edition . new york : mcgraw - hill , 2007 . | in this paper , a novel carrier frequency offset estimation approach , including preamble structure , carrier frequency offset estimation algorithm , is proposed for hexagonal multi - carrier transmission ( hmct ) system .
the closed - form cramer - rao lower bound of the proposed carrier frequency offset estimation scheme is given .
theoretical analyses and simulation results show that the proposed preamble structure and carrier frequency offset estimation algorithm for hmct system obtains an approximation to the cramer - rao lower bound mean square error ( mse ) performance over the doubly dispersive ( dd ) propagation channel .
shell : bare demo of ieeetran.cls for journals hexagonal multi - carrier transmission system ; preamble structure ; carrier frequency offset estimation ; cramer - rao lower bound ; |
narrow - line seyfert 1 galaxy ( nls1 ) is a class of active galactic nuclei ( agn ) .
the width of optical emission lines is narrower than that of seyfert 1 galaxies , and nls1s do not exhihit strong x - ray absorption like seyfert 2 galaxy . in the x - ray spectrum , there is often a large soft - excess .
nls1s are identified by the following three characteristics .
1 . fwhm ( h@xmath0 ) @xmath1 2000 kms@xmath2 ( goodrich 1989 ) 2 . @xmath3oiii@xmath4 $ ] / h@xmath5 3 ( osterbrock & pogge 1985 ) 3 .
strong permitted fe ii emission lines ( boroson & green 1992 ) nls1s have a small - mass black hole , but its mass accretion rate is almost as high as eddington limit ( marconi et al .
we can study the evolution of supermassive black holes at the center of galaxies by observing nls1s .
most of nls1s are radio - quiet ( r @xmath1 10 , r : radio loudness , ratio of 5 ghz radio to b - band flux densities ) ( kellermann et al . 1989 ) , but 7 @xmath6 of nls1s are radio - loud ( r @xmath7 10 ) and 2.5 @xmath6 are very radio - loud ( r @xmath7 100 ) ( komossa et al
. 2006 ) , and they are called as radio - loud narrow - line seyfert 1 galaxies ( rl - nls1 ) . with early fermi observation , gev gamma - ray emission has been discovered from pmn j0948 + 0022 , one of rl - nls1s ( abdo et al .
after that , gev gamma - ray emission was found from other three rl - nls1s ( abdo et al .
2009b ) , so it is found that rl - nls1s generally emit gev gamma - rays . from past observations ,
the multi - wavelength spectrum of rl - nls1 is similar to that of typical blazars ; the synchrotron emission in the lower energy band up to the optical band and the inverse compton scatterred x - ray and gamma - ray emission of low energy photons from disk , torus , and broad line region .
x - ray band is a transition region between the synchrotron and the inverse compton , and also there is a possible disk / corona emission . as described above , x - ray emission mechanism of rl - nls1s is uncertain .
therefore , we studied the energy - dependence of time variability of the x - ray emission of a rl - nls1s 1h 0323 + 342 , which have been observed by suzaku in 2009 and 2013 .
fig.[lc13 ] is a light curve of 1h 0323 + 342 observed by suzaku in 2013 .
1h 0323 + 342 varies with a time scale of several ks , but the soft x - ray band below 1 kev shows a independent behavior from the other bands .
fig.[multi ] shows correlations of count rates between 23 kev and other bands . the lower energy below 2 kev and the higher energy above 7 kev show a different variability from the middle energy band , suggesting that there are at least two spectral components in the x - ray band .
first , we fit the x - ray spectrum of 1h 0323 + 342 with the absorbed power - law model , but there remain redisuals at the higher and lower energy region ( @xmath8 = 1017.26/528 ) and thus the spectrum is not a simple power - law shape ( fig . [ redisual ] ) . to reduce the redisual in fig.[redisual ] , we added a broken power - law model this model of one hard power - law and one broken power - law represents a jet emission of inverse compton and synchrotron , respectively .
as in fig.[bkn09all ] and fig.[bkn13all ] , x - ray spectra can be fitted with this model for 2009 and 2013 ( table i ) .
the breaking energy becomes around 0.7 kev .
.fitting parameters of jet emission model in 2009 and 2013 [ cols="^,^,^,^,^,^ " , ] we also try to fit the spectra of two periods defined in fig [ lc13 ] with disk emission model , as jet emission model . as a result ,
a `` powerlaw '' component varied .
if disk model is correct , the variability is almost attributed to the disk / corona component and the reflection component is stable .
this behavior is similar to other seyfert galaxies .
we suggests that x - ray emission of 1h 0323 + 342 has at least two emission components , based on energy - dependence of time variability .
if the x - ray emission of 1h 0323 + 342 is dominated by jet emission , a variable component is a synchrotron , and it is similar to other blazers .
if the emission is dominated by disk / corona emission , disk / corona emission varies while disk reflection is stable , this is similar to seyfert galaxies .
we can not distinguish these two models by current data .
therefore an extensive study by astro - h observations with high energy resolution and good sensitivity in wide x - ray band is hopeful . | 1h0323 + 342 is one of narrow - line radio - loud seyfert 1 galaxies ( rl - nls1 ) , which is a new class of gamma - ray emitting agns .
narrow - line seyfert 1 galaxies ( nls1 ) have a small - mass black hole , but its mass accretion rate is almost as high as eddington limit .
therefore , by observing nls1s , we can know the evolution of supermassive black holes at the center of galaxies . some of nls1s are radio - loud and we call them rl - nls1 . from past observations ,
multi - wavelength spectrum of rl - nls1s is similar to that of typical blazars ; the synchrotron emission in the lower energy band up to the optical band , and inverse compton scattering of low energy photons from disk , torus , and broad line region .
x - ray band is a transittion region between the synchrotron and inverse compton , and also there is a possible disk / corona emission .
therefore , we studied the energy - dependence of time variability of the x - ray emission of 1h0323 + 342 , which have been observed by suzaku in 2009 and 2013 , in order to constrain the emission mechanism .
we found that the lower energy below 1 kev and the higher energy above 7 kev show a different variability from the middle energy band , indicating at least two emission components in the x - ray band .
x - ray spectrum is not a simple power - law , but requires an additional features ; a broken power - law plus flat hard component , or a power - law plus a relativistic reflection component . each spectral component seems to vary independently . |
the interstellar medium of our galaxy , consisting of large regions of ionized gas with free electrons , can be modeled as a cold tenuous plasma which alters the characteristics of any electromagnetic wave passing through it .
pulsar signals are ideal tools to study these effects because of the pulsed nature of their signal , which allows one to probe the effects of dispersion and scattering .
furthermore , the compact nature of the source allows one to observe the effects of scintillation . on the other hand
, the same dispersion and scattering effects can be a hindrance to the detection and detailed study of short period pulsars , especially at low radio frequencies and typical galactic distances ( see @xcite and @xcite for a review of these effects ) .
this becomes very important for low frequency radio telescopes such as the gmrt .
the gmrt located in khodad near pune , india is a multi - element aperture synthesis telescope consisting of 30 antennas , each with a diameter of 45 metres , and spread over a region of 25 km diameter ( @xcite ) .
it currently operates in 5 frequency bands centered around 150 mhz , 235 mhz , 325 mhz , 610 mhz and 1000 - 1450 mhz , with a maximum available bandwidth of 32 mhz .
the high sensitivity of the gmrt at low frequencies , as well as its wide frequency coverage with reasonably large bandwidths makes it a powerful instrument for pulsar studies .
pulsar observations can be carried out in the array mode in two ways : the incoherent array ( ia ) mode or the phased array ( pa ) mode(@xcite ) . in the ia mode , the power signals from each of the antennas are added to produce the final total intensity signal , whereas in the pa mode , voltage signals from the antennas are phased , added and then detected to produce the final signal .
the intensity signals in both these modes have 256 or 512 spectral channels , leading to minimum time resolutions of 16 and 32 microseconds , respectively .
these can be integrated to produce the desired time resolution , and then recorded on disk as intensities in multiple spectral channels .
the pa mode also has a voltage beam mode where the raw voltage from a single phased array beam can be recorded on a disk , as the real and imaginary parts from the output of a fast fourier transform ( fft ) on the original voltage time series , with 256 or 512 spectral channels .
pulsar observations at low frequencies are distorted by the deleterious effects of dispersion , which can limit the utility of such an instrument .
when an electromagnetic wave travels through the ionized interstellar medium , low frequency components of the signal travel slower than the higher frequency components due to dispersion , which causes a broadband sharp pulse from the source to be smeared out in time when detected with a receiver having a finite bandwidth . this effect becomes more significant at lower frequencies ( as the dispersion delay for a frequency @xmath0 increases as @xmath1 ) , and hence proper correction techniques are vital for pulsar observations in this regime .
the correction for dispersion involves accounting for the frequency dependent velocity of the electromagnetic waves in the interstellar medium , and applying appropriate time delays , or alternatively , phase corrections in the frequency domain to correct for dispersion .
this leads to two techniques for implementing such modifications : incoherent and coherent dedispersion ( @xcite ) . in the former , the observable bandwidth
is split into a number of smaller channels , and the signal is detected separately in each of the channels .
appropriate time delays are then applied to align the pulses in the separate frequency channels , which are then added to produce a dedispersed time series .
it is evident that this technique does not remove the intra - channel dispersion time , which can only be reduced by using a larger number of channels across the bandwidth .
however , this comes at the cost of reduced time resolution since the nyquist time resolution of a channel of width @xmath2 is @xmath3 .
coherent dedispersion ( @xcite ) can provide significant improvements over incoherent dedispersion by completely removing the effects of interstellar dispersion , and recovering the original nyquist time resolution of the sampled voltage signal .
this is done by deconvolving the received voltage signal at the telescope with the transfer function of the interstellar medium , which is treated as a linear filter .
the length of the impulse response to be deconvolved is roughly proportional to the inverse cube of the frequency of observation , linearly proportional to the dispersion measure ( dm ) along the line of sight , and has a quadratic dependence on the bandwidth of observation .
since convolutions are efficiently computed in the frequency domain by using efficient fft algorithms , it is evident that implementing such a technique at low frequencies and large bandwidths is a challenging task due to the computation of the long ffts ( the length of which has to be larger than the impulse response ) that are involved in the process .
the inaccuracies of the incoherent dedispersion technique can be a significant drawback for observations at low frequencies requiring high time resolution .
coherent dedispersion is thus very useful for pulsar observations with low frequency telescopes such as the giant metrewave radio telescope ( gmrt @xcite ) , the long wavelength array ( lwa @xcite ) , the low frequency array ( lofar @xcite ) and the murchison widefield array ( mwa @xcite ) . as a demonstration of the merits of the coherent dedispersion technique , we present incoherently and coherently
dedispersed folded profiles of the pulsar b1937 + 21 , observed with the gmrt at 325 mhz and dedispersed with our offline coherent dedispersion system , in figure [ fig:1937 + 21 ] .
since this pulsar has a relatively large dm of 71 pc / cc and a very short period of 1.55 ms , the effects of intra - channel dispersion time are clearly evident in terms of smearing of the folded profile in the incoherently dedispersed case . on the other hand ,
the coherently dedispersed profile shows sharper pulses and a clear indication of a scattering ` tail ' ( described in section [ scattering ] ) .
we describe the details of the coherent dedispersion technique in section [ cd ] .
@c@ + ( a ) incoherently dedispersed and normalized folded profile for b1937 + 21 .
@c@ + ( b ) coherently dedispersed and normalized folded profile for b1937 + 21 .
coherent dedispersion necessarily requires one to process the original nyquist sampled voltage signal from the phased array beam .
since this voltage data can not be integrated , it either has to be stored on a disk for offline processing , or processed in real - time .
recording voltage data in such cases can require very large amounts of disk space ( particularly for long term programs such as high precision timing ) , and may be a difficult task as one goes for larger bandwidths of observation .
for example , with the current gmrt software back - end ( @xcite ) , the data recording rate is 120 gb / hour in the 16 mhz mode and 240 gb / hour in the 32 mhz mode of observation .
for the 200 mhz bandwidth of the upgraded gmrt wideband back - end , the data recording rate required would be 1500 gb / hour , which would be impractical for recording followed by offline processing , especially for long term observing programs .
the data rates would be similar for other low frequency telescopes with similar bandwidths . for higher frequency telescopes like the green bank telescope ( @xcite ) where the bandwidths may be larger
, the data rates would be proportionately higher .
implementing coherent dedispersion in real - time provides a solution to overcome these drawbacks .
however , the process of coherent dedispersion itself is computationally expensive ( especially at low radio frequencies ) , requiring the computation of large ffts ( often longer than a million points ) , and hence requires a large amount of computing power to implement in real - time . an effective way to reduce the computational load of the process
is dividing the bandwidth of the receiver into a number of channels , coherently dedispersing each sub - band , followed by alignment of the individual sub - bands in time ( the individual sub - bands thus do not contain any intra - channel dispersion ) . using smaller sub - bands
reduces the length of the impulse response to be convolved with , thereby reducing the required length of the ffts and the overall computational load
. however , this comes at the cost of reduced time resolution , which is now limited by the inverse of the bandwidth of the individual smaller sub - bands .
one of the first real - time coherent dedispersion systems was developed in the mid-1980s by hankins and rajkowski ( @xcite ) at the arecibo observatory for a modest bandwidth of 1.5 mhz centered at a frequency of 1414 mhz using a hardware based implementation . before this ,
coherent dedispersion was most commonly implemented by recording data on magnetic tape , followed by offline processing in hardware / software based systems .
the recent emergence of parallel computing on cpu - based clusters and gpus has provided a powerful tool to realize the associated computational challenges and develop real - time coherent dedispersion systems .
the puma - ii back - end developed at the westerbork synthesis radio telescope ( wsrt @xcite ) demonstrated the use of a cpu cluster to implement a near real - time coherent dedispersion system for a bandwidth of 160 mhz , by dividing the bandwidth into 8 sub - bands , each with a bandwidth of 20 mhz , and then dedispersing and combining the signal .
the system first records the baseband signal into a cluster of disks , which is then processed with the a 32 node cluster in near real - time .
it has been implemented with the code available from the open - source software library digital signal processing software for pulsar astronomy ( dspsr @xcite ) .
the dspsr was developed for pulsar observations at the parkes telescope and allows coherent dedispersion to operate in multiple modes involving both cpus and gpus .
as with the dedispersion system in puma - ii , dspsr implements coherent dedispersion by forming a filter - bank of smaller bandwidths , which are coherently dedispersed , detected and aligned in time .
the green bank ultimate pulsar processing instrument ( guppi @xcite ) , functioning on fpgas , a cpu cluster and a gpu cluster of 8 gpus includes a wide bandwidth coherent dedispersion system supporting a total bandwidth of 800 mhz .
the total bandwidth is divided into 8 smaller sub - bands , each of which can be processed in one gpu of the 8 gpu cluster .
the nancay telescope has recently developed a real - time gpu - based coherent dedispersion back - end ( @xcite ) .
a gpu cluster consisting of 4 nvidia geforce 8800 gtx cards is used to coherently dedisperse a bandwidth of 128 mhz ( after sub - banding into four 32 mhz bands ) , at frequencies in the l - band .
the range of fft lengths implemented by these systems ( typically of the order of 1 million points ) are however much smaller than that required for dedispersing signals at low frequencies with similar bandwidths and dms .
scattering occurs due to electron density fluctuations in the interstellar medium , which causes part of the signal from the source to be deflected from its original path of propagation and travel along longer nearby paths . in the case of pulsars ,
this time delay causes narrow pulses to be spread out into an exponential ` tail ' ( see @xcite ) .
the effective time resolution obtainable from coherent dedispersion is thus limited by scatter broadening in the interstellar medium which , unlike dispersion , can not be removed by post - processing techniques .
hence , it is a useful exercise to estimate the best time resolution obtainable in the presence of scattering .
to estimate the effect of scattering for the frequency bands in use at the gmrt , we use the empirical relation determined by bhat et al .
( @xcite ) relating the scattering time @xmath4 to the dispersion measure ( dm ) of the pulsar .
we compare the estimated scattering time @xmath4 as a function of dm for a fixed observing frequency ( center frequencies of 325 mhz and 610 mhz ) with the intra - channel smear time for incoherent dedispersion using 256 channels ( as implemented at the gmrt ) for a fixed bandwidth of observation .
the results are shown in figure [ fig : scattering ] for two sets of parameters .
note that the @xmath4 - dm relation has a significant scatter in observations ( the plots show a scatter of one order of magnitude , which is typically observed @xcite ) , and hence the plots are only meant to give approximate estimates . @c@ + ( a ) center frequency of 325 mhz @c@ + ( b ) center frequency of 610 mhz as can be seen from figure [ fig : scattering ] , for small values of dm , the scattering time is significantly smaller than the intra - channel dispersion time for incoherent dedispersion .
as a result , coherent dedispersion can be useful in this regime for obtaining better time resolution .
the scattering curve still provides a lower limit to the obtainable time resolution ( instead of the nyquist resolution ) after coherent dedispersion has been performed .
however , above a particular value of dm ( depending on the frequency of observation ) , the scattering time dominates over the intra - channel smear time , and hence coherent dedispersion would provide no improvement over incoherent dedispersion in this regime . for 610 mhz observations ,
this limiting value of the dm is around 100 pc / cc , and it is a bit less than that for 325 mhz observations .
the action of the interstellar medium on a propagating electromagnetic wave can be considered as that of a linear filter with a transfer function @xmath5 , where @xmath0 is the observing frequency .
let @xmath6 and @xmath7 be the fourier transforms of the electromagnetic wave emitted at the pulsar and that received at the telescope respectively . considering a real - sampled signal ,
let the bandwidth of observation extend from a frequency of @xmath8 to @xmath9 , where @xmath10 is the lower edge frequency of the band , and @xmath11 is the bandwidth .
if @xmath12 be the deviation from the lower edge of the band so that @xmath13 , then the expression for the received wave can be expressed as @xmath14 .
the phase picked up by an electromagnetic wave travelling though the ism is given by ( see @xcite ) , @xmath15 where l is the distance travelled in the ionized plasma , @xmath16 is the plasma frequency of the medium and @xmath17 is the speed of light .
hence , the transfer function can be written as @xmath18 , which , using equation [ new_eq ] , can be rewritten as @xmath19}\label{eq:1}\ ] ] the first term in the square brackets is a constant phase offset which is lost in detection , while the second term is a linear frequency gradient which corresponds to a time delay in the arrival of the pulse .
the third term causes the dispersion within the band , and needs to be corrected for .
hence the transfer function to correct for the dispersion within the band is the inverse of the third term in equation [ eq:1 ] . in terms of the dispersion measure of the pulsar
, the third term can be expressed as , @xmath20 where @xmath21 is a constant equal to @xmath22 cc / pc mhz .
the impulse response associated with the transfer function in equation [ eq:2 ] has been numerically computed for two sets of parameters and are shown in figure [ fig : impulse ] ( a ) , with the normalized amplitude of the response plotted as a function of time .
the necessary parameters in this case are the observing frequency , bandwidth of observation and the dm along the line of sight to the pulsar .
the length of the response is equal to the dispersion time in the observing band .
the discontinuous termination of the transfer function in the frequency domain ( corresponding to multiplication of the transfer function with a rectangular window , with a width equal to the bandwidth of the receiver ) produces ripples in the time domain impulse response . these ripples can produce unreal artifacts in the data , and hence need to be eliminated with the use of an appropriate window or taper function in the frequency domain .
the modified impulse response for two well known window functions , the hanning and welch windows are shown in figure [ fig : impulse ] ( b ) , which show reduced levels of oscillations , however , at the cost of reduced sensitivity to the edges of the band .
the taper function multiplied with the inverse transfer function is known as the _ chirp function_. @c@ + ( a ) normalized impulse response amplitude .
@c@ + ( b ) effect of applying a window function .
deconvolution of the ism impulse response requires proper book - keeping of edge effects to prevent aliasing in the computed time series .
the overlap - save method is a standard method of computing long convolutions , as described in @xcite , where the number of samples corresponding to the length of the impulse response is discarded from the beginning of the output of a time series segment which has been convolved , while the same number of samples at the end is overlapped with the next such segment in the data to maintain continuity in the time series . at the gmrt ,
the voltage beam data is recorded as 8-bit integers from the output of a fft performed on the original sampled voltage from the phased array .
consequently , one needs to perform an inverse fft on the beam data to get back the original time series , and then perform a longer fft to implement the deconvolution algorithm of coherent dedispersion .
the length of the long fft ( designated with the letter ` n ' henceforth ) has to be at least greater than the number of samples corresponding to the length of the impulse response , i.e. , the dispersion time in the band .
since ffts can be efficiently computed if the length is a power of 2 , we choose the number of samples to be the smallest power of 2 greater than twice the number of samples corresponding to the dispersion time .
the phase changes are then applied to the spectral output from the long fft , and the overlap - save method is used to get back dedispersed data in the time domain .
the final algorithm implemented for the gmrt is schematically shown in figure [ fig : coherent_algorithm ] .
our initial attempts towards developing a coherent dedispersion system for the gmrt were centered around designing an optimized cpu - based implementation of the algorithm ( see @xcite for a cluster based cpu implementation ) and comparing its performance with respect to that required for a real - time pipeline .
we report the steps involved in implementing and optimizing such a cpu - based coherent dedispersion system , as well as our results on its performance as compared to real - time processing rates .
the computing resources used include an intel xeon x5550 cpu @ 2.67 ghz with 4 cores ( supporting a maximum of 8 threads simultaneously ) with on board ram of 4 gb .
we have based our cpu implementation on the fastest fourier transform in the west ( fftw ) package ( @xcite ) , which provides optimized routines to perform forward and inverse ffts on sampled real data .
all our routines have been coded using the double precision floating point format , since we found that the use of single precision resulted in clearly visible artifacts in the dedispersed time series .
we believe that this is due to the fact that the single precision floating point fftw routines produce increasingly inaccurate results ( which are much larger than those for double precision computations ) when computing longer ffts ( see @xcite ) .
see section [ cufft ] for an example of this effect . given the different steps involved in the dedispersion algorithm , the first step towards optimization would involve dividing the algorithm into different and independent sections which can be executed in parallel on a multiple core processor .
however , as shown in figure [ fig : coherent_algorithm ] , each of the steps in the algorithm is dependent on the earlier step , and hence one is required to proceed by performing the different steps in parallel , but on consecutive blocks of data . in this case
, the blocks of data correspond to each of the long n point segments on which the fft is to be performed .
we have designed a scheme where the process has been divided into 3 different sections , such that the respective routines can process different blocks of data and then transfer these to the next process in the pipeline via shared memory .
the division of the process into three different sections follows logically from the schematic shown in figure [ fig : coherent_algorithm ] , with the individual sections ( processes ) corresponding to : * inverse transforming the voltage fft data * performing coherent dedispersion on the voltage time series * detecting and writing the dedispersed intensity signal to a disk seven of the eight available threads on the system are used .
the number of threads assigned to each process has been selected based on tests to determine the best configuration to balance the computational load of different routines .
the optimal algorithm derived is depicted in figure [ fig : cpu_scheme ] . + the details of the individual routines are as follows : + _ inverse transform ( block i ) : _ the spectral data from the gsb is from a 1024 point or 2048 point fft on the original nyquist sampled voltage data .
the gsb is however configured to output this data in larger blocks , consisting of a number of these fft outputs in series .
hence , to improve the processing rate , we use 2 threads on the processor to simultaneously inverse transform 2 halves of a given block into a continuous time series . the data is then written into a ping - pong scheme double buffer in shared memory .
the use of the double buffer in shared memory allows the next routine in the pipeline to read from a shared memory segment , while the other is being simultaneously written into by the current routine .
+ _ coherent dedispersion ( block i-1 ) : _ the chirp function is initialized in this routine , and the voltage time series ( block i-1 ) from the inverse transform is read from the shared memory double buffer .
multi - threaded fftw using 4 threads is used to compute the long n - point fft , the chirp function is applied and then the data is inverse transformed into a time series . this time series is written to a single shared memory segment for the next step of processing . + _ write to disk ( block i-2 ) : _ this routine ( using 1 cpu thread ) reads the deconvolved voltage data ( block i-2 ) written to a shared memory segment by the coherent dedispersion routine , squares it , and integrates it to a given number of samples .
this intensity output is written to a file on the disk .
+ the algorithm has been characterized by measuring the processing rate of the inverse transform and coherent dedispersion routines ( which perform the maximum amount of computation ) , as well as of the entire pipeline itself , as a function of the length of the dedispersion transform ( twice the number of samples corresponding to the dispersion time in the band ) .
we have performed the tests by simulating random data for the voltage stream , storing it in the shared memory of the system in a structure identical to real - time observations , followed by processing the data with the routines described above .
hence , these tests did not involve any disk access and are not i / o dominated .
the lengths of the dedispersion transforms bench - marked are typical lengths of the transforms ( chosen to be powers of 2 ) that are required for the observing frequencies and bandwidths available at the gmrt .
the rate has been expressed as a multiple of the real - time data acquisition rate and in terms of the number of gflops achieved in run - time .
the rate of a routine must be higher than 1 to be able to process data in a real - time application .
the number of gflops has been estimated by calculating the number of floating point operations for each of the two routines .
for the inverse transform routine , this is given by : @xmath23 where n is the length of the original fft on the voltage time series and @xmath24 is the time in nanoseconds taken to compute the inverse transform . for the coherent dedispersion routine ,
the same is given by : @xmath25 where n is the length of the long fft to be computed for coherent dedispersion and @xmath24 is the time taken in nanoseconds to perform the computations . here
the the first term in the numerator corresponds to the computation for the forward and inverse ffts , while the second term comes from applying the complex phases to the @xmath26 frequency channels .
@c@ + ( a ) relative processing rate for the 16mhz mode .
@c@ + ( b ) relative processing rate for the 32 mhz mode .
the results are shown in figures [ fig : cpu_rate ] and [ fig : cpu_perf ] .
it is evident that the processing rate of the optimized cpu - based implementation is significantly below that required for a real - time application ( with the given computing resources ) .
the performance of the full pipeline is limited by the slowest routine in the process , i.e. , coherent dedispersion , and hence the full pipeline performs at roughly the same rate as the coherent dedispersion routine .
our benchmarks indicate that the full cpu - based pipeline performs 3.25 x slower ( on average ) than real - time data acquisition in the 16 mhz mode .
consequently , we have reserved our cpu - based implementation exclusively for offline processing of data , and developed a real - time pipeline using a gpu - based implementation of the algorithm .
we describe the working of the real - time pipeline in the next section .
a real - time coherent dedispersion pipeline provides the ideal platform for harnessing the power of parallel computation on graphics processing units ( gpus ) . for the current implementation
, we have used a nvidia tesla c2075 gpu , with 448 processor cores , each with a clock speed of 1.15 ghz , and a total internal memory of 6 gb , mounted on the same host machine as used for the cpu based implementation .
we present below our implementation of a real - time gpu based coherent dedispersion pipeline , which has been optimized such that the most compute intensive portions of the algorithm are executed on the gpu .
the routines for the gpu - based algorithm have been coded in c under the nvidia cuda^tm^ programming environment .
our benchmarks of the cpu - based implementation clearly indicate that the coherent dedispersion routine is the slowest process in the pipeline , evidently due to the computation of the long fft . the cufft library ( @xcite ) , available with the cuda^tm^ platform , provides routines to compute ffts much faster than standard cpu - based fft libraries , using the computational power of hundreds of processor cores available on standard gpus .
our tests ( explained later ) clearly demonstrate the improvement in the performance of the dedispersion process when using the cufft library . we have based our routines on the double precision floating point cufft routines since we found that the use of single precision resulted in clearly visible artifacts in the dedispersed time series .
similar to the case of the cpu implementation , we believe that this is again due to the fact that the single precision floating point cufft routines produce increasingly inaccurate results when computing longer ffts , which are significantly larger than those produced by the double precision versions ( see @xcite ) . as a demonstration of the artifacts observed
when using the single precision fft routines ( both in the cpu - based fftw and the gpu - based cufft ) , we show in figure [ fig : floatvsdouble ] coherently dedispersed folded profiles from an observation of the pulsar b0329 + 54 at 325 mhz with the gmrt .
figure [ fig : floatvsdouble](a ) shows the coherently dedispersed folded profile using the double precision mode of the pipeline , and is in good agreement with profiles published elsewhere ( @xcite ) .
figure [ fig : floatvsdouble](b ) shows the same data coherently dedispersed in the single precision mode , where one clearly observes the additional artifacts produced in and around the main pulse region .
c + \(a ) folded profile produced from the double precision version of the coherent dedispersion pipeline .
c + \(b ) folded profile produced from the single precision version of the coherent dedispersion pipeline .
the additional artifacts are clearly visible in the profile .
the availability of thousands of threads on a gpu provides further opportunity to parallelize the dedispersion algorithm to improve its performance .
the memory of the gpu and the host cpu are separate , unless one uses pinned memory on the host .
a large amount of allocated pinned memory , however , degrades system performance since it leaves less memory available for paging .
since the dedispersion process requires computation on a large amount of data at once , this data has to be transferred to the gpu memory from the host memory before any computation can be performed .
memory transfers turn out to be relatively expensive as compared to kernel executions , and hence parallelizing memory transfers and kernel executions can improve the performance of an algorithm as compared to a serial implementation .
cuda^tm^ allows for simultaneous memory transfers and kernel executions with the use of streams . for our algorithm
, we have parallelized the memory transfers and kernel executions on the gpu such that at any iteration , there are 3 streams functioning on the gpu : * transferring an already deconvolved block of data from the gpu memory to the host memory . * computing the long fft , applying the chirp function and then computing an inverse fft on the current data block to get deconvolved voltage data in gpu memory .
the chirp function is applied by initializing a kernel which applies the chirp function to each frequency channel in parallel by calling the required number of threads .
* asynchronously transferring the next data block to be processed from the host memory to gpu memory .
the process is depicted in figure [ fig : gpu_streams ] for a sequence of 4 consecutive data blocks ( n point ffts ) .
the real - time coherent dedispersion algorithm is similar to the cpu based algorithm , with the exception of the long ffts for the deconvolution being performed on the gpu to improve the processing rate of the pipeline .
it operates on the following key principles : * the entire process again logically divides into 3 different processes ( as described in section [ cpu_optimize ] ) , which can be run in parallel on a multi - processor system , but on consecutive blocks of data .
the individual sections correspond to the inverse transform , coherent dedispersion and detection followed by writing to disk respectively . *
as described in section [ gpu_optimize ] , the process handling the computation in the gpu can be further sub - divided into 3 parallel processes , handling data transfer and computation inside the gpu .
this is done to reduce the overhead of memory transfers to and from the gpu .
overall , this leads to a 5-way parallelism in the final algorithm . * transferring the coherent dedispersion process over to the gpu
leaves a larger number of free threads in the cpu , which can be now used to speed up other processes in the pipeline .
consequently , the inverse transform process now uses 4 threads instead of 2 as in the cpu based implementation , which leads to a significant improvement in the processing rates ( described below in section [ gpu_performance ] ) . * the configuration of threads assigned to different processes has been restructured to allow optimal performance in the real - time pipeline by balancing the computation load of different routines .
as described below , the overall coherent dedispersion pipeline now uses only 6 of the 8 available threads , with the 2 free threads being left to accommodate 2 additional processes which are involved in real - time operations : the process writing the voltage beam data to shared memory , and the process synchronising the dedispersion process for the 2 polarisations of the beam ( on different systems ) via an mpi based routine . in the voltage beam mode of the gsb ,
the voltage data ( in the form of fft outputs from the original sampled voltage ) is written into a shared memory on two different systems , each receiving the voltage data for two polarisations and synchronized with an mpi based routine to allow one to combine the signals from the two polarisations after detection .
the real - time algorithm is depicted in figure [ fig : realtime_algo ] .
there are again 3 routines which run in parallel , and simultaneously handle different consecutive blocks of data , which are transferred via shared memory . in this case , with the availability of streams on the gpu , we have designed an algorithm with 5-way parallelism to reduce the overhead of memory transfers to and from the gpu .
the voltage beam data is read directly from the shared memory on both the systems which are receiving the data .
the pipeline runs by performing the following tasks in parallel ( the respective routine names are shown in bold ) : 1 . *
inverse transform * : ( using 4 cpu threads ) inverse fft on beam data for the i^th^ block , by splitting the data block into 4 sections and inverse transforming the 4 blocks in parallel .
this allows the simultaneous utilisation of 4 threads on the cpu ( note that we could use only 2 threads for this process in the cpu - based implementation since the coherent dedispersion routine used up 4 threads ) to complete this task . the data is written to a shared memory double buffer .
the use of the shared memory double buffer allows the next routine in the pipeline to read from one segment in the shared memory while the next block is still being written into the other segment simultaneously .
* coherent dedispersion * : ( using 1 cpu thread ) 1 . copy ( i-1)^th^ block from shared memory to gpu memory 2 .
perform long fft on ( i-2)^th^ , apply chirp function and then produce dedispersed time series with appropriate overlap . for applying the chirp function , n/2 threads of a particular kernel ( corresponding to the choice of a particular taper function )
are initialised , corresponding to each of the n/2 frequency channels produced by the forward fft in the gpu .
3 . copy ( i-3)^th^ block from gpu to host shared memory in a double buffer . + the use of this 3 way parallelism inside the gpu allows us to eliminate overheads of data transfer to and from the gpu , by simultaneously processing and copying different segments of data from the voltage beam . +
the overlap - save method of computing convolutions requires one to reject a certain number of samples at the end of a given block , and overlap the same length with the previous block ( corresponding to the length of the impulse response ) .
hence , with every input block of the long fft , one loses samples from the end of the block . to ensure continuity in the time series
, we have implemented the gpu routine such that the number of samples lost in this process is then prefixed to the next block in the input data stream .
write to disk * : ( using 1 cpu thread ) read deconvolved voltage data from the shared memory double buffer .
square , integrate to certain number of samples , and write intensity to a disk .
again , the use of the double buffer allows this detection routine to read from a shared memory segment simultaneously while the next segment is being written to by the gpu routine coherent dedispersion .
the dspsr ( @xcite ) also uses some of the techniques we have used in our implementation , including the use of ring buffers in shared memory for real - time signal processing .
further , the thread - safe input buffering technique in dspsr is similar to the technique we have used to account for the samples lost in the overlap - save method of computing convolutions .
however , the process of initially inverse transforming the data is an additional task that is required for the gmrt voltage beam and has been a part of our implementation .
further , since dspsr was designed primarily to handle larger bandwidths at higher frequencies , it uses the filter - bank scheme to split the full band into narrower bands on which coherent dedispersion is performed .
since we have relatively smaller bandwidths to deal with at the gmrt , we do not use sub - banding which , in principle , allows us to achieve higher time resolutions .
similar to the analysis done for the cpu based implementation described in section [ cpu_analysis ] , we have benchmarked the performance of our gpu - based real - time pipeline , using the same equations to measure the processing rates .
the improvements clearly show up as a significant increase in the computation rate , and in the number of gflops achieved in run - time .
the results are shown in figures [ fig : gpu_rate ] and [ fig : gpu_perf ] . @c@ + ( a ) relative processing rate for the 16mhz mode .
@c@ + ( b ) relative processing rate for the 32 mhz mode .
the performance of the gpu - based pipeline shows improvements on a number of fronts as compared to the cpu based implementation .
firstly , the processing rate of the inverse transform routine shows about a factor of 3 improvement in processing rate as compared to the cpu - based implementation ( and overall 3 times faster than real - time data acquisition rate in 32 mhz mode ) , both in time taken and the number of gflops achieved .
though this routine still runs on the cpu , this marked improvement arises due to the routine now using 4 threads , instead of 2 in the cpu - only based implementation .
the most compute intensive portion of the algorithm , i.e. the coherent dedispersion routine , shows a factor of about 9 improvement in processing rate ( and overall 1.5 times faster than real - time data acquisition rate in 32 mhz mode ) on the gpu , which was what we originally aimed to achieve by moving our implementation to a gpu .
clearly , porting over the algorithm to a gpu has now allowed us to achieve better than real - time processing rates for the coherent dedispersion pipeline .
our benchmarks indicate that the optimized full pipeline can perform coherent dedispersion 1.4 x faster ( on average ) than real - time data acquisition in the 32 mhz mode and 2.5 x faster ( on average ) than real - time data acquisition in the 16 mhz mode .
the processing rate of the pipeline is thus well - suited for a real - time application , which has been subsequently verified in real - time tests .
we present these results in the next section .
the coherent dedispersion pipeline has been tested in real - time at the gmrt , running in parallel on two systems with identical specifications , each receiving one polarisation of the voltage beam data .
we present here the first results obtained from the real - time pipeline . in figure [ fig:0329_realtime ]
( a ) , the folded profile for the pulsar b0329 + 54 observed at 325 mhz with the real - time coherent dedispersion pipeline is shown .
figure [ fig:0329_realtime ] ( b ) shows a sample single pulse from the same data set , exhibiting clear evidence for the occurence of micropulses .
figure [ fig : realtime_2 ] shows folded profiles for the pulsars b0833 - 45 and b2016 + 28 ( zoomed into the on pulse region ) .
the observations parameters are given in table [ test_table ] , and have also been mentioned in the figure captions .
the folded profiles agree with the ones available in the databases provided in @xcite and @xcite .
these tests and sample results show that the pipeline works stably , maintaining real - time performance without losing any blocks of data segments of the voltage time series which are written as individual blocks in the shared memory by the gsb .
if the pipeline does not process data fast enough , blocks of data would be lost in real - time analysis , leading to gaps in the observed time series .
these gaps would show up as multiple pulse features at different phases in the folded profile , which is not the case .
c + \(a ) the folded profile of the pulsar b0329 + 54 .
the profile has been zoomed into the on - pulse region .
c + \(b ) a single pulse from the data stream corresponding to the folded profile in ( a ) , clearly showing small time - scale structure in the pulse intensity .
the figure has been zoomed into the region of the central component of the pulse in phase .
p11 cm c + \(a ) folded profile of the pulsar b0833 - 45 observed at 235 mhz with a bandwidth of 16 mhz for a duration of 5 minutes .
the pulsar has a dm of 67.99 pc / cc and a period of 89.3 ms .
owing to the high dm and low frequency of observation , the scattering tail is very prominent and extends throughout the profile .
time resolution used is 30 @xmath27 .
c + \(b ) folded profile of the pulsar b2016 + 28 observed at 325 mhz with a bandwidth of 32 mhz for a duration of 15 minutes .
the profile has been zoomed into the on - pulse region .
the pulsar has a dm of 14.172 pc / cc and a period of 557.9 ms .
time resolution used is 15 @xmath27 .
currently , the range of dms for a given frequency of observation that can be dedispersed in real - time using our pipeline are limited only by the memory of the gpu . for the observing frequency of 610 mhz ,
the highest dm that can be dedispersed is @xmath28 3000 pc / cc and @xmath28 750 pc / cc for the 16 mhz and 32 mhz modes respectively . for the observing frequency of 325 mhz , the corresponding maximum dms
are @xmath28 480 pc / cc and @xmath28 120 pc / cc for the 16 mhz and 32 mhz modes respectively . figure & observation frequency ( mhz ) & bandwidth ( mhz ) & pulsar name & pulsar period ( ms ) & dm ( pc / cc ) & duration of observation + [ 0.5ex ] 12 ( a ) & 325 & 32 & b0329 + 54 & 714.5 & 26.77 & 900 s + 12 ( b ) & 325 & 32 & b0329 + 54 & 714.5 & 26.77 & 19.8 ms + 13 ( a ) & 235 & 16 & b0833 - 45 & 89.3 & 67.99 & 300 s + 13 ( b ) & 325 & 32 & b2016 + 28 & 557.9 & 14.172 & 900 s +
we have developed a real - time coherent dedispersion pipeline for the voltage beam mode of the gmrt using a gpu - based implementation of the algorithm .
our benchmarks indicate that , using a tesla c2075 gpu , the pipeline is able to perform coherent dedispersion at better than real - time data acquisition rates for the currently available bandwidths at the gmrt , of 16 mhz and 32 mhz , by factors of approximately 2.5 & 1.4 , respectively .
this real - time pipeline has been tested fairly extensively and is found to work reliably , without losing any data .
a cpu based implementation of the pipeline has also been developed and released as an offline processing package , along with an offline version of the gpu - based implementation .
the first results from the pipeline clearly demonstrate the potential of this system as a powerful tool for low frequency pulsar observations at the gmrt , covering aspects such as studies of pulsar microstructure , accurate profiles of millisecond pulsars at low radio frequencies and development of a high fidelity pulsar timing machine .
our coherent dedispersion pipeline has some unique features as compared to existing coherent dedispersion systems ( described in section [ compare ] ) : firstly , our system works completely in real - time , without the requirement of recording data on disk ( as implemented in puma - ii ) .
further , our system does not split the bandwidth of observation to reduce the computational load of the process , which allows one to achieve higher time resolutions than that obtainable by sub - banding .
also , we have implemented the process with transform lengths much larger than the ones explored by other systems ( e.g. guppi ) , as the impulse response duration of the ism is much longer at the low frequencies our pipeline operates at .
this is thus one of the first cases of the development of a real - time coherent dedispersion pipeline for a large , low frequency multi - element radio telescope like the gmrt .
our current design , matched to the existing gmrt system , is easily extensible to a larger bandwidth system .
in particular , we plan to extend the implementation of this pipeline to the gmrt wideband back - end ( gwb ) that is being developed for the upgraded gmrt .
for the large bandwidths of operation of the gwb ( 100 , 200 and 400 mhz ) , recording full time resolution voltage data ( if possible at all ) will require a prohibitively large amount of disk storage space , even for relatively small duration observations .
in such a case , a real - time pipeline for coherent dedispersion becomes a necessity , if one wants to fully exploit the wider bandwidths of the upgraded gmrt and make it an outstanding instrument to study pulsars at low frequencies .
this should be fairly easy to do , as we show below .
the computational complexity for implementing the real - time coherent dedispersion pipeline stems from two aspects : the first is that the length of the ffts involved in the deconvolution increases as @xmath29 ( where @xmath30 is the bandwidth of the observations ) , since both the sampling rate and dispersion time across the band increase approximately linearly with @xmath30 .
this would amount to a factor of 100 times increase over that of the current gsb implementation , and one may run into memory size limitations on the gpu .
this problem can be partially overcome by the trade - off of splitting the wideband into a few narrower sub - bands , the coherent dedispersions of each of which can fit within the memory constraints of the gpu , and then combining the coherently dedispersed signals from all the sub - bands to get a time resolution that is the inverse of the bandwidth of one sub - band .
the second is the fact that even for a given length of the fft , the total computational load would go up as @xmath31 , implying a factor of approximately 30 increase in the required computation rate .
since the current system is built on a relatively old gpu ( tesla c2075 ) , it should be possible to overcome both these challenges with the future generations of the gpus that are expected to become available in the near future , and thus implement a real - time coherent dedispersion system for the upgraded gmrt
. we would like to thank the staff members from ncra who have helped at various stages of this work .
we acknowledge the support of the back - end team at the gmrt who have provided support for the computing infrastructure that was required for this work .
we would particularly like to thank mr .
s. s. kudale from the gmrt digital back - end team for useful inputs on the voltage - beam mode of the telescope .
the gmrt is run by the national centre for radio astrophysics of the tata institute of fundamental research .
+ * compliance with ethical standards * + the work was carried out as a part of the student training programme of the national centre for radio astrophysics ( ncra ) , pune , using equipment and facilities funded by ncra .
kishalay de was visiting yashwant gupta at ncra , pune during the working period .
kishalay de s participation in the project was supported by a fellowship from the kishore vaigyanik protsahan yojana ( kvpy ) scheme of the department of science and technology , government of india , which promotes research careers for undergraduate students .
ait allal , d. , weber , r. , cognard , i. , desvignes , g. and theureau , g. , 2009 , _ rfi mitigation in the context of pulsar coherent dedispersion at the nancay radio astronomical observatory _
, 17th european signal processing conference , glasgow , scotland .
bhat , n. d. ramesh , cordes , james m. , camilo , fernando , nice , david j. and lorimer , duncan r. , 2004 , _ multifrequency observations of radio pulse broadening and constraints on interstellar electron density microstructure _ , the astrophysical journal , * 605 * , 759 - 783 .
gupta , y. , gothoskar , p. , joshi , b. c. , vivekanand , m. , swain , r. , sirothia , s. and bhat , n. d. r. , 2000 , _ pulsar research with the gmrt : a status report _
, pulsar astronomy - 2000 and beyond , asp conference series , proceedings of the 177th colloquium of the iau held in bonn , germany , * 202 * , 277 .
karuppusamy , ramesh , stappers , m. , and van straten , w. , 2008 , _ puma - ii : a wide band pulsar machine for the westerbork synthesis radio telescope _ , publications of the astronomical society of the pacific , * 120 * , 191 - 202 .
swarup , g. , ananthakrishnan , s. , subrahmanya , c. r. , rao , a. p. , kulkarni , v. k. and kapahi , v. k. , 1997 , _ the giant metrewave radio telescope _ ,
high - sensitivity radio astronomy , proceedings of a meeting held at jodrell bank , university of manchester , january 22 - 26 , 1996 , 217 .
whitney , alan , booler , t. , bowman , j. , emrich , d. , goeke , b. and remillard , r. , 2011 , _ the murchison widefield array ( mwa ) : current status and plans _ , bulletin of the american astronomical society , * 43*. | a fully real - time coherent dedispersion system has been developed for the pulsar back - end at the giant metrewave radio telescope ( gmrt ) .
the dedispersion pipeline uses the single phased array voltage beam produced by the existing gmrt software back - end ( gsb ) to produce coherently dedispersed intensity output in real time , for the currently operational bandwidths of 16 mhz and 32 mhz .
provision has also been made to coherently dedisperse voltage beam data from observations recorded on disk .
we discuss the design and implementation of the real - time coherent dedispersion system , describing the steps carried out to optimise the performance of the pipeline .
presently functioning on an intel xeon x5550 cpu equipped with a nvidia tesla c2075 gpu , the pipeline allows dispersion free , high time resolution data to be obtained in real - time .
we illustrate the significant improvements over the existing incoherent dedispersion system at the gmrt , and present some preliminary results obtained from studies of pulsars using this system , demonstrating its potential as a useful tool for low frequency pulsar observations .
we describe the salient features of our implementation , comparing it with other recently developed real - time coherent dedispersion systems .
this implementation of a real - time coherent dedispersion pipeline for a large , low frequency array instrument like the gmrt , will enable long - term observing programs using coherent dedispersion to be carried out routinely at the observatory .
we also outline the possible improvements for such a pipeline , including prospects for the upgraded gmrt which will have bandwidths about ten times larger than at present . |
the dicke hamlitonian describes a system of @xmath0 two - level atoms interacting with a single monochromatic electromagnetic radiation mode within a cavity .
it is described in the accompanying article @xcite .
the purpose of this second part is to show that the benefits to employ the coherent basis are valid for a large region of the hamiltonian parameter space , not only to obtain converged values of the energy , but also for the wave function , for the ground state and for a significative part of the energy spectra .
it can be particularly useful to study the presence of chaos @xcite and of excited states phase transitions @xcite in this model .
the interaction between a system of @xmath0 two - level atoms and a single mode of a radiation field can be described by the dicke hamiltonian : @xmath1 the frequency of the radiation mode is @xmath2 , which has an associated number operator @xmath3 . for the atomic part @xmath4
is the excitation energy , meanwhile @xmath5 , @xmath6 , @xmath7 , are collective atomic pseudo - spin operators which obey the su(2 ) algebra .
it holds that if @xmath8 is the eigenvalue of @xmath9 , then @xmath10 defines the symmetric atomic subspace which includes the ground state .
the interaction parameter @xmath11 depends principally on the atomic dipolar moment .
we compare the minimal truncation needed to obtain convergence of the solution , using the two basis defined in ref .
@xcite : the coherent basis @xmath12 and the fock basis @xmath13 .
the wave functions , expanded in the truncated fock ( f ) and coherent ( c ) basis are , for a given @xmath14 : @xmath15 where @xmath16 for @xmath17 , and @xmath18 for @xmath19 , and @xmath20 enumerates the eigenstates ordered by their energies @xmath21 with @xmath22 assigned to the ground state .
the probability @xmath23 of having @xmath24 photons in the @xmath25-th state in the fock basis , or @xmath26 of having @xmath27 excitations in the coherent basis is : @xmath28 where @xmath29 for @xmath30 , respectively .
the ground state probability distribution @xmath31 is shown as a function of @xmath24 or @xmath27 up to @xmath32 or @xmath33 , for @xmath34 and @xmath35 , and @xmath36 in figs .
[ fig1 ] and [ fig2 ] .
both wave functions were calculated with the truncation necessary to have the energy converged with @xmath37 . from fig .
[ fig1 ] and [ fig2 ] it is clear that many components which contributes very little to the wave function must be included in the calculations to obtain the desired precision in the ground state energy . it can also be observed in the figures that for @xmath34 , which is @xmath38 in this case , the largest probability is to have no photons in the fock basis , or no excitations in the coherent basis .
the situation is different in the superradiant region , @xmath39 , where in the fock basis the distribution of photons resembles a gaussian curve , with its maximum at a photon number proportional to the number of atoms , while in the coherent basis the probability of having zero excitations remains dominant .
this is the power of the coherent basis , which allows to obtain numerically exact ground state wave functions for numbers of atoms which are intractable in the fock basis . to study the convergence in the wave function we define its precision @xmath40 @xcite as @xmath41 where @xmath29 for @xmath30 , respectively .
this @xmath42 criteria demands less computing resources than the @xmath43 criteria @xcite , because it requires only the information about one truncation value ( @xmath44 ) instead of two .
[ fig3 ] displays the plots of @xmath45 as a function of @xmath32 , and of @xmath46 as a function of @xmath33 . a linear fit , for @xmath47
give us the following relation between @xmath33 and @xmath48 : @xmath49
in this section we extend the analysis to the excited states . to accurately evaluate a significative part of the energy spectrum
is a necessary ingredient in the study of quantum chaos @xcite and of excited state quantum phase transitions ( esqpt ) @xcite . in figure [ fig4 ]
we display plots of @xmath42 as a function of the state @xmath25 , for @xmath47 , @xmath34 , @xmath50 , @xmath51 and @xmath52 . in the upper figures we show the @xmath53 and in the lower ones @xmath48 . on the left
the vertical scale is linear and all states are listed in the horizontal axis , while on the right hand side the vertical scale is logarithmic and only the 150 states with lower energies are included .
the horizontal green line depicts the tolerance @xmath54 . of all states as a function of the state number .
details are given in the text.,title="fig : " ] of all states as a function of the state number .
details are given in the text.,title="fig : " ] of all states as a function of the state number .
details are given in the text.,title="fig : " ] of all states as a function of the state number .
details are given in the text.,title="fig : " ] it is indeed remarkable to observe in fig .
[ fig4 ] that a few hundred states calculated in the coherent basis have their wave function converged , and @xmath55 , for these states , grows in a smooth and nearly monotonous way as a function of the @xmath25 index .
this is not the case in the fock basis , where @xmath56 fluctuates by orders of magnitude between a given state and the following one .
it is worth to compare the convergence criteria based in the wave function and described above , with the more standard convergence in energy , which was described in the previous article @xcite . in figure [ fig5 ]
we show @xmath48 versus @xmath57 for the first 250 excited states , @xmath25 , whose energies converged in the coherent basis with @xmath58 .
vs @xmath57 for the first 250 states , which convergence is under @xmath59 . ]
a linear fit of these data results in @xmath60=0.71077 - 1.10337\,\,log_{10}\left[\delta e_{c}(k)\right ] \\ & \rightarrow \,\,\ , \delta p_{c}=0.19464\,\ , \left[\delta e_{c}(k)\right]^ { 1.10337 } \nonumber\end{aligned}\ ] ] the number of states whose @xmath42 is smaller than a tolerance @xmath54 for @xmath61 and @xmath32 given for the fock basis and @xmath33 for the coherent basis is presented in table [ tab2 : ] .
the two tolerances selected are @xmath62 and @xmath59 , with @xmath34 and @xmath63 .
the advantages associated with the use of the coherent basis are even more clear in this case , because the number of states whose wave function has converged with the selected tolerance is larger than those whose energies have converged .
it should be mentioned , however , that the tolerances in @xmath42 are absolute , because its best case value of a fully converged state is zero , and the worst situation , for completely different wave functions , is one . on the contrary ,
the energy scale is arbitrary , and can have positive and negative values , even some levels with energies very close to zero .
it makes the use of the relative error employed in ref @xcite dangerous when the reference energy is very small .
but for excited states a fixed value of @xmath54 implies the need of more precise digits in the calculated energy , making more difficult the convergence for higher energies .
for this reason our @xmath43 criteria is more stringent than the @xmath42 one : every excited state with converged energy has guaranteed the convergence of its wave function . as the coherent basis provides many converged states with a single truncation value , it is promising to study the presence of excited states quantum phase transitions ( esqpt ) , predicted in dicke - like systems and spin systems for @xmath11 values deeply in the superradiant phase @xcite .
to obtain the eigenvalues and eigenvectors of the dicke hamiltonian for a finite number of atoms it is necessary to perform a numerical diagonalization , employing a truncated boson number space .
two basis , associated with the two integrable limits of the hamiltonian , are used along this work . in the present article
we have shown that , in most of the hamiltonian s parameter regions including the qpt , the coherent basis requires a significative smaller truncation .
we extended the analysis to the convergence in the wave function , exhibiting both convergence criteria as equivalent , and presented the numerical relationships between them .
the study of the probability distributions of the number of bosons was helpful in understanding the differences between the two basis , and the advantages of the coherent basis .
the convergence of the energies and the wave functions was also investigated for the excited states , showing that the coherent basis is very powerful also in this case , allowing to obtain hundreds of converged states with a single truncation value .
this findings can be very useful in order to observe the presence of quantum chaos around the phase transition , as well as to study the excited states quantum phase transitions .
we thank o. castaos , r. lpez - pea and e. nahmad for many useful and interesting conversations.this work was partially supported by conacyt - mxico and papiit - unam 102811 .
99 chen q h , zhang y y , liu t and wang k l 2008 _ phys . rev .
a _ * 78 * 051801 ; liu t , zhang y y , chen q h and wang k l 2009 _ phys .
rev . a _ * 80 * 165308 bastarrachea - magnani m a and hirsch j g , this proceedings .
emary c and brandes t 2003 _ phys .
e _ * 67 * 066203 ; 2003 _ phys .
lett . _ * 90 *
044101 lambert n , emary c and brandes t 2005 _ phys .
lett . _ * 92 * 073602 prez - fernndez p , cejnar p , arias j m , dukelsky j , garca - ramos j e , and relao a 2011 _ phys .
a _ * 83 * 033802 prez - fernndez p , relao a , arias j m , cejnar p , dukelsky j , and garca - ramos j e 2011 _ phys . rev
* 83 * 046208 bastarrachea - magnani m a and hirsch j g 2011 _ rev .
s _ * 57 * 0069 ; _ ibid _ 2012 _ aip conf .
proc . _ * 1488 * 418 ; _ ibid _ 2013 _ aip conf . | an extended bosonic coherent basis has been shown by chen _ et al _ @xcite to provide numerically exact solutions of the finite - size dicke model .
the advantages in employing this basis , as compared with the photon number ( fock ) basis , are exhibited to be valid for a large region of the hamiltonian parameter space and many excited states by analyzing the convergence in the wave functions .
pacs numbers : 3.65.fd , 42.50.ct , 64.70.tg |
the existence of a critical point in the phase diagram of qcd , for nonzero baryonic density , is of fundamental significance for our understanding of strong interactions and so its experimental verification has become an issue of high priority @xcite . for this purpose an extensive programme of event - by - event searches for critical fluctuations in the pion sector
is in progress in experiments with heavy ions from sps to rhic energies @xcite . in @xcite
we have emphasized , however , that in order to reveal critical density fluctuations in multiparticle environment , one has to look for unconventional properties in the momentum distribution of reconstructed dipions ( @xmath2-pairs ) with invariant mass just above the two - pion threshold .
in fact , the qcd critical point , if it exists , communicates with a zero mass scalar field ( @xmath0-field ) which at lower temperatures ( @xmath3 ) may reach the two - pion threshold and decay in very short time scales owing to the fact that its coupling to the two - pion system is strong .
obviously , the fundamental , underlying pattern of @xmath0-field fluctuations , built - up near the critical point by the universal critical exponents of qcd @xcite , is phenomenologically within reach if and only if the study of correlated sigmas , reconstructed near the two - pion threshold , becomes feasible . in the present work
we perform a detailed feasibility study of the observables related to the detection of the qcd critical point in nuclear collisions . in order to proceed we
summarize , first , the principles on which the behaviour of a critical system of sigmas is based @xcite .
\(a ) the geometrical structure of the critical system in transverse space ( after integrating in rapidity ) consists of @xmath0-clusters with a fractal dimension @xmath4 leading to a power law , @xmath5 , for the @xmath0-field fluctuations , within each cluster ( @xmath6 : isotherm critical exponent ) \(b ) in transverse momentum space the @xmath0-fluctuations obey a power law @xmath7 leading to observable intermittent behaviour of factorial moments : @xmath8 where @xmath9 is the number of @xmath10 cells @xcite .
\(c ) the density fluctuations of dipions ( @xmath11-pairs ) with invariant mass close to two - pion threshold @xmath12 incorporate the sigma - field fluctuations , @xmath13 , at the critical point , under the assumption that the sigma mass reaches the two - pion threshold @xmath14 in a time scale shorter than the relaxation time of critical fluctuations .
\(d ) the qcd critical point belongs to the universality class of the @xmath15 ising system in which @xmath16 . on the basis of these principles and the fact that critical clusters in the above universality class interact weakly @xcite
, one may construct a monte - carlo generator ( critical monte - carlo : cmc ) able to simulate events of critical sigmas , correlated according to the above prescription @xcite .
then , if the corresponding invariant mass distribution is known , we can also simulate the decay of the critical sigmas to pions which are experimentally observable . to complete the numerical experiment we use the momenta of the charged pions , obtained from the decay of the critical sigmas , to form , in an event - by - event basis , neutral dipions ( @xmath2 pairs ) and to look for fingerprints of the original critical sigma fluctuations in their momentum distribution . to check the applicability of our approach
we perform also a comparative study between the revealed critical correlations - fluctuations in cmc and the corresponding behaviour of a conventional monte - carlo ( hijing ) .
the input parameters of the simulation are the size of the system in rapidity ( @xmath17 ) and transverse space ( @xmath18 ) as well as the proper time scale ( @xmath19 ) characteristic for the formation of the critical system . in
what follows we will use exclusively @xmath20 for the rapidity size , adapted to the sps - na49 conditions ( @xmath21 ) .
the parameters @xmath18 , @xmath19 can then be tuned in order to fit the mean charged pion multiplicity of systems of different size ( @xmath22 , @xmath23 and @xmath24 ) studied experimentally at this energy @xcite . for each choice of input parameters
we produce , using the cmc generator , a large set of critical sigma events . in order to generate the corresponding pion sector we assign to the sigmas the invariant mass distribution @xmath25 and
then we let them decay into pions with a branching ratio @xmath26 for neutral to charged .
the choice of @xmath25 is determined by the requirement that the resulting inclusive neutral dipion invariant mass distribution @xmath27 resembles to a large extent the corresponding distribution obtained in experiments with heavy ion collisions at high energies . in order to perform our analysis in terms of observable quantities
we choose the mean multiplicity of positive charged pions per event @xmath28 as the basic parameter ( instead of @xmath18 , @xmath19 ) characterizing different @xmath29 systems .
we note that within the cmc approach the property @xmath30 is exactly fullfiled .
we will investigate three different cases : low ( @xmath31 ) , intermediate ( @xmath32 ) and high ( @xmath33 ) positive charged pion multiplicity resembling the @xmath34 , @xmath35 and @xmath36 system at @xmath37 respectively .
first we consider the case of low pion multiplicity . before going on with the analysis it is worth to emphasize that , in general , the determination of the critical sigma sector using the observed momenta of charged pions , is a very difficult task .
this is due to the absence of a characteristic pattern in the inclusive dipion invariant mass distribution attributed to the presence of critical sigmas . to illustrate this property we show in fig . 1 the inclusive distribution of the sigma invariant mass calculated before the decay of the sigmas into pions ( solid line ) as well as the corresponding distribution for neutral dipions recontsructed from the final pion momenta ( full circles ) for @xmath38 cmc events with @xmath39 ( dataset i )
. both distributions are equally normalized .
apart from a peak at @xmath40 due to kinematics there is no other structure in the invariant mass profile .
therefore the detection of the critical sigma sector has to go through the , more subtle , study of density fluctuations in momentum space .
the critical sigma invariant mass distribution @xmath25 ( solid line ) as well as the corresponding neutral dipion distribution @xmath27 ( full circles ) for @xmath38 cmc events with @xmath39 . ] in order to reveal the underlying critical fluctuations , at the level of observation , one has first to perform factorial moment analysis in small cells of the momentum space @xcite .
we have chosen transverse momenta for this analysis in order to avoid additional assumptions about the role of longitudinal rapidity in the description of the statistical mechanics of the system @xcite . applying factorial moment analysis to the transverse momenta @xmath41 of the negative pions in the sample of the @xmath38 critical events we obtain for the second moment a weak intermittency effect : @xmath42 with @xmath43 much smaller than the expected to occur in the critical system ( @xmath44 ) .
we note with @xmath45 the number of bins in each momentum space component .
the corresponding factorial moment for the sigmas , before their decay , follows closely the theoretical prediction : @xmath46 .
this behaviour is displayed in fig .
the second factorial moment @xmath47 in transverse momentum space of both the negative pions as well as the critical sigmas .
we use the same dataset as in fig . 1 .
] the reason for the suppression of fluctuations in the pionic sector is the kinematical distortion of the self - similar pattern formed in the sigma sector due to the sigma - decay .
the strength of this distortion increases with momentum transfer : @xmath48 and becomes negligible near the two - pion threshold ( @xmath49 ) . here
@xmath50 where @xmath51 are the four momenta of the charged pions produced through the sigma - decay .
thus the search for critical fluctuations is based on an accurate reconstruction of the momenta of the decaying sigmas . in particular @xmath52 pairs with invariant mass close to the two - pion threshold
are the best candidates to carry potentially the geometrical features of the critical isoscalars .
the great advantage of our approach is that it allows a self - consistency test for the reconstruction of the sigma sector at the level of density fluctuations due to our exact knowledge of the sigma momenta and the corresponding fluctuation pattern .
in practice the reconstruction of the critical momentum fluctuations is performed by looking , event by event , for @xmath52 pairs fullfiling the criterion a : @xmath53 with @xmath54 .
the momentum of the corresponding neutral dipion is then obtained as : @xmath55 . in order to ensure that * all * the available critical sigmas within the above kinematical region are recovered in the reconstruction we have to use * full * pairing forming * all * possible pairs @xmath52 fullfiling ( [ eq : eq1 ] ) for a given @xmath56 .
however the full pairing introduces as a side effect a combinatorial background which has to be treated appropriately .
we will come back to this point later on .
thus for each value of @xmath57 we obtain a set of events including dipion momenta .
we perform factorial moment analysis in transverse momentum space for each such dataset . according to the previous discussion , for decreasing values of @xmath57 the fluctuations measured by the intermittency exponent @xmath58 of the corresponding factorial moment should increase leading to @xmath59 as @xmath60 . in fig .
3a we show the second moment in transverse momentum space obtained from data sets of reconstructed dipions for three different values of @xmath57 ( @xmath61 and @xmath62 ) .
we observe an increase of the slope @xmath58 with decreasing @xmath57 .
however this increase has two different origins : ( i ) the presence of nontrivial fluctuations of dynamical origin and ( ii ) kinematically induced fluctuations through the constraint ( [ eq : eq1 ] ) which become of the same order as the dynamical ones for @xmath63 .
it must be noted that the functional forms of @xmath47 for the different @xmath57 values are not exact power - laws and the exponent @xmath58 is an effective one .
increasing @xmath57 the deviations from a power - law description increase too . in order to control the kinematical fluctuations we form
datasets with mixed events through appropriate shuffling of the momenta in the original cmc events .
then we calculate the second factorial moments for the datasets consisting of mixed events .
we show in fig .
3b the results of this analysis .
also in the case of mixed events we observe a clear increase of the effective slope @xmath64 is used to indicate mixed events ] for decreasing @xmath57 . due to the absence of dynamical fluctuations in the mixed events it is natural to conclude that the observed behaviour in @xmath65 is attributed to the kinematical fluctuations .
( a ) the second factorial moment @xmath47 in transverse momentum space of reconstructed dipions for @xmath66 and @xmath62 ( log - log plot ) .
we use the same 30000 cmc events as in fig . 1 .
in ( b ) we show for comparison the corresponding factorial moments for datasets consisting of mixed events . ]
these can be removed if we introduce the difference @xmath67 between the factorial moments of real and mixed events ( for any order @xmath68 ) . for @xmath69
the quantity @xmath70 represents the correlator of the corresponding dipions .
thus in the difference @xmath71 the combinatorial background representing the uncorrelated part of the reconstructed dipions is suppressed and the critical fluctuations associated with the correlated part of the dipion sector are recovered .
we illustrate how this subtraction succeeds in practice in fig . 4 where we present @xmath70 for @xmath72 and @xmath39 .
although the moments @xmath47 and @xmath65 are not exact power - laws , as indicated above , the difference @xmath70 is significantly better fitted by a power - law and the corresponding slope @xmath73 is very close to the slope obtained from the dataset of the sigmas before their decay ( see fig . 1 ) .
( a ) the log - log plot of the correlator @xmath70 in transverse momentum space of reconstructed dipions for @xmath72 obtained using 30000 cmc events ( the same as in fig . 1 ) and the corresponding mixed events .
] the fluctuations in the cmc events , as described by @xmath70 , are induced by strongly correlated critical qcd dynamics and are expected to be at least one order of magnitude greater than fluctuations originating from conventional hadronic dynamics . to check the validity of this statement we have calculated @xmath70 using @xmath74 events obtained from the hijing monte - carlo generator @xcite simulating the sps @xmath34 system at @xmath37 and involving only noncritical qcd dynamics . in figs .
5a - d we show the results of our calculations for four different values of @xmath57 . for comparison we plot in the same figure the correlator for the @xmath38 cmc events with @xmath39 .
we observe that for all values of @xmath57 the correlator @xmath70 for the hijing system remains flat fluctuating around zero , while for the cmc events as @xmath57 decreases the characteristic , for the critical system , power - law behaviour is setting on according to the previous discussions .
the correlator @xmath70 for the cmc and the hijing system using four different values of @xmath57 : ( a ) 5 mev , ( b ) 20 mev , ( c ) 50 mev and ( d ) 120 mev . ]
it is worth investigating in more detail the role of the parameter @xmath57 in our analysis .
practically the desired limit @xmath75 is not accessible for several reasons .
first of all looking at fig
. 1 we observe that in this limit the number of initial critical sigmas as well as the number of reconstructed dipions vanishes .
we practically need infinite statistics in order to extend our analysis in this kinematical region .
in addition , in the range of very small @xmath57 values the fluctuations induced by the kinematical constraint ( [ eq : eq1 ] ) become of the same order as the dynamical fluctuations due to the critical sigmas .
our entire treatment is based on an accurate cancellation of the kinematical fluctuations in the difference @xmath70 and this requires again very high statistics if @xmath63 .
in fact , using @xmath57 values smaller than @xmath76 in the analysis of the @xmath38 cmc events considered so far , the statistical errors increase rapidly and prohibit a reliable reveal of the critical fluctuations .
on the other hand , if we increase @xmath57 , the combinatorial background due to the reconstructed dipions is increasing rapidly and the relative measure of the critical sector goes to zero .
therefore we must search for an optimal region of @xmath57 values to perform our analysis . in order to achieve a more quantitative criterion for the determination of the appropriate region of @xmath57-values we consider more carefully the sector of reconstructed dipions . in the cmc case
the reconstructed dipions are divided into two subsets : the set of real sigmas and the set of fake sigmas .
let us denote by @xmath77 the mean number of real sigmas per event with invariant mass in the kinematical window ( [ eq : eq1 ] ) for a given value of @xmath57 . within our approach
we investigate only the part of the sigmas which decays into a pair of opposite charged pions .
therefore we have : @xmath78 where @xmath79 is the mean number of positive ( negative ) charged pions per event produced through the decay of the real sigmas with invariant mass in the region ( [ eq : eq1 ] ) .
obviously to a good approximation the corresponding number of fake sigmas is given by : @xmath80 the first term on the right hand side of eq.([eq : eq3 ] ) is dominated by uncorrelated @xmath52 pairs .
therefore we can write : @xmath81 und using the property : @xmath82
we finally obtain : @xmath83 a natural constraint ensuring the dominance of the real sigmas over the fake ones is to use in our analysis @xmath57 values fullfiling the condition : @xmath84 which simplifies to : @xmath85 .
this is fullfiled for example for the value @xmath72 which we have used so far in our analysis since in this case @xmath86 .
we also observe that @xmath87 for any value of @xmath57 , where @xmath88 is the mean number of dipions per event obtained through the reconstruction in the kinematical window ( [ eq : eq1 ] ) .
the upper bound in @xmath89 corresponds to an upper bound for @xmath57 .
the lower bound both for @xmath57 as well as @xmath89 is determined by the statistics according to the discussion in the previous paragraph . for practical purposes one applies the stronger bound @xmath90 for the estimation of the appropriate region of @xmath57-values to be used in the data analysis .
the great advantage of restricting @xmath88 instead of @xmath57 becomes more clear when we compare the analysis in cmc datasets simulating @xmath29 processes of different size . in this case
the dependence of @xmath88 on @xmath57 varies from system to system . to illustrate this
we have produced two additional cmc datasets , with @xmath91 ( dataset ii ) and @xmath92 ( dataset iii ) respectively , each consisting of @xmath38 events . in fig .
6 we show the function @xmath88 for the cmc datasets i , ii and iii .
the horizontal lines correspond to the values @xmath93 , @xmath94 and @xmath95 .
the function @xmath96 for the three different cmc datasets described in text . ] for increasing size of the system ( increasing @xmath28 ) the @xmath57 values leading to a given @xmath88 decrease .
the complete analysis within our approach for these datasets involves the reconstruction of the isoscalar sector as well as the calculation of the corresponding correlator @xmath70 .
however a comparative study between the different datasets in terms of factorial moments is possible only for classes of events characterized by almost the same multiplicity since in the opposite case artificial fluctuations are induced @xcite .
thus , to compare the systems of different size , we have to calculate the correlator for fixed multiplicity of reconstructed dipions @xmath97 choosing a suitable value of @xmath57 in each case . as in the case of dataset
i we use the effective slope ( or critical index ) @xmath98 of the correlator ( @xmath99 ) as a measure of the fluctuations in the corresponding dataset . in fig .
7 we show the results of our calculations for the three datasets i , ii and iii using @xmath100 ( full circles ) , @xmath101 ( open stars ) and @xmath102 ( crosses ) .
we plot @xmath98 as a function of the mean number of positive pions per event @xmath28 respectively .
it is clearly seen in fig .
7 that the critical fluctuations , within the limitations of our analysis due to statistics , are at best recovered for cmc systems involving medium to small size nuclei and using in the analysis @xmath57 values fullfiling the constraint @xmath103 .
the critical index @xmath98 approaches the theoretically expected value @xmath104 .
in fact we see that for @xmath102 and @xmath39 we obtain exactly the critical qcd prediction .
this is the main result of the present work providing us with a useful guide for the search of qcd critical fluctuations in relativistic ion collisions .
the slope @xmath98 as a function of @xmath28 for the cmc datasets i , ii and iii using @xmath93 ( full circles ) , @xmath94 ( open stars ) and @xmath95 ( crosses ) . ]
in conclusion , we have shown that a set of well prescribed observables ( factorial moments , correlators , intermittency exponents ) associated with the existence of a critical point in quark matter , can be established in nuclear collisions .
these observables belong to the reconstructed isoscalar sector describing massive dipions ( @xmath2 pairs ) near the two - pion threshold and their behaviour reveals strong critical effects suggested by @xmath0-field fluctuations near the critical point .
we claim that a search for such a critical behaviour in heavy ion experiments is feasible within the framework of a reconstruction procedure of the momenta of massive dipions , discussed in this work .
the critical effects in this sector although independent of the system size can be at best recovered , through the proposed reconstruction algorithm , in collisions of relatively small nuclei . the appropriate kinematical window to look for these effects
is also determined .
our study has also shown that although it is not possible to reveal any conventional sign of the sigma itself in experiments with nuclei at high energies , nevertheless its density fluctuations associated with the critical point are observable and can be measured and studied in a systematic way .
therefore our proposal is to study , using the above observables , different processes at the sps and rhic with the aim to scan a substantial area of the phase diagram , in a systematic search for the qcd critical point in collisions of nuclei .
we thank the na49 collaboration and especially k. perl and r. korus for their help in the treatment of the hijing data for the @xmath34 system .
one of us ( n.g . a. ) wishes to thank larry mclerran and dmitri kharzeev for the kind hospitality at the brookhaven national laboratory .
this work was supported in part by the epeaek research funding program pythagoras ( 70/3/7315 ) and also by the research funding program of the university of athens `` kapodistrias '' .
f. wilczek , hep - ph/0003183 ; m. stephanov , k. rajagopal , e. shuryak , phys . rev . lett . * 81 * , 4816 ( 1998 ) ; j. berges and k. rajagopal , nucl . phys . *
b538 * , 215 ( 1999 ) ; j. berges , n. tetradis , c. wetterich , phys . rep . *
363 * , 223 ( 2002 ) ; z. fodor and s.d .
katz , jhep * 0203:014 * , 2002 ; m.a .
stephanov , prog .
. phys . suppl . *
153 * , 139 ( 2004 ) ; hep - ph/0402115 . | a detailed study of correlated scalars , produced in collisions of nuclei and associated with the @xmath0-field fluctuations , @xmath1 , at the qcd critical point ( critical fluctuations ) , is performed on the basis of a critical event generator ( critical monte - carlo ) developed in our previous work .
the aim of this analysis is to reveal suitable observables of critical qcd in the multiparticle environment of simulated events and select appropriate signatures of the critical point , associated with new and strong effects in nuclear collisions . |
the principal roadblock impeding progress in disentangling the physics of the cuprate superconductors is arguably our inability to access quantum ground states of two - dimensional itinerant electrons which are qualitatively distinct from a landau fermi liquid . overcoming this obstruction is of paramount importance , indispensable in explaining the strange metal behavior observed near optimal doping and a likely requisite to account for the emergent pseudo - gap at lower energies .
a putative underlying paramagnetic mott insulator provides the scaffolding for a popular class of theories , which view the pseudo - gap as a lightly doped spin liquid .
@xcite significant progress has been made in developing the groundwork and there now exists a well established theoretical framework to describe a myriad of distinct spin liquids . for the cuprates
the most promising spin liquids are described in terms of fermionic spinons minimally coupled to a compact @xmath1 gauge field and moving in various background fluxes . upon doping formidable challenges
bosonic holons carrying the electron charge become mobile carriers and lead to electrical conduction . but at low temperatures bose condensation appears inevitable and this leads to fermi liquid behavior , either a metal with conventional landau quasiparticles or a bcs superconductor if the spinons are paired . accessing a pseudo - gap or a strange metal which conduct electricity despite the absence of long - lived landau quasiparticles requires doped holons that form an uncondensed quantum bose fluid rather than a condensed superfluid . but is this possible , even in principle ?
if possible , what properties would such a putative `` bose metal '' exhibit?@xcite and what theoretical framework is appropriate ?
the slave - particle gauge theory approach has been stymied by this stumbling block for over 15 years . in this paper
we provide ( some ) answers to these questions by constructing explicit examples of such unusual phases of bosons that may offer some routes out of the conundrum . our goal ,
then , is to access and explore uncondensed quantum phases of 2d bosons which are conducting fluids but not superfluids .
specifically we have in mind hard core bosons moving on a 2d square lattice , but seek to construct states which do not require particular commensurate densities . while our construction can be implemented for bosons moving in the 2d continuum euclidean plane , the states will only possess the reduced point group symmetry of the square lattice . despite our interest in time reversal invariant quantum ground states ,
our technical approach will be strongly informed by theories of the fractional quantum hall effect ( fqhe ) . in some regards ,
the quantum phases that we construct are time reversal invariant analogs of the laughlin state for bosons in a half - filled landau level .
but the physical properties of the phases will be dramatically different from the incompressible fqhe states , and will have gapless excitations and metallic transport for example . to motivate and illustrate our approach , it will be helpful to briefly revisit the bosonic fqhe .
consider the laughlin wavefunction for bosons in a half - filled landau level , @xmath2 much of the physics of the laughlin state has its origin in the structure of zeroes , which reveals that any two particles upon close approach in real space are in a relative two - body @xmath3 state , @xmath4 .
our first objective is to construct a time reversal invariant ( real ) wavefunction in which particle pairs are similarly in a relative @xmath5 state .
a clue is offered by noting that the laughlin wavefunction is the square of a vandermonde determinant , @xmath6 . in the vandermonde determinant ,
all pairs of particles are in a relative @xmath7 state , @xmath8 . upon squaring ,
the two @xmath7 states combine to form a single @xmath3 state , which is essentially just addition of angular momentum .
this suggests constructing a time reversal invariant boson wavefunction in zero magnetic field by simply squaring a determinant constructed from momentum states within a fermi sea , @xmath9 let s examine the nodal structure .
the generic behavior of each fermion determinant when any two particles are taken close together is dictated by fermi statistics and reality of the wavefunction and has the functional form @xmath10 .
the unit vector @xmath11 will depend in a complicated way on the location of all the other particles ( see ref . for a discussion and illustrations of the free fermion nodes ) .
when @xmath12 , this is a @xmath13 form vanishing along a nodal line ( in the relative coordinate ) parallel to the @xmath14-axis .
unfortunately , in contrast with the laughlin case , squaring the determinant leads here to an extended " @xmath15-wave form with a quadratic nodal line rather than the desired @xmath16-wave . but consider instead multiplying together two fermion determinants , each constructed by filling up a fermi sea of momentum states , but with different fermi surfaces . a wavefunction with @xmath5 two - particle correlations
can be constructed by choosing two elliptical fermi surfaces , one with its long axis along the @xmath14-axis and the other rotated by 90 degrees , as illustrated in fig .
[ fig : introfs ] , @xmath17 where the short - hands @xmath18 and @xmath19 represent the corresponding slater determinants . in the limit of extreme eccentricity of the ellipses , the wavefunction will have the desired @xmath5 form when two particles are brought close together , @xmath20 . away from this limit
, the two nodal lines will not align precisely with the @xmath14 and @xmath21 axes , but upon taking one particle around the other the wavefunction will exhibit the same sign structure as a @xmath5 orbital , @xmath22 , changing sign twice . a picture of such sign structure as seen by a test particle is shown in fig .
[ fig : nodes ] . ) is obtained by multiplying two slater determinants corresponding to distinct but symmetry related fermi surfaces , which can be viewed as fermi surfaces of @xmath23 and @xmath24 slave fermions .
the left panel shows an example with closed elliptical fermi surfaces , while the right panel is a case with open fermi surfaces in the first brillouin zone of the square lattice . ]
-2 mm pattern upon encircling a target particle .
, width=288 ] -2 mm our principal thesis is that this wavefunction captures qualitative features of a new quantum liquid phase of bosons , which we call `` d - wave - correlated bose liquid '' ( dbl ) . but
a variational wavefunction does not provide a complete characterization of a quantum phase , and can not be used to address its stability .
as for the fqhe , a field theoretic approach , such as chern - simons gauge theory , is both desirable and ultimately necessary .
since we require time reversal invariance , chern - simons is inappropriate , but a tractable field theoretic framework for the dbl will nevertheless turn out to be a gauge theory . indeed , we follow closely the gauge theory approaches to spin liquids in quantum antiferromagnets,@xcite which after all are lattice bosonic systems .
but there are some notable and important differences for the dbl which we discuss below . the variational wavefunction in eq . ( [ psi_dbl ] )
, being a product of two fermion determinants , naturally suggests expressing the boson creation operator as a product of two fermion operators , @xmath25 with the 2d position @xmath26 either continuous or denoting the discrete sites of a square lattice . in a general case
, this decomposition introduces a local @xmath27 gauge redundancy well - known in the slave - fermion treatments of the spin-1/2 antiferromagnet ( see ref . for a recent comprehensive discussion ) .
the s - type wavefunction corresponds to a so - called @xmath27 liquid , while the dbl wavefunctions are @xmath1 liquids and require a @xmath1 gauge field which is minimally coupled to the two fermions with opposite gauge charges ( the product @xmath28 is then a gauge neutral composite , the physical boson ) . within slave particle theory a mean field state with a fixed background gauge
magnetic " flux is chosen .
this choice is constrained by symmetry : all the physical symmetries of the boson model , even if not respected by the mean field ansatz , must be present in the full gauge theory once gauge transformations are allowed .
the collection of these symmetry transformations in the gauge theory is called the projective symmetry group ( psg).@xcite for lattice boson theories many different mean field states with different psg s are possible , while little is offered to guide which mean field state is the correct one . in the continuum ,
the symmetries of the euclidean plane together with time reversal invariance are more restrictive and we can only offer two psg mean field ansatze in this case .
one is a mean field state with zero gauge flux and spherical fermi seas for both fermions ; this is an @xmath27 ansatz and corresponds to the s - type wavefunction considered earlier .
a different @xmath1 ansatz is obtained by considering @xmath23 and @xmath24 fermions each moving in a uniform field but of opposite signs for the two ; in this case , the boson wavefunction is schematically @xmath29 , e.g. , @xmath30 to give a concrete example ; this wavefunction is again non - negative and has off - diagonal quasi - long - range order.@xcite for bosons moving in continuous space but with only the square lattice point group symmetry , we can offer only four psg mean fields ; all of them have zero flux , but differ in the way the 90 degree rotation and mirror symmetries are realized .
one example has both @xmath23 and @xmath24 fermi surfaces individually respecting the symmetries but otherwise independent of each other .
two more examples are the already introduced d@xmath31 state , eq .
( [ psi_dbl ] ) and fig .
[ fig : introfs ] , with elliptical fermi surfaces elongated along the @xmath14 and @xmath21 axes , and a similarly constructed d@xmath32 state with ellipses along @xmath33 ; in both cases , the @xmath23 and @xmath24 fermi surfaces are related by the 90 degree rotation , but the two states differ in the way the mirror symmetries are realized . in the fourth
example , each fermi surface is invariant under the rotation but not under the mirrors , which instead are realized by interchanging the two fermions . each of the four presented ansatze can be loosely called a d - wave bose liquid as far as the nodal pictures like that in fig .
[ fig : nodes ] are concerned ; the gauge theory analysis would also be rather similar . from now on ,
we will focus on the d@xmath31 boson liquid .
the associated mean field wavefunction once gutzwiller projected into the physical hilbert space , @xmath34 , is precisely the variational wavefunction in eq .
( [ psi_dbl ] ) . a full incorporation of gauge fluctuations about the mean field performs , in principle , this projection exactly .
but in practice the gauge theory approach allows incorporation of slowly varying gauge fluxes and is different in this respect from the wavefunction .
provided the gauge theory is not in a confined state and no physical symmetries are broken , an exact description of the putative dbl phase is thereby obtained .
herein , we introduce a large @xmath35 generalization of the gauge theory which allows a controlled and systematic treatment of the gauge fluctuations order by order in powers of @xmath36 .
we can then perturbatively extract physical properties of the dbl phase . whether the dbl phase survives down to @xmath37 is not something that we can reliably address .
in describing the d@xmath31 liquid phase for lattice bosons there is considerable freedom when choosing the precise form of the slave fermion hopping in the mean field hamiltonian .
the simplest choice is near neighbor hopping with different amplitudes in the @xmath14 and @xmath21 directions .
but even with this restriction , there are two different possible fermi surface topologies , closed or open , as depicted in fig .
[ fig : introfs ] .
the fermi surface topology has rather dramatic consequences for the nature of the associated d - wave bose fluid .
when the fermi surfaces are open and have no parallel fermi surface tangents , the resulting phase has a quasi - local character the boson green s function is found to fall off exponentially in space .
this is a distinct quantum phase which we refer to as a d - wave local bose liquid ( dlbl ) .
the dlbl is intrinsically very stable against gauge fluctuations and we are fairly confident that when present it will be a stable quantum phase .
the effect of gauge fluctuations on the dbl phase , however , are more subtle . within our systematic large @xmath35 approach
we do find a regime where the dbl is stable , but it is less clear that this will survive down to @xmath37 .
the gutzwiller wavefunction does appear to describe a dbl phase with properties that are consistent with those inferred from the gauge theory .
this provides some support for the stability of the dbl phase .
together the gauge theory and the variational wavefunction provide a consistent and rather complete picture of both the dbl and the dlbl . here
we briefly highlight some of the important characteristics . in the dbl
, the single particle boson green s function , @xmath38 , decays as an oscillatory power law at equal times , @xmath39 } { |{\bf r}|^{4-\eta } } \nonumber \\ & + & \frac{\cos[({\bf k}_{f_1 } + { \bf k}_{f_2 } ) \cdot { \bf r } - 3\pi/2 ] } { |{\bf r}|^4 } ~ , \label{gb_final}\end{aligned}\ ] ] where the two wave vectors @xmath40 depend on the observation direction @xmath41 , as does the anomalous exponent @xmath42 .
as the direction of @xmath41 is rotated in real space , the wave vectors @xmath40 and @xmath43 trace out closed momentum space curves , and for the dbl in the continuum these curves will have the topology of a circle .
if one measures the boson green s function and finds it to be of the above form , it is natural to refer to the @xmath43 as `` bose surfaces . '' within the gauge theory description the origin of these singular surfaces can be traced to the fermi surfaces of the constituent fermions .
specifically , @xmath40 are locations on the two fermi surfaces where the surface normals are along the observation direction @xmath41 , and the momentum space areas enclosed by each of these surfaces will equal @xmath44 . note that the @xmath40 surfaces can be uniquely reconstructed from the measured bose surfaces @xmath43 . in this sense ,
the bose surfaces in the dbl contain information analogous to luttinger s fermi surface volume theorem in the fermi liquid .
the bose surface can also be extracted directly from the dbl wavefunction by using variational monte carlo to compute the equal time boson green s function , and can also be indirectly inferred by using the shape of the two fermi surfaces input into the @xmath18 and @xmath19 factors .
it is both reassuring and quite remarkable that these two coincide .
the spatially local green s function , @xmath45 , in the dbl falls off in time as a power law which is at least as slow as @xmath46 corresponding to the local boson tunneling density of states @xmath47 ; these are mean field results obtained by combining two fermions each with a finite density of states , but we suspect the time decay may actually be slower upon including gauge fluctuations .
the behavior of the boson green s function in the dlbl is quite different .
the time decay at @xmath48 is a power law , @xmath49 , which is an exact result insensitive to lattice scale details .
but the equal time green s function falls off exponentially in space in the dlbl , @xmath50 . despite this , a two boson box correlator , @xmath51 falls off as a ( non - oscillatory ) power law at large distances in the dlbl , @xmath52 , with both the sign ( negative ) and the exponent being universal and insensitive to lattice scale physics .
paradoxically , this seems to imply that a pair of bosons injected into the dlbl on opposite corners of a square box can move more readily than a single injected boson .
but in a strongly interacting quantum state such a single - particle interpretation can be misleading the dynamics of an injected boson is not that of a weakly interacting quasiparticle . in both
the dbl and the dlbl the density - density correlator @xmath53 behaves at large distances as , @xmath54 } { |{\bf r}|^{4-\gamma_\alpha } } + \frac{1}{|{\bf r}|^4 } ~ , \label{db_final}\ ] ] where again both the wave vectors @xmath40 and the anomalous exponent @xmath55 depend on the observation direction @xmath41 .
measurement of such correlation can be used to extract @xmath40 in the dlbl , which are not accessible from the boson green s function here . upon rotating the unit vector @xmath41 in real space ,
@xmath40 will again trace out the fermi surfaces of the underlying fermions which for the dlbl will be open . both the dbl and the dlbl are conductors with a resistance varying with temperature as @xmath56 .
this is particularly striking for the dlbl where the boson green s function is short - ranged .
evidently , it is not possible to understand the metallic transport in terms of the motion of weakly interacting bosons . in a sense it is the fermionic constituents which transport the charge , but even this is not quite right since the @xmath57 fermions do not exist as well defined quasiparticle excitations , being strongly scattered by the gauge fluctuations .
the lattice gauge theory formulation can be used to motivate a lattice boson hamiltonian which might plausibly have the dbl / dlbl as a ground state .
the simplest such model consists of a near neighbor boson hopping term supplemented by a four - site ring " exchange term : @xmath58 with @xmath59 .
this model is fully specified by two dimensionless numbers , the ratio @xmath60 and the boson filling @xmath61 .
this ring hamiltonian with @xmath62 and @xmath63 was introduced and studied by paramekanti,@xcite and later studied extensively with quantum monte carlo by sandvik,@xcite melko,@xcite and rousseau@xcite there is no sign problem in this regime , and it was possible to access large system sizes and low temperatures .
but for @xmath62 and @xmath64 , there is a sign problem since the hamiltonian does not satisfy the marshall sign conditions , and one expects the ground state wavefunction to take both positive and negative values . for this ring model
we have evaluated the variational energetics for the dbl / dlbl wavefunctions and compared these to the energy of a superfluid wavefunction of the usual jastrow form . within this necessarily limited energetics study
we do find a region of parameter space with small @xmath65 and near half filling where the dlbl wavefunction has the lowest energy . in view of its quasi - local character
, we suspect that dmrg studies could be fruitful in helping establish whether or not the dlbl is present in the phase diagram of this ( or related ) ring hamiltonian .
the paper is organized as follows . in sec .
[ sec : motiv_wavefnc ] we discuss in more detail the wavefunction motivation for considering the dbl states . in sec .
[ sec : mft ] we introduce the lattice gauge theory description and study properties of the dbl state in the mean field theory that ignores gauge fluctuations .
this provides an initial guide to the singular ( bose ) surfaces , and is followed in sec .
[ sec : wavefnc_props ] with numerical characterization of the properties of the actual dbl wavefunctions . in sec .
[ sec : gauge_fluct ] , we consider the full gauge theory description , focusing on the effects of the gauge fluctuations on the singularities across the bose surfaces , and in particular obtain the long - distance properties such as eq .
( [ gb_final ] ) and eq .
( [ db_final ] ) to order @xmath36 in our systematic large @xmath35 approach . in sec .
[ sec : eff_ft ] we address the issue of the stability of the dbl by considering a putative fixed point theory .
we conclude in sec .
[ sec : concl ] with a discussion of physical properties and possible future directions .
in this section , we expand on the initial motivation for considering the dbl states as a way to perform ( a kind of ) flux attachment transformation in a time reversal invariant manner .
we also discuss the nodal structure of the bosonic wavefunctions , providing some justification for the qualifier `` d - wave '' in the suggested wavefunction names in this work . to this end
, we consider a relative single particle wavefunction " more precisely , a cross - section of the many - body wavefunction defined as follows . fixing the positions of all the particles except one ,
we define a function which depends explicitly on the coordinates of the one test " particle , and implicitly on the coordinates of all the other particles , @xmath66 for notational ease we will henceforth drop the implicit dependence on the @xmath67 spatial coordinates and just use the notation @xmath68 .
consider the laughlin wavefunction for bosons in a half - filled landau level , eq .
( [ psi_nu_1/2 ] ) . in this case , the relative single particle wavefunction @xmath70 is complex and has double strength zeroes at the positions of all the other @xmath71 particles . upon close approach to a specific particle ,
say @xmath72 , the function @xmath73 is of a @xmath3 form .
it is in this sense that the @xmath69 laughlin state can be viewed as a d - wave fluid : all pairs of particles , upon close approach in real space , are in a two - body @xmath3 state .
our goal is to construct a time reversal invariant analog of the @xmath69 laughlin state in which particle pairs are similarly in a relative @xmath5 or @xmath74 state . in thinking about laughlin states
, it has been particularly instructive to view them in terms of composite particles created by flux attachment . " for example , if one re - expresses the @xmath69 laughlin state as , @xmath75 the composite fermion " wavefunction is simply a ( vandermonde ) determinant , @xmath76 , of filled landau level orbitals , @xmath77 with @xmath78 . in a second quantized framework
, flux attachment can be achieved by introducing a chern - simons gauge field@xcite .
chern - simons gauge theory has been a useful tool to describe the full haldane - halperin hierarchy of fractional quantum hall states , encoding the fractional charge and statistics of the quasiparticles as well as the structure of the edge excitations@xcite .
but in discussing bosons in a time reversal invariant setting ( zero magnetic field ) , flux - attachment techniques are problematic . for instance
, the usual flux smearing mean field approximation is likely to inadvertently break time reversal invariance . in seeking to avoid chern - simons theory ,
it is worth noting that the @xmath69 state is a perfect square
a square of the vandermonde determinant , @xmath79 this suggests that the @xmath69 laughlin state can be fruitfully described within a slave particle framework.@xcite consider decomposing the boson creation operator as a product of two fermions , @xmath80 . as discussed in the introduction
, there is a local gauge redundancy , and a correct treatment requires the presence of a gauge field minimally coupled to both fermions . in the slave particle mean field approach to the @xmath69 boson problem ,
one simply drops the gauge field , obtaining a problem of two fermion flavors each moving in a magnetic field .
if the electrical charge of the boson is divided equally , each fermion flavor is effectively in a full landau level .
the mean field wavefunction is , @xmath81 to obtain a wavefunction for the bosons it is necessary to project into the physical hilbert space , and this is achieved here by simply setting @xmath82 .
one thereby recovers the laughlin state as a square of the vandermonde determinant , eq .
( [ detv2 ] ) . within a first quantized framework , the relative simplicity of a time reversal invariant bosonic superfluid as compared to the laughlin state is the nodelessness of the ground state wavefunction . as feynman argued many years ago , for non - relativistic bosons moving in the continuum with an interaction only depending on the particle positions , the kinetic energy of any bosonic wavefunction which has sign changes could be reduced by making all the signs positive while keeping the magnitude of the wavefunction fixed to leave the potential energy unchanged .
the ground state wavefunction should thus be everywhere non - negative .
for the non - interacting bose gas the ground state wavefunction is just @xmath83 , but in the presence of interactions a popular variational wavefunction is of the jastrow form,@xcite @xmath84 with variational freedom in the two - particle pseudopotential @xmath85 , which is usually taken to approach zero as @xmath86 at large separations . as in the case with a magnetic field present
, we again consider the relative single particle wavefunction . for a bose condensed superfluid in a time reversal
invariant system @xmath68 can be taken as real and is everywhere non - negative .
if there are repulsive interactions between the bosons in the superfluid , the amplitude of @xmath68 will be reduced when the test particle is taken nearby another particle , which is implemented by the jastrow pseudopotential in eq .
( [ psi_jastrow ] ) .
but the sign of @xmath68 will remain positive , so in some sense all of the particle pairs are in a relative s - type state . in the special case of a hard core interaction ,
@xmath87 for @xmath88 , so one can then view this as an extended " s - type wavefunction .
motivated by the preceding discussion of the @xmath69 laughlin state , in eq .
( [ psi_det - squared ] ) we introduced a simple time reversal invariant wavefunction for hardcore bosons which is the square of a determinant for free fermions filling a fermi sea . by construction
this wavefunction is non - negative .
moreover , the wavefunction will have zeroes which coincide with the nodes of the filled fermi sea , @xmath89 . with time
reversal invariance , a relative single fermion wavefunction , @xmath90 , can be taken as real . as a result
, the nodal structure of @xmath91 will be qualitatively different than for the filled landau level state , vanishing along nodal lines rather than at isolated points .
these nodal lines will pass through the positions of all of the other fermions.@xcite upon taking the test " particle across a nodal line , the function @xmath91 changes sign , vanishing linearly upon approaching the nodal line .
if one takes the position of the test particle close to another particle , @xmath91 will have a @xmath92-wave character in particular a @xmath13 form if we define the @xmath14-axis as being parallel to the nodal line . despite the @xmath92-wave character of the free fermion determinant
, the determinant - squared wavefunction will not have a @xmath16-wave character .
rather , the relative single boson wavefunction will vanish quadratically upon crossing the nodal lines of the fermion determinant .
when the test particle is taken near another particle @xmath93 , the function vanishes quadratically , @xmath94 .
the pair " wavefunction is thus of an extended s - type " form , vanishing along the residual nodal line .
let us now consider the dbl many - body wavefunction eq .
( [ psi_dbl ] ) , which is the product of two different slater determinants for fermions that fill elliptical fermi surfaces as shown in fig .
[ fig : introfs ] .
the nodal structure is revealed by exploring @xmath68 .
this function will have two sets of nodal lines , one from each of the determinants . due to the elliptical nature of the respective fermi surfaces ,
the two sets of lines , both of which pass through all of the particles , will generally not coincide with one another .
indeed , the fermion determinant coming from an elliptical fermi surface will have nodal lines running preferentially perpendicular to the long axis of the ellipse . focusing on the behavior of @xmath68 near a target particle @xmath93 ,
one anticipates a behavior of the @xmath5 form , @xmath95 .
here we have assumed that the two nodal lines actually coincide with the @xmath14 and @xmath21 axes . in general , for a typical configuration of fixed particle coordinates and a given target particle , this will not precisely be the case .
more generically , the two nodal lines will intersect a particle at two angles which are not aligned with the axes . but the sign of the relative wavefunction when the test particle encircles the target particle will still behave as @xmath22 , the same sign structure as a d@xmath31 or d@xmath32 orbital .
this is illustrated in fig .
[ fig : nodes ] .
since the dbl many - body wavefunction is not nodeless , it can not be the ground state of a continuum hamiltonian of bosons .
if we put the coordinates on the sites of a 2d square lattice , the ground state wavefunction is only assured to be non - negative if the sign and form of the hopping matrix in the lattice tight binding hamiltonian is such that it satisfies the marshall sign conditions the requirement that a choice of gauge is possible to make all of the off - diagonal matrix elements negative . in sec .
[ sec : energetics ] we consider a particular marshall sign violating lattice boson hamiltonian which might exhibit a ground state of the proposed d - wave form .
before focusing solely on the dbl states , we want to mention that our construction of time reversal invariant bosonic wavefunctions as a product of two distinct determinants has nice precedents in the studies of spin liquids on the triangular lattice .
one can view the triangular heisenberg antiferromagnet as a system of hard - core bosons at half - filling in the background field of flux @xmath97 through each triangle .
kalmeyer and laughlin@xcite proposed to view this in the continuum , obtaining a boson system at @xmath69 , and their chiral spin liquid wavefunction is precisely the lattice analog of the @xmath69 state eq .
( [ psi_nu_1/2 ] ) . alternatively , using a slave fermion approach,@xcite @xmath98 , we divide the boson charge equally between the two fermions , so each sees on average a flux of @xmath99 per triangle ; the mean field where the @xmath23 and @xmath24 see the same static flux of @xmath99 per triangle gives a filled landau level for each fermion and reproduces precisely the laughlin - kalmeyer chiral spin liquid . however , we can be more creative about the fluxes seen by the slave fermions while maintaining the average of @xmath99 flux per triangle .
one example is to take different flux patterns for the two species as follows : for the @xmath23 fermions , put @xmath100 flux through all up - pointing triangles and @xmath97 flux through all down - pointing triangles , while for the @xmath24 fermions interchange the locations of the @xmath100 and @xmath97 fluxes .
this state is in fact identical to the so - called @xmath101 spin liquid found in ref . .
it is a time reversal invariant gapless algebraic spin liquid ( asl ) with dirac nodes in the spinon spectrum ; it has very good energetics for the nearest neighbor triangular antiferromagnet , starting from the isotropic lattice and all the way to the limit of weakly coupled chains ( e.g. , ref .
found a different gauge - equivalent formulation of this state without realizing its asl character ) .
another example is obtained by taking yet different flux patterns : select one lattice direction chain direction in the anisotropic lattice case . for the @xmath23 fermions ,
put @xmath100 flux through triangles siding even chains and @xmath97 flux through triangles siding odd chains , while for the @xmath24 fermions interchange the locations of the @xmath100 and @xmath97 fluxes .
this is again a time reversal invariant asl and is listed as @xmath102 in ref . .
unlike the @xmath103 state , there is no isotropic @xmath104 liquid , but the energetics performance of @xmath104 and @xmath103 spin liquids is almost indistinguishable for weakly coupled chains and matches that of competing magnetically ordered states ; in this context , the time reversal invariant @xmath103 and @xmath104 liquids are significantly better than the chiral laughlin - kalmeyer variant .
our construction of the dbl states differs in that we do not require any special filling for the bosons and there is no special heisenberg spin symmetry .
neither of these special conditions of the spin model setting are needed for the construction and subsequent gauge theory analysis to go through .
given the growing belief@xcite that critical spin liquids do exist , we do not see any reasons why the situation should be any different for our boson liquids at arbitrary incommensurate densities . in our construction
, such liquids will generically have some underlying partially filled bands and therefore fermi surfaces of slave fermions .
of course , whether a particular state is realized in a given model requires a detailed case by case study , and in sec .
[ sec : energetics ] we suggest some frustrated boson models that may stabilize the dbl phase . our primary goal in the next secs .
[ sec : mft]-[sec : eff_ft ] will be to characterize the dbl states without worrying where to find them . +
we next consider the challenge of constructing a field theory which can access such a @xmath105-bose liquid ( dbl ) state . due to our inability to implement flux attachment in a tractable time reversal invariant manner , we follow instead the slave particle approach .
as above , we decompose the hard core lattice boson as a product of two fermions , @xmath98 .
consider then a lattice @xmath1 gauge theory on the square lattice in terms of these slave particles : @xmath106 with the fermion hopping hamiltonian of the form , @xmath107\\ & & -\sum_{\bf r } \left [ t_\perp e^{-i a_x({\bf r } ) } d_2^\dagger({\bf r } ) d_2({\bf r } + \hat{\bf x } ) + t_\parallel e^{-i a_y({\bf r } ) } d_2^\dagger({\bf r } ) d_2({\bf r } + \hat{\bf y})+ h.c .
\right ] ~.\end{aligned}\ ] ] the @xmath23 fermion hopping amplitudes in the @xmath108 and @xmath109 directions are @xmath110 and @xmath111 respectively , while the two amplitudes are interchanged for the @xmath24 fermions . in the following ,
we take @xmath112 for concreteness .
note also that the two fermion species carry opposite gauge charges .
the gauge field hamiltonian is simply @xmath113 ~ , \label{hgauge}\ ] ] where the lattice magnetic " field is @xmath114 the integer - valued electric " field @xmath115 is canonically conjugate to the compact gauge field @xmath116 on the same link .
the above hamiltonian is supplemented by a gauge constraint on the physical states , which must satisfy gauss law , @xmath117 with @xmath118 in the limit @xmath119 , the electric field vanishes , and gauss law reduces to @xmath120 which projects back into the physical boson hilbert space . in this strong coupling limit ,
it is possible to perturbatively eliminate the gauge field to obtain a hamiltonian for hard core bosons hopping on the square lattice with additional ring exchange terms .
we pursue this in section [ sec : energetics ] where we compare the energetics of the dbl wavefunction with other states . here
we instead focus on the weak coupling limit , @xmath121 , which suppresses the magnetic flux .
the usual slave particle mean field treatment corresponds to simply setting @xmath122 equal to a constant .
the simplest mean field state , and the one which should correspond to the wavefunctions in the previous section , is with zero flux through all plaquettes .
the corresponding mean field hamiltonian describes non - interacting slave fermions .
each species has anisotropic near - neighbor hopping amplitudes , but the two are related under the 90 degree rotation , thus producing a boson liquid that respects the symmetries of the square lattice . in order to focus on the effects of the underlying fermi surfaces without the complications of lattice physics such as brillouin zone folding , we first consider fermions in the 2d continuum with anisotropic effective masses .
the resulting elliptical fermi surfaces are , @xmath123 where the parameter @xmath124 characterizes the degree of eccentricity ( the conventional eccentricity of the ellipses is given by @xmath125 ) .
in the d - wave bose liquid , @xmath126 signifies the mismatch between the two fermi surfaces , and we will refer to this measure as d - eccentricity " .
we study equal time correlation functions since these can be compared directly with the properties of the wavefunctions , which is done in sec .
[ sec : wavefnc_props ] ; we also consider temporal dependencies within the mean field as a measure of spectral properties .
it is useful to have in mind that much of the following analysis of long - distance properties needs only the knowledge of relevant fermi surface patches and not of the full surfaces .
consider first the one - particle off - diagonal density matrix ( or green s function ) for the boson , @xmath127 , with @xmath128 the average boson density .
it will also be of interest to consider the momentum occupation probability , @xmath129 within the mean field theory , the natural approximation for the order parameter correlation is @xmath130 where @xmath131 are the mean field ( bare ) fermion green s functions .
this approximation satisfies @xmath132 . here and below , the imaginary time @xmath133 is understood to be zero if not explicitly present .
the fermion green s functions are readily calculated .
thus , at long distances @xmath134 , the main contribution to @xmath135 comes from the fermi surface patches where the group velocity is parallel or antiparallel to the observation direction @xmath136 . with inversion symmetry
, we can denote the corresponding patch locations as @xmath137 and the fermi surface curvature as @xmath138 , and obtain @xmath139 it is important to remember that @xmath140 , and @xmath141 depend implicitly on the direction @xmath41 and are different in general when the @xmath23 and @xmath24 fermi surfaces do not coincide . for the equal - time boson green s function we thus get @xmath142 } { c_1^{1/2 } c_2^{1/2 }
|{\bf r}|^3 } \nonumber \\ & + & \frac{\cos[({\bf k}_{f_1 } + { \bf k}_{f_2 } ) \cdot { \bf r } - 3\pi/2 ] } { c_1^{1/2 } c_2^{1/2 } |{\bf r}|^3 } ~ , \label{gb_mf}\end{aligned}\ ] ] which decays algebraically while oscillating with @xmath41-direction - dependent wave vectors @xmath143 and @xmath144 .
such wave vectors , which are constructed by considering patches on the two fermi surfaces that are parallel to each other , will span some new loci in the momentum space as illustrated in fig .
[ fig : loci2kf ] .
the above large - distance behavior corresponds to singularities @xmath145 across these lines .
in the zero eccentricity limit ( i.e. , when @xmath146 ) , the two fermi surfaces coincide , giving @xmath147}{r^3}~ , \quad\quad \text{(s - type ) } ~.\ ] ] compared to the general case with @xmath148 , the @xmath144 locus shrinks here to zero momentum , while the singularity in @xmath149 is hardened to @xmath150 . as we show in sec . [
sec : gauge_fluct ] , the power law decay of the mean field boson green s function survives in the presence of gauge fluctuations ( but with modified exponents ) .
the algebraic decay is a consequence ( and a measurable indication ) of the `` criticality '' of the dbl and its gapless excitations .
of course , by the very construction we expect many gapless excitations because of the underlying fermi surfaces .
some measure of the low - energy spectrum is contained in the time dependence of the local boson green s function , @xmath151 where @xmath152 is the density of states at the fermi energy for each species .
the corresponding local boson spectral function is @xmath153 in the discussion of the dbl phase on the lattice in sec .
[ subsec : openfs ] , we will encounter a situation where the boson green s function decays exponentially in space because of the topology of the fermi surfaces , while the above spectral signature of the low - energy excitations depends only on the density of states which is a property of the entire fermi surface and is insensitive to the topology otherwise .
the most natural instability of the dbl is towards a superfluid state . as we discuss in section [ sec : eff_ft ] , in the limit of vanishing d - eccentricity ( @xmath154 , the resulting s - type bose liquid phase obtained in mean field theory is , in the presence of gauge fluctuations , most probably unstable to superfluidity .
moreover , our sec .
[ sec : wavefnc_props ] analysis of the determinant squared wavefunction appropriate to the s - bose liquid is consistent with off - diagonal long - range order .
these results strongly suggest that the mean field s - type bose liquid phase probably can not exist as a stable quantum phase . however , with non - vanishing d - eccentricity , both the gauge theory analysis in sec .
[ sec : eff_ft ] and the properties of the corresponding @xmath155 wavefunction which we explore in sec .
[ sec : wavefnc_props ] suggest that the dbl is a stable critical quantum phase .
it is also instructive to examine several other correlation functions within the present mean field treatment .
specifically , consider the boson density - density correlation function , @xmath156 where @xmath35 is the total number of particles .
as defined , @xmath157 approaches zero for large separations , while negative values at small distances signify a correlation hole .
the density structure factor can be calculated as @xmath158 where @xmath159 - n\delta_{\bf k,0}$ ] and @xmath160 is the system volume .
microscopically , since @xmath98 , for each boson added to the system both a @xmath23 and a @xmath24 fermions are added . as such
, the boson and fermion densities are equal : @xmath161 . however , in the mean field treatment , the two fermion flavors propagate independently with different fermi surfaces , so the corresponding fermion density - density correlation functions , which we denote as @xmath162 , do not coincide .
this ambiguity in the density correlation is an intrinsic deficiency of the mean field theory . as a crude measure , we approximate the boson density correlation function as an average of that of the @xmath23 and @xmath24 fermions , @xmath163 ~. \label{boson_mf_density_corr}\ ] ] the individual density correlation functions in the mean field theory are simply , @xmath164 } { c_\alpha |{\bf r}|^3 } ~ , \label{dd_mf}\ ] ] where again @xmath165 is the place on the @xmath166 fermi surface where the normal is parallel to the observation direction @xmath41 .
we recognize the @xmath167 oscillation with power law envelope , which in the momentum space translates to a singularity in the structure factor @xmath168 across the @xmath169 line ( such @xmath169 surfaces are illustrated in fig .
[ fig : loci2kf ] ) , while there is also a @xmath150 singularity at zero momentum . as obtained from eq .
( [ boson_mf_density_corr ] ) , the mean field boson density correlator @xmath170 has singularities at both @xmath171 and @xmath172 , and knows " about the presence of both fermi surfaces .
remarkably , this mean field approximation is consistent with our analysis of the @xmath155 boson wavefunction in sec .
[ sec : wavefnc_props ] , which also reveals singular behavior at both @xmath173 and @xmath172 . and
@xmath24 fermi surfaces for the parameter @xmath174 , cf .
( [ wparam ] ) . in the left panel ,
the @xmath43 loci are constructed by considering a pair of @xmath23 and @xmath24 fermions with parallel ( @xmath175 sign ) or antiparallel ( @xmath176 sign ) group velocities .
the axes are scaled by an `` average '' @xmath177 that would obtain if the fermi surfaces were circular for the same particle density . ]
-2 mm let us now consider the two - boson correlator , @xmath178 which injects a pair of bosons at @xmath179 and removes a pair at @xmath180 .
consider first the limit where the separation between the two injected bosons , @xmath181 , and the two removed bosons , @xmath182 , are both small compared to the mean inter - particle spacing , @xmath183 .
moreover , take the separation @xmath184 between the injected and removed pair to be much larger than the inter - particle spacing , @xmath185 . in this limit the mean field 2-boson correlator can be expressed as , @xmath186^{5/2 } } \frac{\psi_{xy}({\bf r } ; \hat{\bf r } ) \psi_{xy}^*({\bf r}^\prime ; \hat{\bf r } ) } { ( k_f r)^6 } ~,\ ] ] with pair wavefunction " , @xmath187 here for concreteness we specialized to the elliptical fermi surfaces , eq .
( [ wparam ] ) , and @xmath188 characterizes the d - eccentricity , @xmath189 @xmath190 , is the unit vector pointing from one pair to the other . with vanishing d - eccentricity , @xmath191 ,
the pair wavefunction is of an
extended-@xmath15 " form , for example @xmath192 for @xmath193 .
this corresponds to a quadratic nodal line in the pair wavefunction . with @xmath191
the directions of the nodal lines for each pair are aligned perpendicular to the vector connecting one pair to the other . on the other hand , in the limit of very large d - eccentricity , @xmath194
, the pair wavefunction takes the @xmath5 form , @xmath195 , corresponding to two nodal lines along the @xmath14 and @xmath21 axes . upon rotating the internal coordinate , the sign of the pair wavefunction takes the usual @xmath5 form , precisely as for the cooper pair wavefunction in a @xmath5 superconductor .
thus , the dbl has quasi long - range order in the @xmath5 boson pair channel , although the power law decay exponent within mean field theory is very large .
it is also interesting to examine the time decay of a local " two - boson green s function , @xmath196 with @xmath197 .
a pair of bosons is injected at positions @xmath26 and @xmath198 and is removed at later times at positions @xmath199 and @xmath198 . within mean field theory , @xmath200 ~.\ ] ] for large d - eccentricity ,
@xmath194 , this factorizes into a product of two pair wavefunctions " of the @xmath5 form .
this correlator can be used to extract a local pair tunneling density of states , @xmath201 where @xmath202 injects a @xmath5 pair centered at the origin , @xmath203 with pair size " @xmath204 . within mean field theory one
obtains a power law tunneling density of states @xmath205 , with an amplitude that grows with the d - eccentricity parameter @xmath206 .
the tunneling density of states for an @xmath15-wave pair also vanishes as @xmath207 but the amplitude is independent of @xmath206 .
this pair tunneling density of states is perhaps the best diagnostic for measuring the degree of local @xmath16-wave two - boson correlations in the dbl .
finally , we consider a box correlator for bosons which is defined in the introduction , eq .
( [ bb_def ] ) . within mean field
this factorizes as , @xmath208 , and from wick s theorem and 90-degree rotational invariance , @xmath209 thus , @xmath210 ^ 2 \le 0 ~.\ ] ] notice that in the case with zero eccentricity the mean field box correlator vanishes , but since the @xmath211 wavefunction is non - negative , one expects @xmath212 to be positive upon inclusion of gauge fluctuations . with non - zero d - eccentricity ,
the mean field box correlator is negative for all spatial separations , which reflects the underlying @xmath5 nodal structure , and decays as @xmath213 .
it is possible that the inclusion of gauge fluctuations will again modify the mean field result in the case with closed fermi surfaces because of the pairing tendencies of the @xmath23 and @xmath24 fermions . in the case with open fermi surfaces to be discussed next , we conjecture that the exact box correlator will be negative at large spatial separations while decaying to zero as the box size is taken to infinity .
this conjecture appears to be consistent with the box correlator extracted from the @xmath155 wavefunction with open fermi surfaces in sec .
[ subsec : lattice_wavefnc ] .
the preceding analysis holds for arbitrary fermi surfaces , in particular , for the lattice bands obtained from eq .
( [ ht ] ) .
we only need to remember that the fermion green s function eq .
( [ gd_mf ] ) is determined by the fermi surface patches with the group velocities that are parallel or antiparallel to @xmath41 .
once @xmath214 and @xmath215 are known , the other correlation functions follow .
an interesting situation occurs on the lattice when the ratio @xmath216 is such that the @xmath23 and @xmath24 fermi surfaces are open , which is illustrated in the right panel of fig .
[ fig : introfs ] . in this case , for an observation direction close to , say , the @xmath21-axis , there are no @xmath23 fermi surface patches with normals in this direction , so the @xmath23 green s function has an exponential decay in this direction instead of the power law eq .
( [ gd_mf ] ) . since the real - space boson green s function is the product of the two fermion green s functions , eq .
( [ gb_mfdef ] ) , we conclude that it decays exponentially in the directions near the @xmath14- and @xmath21-axes
. there may still still be directions near @xmath33 in which both @xmath214 and @xmath215 show power law behavior , and so will the boson green s function .
eventually , for large enough @xmath216 , the two fermi surfaces will have no parallel patches , and the boson green s function will decay exponentially in all directions .
we will call this phase dlbl for d - wave correlated local boson liquid . note , however , that the system is still gapless and critical , as can be measured , e.g. , from the local spectral function eq .
( [ aloc ] ) . also , the boson density correlations still have the power law envelope , eq .
( [ dd_mf ] ) , if one fermion field can propagate in the observation direction .
furthermore , the boson box correlator , eq . ( [ bb_mf ] ) , exhibits power law behavior even though the single and pair boson green s functions are exponentially decaying . from the boson field perspective
, the system is local and bosons have hard time to propagate ; nevertheless , the system has power law correlations in other properties and , in particular , charges can propagate . based on an analysis of gauge fluctuations in sec .
[ sec : gauge_fluct ] , we conjecture that the @xmath23 and @xmath24 systems effectively decouple at low energies in the dlbl , and at large distances the exact box correlator is negative and decays to zero as a non - oscillatory power law with an integer exponent which is independent of non - universal lattice scale physics , @xmath217 .
this is the same as in the mean field theory except with a larger power .
it is also worth mentioning the limiting case @xmath218 that gives completely flat fermi surfaces , which we will call `` extremal dlbl . ''
the @xmath23 fermions can move only along the @xmath14-axis , @xmath219 while the @xmath24 fermions can move only along the @xmath21-axis . as a result
, the boson field can not propagate at all , not even one lattice spacing . however , this special system still has power law density - density correlations , e.g. , @xmath220 as well as power law box correlation , @xmath221 from the numerical study of the extremal dlbl wavefunction , sec .
[ subsubsec : xtrmdlbl ] , we conjecture that these mean field power laws also hold upon the gutzwiller projection . in the gauge theory context , we would say that the fermions remain unaffected by the gauge field fluctuations .
this extremal dlbl wavefunction is of interest because of its similarity to the so - called excitonic bose liquid ground state of a pure ring exchange model we discuss this in sec . [
sec : energetics ] .
in this section we study the @xmath222 wavefunctions directly and measure numerically their properties such as the boson green s function and the density correlation defined in sec .
[ subsec : mf_props ] .
a detailed comparison is made with the mean field , which provides an initial guide .
this is followed by interpretations of the observed deviations from the mean field using simple `` amperean interaction '' rules of thumb ; the actual calculations behind these rules within the gauge fluctuations theory are given in sec .
[ sec : gauge_fluct ] .
we first consider wavefunctions in the continuum so as to avoid lattice effects and focus on the consequences of the underlying fermi surfaces .
it appears that the s - type wavefunction , @xmath223 , has off - diagonal long - range order , while a generic dbl wavefunction with non - zero d - eccentricity has only power law boson correlations .
we then consider wavefunctions on the lattice where we can access the dlbl phase with open fermi surfaces described in sec .
[ subsec : openfs ] ; this features boson green s function that decays exponentially in real space . finally , we discuss in some detail the extremal wavefunction that is obtained when the fermi surfaces are completely flat .
the calculations with the wavefunctions are performed numerically using standard determinantal vmc ( variational monte carlo ) techniques.@xcite the system used in all calculations is a square box with periodic boundary conditions . in each case , the particle number is chosen so as to fill complete momentum shells under the fermi surfaces . to facilitate the comparison , the presented mean field is calculated for the same finite systems . before proceeding with the numerics ,
the dbl structure factors can be found analytically in some ranges in the momentum space : thus , by expanding the boson wavefunction in terms of the orbitals that form each determinant , it is easy to see that @xmath149 vanishes outside the @xmath224 surface of fig .
[ fig : loci2kf ] , while @xmath225 vanishes outside the @xmath226 surface .
we begin by considering the limit with zero d - eccentricity , in which case the boson wavefunction is positive everywhere except for the nodes . figure [ fig : s - type ] shows boson green s function measured for two systems with @xmath227 and @xmath228 particles .
it appears that at large separations @xmath229 approaches a finite positive value which is roughly the same for both system sizes . to be more quantitative , the @xmath230 mode contains @xmath231 and @xmath232 fractions of bosons in the two systems .
the present data extracted from the continuum wavefunction can not rule out the possibility that the green s function vanishes in the large distance limit .
but at the very least , the result of the projection is clearly dramatic in this case with matched fermi surfaces , since the fall - off of the boson correlation is very slow ( if any ) .
we also remark that we unambiguously find off - diagonal long - range order when this wavefunction is studied on a lattice at fixed boson density the extremely strong enhancement of the boson correlation over the mean field prediction seen in fig .
[ fig : s - type ] can be understood qualitatively as a result of the pairing of the @xmath23 and @xmath24 fermions mediated by the gauge field . indeed , as we discuss in sec .
[ sec : gauge_fluct ] , the constituents of such a ( zero momentum ) `` cooper pair '' with oppositely directed group velocities and opposite gauge charges produce parallel gauge currents and therefore experience amperean attraction mediated by the gauge field .
this `` amperean attraction '' rule of thumb for the enhancements in correlations appears to be taken to the extreme in the @xmath211 wavefunction , where we can crudely picture the fermions paired back into @xmath233 and condensed , giving rise to long - range order in @xmath234 .
we also point out that this wavefunction has rather unusual density - density correlations which are singular around the @xmath167 circle .
since the presence of the off - diagonal order makes this state less interesting to us , we do not consider such details any further .
the observation of the order suggests that the s - type state is unstable ; the ultimate phase in this case is likely a conventional superfluid , and a good superfluid wavefunction needs to be constructed differently , e.g. , using jastrow approach that builds in proper density correlations .
wavefunction for two systems with @xmath227 and @xmath228 particles .
the distance is measured in units of @xmath235 , and the plots are cut when the separation exceeds half of the box size . since @xmath229 maintains a sizable limiting value that depends only weakly on @xmath35 , we suggest that there is off - diagonal long - range order in this wavefunction .
the mean field result is also plotted , and the vertical scale is chosen to accommodate its fast decay , @xmath236 . ]
-2 mm we now consider in detail a representative case with non - zero d - eccentricity specifically , with @xmath174 in eq .
( [ wparam ] ) .
the corresponding @xmath23 and @xmath24 fermi surfaces are shown to scale in fig .
[ fig : loci2kf ] together with the singular lines for the order parameter and density correlations identified in the mean field , sec .
[ sec : mft ] . figure [ fig : allq ] gives an overall view of @xmath149 and @xmath225 in the two - dimensional momentum space , and also shows a one - dimensional cut in the @xmath237 k - space direction together with the mean field predictions . and
density structure factor @xmath225 measured in the continuum dbl wavefunction with @xmath238 ( note that our @xmath225 defined in eq .
[ db_k ] vanishes for large @xmath239 ) .
the boson number is @xmath240 ; the underlying fermi surfaces and all singular lines are shown to scale in fig .
[ fig : loci2kf ] .
bottom : cut through the data in the ( 1,0 ) direction in the k - space and comparison with the mean field results . in the left panel , the arrows point the appropriate @xmath241 momenta ( see fig .
[ fig : loci2kf ] ) , while in the right panel the arrows point the @xmath242 and @xmath243 momenta . ] -10 mm and density structure factor @xmath225 measured in the continuum dbl wavefunction with @xmath238 ( note that our @xmath225 defined in eq .
[ db_k ] vanishes for large @xmath239 ) .
the boson number is @xmath240 ; the underlying fermi surfaces and all singular lines are shown to scale in fig .
[ fig : loci2kf ] .
bottom : cut through the data in the ( 1,0 ) direction in the k - space and comparison with the mean field results . in the left panel , the arrows point the appropriate @xmath241 momenta ( see fig . [
fig : loci2kf ] ) , while in the right panel the arrows point the @xmath242 and @xmath243 momenta . ] -2 mm consider first the mode occupation function @xmath149 . upon projection , this develops a squarish top with somewhat sharper edges as compared with the mean field .
the latter is not shown in the full k - space but looks more smooth , while a ( 1,0 ) cut can be seen in the bottom panel of fig .
[ fig : allq ] . coming from large momenta , significant deviations from the mean field set in near the @xmath244 line of fig .
[ fig : loci2kf ] ( the corresponding location in the 1d cut is indicated with an arrow ) .
this surface is singular in the mean field , but the singularities are weak and are almost not visible , while they become more pronounced upon the projection . on the other hand , we observe no such enhancement near the @xmath224 line .
this difference between the @xmath244 and @xmath224 is also visible when examining the boson green s function in real space as shown in the left panels of figures [ fig : real - space-10 ] and [ fig : real - space-11 ] .
the mean field @xmath245 , eq .
( [ gb_mf ] ) , has both @xmath244 and @xmath224 components equally present , while after the gutzwiller projection one finds that the @xmath244 component dominates . indeed , for the boson green s
function measured along the @xmath14-axis , fig . [ fig : real - space-10 ] , the relevant fermi surface patches have normals in the @xmath246 direction and are easily located ( the relevant momentum space cut of @xmath149 is also shown in the left panel of fig .
[ fig : allq ] ) . by comparing with the mean field eq .
( [ gb_mf ] ) , one finds that after the projection the amplitude of the oscillation with the smaller wave vector @xmath247 wins over that with the larger wave vector @xmath248 .
slightly more care is needed when interpreting @xmath249 in the @xmath250 diagonal direction , fig .
[ fig : real - space-11 ] .
in this case , the appropriate locations on the @xmath244 surface are at the `` cusp '' points with @xmath251 in fig . [
fig : loci2kf ] , since this is where the normal to the singular surface is parallel to the observation direction @xmath41 .
this component does not oscillate when moving along the @xmath250 diagonal in real space , and when it is enhanced , the oscillations due to the other @xmath224 component become less visible , which is what what we see in fig .
[ fig : real - space-11 ] . and
density correlation @xmath252 measured along the @xmath253 direction for the dbl wavefunction with @xmath240 bosons and d - eccentricity characterized by @xmath238 ( this is the same system as in fig . [
fig : allq ] ) .
the distance is measured in units of @xmath235 , and the plots are cut roughly where the signal in @xmath229 approaches the noise level . note the logarithmic scales used , so it is the absolute values that are plotted , while the positive / negative signs of the data are indicated with filled / open symbols respectively . in the case of the mean field data , the signs are indicated using solid / broken lines ; @xmath254 is always negative , see eq .
( [ dd_mf ] ) . ] -2 mm but in the @xmath255 diagonal direction in real space . ] -2 mm unfortunately , from this data we can not attest whether the enhanced singularities are characterized by new exponents or whether we see only an amplitude effect . the locations where the enhancements occur are consistent with the amperean rules of thumb , sec .
[ sec : gauge_fluct ] .
indeed , the @xmath23 and @xmath24 constituents of the boson operator @xmath98 move in the opposite directions when contributing to the correlation at @xmath244 and therefore experience amperean attraction and enhancement ( remembering that @xmath23 and @xmath24 carry opposite gauge charges ) , while they move in the same direction when contributing at @xmath224 and therefore experience amperean suppression .
these pictures are made more precise in sec .
[ sec : gauge_fluct ] , where we find that within the gauge fluctuations theory , the enhanced correlations are characterized by new exponents that depend on the observation direction because of the varying degree of the fermi surface curvature matching in the @xmath23-@xmath24 pairing channel .
the strongest such enhancement is expected along the diagonals @xmath256 , which is roughly what we find in the projected wavefunctions ( we repeat again that we can not make any statement about the exponents from the data other than that we see increased numerical correlations compared with the mean field ) .
consider now the density correlation @xmath252 and the corresponding structure factor @xmath257 . in fig .
[ fig : allq ] , we can see the singular @xmath258 and @xmath259 lines , cf .
[ fig : loci2kf ] ; the enhancements are peaked where the two curves cross .
the singular @xmath169 points are also visible in the ( 1,0 ) k - space cut , bottom right panel of fig .
[ fig : allq ] . on the other hand ,
the @xmath150 singularity near zero momentum is not enhanced .
examination of the density correlation in real space , right panels of figures [ fig : real - space-10 ] and [ fig : real - space-11 ] , shows an overall increase and dominance of the oscillatory components over the zero - momentum component as compared with the mean field eq .
( [ dd_mf ] ) .
again , we can not tell whether there is a change in the exponents or just an amplitude effect
. the @xmath169 enhancements of the density correlation agree qualitatively with the gauge theory expectations .
indeed , consider the density operator @xmath260 .
the particle and hole constituents ( which are oppositely gauge charged ) have antiparallel group velocities when contributing to the @xmath167 component and therefore experience amperean attraction , while they move in the same direction when contributing at zero momentum and therefore repel each other .
again , these rules of thumb are made more precise in sec .
[ sec : gauge_fluct ] , where such enhancements in the particle - hole channel are characterized by new direction - dependent exponents ( the density correlations are found to decay more slowly when less curved patches are involved ) . note that the projection imposes @xmath261 , and we expect @xmath262 to acquire both the @xmath242 and @xmath243 signatures ( in the gauge theory , @xmath263 and @xmath264 imprint on each other via non - singular high - energy connections that are not manifest in our low - energy effective description of sec .
[ sec : gauge_fluct ] ) .
to summarize , the dbl wavefunction clearly knows about the underlying fermi surfaces ; it also contains germs of the gauge fluctuations theory , since the enhancements / suppressions of the various `` @xmath167 '' lines appear to agree with the amperean rules .
however , there are no reasons to believe that the wavefunction and the gauge theory will have the same long - distance properties . in particular
, our measurements can not tell whether the long - distance power laws are changed upon the projection compared with the mean field .
still , it is gratifying to see that the amperean rules work for the dbl wavefunction .
we now describe our results for various states on the lattice . as we have already mentioned , s - type wavefunction , @xmath223 , with finite boson density per site , have off - diagonal long - range order , which we confirm unambiguously using finite - size scaling . on the other hand , wavefunctions with non - zero d - eccentricity
do not show such order .
while we have not performed detailed studies , we expect that the case with closed fermi surfaces is similar to the already described continuum dbl wavefunction , with a slight complication that one needs to do proper brillouin zone folding when considering singular surfaces in the momentum space . of particular interest
is the case with open fermi surfaces , which can only be realized on the lattice .
specifically , we studied a @xmath265 system with @xmath266 bosons ( @xmath267 per site ) and the fermion hopping parameters @xmath268 , @xmath269 . in this case , the @xmath23 and @xmath24 fermi surfaces have no parallel patches ; we therefore expect that the boson green s function decays exponentially .
when we measure @xmath229 , we find that it drops below our noise level already at 3 lattice spacings , so the corresponding plots are not particularly informative and are not shown here . where we can measure reliably , the values after the projection are of the same order as the mean field values in the same system . since the latter decay exponentially at large distances
, we conjecture the same behavior in the dlbl wavefunction
. on the other hand , the boson density correlation along the @xmath14 and @xmath21 axes shows oscillations with a clear power law envelope , while the correlations are much smaller in the diagonal @xmath256 directions ; this behavior is again qualitatively consistent with the mean field , since neither @xmath23 nor @xmath24 fermi surfaces have normals in the diagonal directions , while one or the other has normals along the @xmath14 or @xmath21 axis .
finally , let us discuss the extremal case when the fermi surfaces are completely flat .
the @xmath23 fermions can move only along the @xmath14-axis and @xmath24 fermions only along the @xmath21-axis . using fermion orbitals that are localized on individual rows for @xmath23 ( or columns for @xmath24 )
, one can see that the boson wavefunction is nonzero only when the number of particles on each row ( or on each column ) is the same .
thus , the bosons can not propagate and their inter - site correlation is identically zero . as we discuss in sec .
[ sec : energetics ] , pure ring hamiltonian conserves boson number on each row / column , and the extremal dlbl wavefunction may be useful in this context .
we can still use the box correlator , eq .
( [ bb_def ] ) , to characterize the state and see some `` gaplessness '' in the system . the measurement is shown in fig .
[ fig : corrbox ] , where we also plot renormalized mean field result .
the latter is obtained by dividing eq .
( [ bb_mf ] ) by @xmath270 , and a crude justification for such procedure@xcite is as follows : each @xmath23 or @xmath24 mean field box calculation contains implicitly a weight of order @xmath270 , since for the ring operator to be nonzero , two specific sites need to be occupied and two need to be empty .
however , after the projection it is enough to require that only the @xmath23 configuration is `` correct '' since the @xmath24 fermions are tied to @xmath23 . from fig .
[ fig : corrbox ] , we see that the box correlator is negative and tracks closely the renormalized mean field values . in the extremal dlbl wavefunction with flat fermi surfaces on a @xmath265 lattice with @xmath266 bosons ( the lines are guide to the eye ) .
note that the box correlator is negative in the mean field and is found to be negative in the dlbl state ; we plot @xmath271 to be able to use the logarithmic scale . the mean field eq .
( [ bb_mf ] ) is renormalized as explained in the text using a numerical factor that removes double weighting of configurations .
the measured values below @xmath272 are at the noise threshold . ]
-2 mm . left panel : real space dependence along the @xmath14-axis , @xmath273 , together with the mean field expectation .
right panel : density structure factor in the momentum space . ]
-2 mm we can also characterize the extremal dlbl state by measuring the density correlations and comparing with the mean field prediction , eq .
( [ dd_mf_xtrm ] ) .
the results are in fig .
[ fig : denscorr ] .
the left panel shows @xmath273 as a function of the distance along the @xmath14-axis ; it has stronger oscillatory component than the mean field and swings back and forth across the zero line while the mean field only touches it , but the overall magnitudes are comparable and decay as @xmath274 .
we also find that the density correlation in the @xmath256 diagonal directions decays exponentially ( not shown ) ; the mean field predicts zero correlation unless strictly along the axes , and we expect that after the projection this corresponds to exponential decay . in the right panel of fig .
[ fig : denscorr ] , we show the density structure factor @xmath225 .
several features are clearly visible : rising `` towers '' draw attention to the lines @xmath275 and @xmath276 ; one can also see a `` cross '' formed by the lines @xmath277 and @xmath278 that run along the axes .
as far as we can say , the character of the singularities across these lines remains the same as in the mean field ; there is an amplitude enhancement of the @xmath169 , but no qualitative difference otherwise . in particular ,
near the line @xmath277 , we observe @xmath279 with @xmath280 which is independent of @xmath281 as long as @xmath282 .
we thus conjecture that the extremal dlbl wavefunction is adequately described by the free fermion mean field .
some of our findings , e.g. , the cross singularity that has a long wavelength character , can be understood semi - analytically , since the absolute value of the wavefunction has a jastrow form with a peculiar pseudo - potential , @xmath283 , it seems plausible that one can calculate other properties of this wavefunction analytically .
it is interesting to note that the above suggests a free fermion description of this boson state , perhaps with some constraints that become irrelevant in an infinite system .
we also note that the extremal dlbl appears to be a relative of the so - called excitonic bose liquid ( ebl ) phase predicted in a pure ring model on the square lattice@xcite that we discuss in sec .
[ sec : energetics ] ; the above therefore suggests that there may be some such description of the ebl in terms of fermions carrying fractional @xmath284 charge .
we now study the gauge theory using analytic techniques , focusing on the large @xmath285 limit where it is reasonable to ignore the presence of magnetic monopoles in space - time and to expand the cosine term in eq .
( [ hgauge ] ) treating the gauge field as non - compact .
moreover , in the @xmath286 limit , the fermions become free and one can put them into a fermi sea state .
a fermi surface of fermions minimally coupled to a non - compact @xmath1 gauge field has a considerable history.@xcite the @xmath287d such system has been studied most notably as a theory for the spin sector in the uniform rvb phase in the slave boson approach to the high-@xmath288 superconductors .
it has been argued that such fermion systems have a stable phase that in some crude aspects is similar to a fermi liquid for example , it has a finite long - wavelength compressibility and spin susceptibility .
however , the system is strikingly different in other aspects and is described by a new non - fermi liquid fixed point .
a scaling description of this fixed point was developed in refs . .
while we largely follow the earlier work , we find that it is convenient to consider a slight reformulation in which the only uncontrolled approximation is to assume that the gauge field dynamics can be described correctly by the rpa approximation , retaining only terms quadratic in the gauge field with a singular quadratic kernel .
a virtue of this approach is that one thereby obtains a theory which has an @xmath35 flavor extension which is soluble at @xmath289 .
moreover , a controlled and systematic perturbation expansion in powers of @xmath36 can be implemented , which allows one to compute physical properties in terms of non - universal bare " parameters such as the shapes of the fermi surfaces .
we employ this approach to calculate the leading @xmath36 behavior for both the boson correlator , @xmath249 , and the density - density correlator , @xmath157 , in the d - wave bose liquid . in the next sec .
[ sec : eff_ft ] , we will consider an effective field theory approach , which allows us to check for the stability of the dbl phase that is present at large @xmath35 .
specifically , we study the effects of residual short range attractive interactions between the two fermion flavors . with vanishing d - eccentricity
when the fermi surfaces for both fermion species become the same , the @xmath15-wave cooper channel is nested " , and a possible instability towards a paired bcs state seems likely .
the resulting state is a bosonic condensate , since @xmath290 .
however , a non - vanishing d - eccentricity of the fermi surfaces destroys the bcs nesting and the possible pairing instability .
indeed , we will find that the dbl with a large d - eccentricity can exist as a stable phase .
( for smaller d - eccentricity , an instability towards a finite momentum bose condensate or an incommensurate charge density wave co - existing with the dbl is a possibility . ) as is discussed in sec .
[ sec : wavefnc_props ] , this is nicely in line with the properties of the associated gutzwiller wavefunctions : the @xmath211 wavefunction that obtains in the limit of zero d - eccentricity appears to have off - diagonal long range order , whereas the general @xmath155 wavefunction exhibits power law correlations consistent with the dbl , as extracted from the gauge theory .
being interested in the low energy properties , it is legitimate to focus on the fermions living near the fermi surfaces , just as in fermi liquid theory .
moreover , the longitudinal coulomb interactions mediated by the gauge field are readily screened out within an rpa treatment , generating short - ranged screened density - density interactions . but the transverse fluctuations of the gauge field the photon are incompletely screened by the fermion particle - hole excitations .
it is thus adequate to just retain the transverse components of the vector potential .
working in the coulomb gauge with @xmath291 , the vector potential reduces to a scalar , e.g. , @xmath292 .
moreover , for each fermion species , it will be sufficient to focus on a pair of fermi surface patches with normals parallel to some axis , say @xmath108 , as shown in fig .
[ fig : patches ] .
it is the @xmath293 component of the vector potential which is minimally coupled to these fermions .
the important wave vectors of the gauge field @xmath294 that are strongly landau damped by the fermions in these patches satisfy @xmath295 ( see fig . [
fig : patches ] ) , and in this region of momentum space @xmath296 .
conversely , the modes of the gauge field @xmath297 with @xmath295 feed back and scatter the patch fermions .
thus , we can focus on fermion fields @xmath298 and the gauge field @xmath297 that are confined to their respective patches in momentum space , as shown schematically in fig .
[ fig : patches ] . for concreteness
, we assume that for each fermion species there are two relevant patches on the opposite sides of the fermi surface that are labelled @xmath299 for the right / left patch as indicated in fig .
[ fig : patches ] .
we then decompose the fermion fields into right and left movers by writing , @xmath300 here , @xmath165 denote the locations on the two fermi surfaces which have normals aligned along the @xmath14-axis , and the fields @xmath301 are assumed to be slowly varying on the scale of @xmath302 .
the full low energy action consists of three terms describing the dynamics of the gauge field , of the fermions , and of their interactions : @xmath303 ~.\ ] ] the patch " lagrangian density for the fermions is simply , @xmath304 d_{\alpha s } ~,\ ] ] which is characterized by the local fermi velocities , @xmath305 , ( @xmath306 ) , and the local fermi surface curvatures , @xmath138 .
the curvature is crucial in the fermion - gauge problem since there is an effective decoupling of the fermions outside the patch from the gauge field , @xmath307 , which is parallel to the patch normals . here
@xmath308 specifies the sign of the group velocity while @xmath309 will always denote the absolute value .
the dynamics for the transverse gauge field describes the free photon , @xmath310 ~,\ ] ] whereas the gauge - fermion interactions are of the usual minimal coupling form : @xmath311 here , @xmath312 with @xmath313 , but it will be convenient for bookkeeping purposes to retain @xmath314 as a dimensionless parameter in the theory .
we emphasize that both the fermion fields and the gauge field , which enter the above action in real space , are slowly varying fields corresponding to momentum modes inside their respective patches ; in what follows , we always specify the wave vectors of the slow fields relative to their respective patch centers .
direction ; the required properties are the fermi velocities and surface curvatures ; the latter are crucial since differently oriented patch systems effectively decouple at low energies . ]
-2 mm the above theory is strongly coupled and can not be treated perturbatively in @xmath314 . as such , it is necessary to resort to approximations to make any meaningful progress . here
we make our central approximation , namely replacing the dynamics of the gauge field by a landau damped form which results if one sums the rpa bubble of particle - hole excitations : @xmath315 with @xmath316 , with the momentum restricted to a region satisfying @xmath317 with @xmath318 .
the magnitude of the landau damping coefficient , @xmath319 , will be determined by the low - energy fermions near the specified fermi surface patches
. within the rpa approximation one obtains , @xmath320 but we will retain @xmath319 as an independent parameter since beyond this point we will not be using rpa in any event .
it will sometimes prove convenient to consider the limit in which the dimensionless gauge charge , @xmath321 is small . here
@xmath322 is a characteristic fermi surface curvature . indeed , within the @xmath35-flavor extension discussed below , while @xmath36 will be the small parameter that leads to a controlled analysis , already at order @xmath36
there are some physical properties which we can only compute analytically when @xmath323 is simultaneously taken as a small parameter .
the parameter @xmath324 that enters in @xmath325 is a `` stiffness '' which has to be positive for stability .
starting from the lattice gauge theory formulation , @xmath324 gives a measure of the energetic cost of setting up static internal fluxes .
one can very crudely estimate @xmath324 using the expression for the diamagnetic response of free fermions , @xmath326 where @xmath327 is some effective mass .
we reiterate that this stiffness comes from some energetics that is assumed to stabilize the fermi surfaces of the @xmath16 fermions in the first place .
therefore , from the low - energy perspective , @xmath324 should be viewed as a phenomenological parameter that encodes some high - energy physics .
furthermore , the stiffness @xmath324 is independent of the patch orientation in momentum space for the transverse gauge field @xmath297 , since gauge invariance requires that the field energy be proportional to @xmath328 . in any event
, the magnitude of @xmath324 will play a minor role below .
indeed , for correlation functions with power law decay as we find below , changing @xmath324 will only modify the amplitude but not the form of the correlators . before proceeding with an analysis of the full theory , @xmath329 , it is important to note that within the above approximation the landau damped dynamics of the gauge field described by the action @xmath325 is both singular and harmonic .
we have ignored possible terms in the action which involve higher powers of the gauge field .
as such , at this stage we could integrate out the gauge field exactly generating a non - local and retarded four - fermion interaction . it is the absence of non - linearities for the gauge field dynamics which allows us to introduce a large-@xmath35 generalization which is soluble at @xmath289 and which facilitates a formal and systematic @xmath36 expansion , as we now describe .
purely for technical reasons , then , we generalize the theory to @xmath35 flavors of fermions at the two patches on each fermi surface , @xmath330 . here
@xmath331 labels the flavor index , while @xmath332 labels the patch location and @xmath333 the two fermion species .
the action for the fermions is simply generalized as , @xmath334 the real scalar gauge field @xmath335 is likewise generalized , but now as an @xmath336 hermitian matrix of complex fields , @xmath337 .
when expressed in momentum space , each component of the matrix , @xmath338 , has dynamics with a landau damped form , @xmath339 , this can also be re - written as a trace over the flavor indices of the matrix @xmath340 : @xmath341 ~.\ ] ] finally , the fermion - gauge field interaction is taken as , @xmath342 notice that the magnitude of the interaction strength ( the charge ) has been taken to scale as @xmath343 . at @xmath37
the theory reduces to that discussed in the previous subsection , with the assumption of a harmonic landau damped form for the gauge field dynamics as the only approximation .
it is worth pointing out that the full action has an enlarged global symmetry being invariant under , @xmath344 for arbitrary phases @xmath345 , @xmath346 , and @xmath347 , @xmath348 .
thus , the densities @xmath349 are independently conserved for any @xmath350 , and also @xmath351 are independently conserved for any @xmath352 .
note also that the dynamics of each of the @xmath35 flavor fermions @xmath353 is identical to one another , and similarly for all the @xmath354 fermions . before solving the full @xmath35-flavor theory , @xmath355 , exactly at @xmath289
, we define large @xmath35 generalizations of the various correlation functions . as in the gauge mean field theory
, we anticipate that the exact fermion green s function for ( any ) one of the @xmath35 fermion flavors , which we denote @xmath356 , will be dominated at large distances by the patch fermions with normals parallel or antiparallel to the observation direction @xmath41 .
we can thus expand @xmath356 for large @xmath357 in terms of the patch " fermion green s functions as , @xmath358 with the definition , @xmath359 note that there is no summation over @xmath350 ; throughout , any summation over indices will be shown explicitly .
there are now @xmath35 flavors of bosons with creation operators , @xmath360 with all @xmath35 boson flavors having identical dynamics .
the large @xmath35 generalization of the boson correlator is then simply , @xmath361 due to the global symmetries , the correlators which are off - diagonal in the flavor indices vanish , leaving a uniquely defined boson green s function , @xmath362 .
as desired , @xmath363 reduces at @xmath37 to the boson correlator defined in sec .
[ subsec : mf_props ] .
the boson density - density correlator , @xmath364 , is readily defined in terms of ( any ) of the boson densities @xmath365 as in eq .
( [ densitycorr : def ] ) .
similarly , we can define the fermion density - density correlators for each species , @xmath366 , in terms of any one of the fermion densities , @xmath367 . as before
, we assume that the boson density - density correlator can be approximated as an average over @xmath352 of the fermion density - density correlators , @xmath368 . in order to extract the above correlation functions , it is convenient to define patch fermion particle - hole and particle - particle bubbles , which we denote as @xmath369 and @xmath370 , respectively .
the exact bubble correlators can be formally expressed in terms of the exact patch fermion green s functions and the exact particle - hole and particle - particle vertices , @xmath371 here @xmath372 and @xmath373 denote the fully renormalized vertices , and @xmath374 and @xmath375 label the patch locations of the two fermion lines that come out of the vertex .
the total momentum and frequency running through the bubble is @xmath376 , and is divided between the two fermions as shown above . at @xmath289
there are no vertex corrections , @xmath377 and @xmath378 , and each bubble is just a convolution of the two green s functions . the boson correlator at large distances and long times can be readily expressed in terms of the bubbles , @xmath379 and similarly for the fermion density - density correlator , @xmath380 at @xmath289 the fermion green s function @xmath381 can be readily obtained by summing the nested rainbow diagrams to give , @xmath382 with @xmath383 because of the gauge interactions , each fermion species obtains an anomalous self - energy of the form @xmath384 which dominates over the bare term @xmath385 on energy scales below @xmath386 .
physically , the fermions cease to propagate as free particles due to the strong scattering from the dynamical gauge field , becoming
incoherent " .
we first consider the long distance and time behavior of @xmath387 . as in the mean field sec .
[ sec : mft ] , the green s function is dominated at large distances by the patches of the fermi surface with normals along @xmath388 . for the equal - time green s function ,
we obtain , @xmath389 where the velocity @xmath305 and curvature @xmath138 characterize the @xmath166 fermi surface patches with normals along @xmath388 , and @xmath390 is given in eq .
( [ gd_mf ] ) . on the other hand ,
the long time dependence of the local fermion green s function is not affected by the anomalous self - energy and is determined solely by the density of states @xmath152 at the fermi energy , @xmath391 turning to the physical correlators , at @xmath289 there are no vertex corrections to the fermion bubble , so we have simply , @xmath392 which gives for the @xmath289 equal - time density correlator , @xmath393 where @xmath394 is given in eq .
( [ dd_mf ] ) .
similarly , in the absence of vertex corrections at @xmath289 in the particle - particle bubble , one has for the equal - time boson green s function , @xmath395 where @xmath245 is given in eq .
( [ gb_mf ] ) . however , the time dependence of the local boson green s function @xmath45 is unchanged from the mean field eq .
( [ gtau ] ) .
finally , the box correlator is given by eq.([bb_mf ] ) with appropriate @xmath289 fermion green s functions and decays with a power law envelope of @xmath396 .
notice that both the boson and the boson density - density correlators at @xmath289 fall off in space as @xmath397 , which is faster than their mean field counterparts . injecting
a boson corresponds to creating two fermions . in the mean field , these fermions propagate independently and as free particles . on the other hand , in the present @xmath289 theory that sums the nested rainbow diagrams ,
both fermions propagate incoherently " due to their scattering from the gauge field , which effectively reduces the ability of the created boson to propagate , thereby leading to a faster decay of the boson correlator .
at this level of approximation , the two injected fermions , while scattered by the gauge fluctuations , do not scatter off one another .
the effects of interactions between the two injected fermions appear at order @xmath36 , entering as vertex corrections .
since the two fermion species have opposite gauge charge , one expects that they will attract one another . as we shall see , though
, they do not form a composite boson ( a bound state , for example ) .
but their motion becomes strongly correlated allowing them to propagate more effectively when close by spatially .
the net result , as we find below , is that the boson can propagate more effectively when the motion of the fermion pair is correlated , leading to a slower decay of the correlator compared to the @xmath289 result .
thus , the effect on the boson dynamics due to the decoherence experienced by the fermions from scattering off the gauge fluctuations is compensated , perhaps only partially , by the attractive interaction between the pair of fermions .
the @xmath36 contributions to the fermion green s function are obtained by considering the nested rainbow diagrams and replacing one gauge field propagator @xmath398 with the bubble correction @xmath399 .
formally , we can first calculate the rainbows using @xmath400 everywhere and then extract the @xmath36 piece . as we now argue , the functional form of the fermion green s function in eq .
( [ n = infinity_greens ] ) remains unchanged except for finite @xmath36 shifts in the parameters .
indeed , evaluating the @xmath289 particle / hole bubble at small wave vector and frequency gives @xmath401 , which only leads to a finite @xmath36 shift in @xmath319 .
this is not surprising , since the bare singular gauge propagator postulated at the outset in our theory is motivated by the free fermion particle - hole bubble , and one can verify that the bubble remains unchanged also in the presence of an arbitrary fermion self - energy that depends on the frequency only .
then , using this renormalized gauge propagator to evaluate the rainbow diagram gives a contribution to the fermion green s function of the form , @xmath402 ^ 2 $ ] .
this will at most give finite @xmath36 shifts to @xmath403 .
next we consider the vertices at @xmath36 , which undergo more dramatic modifications .
consider first dressing the bare vertices with a single gauge propagator .
we denote the value of the corresponding feynman diagrams as @xmath404 and @xmath405 , where @xmath406 stands for one ( landau damped ) photon exchanged .
these diagrams are the first ones in figs .
[ fig : vph ] and [ fig : vpp ] . as we will discuss below ,
the full structure at @xmath36 is more involved , but the calculation of @xmath407 already captures the main qualitative physics . for simplicity ,
we first focus on the situation with zero external @xmath21 momentum and frequency , defining , @xmath408 where @xmath409 , @xmath410 denote the momentum / frequencies in and out of the two legs and @xmath411 is the momentum running through the vertex . for notational simplicity we have suppressed the corresponding flavor labels , @xmath412 .
evaluating the feynman diagrams gives , @xmath413 revealing a logarithmic divergence cut off by the total @xmath14-momentum running through the vertex .
explicit expressions for the dimensionless non - negative coefficients @xmath414 and @xmath415 are given below .
notice that both the particle - hole and the particle - particle vertex corrections vanish unless the two patches have opposite group velocities .
moreover , the sign of the correction is given by the sign of the product of the two charges and differs for the particle - hole and particle - particle channels .
the origin of these signs is essentially ampere s law familiar from electrodynamics .
two moving charges producing parallel charge currents experience an attractive interaction , bringing them closer together and enhancing their interaction strength , which is encapsulated by the magnitude of the interaction vertex .
for example , a particle and a hole of the same fermion species ( @xmath416 ) residing on the opposite fermi surface patches ( @xmath417 ) have an enhanced interaction strength , which is encoded in @xmath418 .
on the other hand , the interaction vertex is suppressed when the charge currents are anti - parallel .
for example , a pair of particles with the same charge ( @xmath419 ) experience a suppressed vertex interaction , @xmath420 , when their group velocities are antiparallel .
the absence of singular vertex corrections when the two patch fermions have parallel group velocities , @xmath421 , is , we believe , not just a peculiarity of this lowest order contribution , but will be valid generally for the exact vertices at arbitrary @xmath35 .
this expectation is based on the specific form of the propagators : the fermion propagator has simple poles in the complex @xmath422 plane while the gauge field propagator is independent of @xmath422 , the latter a reflection of the transverse nature of the gauge field .
for any given feynman diagram , it is the sign of the internal frequency variables running around the loops together with @xmath423 which will determine whether the poles in the integrand will be in the upper or the lower complex @xmath422 planes . for vertices with @xmath421
all of the fermion lines internal to any diagram will have the same group velocity , and with energy ( frequency ) conservation one expects that at least one of the @xmath422 variables will only have poles in the upper ( or lower ) half - plane , so that the contour can be appropriately deformed to show that the diagram vanishes .
thus , we henceforth restrict attention exclusively to vertices with @xmath417 , and to ease the notation will drop the explicit @xmath423 indices in the following .
the dimensionless magnitudes of the vertex enhancements in the particle - hole and the particle - particle channel , @xmath414 and @xmath415 respectively , are given by , @xmath424 ~ , \label{lambda_ph } \\
\lambda_{pp } & = & \frac{g^2}{\sqrt{3}\pi \gamma^{1/3 } \chi^{2/3 } \left ( \frac{\omega^{1/3}}{v } + \frac{\omega^{\prime 1/3}}{v^\prime } \right ) } { \cal e}[\zeta_{pp } ] ~ ; \label{lambda_pp } \\ \lambda_{ph } & = & { \cal e}[\zeta_{ph } ] ~ , \quad\quad \lambda_{pp } = { \cal e}[\zeta_{pp } ] ~. \label{lambda_ninfty}\end{aligned}\ ] ] in the last line , we specialized to the @xmath289 expressions for @xmath386 and @xmath425 ( eq . [ omega ] ) , which is valid for extracting the order @xmath36 vertex corrections . here ,
@xmath426 $ ] is a dimensionless function of a dimensionless argument given by @xmath427 = \frac{3\sqrt{3}}{2\pi } \int_0^\infty \frac{t dt}{(1+t^3 ) ( 1 + \zeta^2 t^4 ) } = \frac{3\pi\zeta^5 + 8\pi(1-\zeta^4)/\sqrt{3 } - 3\sqrt{2}\pi(\zeta^{1/2 } - \zeta^{7/2 } ) - 6\zeta^2 \ln\zeta } { 8\pi ( 1+\zeta^6 ) / \sqrt{3 } } ~ , \label{ffunc}\ ] ] and has been normalized so that @xmath428=1 $ ] .
this function is monotonically decreasing with increasing argument varying as @xmath426 \approx \frac{3 \sqrt{3}}{8 \zeta}$ ] for @xmath429 .
the parameters @xmath430 and @xmath431 are given by , @xmath432 in the last line , we again used the @xmath289 values for @xmath433 .
notice that the vertex interaction strength between two fermions on opposing patches depends on their particular fermi surface curvatures @xmath434 .
moreover , in the particle - particle channel the interaction strength is maximal when the curvatures are equal .
this reflects a particle - particle nesting , which is responsible for the bcs pairing instability in the fermi liquid context . at @xmath36 .
, width=192 ] -2 mm at @xmath36 .
, width=192 ] -2 mm having evaluated the one - photon corrections @xmath407 to the bare vertices , we now discuss the full set of diagrams that appear at order @xmath36 in each case .
first consider the fermion particle - hole vertex which enters into the density - density correlator : @xmath435 .
the @xmath36 diagrams are shown in fig .
[ fig : vph ] and contain an arbitrary number of non - crossing photon lines,@xcite beginning with the one - photon diagram @xmath436 , eq .
( [ vph1a ] ) .
we shall denote the contribution from the diagram with @xmath437 photon lines as @xmath438 .
the magnitude of the singular contributions from each diagram in the ladder sum will depend on a dimensionless gauge charge , @xmath439 above we found @xmath440 \ln(1/q_x^- ) \equiv ( 1/n ) a^{(1)}_1 \ln(1/q_x^-)$ ] .
the singular contributions from the diagram with two photon lines can also be extracted from the feynman diagram , @xmath441 $ ] , with @xmath442 universal dimensionless functions of @xmath443 . from this
we infer the general structure for the diagram with @xmath437 photon lines , @xmath444 this form suggests that the full set of @xmath36 ladder diagrams can be exponentiated , @xmath445 \sim \frac{1}{n } |q_x - q_x^\prime|^{-\gamma_\alpha } ~ , \label{ph_ladder_sum}\ ] ] with @xmath446 .
the validity of this exponentiation can be justified by formally summing the ladder series to obtain an integral equation for the full vertex .
the integral equation has a singular kernel and its solution is expected to have the power - law form with some exponent @xmath55 which will depend on the dimensionless charge @xmath443 .
unfortunately , we have been unable to solve the integral equation to obtain @xmath55 . however , we observe that for small charge @xmath443 , @xmath447 .
this gives the leading behavior even though @xmath448 is not analytic in @xmath443 , as can be seen from eq .
( [ ffunc ] ) ; more generally , one finds @xmath449 .
thus , assuming small parameter @xmath443 , the leading contribution to @xmath55 comes just from the single photon diagram , @xmath450 for @xmath443 of order one , the exponent @xmath55 , while obtained by summing diagrams at order @xmath36 , is itself of order one .
it is the amplitude of the vertex in eq .
( [ ph_ladder_sum ] ) which is of order @xmath36 .
this reflects the particular structure of the large @xmath35 theory , which , constructed to sum the nested rainbows for the fermion self energy at leading order , also exponentiates the ladder sum for the particle - hole vertex at order @xmath36 .
notice that @xmath55 monotonically decreases as the curvature of the fermi surface is taken large , due to the increasingly restrictive phase space requirements on the particle - hole vertex in this limit .
we expect that this trend is valid more generally , for larger @xmath443 and for arbitrary @xmath35 . summarizing the discussion of the particle - hole vertex ,
we can also write a scaling form for more general external momenta and frequencies , @xmath451 we next consider the vertex which enters in the boson - boson correlator , @xmath452 . the full set of diagrams that contribute to the particle - particle vertex are shown in fig .
[ fig : vpp ] and contain an arbitrary number of maximally - crossed photon lines.@xcite the first term is the one - photon diagram that we evaluated in eq .
( [ vpp1a ] ) , @xmath453 \ln(1/q_x^- ) \equiv ( 1/n ) b_1 \ln(1/q_x^-)$ ] ( plus non - singular contributions ) .
an evaluation of the two - photon diagram gives the same form , @xmath454 .
in contrast to the particle - hole case , here the two - photon diagram does not contain a log - squared term , which indicates that the crossed ladder series can not be exponentiated . while we can not evaluate the higher order diagrams
, we expect that each term will have a log contribution with some dimensionless amplitudes @xmath455 .
if this is the case , the full structure of the particle - particle vertex up to order @xmath36 is , @xmath456 with @xmath457 . to check whether the latter series can be defined in principle
, we again take the limit of small dimensionless charges @xmath443 .
consider the least favorable case with matched fermi surface curvatures , @xmath458 , so @xmath459 .
we find that the first term is independent of @xmath323 , @xmath460 .
however , evaluating the two - photon diagram gives @xmath461 for small @xmath323 .
the diagrams are expected to be smaller when @xmath462 ( for example , even the first term @xmath463 $ ] can be made small when @xmath464 ) .
we thus conjecture that the series converges for small enough @xmath443 .
provided it is legitimate to exchange the order of limits @xmath465 and @xmath466 in eq .
( [ bosonvertex_1/n ] ) , the boson vertex can be written in a power law form , @xmath467 with @xmath468 we will discuss the legitimacy of this exponentiation procedure in sec . [
sec : eff_ft ] . for more general external momenta and frequencies
the full boson vertex will satisfy a scaling form , @xmath469 where we have used the fact that the vertex depends only on the total @xmath14-momentum running through it , @xmath411 .
the scaling function will vary as , @xmath470 when any of its arguments @xmath471 .
for example , one has , @xmath472 .
as will be detailed shortly , the singular vertices eq .
( [ vrho;scaling ] ) and eq .
( [ vb;scaling ] ) will lead to anomalous decay exponents for the boson density - density correlator and the boson green s function .
since there are no singular @xmath36 corrections to the fermion green s functions , to extract the boson and density correlators up to order @xmath36 one can use the @xmath289 fermion green s functions when evaluating the bubble diagram . using eq .
( [ n = infinity_greens ] ) and the scaling form eq .
( [ vb;scaling ] ) for the @xmath473 vertex enables the particle - particle bubble to be expressed as , @xmath474 where @xmath475 is an appropriate scaling function .
one can then scale out the @xmath422 dependence under the integrals to obtain , @xmath476 we thereby arrive at the desired expression for the boson correlator , @xmath477 with exponent @xmath478 given in eq .
( [ eta_1/n ] ) .
while this result was obtained systematically to order @xmath36 , we expect it will remain valid more generally , and in particular will involve the same scaling combinations @xmath479 , @xmath480 .
the scaling function @xmath481 and the dependence of the exponent @xmath478 on the bare parameters such as the fermi surface curvatures will presumably vary with @xmath35 . near the other momentum @xmath143 , the boson correlator will satisfy a similar scaling form except with @xmath482 . as argued before , we expect that the anomalous exponent at momentum @xmath483 will vanish in general , not just to leading order in @xmath36 .
the large distance behavior of the equal - time boson green s function in the dbl follows directly from eq .
( [ gb_scaling ] ) and was already listed in the introduction , eq .
( [ gb_final ] ) .
we remark that simple `` power counting '' using eq .
( [ gb_scaling ] ) works here because the relevant frequencies and momenta are indeed in the `` scaling regime '' ; however , it should be kept in mind that we are dealing with line singularities and should exercise more caution with such arguments in general . once again , both the fermi wave vectors @xmath140 and the scaling exponent @xmath478 will depend on the location and shape of the fermi surface patches with normals along the observation direction @xmath41 . at wave vector @xmath144 ,
the green s function decay becomes more slow due to the amperean attraction between the @xmath23 and @xmath24 fermions .
the fermions tend to move as pairs , but unlike a cooper pair their motion is not phase coherent the pairs are uncondensed .
the boson correlator , while decaying more slowly with positive @xmath478 , does not exhibit odlro in the dbl phase . as we found to leading order in @xmath36 , eq .
( [ eta_1/n ] ) , @xmath478 is expected to be largest when the two fermi surface curvatures are equal . in the dbl phase which has closed fermi surfaces , the characteristics of the @xmath23 and @xmath24 patches with normals along either diagonal @xmath33 are always equal due to the symmetry of the square lattice .
moreover , the decay of the boson green s function @xmath249 will be non - oscillatory along the diagonals , @xmath484 a prediction that we find consistent with the properties of the @xmath155 wavefunctions studied in sec . [
sec : wavefnc_props ] ( see fig . [
fig : real - space-11 ] ) . for @xmath485 ,
a very rough estimate can be extracted from the leading behavior , @xmath486 by setting @xmath487 giving @xmath488 .
this would shift the @xmath289 decay exponent back to its mean field value of @xmath489 . as noted earlier , the gauge fluctuations encapsulated already at @xmath289 strongly modify the motion of each fermion leading to a boson that moves less coherently decaying with a larger exponent than in the mean field theory .
the amperean attraction between the pair of fermions which enters at order @xmath36 compensates this effect , leading to a slower decay of the boson correlator . for @xmath490
the two competing effects exactly compensate one another .
next we consider the density - density correlator .
just as for the boson correlator , the density correlator will satisfy a scaling form : @xmath491 with an anomalous exponent that depends on the fermi surface patches , see eq .
( [ anomalous;density;exponent ] ) . back in real space ,
the dominant density correlator oscillates with wave vector @xmath492 , with an envelope decaying as a power law with the anomalous exponent , @xmath493 } { |{\bf r}|^{4-\gamma_\alpha } } - \frac{1}{|{\bf r}|^4 } ~.\ ] ] we have also indicated that the zero momentum component is not modified relative to the @xmath289 behavior .
again , both @xmath165 and @xmath55 are particular to the fermi surface patches with normals parallel to @xmath388 .
in contrast to the particle - particle channel , the particle - hole channel is not nested so that we suspect the exponents @xmath55 will be smaller than @xmath478 for the physically relevant case with @xmath37 . in sum , at leading order in our systematic @xmath36 expansion , the boson correlator was found to exhibit power law decay with an exponent that depends on the bare fermi surface curvatures , varying continuously around the fermi surface .
similar behavior was found for the density - density correlator .
provided this qualitative behavior persists down to the physically relevant case of @xmath37 , we conclude that the dbl phase is described by a manifold of scale invariant theories rather than an isolated fixed point . within the @xmath36 expansion , the power law form obtained for the boson correlator relied on the exponentiation of the logarithmic behavior on momentum and frequency of the leading @xmath36 correction . in the next section
, we check the legitimacy of this procedure by revisiting the renormalization group approach developed earlier to describe the low energy physics of a sea of fermions coupled to a @xmath1 gauge field .
we conclude this section by pointing out that throughout we worked on the assumption of having both @xmath23 and @xmath24 parallel patches present as in fig .
[ fig : patches ] . in the case with open fermi surfaces shown in the right panel of fig .
[ fig : introfs ] , we encounter situations when for a given observation direction @xmath41 only one fermi surface ( or even none at all ) has patches with normals in this direction . in such cases , the boson correlator will decay exponentially ; there is no inter - species gauge interactions ( one species is simply absent ) , but the fermions that are present are still strongly affected by the gauge field , and in particular the preceding analysis of their density correlations remains the same . in the dlbl phase , there are no parallel patches on the entire two fermi surfaces , so the @xmath23 and @xmath24 fermions effectively decouple from each other .
it is also interesting to remark that in the limit of extreme d - eccentricity when the fermi surfaces are completely flat , we expect that the relevant gauge field is very strongly damped .
this can be seen , e.g. , from the naive rpa approximation for the landau damping coefficient @xmath319 , eq . ( [ gamma_rpa ] ) , in the limit of vanishing curvatures . in this limit , the energy scale @xmath386 , eq .
( [ omega ] ) , below which fermions become incoherent , goes to zero , which leads us to speculate that perhaps in this case the @xmath23 and @xmath24 fermions behave as essentially free .
polchinski@xcite and others@xcite argued that 2d fermions interacting with a @xmath1 gauge field can be fruitfully studied within a renormalization group analysis of a particular effective field theory . in the present case
, the fixed point theory has an action given by , @xmath494 with @xmath495 while the gauge field part @xmath325 is given in eq .
( [ sa ] ) and the fermion - gauge coupling is specified in eq .
( [ lint ] ) . within this fixed point ansatz , both the gauge field and the fermions have singular propagators , but their interaction is local .
the advantage of this effective field theory is that it is possible to define a simple renormalization group transformation which leaves the gaussian part of the theory invariant , and under which the interaction strength @xmath314 is a marginal perturbation .
if @xmath314 is assumed to be small , it can be treated via a conventional perturbative rg approach as we describe below .
one finds that the full effective field theory is invariant under the rg , i.e. , it is at a fixed point .
this strongly suggests that exponentiating the logarithmic singularities present in the large @xmath35 approach , as we did to order @xmath36 above , is a correct procedure at all orders in @xmath36 .
if it is , then at order @xmath496 one must find @xmath497 singularities with particular coefficients such that the exponentiation procedure to obtain a power law behavior is consistent .
an unsettling drawback with this effective field theory approach is that it leaves unclear what constraints must be placed on additional interactions that can be added to the theory . at the very least these interactions must be gauge invariant and consistent with momentum conservation , but
is it legitimate to require that the interactions be local in the fermion fields ?
for example , can one require that four - fermion interactions be local with non - singular coefficients , or will singular interactions necessarily be generated ?
indeed , if one were to formally integrate out the gauge field with its singular propagator , one would generate four - fermion interactions with a particular singular form .
in what follows we will ignore these subtleties , exploring the possible perturbative instabilities driven by non - singular fermion interactions .
specifically , we will consider all non - singular quartic interactions involving four fermions living near the fermi surfaces that are consistent with the relevant symmetries , most importantly momentum conservation .
the fixed point action , as it stands , has 9 parameters , @xmath498 but with an appropriate rescaling of the fermion fields and the gauge field together with the momenta and frequency , it is possible to set 5 of these parameters to unity , @xmath499 . the effective field theory is then specified by 4 dimensionless parameters , @xmath500 and @xmath501 . evidently , this theory is not describing a fixed point " per se , but constitutes a four - dimensional manifold of theories which are invariant under the rg . establishing definitively that a particular bare ( lattice ) gauge theory hamiltonian is attracted to this manifold
is exceedingly difficult . arguably , it is even harder to deduce where on this manifold the theory flows .
the values of the dimensionless parameters which are obtained from the leading large @xmath35 analysis can , nevertheless , be used as a rough guide in addressing both questions .
the rg analysis of the fermion - gauge action , @xmath502 , proceeds as follows . at each stage ,
the fermion fields reside in the appropriate momentum space patches @xmath503 , @xmath504 , where @xmath505 is the shell width in the direction normal to the fermi surface .
the corresponding restriction on the frequency is @xmath506 .
the gauge fields reside in similar momentum - frequency regions but centered around zero momentum , and the overall setup is illustrated in fig .
[ fig : patches ] . if the dimensionless parameters exhibited earlier are of order one ( which is the case if the parameters are taken from the large-@xmath35 analysis ) , the corresponding regions are roughly similar for all fermion and gauge fields . also , in practical rg calculations beyond the tree level it is convenient to keep the cutoff only on the frequencies and perform unrestricted integrations over the momenta .
we integrate out the high - energy fields from the shell between @xmath505 and @xmath507 and then rescale the momenta and frequencies in order to recover the initial cutoff : @xmath508 , @xmath509 , and @xmath510 .
we also perform appropriate rescaling of the fields : @xmath511 , @xmath512 . upon such tree - level scaling ,
the fixed - point action remains exactly as before , i.e. , all couplings remain unchanged .
it is also useful to bear in mind that even though we nominally restore all cutoffs as we proceed , in terms of the original momenta the patches become more and more elongated in the @xmath109 direction in fig .
[ fig : patches ] . in particular , any overlap between two patches that correspond to nearby but non - parallel tangents @xmath109 and @xmath513 goes to zero in the low - energy limit . to proceed with the analysis beyond the tree level , we will work perturbatively in the ( dimensionless ) fermion - gauge coupling , assuming that it is small .
quite generally , since the momentum shell rg can not produce terms that are singular at small frequencies and momenta , the couplings @xmath514 and @xmath319 will not renormalize at any order in the perturbation expansion . on the other hand , the parameters @xmath305 , @xmath138 , @xmath324 , and @xmath314 , can potentially flow .
however , a direct examination shows that they do not renormalize at the lowest ( one - loop ) order , and we strongly suspect that this remains true at all orders .
the implication is that the fermion - gauge interaction is exactly marginal and that @xmath502 describes a manifold of rg fixed points parameterized by 4 dimensionless couplings .
although the fermion - gauge coupling need not be small , we expect that this rg - invariant manifold will extend outside the perturbatively accessible regime .
this manifold describes the putative dbl phase . in order to establish the stability of the dbl , however , we need to consider the effects of all symmetry allowed perturbations that can be added to @xmath502 .
stability requires that all such perturbations are irrelevant under the rg . restricting ourselves to local terms , we focus now on the four - fermion interactions which are most likely to destabilize the fixed point manifold .
let us consider a general quartic term , @xmath515 which can contain different fermion species labelled by @xmath352 and involve different patches labelled by @xmath15 .
the fermion / patch labels are assumed to be such as to satisfy the momentum conservation .
also , we do not show explicitly the ( conserved ) frequency and momentum of the fermion fields .
the amplitude for this four - fermion interaction , @xmath206 , measures the strength of a direct scattering between two fermions . at the tree level in the rg , all such short - range interactions are irrelevant , @xmath516 scaling towards zero at low energies .
physically , the incoherent motion of the fermions , manifest in their @xmath517 self energy , leads to a reduced phase space for direct two - body interactions relative to the case for free fermions .
-2 mm at one loop , a gauge propagator can connect any two legs of the quartic vertex which involve fermions located on patches with parallel or antiparallel fermi surface normals .
two such physically distinct diagrams are shown in fig .
[ fig:4ferm ] .
both processes correspond to an interaction between the two fermions mediated by the landau damped gauge field .
the one - loop integrals needed to evaluate these diagrams are identical to those for the vertex corrections in subsection [ subsubsec : v1a ] , and give the following contributions to the four - fermion interaction before any rescaling : @xmath518 here , the primed and unprimed parameters refer to the possibly distinct species of the two fermions connected by the gauge propagator .
the anomalous exponents @xmath414 and @xmath415 are given in eqs .
( [ lambda_ph ] ) and ( [ lambda_pp ] ) these formulas are general and do not assume any relations among the parameters , as appropriate in the present rg setup ; sometimes , which will be indicated explicitly , we will use the @xmath289 expressions eq .
( [ lambda_ninfty ] ) to get a crude estimate of the numerical values .
similarly to the analysis in sec .
[ subsubsec : v1a ] , either diagram gives zero if the fermions involved have parallel group velocities , whereas a nontrivial contribution is found when the two fermion fields reside on the opposite patches with antiparallel group velocities .
the sign is again determined by ampere s law : two fermions emerging from a scattering process with parallel charge currents attract one another , spend more time close together , and consequently enhance the amplitude for the scattering strength @xmath206 .
conversely , the interaction strength is suppressed when the pair of scattered fermions have anti - parallel charge currents .
the difference in the absolute values of the contributions from the diagrams a ) and b ) is similar to the distinction between the particle - hole ( cdw ) and particle - particle ( bcs ) processes in a fermi liquid .
thus , in the situation when @xmath322 and @xmath519 have the same sign , the particle - hole process is not `` nested '' because of the curvature of the fermi surface . on the other hand ,
the particle - particle process is somewhat better nested and becomes perfectly nested when @xmath520 , which is the familiar phase space reasoning in the theory of the bcs instability .
this distinction between the processes can be clearly seen from the arguments @xmath430 and @xmath431 , eqs .
( [ zeta_ph ] ) and ( [ zeta_pp ] ) , of the function @xmath426 $ ] in the two cases . with the above general expressions in hand ,
we now consider specific four - fermion interactions .
of interest is the one - loop contribution @xmath521 to the eigenvalue of such perturbations at the fixed point , @xmath522 the @xmath523 terms , which in general mix different four - fermion terms , can be deduced similarly to the rg treatment of quartic interactions around the fermi liquid fixed point@xcite and in fact have very similar structure . in the fermi liquid case ,
the interactions are marginal and there is no term linear in @xmath206 , so the @xmath523 terms determine the physics . in the fermion - gauge case , on the other hand , the interactions can be either relevant when @xmath524 or irrelevant when @xmath525 . for @xmath525 ,
the dbl phase is stable to weak perturbations , but sufficiently strong bare interactions might still drive the dbl through a quantum phase transition into another phase . in this case
the @xmath523 terms could be helpful in discerning the nature of the new phase .
as in fermi liquid theory , the strongest constraint on the allowed quartic interactions comes from momentum conservation , since all four fermions have to live near the fermi surfaces . in fermi
liquid theory there are only two allowed types of vertices , the forward scattering interactions which contribute to the landau parameters and the cooper vertices .
we first consider interactions in which all four fermions are of the same species .
a general forward scattering interaction then takes the form , @xmath526 where @xmath527 are two wave vectors on the species @xmath352 fermi surface .
consider first the generic case with @xmath528 . since this interaction does not involve opposite patches on the fermi surface ,
the one - loop contribution will vanish , @xmath529 .
this is also the case when @xmath530 .
the forward scattering vertex with @xmath531 is part of the cooper channel to which we now turn . for a single species of fermion
, there exists only the odd angular momentum cooper pairing channel ( `` triplet pairing '' ) , @xmath532 with @xmath533 .
the one loop contribution is negative , @xmath534 for @xmath528 , due to the amperean repulsion between the two particles ( and between the two holes ) with antiparallel group velocities .
for the special case @xmath535 one has @xmath536 since there is now also an amperean attraction between particles and holes on opposite patches of the fermi surface .
but since we expect @xmath537 due to the nesting in the particle - particle channel , one has @xmath538 in this case as well .
thus , the one - loop contribution makes this triplet pairing vertex more irrelevant .
physically the repulsive gauge interaction between two @xmath166 fermions is unfavorable for their pairing .
we now turn to vertices which involve two @xmath23 fermions and two @xmath24 fermions . for simplicity , we first focus on the situation with vanishing d - eccentricity , so that the two fermi surfaces coincide .
consider first the forward scattering interactions coupling the densities of the two fermion species , @xmath539 the only non - vanishing one loop correction is when @xmath531 , and gives @xmath540 .
if this is greater than one , it would signal a possible instability , but in any event we will see that this is smaller than the eigenvalue in the conventional bcs singlet " pairing channel that we consider next .
the singlet bcs interaction is of the form , @xmath541 with zero d - eccentricity as assumed here , the bcs interaction is nested .
moreover , since the two particles ( or the two holes ) here carry opposite gauge charge and therefore have parallel gauge currents , there will be an amperean attraction giving a positive contribution to the cooper vertex : @xmath542 , for @xmath528 . for a circular fermi surface , symmetry dictates that @xmath543 will be @xmath544-independent , but this is actually the case generally for such one - loop contribution in the case with matched @xmath23 and @xmath24 fermi surfaces since the curvatures on the opposing patches are equal .
thus , as claimed earlier , the eigenvalue of the singlet bcs pairing interaction , @xmath545 , is larger than the corresponding eigenvalue of the forward scattering interaction coupling the densities of the two species on opposite sides of the fermi surface , @xmath546 .
for the special case @xmath547 , there will be an additional positive contribution to the eigenvalue of the singlet bcs pairing interaction due to the amperean attraction between particles and holes , @xmath548 . but being a set of measure zero , this will not contribute to a bcs - type pairing instability . at the present it is not clear what state would be preferred by a large interaction of this form and what significance this might have .
it is interesting to use the @xmath289 parameters to deduce an approximate value for the one - loop contribution to @xmath549 . with zero d - eccentricity ,
the two fermi surfaces coincide , so the curvatures are equal , @xmath485 , and @xmath550 = 1 $ ] for all @xmath544 .
this gives @xmath551 and implies that the bcs interaction is strongly relevant , @xmath552 within this approximation .
since we expect some bare short - range attraction between the @xmath23 and @xmath24 particles , the runaway flows will lead to bcs pairing .
the composite boson @xmath98 will condense , resulting in a superfluid phase .
thus , based on this analysis , we suspect that the putative s - type bose liquid phase accessed from the gauge theory with identical fermi surfaces for @xmath23 and the @xmath24 will be generically unstable towards superfluidity .
we now consider the situation with non - zero d - eccentricity . in this case ,
@xmath23 and @xmath24 have different fermi surfaces and the cooper vertex will no longer be nested .
this precludes a conventional weak coupling bcs instability for the d - wave bose liquid phase .
the remaining channels that could potentially drive a weak coupling instability are the forward scattering interactions given in eq .
( [ forward_scattering ] ) with the locations @xmath544 and @xmath553 on the two fermi surfaces tuned to have anti - parallel patch normals . for bosons on the square lattice , with increasing d - eccentricity
the two closed fermi surfaces will eventually open up , and above some critical eccentricity the requirement of anti - parallel @xmath23 and @xmath24 fermi surface normals will be no longer possible . in this large d - eccentricity regime , when the two fermi surfaces are so dissimilar , there are no such forward scattering interactions coupling the two species which can be enhanced by the gauge fluctuations
. sometimes one can find interactions that can be enhanced via the particle - hole attraction within the same species , e.g. , @xmath554 , which requires tuning both @xmath553 and @xmath555 to satisfy momentum conservation but this is not always possible . in any event ,
the corresponding one - loop @xmath556 is not likely to change the irrelevance of this term from the tree level ; e.g. the @xmath289 estimate of @xmath557 , eq .
( [ lambda_ninfty ] ) , is always smaller than 1 .
regarding other interactions , none of the intra - species four fermion terms have positive eigenvalues .
we conclude that in this large d - eccentricity regime the d - wave local bose liquid ( as specified in sec .
[ subsec : openfs ] ) will exist as a stable phase .
returning to the case with smaller d - eccentricity , the stability of the dbl phase will depend on the eigenvalues @xmath558 of the forward scattering interaction that couples a patch on one fermi surface with normal @xmath559 to a patch on the other fermi surface with anti - parallel normal .
the particles on opposing patches experience a strong amperean attraction , whereas a particle and hole experience a weaker repulsion , so that the eigenvalues @xmath560 will be positive for all orientations of @xmath559
. however , the magnitude of @xmath558 depends on the fermi surface curvatures and will vary with orientation .
this can be made explicit by evaluating the one - loop contribution using the @xmath289 values of the various parameters , thereby we obtain an approximate expression for the eigenvalue , @xmath561 - 2{\cal e } \left [ \frac{\sqrt{3}\pi}{2 g^2 } \gamma ( c_1 + c_2 ) \right ] ~.\end{aligned}\ ] ] the first contribution is maximal and equal to @xmath562 when the curvatures on the two fermi surfaces coincide , which they will when @xmath559 is along a diagonal of the square lattice
. we have no such estimate of the second contribution since @xmath563 are independent in our @xmath289 theory ; if we use in addition a crude rpa approximation eq .
( [ gamma_rpa ] ) , we get the total @xmath564 = 1.53 > 1 $ ] . in this case
, the d - wave bose liquid in this regime of d - eccentricity will presumably be unstable , driven by the forward scattering interactions with @xmath565 along the diagonals of the square lattice .
the above numerical estimate is very uncontrolled and is given only to see possible trends . still , let us speculate what the resulting phase might if there is indeed such instability .
most naively , there are two guesses , obtained by a mean field decoupling of the four fermion interaction @xmath566 , where @xmath544 and @xmath553 are the locations of the two patches with normals @xmath559 and @xmath567 that give the largest @xmath568 .
we can assume a non - vanishing particle - particle or particle - hole condensate , @xmath569 or @xmath570 , respectively .
both condensates carry non - zero momentum , @xmath571 for the particle - particle condensate , and @xmath572 for the particle - hole .
but since the interaction is repulsive in the particle - hole channel , such a condensate seems rather unlikely . also , the combination @xmath573 carries a non - zero gauge charge , so is unphysical and can not serve as a legitimate order parameter [ but the product @xmath574 is gauge invariant , and if condensed would correspond to an energy density wave at momentum @xmath575 .
if there is an instability , we think it is more likely to occur in the particle - particle channel .
the order parameter @xmath569 is gauge invariant and correspond physically to a finite momentum bose condensate .
the situation is analogous to the fflo problem , and as there , provided the condensate is not too strong , it will only gap out parts of the fermi surfaces . as such , this would still be a very unusual type of superfluid , which in a mean field description would have residual gapless fermionic excitations . beyond the mean field , the gauge fluctuations would still scatter the gapless fermions , rendering them incoherent .
this state would thus correspond to finite momentum bose condensation co - existing with the dbl . in sum
, whether or not the dbl for small d - eccentricity is unstable to such finite momentum pairing is a quantitative issue which will depend on the values of the anomalous dimensions .
an analysis of the actual wavefunctions in sec .
[ sec : wavefnc_props ] reveals no such tendencies , so might be taken as an indication against the instability .
indeed , as we saw in sec .
[ sec : wavefnc_props ] , the s - wave bose liquid wavefunction does appear generically unstable towards a conventional superfluid , consistent with the expectation of a zero momentum bcs instability in the gauge theory . in the same spirit , we will see in the next section that for hard - core bosons at half - filling the dbl wavefunction in the extreme d - eccentricity limit ( corresponding to fermions which can only hop in one of the two directions on the square lattice ) reveals a co - existence of a commensurate @xmath576 cdw with a gapless dbl fermi surface .
this indicates that the gutzwiller wavefunctions for the dbl , at least in some special instances , can also reveal translational symmetry breaking instabilities , while we repeat that no such instabilities are observed for generic dbl wavefunctions .
this concludes our discussion of the effective field theory description and possible instabilities of the dbl phase .
as we have seen , the main potential instability involves pairing @xmath23 and @xmath24 fermions moving with opposite group velocities , but such bcs channel is suppressed for mismatched fermi surfaces and is completely eliminated in the dlbl regime , which appears to be particularly stable .
in this section we ask what energetics may stabilize the dbl or dlbl phases . specifically , we motivate and study the @xmath577 hamiltonian eq .
( [ hring ] ) with competing boson hopping and ring exchange terms .
when @xmath62 and @xmath64 , this hamiltonian does not satisfy the marshall sign conditions , so one expects the ground state wavefunction to take both positive and negative values .
the hamiltonian ( [ hring ] ) is motivated by considering the gauge theory description eq .
( [ hu1 ] ) of the dbl phase in the strong coupling limit of the gauge theory , @xmath119 . in this limit
, one can perturbatively eliminate the gauge field ; the resulting boson hamiltonian contains , among other terms , both @xmath578 and @xmath579 with @xmath580 and @xmath581 with positive numerical coefficients @xmath582 .
only the signs of the contributions to @xmath583 and @xmath584 are of interest here , since in what follows both @xmath583 and @xmath584 are taken as free parameters of the hamiltonian eq .
( [ hring ] ) . coming from the gauge theory , we have @xmath62 and @xmath64 in the regime of primary interest when @xmath585 and @xmath586 .
to define the system , we also need to specify the boson density per site @xmath587 .
the hard - core boson model has particle - hole symmetry , so it is enough to consider @xmath588 . before proceeding
, we note that the above hamiltonian with @xmath62 and @xmath63 was introduced and analyzed in ref . .
since there is no sign problem in this case , extensive quantum monte carlo studies of refs . were able to map out the phase diagram of the model .
in addition to a bosonic superfluid phase , the numerical studies found two ordered phases at half - filling , a bond - ordered stripe phase and a staggered charge density wave ( cdw ) . when the monte carlo was performed at fixed chemical potential,@xcite varying the chemical potential drove first - order transitions out of the commensurate solid phases into the superfluid . on the other hand , when the monte carlo was performed at fixed density,@xcite the superfluid phase was observed as soon as the density dropped below half - filling , while a phase separation occurred when the density was decreased further . the case @xmath62 and @xmath64 was not studied in monte carlo since there is a sign problem .
below , we present a rudimentary variational energetics study of this regime .
the results are summarized in fig .
[ fig : tkphased ] and lend some support that the dlbl phase of sec . [
subsec : openfs ] may be stabilized by such competing hopping and ring exchange terms .
consider first the hamiltonian with the ring term only . in this case , we can change the sign of @xmath584 by dividing the square lattice into four sublattices and performing a transformation @xmath589 on one sublattice .
the cited quantum monte carlo studies did not consider the pure ring model , but extrapolating these to @xmath590 , it seems likely that at half - filling the ground state is the @xmath576 cdw .
however , we also expect that there is a regime away from half - filling that realizes a novel `` excitonic bose liquid '' ( ebl ) phase discovered and studied by paramekanti,@xcite which we now discuss .
the ebl was accessed by considering a rotor version of the ring hamiltonian , @xmath591 where the phase @xmath592 and the boson density @xmath593 are canonically conjugate .
when the cosine in the ring term is expanded , i.e. , in the phase with no topological defects , one finds a quasi
one - dimensional dispersion relation for the excitons ( density waves ) , @xmath594 .
this gapless normal bose state is expected to be stable against becoming an insulator when @xmath595 dominates over @xmath596 .
the ring only model on the square lattice has special conservation laws : particle numbers on each column and each row are individually conserved .
it is this property that is responsible for the vanishing of @xmath597 along the lines @xmath598 and @xmath599 in the ebl phase .
paramekanti@xcite showed that due to the gapless lines , ebl is a critical ( power - law ) 2d quantum phase with continuously varying exponents . for our energetics study , it is useful to have a good trial wavefunction for the ebl phase of hard - core bosons
. motivated by the described `` spin wave '' theory , we write the following wavefunction @xmath600 ~,\ ] ] with @xmath601 in the spin wave theory , we have @xmath602 , but more generally @xmath603 is treated as a variational parameter .
the summation over @xmath544 excludes the lines where either @xmath598 or @xmath604 .
such wavefunction can be defined in any sector with fixed number of particles on each row and column as is appropriate in the study of the ring - only model .
we find that for moderate @xmath603 and away from half - filling , this wavefunction has a density structure factor of the form , @xmath605 at long wavelengths as expected in the ebl .
furthermore , the ebl theory predicts that the box correlator defined in eq .
( [ bb_def ] ) decays as a power law with a continuously varying exponent , and this is what we observe for the wavefunction . in the energetics study ,
the parameter @xmath603 is varied to minimize the ring energy @xmath579 .
returning to the model with @xmath606 , the trial wavefunction is @xmath607 ~ , \label{psi_ebl}\ ] ] where @xmath608 is the number of bosons on the sublattice @xmath609 of the four sublattices .
one can readily translate the preceding results to this wavefunction , which has the sign structure appropriate for the pure ring exchange model with @xmath606 .
interestingly , the dlbl wavefunction in the extreme d - eccentricity limit , which was introduced towards the end of sec .
[ subsec : openfs ] and discussed further in sec .
[ subsubsec : xtrmdlbl ] , also has the correct sign structure for the ring hamiltonian .
this follows from an observation that this wavefunction takes opposite signs for any two boson configurations that are connected by @xmath579 .
it is important here that @xmath579 contains only unit square plackets , and one subtlety is that such unit ring moves are not ergodic in the sectors mentioned earlier with fixed boson numbers on each row and column .
the signs of the ebl and extremal dlbl wavefunctions as written will not agree everywhere .
however , in the study of the ring model , it is natural to consider smaller sectors comprising states that are connected by repeated applications of @xmath579 , and in such sectors the signs of the two wavefunctions agree .
thus , one needs to carefully specify what sector restriction is being made when working with these wavefunctions .
as far as the @xmath584 energetics is concerned , we find essentially no difference between the wavefunctions obtained by restricting to the smaller ( ergodic ) sectors or the larger sectors which fix only the boson numbers on each row and column .
the presented results are all for the latter choice of sectors ; the monte carlo sampling is performed using arbitrary rectangular ring moves , which are ergodic in this sector .
figure [ fig : etrial_ring ] shows the ring exchange energy per site for the optimized ebl and extremal dlbl wavefunctions plotted vs boson density .
the two trial energies are fairly close ; the ebl state has somewhat lower energy for @xmath610 , and the difference is largest near @xmath611 .
note that the extremal dlbl wavefunction has no variational parameters , while the ebl state has one parameter .
the trial dlbl energy can be somewhat improved by maintaining the same sign structure but taking a variable power of the determinants ; such trial energies ( which we do not show ) approach closer to the optimal ebl energies .
figure [ fig : etrial_ring ] suggests that the dlbl wavefunction not only has the right sign structure , but also has similar short range correlations to the ebl state .
of course , additional jastrow - type factors can be used to further improve the dlbl energy , but here and below we emphasize the good sign structure and correlations that are already present in the determinantal construction itself .
system ; the boson number is always a multiple of 24 since the extremal dlbl requires that the number of particles is exactly the same in each row / column .
away from half - filling the optimized ebl and dlbl wavefunctions produce uniform liquids , but exactly at half - filling we find a long - range @xmath612 cdw order in both states , which is emphasized with large box symbols .
the thin lines are guides to eye , while the thick lines are the result of the maxwell construction for the two trial energies that addresses the possibility of the phase separation at low densities . from this construction ,
the uniform liquids appear to be stable in the the density windows @xmath613 . ]
-2 mm the similarity between the ebl and extremal dlbl wavefunctions is also revealed in their density structure factors , see also sec .
[ subsubsec : xtrmdlbl ] . in both states
away from half - filling , for small @xmath614 , we can write @xmath615 . in the dlbl state
, @xmath616 appears to be independent of @xmath281 . on the other hand , in the ebl state , @xmath617
thus , the character of the @xmath618 singularity is the same in both states for any non - zero @xmath281 , and it is only in the limit @xmath619 that the two differ . interestingly , both wavefunctions also exhibit `` @xmath167 '' singularities in the density structure factor .
such singularities are not apparent in the naive continuum `` spin wave '' ebl theory , but manifest themselves in the lattice hard - core boson ebl wavefunction .
the study of density correlations also reveals that these optimized states at half - filling in fact have spontaneous @xmath576 charge order , while away from half - filling no order is observed .
thus , even though in the present crude energetics work we do not consider directly conventional ordered states of bosons such as charge or bond density waves , the ebl and dlbl wavefunctions provide some access to the energetics of such ordered states .
( note , however , that these trial states likely have some degree of ebl - ness , e.g. , finite compressibility ; this is analogous to the well - known observation that when a conventional jastrow wavefunction is driven into a regime of charge order , it in fact represents a supersolid that also supports off - diagonal long range order . ) as we have already mentioned , monte carlo studies@xcite suggest staggered cdw at @xmath611 , and our finding of the same tendency in the considered wavefunctions lends some support to the goodness of the variational approach .
conventional boson orders are potentially important near other commensurate boson densities but are less important at generic @xmath587 , and we will not consider charge ordered states further . leaving the charge orders aside , an important feature is prominent in the energy plots in fig .
[ fig : etrial_ring ] for the considered uniform liquids .
we see regions where @xmath620 is concave , which signals an instability towards phase separation . using the maxwell construction for the available data ,
we conclude that the system phase separates for boson densities @xmath621 . at low densities ,
the system will spontaneously separate into an empty phase and the ebl phase at density @xmath622 . in the region near @xmath611
, the considered uniform liquids appear to be stable , except perhaps very close to half - filling .
the tendency in the ring model towards phase separation at low densities was noted by rousseau@xcite indeed , if we consider two bosons on an otherwise empty lattice , the lowest ring energy is achieved when the two reside on the same square placket , since only in this case the ring term is nonzero .
thus , two bosons can be bound together by the ring energy . by performing exact diagonalization of small clusters up to @xmath623 with open boundary conditions and for different boson numbers ,
we find that the energy per particle is the lowest when the density is near half - filling ; also , larger open clusters achieve lower energies per boson .
one can construct eigenstates of the ring hamiltonian at low densities composed of such isolated clusters , and the discussed tendencies suggest that not only two but many bosons tend to clump together in the presence of the ring exchanges . one more notable feature in the energy plots occurs near half - filling . we show only the @xmath588 part , while the @xmath624 part would be the mirror image due to particle - hole symmetry .
therefore , @xmath620 shows a cusp precisely at half - filling at least in the ebl case and for the system sizes studied . since @xmath625
, the cusp may be interpreted as an evidence for a charge gap at this density , which appears to be in line with the proposal of the cdw order by the monte carlo studies .
we do not know whether the above observations persist in the limit of large systems .
thus , it may also happen that there will be no regime of the ebl phase in the specific model if the system chooses to phase separate into empty space and a half - filled charge insulator .
if there is a stable ebl regime over some density window near half - filling , then the phase separation at lower densities is likely to be into empty space and an ebl phase of an appropriate density determined from the maxwell construction .
the existence of an ebl phase near half - filling could presumably be checked by careful monte carlo simulations for the pure ring model .
it is also plausible that the regime of the stable uniform ebl phase could be extended by adding interactions that disfavor phase separation , but we have not explored this in the present work .
we now turn to the case with non - zero boson hopping @xmath626 .
consider first small @xmath583 , and suppose we start in the uniform ebl or dlbl regime assuming such exists .
as we have discussed , the hopping and ring terms frustrate one another . at present
, we do not know how to include the effects of hopping if we begin with the ebl .
thus , the ebl wavefunction is defined in sectors with fixed boson numbers on each row / column , but what relative signs should we take for the different sectors that are now mixed by the boson hopping ? if , for example , we naively extend eq .
( [ psi_ebl ] ) to all boson configurations , with perhaps modified jastrow factors , the expectation value of @xmath578 remains precisely zero .
it seems that one needs to change the sign structure of the wavefunction in order to treat the frustration accurately . on the other hand
, the dlbl wavefunction provides a natural continuation from the extremal case of the ring only model , and we find that it is indeed capable of utilizing both the ring energy and the hopping energy .
more precisely , for non - zero @xmath583 the optimized d - eccentricity is finite and varies as @xmath583 is increased , moving towards zero d - eccentricity at large @xmath583 . within our restricted variational energetics study
this is the strongest evidence we can offer that the ground state of the ring hamiltonian in this regime is in the dlbl phase . model .
numerics is done for the same sizes as in fig .
[ fig : etrial_ring ] .
the phase boundaries are obtained by examining the dbl state , allowing for the possibility of phase separation , and the superfluid state ( the latter appears to be always uniform in our study ) .
the optimal values of @xmath627 throughout the whole dbl phase are such that the fermions have open fermi surfaces , i.e. , this is the dlbl phase of sec .
[ subsec : openfs ] . ] -2 mm some details of the variational study are summarized in the `` phase diagram '' in fig .
[ fig : tkphased ] . at a given density ,
the trial energy of the dbl state is optimized using the ratio @xmath627 as a variational parameter . similarly to the pure ring model , we also analyze the possibility of phase separation , which is done in the spirit of fig .
[ fig : etrial_ring ] .
we find a stable uniform dbl phase in the density range roughly similar to the ring only model .
physically , we expect the phase separation to be suppressed when the hopping becomes non - zero , but our study is too limited to explore this systematically .
the phase diagram fig .
[ fig : tkphased ] also includes a conventional superfluid state of bosons , which is expected in the regime of large @xmath583 .
we take the jastrow - type form for the trial wavefunction , eq .
( [ psi_jastrow ] ) , and use the pseudopotential @xmath628 with two variational parameters @xmath206 and @xmath92 . allowing more parameters such as an independent nearest - neighbor pseudopotential does not visibly change the optimized energies .
interestingly , as deduced from the energetics , the superfluid - dbl phase boundary roughly coincides with the limit of vanishing d - eccentricity @xmath629 .
the resulting @xmath211 wavefunction is positive , and , as we discussed at length in sec .
[ sec : wavefnc_props ] , this state has off - diagonal long range order .
of course , it does not have the same long - wavelength correlations as the superfluid ( e.g. , it has unusual @xmath169 density correlations ) , but it appears to have reasonable short - range correlations . finally , we also remark that in the present @xmath577 study , the optimal dbl phase when it wins over the superfluid is in fact the dlbl phase with open fermi surfaces discussed in sec .
[ subsec : openfs ] and is in the regime where the boson green s function decays exponentially in all directions . as we discussed in secs .
[ sec : mft ] and [ sec : wavefnc_props ] , despite such `` locality '' in the boson correlator , this unusual phase is still critical with many low energy excitations .
summarizing , the dbl wavefunction is able to interpolate between the ebl and the superfluid regimes in the frustrated @xmath577 model .
this is the main argument supporting our proposal that the ground state of the ring hamiltonian in the intermediate regime is in the dlbl phase .
our main results were summarized in the introduction , so here we only highlight the most interesting points and discuss possible future directions . we are searching for examples of uncondensed quantum boson liquids that respect time reversal and occur at generic continuously varying densities . in this paper , we proposed several wavefunctions that produce such liquids and studied in detail the specific states dubbed dbl and dlbl .
while the initial wavefunction construction appears somewhat ad - hoc , we argued that it is no more so than the slave particle construction of spin liquid states in frustrated antiferromagnets , which has been developed to maturity over the last two decades .
we introduced a particular slave fermion treatment of hard core bosons on a square lattice that corresponds to the proposed dbl wavefunctions .
this approach allows to go beyond the trial ground state wavefunction , suggesting natural excitations , and eventually leads to a gauge theory description of the low - energy properties of the putative boson liquid phase . using techniques previously developed for the so - called uniform rvb spin liquid with a fermi sea of spinons , we argued for the possible stability of the new dbl phases to gauge fluctuations and found the dlbl regime to be particularly stable . within the gauge theory formulation
, we also analyzed boson and boson density correlators at long distances , which reveal special singular surfaces in the momentum space .
the singular surfaces were confirmed by measuring the properties of the proposed dbl / dlbl wavefunctions directly ; in the same numerics , no potential orders were observed , indicating stability of the states .
we remark here that the gutzwiller wavefunctions and the gauge theory should not be viewed as identical descriptions , and , in particular , the detailed long - distance properties may be different .
the gutzwiller wavefunctions can be used for crude energetics and qualitative characterization , but in our opinion the gauge theory is more complete as far as the actual quantum phase is concerned .
indeed , the wavefunctions as constructed do not incorporate any dynamics of the gauge fluxes , which are clearly additional low - energy `` variational '' degrees of freedoms that the quantum system is going to explore and utilize .
from the singular momentum surfaces in the boson properties , one can in principle recover the underlying fermi surfaces of the slave particle construction ; the latter have correct volumes for the boson density , offering a tantalizing possibility of some `` luttinger theorem '' for uncondensed liquids of interacting bosons .
some striking thermodynamic properties of the proposed boson liquids follow from their fermion - gauge character.@xcite in both the dbl and dlbl phases , because of the gapless fermi surfaces and the gauge interactions , the specific heat at low temperatures is expected to behave as @xmath630 , while the resistivity is expected to vanish as @xmath631 .
the last result in the dbl phase with closed fermi surfaces is simply a quote from earlier studies of such fermi sea - gauge systems,@xcite while the case with open fermi surfaces requires slightly more care .
our argument is based on the application of the ioffe - larkin rule@xcite which gives the boson resistivity as @xmath632 . for the fermi surfaces as in the right panel of fig .
[ fig : introfs ] , and the conductivity measured , say , in the @xmath108 direction , the small - angle scattering by the gauge field can effectively degrade the @xmath24 fermion current , therefore the standard result @xmath633 . on the other hand
, such scattering can not completely degrade the @xmath23 fermion current , so @xmath634 , and @xmath631 follows .
( parenthetically , the @xmath23 current can be degraded by quartic or more particle interactions , but such contributions vanish as @xmath635 or faster . ) intriguingly , in the dlbl phase , the boson correlators are expected to be short - ranged despite such manifest thermodynamic signatures of gaplessness and criticality . on the other hand ,
the boson box correlator in the dlbl decays as a power law and is negative , @xmath396 , offering perhaps a more physical glimpse of the fermionic partons in this phase .
indeed , consider inserting bosons at @xmath636 and @xmath637 and removing at @xmath638 and @xmath639 .
the dominant contribution to the box correlator comes from the @xmath23 fermions propagating @xmath640 and @xmath641 , while the @xmath24 paths are @xmath642 and @xmath643 , which is illustrated in fig . [
fig : partons ] . the minus sign of the box correlator is then due to one fermionic exchange needed so that the two pairs of injected fermions are removed with the same pairing .
the proposed dbl / dlbl wavefunctions do not satisfy marshall signs , and their interesting sign structure is brought out by the nodal pictures in the continuum such as fig . [
fig : nodes ] , which give some caricature of the wavefunction signs also on the lattice even though the nodes are not sharply defined in this case .
clearly , these wavefunctions can not be ground states of boson models without frustration . to address the question of what hamiltonians may stabilize such phases , we considered a particular frustrated hard core boson model with competing hopping and four - site ring exchange terms , and our energetics study suggests that the dlbl state is a good candidate in this model near half - filling .
unfortunately , the frustrated nature of the boson motion makes this model not suitable for large system quantum monte carlo studies . we still suggest it as an interesting model for numerical studies such as exact diagonalization of small systems or dmrg
; the gapless nature of the boson liquid may make results hard to interpret , but the short - range character of the correlations and particularly the quasi - local nature of the dlbl phase can perhaps facilitate dmrg to access larger systems . looking ahead
, we now describe one of the main drives behind the study of uncondensed boson liquids , despite its own intrinsic interest .
we are searching for electronic conducting non - fermi liquids , and the ideas of the present work suggest some avenues in this direction . as a specific example , consider the slave boson approach that is popular in the context of the @xmath644 model of high-@xmath288 superconductors .
the electron operator is written as @xmath645 , leading to a theory of spinons and slave bosons strongly coupled via an emergent fluctuating gauge field .
phenomenologically , a fermi sea of spinons is very appealing in the strange metal , but what about the slave bosons ?
if they condense , we would recover the conventional fermi liquid .
it has been argued that the fluctuating gauge field frustrates the motion of the bosons and suppresses the tendency to condense.@xcite if this suppression could persist to zero temperature , we would obtain a conducting non - fermi liquid state .
such a possibility is precisely what we are trying to establish in the present work , and our approach would be to write the slave boson in terms of new ( second generation ) slave fermions , @xmath98 . on the level of wavefunctions
, we would then write an electronic state of the form : @xmath646 ( \up , \dn ) \,\ , \det(\up ) \ , \det(\dn ) ~,\end{aligned}\ ] ] where schematically @xmath647 or @xmath648 denotes locations of all electrons with spin - up or spin - down respectively .
each determinant in the slave boson wavefunction @xmath649 evaluates appropriate @xmath16-particle orbitals at the locations of all electrons irrespective of their spin , which assures the no - double occupancy constraint .
there is significant freedom in specifying these orbitals , and whether such wavefunctions are useful for any electronic system requires detailed energetics studies of specific hamiltonians .
one can nevertheless develop a low - energy theory of such a @xmath650 phase in the spirit of the present paper , thereby obtaining an itinerant non - fermi liquid conducting phase .
if the dbl or dlbl bosonic states studied in this work appear as useful caricatures of the behavior of the electronic charge , we may even call the resulting phase a `` d - wave metal '' ! such all - fermion description of electrons on the square lattice with no double occupancy has some phenomenological appeal , but detailed explorations are left for future work .
we would like to acknowledge discussions with t. senthil and a. vishwanath , and thank jason alicea for help with some calculations .
the work at kitp was supported by the national science foundation through grants phy-9907949 and dmr-0529399 .
the structure of the @xmath36 diagrams for the particle - hole and particle - particle vertices , figures [ fig : vph ] and [ fig : vpp ] , is specific to our @xmath35-flavor extension defined in sec .
[ subsec : n_flavor ] , while different extensions can lead to different organization of diagrams .
j. polchinski , effective field theory and the fermi surface , _ in _ recent directions in particle theory , proc .
1992 tasi , eds .
j. harvey and j. polchinski ( world scientific , singapore , 1993 ) ; hep - th/9210046 | we develop a description of a new quantum liquid phase of interacting bosons confined in two dimensions which possesses relative d - wave two - body correlations .
we refer to this stable quantum phase as a d - wave bose liquid ( dbl ) .
the dbl has no broken symmetries , supports gapless boson excitations which reside on bose surfaces " in momentum space , and exhibits power law correlation functions characterized by a manifold of continuously variable exponents .
while the dbl can be constructed for bosons moving in the 2d continuum , the state only respects the point group symmetries of the square lattice . on the square lattice
the dbl respects all symmetries and does not require a particular lattice filling . but lattice effects do allow for the possibility of a second distinct phase , a quasi - local variant which we refer to as a d - wave local bose liquid ( dlbl ) . remarkably , the dlbl has short - range boson correlations and hence no bose surfaces , despite sharing gapless excitations and other critical signatures with the dbl
. moreover , both phases are metals with a resistance that vanishes as a power of the temperature .
we establish these results by constructing a class of many - particle wavefunctions for the dbl , which are time reversal invariant analogs of laughlin s quantum hall wavefunction for bosons in a half - filled landau level .
a gauge theory formulation leads to a simple mean field theory and a suitable @xmath0flavor generalization enables incorporation of gauge field fluctuations to deduce the properties of the dbl / dlbl in a controlled and systematic fashion .
various equal time correlation functions thereby obtained are in qualitative accord with the properties inferred from the variational wavefunctions .
we also identify a promising microscopic hamiltonian which might manifest the dbl or dlbl , and perform a variational energetics study comparing to other competing phases including the superfluid .
we suggest how the d - wave bose liquid wavefunction can be suitably generalized to describe an itinerant non - fermi liquid phase of electrons on the square lattice with a no double occupancy constraint , a d - wave metal phase . |
a fano manifold is a smooth complex projective variety @xmath0 with ample anticanonical class .
these varieties have a rich structure , and play an important role in higher dimensional geometry under the viewpoint of mori theory , because they appear as fibers of mori contractions of fiber type of smooth projective varieties .
after @xcite we know that @xmath0 is rationally connected and simply connected . moreover
smooth fano varieties of dimension @xmath11 form a limited family , so they have only a finite number of possible topological types . however in general very little is known about their topological invariants . in particular we consider here the second betti number @xmath12 of @xmath0 , which coincides with the picard number @xmath9 .
recall that a del pezzo surface @xmath13 has @xmath14 .
fano @xmath15-folds have been classified by iskovskikh , mori , and mukai , see @xcite and references therein .
thus we know that a fano @xmath15-fold @xmath0 has @xmath16 .
in fact , more is true : as soon as @xmath17 , @xmath0 is a product of a del pezzo surface with @xmath18 ( * ? ? ? * theorem 2 ) .
starting from dimension @xmath19 , we do not have a bound on @xmath9 .
the known examples with largest picard number are just products of del pezzo surfaces with picard number @xmath20 , which gives @xmath21 .
optimistically one could think that fano varieties with large picard number are simpler , maybe a product of lower dimensional varieties .
this would yield a linear bound ( in the dimension @xmath11 ) for @xmath9 , in fact one could expect precisely @xmath22 ( see @xcite ) .
this is actually what happens in the toric case : if @xmath0 is a smooth toric fano variety of dimension @xmath11 , then @xmath23 , and equality holds if and only if @xmath11 is even and @xmath0 is @xmath24 , @xmath13 the blow - up of @xmath25 in three non collinear points ( see @xcite ) .
let us also recall that there is a class of fano varieties for which a stronger linear bound on @xmath9 is expected .
these are fano varieties which do not contain curves of anticanonical degree @xmath26 , e.g. fano varieties of index @xmath27 . in this case
it is expected that @xmath28 , with equality only for @xmath29 .
this is a generalization of a conjecture by mukai , and has been proved in dimension @xmath30 and in the toric case , see @xcite .
a strategy in this direction is to look for a contraction @xmath31 of fiber type on @xmath0 , and try to bound @xmath9 in terms of @xmath32 and @xmath33 , where @xmath3 is a general fiber .
there are ( at least ) two difficulties in this approach .
first , @xmath7 may not be fano , so that we do not know how to bound @xmath32 .
second , surely @xmath3 is fano , but in general @xmath33 is much smaller than @xmath34 ( for instance any del pezzo surface @xmath13 admits a contraction @xmath35 with general fiber @xmath18 ) .
this problem could be avoided by considering only elementary contractions of fiber type , for which @xmath36 .
however this is not very satisfactory , because we do not necessarily expect a fano variety with large picard number to have an elementary contraction of fiber type ( think of a product of del pezzo surfaces ) .
this was our motivation to introduce the notion of `` quasi elementary '' contraction of fiber type .
this is a contraction of fiber type as above , such that if @xmath37 is a general fiber , then the image of @xmath38 contains all numerical classes of curves contracted by @xmath4 ( see definition [ quasiel ] ) .
this implies that @xmath2 .
in particular any elementary contraction of fiber type is quasi elementary .
if a contraction is a smooth morphism ( such as a projection @xmath39 ) , then it is quasi elementary , see lemma [ smoothness ] .
we study several properties of quasi elementary contractions of smooth fano varieties , in particular when the target has small dimension .
the following is our main result .
[ result ] let @xmath0 be a smooth complex fano variety , @xmath1 a quasi elementary contraction of fiber type , and @xmath3 a general fiber . 1 .
suppose that @xmath40 .
then @xmath7 is a smooth del pezzo surface , @xmath4 is equidimensional , and @xmath41 . + if moreover @xmath42 , then @xmath43 and @xmath4 is the projection on the first factor .
2 . suppose that @xmath44
. then @xmath45 .
+ if moreover @xmath6 , then @xmath7 is smooth and fano .
+ if @xmath8 , then @xmath46 and @xmath47 , where @xmath13 is a del pezzo surface , and @xmath48 a smooth fano variety with a quasi elementary contraction onto @xmath18 . in the case
@xmath49 , @xmath50 has been shown in ( * ? ? ?
* proposition 8) under the more general assumption that @xmath1 is an equidimensional contraction onto a surface .
see section 7 ( p. ) for examples concerning the sharpness of the statement .
the following are some applications to fano @xmath19-folds and @xmath51-folds .
[ 4fold ] let @xmath0 be a smooth complex fano 4-fold . if @xmath0 has a non trivial quasi elementary contraction of fiber type , then @xmath52 , with equality if and only if @xmath53 , @xmath54 del pezzo surfaces with @xmath55 .
if @xmath0 has an elementary contraction onto a surface @xmath13 and @xmath56 , then @xmath57 . if @xmath0 has an elementary contraction onto a threefold and @xmath58 , then either @xmath59 , or @xmath60 , @xmath13 a del pezzo surface .
[ 5fold ] let @xmath0 be a smooth complex fano 5-fold . if @xmath0 has two distinct elementary contractions of fiber type , then @xmath61 .
if @xmath0 has an elementary contraction onto @xmath7 with @xmath5 , then @xmath62 .
suppose that @xmath0 has an elementary contraction @xmath1 with @xmath63 .
if @xmath0 has another elementary contraction @xmath64 of type ( 3,0 ) , ( 4,0 ) , ( 4,1 ) , or such that @xmath65 , then @xmath61 .
finally we give an application to fano varieties with two quasi elementary contractions .
[ appl ] let @xmath0 be a smooth complex fano variety of dimension @xmath11 .
let @xmath66 and @xmath67 be two quasi elementary contractions of fiber type with @xmath68 .
let @xmath69 be a general fiber of @xmath70
. then @xmath71 , moreover : 1 .
@xmath72 implies @xmath73 2 .
@xmath74 implies @xmath75 3 .
@xmath76 implies @xmath77 4 .
if @xmath78 and @xmath79 is elementary , then @xmath80 .
if @xmath70 is elementary one can replace @xmath81 by @xmath26 in the statement , see remark [ elem ] .
these corollaries should be compared to the following result by winiewski , involving an arbitrary number of elementary contractions of fiber type .
let @xmath0 be a smooth complex fano variety of dimension @xmath11 .
suppose that @xmath0 has @xmath82 distinct elementary contractions of fiber type , and let @xmath83 be the general fibers
. then @xmath84 , moreover : 1 .
@xmath85 implies @xmath86 2 .
@xmath87 implies @xmath88 . in particular @xmath89 , and if @xmath90 then @xmath28 .
our main tool is mori theory , in particular we use many properties of contractions of fano varieties shown in @xcite . in section [ prel ]
we recall some basic notions and properties . in [ lifting ] we show that when a variety @xmath7 is the target of a contraction @xmath1 where @xmath0 is fano , @xmath7 shares with @xmath0 a good behaviour with respect to mori theory , see lemma [ easy ] . in section [ qec ]
we define quasi elementary contractions of fiber type @xmath1 and give some related properties and examples .
we study the singularities of @xmath7 in lemma [ target ] , generalizing results known in the elementary case and using in particular results from @xcite .
then in [ boh ] we study elementary contractions of @xmath7 by means of their liftings to @xmath0 .
this is a key ingredient in the proof of theorem [ result ] .
finally in [ div ] we show that if @xmath0 is fano , @xmath91 and @xmath7 contains a prime divisor @xmath92 with @xmath93 , then @xmath94 , so @xmath95 where @xmath3 is a general fiber of @xmath4 .
this is a generalization of results from @xcite and @xcite . in section [ quasiunsplit ]
we apply results from @xcite to deduce that some contractions of @xmath0 always induce contractions of @xmath7 .
this is needed in the proof of theorem [ result ] @xmath50 , moreover it extends the applicability of theorem [ result ] , see section [ ea ] .
section [ surf ] contains the proof of theorem [ result ] @xmath50 , which relies on the results of sections [ qec ] and [ quasiunsplit ] . in section [ 3fold ]
we show theorem [ result ] @xmath96 .
this is based on a detailed analysis of the possible elementary contractions of the target @xmath7 .
we need the classification of smooth fano @xmath15-folds by mori and mukai , and we imitate the strategy for the classification of imprimitive smooth fano @xmath15-folds ( see @xcite ) to get some results about the singular case .
we also use the existence of a smoothing of a terminal fano @xmath15-fold @xmath97 shown in @xcite , and some relations among @xmath97 and its smoothing shown in @xcite , see lemma [ conodimori ] .
finally in section [ ea ] we prove corollaries [ 4fold ] , [ 5fold ] , and [ appl ] , we give some other application and related examples .
i wish to thank laurent bonavero and stphane druel for many discussions about fano varieties and their picard number , and about families of rational curves .
my thanks also to rita pardini , for her interest in this subject and for the conversations we had about it , and to priska jahnke , from whom i learnt a lot about singular fano threefolds .
finally i thank the referee for useful remarks .
we work over the field of complex numbers .
let @xmath0 be a normal projective variety of dimension @xmath11 .
we denote by @xmath98 the smooth locus of @xmath0 .
let @xmath99 be the vector space of @xmath26-cycles in @xmath0 with real coefficients , modulo numerical equivalence .
the dimension of @xmath99 is equal to the picard number @xmath9 of @xmath0 . inside @xmath99
we denote by @xmath100 the convex cone generated by classes of effective curves , and by @xmath101 its closure .
if @xmath102 is a curve , its numerical class is @xmath103\in{{\mathcal}{n}_1}(x)$ ] . if @xmath97 is a closed subset of @xmath0 , call @xmath104 the inclusion , and consider the linear map @xmath105 we denote by @xmath106 the image of @xmath107 in @xmath99 .
thus we have : @xmath108 a _ contraction _ of @xmath0 is a surjective morphism with connected fibers @xmath1 onto a projective and normal variety @xmath7 .
the push - forward of @xmath26-cycles defined by @xmath4 gives a surjective linear map @xmath109 so that @xmath110 .
we also consider the convex cone @xmath111 in @xmath99 generated by classes of curves contracted by @xmath4 , that is @xmath112 the contraction @xmath4 is determined ( up to isomorphism ) by @xmath111 , see ( * ? ? ?
* proposition 1.14 ) .
we say that @xmath4 is _ of fiber type _ if @xmath113 , otherwise @xmath4 is birational .
when @xmath4 is of fiber type , we say that @xmath4 is non trivial if @xmath114 .
we denote by @xmath115 the exceptional locus of @xmath4 , i.e. the locus where @xmath4 is not an isomorphism .
we say that @xmath4 is _ divisorial _
if @xmath115 is a divisor , _ small _ if @xmath115 has codimension bigger that @xmath26 .
more generally we say that @xmath4 is _ of type _ ( a , b ) if @xmath116 and @xmath117 .
finally @xmath4 is _ elementary _ if @xmath118
. we will need to work with singular varieties ; we refer the reader to @xcite for the definitions and properties of terminal and canonical singularities .
we say that @xmath0 is @xmath119-factorial if every weil divisor is @xmath119-cartier .
suppose that @xmath0 has canonical singularities ( in particular @xmath120 is @xmath119-cartier ) .
we say that @xmath1 is a _ mori contraction _ if it is a contraction and moreover @xmath121 for every curve @xmath102 contracted by @xmath4 .
we recall two important properties of mori contractions : 1 .
@xmath122 , namely @xmath123 is the linear subspace generated by @xmath111 ; 2 . for any @xmath124 one has @xmath125 if and only if @xmath126 for every curve @xmath102 contracted by @xmath4 .
( see ( * ? ? ? *
theorem 3.7 ( 4 ) ) for the second statement , which implies the first one . )
suppose that @xmath100 is closed and polyhedral . by a _
face _ of @xmath100
we just mean a face in the geometrical sense . for any contraction @xmath4 of @xmath0 , @xmath111 is a face of @xmath100 .
extremal ray _ is a one dimensional face , with no assumptions on the intersection of @xmath120 with its elements .
we will use greek letters @xmath127 , @xmath128 , etc . to denote faces of @xmath100 .
if @xmath127 is an extremal ray and @xmath92 a @xmath119-cartier divisor on @xmath0 , we will say that @xmath129 , @xmath130 , or @xmath131 , if respectively @xmath132 , @xmath133 , or @xmath134 for a non zero element @xmath135 .
we denote by @xmath136 the union of all curves in @xmath0 whose numerical class is in @xmath127 .
let @xmath0 be a projective variety with canonical singularities and @xmath120 cartier .
we say that @xmath0 is _ fano _ if @xmath137 is ample .
if so , the cone @xmath100 is closed and polyhedral , and any contraction of @xmath0 is a mori contraction .
moreover for any face @xmath127 of @xmath100 there exists a contraction @xmath4 of @xmath0 such that @xmath138 .
this follows from the contraction theorem , see ( * ? ? ?
* theorem 3.7 ) . [ wellknown ]
let @xmath0 and @xmath7 be factorial projective varieties , and @xmath139 the blow - up of a smooth subvariety @xmath140 .
suppose that @xmath0 is fano and let @xmath141 be an irreducible curve such that @xmath142 and @xmath143 .
let @xmath144 be the proper transform of @xmath145 in @xmath0 . then : @xmath146 in particular @xmath147 .
in fact if @xmath148 we have @xmath149 and @xmath150 , so @xmath151 .
moreover @xmath152 where @xmath153 , so @xmath154 [ divisorial ] let @xmath0 be a normal and @xmath119-factorial projective variety and @xmath1 an elementary divisorial contraction . then @xmath115 is an irreducible divisor and @xmath155 .
in fact let @xmath156 be an irreducible component of @xmath115 , then @xmath157 ( see for instance ( * ? ? ?
* lemma 3.39 ) applied to @xmath158 ) . now
if @xmath102 is an irreducible curve contracted by @xmath4 , we have @xmath159 , thus @xmath160 . hence @xmath161 . [ lifting ] * targets of contractions of smooth fano varieties .
* let @xmath0 be a smooth fano variety and @xmath1 a contraction .
consider the push - forward @xmath162 .
we observe that @xmath163 namely _
@xmath164 is the linear projection of @xmath100 from the face @xmath111_. this simple remark implies many properties of @xmath164
. for instance it is closed and polyhedral , since @xmath100 is .
moreover , faces of @xmath164 are in bijection ( via @xmath165 ) with faces of @xmath100 containing @xmath111 .
in fact this description is the same as the one involving the `` star of a cone '' in toric geometry , see @xcite .
let s consider a face @xmath127 of @xmath164 , and let @xmath166 be the unique face of @xmath100 containing @xmath111 and such that @xmath167 . then @xmath168 . since @xmath111 is a face of @xmath166 , we can choose another face @xmath169 of @xmath166 with the properties : @xmath170 observe that the choice of @xmath169 will not be unique in general , and that it can very well be @xmath171 . when @xmath127 is an extremal ray ,
@xmath169 is an extremal ray of @xmath100 , and @xmath172 .
there is a rational curve @xmath102 such that @xmath103\in\widetilde{\alpha}$ ] , hence @xmath173 is a rational curve in @xmath7 with numerical class in @xmath127 .
since @xmath0 is fano , there exist contractions @xmath174 and @xmath175 such that @xmath176 and @xmath177 : @xmath178_f\ar[dr]^{h}\ar[r]^{{\varphi } } & w \\ y & z}\ ] ] now by rigidity ( see for instance ( * ? ? ?
* lemma 1.15 ) ) there exist contractions @xmath179 and @xmath180 that make the following diagram commute : @xmath178_f\ar[dr]^{h}\ar[r]^{{\varphi } } & w\ar[d]^g
\\ y\ar[r]_{\psi } & z}\ ] ] it is not difficult to check that @xmath181 and that @xmath182 .
we will say that @xmath174 is a _ lifting _ of @xmath183 .
summing up , we have proved the following .
[ easy ] let @xmath0 be a smooth fano variety and @xmath1 a contraction .
then @xmath164 is a closed polyhedral cone , and every extremal ray contains the class of some rational curve .
moreover for every face @xmath127 of @xmath164 there exists a contraction @xmath179 such that @xmath181 and @xmath184 .
let @xmath0 be a smooth variety and @xmath185 a mori contraction of fiber type . recall that :
@xmath186 let @xmath3 be a general fiber and consider @xmath187 .
we have @xmath188 hence @xmath189 .
[ quasiel ] we say that @xmath4 is _ quasi elementary _ if @xmath190 equivalently if @xmath191 .
since @xmath192 , if @xmath4 is quasi elementary we get : @xmath193 [ compo ] let @xmath1 be a mori contraction of fiber type . 1 . if @xmath4 is elementary , then it is also quasi elementary .
+ this is because @xmath194 .
2 . suppose that @xmath195 .
then @xmath4 is quasi elementary if and only if it is elementary .
3 . suppose that @xmath0 is fano and @xmath4 is quasi elementary .
if @xmath196 is an elementary contraction of fiber type , then the composition @xmath197 is quasi elementary .
+ in fact since @xmath183 is elementary we have @xmath198 .
let @xmath199 be a general fiber of @xmath200 , then @xmath201 hence @xmath202 .
choosing a curve @xmath203 not contracted by @xmath4 shows that @xmath204 , so @xmath205 and @xmath200 is quasi elementary .
let s show that the notion of quasi elementary is related to smoothness .
[ smoothness ] let @xmath0 be a smooth variety and @xmath1 a mori contraction of fiber type .
let @xmath206 be an open subset over which @xmath4 is smooth , and set @xmath207 .
if @xmath208 and @xmath7 is @xmath119-factorial , then @xmath4 is quasi elementary .
we remark that the converse to lemma [ smoothness ] does not hold ( take for instance a smooth fano 3-fold @xmath0 with an elementary contraction @xmath209 which is not a smooth morphism ) .
moreover the hypothesis of @xmath119-factoriality on @xmath7 is necessary , see the contraction @xmath210 in example [ delpezzo ] .
set @xmath211 .
following the notation of @xcite , consider @xmath212 since @xmath213 is a mori contraction , @xmath214 is closed and polyhedral by the relative version of the cone theorem , see ( * ? ? ?
* theorem 4 - 2 - 1 ) .
on the other hand @xmath215 , and the inclusion @xmath216 induces a natural injective homomorphism @xmath217 with a slight abuse of notation , we will consider @xmath218 as a subspace of @xmath99 .
for every @xmath219 we have @xmath220 .
consider an extremal ray @xmath127 of @xmath214 .
then ( * ? ? ?
* proposition 1.3 ) says that @xmath221 dominates @xmath222 via @xmath4 , hence @xmath223 for every @xmath219 .
repeating this for every extremal ray of @xmath214 , we get @xmath224 for every @xmath219 , in particular @xmath225 the homomorphism dual to the inclusion @xmath226 is @xmath227 induced by the restriction @xmath228 ( see ( * ? ? ?
* and lemma 3 - 2 - 5 ( 2 ) ) ) .
clearly @xmath229 is surjective .
let @xmath124 be such that @xmath230 for some @xmath231 .
since @xmath7 is @xmath119-factorial , there exists @xmath232 such that @xmath233 , @xmath234 .
then @xmath235 is trivial on @xmath236 , and by our hypothesis trivial on @xmath0 .
thus @xmath229 is an isomorphism .
dually , this is gives @xmath237 , so @xmath4 is quasi elementary .
observe that shows that @xmath238 does not depend on @xmath219 , so that the condition @xmath239 can be checked for an arbitrary @xmath219 .
if @xmath1 is quasi elementary , then @xmath240 for _ every _ fiber @xmath241 of @xmath4 ( with the reduced structure ) .
in fact we have @xmath242 .
moreover if @xmath243 is as in the proof of lemma [ smoothness ] , @xmath214 has dimension @xmath34 .
for any fixed extremal ray @xmath127 of @xmath214 , ( * ? ? ?
* proposition 1.3 ) says that @xmath221 dominates @xmath222 .
then taking a family of curves whose class is in @xmath127 and their degenerations , we see that @xmath244 .
this implies that @xmath245 .
suppose that @xmath0 is smooth and fano , @xmath1 a contraction of fiber type , and @xmath3 a general fiber .
in general the push - forward @xmath246 does not need to be injective : for instance there are smooth fano threefolds that have an elementary contraction onto @xmath18 , with fibers del pezzo surfaces with @xmath247 .
this is related to the monodromy of the fibration @xmath4 .
consider an open subset @xmath222 as in lemma [ smoothness ] and let @xmath219 .
the dimension of @xmath238 is equal to the dimension of the image of the restriction @xmath248 in turn this is equal to the dimension of the linear subspace of @xmath249 which is invariant for the monodromy action of @xmath250 ( see for instance ( * ? ? ?
* chapter 15 ) ) .
hence @xmath251 if and only if the monodromy action is trivial .
let @xmath0 be a smooth fano variety , @xmath1 a non trivial contraction of fiber type with @xmath7 smooth , and @xmath3 a general fiber .
suppose that @xmath4 is smooth outside a finite number of points of @xmath7 , and that there are no fibers of codimension @xmath26 .
then @xmath7 is fano , @xmath4 is quasi elementary , and @xmath252 .
in fact @xmath7 is fano by ( * ? ? ? * theorem 3 ) , in particular it is simply connected , and @xmath253 .
then @xmath254 stays simply connected , so the monodromy action on @xmath255 is trivial , and @xmath256 . on the other hand
@xmath4 is quasi elementary by lemma [ smoothness ] , so @xmath257 .
the following two lemmas give some basic properties of quasi elementary contractions .
[ nonequilocus ] let @xmath0 be a smooth variety and @xmath1 a quasi elementary mori contraction of fiber type . 1 . if @xmath92 is a prime divisor in @xmath0 such that @xmath258 , then @xmath259 is a cartier divisor , and @xmath260 .
the locus where @xmath4 is not equidimensional has codimension at least @xmath15 in @xmath7 .
since @xmath4 is quasi elementary and @xmath92 is disjoint from the general fiber , we have @xmath261 for every curve @xmath102 contracted by @xmath4 . by property ( [ prel ] . ) of mori contractions , there exists an effective cartier divisor @xmath262 on @xmath7 such that @xmath263 .
moreover @xmath264 is supported on @xmath259 , so there exists @xmath265 such that @xmath266 .
set @xmath267 and observe that @xmath268 has codimension at least two in @xmath0 . in @xmath269 the intersection @xmath270 is a ( non trivial ) cartier divisor and @xmath271 .
since @xmath0 is smooth , there exists an effective divisor @xmath272 in @xmath0 such that @xmath273 . then @xmath274 , so @xmath275 and @xmath276 because @xmath92 is prime . let s show @xmath96 .
let @xmath277 be an irreducible closed subset such that the general fiber of @xmath4 over @xmath278 has dimension at least @xmath279 .
then @xmath280 so @xmath281 . if by contradiction @xmath282 , then @xmath283 . consider a prime divisor @xmath92 contained in @xmath284 : we have @xmath285 , which contradicts @xmath50 .
we generalize to quasi elementary contractions some known properties of elementary contractions of fiber type . in particular @xmath96
is shown in @xcite in the elementary case .
[ target ] let @xmath1 be a mori contraction of fiber type , with @xmath0 smooth . 1 . if @xmath4 is quasi elementary , then @xmath7 is factorial and has canonical singularities .
if @xmath7 is a surface and @xmath4 is equidimensional , then @xmath7 is smooth .
if @xmath7 is a surface and @xmath4 is quasi elementary , then @xmath4 is equidimensional and @xmath7 is smooth .
4 . if @xmath44 and @xmath4 is quasi elementary , then @xmath7 has isolated singularities .
@xmath50 let @xmath92 be a prime weil divisor in @xmath7 and let @xmath262 be a prime divisor in @xmath0 such that @xmath286 .
then lemma [ nonequilocus ] @xmath50 yields that @xmath92 is cartier .
thus @xmath7 is factorial , in particular @xmath287 is cartier .
it is known that @xmath7 has rational singularities , see ( * ? ? ?
* corollary 7.4 ) .
then @xmath7 has also canonical singularities , see ( * ? ? ?
* corollary 5.24 ) .
@xmath96 when @xmath4 is elementary , this is ( * ? ? ?
* proposition 1.4.1 ) . in general , observe that @xmath7 is a normal surface with rational singularities , in particular it is @xmath119-factorial and has isolated singularities .
we want to show that in fact @xmath7 has quotient singularities , using results from @xcite and @xcite .
more precisely , let @xmath288 be a singular point . in ( * ? ?
* definition 1.4 ) and ( * ? ? ?
* definition 1.25 ) one can find the definition of the plurigenera @xmath289 of @xmath7 in @xmath290 , for @xmath291 .
since @xmath4 is equidimensional , then it is `` non degenerate '' in the sense of ( * ? ?
* definition 1.14 ) .
then ( * ? ?
? * corollary 1.27 ) gives @xmath292 for every @xmath265 .
this is equivalent to saying that @xmath290 is a quotient singularity by ( * ? ? ?
* theorem 3.9 ) .
hence @xmath7 has quotient singularities , and we can apply ( * ? ? ?
* proposition 1.4 ) to deduce that @xmath7 is actually smooth . observe that in @xcite the contraction @xmath4 is assumed to be elementary , but the proof works word for word in the case of an equidimensional mori contraction .
let us also remark that by the definition of quotient singularities , we can cover @xmath7 by open subsets in the complex topology which are quotients of an open subset of @xmath293 by a finite group .
hence in the proof of ( * ? ? ?
* proposition 1.4 ) one has actually to work in the analytic category , however everything works in the same way .
@xmath294 this follows immediately from lemma [ nonequilocus ] and @xmath96 .
@xmath295 by lemma [ nonequilocus ] , @xmath4 can have at most isolated fibers of dimension @xmath296 .
let @xmath297 be a general hyperplane section and @xmath298 .
then @xmath13 is a normal surface ( see ( * ? ? ?
* lemma 5.30 ) ) and @xmath299 is equidimensional .
moreover @xmath92 is general in a base point free linear system , so it is smooth . since @xmath92 is disjoint from the general fiber of @xmath4 and @xmath4 is quasi elementary , we have @xmath261 for every curve @xmath102 contracted by @xmath4 . in particular
if @xmath300 we get @xmath301 so @xmath302 is an equidimensional mori contraction .
then @xmath13 is smooth by @xmath96 , and this yields @xmath303 .
thus the target of a quasi elementary contraction has reasonable singularities .
the following simple remark will be very useful .
[ fano ] let @xmath0 be a smooth fano variety and @xmath1 a quasi elementary contraction , so that @xmath7 is factorial and has canonical singularities by lemma [ target ] .
then @xmath7 is fano if and only if @xmath304 for every extremal ray @xmath127 of @xmath164 .
equivalently , @xmath7 is fano if and only if every elementary contraction @xmath179 is a mori contraction .
this is a straightforward consequence of lemma [ easy ] .
[ boh ] * elementary contractions of the target .
* let @xmath0 be a smooth projective variety and @xmath1 a quasi elementary mori contraction of fiber type .
recall that @xmath7 is factorial with canonical singularities by lemma [ target ] .
let @xmath179 be a contraction .
we say that _ @xmath183 has a lifting _ if there exist a mori contraction @xmath305 such that @xmath306 , @xmath307 , and a commutative diagram : @xmath308_f \ar[r]^{{\varphi } } & w\ar[d]^g \\
y\ar[r]_{\psi } & z}\ ] ] we will also say that @xmath64 is a _ lifting _ of @xmath183 .
notice that @xmath183 is elementary if and only if @xmath64 is elementary .
we are interested in comparing properties of @xmath64 and @xmath183 .
when @xmath0 is fano , any contraction @xmath179 has a lifting , as explained in [ lifting ] .
conversely , in theorems [ face ] and [ face2 ] we will give conditions on a mori contraction @xmath174 to be a lifting of some @xmath183 .
[ secondo ] let @xmath0 be a smooth projective variety and @xmath1 a quasi elementary mori contraction of fiber type .
let @xmath179 be a birational elementary contraction with fibers of dimension at most @xmath26 .
assume that @xmath183 has a lifting @xmath174 as in .
then the following holds : 1 .
@xmath48 is smooth and @xmath64 is the blow - up of a smooth subvariety of codimension @xmath309 ; 2 .
@xmath310 is a quasi elementary mori contraction and @xmath97 is factorial with canonical singularities ; 3 .
@xmath183 is divisorial , @xmath311 , and @xmath312 .
4 . if @xmath44 , then @xmath183 is the blow - up of a smooth curve @xmath313 .
5 . if @xmath40 , then : 1 . @xmath7 and @xmath97 are smooth , @xmath183 is the blow - up of @xmath314 , and @xmath310 has smooth fiber @xmath241 over @xmath315 ; 2 .
@xmath64 is the blow - up of @xmath241 , @xmath316 , and @xmath317 , @xmath318 are the two projections ; 3 . if @xmath0 is fano , then @xmath48 is fano .
let @xmath3 be a non trivial fiber of @xmath64 .
then @xmath4 is finite on @xmath3 and @xmath319 is contained in a non trivial fiber of @xmath183 , hence @xmath320 and @xmath321 . since @xmath322 is covered by non trivial fibers of @xmath64 , we get @xmath323 .
in particular @xmath64 is birational with fibers of dimension at most @xmath26 , so ( * ? ? ?
* theorem 1.2 ) yields that @xmath48 is smooth and @xmath64 is the blow - up of a smooth subvariety of codimension @xmath309 .
let @xmath324 be the exceptional divisor of @xmath64 , then @xmath325 is contained in @xmath326 and it is a divisor by lemma [ nonequilocus ] @xmath50 .
hence @xmath183 is elementary and divisorial , and @xmath326 is irreducible by remark [ divisorial ] , that is @xmath327
. then again lemma [ nonequilocus ] @xmath50 gives @xmath328 .
we have @xmath329 let @xmath145 be an irreducible curve contracted by @xmath310 and @xmath144 an irreducible curve in @xmath0 such that @xmath330 . then @xmath331 with @xmath265 .
moreover @xmath332 is a point , so @xmath333 .
then : @xmath334 so @xmath310 is a mori contraction .
we have @xmath335 .
moreover @xmath336 is an extremal ray of @xmath100 not contained in @xmath111 , hence @xmath337
. then @xmath338 , on the other hand @xmath339 , so equality holds .
observe that @xmath64 is an isomorphism over the general fibers of @xmath4 and @xmath310 .
let @xmath340 be a general fiber of @xmath4 , so that @xmath341 because @xmath4 is quasi elementary .
then @xmath342 is a general fiber of @xmath310 , and @xmath343 .
thus @xmath310 is quasi elementary
. then @xmath97 is factorial with canonical singularities by lemma [ target ] @xmath50 .
let s show that @xmath344 .
write @xmath345 with @xmath346 , and let @xmath347 be a curve contracted by @xmath183 .
then @xmath348 by remark [ divisorial ] , and @xmath349 .
hence we have to show that @xmath350 .
let @xmath351 be a resolution of singularities of @xmath7 , and consider the composition @xmath352 .
call @xmath353 the proper transform of @xmath326 in @xmath354 .
then @xmath229 is the coefficient of @xmath353 in @xmath355 , thus @xmath350 because @xmath97 has canonical singularities .
assume now that @xmath44 and set @xmath356 .
let s first notice that @xmath357 .
this is because if @xmath358 , then @xmath290 must be smooth for @xmath7 too , because @xmath13 is a cartier divisor in @xmath7 .
on the other hand let @xmath359 and let @xmath360 be a local equation for @xmath13 in @xmath361 . since @xmath362 , @xmath363 is a local equation for @xmath156 near the fiber over @xmath361 . take @xmath364 in this fiber .
since @xmath156 is smooth , the differential @xmath365 is non zero , and since @xmath366 , the differential @xmath367 must be non zero too .
thus @xmath13 is smooth at @xmath361 .
recall that @xmath7 has rational singularities , in particular it is cohen - macaulay ( see ( * ? ? ?
* corollary 7.4 ) and ( * ? ? ? * theorem 5.10 ) ) . since @xmath13 is a cartier divisor in @xmath7 , it is cohen - macaulay too . on the other hand , lemma [ target ]
@xmath295 implies that @xmath7 and @xmath13 have isolated singularities .
then @xmath13 is normal by serre s criterion . by lemma [ nonequilocus ]
@xmath4 can have at most isolated fibers of dimension @xmath296 .
let s show that @xmath4 is equidimensional over @xmath13 . if @xmath368 is an irreducible component of a fiber of @xmath4 with @xmath369 , then @xmath370 and @xmath371 , hence @xmath372 .
this gives @xmath373 a contradiction .
now @xmath156 is smooth , @xmath13 is normal , and @xmath374 has connected fibers . as in the proof of lemma [ target ] @xmath295 we see that @xmath375 is a mori contraction , and by lemma [ target ] @xmath96 the surface @xmath13 is smooth , so that @xmath376 .
the general fiber @xmath377 of @xmath378 is a rational curve , and it is smooth because @xmath13 is , hence @xmath379 .
let @xmath380 be a fiber of @xmath64 such that @xmath381 .
then @xmath382 which gives @xmath383 and @xmath384 .
this shows also that @xmath385 is a @xmath18-bundle .
therefore @xmath183 is the blow - up of a smooth curve contained in the smooth locus of @xmath97 , and @xmath97 has the same singularities as @xmath7 . finally , let s suppose that @xmath40
. then @xmath7 and @xmath97 are smooth by lemma [ target ] @xmath294 , and @xmath4 and @xmath310 are equidimensional . hence @xmath183 is the blow - up of a point @xmath314 .
we have @xmath386 so the center of @xmath64 is @xmath387 .
since @xmath328 , @xmath156 is the ( schematic ) fiber of @xmath200 over @xmath315 , and it is reduced . thus @xmath388 has reduced fiber over @xmath315 , and the same must hold for @xmath310 .
now @xmath310 is an equidimensional morphism between smooth varieties , whose fiber @xmath241 over @xmath315 is reduced and smooth .
this implies that @xmath310 is smooth over @xmath315 , and hence the normal bundle of @xmath241 in @xmath48 is trivial .
since @xmath156 is the exceptional divisor of the blow - up of @xmath241 , we deduce that @xmath389 , and @xmath375 and @xmath390 are the two projections .
suppose that @xmath0 is fano and @xmath48 is not .
then ( * ? ? ?
* proposition 3.4 ) says that there exists an extremal ray @xmath127 of @xmath100 , different from @xmath336 , such that @xmath391 .
this implies that @xmath127 is not contained in @xmath111 , because @xmath392 for every curve @xmath145 contracted by @xmath4 .
however @xmath393 , so every curve in @xmath156 is numerically equivalent to a linear combination with coefficients in @xmath394 of a curve in @xmath111 and a curve in @xmath336 .
thus no curve in @xmath156 can have numerical class in an extremal ray @xmath127 not contained in @xmath395 , and we have a contradiction .
[ div ] * divisors @xmath396 with small @xmath397 . *
let @xmath0 be a smooth fano variety of dimension @xmath11 , and @xmath92 a prime divisor in @xmath0 .
if @xmath92 is simple , e.g. if @xmath398 is very small , then one can hope to deduce informations on @xmath0 itself .
for instance in @xcite the authors classify the possible pairs @xmath399 when @xmath400 and @xmath401 .
this is equivalent to asking that @xmath0 is obtained by blowing - up a smooth variety in a point .
this classification has been generalized in @xcite to the case @xmath400 and @xmath402 with @xmath403 . with the same techniques (
* proposition 5 ) shows that if @xmath0 contains a prime divisor @xmath92 with @xmath93 and @xmath404 , then @xmath405 .
the proof of ( * ? ? ?
* proposition 5 ) can be generalized in order to obtain the following .
[ divisor ] let @xmath0 be a smooth fano variety , and let @xmath1 be either a quasi elementary contraction of fiber type or an isomorphism .
suppose that @xmath91 and that there exists a prime divisor @xmath406 such that @xmath407
. then @xmath94 . in the case where @xmath4 is an isomorphism
, this says that if @xmath404 and @xmath0 contains a prime divisor @xmath92 with @xmath408 , then @xmath405 .
in particular if @xmath409 and @xmath0 has an elementary contraction of type @xmath410 , then @xmath405 .
recall that @xmath7 is factorial by lemma [ target ] @xmath50 .
moreover @xmath164 is closed and polyhedral , and every extremal ray can be contracted , by lemma [ easy ] .
let s assume that @xmath411 .
there exists at least one extremal ray @xmath412 of @xmath164 such that @xmath413 , let @xmath414 be its contraction .
then @xmath415 and @xmath92 must intersect every non trivial fiber of @xmath416 .
if @xmath417 , then every curve in @xmath92 has numerical class contained in @xmath412 , hence @xmath418 is a point . then @xmath416 can not be of fiber type ( otherwise @xmath419 is a point ) , so @xmath416 is birational and @xmath420 .
this implies @xmath421 by remark [ divisorial ] , a contradiction . hence @xmath422 .
consider a non trivial fiber @xmath3 of @xmath416 .
then @xmath423 , on the other hand @xmath424 otherwise we would get a curve in @xmath92 with numerical class in @xmath412 . since @xmath7 is factorial , we get @xmath425 so @xmath3 has dimension one . if @xmath416 is of fiber type , then @xmath426 is surjective , so @xmath427 and @xmath428 .
assume that @xmath416 is birational and consider a lifting of @xmath416 as in [ lifting ] : @xmath429^h \ar[r]^{{\varphi}_1}\ar[d]_f & { w_1}\ar[d]^g \\ y\ar[r]_{\psi_1 } & { z_1 } } \ ] ] since @xmath0 is fano , @xmath430 is a mori contraction , and theorem [ secondo ] applies .
thus @xmath310 is a quasi elementary mori contraction , @xmath431 is smooth , @xmath430 is a blow - up , @xmath416 is divisorial , @xmath419 is factorial , and @xmath432 . since @xmath416 is finite on @xmath92
, @xmath433 is a cartier divisor in @xmath419 .
moreover @xmath434 is surjective , hence @xmath435 .
recall also that @xmath92 intersects every non trivial fiber of @xmath416 , hence @xmath436 .
again by lemma [ easy ] , @xmath437 is closed and polyhedral , and every extremal ray can be contracted .
thus consider an extremal ray @xmath438 of @xmath437 such that @xmath439 , and let @xmath440 be the associated contraction .
as before , we see that if @xmath441 is of fiber type then @xmath442 and @xmath94 .
if @xmath441 is birational , then @xmath443 and again every non trivial fiber of @xmath441 has dimension one .
we show that this case leads to a contradiction .
let @xmath444 be the unique extremal face of @xmath100 such that @xmath444 contains @xmath445 and @xmath446 ( see [ lifting ] ) .
notice that @xmath447 is an extremal ray of @xmath445 .
let s make an easy remark on convex polyhedral cones .
if @xmath448 is a convex polyhedral cone , @xmath449 is a proper face of @xmath450 , and @xmath451 is an extremal ray of @xmath449 , then there exists an extremal ray @xmath452 of @xmath450 , not contained in @xmath449 , and such that @xmath453 is a face of @xmath450 .
the reader who is not familiar with convex geometry may easily prove this by induction on @xmath454 .
applied to our situation , this says that there is an extremal ray @xmath455 of @xmath444 such that @xmath456 and @xmath457 is a face of @xmath444 .
then @xmath458 is an extremal ray of @xmath459 , whose contraction @xmath460 yields a commutative diagram : @xmath461^h\ar[r]^{{\varphi}_1}\ar[d]_f & { w_1}\ar[d]^g\ar[r]^{{\varphi}_2 } & { w_2}\ar[d ] \\
y\ar[r]_{\psi_1 } & { z_1}\ar[r]_{\psi_2 } & { z_2 } } \ ] ] observe that @xmath462 and @xmath463 , so the morphism @xmath464 exists by rigidity , see [ lifting ] .
since @xmath441 is birational and has fibers of dimension at most one , it is easy to see that the same holds for @xmath465 .
suppose that there exists a non trivial fiber @xmath145 of @xmath465 contained in @xmath466 .
let @xmath144 be an irreducible curve in @xmath467 such that @xmath468
. then @xmath469 and @xmath470 .
on the other hand , since @xmath471 is a point , @xmath472 must also be a point , namely the numerical class of @xmath473 must be in @xmath474 .
but this contradicts @xmath443 .
therefore if @xmath145 is a non trivial fiber of @xmath465 , we have @xmath475 , and this implies that @xmath476 .
thus @xmath465 is a mori contraction , and theorem [ secondo ] yields that @xmath477 is smooth , @xmath465 is the blow - up of a smooth subvariety of codimension @xmath309 in @xmath477 , and @xmath478 . in particular @xmath479 . now using remark [ wellknown ] we see that any non trivial fiber @xmath145 of @xmath465 can not intersect @xmath466 , hence : @xmath480 recall that @xmath432 , which gives @xmath481 .
therefore @xmath482 namely @xmath483 .
let s notice that @xmath484 , so there exists a curve @xmath485 , and @xmath486 . on the other hand @xmath487 and @xmath435 .
this implies that @xmath488 for every curve @xmath489 .
but this is impossible , because @xmath490 and @xmath491 are distinct prime divisors with non empty intersection .
[ dettaglio ] more precisely we have shown that in the hypotheses of proposition [ divisor ] , either @xmath492 , or one of the following occurs : 1 .
@xmath493 and @xmath7 has an elementary contraction of fiber type with @xmath26-dimensional fibers ; 2 .
@xmath494 and @xmath7 has a divisorial elementary contraction such that every fiber has dimension at most @xmath26 .
let @xmath0 be any normal projective variety .
consider an irreducible closed subset @xmath495 of @xmath496 such that : 1 . for @xmath497 general , the corresponding cycle @xmath498 is an irreducible , reduced , and connected rational curve ; 2 .
every point of @xmath0 is contained in @xmath499 for some @xmath497 .
we call @xmath495 a _ covering family of rational curves _ in @xmath0 .
such family induces an equivalence relation on @xmath0 ( as a set ) , called @xmath495-equivalence , as follows .
two points @xmath500 are @xmath495-equivalent if there exist @xmath501 such that @xmath502 is connected and contains @xmath503 and @xmath504 .
this notion was originally introduced by campana @xcite and is by now well known , see @xcite , @xcite , @xcite .
we refer specifically to @xcite for the set up and for precise references .
in particular it is known that there exist : an open subset @xmath236 which is closed under @xmath495-equivalence , a normal quasi - projective variety @xmath222 , and a proper , equidimensional morphism @xmath505 , such that every fiber of @xmath506 is a @xmath495-equivalence class . in general
, there are no morphisms defined on the whole @xmath0 which extend @xmath506 .
if @xmath1 is a mori contraction of fiber type , then one can find a family @xmath495 as above such that @xmath507 ( see the proof of theorem [ face ] ) .
using the properties of this family , we can apply the results of @xcite to deduce the following .
[ face ] let @xmath0 be a smooth variety of dimension @xmath11 and consider two mori contractions @xmath508_f\ar[r]^{{\varphi } } & w\\ y & } \ ] ] where @xmath4 is elementary of fiber type and @xmath509 .
let @xmath510 and @xmath511 be the dimensions of the general fibers of @xmath4 and @xmath512 respectively .
assume that we are in one of the following situations : 1 .
@xmath513 is quasi elementary of fiber type , and @xmath514 ; 2 .
@xmath513 is elementary and divisorial , and @xmath5 ; 3 .
@xmath513 is elementary and divisorial , @xmath65 , and @xmath515 .
then there exists a commutative diagram : @xmath516^{{\varphi}}\ar[d]_f\ar[dr]^{h } & w\ar[d]^g \\
y \ar[r ] & z } \ ] ] where @xmath180 is an elementary mori contraction and @xmath517 .
let us point out that we do not know whether the hypotheses on @xmath518 , @xmath519 , and @xmath511 respectively are really necessary for the statement to hold . in the first part of the proof we will just assume that @xmath4 is quasi elementary and that @xmath509 .
we first construct a suitable covering family @xmath495 of rational curves in @xmath0 , such that @xmath4 is the quotient for @xmath495-equivalence over an open subset of @xmath0 .
let @xmath3 be a general fiber of @xmath4 , then @xmath3 is a smooth fano variety of dimension @xmath510 . in particular
@xmath3 is rationally connected , and there exists a smooth rational curve @xmath520 which is very free , namely with ample normal bundle in @xmath3 : @xmath521 ( see ( * ? ? ? * and 5.6 ) ) .
since @xmath522 is trivial , we have @xmath523 hence @xmath524 is a free curve in @xmath0 .
this means that the deformations of @xmath524 cover the whole @xmath0 : by ( * ? ? ?
* proposition 4.8 ) there exists a covering family @xmath495 of rational curves in @xmath0 such that @xmath525 for some @xmath526 .
clearly all curves parametrized by @xmath495 are numerically equivalent in @xmath0 , and their numerical class is contained in @xmath111 , because @xmath527 is a point .
since @xmath111 is a face of @xmath100 , we deduce that every irreducible component of every curve parametrized by @xmath495 is contained in some fiber of @xmath4 .
this implies that every @xmath495-equivalence class is contained in some fiber of @xmath4 . consider the quotient for @xmath495-equivalence
@xmath505 , let @xmath241 be a general fiber , and let @xmath528 general
. then @xmath529 and we have : @xmath530 on the other hand , by ( * ? ? ?
* ii.3.9.2 ) the number of trivial summands in @xmath531 is at most the number of trivial summands in @xmath532 , so that @xmath533 and @xmath534 .
this implies that @xmath535 and hence all fibers of @xmath4 of dimension @xmath510 are @xmath495-equivalence classes .
we observe that @xmath495 induces in a natural way a covering family of rational curves on @xmath48 .
first of all , the hypothesis @xmath536 implies that @xmath513 does not contract any irreducible component of any curve of @xmath495 .
consider now the incidence diagram associated to @xmath495 : @xmath537\ar[r ] & x \\ v & } \ ] ] we proceed as in ( * ? ? ?
* proof of lemma 2 ) .
let @xmath538 be the normalization of @xmath539 and @xmath540 be the stein factorization of the composite map @xmath541 .
then @xmath542 is normal , the general fiber of @xmath543 is @xmath18 , and the composite map @xmath544&{\mathcal{c}}\ar[r]&x\ar[r]^{{\varphi } } & w } \ ] ] yields a family of @xmath26-cycles in @xmath48 , thus a morphism @xmath545 .
we call @xmath546 the image of this morphism . by the construction , for every irreducible component @xmath145 of a curve parametrized by @xmath495 ,
the image @xmath547 is a component of some curve parametrized by @xmath546 , and conversely . in particular ,
if @xmath500 are @xmath495-equivalent , then @xmath548 are @xmath546-equivalent . [ dimension ]
let @xmath549 be a general @xmath546-equivalence class .
then @xmath550 , and @xmath551 if @xmath65 .
observe that the second case always holds if @xmath64 is of fiber type .
the inverse image @xmath552 is closed for @xmath495-equivalence .
let @xmath3 be a general @xmath495-equivalence class contained in @xmath552 , so that @xmath3 is a fiber of @xmath4 . since @xmath513 is finite on @xmath3 and @xmath553 , we have @xmath554 , while @xmath555 , which gives the first statement .
if @xmath64 is of fiber type , then @xmath556 , so the same argument gives @xmath551 .
let s assume that @xmath64 is birational and that @xmath322 dominates @xmath7 via @xmath4 .
then @xmath557 , so @xmath558 , and since @xmath559 is general , every @xmath546-equivalence class intersects @xmath560 .
this means that @xmath552 contains a general fiber @xmath561 of @xmath318 .
let @xmath562 be an irreducible component of @xmath561 with @xmath563 .
there exists a non empty open subset @xmath564 of @xmath565 such that every fiber of @xmath4 over @xmath564 is an irreducible @xmath495-equivalence class
. then @xmath566 is irreducible of dimension @xmath567 , and it must be contained in @xmath568 because @xmath552 is closed for @xmath495-equivalence .
moreover @xmath569 is birational , so @xmath570 .
consider now the linear subspaces @xmath571 of @xmath99 and @xmath572 of @xmath573 generated by the numerical classes of all irreducible components of all curves in @xmath495 and @xmath546 respectively .
then we have @xmath574 , moreover @xmath575 and @xmath576 by ( * ? ? ?
* proposition iv.3.13.3 ) ( see also ( * ? ? ?
* remark 1 ) ) . since @xmath4 is quasi elementary , we get @xmath577 , hence @xmath578 .
then we have : @xmath579 we observe that @xmath513 is either elementary and divisorial , or quasi elementary of fiber type .
then in any case @xmath48 is @xmath119-factorial and has canonical singularities , by lemma [ target ] and ( * ? ? ?
* proposition 7.44 ) .
let s assume now that @xmath4 is an elementary contraction .
then @xmath580 , and this means that @xmath495 and @xmath546 are quasi - unsplit in the sense of @xcite .
moreover , using the claim we see that in any case the general @xmath546-equivalence class has codimension at most @xmath15 .
hence ( * ? ? ?
* theorem 2 ) yields the existence of an elementary mori contraction @xmath180 such that every fiber of @xmath310 is a @xmath546-equivalence class .
then @xmath581 , and by rigidity ( see for instance ( * ? ? ?
* lemma 1.15 ) ) there exists a morphism @xmath582 as in the statement .
we observe that with a slight modification of the argument , we can prove a different version of the previous theorem .
namely we can allow @xmath4 to be quasi elementary instead of elementary , if we impose a stronger condition on @xmath510 and @xmath511 .
[ face2 ] let @xmath0 be a smooth variety of dimension @xmath11 and consider two mori contractions @xmath1 , @xmath174 such that @xmath509 and @xmath4 is quasi elementary of fiber type .
let @xmath510 and @xmath511 be the dimensions of the general fibers of @xmath4 and @xmath512 respectively .
assume that we are in one of the following situations : 1 .
@xmath513 is quasi elementary and @xmath583 ; 2 .
@xmath513 is elementary and divisorial , and @xmath584 ; 3 .
@xmath513 is elementary and divisorial , @xmath65 , and @xmath585 .
then there exists a commutative diagram as where @xmath175 is a contraction , @xmath586 , and @xmath587 ( equality holds except possibly in @xmath50 ) .
we perform the same construction as in the previous proof , so that @xmath48 is a normal and @xmath119-factorial projective variety , and @xmath546 is a covering family of rational curves in @xmath48 . moreover by the claim and the assumptions , we see that now the general @xmath546-equivalence class has codimension at most @xmath309 in @xmath48 . on the other hand ,
@xmath546 is not quasi - unsplit if @xmath4 is not elementary .
however implies the following property .
_ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath92 be a weil divisor in @xmath48 whose support is disjoint from the general @xmath546-equivalence class .
then @xmath261 for every irreducible component of every curve in @xmath546 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we claim that we can apply ( * ? ? ?
* proposition 1 ) to the family @xmath546 on @xmath48 , even if @xmath546 is not quasi - unsplit .
indeed the quasi - unsplit assumption is used in the proof of this proposition uniquely to deduce the property above .
one can think that the property above generalizes the quasi - unsplit property , in the same way as quasi elementary contractions generalize elementary contractions of fiber type .
now as in the proof of ( * ? ? ?
* theorem 1 ) , we get the existence of a normal projective variety @xmath97 and a surjective morphism @xmath180 such that every fiber of @xmath310 is a @xmath546-equivalence class ( even if @xmath310 is not necessarily a mori contraction ) . as before this implies the existence of the diagram by rigidity , and @xmath588 is a contraction .
moreover it is clear that @xmath589 and @xmath590 observe that the hypothesis @xmath591 does not imply @xmath592 , unless we know that one of @xmath4 or @xmath64 is elementary , as in @xmath96 or @xmath294 .
thus @xmath593 , and equality holds except possibly in @xmath50 .
[ dim ] in both theorems [ face ] and [ face2 ] , we have in fact proved that @xmath594 in case @xmath96 , and @xmath595 in cases @xmath50 and @xmath294 .
thus the contraction @xmath582 is of fiber type in case @xmath50 , and elementary of fiber type in case @xmath294 .
in this section we show theorem [ result ] @xmath50 .
so let s consider a smooth fano variety @xmath0 of dimension @xmath596 , and a quasi elementary contraction @xmath1 onto a surface .
we know by lemma [ target ] @xmath294 that @xmath4 is equidimensional and @xmath7 is smooth .
moreover @xmath7 is rational , for instance because @xmath0 is rationally connected .
[ nuoto ] let @xmath179 be a birational elementary contraction .
then @xmath97 is smooth , @xmath183 is the blow - up of a point @xmath597 , and there exists a diagram @xmath598_f \ar[r]^{{\varphi } } & w\ar[d]^g \\
y \ar[r]^{\psi } & { z}}\ ] ] where @xmath48 is a smooth fano variety , @xmath310 a quasi elementary contraction with smooth fiber @xmath241 over @xmath315 , @xmath64 is the blow - up of @xmath241 , @xmath316 , and @xmath317 , @xmath318 are the two projections .
in fact @xmath183 has a lifting as explained in [ lifting ] , which gives the diagram above . since @xmath0 is fano , @xmath64 is a mori contraction
. then theorem [ secondo ] @xmath599 gives the statement .
notice moreover that the lifting @xmath64 is uniquely determined by @xmath183 , because @xmath600 and @xmath336 is determined by the curves @xmath601 in @xmath0 . as described in [ lifting ] , every choice of an extremal ray @xmath169 in @xmath602 such that @xmath603 , gives rise to a lifting of @xmath183 .
we deduce that such an extremal ray @xmath176 is unique , and that @xmath604 .
if @xmath492 then @xmath605 , so we can assume that @xmath411 . in order to show that @xmath7 is del pezzo , it is enough to show that any elementary contraction of @xmath7 is a mori contraction , see remark [ fano ] .
let s fix such a contraction @xmath179 .
if @xmath183 is birational , then remark [ nuoto ] says that @xmath97 is smooth and @xmath183 is a blow - up , hence a mori contraction . let s consider the case where @xmath183 is not birational , so that @xmath606 .
since we have assumed @xmath411 , we have @xmath607 , and @xmath7 is a smooth rational surface with @xmath493 . thus @xmath7 is a hirzebruch surface and @xmath608 . if we consider the other elementary contraction of @xmath7 , we find two possibilities : either it is again a contraction over @xmath18 and @xmath609 , or it is birational . in this case , by remark [ nuoto ] it must be a smooth blow - up , so @xmath610 .
hence @xmath7 is a del pezzo surface , and we have the first part of the statement .
let s assume that @xmath42 and prove the second part of the statement .
since @xmath7 is del pezzo , there is a morphism @xmath611 which is a blow - up of @xmath612 distinct points @xmath613 .
thus we can apply remark [ nuoto ] to each blow - up and get a smooth fano variety @xmath236 with a quasi elementary contraction @xmath614 , such that @xmath0 is obtained from @xmath236 by blowing - up the fibers of @xmath213 over @xmath613 .
therefore it is enough to prove the statement in the case @xmath494 , where @xmath7 is the blow - up of @xmath615 in two points @xmath616 and @xmath617 . call @xmath618 and @xmath619 the two corresponding exceptional curves , and @xmath620 the proper transform in @xmath7 of the line through @xmath616 and @xmath617 in @xmath25 .
these are all the @xmath621-curves in @xmath7 .
call @xmath622 the extremal ray of @xmath164 containing the numerical class of @xmath623 ; then @xmath624 is simplicial .
let s make a preliminary remark on the cone @xmath100 .
the surface @xmath7 has three elementary contractions , which are blow - ups with exceptional curves @xmath618 , @xmath619 , and @xmath620 . recall from remark [ nuoto ] that the lifting of each of these blow - ups is unique .
this means that for every @xmath625 there is a unique extremal ray @xmath626 of @xmath100 such that @xmath627 is a face of @xmath100 and @xmath628 .
let s show that @xmath0 has no small elementary contraction @xmath629 such that @xmath630 . by contradiction ,
let @xmath629 be such a contraction .
since @xmath4 is finite on fibers of @xmath631 , these fibers have dimension at most @xmath309 .
if @xmath241 is an irreducible component of a fiber of @xmath631 with @xmath632 , then @xmath633 , so @xmath634 is surjective .
this is impossible because @xmath635 while @xmath494 . on the other hand
, a small contraction can not have only @xmath26-dimensional fibers , see ( * ? ? ?
* corollary on p. 145 ) .
thus we have a contradiction .
consider now an extremal ray @xmath128 of @xmath100 which is not contained in @xmath111 . since the contraction of @xmath128 is not small , theorem [ face2 ] @xmath96 implies that there is face @xmath445 ( @xmath636 as in ) of @xmath100 , of dimension @xmath637 , containing both @xmath128 and @xmath111 .
then @xmath638 is an extremal ray of @xmath164 , say @xmath412 .
this implies that @xmath639 . summing up
, we have shown that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ @xmath640 , @xmath641 , and @xmath642 are the only extremal rays of @xmath100 outside @xmath111_. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let s show that @xmath0 is a product . applying remark [ nuoto ] we get the following sequence : @xmath598_f\ar[r]_{{\varphi}_1}\ar@/^1pc/[rr]^{\zeta } & { w_1}\ar[d]\ar[r ] & { x_0 } \ar[d]^{f_0}\\ y\ar[r]_{\psi_1 } & { \mathbb{f}_1}\ar[r ] & { { \mathbb{p}}^2 } } \ ] ] where @xmath643 is the contraction of @xmath618 and @xmath644 is the contraction of ( the image of ) @xmath619 , hence @xmath645 is the contraction of @xmath646 . since @xmath647
, @xmath648 is a proper face of @xmath649 , and there is at least one extremal ray @xmath128 of @xmath649 outside @xmath648 . as explained in [ lifting ] , @xmath128 must be the image via @xmath650 of some extremal ray @xmath455 of @xmath100 not lying in @xmath651 .
since @xmath652 and @xmath653 , it must be @xmath654 .
thus the only possibility is @xmath655 and @xmath656 .
let @xmath657 be the contraction of @xmath128 .
observe that @xmath658 , because @xmath636 is finite on fibers of @xmath213 .
our goal is to show that @xmath659 and @xmath213 is the projection , which implies the statement .
the first step is to show that @xmath660 . for @xmath625 let @xmath661 be the smooth blow - up obtained by contracting @xmath626 .
let @xmath662 be the exceptional divisor of @xmath663 , @xmath3 the fiber of @xmath4 over @xmath664 , and @xmath665 the one over @xmath666 .
by remark [ nuoto ] we have @xmath667 , so @xmath668 , let s call it @xmath3 .
thus @xmath669 , @xmath670 , while @xmath671 and @xmath672 are fibers of @xmath4 .
let @xmath673 be a curve corresponding to @xmath674 .
then @xmath675\in\widetilde{\alpha}_i$ ] , @xmath676 , @xmath677 , and @xmath678 .
this gives : @xmath679 finally set @xmath680 and @xmath681 , so that @xmath682\in\beta$ ] .
observe that @xmath683 is a divisor in @xmath236 and there are @xmath684 such that @xmath685 then @xmath686 so @xmath687 , and similarly @xmath688 .
hence @xmath689 this says that @xmath690 is a minimal element in the extremal ray @xmath128 , namely that for every curve @xmath691 such that @xmath103\in\beta$ ] , we have @xmath692 with @xmath265 .
moreover @xmath693 which gives @xmath694 this says that the length @xmath695 of the extremal ray @xmath128 , that is the minimal anticanonical degree of rational curves whose class is in @xmath128 , is @xmath15
. now ( * ? ? ?
* theorem 1.1 ) yields that for every non trivial fiber @xmath3 of @xmath636 we have @xmath696 on the other hand @xmath213 is finite on fibers of @xmath636 , so @xmath697 .
therefore @xmath698 , every fiber of @xmath636 has dimension @xmath309 , and @xmath660 . now @xmath213 and
@xmath636 induce a finite morphism @xmath699 , let @xmath700 be its degree .
let @xmath701 be a general point and @xmath702 .
then @xmath703 is a smooth surface and @xmath704 is finite of degree @xmath700 .
since in @xmath0 the divisor @xmath705 intersects trasversally @xmath706 and @xmath707 , @xmath708 is an isomorphism , so that @xmath709 and @xmath710 is the first projection .
observe that @xmath711 is a line @xmath377 ( the line through @xmath616 and @xmath617 ) , and @xmath712 is surjective , so it factors as @xmath713 where @xmath714 is the second projection and @xmath631 is a finite morphism .
we have @xmath715 on the other hand @xmath716 is the support of an ample divisor in @xmath703 and hence it is connected .
this implies that @xmath631 is an isomorphism and that @xmath717 . then @xmath718 ; on the other hand @xmath719
is an isomorphism , so @xmath720 .
now we have @xmath721 so @xmath722 and we are done .
in this section we prove theorem [ result ] @xmath96 .
let s consider a smooth fano variety @xmath0 of dimension @xmath723 , and a quasi elementary contraction @xmath1 with @xmath44 .
recall that @xmath7 is factorial and has at most isolated canonical singularities by lemma [ target ] @xmath50 and @xmath295 .
let @xmath179 be an elementary contraction of @xmath7 .
we consider all possibilities for @xmath183 . if @xmath183 is of type ( 3,0 ) , then @xmath97 is a point and @xmath7 is fano with @xmath492 .
if @xmath183 is of type ( 3,1 ) , then @xmath724 and @xmath493 . in this case
the other elementary contraction of @xmath7 can not be again of type ( 3,1 ) , because non trivial fibers of distinct elementary contractions can intersect at most in finitely many points , and @xmath7 is factorial .
[ ( 3,2 ) ] if @xmath179 is of type ( 3,2 ) , then @xmath183 is equidimensional , and we have the following possibilities : 1 . @xmath493 and @xmath725 ; 2 . @xmath494 and @xmath97 is either @xmath726 or @xmath727 ; 3 .
@xmath97 is a smooth del pezzo surface with @xmath728 , @xmath729 , @xmath183 is the projection , and @xmath730 , where @xmath48 is a smooth fano variety of dimension @xmath296 with a quasi elementary contraction onto @xmath18 .
we have @xmath731 and @xmath732 is quasi elementary ( see example [ compo ] ) . by theorem [ result ] @xmath50 , @xmath200 is equidimensional and @xmath97 is a smooth del pezzo surface
. then @xmath183 must be equidimensional too .
if @xmath733 , we get the first two cases .
if @xmath728 , again by theorem [ result ] @xmath50 we see that @xmath734 where @xmath48 is a smooth fano variety of dimension @xmath296 , and @xmath200 is the projection onto @xmath97 . then any intermediate contraction @xmath735 must be onto a product @xmath736 , hence we get @xmath737
. moreover @xmath4 induces a contraction @xmath738 .
since @xmath4 is quasi elementary , it is easy to see that @xmath310 is quasi elementary too . [ ( 2,0 ) ] if @xmath183 is of type ( 2,0 ) , then we have the following possibilities : 1 .
@xmath493 and @xmath7 has an elementary contraction of type ( 3,2 ) ; 2 .
@xmath494 and @xmath7 has a birational elementary contraction with fibers of dimension at most @xmath26 .
the divisor @xmath326 satisfies @xmath739 , hence proposition [ divisor ] yields @xmath94 .
however @xmath7 has a non trivial elementary contraction , hence @xmath411 .
then the statement follows from remark [ dettaglio ] . [ curry ]
let @xmath179 be an elementary contraction which is birational with fibers of dimension at most @xmath26 .
then @xmath183 is the blow - up of a smooth curve @xmath313 , and we have a commutative diagram @xmath598_f \ar[r]^{{\varphi } } & w\ar[d]^g \\
y \ar[r]^{\psi } & { z}}\ ] ] where @xmath48 is smooth , @xmath64 is the blow - up of a smooth subvariety of codimension @xmath309 , @xmath311 , and @xmath310 is a quasi elementary mori contraction .
recall that @xmath183 has a lifting as explained in [ lifting ] , so we have a diagram as above . since @xmath0 is fano , @xmath64 is a mori contraction .
then theorem [ secondo ] yields the statement .
in particular claim [ curry ] implies that _ @xmath7 has no small elementary contractions_. [ basta ] in the setting of claim [ curry ] , suppose that @xmath7 is fano . set @xmath740 and @xmath356 .
then the following statements are equivalent : 1 .
@xmath48 is not fano ; 2 .
@xmath97 is not fano ; 3 .
@xmath741 with normal bundle @xmath742 , and there is an extremal ray @xmath128 of @xmath164 , different from @xmath743 , such that @xmath744 .
@xmath745 if @xmath48 is not fano , by ( * ? ? ?
* proposition 3.4 ) there exists an extremal ray @xmath127 of @xmath100 such that @xmath746 and @xmath747 .
observe that @xmath127 is not contracted by @xmath4 , because @xmath748 for every curve @xmath102 contracted by @xmath4 .
let @xmath749 be the contraction of @xmath127 and let @xmath241 be an irreducible component of a non trivial fiber of @xmath465 .
then @xmath368 and @xmath4 is finite on @xmath241 , hence @xmath750 and @xmath751 .
moreover if @xmath752 then @xmath753 .
this would imply that @xmath754 , while @xmath755 .
thus @xmath756 .
now ( * ? ? ?
* theorem 1.2 ) yields that @xmath465 is a smooth blow - up with exceptional divisor @xmath156 .
observe that @xmath13 is a smooth @xmath18-bundle over @xmath757 .
it is not difficult to see that the @xmath18-bundle @xmath758 induces a second rational fibration on @xmath13 , so that @xmath741 .
moreover if @xmath759 is a non trivial fiber of @xmath465 , then @xmath760 and this gives @xmath761 .
in particular @xmath757 is a curve of anticanonical degree @xmath762 in @xmath97 , so @xmath97 is not fano .
@xmath763 suppose that @xmath97 is not fano .
observe that @xmath7 and @xmath97 may be singular , however they are factorial and @xmath764
. then reasoning as in ( * ? ? ?
* proposition 3.4 ) one gets @xmath294 .
@xmath765 the contraction of @xmath128 is birational with fibers of dimension at most @xmath26 ; let @xmath749 be the smooth blow - up given by claim [ curry ] .
then @xmath766 , and @xmath767 .
if @xmath759 is a non trivial fiber of @xmath465 , then @xmath768 , which yields @xmath769 . hence @xmath48 is not fano .
[ pollock ] suppose that @xmath7 is fano , and let @xmath770 and @xmath771 be two distinct extremal rays of @xmath164 with divisorial loci @xmath772 . if @xmath773 , then the contraction @xmath179 of one of the @xmath774 s is the blow - up of a smooth curve @xmath313 , and @xmath97 is fano . let @xmath775 be the contraction of @xmath774 . since @xmath773 and @xmath7 is factorial , we have @xmath776 . thus @xmath416 and @xmath441 can not be both of type ( 2,0 ) .
so let s assume that @xmath416 is of type ( 2,1 ) . if @xmath419 is fano , we are done . if @xmath419 is not fano , claim [ basta ] yields the existence of a second extremal ray @xmath777 of @xmath164 , distinct from @xmath770 , with @xmath778 .
then @xmath779 is generated by @xmath770 and @xmath777 , and no other extremal ray of @xmath164 can be contained in @xmath779 .
thus @xmath441 can not be of type ( 2,0 ) .
let s show that @xmath780 must be fano .
if not , by claim [ basta ] there is another extremal ray @xmath781 of @xmath164 , distinct from @xmath771 , with @xmath782 . since @xmath771 and @xmath781
are not contained in @xmath779 , we have @xmath783 and @xmath784 .
now if @xmath785 is an irreducible curve , we get @xmath786 because @xmath787 and @xmath788 . on the other hand @xmath789 because @xmath790 and @xmath784 .
thus we have a contradiction .
[ 2,1fano ] in the setting of claim [ curry ] , suppose that @xmath7 is fano
. then we are in one of the following situations : 1 .
@xmath48 and @xmath97 are fano ; 2 .
@xmath42 and @xmath7 has another elementary contraction @xmath791 as in claim [ curry ] , such that the corresponding @xmath792 , @xmath793 are fano ; 3 .
@xmath494 and @xmath7 has an elementary contraction of type ( 3,2 ) .
let s assume that we are not in @xmath50 , so @xmath48 and @xmath97 are not fano . by claim [ basta ]
we have @xmath741 , and there is a second extremal ray @xmath128 of @xmath164 with @xmath794 .
there exists an extremal ray @xmath455 of @xmath164 such that @xmath795 , let @xmath796 be its contraction .
clearly @xmath455 is distinct from @xmath743 and @xmath128 , and the elements of three distinct extremal rays must by linearly independent in @xmath797 .
hence @xmath42 and @xmath798 , which implies that @xmath791 is finite on @xmath13 .
on the other hand @xmath13 must intersect every non trivial fiber of @xmath791 , hence @xmath791 has fibers of dimension at most @xmath26 . if @xmath791 is of fiber type , then it is of type ( 3,2 ) .
let s consider @xmath799 since @xmath800 is the line spanned by @xmath455 in @xmath801 , we have @xmath802 and @xmath803 is injective . on the other hand @xmath804 ,
hence @xmath803 is surjective too .
this gives @xmath805 and @xmath494 , and thus @xmath294 .
suppose that @xmath791 is birational .
then @xmath806 , and claim [ pollock ] implies that @xmath793 is fano .
finally also @xmath792 is fano by claim [ basta ] , and we get @xmath96 .
[ m ] let @xmath807 be a contraction onto a surface .
let @xmath808 be the extremal rays of @xmath809 and @xmath775 the contraction of @xmath622 .
then @xmath810 for some @xmath811 .
set @xmath812 , and assume by contradiction that @xmath813 for every @xmath814 .
then each @xmath815 is birational and @xmath816 for every @xmath817 by remark [ divisorial ] ; moreover @xmath818 for every @xmath811 .
let @xmath141 be an irreducible curve which is contracted by @xmath819 but not contained in @xmath820
. then @xmath821\in\alpha_i$ and $ \lambda_i\in{\mathbb{q}}_{\,\geq 0}$ for every $ i=1,\dotsc , m$.}\ ] ] so @xmath822 implies @xmath823 for every @xmath824 , a contradiction .
let s suppose that @xmath6 . by claims [ ( 3,2 ) ] , [ ( 2,0 ) ] , and [ curry ] , any elementary contraction of @xmath7 is either a smooth blow - up or a @xmath18-bundle .
this implies that @xmath7 is fano ( but possibly singular ) by remark [ fano ] .
suppose now that @xmath7 is smooth and that @xmath8 .
then @xmath825 with @xmath13 a del pezzo surface by ( * ? ? ?
* theorem 2 ) .
thus @xmath826 is a quasi elementary contraction ( see example [ compo ] ) , and applying theorem [ result ] @xmath50 as in the proof of claim [ ( 3,2 ) ] we easily get the statement . thus we are left to prove that if @xmath6 , then @xmath7 must be smooth . by contradiction ,
assume that @xmath7 is singular and @xmath6 . by claims [ ( 3,2 ) ] ,
[ ( 2,0 ) ] , and [ 2,1fano ] , we can construct a sequence @xmath827_f \ar[r ] & { x_{\rho_y-1 } } \ar[r ] \ar[d ] & { \cdots}\ar[r ] & { x_4 } \ar[d]^{f_4 } \\ { y = y_{\rho_y } } \ar[r]^{\psi_{\rho_y } } & { y_{\rho_y-1 } } \ar[r ] & { \cdots}\ar[r]^{\psi_5 } & { y_4 } } \ ] ] where for every @xmath828 @xmath829 is smooth and fano , @xmath70 is a quasi elementary contraction , @xmath830 is fano with @xmath831 , and @xmath815 is the blow - up of a smooth curve contained in the smooth locus of @xmath832 . in particular each
@xmath830 is singular , because @xmath7 is . since @xmath833 is singular and @xmath834 , claim [ ( 3,2 ) ]
implies that @xmath833 has no elementary contraction of fiber type .
[ final ] @xmath833 has no contraction onto a surface
. by contradiction let @xmath835 be such a contraction .
then @xmath819 can not be elementary , so @xmath836 .
the cone @xmath809 contains @xmath837 extremal rays , whose contractions are birational .
call @xmath838 their exceptional loci . by claim [ m ]
they can not be all disjoint , so we can assume that @xmath839 .
using claim [ pollock ] , we can assume that the elementary contraction @xmath840 with exceptional locus @xmath841 is the blow - up of a smooth curve @xmath842 , and that @xmath843 is fano .
we observe that @xmath843 is singular because @xmath833 is . as before we get @xmath844_{f_4}\ar[r ] & { x_3}\ar[d]^{f_3}\\ { y_4 } \ar[r]^{\psi_4 } & { y_3 } } \ ] ] where @xmath845 is smooth and fano and @xmath846 is a quasi elementary contraction .
moreover @xmath819 induces a contraction @xmath847 .
if @xmath848 is elementary , then @xmath849 , @xmath848 is equidimensional , and @xmath559 is either @xmath726 or @xmath727 by claim [ ( 3,2 ) ] .
hence @xmath848 is a conic bundle ( see @xcite for the definition and properties of conic bundles ) .
let @xmath850 be the discriminant locus of @xmath848 , which is non empty because @xmath843 is singular .
we recall that @xmath851 can not intersect any curve of anticanonical degree @xmath26 by remark [ wellknown ] .
thus @xmath620 can not be a component of a reducible fiber of @xmath848 .
if @xmath620 is an irreducible fiber , let @xmath852 and call
@xmath853 the blow - up of @xmath559 in @xmath854 : then @xmath833 has an elementary contraction onto @xmath853 , which is impossible .
hence @xmath855 is a curve in @xmath559 , and it is disjoint from @xmath856 because @xmath620 can not intersect singular fibers of @xmath848 . if @xmath857 , the only possibility is that @xmath856 is a union of fibers of a projection @xmath858 , hence a disjoint union of smooth rational curves .
but this is impossible , see for instance ( * ? ? ?
* lemma 5.3 ) .
consider now the case where @xmath859 .
then we have a contraction @xmath860 , which has a second factorization @xmath861 .
it is not difficult to see that @xmath862 is birational with fibers of dimension @xmath863 , so by claim [ curry ] it is again a blow - up of a smooth curve @xmath864 , and we have a diagram : @xmath865_{f_3}\ar[r ] & { x_2}\ar[d]^{f_2}&\\ { y_3 } \ar[r]^{\psi_3 } & { y_2 } \ar[r]^{\xi } & { { \mathbb{p}}^2 } } \ ] ] where @xmath866 is smooth , @xmath867 is singular with @xmath868 , and @xmath79 is a quasi elementary mori contraction .
the only curve in @xmath867 which could have non positive anticanonical degree is @xmath619 .
thus @xmath869 is a conic bundle with non empty discriminant locus @xmath870 .
if @xmath871 is a curve in @xmath25 , then @xmath619 must intersect some singular fiber , which is again impossible by remark [ wellknown ] .
thus @xmath619 is a smooth fiber of @xmath631 , @xmath867 is fano , and @xmath866 is fano too by claim [ basta ] .
let s consider the other elementary contraction @xmath872 of @xmath867 .
if @xmath873 is again of type ( 3,2 ) , reasoning as above we get that @xmath873 must contract @xmath619 , a contradiction because @xmath874 .
if @xmath873 is of type ( 3,1 ) , it is a fibration in del pezzo surfaces over @xmath18 .
then @xmath867 is a finite cover of @xmath875 , and @xmath619 is the inverse image of @xmath601 .
this implies that @xmath843 is a finite cover of @xmath876 , which gives a surjective morphism @xmath877 .
taking the stein factorization we get an elementary contraction @xmath878 of type ( 3,2 ) , where @xmath853 is a finite cover of @xmath726 . by claim
[ ( 3,2 ) ] @xmath853 can be @xmath726 or @xmath727 , but @xmath853 has two distinct fibrations , thus @xmath879 .
we have already excluded this possibility .
therefore @xmath873 is birational , let @xmath156 be its exceptional divisor .
since @xmath880 by remark [ divisorial ] , it must be @xmath881 , hence @xmath882 and @xmath156 intersects @xmath619 . in particular @xmath156
can not be covered by curves of anticanonical degree @xmath26 , thus @xmath873 is of type ( 2,0 ) .
moreover @xmath883 , because @xmath631 has singular fibers .
the composite contraction @xmath884 has a second factorization @xmath885 it is not difficult to see that @xmath450 has exceptional locus @xmath886 and is the blow - up of a smooth curve @xmath887 .
since the image of @xmath886 in @xmath97 is @xmath888 , @xmath889 is contracted by @xmath890 .
then we see that @xmath890 is again a mori contraction of type ( 2,1 ) with exceptional divisor @xmath891 , and @xmath892 is fano .
in particular @xmath890 is again a blow - up of a smooth curve contained in @xmath893 , so that @xmath315 is a smooth point .
moreover @xmath886 is contained in the smooth locus of @xmath843 , so that @xmath894 .
therefore @xmath873 is just the blow - up of @xmath315 and @xmath895 .
if @xmath156 intersects @xmath619 in at least two distinct points , take @xmath377 a line in @xmath156 through these two points .
let @xmath896 be the proper transform of @xmath377 in @xmath843 .
then @xmath897 and @xmath898 , so by remark [ wellknown ] we get @xmath899 , a contradiction . if @xmath156 intersects @xmath619 in a single non reduced point @xmath290 , similarly as before take @xmath377 a line in @xmath895 through @xmath290 .
then the schematic intersection @xmath900 in non reduced along the fiber of @xmath901 over @xmath290 , thus again @xmath898 gives a contradiction .
this concludes the case where @xmath849 .
we still have to exclude the case where @xmath902 and @xmath843 has no elementary contractions of type ( 3,2 ) .
reasoning as for @xmath833 , we see that one of the two extremal rays of @xmath903 yields a blow - up @xmath904 of a smooth curve @xmath905 , and @xmath780 is fano with @xmath906 .
now @xmath848 yields an elementary contraction @xmath907 .
claim [ ( 3,2 ) ] gives @xmath908 , and @xmath909 is non empty . as before we easily get a contradiction . using claims [ final ] , [ ( 2,0 ) ] , and [ 2,1fano ]
, we get a sequence @xmath910^{\psi_4 } & { y_3 } \ar[r]^{\psi_3 } & { y_2 } \ar[r]^{\psi_2 } & { y_1 } } \ ] ] where each @xmath830 fano and each @xmath911 is the blow - up of a smooth curve @xmath912 . in particular @xmath913 is factorial with isolated canonical singularities , singular , and @xmath914 .
we first show that @xmath913 must have terminal singularities , using the following lemma .
[ canonical ] let @xmath97 be a @xmath15-dimensional @xmath119-factorial projective variety , with isolated canonical singularities and @xmath915 cartier .
suppose that @xmath97 is fano with @xmath916 , and that the singularities of @xmath97 are not terminal .
then one of the following occurs : 1 .
@xmath97 contains a @xmath26-dimensional family of curves of anticanonical degree @xmath26 passing through a singular point ; 2 .
@xmath97 is covered by a family of curves of anticanonical degree @xmath917 passing through a singular point .
we postpone the proof of lemma [ canonical ] and carry on with the proof of theorem [ result ] @xmath96 .
suppose that @xmath913 has at least one non - terminal singular point : then lemma [ canonical ] applies to @xmath913 .
if @xmath50 holds , let @xmath918 be a surface covered by curves of anticanonical degree @xmath26 .
since @xmath914 , @xmath13 is ample , and @xmath919 . observe that even if @xmath920 , @xmath618 does not contain any singular point , hence it can not be a member of the family given by @xmath50 .
thus @xmath618 intersects some curve of anticanonical degree @xmath26 , which is impossible by remark [ wellknown ] .
suppose now that @xmath96 holds for @xmath913 .
if @xmath618 is a component of some reducible curve @xmath921 of the family , it must be @xmath922 with @xmath923 , which gives again a contradiction .
again @xmath618 can not be a member of the family , because it does not contain singular points .
hence @xmath618 is not contained in any member of the family ; let @xmath559 be an irreducible surface containing @xmath618 such that through every point of @xmath618 there is a curve of anticanonical degree @xmath924 contained in @xmath559 .
let @xmath925 be the proper transform of @xmath559 in @xmath867 .
then through every point of @xmath926 there is a curve of anticanonical degree @xmath26 contained in @xmath925 ( see remark [ wellknown ] ) .
consider @xmath927 .
if @xmath928 is a point , then @xmath619 must intersect some curve of anticanonical degree @xmath26 contained in @xmath925 . on the other hand
if @xmath928 is a curve , then it must intersect @xmath559 , thus @xmath619 must intersect @xmath929 . in any case
@xmath619 will intersect some curve of anticanonical degree @xmath26 , which gives a contradiction .
hence @xmath913 has terminal singularities , and the same holds for each @xmath830 .
consider in particular @xmath833 . by @xcite
@xmath930 has a smoothing , that is an integral complex space @xmath931 , with a projective flat morphism @xmath932 onto the complex unit disc , such that @xmath833 is the fiber over @xmath762 while the fiber @xmath933 over @xmath934 is a smooth fano threefold .
it is proven in @xcite that @xmath935 , in particular @xmath936 .
then we know by mori and mukai s classification that @xmath933 has a conic bundle structure ( not necessarily elementary ) , see ( * ? ? ?
* theorem on p. 141 ) .
our goal is to deduce from this that @xmath833 must have a contraction onto a surface , contradicting claim [ final ] . for this
we need the following lemma , based on @xcite .
[ conodimori ] let @xmath97 be a @xmath15-dimensional factorial fano variety with terminal singularities and @xmath937 a smoothing .
consider the inclusions @xmath938 and @xmath939 .
then the push - forwards @xmath940 and @xmath941 induce bijections among the cones @xmath942 and every contraction of @xmath943 or of @xmath944 is the restriction of the contraction of the corresponding face of @xmath945 .
by lemma [ conodimori ] , the conic bundle on @xmath933 induces a contraction @xmath946 .
this restricts to a contraction @xmath947 onto a surface , which contradicts claim [ final ] . this concludes the proof of theorem [ result ] .
let @xmath948 be a partial crepant resolution such that @xmath949 has terminal and factorial singularities ( see @xcite ) .
hence @xmath950 , @xmath951 is nef and big , and @xmath952 because @xmath183 is not an isomorphism .
since @xmath97 is @xmath119-factorial , for every non terminal point @xmath953 the inverse image @xmath954 has pure dimension @xmath309 , therefore @xmath326 is a divisor . moreover any irreducible curve contained in @xmath326 has anticanonical degree zero .
there is at least one elementary mori contraction @xmath955 .
observe that @xmath4 must be finite on @xmath326 , because any curve contracted by @xmath4 has positive anticanonical degree .
hence any fiber @xmath3 of @xmath4 such that @xmath956 has dimension at most @xmath26 .
suppose that @xmath4 is of fiber type
. then it must be of type ( 3,2 ) and ( * ? ? ?
* theorem 7 ) says that @xmath48 is a smooth surface and @xmath4 is a conic bundle .
moreover if @xmath957 is an irreducible component , then @xmath958 , so that every fiber of @xmath4 intersects @xmath156 . then the fibers of @xmath4 give a covering family of curves of anticanonical degree @xmath917 in @xmath97 , all passing through the singular point @xmath959 , and we get @xmath96 .
suppose that @xmath4 is birational . since @xmath949 is gorenstein , @xmath4 is divisorial ; set @xmath960 .
we claim that @xmath92 can not be disjoint from @xmath326 .
in fact if so , @xmath961 would be a non nef cartier divisor in @xmath97 , which is impossible because @xmath916 .
hence @xmath4 must be of type ( 2,1 ) and by ( * ? ? ?
* theorem 4 ) we have @xmath962 for the general fiber @xmath377 of @xmath4 . again if @xmath156 is an irreducible component of @xmath326 intersecting @xmath92 , every fiber @xmath377 of @xmath4 must intersect @xmath156 .
so we get a one - dimensional family of curves of anticanonical degree @xmath26 in @xmath97 , passing through the singular point @xmath959 . by (
* proposition 1.1 ) @xmath963 has at most isolated terminal factorial singularities at the singular points of @xmath944 .
we refer to @xcite for the notation in the relative situation .
observe that for a projective morphism in the analytic category the standard results of mmp hold , see ( * ? ? ?
* example 2.17 ) and references therein .
in particular since each fiber of @xmath937 is fano , @xmath945 is closed and polyhedral by the relative version of the cone theorem .
we first observe that the linear maps @xmath964 are isomorphisms .
in fact they are dual to the restrictions @xmath965 which are isomorphism by ( * ? ? ? * theorem 1.4 ) .
moreover we have @xmath966 and @xmath967 . up to shrinking @xmath968
, we can assume that for every extremal ray @xmath127 of @xmath945 either @xmath221 dominates @xmath968 , or @xmath221 is contained in @xmath944 .
let s show that the second case can not happen .
assume by contradiction that @xmath221 is contained in @xmath944 and consider the contraction @xmath64 of @xmath127 : @xmath969\ar[r]^{{\varphi } } & { \mathcal{w}}\ar[dl ] \\ { \delta } & } \ ] ] then @xmath64 is an isomorphism outside the central fiber , and restricts to a contraction @xmath970 on the central fiber .
the morphism @xmath971 is projective and the general fiber has dimension @xmath15 , hence @xmath972 and @xmath973 is birational .
this means that @xmath974 has dimension at most @xmath309 , namely @xmath64 is a small contraction of @xmath963 and it must be of type ( 2,0 ) . hence @xmath975 is not factorial at @xmath976 .
since @xmath941 is injective , @xmath973 is an elementary contraction of type ( 2,0 ) of @xmath944 , and @xmath156 is irreducible .
moreover @xmath977 has terminal singularities and is fano . if @xmath977 were gorenstein , @xmath975 should be factorial again by ( * ? ? ?
* proposition 1.1 ) , which is not the case : so @xmath978 is not cartier . the possibilities for @xmath973 are given in ( * ? ? ?
* theorem 5 ) , and the only case where @xmath978 is not cartier is @xmath895 with normal bundle @xmath979 . since @xmath156 is a smooth prime divisor in the factorial variety @xmath944 ,
it is contained in @xmath980 and hence in @xmath981 .
now ( * ? ? ?
* theorem 2.1 ) yields @xmath982 , a contradiction .
hence @xmath221 dominates @xmath968 , which means that @xmath127 is contained in both cones @xmath983 and @xmath984 . repeating this for every extremal ray of @xmath945
, we get the statement .
in this section we prove the corollaries stated in the introduction , and some other applications .
we also give some related examples .
suppose that @xmath0 is a smooth fano variety with a quasi elementary contraction @xmath1 .
if @xmath0 has other suitable contractions , one can use theorems [ face ] and [ face2 ] to get a quasi elementary contraction @xmath175 with @xmath517 , and then apply theorem [ result ] .
corollaries [ appl ] , [ cor1 ] , and [ cor2 ] are obtained in this way .
recall that @xmath985 because @xmath70 is quasi elementary .
suppose that @xmath986 . by theorem [ face2 ] @xmath50 and remark [ dim
] , there exists a contraction @xmath175 where @xmath987 and @xmath988 this immediately gives @xmath71 , and if equality holds @xmath97 is a point so @xmath989 .
moreover @xmath74 implies @xmath990 and @xmath916 .
let s notice that @xmath636 is quasi elementary too .
in fact let @xmath199 be a general fiber of @xmath636 , then @xmath199 contains general fibers of @xmath991 and @xmath79 .
since both @xmath70 s are quasi elementary , @xmath992 contains both @xmath993 and @xmath994 . on the other hand we have @xmath995 as shown in the proof of theorem [ face2 ] , so @xmath996 and @xmath636 is quasi elementary .
finally suppose that @xmath78 and that @xmath79 is elementary .
similarly to the previous case theorem [ face ] @xmath50 gives a quasi elementary contraction @xmath175 with @xmath998 and @xmath999 .
if @xmath586 we proceed as before .
if @xmath1000 then @xmath1001 by theorem [ result ] @xmath96 , so we are done .
recall that @xmath2 because @xmath4 is quasi elementary .
since @xmath4 is finite on fibers of @xmath64 , they have dimension at most @xmath519 .
if there is a fiber with the same dimension as @xmath7 , let @xmath241 be an irreducible component with @xmath1005 . then @xmath633 , so @xmath1006 is surjective and @xmath1007 .
if @xmath64 has a fiber of dimension @xmath1004 , let @xmath241 be an irreducible component with @xmath1008 .
then @xmath1009 is a prime divisor in @xmath7 with @xmath1010 , so proposition [ divisor ] implies that @xmath94 .
the proof is very similar to that of corollary [ appl ] .
we apply theorem [ face2 ] @xmath294 and theorem [ result ] @xmath50 in the first case , theorem [ face ] @xmath294 and theorem [ result ] @xmath96 in the second case .
the first two statements are a direct consequence of theorem [ result ] @xmath50 . for the last statement , suppose that @xmath1014 and that @xmath1 is an elementary contraction with @xmath44
. then @xmath8 , so theorem [ result ] @xmath96 says that @xmath1015 where @xmath13 , @xmath1016 are del pezzo surfaces and @xmath1016 has an elementary contraction onto @xmath18 .
then @xmath1017 or @xmath1018 .
the first two statements follow from corollary [ appl ] and theorem [ result ] .
let @xmath1 be an elementary contraction with @xmath63 , and @xmath174 another elementary contraction . if @xmath64 has a fiber of dimension at least @xmath15 , then @xmath1019 by corollary [ cor1 ] .
in particular this holds if @xmath64 is of type ( 3,0 ) , ( 4,0 ) , or ( 4,1 ) .
finally suppose that @xmath65 .
if @xmath64 is of fiber type , we have @xmath61 by the previous part .
if @xmath64 is birational , then it must be divisorial , so @xmath61 by corollary [ cor2 ] .
1 . @xmath1022 , @xmath1023 , and @xmath92 is a fiber of @xmath4 ; 2 .
@xmath1019 , @xmath94 , and @xmath4 is elementary ; 3 .
@xmath16 , @xmath1024 , @xmath7 is a del pezzo surface , @xmath1025 , and @xmath1026 ; 4 .
@xmath1003 , @xmath492 , and @xmath1027 .
suppose that @xmath258 .
then lemma [ nonequilocus ] says that @xmath1028 ; in particular @xmath92 contains a fiber @xmath3 , so @xmath1029 and @xmath1030 . if @xmath1031 , then @xmath1032 , so @xmath92 is a fiber and @xmath1033 .
thus we are in @xmath50 .
assume now that @xmath1027 .
then the restriction @xmath1035 is surjective , so @xmath1036 . if equality holds , then @xmath1037 and hence @xmath1038 .
this implies that @xmath302 is finite , so @xmath4 has @xmath26-dimensional fibers .
then @xmath4 is elementary and @xmath1039 , so we are again in @xmath96 .
finally if @xmath492 we get @xmath295 . [ firstex ] it is not difficult to write down examples of smooth fano varieties of dimension @xmath596 which are not products , but have an elementary contraction of fiber type over @xmath25 , @xmath726 , or @xmath727 . thus the condition @xmath42 in the second part of theorem [ result ] @xmath50 is necessary .
a different example is given by @xmath1043 , which has a quasi elementary contraction onto @xmath1044 with fiber @xmath1045 . in this case
the variety is a product , but it is not the product of the fiber and the target of the contraction . let @xmath1046 , so that @xmath7 is fano with @xmath494 .
observe that @xmath7 is the divisorial resolution of a quadric @xmath1047 with an isolated singularity .
let @xmath1048 be the pull - back of @xmath1049 , and consider @xmath1050 then @xmath0 is fano with dimension @xmath11 and @xmath1051 , and it is not a product .
one can write down analogous examples with @xmath1052 .
we do not know whether there are similar examples with @xmath1053 .
let us point out that smooth fano @xmath15-folds @xmath7 with @xmath1054 ( respectively @xmath1055 ) which are not products are given by just two families ( respectively @xmath1056 ) , after @xcite .
if @xmath4 is a conic bundle , then @xmath7 is smooth ( see ( * ? ? ?
* theorem 3.1 ) and ( * ? ? ?
* proposition 4.1 ) ) and if @xmath1058 then @xmath7 is also fano , by ( * ? ? ?
* corollary on p. 156 ) .
however this is not true in higher dimensions , see ( * ? ? ?
* example on p. 156 ) .
if @xmath4 is equidimensional then @xmath7 is smooth and fano , see example [ conicbdl ] .
however it is well known that @xmath4 can have isolated @xmath309-dimensional fibers , which are classified , see @xcite . in @xcite
we find several examples where @xmath4 is not equidimensional and @xmath1059 ; in particular in this case @xmath0 is not a product .
however we are not aware of similar examples with @xmath7 singular .
[ delpezzo ] in the toric case , smooth fano varieties are classified up to dimension @xmath1060 , the cases of dimensions @xmath51 , @xmath1061 and @xmath1060 being quite recent @xcite .
they are a good source of explicit examples .
after the classification in @xcite ( see also @xcite ) , toric fano 4-folds have picard number at most @xmath1062 .
the ones with @xmath1063 are just @xmath1064 and @xmath1065 , @xmath1066 and @xmath1067 being the blow - up of @xmath25 in two points and in three non collinear points respectively . among the ones with @xmath1068
there is a case with no ( non trivial ) quasi elementary contractions , the toric del pezzo 4-fold @xmath495 ( n. 118 in @xcite ) . the mori cone @xmath1069 has dimension @xmath1061 but has @xmath1070 extremal rays .
_ every _ elementary contraction is a small contraction with exceptional locus a @xmath25 with normal bundle @xmath1071 .
every such exceptional @xmath25 intersects three others in a point .
one can see that @xmath495 has a contraction of fiber type @xmath210 with @xmath44 and @xmath492 , so @xmath4 is not quasi elementary .
there are @xmath1061 two - dimensional fibers , which are unions of two exceptional @xmath25 s intersecting in one point .
moreover @xmath7 has @xmath1061 isolated non @xmath119-factorial points in the images of these fibers .
this example has an analog @xmath1072 in each even dimension @xmath1073 .
this is a smooth toric fano variety with @xmath1074 and @xmath1075 extremal rays .
every elementary contraction is a small contraction with exceptional locus a @xmath1076 with normal bundle @xmath1077 .
the varieties @xmath1078 are called toric del pezzo varieties and were introduced in @xcite .
laurent bonavero , frdric campana , and jarosaw a. winiewski , _ varits projectives complexes do nt lclate en un point est de fano _ , comptes rendus de lacadmie des sciences * 334 * ( 2002 ) , 463468 .
yuri g. prokhorov , _ on the degree of fano threefolds with canonical gorenstein singularities _ , matematicheski sbornik * 196 * ( 2005 ) , 81122 , russian .
english translation : sbornik : mathematics * 196 * ( 2005 ) , 77114 .
v. g. sarkisov , _ on conic bundle structures _
, izvestiya akademii nauk sssr seriya matematicheskaya * 46 * ( 1982 ) , 371408 , russian .
english translation : mathematics of the ussr izvestiya * 20 * ( 1982 ) , 355390 . v. e. voskresenski and alexander klyachko , _ toric fano varieties and systems of roots _ , izvestiya akademii nauk sssr seriya matematicheskaya * 48 * ( 1984 ) , 237263 , russian .
english translation : mathematics of the ussr izvestiya * 24 * ( 1985 ) , 221244 . | let @xmath0 be a smooth complex fano variety .
we study `` quasi elementary '' contractions of fiber type of @xmath0 , which are a natural generalization of elementary contractions of fiber type .
if @xmath1 is such a contraction , then the picard numbers satisfy @xmath2 , where @xmath3 is a general fiber of @xmath4 .
we show that if @xmath5 and @xmath6 , then @xmath7 is smooth and fano ; if moreover @xmath8 , then @xmath0 is a product . this yields sharp bounds on @xmath9 when @xmath10 and @xmath0 has a quasi elementary contraction of fiber type , and other applications in higher dimensions . |
in this note , we give definite analytical results concerning the process of mott scattering in the presence of a strong laser field and compare our results to previous theoretical works , namely the work of s.m .
et al _ @xcite and the work of y. attaourti _
et al _ @xcite .
we hope this contribution will be useful to all researchers working in this field and will bring an end to the controversy raised by the expression found by c. szymanowski _
@xcite for the case of a circular polarization of the laser field .
an analytical expression for the spin - unpolarized differential cross section is derived using trace calculations .
the electric field strength as well as the frequency of the laser field and the kinetic energy of the incoming electron being key parameters , the study of the process of mott scattering in the presence of an elliptically polarized laser field introduces a new key parameter , namely the degree of ellipticity @xmath0 . the cross section dependency on this new key parameter
is reported .
the general features of the mott scattering process are qualitatively modified when a laser field is present and this is particularly true when one study the spin - dependent relativistic mott scattering .
not only it is important to take care of the fact that the electron is a fermion but also to describe this particle by the appropriate wave function in a non - perturbative way .
this is done by using the dirac - volkov wave functions @xcite which contain the interaction of the electron with the laser field to all orders .
the organization of this paper is as follows . in section
ii , we present the theory in the first born approximation . in section iii , we discuss the analytical results for the spin - unpolarized differential cross section modified by the laser field and analyze their dependencies on the new relevant parameter , that is the degree of ellipticity @xmath0 .
we end by a brief summary and conclusion in section iv . throughout this work ,
we use atomic units @xmath1 and work with the metric tensor @xmath2 .
we treat the laser field classically since we are considering intensities that do not allow pair creation @xcite .
the four potential corresponding to the laser field satisfies the lorentz condition @xmath3 and is given by @xmath4 with @xmath5 and where @xmath6 is the degree of ellipticity of the laser field .
the four vectors @xmath7 and @xmath8 satisfy the following relations @xmath9 @xmath10 and the lorentz condition @xmath11 implies @xmath12 .
the linear polarization is obtained for @xmath13 and the circular polarization is obtained for @xmath14 . the electric field associated with the potential of the laser field
is @xmath15 the interaction potential is the coulomb potential of a target nucleus of charge @xmath16
@xmath17 and in the first born approximation , the transition matrix element for the transition ( @xmath18 ) is @xmath19 the dirac - volkov wave functions @xmath20 and @xmath21 describe the incident and scattered electron respectively .
such wave functions normalized to the volume @xmath22 are @xcite @xmath23 where @xmath24 represents a dirac bispinor normalized as @xmath25 and @xmath26 is the quasi - impulsion acquired by the electron in the presence of the laser field @xmath27 the quantity @xmath28 is defined by @xmath29 , and @xmath30 , where the feynmann slash notation is used @xcite : for a given four vector @xmath31 , we have @xmath32 .
finally the averaged squared potential @xmath33 is given by @xmath34 from which one deduces @xmath35 for the case of a circular polarization of the laser field and @xmath36 for the case of a linear polarization of the laser field .
the argument @xmath37 of the ordinary bessel functions is @xmath38 with @xmath39 and @xmath40\tan ( \eta /2 ) .
\label{10}\ ] ] a useful parameter that intervenes in the expression of the dcs is @xmath41 . using the standard procedures of qed @xcite
, we obtain for the spin - unpolarized differential cross section evaluated for @xmath42 @xmath43 with @xmath44 using reduce for the trace calculations @xcite , we obtain @xmath45 the eight coefficients @xmath31 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath52 are given respectively by @xmath53 @xmath54 @xmath55 @xmath56 @xmath57 @xmath58 @xmath59 @xmath60 where @xmath61 . in the absence of the laser field , all the contributions coming from the sum over @xmath62 of the various ordinary bessel functions vanish except for @xmath63 where @xmath64 and we recover the well known formula for mott scattering in the absence of the laser field @xcite @xmath65 where @xmath66 . it can easily be checked that for the case of a linear polarization of the laser field ( @xmath13 ) , the phase @xmath67 and the results of s.m .
et al _ @xcite are straightforward to obtain whereas for the circular polarization ( @xmath14 ) , we find the results previously found by y. attaourti _ et al _ @xcite .
( [ 11 ] ) is the relativistic generalization of the bunkin and fedorov @xcite treatment and it contains the degree of ellipticity of the laser field as a new key parameter .
in this section , we discuss the numerical simulations for the differential cross sections of the mott scattering by an elliptically polarized laser field .
we assume without loss of generality that the target is a proton having a charge @xmath68 .
the @xmath37 axis is set along the direction of the field wave vector @xmath69 , @xmath70 and @xmath71 with the vectors @xmath72 and @xmath73 such that @xmath74 and @xmath75 from which one deduces that @xmath76 .
thus , as found by s.m .
li @xcite for the linear polarization , we have @xmath77 and for the case of circular polarization @xcite , @xmath78 . the laser frequency used is @xmath79 @xmath80 which corresponds to a photon energy @xmath81 @xmath82 .
the incident electron kinetic energy is @xmath83 @xmath84 which corresponds to a relativistic parameter @xmath85 and the electric field strength has the value @xmath86
@xmath87 @xmath88 @xmath89 .
the various differential cross sections are plotted as functions of the angle @xmath90 . for small scattering angles ,
typically ( @xmath91 ) , ( @xmath92 ) , the summed spin - unpolarized differential cross sections are sharply peaked around @xmath93 and are all close to the corresponding unpolarized laser - free differential cross section given in eq .
( [ 22 ] ) .
@xmath84 as a function of the angle @xmath90 scaled in @xmath94 .
the electric field strength is @xmath86 @xmath80 and the laser frequency is @xmath79 @xmath80 . the corresponding mott - scattering geometry is given in the text.,width=4,height=4 ] the three dcss corresponding to @xmath13 , @xmath95 and @xmath14 are given in fig .
1 together with the laser - free dcs for the geometry ( @xmath96 ) , ( @xmath97 ) . we obtain four almost indistinguishable curves .
thus , at small angles , the summed differential cross sections are almost unmodified by the laser field and its polarization does not play a key role .
the physical explanation of this observation is that classically , when the particles are close to the small angle scattering region , this corresponds to large impact parameters and the incident electron does not deviate notably from its trajectory . for other scattering angles , ( @xmath98 ) , ( @xmath99 ) ,
the situation changes drastically since for medium and large scattering angles , the momentum transfer during the mott scattering is large and a significant number of photons can be exchanged with the laser field . in fig .
2 , we compare the three dcss corresponding to @xmath100 , @xmath95 and @xmath14 together with the laser - free dcs for the geometry ( @xmath101 ) , ( @xmath102 ) for an exchange of @xmath103 photons . in this geometry , the effect of the laser field polarization is clearly shown since the three dcss are now well distinguishable .
one has to sum over a very large number of photons to recover the laser - free dcs .
furthermore , all numerical simulations have shown the following .
the dcs for linear polarization is always higher than the two others .
the dcs for elliptical polarization is lower or higher than the dcs for circular polarization depending on the value of the degree of ellipticity @xmath0 .
photons scaled in @xmath104 .
as in fig . 1 , @xmath86 @xmath80 and @xmath79 @xmath80
. the corresponding mott - scattering geometry is explained in the text.,width=5,height=3 ] to have an idea about the behaviour of the dcs as a function of the degree of ellipticity @xmath0 , we have obtained a three dimensional curve corresponding to the same geometry as in fig . 2 but for a degree of ellipticity @xmath0 varying from @xmath105 to @xmath106 .
there are oscillations of the dcs for the elliptical polarization and the dcs for linear polarization is always the higher dcs .
these oscillations decrease as the number of photons exchanged is increased . the convergence towards the laser - free dcs is faster for the linear polarization of the laser field . for the circular polarization
, this convergence is easily obtained when one finds the value for the convergence corresponding to the linear polarization whereas it is much more difficult to infer from the previous results for which value of the number of photons exchanged , the dcs for the elliptical polarization will converge to the laser - free dcs .
however , depending on the value of @xmath0 and in the non relativistic regime we have chosen , this number is @xmath107 photons . when the incident electron relativistic parameter is increased from @xmath108 to @xmath109 , the previous mentioned results remain valid but the corresponding dcss are very small indicating a small probability that the very fast projectile electron will exchange photons with the radiation field . as for the behaviour of the dcss with respect to the degree of ellipticity ,
the ellipically and circularly polarized laser modified cross sections become more sharply peaked around the angle @xmath110 .
a similar result has been reported [ 3 ] .
in this work , we have extended the study of the mott scattering process of an electron by a charged nucleus to the case of a general polarization .
we have shown that the mott - scattering geometry as well as the key parameters such as the electric field strength and the incident electron kinetic energy influence the behaviour of the dcss . moreover ,
the degree of ellipticity @xmath0 is also a key parameter for the description of the mott scattering process particularly in the region of large momentum transfer and for a number of photons exchanged lower than that for which the dcss tend to the laser - free one .
li , j. berakdar , j. chen , z .- f .
zhou , phys .
a , * 67 * , 063409 , ( 2003 ) .
y. attaourti , b. manaut , phys .
a , * 68 * , 067401 , ( 2003 ) . c. szymanowski , v. vniard , r. taeb , a. maquet , and c.h .
keitel , phys .
rev . a , * 56 * , 3846 , ( 1997 ) .
v. berestetzkii , e. m. lifshitz and l. p. pitaevskii , _ quantum electrodynamics , 2nd ed . _
( pergamon press , oxford , 1982 ) .
w. greiner and j. reinhardt , _ quantum electrodynamics _ , ( springer - verlag , berlin - heidelberg , 1992 ) .
a. g. grozin , _ using reduce in high energy physics _ ( cambridge university press , 1997 ) .
f. v. bunkin and m.v .
fedorov , zh eksp .
teor . fiz . *
49 * , 1215 , ( 1965)[sov .
jetp * 22 * , 844 ( 1966 ) ] . | in the first born approximation and using an elliptically polarized laser field , the mott scattering of an electron by a coulomb potential is investigated using the dirac - volkov states to describe the incident and scattered electrons .
the results obtained are compared with the results of s.m .
li _
et al _
@xcite for the case of a linearly polarized laser field and with the results of y. attaourti _
et al _
@xcite for the case of a circular polarization .
+ + pacs number(s ) : 34.80.qb , 12.20.ds = 18.6pt plus 0.2pt minus 0.1pt |
asteroseismology is at present being developed as an efficient instrument in the study of stellar interiors and evolution .
pulsational frequencies are the most important asteroseismic observational inputs .
it is evident that a meaningful analysis of the observation , in terms of the basic physics of stellar interiors which is the ultimate target of the investigation , requires reliable computation of oscillation frequencies for specified physics .
this is a two - step process , involving first the computation of stellar evolutionary models and secondly the computation of frequencies for the resulting models .
lebreton et al .
( this volume ) provide an overview of the tests of stellar model calculations . here
we consider the computation of the oscillation frequencies .
an evident goal is that the computed frequencies , for a given model , should have errors well below the observational error , which in the case of the corot mission is expected to be below 0.1@xmath1hz @xcite . for the kepler mission ( e.g. , * ? ? ?
* ) , with expected launch in early 2009 , selected stars may be observed continuously for several years and errors as low as @xmath2hz may be reachable , particularly for modes excited by the heat - engine mechanism . since errors resulting from numerical problems
are typically highly systematic , they may affect the asteroseismic inferences even if they are substantially below the random errors in the observed frequencies .
this must be kept in mind in the assessment of the estimates of the numerical errors . during the last decades a lot of codes obtaining numerical solutions of an adiabatic system of differential equations describing stellar oscillations
have been developed . in order to ascertain whether any possible difference in the description of the same observational data by different numerical codes is due to physical descriptions or to different numerical integration schemes ,
the inter - comparison of these oscillation codes in a fixed and homogeneous framework is absolutely necessary .
some effort has been already done in the past but only regarding pairs of codes .
some codes have also developed a lot of internal precision tests .
however , there is a lack of inter - comparison of a large enough set of codes .
we aim in this study try to fix a set of minimum requirements for a code to be sure that any difference found is only due to a different physical assumption . ideally , for a given model there should be a set of ` true ' frequencies with which the results of the different codes could be compared .
this ideal situation could probably be approximated by considering polytropic models for which it is relatively straightforward to calculate the equilibrium structure with essentially arbitrary accuracy ( see also * ? ? ?
* ) . in practice ,
the situation for realistic stellar models is more complex .
owing to the complexity of the stellar evolution calculation the models are often available on a numerical mesh which is not obviously adequate for the pulsation calculation .
the effect of this on the frequency computation depends on the detailed formulation of the equations in the pulsation codes .
these formulations are equivalent only if the equilibrium model satisfies the ` dynamical ' equations of stellar structure , i.e. , the mass equation and the equation of hydrostatic support , and this is obviously not exactly true for a model computed on a finite ( possibly even relatively sparse ) mesh .
one might define a consistent set of frequencies for a given model by interpolating it onto a very dense mesh and resetting it to ensure that the relevant equations of stellar structure are satisfied .
the model is fully characterized by the dependence on distance @xmath3 to the centre of density @xmath4 and the adiabatic exponent @xmath5 , @xmath6 being pressure and the derivative being at constant specific entropy .
thus one could interpolate @xmath7 and @xmath8 to a fine mesh , and recompute the mass distribution and pressure by integrating the mass equation and the equation of hydrostatic equilibrium .
frequencies of this model should then be essentially independent of the formulation of the oscillation equation and would provide a suitable reference with which to compare other frequency calculations .
such a test may be carried out at a later stage in the esta effort . in task 2 , for now , we aim at testing , comparing and optimizing the seismic codes used to calculate the oscillations of existing models of different types of stars . in order to do
so we consider steps in the comparison by addressing some of the most relevant items that must be compared regarding the seismic characterization of the models : * step 1 : comparison of the frequencies from different seismic codes for the same model . * step 2 : comparison of the frequencies from the same seismic code for different models of the same stellar case provided by different equilibrium codes
* step 3 : comparison of the frequencies for specific pulsators .
the work presented here is mostly focused on step 1 . at this step three
different equilibrium models have been used .
two of them have been computed using cesam @xcite , with 902 and 2172 mesh points , and a third one with 4042 mesh points provided by astec @xcite.we present inter - comparisons using the two models with the larger numbers of mesh points .
the same physics and physical constants ( except @xmath0 ) are used for all the oscillation codes .
frequencies in the range of @xmath9 \,\mu$]hz , belonging to spherical degrees @xmath10 and 3 have been calculated , in order to recover most of the possible values we can find in the observational photometric data .
we present in sect .
2 the equilibrium models used and we analyze their main features .
section 3 is devoted to the different oscillation codes used , discussing common requirements and variations in the treatment . in sect .
4 the direct comparison of frequencies for different frequency ranges and spherical degrees is presented in detail . sections 5 and 6
analyze the same inter - comparison for the values of the large and small separations , respectively .
section 7 discusses the dominant effects that contribute to the differences in frequencies determined by the different codes .
conclusions are given in sect .
to ensure that any difference obtained in the inter - comparisons is only due to differences in the numerical schemes , we imposed to all the codes the use of the same equilibrium models .
these models were supplied in several formats : osc , fgong , srox , and famdl .
the first model was required to have 900 mesh points , and it was provided by cesam .
the differences in this case reached unacceptable values of 2 - 3 @xmath1hz when the same integration schemes are compared , and even 10 @xmath1hz when the use or not of the richardson extrapolation are compared .
the maximum difference found for large separations with this model is @xmath11hz .
this showed that either a larger number of mesh points or richardson extrapolation was necessary .
[ tab:1 ] lllllll @xmath12 & @xmath13 & age & @xmath14 & @xmath15 & mesh & @xmath0 + & & ( in my ) & & & points & @xmath16 $ ] + 1.5 & 3.826 & 1355 & 0.4 & 1.731 & 4042 & 6.6716823 + 1.5 & 3.830 & 1368 & 0.4 & 1.724 & 2172 & 6.67232 + the second model was also provided by cesam ( referred from now on as m2k ) .
it uses a grid , with 2172 mesh points , more suitable for asteroseismic purposes .
general characteristics of the model are presented in table 1 .
these are typical of a @xmath17 doradus star showing oscillations in the asymptotic g - mode regime , and also around the fundamental radial mode . a priori
, solar - like pulsations can not be excluded for this type of star and therefore it can be a good candidate for a global study . in fig .
[ fig:1 ] , @xmath18 ( which is a quantity directly related to the brunt - visl frequency @xmath19 : @xmath20 , where @xmath21 is the gravitational acceleration ) is depicted as a function of the relative radius ( @xmath22 ) in a region of steeply varying hydrogen abundance , and hence mean molecular weight @xmath1 , just outside the convective core .
the model is in a phase of a growing convective core .
if diffusion and settling are neglected this leads to a discontinuity in the hydrogen abundance and hence , formally , to a delta function in @xmath18 ; also , there is a region of ` semiconvection ' at the edge of the core .
in fact , the figure shows that three points in this model display an erratic variation in @xmath18 just in the transition region between the convective and the radiative zone ; also , the mesh resolution of this region of rapid variation seems inadequate .
this is emphasized by fig .
[ fig:2 ] which shows the distribution of mesh points of this model along the stellar radius , indicating that there are not enough mesh points in this transition zone .
as discussed below , these features of model m2k give rise to frequency differences in the comparison , particularly for those modes for which this inner part is critical for their physical description .
the third and last model is a 4042 mesh points model ( from now on m4k ) provided by astec .
general characteristics of the model are presented in table 1 .
it has been computed to have overall characteristics similar to the previous cesam model in order to understand better the differences . however , as discussed by christensen - dalsgaard ( this volume ) particular care has been taken in the treatment of the @xmath1-gradient region ; the semiconvective region was replaced by a region with a steep gradient in the hydrogen abundance , defined such as to ensure neutral stability of the temperature gradient . as a result ,
@xmath18 for this model , also presented in fig .
[ fig:1 ] , shows a well - defined and reasonable behaviour .
the distribution of the mesh points can be found in fig .
[ fig:2 ] . in the central and outer parts of the stellar model ,
the distribution is similar to the cesam model m2k .
it is in the inner zones , particularly in the boundary region between the convective core and the @xmath1-gradient zone , that the models present different mesh - point distributions , with m4k providing a far superior resolution of this critical region .
all oscillation codes involved in this task were asked to provide adiabatic frequencies in the range of @xmath9 \,\mu$]hz and spherical degrees @xmath10 and 3 . in addition , the solution of the equations must satisfy the following requirements : [ tab:2 ] lllllll code & e.f . & i.s .
& rich . & i.v . & @xmath0 & reference + & & & & & @xmath16 $ ] & + adipls & @xmath23 & 2 & y , n & @xmath3 & 6.67232 & @xcite + filou & @xmath23 & 2 & n & @xmath3 & 6.67232 & @xcite + graco & @xmath23 , @xmath24 & 2 & y , n & @xmath3 , @xmath25 & 6.6716823 & @xcite + losc & @xmath24 & 4 & n & @xmath3 & 6.67232 & @xcite + nosc & @xmath23 , @xmath24 & 2 & y , n & @xmath3 , @xmath25 & 6.67259 & @xcite + oscrox & @xmath23 & 4 & n & @xmath3 & 6.6716823 & @xcite + posc & @xmath23 & 2 & y , n & @xmath3 & 6.6716823 & @xcite + pulse & @xmath23 & 4 & n & @xmath25 & 6.6716823 & @xcite + lnawenr & @xmath23 & 2 & n & @xmath3 & 6.67232 & @xcite + * to use the mesh provided by the equilibrium model , no re - meshing is allowed . * to set the lagrangian perturbation to the pressure to zero ( @xmath26 ) as outer mechanical boundary condition . * to use the physical constants prescribed in task 1 . * to use linear adiabatic equations
. nevertheless , some other schemes for the numerical solution of the differential equations ( from now on called for simplicity `` degrees of freedom '' ) remain open .
nine oscillation codes of different research groups in the field have been used in this inter - comparison exercise .
a summary of the participating codes and the different degrees of freedom " provided by each one is found in table 2 and include : * set of eigenfunctions : use of the lagrangian or the eulerian perturbation to the pressure ( @xmath24 or @xmath23 ) .
this obviously affects the form of the equations ; in particular , when using @xmath24 the equations do not depend explicitly on @xmath18 .
* order of the integration scheme : most of the codes use a second - order scheme , but some of them have implemented a fourth - order scheme . * richardson extrapolation : some of the codes using a second - order scheme have the possibility to use richardson extrapolation @xcite to decrease the truncation error ; combining a second - order scheme with richardson extrapolation yields errors scaling as @xmath27 , @xmath28 being the number of mesh points ( e.g. * ? ? ?
* integration variable : two integration variables are used : 1 ) the radius ( @xmath3 ) , or 2 ) the ratio @xmath25 .
the latter variable may reduce the effect of rounding errors in the outer layers ( see sect .
[ sec : indep ] ) .
* despite the requirement that the physical constants be fixed at the values for task 1 , the codes used slightly different values of the gravitational constant @xmath0 , as listed in table [ tab:2 ] .
ideally the equilibrium model should have been computed with the prescribed task 1 value ( @xmath29cgs ) which should then have been used for the oscillation calculations . in practice model
m4k was computed with @xmath30cgs .
using different values of @xmath0 in the oscillation equations clearly gives rise to inconsistencies , with potential effects on the frequencies , as discussed further in sect .
[ sec : grav ] . note that most of the oscillation codes put @xmath31 in the general non - radial differential equations to obtain the radial modes , except for losc that uses the lawe differential equation ( linear adiabatic wave equation for radial modes ) , and for graco for which results will be shown for both sets of equations .
it should be noted that the lawe does not depend on @xmath18 .
all the main characteristics and numerical schemes are presented in previous chapters of this volume .
the nice code ( nosc ) has the options of using @xmath32 or @xmath24 as dependent variables , and @xmath3 or @xmath25 as independent variable .
however , all nosc results presented here use @xmath32 and @xmath3 . the use of the mesh provided in the equilibrium model , rather than meshes optimized for the different kinds of modes , may result in inadequate resolution of the rapid variation of high - order p- and g - modes and hence larger truncation errors in the solution of the differential equations . in these cases , therefore
, the second - order scheme may result in unacceptable errors . here
the use of higher - order integration schemes ( a fourth - order scheme or a second - order scheme followed by richardson extrapolation ) is therefore expected to give better results . for low- and intermediate - order modes
we expect little effect of the use of higher - order schemes .
in this section , the results of the direct frequency inter - comparison are presented .
we have structured this study splitting the frequency range in three parts for the non - radial case , only one for the radial case , and comparing codes only with similar selections of degrees of freedom " .
in addition , the influence of using different selections in the same code ( graco in this case ) is shown ; for these tests the value of @xmath0 in the oscillation calculations was the same as was used to compute the equilibrium model .
these are shown in a single frequency range . in fig .
[ fig:3 ] the results obtained using the model m2k are presented . the reference line for all the inter - comparisons
is selected to be graco . in this figure
the reference frequencies have been obtained using @xmath23 , second order , no richardson extrapolation and @xmath3 as independent variable ( see table 2 ) .
two sets of codes can be identified in the figure , adipls - nosc - graco , with differences lower than @xmath33hz , and all with the same degrees of freedom " , and oscrox - pulse - losc - posc with differences for high frequencies lower than @xmath34hz with graco , but with differences among them around @xmath35hz .
this second set of codes differs from the first one in the use of a fourth - order numerical scheme instead of a second - order one . in the present figure , in fig .
[ fig:5 ] and in figs [ fig:7 ] and [ fig:8 ] , showing inter - comparisons with model m2k , posc has been chosen as representative of the codes using second order plus richardson extrapolation ( see table [ tab:2 ] ) .
we can see how the use of this integration procedure provides similar results as the fourth - order solutions .
these differences , around @xmath35hz for codes using the same integration scheme , and @xmath34hz for codes using different schemes , are larger than the expected precision of the coming observational data .
therefore this effect can change any detailed physical description as interpreted by different oscillation codes . also , we point out that the differences between the codes using fourth - order schemes and graco results using a second - order scheme indicate that the model has an insufficient number of mesh points for asteroseismic studies
. figure [ fig:4 ] shows the results obtained for the model m4k . in the top panel models with second order , no richardson extrapolation , @xmath3 ,
are depicted .
all the differences are lower in magnitude than @xmath36hz , i.e. , two orders of magnitude lower than those obtained for model m2k . therefore improving the mesh , including a doubling of the number of points , provides a very substantial improvement in the precision , making these values more acceptable for theoretical modeling . the middle panel of fig .
[ fig:4 ] presents the differences obtained with models providing fourth - order integration solutions or second - order plus richardson extrapolation .
the global precision here is similar to the previous case or even slightly lower , with differences lower than @xmath37hz .
it is interesting to point out that we can not directly distinguish between a fourth - order integration scheme solution ( oscrox ) or a second - order plus richardson extrapolation ( the rest ) .
however , it is noticeable that the graco - oscrox - posc fall in one group and adipls - nosc in a second , with a slight difference in the latter case .
these two groups are distinguished by the value of @xmath0 ( cf .
table [ tab:2 ] ) , with adipls and nosc having similar but not identical values . this pattern will be found in other cases also .
the losc behaviour is discussed in the next paragraph . finally in the bottom panel of fig .
[ fig:4 ] a comparison between the different degrees of freedom " using only the graco code are presented . as reference
we have used the solution with
degrees of freedom " : @xmath38(@xmath31 , no richardson extrapolation , @xmath23 , @xmath3 ) . for each comparison
we have changed only one of these degrees of freedom " at a time , keeping the rest unchanged ( solutions @xmath39 ) .
the most prominent effect arises from the use of richardson extrapolation , which changes the frequencies by nearly @xmath40hz for the highest - order modes , substantially more than the expected observational accuracies . for model m2k ( see fig . [ fig:3 ] ) , we have similarly found a change of @xmath34hz , reflecting the smaller number of mesh points .
this clearly shows that second - order schemes are inadequate , even for the mesh in m4k , for the computation of high - order acoustic modes ; as expected the effect decreases rapidly with decreasing mode order . the use of @xmath25 as integration variable provides small differences , always lower than @xmath41hz .
these differences are of the order of those obtained in the top and middle panels .
the use of the lagrangian perturbation to the pressure as variable ( @xmath24 ) or the use of the lawe differential equation provide very similar differences , lower than @xmath42hz , but with an oscillatory pattern .
this pattern is very similar to that observed for losc , which also uses lawe to obtain the radial modes . as discussed in sect .
[ sec : dep ] , this oscillatory pattern arises from an inconsistency in the thermodynamics of model m4k which affects @xmath18 ; solutions of equations that do not depend on @xmath18 ( i.e. , the lawe or the equations based on @xmath24 ) are insensitive to this effect . therefore , even for model m4k and radial modes , the use of different integration procedures can give different values for the oscillation eigenfrequencies that are non - physical in nature .
these non - physical sources of differences are mainly some inconsistencies in the equilibrium models ( see sect .
[ sec : dep ] ) and the lack of mesh points .
but when the same numerical schemes are used , the different codes provide very similar frequencies . to illustrate the differences appearing in the case of non - radial modes , the spherical degree @xmath43 has been chosen .
we have divided the frequency spectrum into three regions : 1 ) large frequencies ( @xmath44 \,\mu$]hz ) , 2 ) frequencies around the fundamental radial mode ( [ 80,500 ] @xmath1hz ) , and 3 ) low - frequency region ( @xmath45 \,\mu$]hz ) . in all cases
a study similar to that developed in the radial case has been carried out . for the sake of simplicity
the high - frequency differences are not represented since the results are very similar to those presented for the radial case .
only a slightly higher precision is found in this case .
the results of losc present the same pattern as the radial case , owing to the use of @xmath24 as eigenfunction in that code ( see sect .
[ sec : dep ] ) .
figure [ fig:5 ] shows the results obtained for model m2k when comparing @xmath43 frequencies around the fundamental radial mode .
the main differences are smaller than in the high - frequency region , corresponding to the low order of the modes and the consequent lesser sensitivity to the number of mesh points .
the largest differences are found for two modes of frequency near @xmath46 and @xmath47hz showing avoided crossing ; these modes have a mixed character , with fairly substantial amplitude in the @xmath1-gradient zone .
pulse and losc present differences around @xmath48hz , posc around @xmath49hz , oscrox less than @xmath50hz , and the rest do not present significant differences for these two modes .
that is , the largest differences are found in these codes when using a fourth - order integration scheme or a second - order plus richardson extrapolation .
the values of these differences are larger than the expected precision of the satellite data to come .
they clearly reflect the inadequate representation of the @xmath1-gradient zone in m2k , with higher - order schemes being more sensitive to the resulting inconsistency in the model . in the top panel of fig .
[ fig:6 ] the inter - comparisons of the frequencies ( @xmath43 ) for model m4k , using a second - order scheme without richardson extrapolation , are depicted for the same frequency range as in fig .
[ fig:5 ] .
the main differences are two orders of magnitude lower than those obtained for model m2k , and they are also of the same order of magnitude as those obtained for the high - frequency range with m4k .
all these differences remain always lower than @xmath51hz .
graco and posc are extremely close , while filou and adipls provide very similar results , with slightly larger differences for nosc , by about @xmath52hz , relative to these codes .
thus again the differences are related directly to the different values of @xmath0 .
there remain small wiggles for the mixed modes near @xmath53hz but reduced more than two orders of magnitude relative to the largest differences for model m2k , reflecting the superior resolution of the critical region in model m4k . the middle panel of fig .
[ fig:6 ] shows the same inter - comparison as the top panel for codes using a fourth - order scheme or a second - order plus richardson extrapolation .
the precision is similar to the previous case , with differences of the same order of magnitude .
we can distinguish two groups of codes providing very similar results : oscrox - posc - graco and losc - adipls - nosc .
this distribution does not depend on the integration scheme , but again reflects the values of @xmath0 .
the wiggle of the mixed modes is also very similar to that obtained with a second - order scheme without richardson extrapolation .
finally , an inter - comparison of different degrees of freedom " using only graco is presented in order to test the differences obtained for the different choices .
this comparison is depicted in the bottom panel of fig .
[ fig:6 ] .
the use of the richardson extrapolation is not very important , as expected for these modes of low order , with effects generally smaller than @xmath54hz , although a larger value is present for the mixed modes . using @xmath24 as variable gives differences larger than the differences among codes with the same degrees of freedom " .
the use of the integration variable @xmath25 does not introduce significant differences . compared with model m2k
, we find a general reduction of the differences for m4k , the main effects being in the region of mixed modes with an improvement reaching up to three orders of magnitude .
this is obviously not a simple consequence of the doubling of the number of mesh points .
the main reason is likely the inadequate resolution shown by model m2k in the description of the brunt - visl frequency in the region close to the boundary of the convective core , which is not present in model m4k .
the avoided - crossing phenomenon and the behaviour of the mixed modes are very sensitive to the detailed treatment of this region , including the effects of semiconvection .
therefore , an accurate description of @xmath19 is critical for the oscillation codes in order correctly to obtain eigenfrequencies for modes near an avoided crossing .
the direct frequency inter - comparison ends with the study of the low - frequency region ; as above we concentrate on modes of degree @xmath55 . fig .
[ fig:7 ] shows the results obtained for model m2k . in this case
the differences are lower than obtained at higher frequencies for this model ( cf .
figs [ fig:3 ] and [ fig:5 ] ) .
the most surprising behaviour is present in pulse and losc .
posc and oscrox also present some differences in the same region .
to understand the reason for these differences and the region where they appear fig .
[ fig:8 ] shows the period separation @xmath56 between adjacent modes .
the first - order asymptotic g - mode theory predicts a constant separation of the periods in this regime for a given @xmath57 .
however , when the equations are solved numerically , this period spacing presents several minima , and these minima are directly linked with the mode trapping @xcite .
[ fig:8 ] shows that the position of one of these minima is the same as the position of the largest differences .
the modes in this region have somewhat enhanced amplitudes in the region just outside the convective core . in pulse and losc
this apparently happens for modes somewhat different from the remaining codes .
this is the origin of the frequency differences .
as the mode trapping in this region for these stellar models is related to the brunt - visl frequency in the @xmath1-gradient zone , the previously mentioned inadequate treatment of this region in model m2k is likely the reason for these differences . the top panel of fig .
[ fig:9 ] presents inter - comparisons for @xmath43 frequencies calculated with a second - order scheme without richardson extrapolation and for model m4k , for the same frequency range as in fig .
[ fig:7 ] .
posc is again extremely close to graco , within @xmath58hz .
filou and adipls present very similar results , with nosc showing slightly larger differences , ranging from 0.0005 up to @xmath59hz relative to this group .
the differences decrease globally as far as the frequency decreases until reaching a magnitude of @xmath59hz for the smallest frequency studied here . the middle panel of fig .
[ fig:9 ] shows the same comparison for codes using a fourth - order scheme or a second - order plus richardson extrapolation .
once again it can be seen that the precision resembles the results in the previous panel .
here posc and oscrox provide very similar results to graco .
adipls - losc - noscpresent small differences among them , with nosc having a small increasing difference with frequency with respect to the other two . again the general pattern here , and in the top panel ,
largely reflects the differences in @xmath0 . in this case
we can distinguish the codes using fourth - order integration scheme with its apparently noisy profiles as compared with a solution using second - order integration plus richardson extrapolation . in the bottom panel of fig .
[ fig:9 ] , an inter - comparison for the different degrees of freedom " using only graco is presented ; as in earlier corresponding plots the same value of @xmath0 is used as in the computation of the equilibrium model .
as expected the richardson extrapolation has a growing influence as the frequency decreases and the mode order increases , with quite substantial differences , compared with the differences between different codes , for the lowest - frequency modes .
thus , with the mesh provided by the evolution calculation the second - order schemes have inadequate numerical precision . in this case , the differences provided by the use of @xmath25 as integration variable are negligible ; the use of the lagrangian perturbation to the pressure ( @xmath24 ) gives rise to frequency differences exceeding those obtained between the different codes , at the lowest frequencies ; we note , however , that a corresponding comparison between adipls and losc does not show this effect which may therefore be particular to the gracoimplementation .
finally , we want to point out that in the case of model m4k , the large differences appearing in the mode trapping region are not found .
[ fig:10 ] presents the same period separation as fig .
[ fig:8 ] but for this model .
all the codes give quite similar results .
two mode trapping regions appear with the same frequency domain as in fig .
[ fig:8 ] . as
this model does not present any numerical imprecision in the brunt - vsl frequency pattern , the obvious conclusion is that , as for the mixed modes in avoided crossing , any numerical imprecision in the description of @xmath19 coming from the equilibrium model can give rise to large differences in the frequencies calculated by different oscillation codes for the g - modes trapped in the @xmath1-gradient zone .
this section is devoted to the asymptotic behaviour of p - modes through the use of the so called large separations " , that is , the difference between two consecutive modes with the same spherical degree @xmath57 ( @xmath60 , @xmath61 being the radial order of the mode ) .
the structure of the section is similar to the previous one .
we will use the same definitions as in the previous section to study all the frequencies ranges . from now on , the results with the m2k model will not be discussed , since no additional information is found from the further inter - comparisons .
figure [ fig:12 ] shows the results obtained for model m4k . in the top panel differences for the different codes using second order , no richardson extrapolation , @xmath3 ,
are depicted .
posc and graco are again extremely close .
the rest of the codes ( adipls - nosc - filou - graco ) present differences of up to around @xmath59hz , generally sharing an oscillatory pattern , particularly at relatively low frequency ; we have no explanation for this behaviour .
however , the effect is evidently small . in the middle panel of fig .
[ fig:12 ] the ls differences for the different codes , using a fourth - order integration scheme or a second - order plus richardson extrapolation , are presented . with the exception of losc
the global behaviour is similar to that obtained without richardson extrapolation , the nosc - adipls - oscrox - graco differences being always lower than @xmath54hz .
the pattern of the losc differences , presenting differences one order of magnitude larger than the rest of the codes , is clearly related to the corresponding oscillatory pattern found in fig .
[ fig:4 ] ; as before it is explained by the use of the lawe differential equation .
the bottom panel of fig .
[ fig:12 ] shows the ls differences for the radial modes obtained with graco and model m4k when different degrees of freedom " are used .
the richardson extrapolation introduces differences increasing with frequency and mode order , as expected , giving the largest differences , around @xmath62hz .
this again emphasizes the inadequacy of the second - order schemes for the highest - order modes , on the m4k mesh .
the integration variable @xmath25 gives differences slightly lower than @xmath63hz , i.e. , much smaller than that found for different codes using the same numerical techniques , as depicted in the previous panels .
the use of the lagrangian perturbation to the pressure @xmath24 and the lawe differential equation show the same oscillating behaviour and values as those previously observed for the losc results . as discussed above , this is related to the inconsistency in @xmath18 in model m4k ( see sect . [
sec : dep ] ) . to illustrate the differences appearing in the case of non - radial modes , the spherical degree
@xmath43 has been arbitrarily chosen .
we have divided the frequency spectrum in three regions , like in the direct frequency inter - comparison : 1 ) high - frequency region , 2 ) frequencies around the fundamental radial mode and 3 ) low - frequency region . in all cases , a study similar to that developed in the radial case has been carried out . in the low - frequency region ,
the more physical period separation is studied , instead of the frequency separation relevant for acoustic modes . in the first region
the results are very similar to those obtained for the radial case ; therefore the plots are not presented here . as in the direct frequency
inter - comparison case , losc also presents an oscillating pattern , owing to the use of @xmath24 as eigenfunction ( see sect . [
sec : dep ] ) . on the other hand ,
the only noticeable difference , when compared with the radial case in this region , is that for @xmath43 the precision among codes using the same integration procedures is slightly higher .
the results obtained with the codes using a second - order scheme , for the modes around the fundamental radial mode , are depicted in the top panel of fig .
[ fig:14 ] .
posc remains very close to graco , with the other set of codes ( filou - nosc - adipls ) extending the oscillatory pattern in the top panel of fig .
[ fig:12 ] , with the largest difference being @xmath64hz for the mixed modes .
this set of codes agrees to within differences around @xmath65hz ( slightly larger for nosc ) .
the middle panel of fig .
[ fig:14 ] depicts the differences for a fourth - order integration scheme or second - order plus richardson extrapolation .
the precision of the different codes is similar to that given by the second - order scheme without richardson extrapolation .
the maximum difference is lower than @xmath66hz , and most of the codes present differences around @xmath52hz .
once again the largest differences are obtained in the mixed modes .
no different behaviours depending on the integration scheme are found . the bottom panel of fig .
[ fig:14 ] is devoted to the differences obtained with graco when different options for the solution of the differential equations are chosen . in this frequency region
the replacement of @xmath3 by @xmath25 as integration variable introduces the smallest differences , one order of magnitude lower than those obtained using the codes with the same degree of freedom " . using the lagrangian perturbation to the pressure ( @xmath24 ) as eigenfunction causes some difference indicating sensitivity to whether or not @xmath18 is used .
the richardson extrapolation introduces differences lower than those found for the different codes , except in the avoided crossing zone of mixed modes , where changes comparable to the largest one are found .
given the rather substantial variations in the bottom panel , and the fact that most codes show the same variation in the top two panels , one might suspect that the dominant source of this variation is in fact in the graco results used as reference .
the low - frequency region is studied through the period separation for @xmath55 ( @xmath67 in seconds ) , illustrated in fig .
[ fig:16 ] .
the top panel of this figure shows the differences found using only codes with a second - order integration scheme .
adipls and nosc show similar shifts of around 0.1 s relative to graco .
filou is similar at the higher frequencies but shows a small oscillating behaviour in the mode - trapping regions .
finally posc presents a quite noisy pattern , varying around zero .
again the overall grouping of the differences ( adipls - nosc - filouand graco - posc ) reflects the different values of @xmath0 .
results obtained using a fourth - order integration scheme or a second - order plus richardson extrapolation are compared in the middle panel of fig .
[ fig:16 ] .
the values of the differences found in this case are of the same order as in the previous inter - comparison . here
we can distinguish codes using a fourth - order scheme or a second - order plus richardson extrapolation , because of the apparently random pattern they present , with differences one order of magnitude larger than the main values .
the frequency differences obtained with the oscrox and losc results are those presenting a noisy behaviour , when comparing with a second - order plus richardson extrapolation solution ( as graco does ) as the reference line .
posc also presents some differences in the mode - trapped regions when compared with other codes but using the same integration scheme . the bottom panel of fig .
[ fig:16 ] shows the differences obtained with the same code ( graco ) and different choices of the degrees of freedom " .
the use of the richardson extrapolation introduces substantial differences , of the order of seconds , with noticeable wiggles in the two mode - trapping regions .
as noted previously this reflects the inadequacy of the second - order schemes for high - order modes .
the use of the lagrangian perturbation to the pressure as variable ( @xmath24 ) also introduces substantial differences , particularly around the trapped modes near @xmath68hz , related to the frequency differences found with graco in this region when @xmath24 is used ( cf .
[ fig:9 ] ) . using @xmath25
as integration variable gives a small difference when compared with @xmath3 , one order of magnitude lower than the differences found between the codes using the same numerical integration schemes .
in this section we study the inter - comparisons for the small separations ( ss ) @xmath69 .
therefore , two sets of inter - comparisons can be done , one for the modes and another for the ( @xmath70 ) modes . in this case only the high - frequency region is studied , since this is where this quantity has physical meaning . as in the previous section we concentrate on results for model m4k .
the results obtained for the small separation @xmath71 are presented in fig .
[ fig:18 ] .
the top panel of this figure shows the differences obtained with the codes solving the set of differential equations with a second - order integration scheme . in the avoided crossing region
some wiggles occur related to variations for @xmath43 , but these wiggles are of the order of magnitude of the differences obtained for the high - frequency region .
adipls - posc - gracopresent similar values for high frequencies and nosc - filou presents differences around @xmath72hz , and always lower than @xmath73hz .
these differences are much lower than the expected observational errors for corot .
the middle panel shows the same inter - comparison using a fourth - order integration scheme or a second - order plus richardson extrapolation .
once again we can not distinguish the integration scheme used .
just losc shows an oscillatory pattern for high frequencies due to the use of lawe or @xmath24 ( see sect .
[ sec : dep ] ) .
the order of magnitude of the main differences is the same as those obtained using only a second - order scheme .
the wiggles in the avoided crossing regions are still present , and for high frequencies the differences are all in the range @xmath74 \,\mu$]hz . adipls and posc present an almost constant difference with graco , nosc and oscrox show a small noisy behaviour . the bottom panel of fig .
[ fig:18 ] presents the ss differences induced in graco when different choices of degrees of freedom " are selected . in this case
they are all below , or of the same order of magnitude as , the spread between the different codes illustrated in the middle panel .
interestingly , richardson extrapolation introduces differences for higher frequencies far smaller than found for the large separation ( fig .
[ fig:12 ] ) .
the use of the lagrangian perturbation to the pressure ( @xmath24 ) as variable results in the same oscillatory pattern as seen for losc in the middle panel .
the integration variable @xmath25 gives increasing differences in the range [ 0.001,0.003 ] @xmath1hz , similar to the richardson extrapolation , probably reflecting a difference in the sensitivity of radial and nonradial modes to the choice of independent variable in the graco code ; however , the effect is obviously small .
note that in this case the use of the lawe for the radial modes , keeping the default degrees of freedom " ( including the use of @xmath32 ) for @xmath55 , is hardly meaningful ; thus no results are included for lawe . in fig .
[ fig:20 ] the inter - comparison of @xmath75 is done for model m4k .
the top panel shows the differences given by the codes using a second - order integration scheme .
for posc - adipls - filou - gracothe differences are between @xmath76 and @xmath54hz and nosc shows somewhat larger values . in any case
the precision is good and the patterns rather smooth . the codes using a fourth - order integration scheme or a second - order plus richardson extrapolation have produced the results presented in the middle panel of this figure .
the precision is very similar as in the previous panel or even a little higher .
losc shows the oscillatory pattern already obtained in all the previous inter - comparisons .
posc and adipls present smooth difference profiles when compared with graco . on the other hand oscrox and nosc give a low noisy profile .
once again the integration scheme used can not be discriminated .
finally , the bottom panel presents the effect in @xmath75 of using different numerical integration schemes .
using richardson extrapolation gives large differences in the high - frequency region , even larger than the precision of the different codes while , as expected , the effect is small for low - order modes .
if the differential equations are solved with @xmath25 as integration variable , an almost increased difference is introduced , always lower than @xmath77hz , i.e. , of the order of the precision among most of the codes using the same numerical techniques . the use of the lagrangian perturbation to the pressure as variable introduces the same oscillatory pattern as was obtained with losc .
as indicated in table [ tab:2 ] different values of @xmath0 are used by different codes . in many cases
the value differs from the value , @xmath30cgs , which was used in the computation of model m4k . as a result , as seen by the pulsation code the equilibrium model is not strictly in hydrostatic equilibrium , thus potentially causing errors in the computed frequencies . in this section
we examine the consequence of such inconsistencies .
thus we test the influence of the choice of a particular gravitational constant @xmath0 , within always the recent experimental values found in the literature .
the value of this constant is not very accurately known , and the different values we can find in the literature can have an impact on the frequency calculation of the same order or even larger than the differences studied here . to illustrate this , two extreme values of @xmath0 found in the literature have been chosen : @xmath78 cgs ( as fixed for task 1 ) , and @xmath79 cgs @xcite , the most recent one , although with a quoted random error of @xmath80 and a systematic error of @xmath81 cgs it is consistent with the previous value .
the present recommended value of the committee on data for science and technology ( codata ) can be found in the world wide web at physics.nist.gov/constants , and it is closer to that fixed in task 1 . the comparison has been carried out with graco at fixed
degrees of freedom " ( @xmath23 , @xmath82 , no richardson extrapolation , @xmath3 ) . the differences obtained using both values of @xmath0 are shown in fig .
[ fig:25 ] for modes with @xmath31 and 2 .
the equilibrium model used is m4k .
surprisingly , the differences obtained in the region around the fundamental radial mode are much larger than those obtained for the direct frequency inter - comparison for the different codes or for the same code using different integration numerical schemes .
as expected for radial modes the largest difference is obtained for the fundamental radial mode .
the value of this difference ( more than @xmath83hz in magnitude ) is really considerable . in the asymptotic regions
the influence of the value of this parameter decreases until reaching values lower than @xmath84hz . nevertheless , these differences are larger than those obtained by using or not the richardson extrapolation .
the presence of wiggles for @xmath43 for the mixed modes is also remarkable .
the relative difference between @xmath85 and @xmath86 is around @xmath87 .
in contrast , the relative difference between , say , the value chosen for task 1 and the adipls value is only @xmath88 .
thus we might expect this difference to cause a frequency difference of order @xmath89hz . in fact , both the magnitude and shape of the differences between adipls and graco in the middle panel of fig .
[ fig:4 ] are consistent with this estimate . in general ,
as already discussed , it appears that much of the differences between codes using the same numerical scheme results simply from the different values of @xmath0 .
thus , it is obviously important to ensure that consistent values of @xmath0 are used in the evolution and oscillation calculations .
the most dramatic difference between the different codes is the oscillatory variation shown by the computations using the lawe or @xmath24 , relative to the reference graco results ( e.g. , fig . [ fig:4 ] ) .
the oscillatory nature and the ` period ' of the variation indicated that it reflected a sharp difference in some aspect of the model , located approximately in the region of the second helium ionization zone .
after various failed attempts to identify the cause of the variation it transpired that it resulted from a , previously known ( e.g. , * ? ? ?
* ) , inconsistency in the opal equation - of - state tables ( the original tables of * ? ? ?
* ) used in the computation of model m4k with the astec code . to understand this we consider the computation of @xmath90 from the point of view of stellar evolution calculation this is somewhat inconvenient , since the gradient of @xmath4 does not appear directly in the equations of stellar structure
. thus evaluation of @xmath18 would involve numerical differentiation of @xmath4 .
a potentially more convenient formulation is obtained by rewriting the expression in terms of the temperature gradient which is known from the equation of energy transport , given that @xmath4 is a function , known from the equation of state , of pressure , temperature and composition which we characterize solely by the hydrogen abundance @xmath91 .
thus we obtain @xmath92 here , as usual , @xmath93 and @xmath94 is its adiabatic value .
this expression still requires numerical differentiation of @xmath91 , appears directly in the equations ( see , for example , the description of astec by @xcite ) ] but in much of the model @xmath91 varies little or not at all and hence the term in @xmath95 makes a limited contribution .
furthermore , in the bulk of convective zones , with homogeneous composition , eq .
[ eq : atemp ] accurately reflects the small value of @xmath18 resulting from the small value of @xmath96 determined , e.g. , from the mixing - length treatment . as evaluated from eq .
[ eq : arho ] and evaluated from eq .
[ eq : atemp ] as done in astec , for model m4k , against fractional radius .
the differences have been normalized by @xmath97 . ,
title="fig:",width=302 ] the transformation in eq .
[ eq : atemp ] obviously assumes that the thermodynamical quantities used are mutually consistent . as noted by @xcite , this is not the case of the original opal tables provided by @xcite .
the effect of this on @xmath18 is illustrated in fig .
[ fig : compa ] for model m4k , in terms of the difference between the value of @xmath18 calculated directly from eq .
[ eq : arho ] , using numerical differentiation to evaluate @xmath98 , is calculated from the equation of hydrostatic support or through numerical differentiation . ] and the value , @xmath99 , evaluated from eq .
[ eq : atemp ] .
there are obvious systematic differences , concentrated in the ionization zones of hydrogen and helium .
these would indeed affect the oscillations as a ` sharp feature ' and hence cause an oscillatory signature in the frequencies , with the ` period ' seen in fig .
[ fig:4 ] and elsewhere ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . to test further this interpretation of the results we have recomputed frequencies for model m4k , but replacing @xmath99 with @xmath18 computed from eq .
[ eq : arho ] . for this modified model
the differences between adipls and losc are smaller by an almost an order of magnitude than the differences illustrated in fig .
[ fig:4 ] and with none of the systematic oscillatory character .
we finally note that transforming between the equations using @xmath32 and @xmath24 also depends on the equation of hydrostatic equilibrium and hence on a consistent choice of @xmath0 in the oscillation equations ( cf .
[ sec : grav ] ) .
however , within the actual range of @xmath0-values used by the different codes this effect is smaller than the effect of the inconsistency in @xmath18 . we have found a fairly significant dependence on the @xmath0 value used by the oscillation codes for the differences of using the integration variables @xmath3 or @xmath25 .
[ fig : comprp ] shows this dependence in the case of @xmath100 . in this figure
, the differences between the use of @xmath3 or @xmath25 with the three different @xmath0 values presented in table 2 , are depicted .
all of them have been obtained using graco and m4k .
the differences increase with frequency , and their value depend on the difference @xmath101 , where @xmath102 is the @xmath0 value used to obtain the equilibrium model or the oscillation frequencies , respectively . in fact , the transformation between using @xmath3 and @xmath25 as independent variables uses the equation of hydrostatic equilibrium and hence the result is obviously sensitive to whether or not a consistent value of @xmath0 is used . by far the smallest differences are obtained when the same @xmath0 are used in the equilibrium and oscillation codes .
even when the same value of @xmath0 is used , we have found a slight effect of using @xmath25 as independent variable , as implemented in a few of the codes ( see also @xcite ) , instead of @xmath3 ; in fig .
[ fig : comprp ] the maximum difference is @xmath41hz , as also found in fig .
[ fig:4 ] .
the origin of choosing @xmath25 seems to be a concern that near the surface , where @xmath3 varies little between adjacent mesh points , rounding errors in the evaluation of differences in @xmath3 , when representing the differential equations on finite - difference form , could have a significant effect on the results .
given the rapid variation of @xmath6 in the near - surface layers this problem is obviously avoided if @xmath25 is used instead . while there might well be such problems with using @xmath3 when variables are represented in single - precision form ( with four - byte reals ) it seems unlikely to be a problem when using double - precision variables . on the other hand ,
the present test has been carried out with models transferred in the fgong format , where variables are given with ten significant digits . here
one can not exclude that rounding errors might be significant . to test this jc - d computed frequencies with adipls using both the original binary ( double - precision ) version of m4k and the version resulting from conversion to fgong format .
the frequency differences between these two cases were always less than @xmath103hz , strongly indicating that the use of the fgong format is not a concern with the present level of accuracy .
it is obvious that the choice of independent variable also affects the _ truncation _ error , i.e. , the error in the representation of the differential equations on finite - difference form .
this can presumably be the reason for the differences found with graco when comparing the use of @xmath3 and @xmath25 ( e.g. , fig . [ fig:18 ] ) , although the effects are small .
further analysis is required to decide which is the more accurate representation .
a complete inter - comparison of the frequencies calculated by using a set of oscillation codes has been presented .
nine oscillation codes have been used : adipls , filou , nosc , losc , graco , oscrox , pulse , posc and lnawenr .
the only free parameter for each code has been the integration numerical scheme . in order to ensure that the same physics has been imposed to all the codes two equilibrium models of a @xmath104 star
have been supplied , with 4042 ( m4k ) and 2172 ( m2k ) mesh points in their grids , respectively .
no re - meshing has been allowed .
the model m2k presents inadequate numerical resolution in the brunt - visl frequency close to the boundary of the convective core .
the model m4k does not suffer from these problems .
in the present paper we have used these models only to show the sensitivities of the oscillation codes to possible numerical inaccuracies in the equilibrium models . a complete study of the accuracies of the different evolutionary codes is provided by lebreton et al . in this volume .
several inter - comparisons have been performed : 1 ) direct frequency inter - comparison , 2 ) large separation ( ls ) , 3 ) small separation ( ss ) , and 4 ) different experimental values of the gravitational constant @xmath0 .
* when codes using the same numerical integration schemes are compared , the general precision obtained for model m4k ( @xmath37hz ) is higher than the expected precision of the observational data obtained with the corot mission .
this precision is always one order of magnitude better or even more when compared with that obtained for model m2k ( @xmath105hz ) .
the largest differences are obtained for the high - frequency region . * the use of a second - order integration scheme plus richardson extrapolation or a fourth - order integration scheme
does not improve the agreement between the codes when compared with the use of a second - order scheme . in the cases here
presented it was not possible to distinguish between the use of a fourth - order scheme or richardson extrapolation , which always give very acceptable consistency for model m4k .
* however , the use of a fourth - order integration scheme ( or a second - order plus richardson extrapolation ) introduces differences larger than @xmath84hz ( @xmath106hz for m2k and @xmath40hz for m4k ) compared with a second - order scheme , mainly located in the regions of high order ( high frequency for p - modes and low frequency for g - modes ) .
this indicates that the second - order schemes have inadequate numerical precision on the mesh provided in m4k and , obviously , even more so on the sparser mesh in m2k . * for large separations "
the agreement between different comparable codes using model m4k , with every possible numerical scheme is generally better than the expected precision of the data .
however , the equation - of - state inconsistency in model m4k ( see sect . [
sec : dep ] ) introduces a signature that it potentially significant .
the use of the richardson extrapolation together with a second - order integration scheme , or a fourth - order scheme , gives differences for high - order p- and g - modes relative to the use of only a second - order scheme that are comparable to the expected observational accuracies , indicating that the accuracy of the latter scheme is inadequate .
* the `` small separations '' present differences for model m4k lower than the expected observational accuracies everywhere , although the inconsistency in m4k has a noticeable effect . use of richardson extrapolation has a potentially significant effect , indicating that the precision of second - order schemes may be inadequate . *
the use of @xmath25 as integration variable instead of @xmath3 has a small influence when a consistent value of @xmath0 is used .
however , an inconsistency between the @xmath0 values used to obtain the equilibrium model and the oscillation frequencies can give substantial differences between the use of these integration variables , over the range of @xmath0 values in table 2 .
* for model m2k , the mixed modes near the avoided crossing present large differences , reaching even up to @xmath107hz .
these differences are not present when model m4k is used .
the problems with @xmath19 in m2k are undoubtedly the main source of this difference .
the same behaviour is found for the trapped g - modes .
therefore , the correct numerical description of @xmath19 is critical for the value of the frequencies of these trapped modes or in avoided crossing .
* the use of @xmath24 or @xmath23 as eigenfunction , or the solution of the lawe , may have a significant effect if the equilibrium model is based on thermodynamic quantities that are not internally consistent , as is the case for the opal tables used to calculate model m4k .
further effects arise when there are problems with the brunt - visl frequency ( as for model m2k ) ; here a different choice can change the frequencies of some modes since @xmath19 does not appear in the differential equations when @xmath24 or the lawe are used and it does when using @xmath23 .
* the value of the gravitational constant @xmath0 in the oscillation calculations can introduce non - negligible differences as well , if it is not the same as the value used in the equilibrium model .
when the two extreme values found in the literature are used , such inconsistency yields differences in the range @xmath108 \,\mu$]hz for model m4k , larger than those obtained when different numerical integration schemes are used . differences between the values of @xmath0 actually used by the different codes ,
although less extreme , account for much of the difference between the computed frequencies .
therefore , for a proper pulsational study , we require that the number of mesh points and their distribution must be such as to yield an equilibrium model that satisfies the dynamical equations with sufficient accuracy also in the regions of the star where the physical quantities present rapid variations ( e.g. , the outer layers and @xmath1-gradient zones ) .
in addition , the mesh used in the pulsation calculation must properly resolve the eigenfunctions of the highest - order modes considered .
in the present case model m4k , with 4042 points , appears to satisfy these conditions although the use of a fourth - order integration scheme , or a second - order scheme and richardson extrapolation , is still needed in the oscillation calculation ; these higher - order schemes give significant improvements compared with the use of a simple second - order scheme .
the use of a second - order integration plus richardson extrapolation scheme is not distinguishable , in accuracy and precision , from the use of a fourth - order integration scheme .
a correct physical and numerical description of the brunt - visl frequency is essential when @xmath23 is used as eigenfunction ; in particular , inconsistencies in the equation of state can have serious effects on the frequencies .
inconsistency between the value of @xmath0 used in the oscillation calculation and the value used to compute the equilibrium model , within the range of the different values of @xmath0 found in the literature , may lead to substantial errors in the computed frequencies .
we note that the situation is somewhat different if consistent values of @xmath0 are used in the evolution and oscillation calculations .
then the effect on the frequencies is approximately given , according to homology arguments , as a scaling by @xmath109 .
however , since the product @xmath110 is known extremely precisely from planetary motion in the solar system , any variation in @xmath0 should be reflected in a corresponding change in the assumed value of @xmath111 . if this is the case , and if the model is characterized by a given value of @xmath12 ( as is typically the case ) the effect on the frequencies of changes in @xmath0 are very small ( see also * ? ? ? * ) . in further tests
more care is required to secure the full consistency of the models : a consistent equation of state should be used , and the value of @xmath0 should obviously be the same in the equilibrium - model and the pulsation calculations ; indeed , this strongly argues for including the value of @xmath0 as one of the parameters in the model file . the main conclusion of this extensive investigation , however , it positive : with a properly resolved equilibrium model the broad range of oscillation codes likely to be involved in the asteroseismic analysis of data from corot and other major upcoming projects generally give consistent results , well within the expected errors of the observations .
thus , although the remaining problems in the calculation evidently require attention , we can be reasonably confident in our ability to compute frequencies of given models and hence in the inferences concerning stellar structure drawn from comparing the computed frequencies with the observations .
this work was supported by the spanish pne under project number esp 2004 - 03855-c03-c01 , and by the european helio- and asteroseismology network ( helas ) , a major international collaboration funded by the european commission s sixth framework programme .
a.m. and j.m .
acknowledge financial support from the belgian science policy office ( belspo ) in the frame of the esa prodex 8 program ( contract c90199 ) .
mjpfgm is supported in part by fct and feder ( poci2010 ) through projects poci / cte - ast/57610/2004 and poci / v.5/b0094/2005 .
christensen - dalsgaard , j. & berthomieu , g. : theory of solar oscillations . in _ solar interior and atmosphere _ , ( eds ) cox , a. n. , livingston , w. c. & matthews , m. , space science series , university of arizona press , p. 401478 ,
( 1991 ) christensen - dalsgaard , j. , arentoft , t. , brown , t. m. , gilliland , r. l. , kjeldsen , h. , borucki , w. j. & koch , d. : asteroseismology with the _
kepler mission_. in _ proc .
vienna workshop on the future of asteroseismology _ , ( eds ) g. handler & g. houdek , coast * 150 * , 350356 ( 2007 ) gough , d. o. : comments on helioseismic inference . _
progress of seismology of the sun and stars _ , ( eds )
osaki , y. & shibahashi , h. , springer - berlin , _ lecture notes in physics _ , vol .
* 367 * , 283 318 ( 1990 ) | in order to make asteroseismology a powerful tool to explore stellar interiors , different numerical codes should give the same oscillation frequencies for the same input physics .
any differences found when comparing the numerical values of the eigenfrequencies will be an important piece of information regarding the numerical structure of the code .
the esta group was created to analyze the non - physical sources of these differences .
the work presented in this report is a part of task 2 of the esta group .
basically the work is devoted to test , compare and , if needed , optimize the seismic codes used to calculate the eigenfrequencies to be finally compared with observations .
the first step in this comparison is presented here .
the oscillation codes of nine research groups in the field have been used in this study .
the same physics has been imposed for all the codes in order to isolate the non - physical dependence of any possible difference .
two equilibrium models with different grids , 2172 and 4042 mesh points , have been used , and the latter model includes an explicit modelling of semiconvection just outside the convective core . comparing the results for these two models illustrates the effect of the number of mesh points and their distribution in particularly critical parts of the model , such as the steep composition gradient outside the convective core .
a comprehensive study of the frequency differences found for the different codes is given as well .
these differences are mainly due to the use of different numerical integration schemes .
the number of mesh points and their distribution are crucial for interpreting the results .
the use of a second - order integration scheme plus a richardson extrapolation provides similar results to a fourth - order integration scheme .
the proper numerical description of the brunt - visl frequency in the equilibrium model is also critical for some modes .
this influence depends on the set of the eigenfunctions used for the solution of the differential equations .
an unexpected result of this study is the high sensitivity of the frequency differences to the inconsistent use of values of the gravitational constant ( @xmath0 ) in the oscillation codes , within the range of the experimentally determined ones , which differ from the value used to compute the equilibrium model .
this effect can provide differences for a given equilibrium model substantially larger than those resulting from the use of different codes or numerical techniques ; the actual differences between the values of @xmath0 used by the different codes account for much of the frequency differences found here . |
models have become widely adopted as means to implement many machine learning algorithms and represent the state - of - the - art for many image and speech recognition applications @xcite . as the application space for evolves beyond workstations and data centres towards low - power mobile and embedded platforms , so too must their design methodologies .
mobile voice recognition systems , such as apple s _ siri _ , currently remain too computationally demanding to execute locally on a handset . instead
, such applications are processed remotely and , depending on network conditions , are subject to variations in performance and delay @xcite .
are also finding application in other emerging areas , such as autonomous vehicle localization and control , where meeting power and cost requirements is paramount @xcite . , which replace manually engineered computer algorithms ,
must be trained instead of programmed .
this training involves an optimization process where network weights are adjusted with the objective of minimizing the output error .
these adjustments often involve a variation of the method @xcite .
while these training methods have been automated , much of the design process and choice of network hyper - parameters ( e.g. , number of hidden layers , nodes per layer , or choice of activation functions ) has been historically relegated to manual optimization .
this relies on human intuition and expert knowledge of the target application in conjunction with extensive trial and error @xcite .
this process is difficult , considering the vast network hyper - parameter space which includes : the number of convolutional or hidden layers , the quantity of nodes in each layer , the type of nonlinear activation functions , and many others which depend on the system in - hand .
in addition , the problem with manual hyper - parameter tuning is that there is no guarantee that the process will result in optimal configurations .
not only does the diversity of possible hyper - parameters create extremely large design spaces , but time intensive training phases on comprehensive data sets must also be performed prior to evaluating candidate solutions .
this significant computational overhead renders exhaustive searches intractable , and necessitates the use of automated tools to intelligently explore the solution space while limiting the number of candidate models that must be trained and evaluated . with the proliferation of machine learning on embedded and mobile devices
, application designers must now deal with stringent power and cost requirements @xcite .
these added constraints transform hyper - parameter design into a multi - objective optimization problem where no single optimal solution exists . instead , the set of points which are not dominated by any other solution forms a pareto - optimal front .
simply put , this set includes all solutions for which no other is objectively superior in all criteria .
formally , a solution vector @xmath1 ( where @xmath2 and @xmath3 are two objectives for optimization ) is said to dominate another point @xmath4 if @xmath5 and @xmath6 , or @xmath7 and @xmath8 ; the set of points which are not dominated by any other solution constitutes the pareto - optimal front @xcite .
this paper presents an automated method that effectively trains a neural network to design other neural networks , optimizing for both performance and implementation cost .
this meta - heuristic exploits machine learning to predict the performance of candidate solutions ( modelling the response surface ) ; the algorithm learns which points to explore and avoids the lengthy computations involved in evaluating solutions which are predicted to be unfit .
this leveraging of techniques dramatically reduces the proposed algorithm run - time , which can ultimately result in the reduction of product design time , application time - to - market , and overall costs . while there are many different models , the is a well - known form , which rose in popularity with the advent of the back - propagation training algorithm @xcite
. characterized by a series of cascaded _ hidden _
layers , have a single feed - forward path from the input to output layers .
layers are also fully - connected ; each node has a connection to each and every node in adjacent layers .
when illustrated as a directed graph ( shown in figure [ fig : mlp ] ) , graph connections are representative of signals being multiplied by corresponding weights , and graph nodes the summation of all inputs followed by non - linear activation functions .
the evaluation of such elements , commonly or sigmoid functions , is comparatively simple .
therefore , multiply - accumulate operations and memory accesses remain the dominant tasks in terms of cost @xcite . while structurally simple , determining an optimal configuration is a difficult task due to the large number of parameters inherent to even the simplest designs . for example , given the simple shown in figure [ fig : mlp1 ] , with @xmath9 inputs , a single hidden layer with @xmath10 nodes , and @xmath11 nodes in the output layer , @xmath12 operations must be evaluated at each inference pass and an equal number of weights must be accessed in memory .
starting from this initial configuration , a designer may then choose to include a second hidden layer , as illustrated in figure [ fig : mlp2 ] ; the first will then act as a feature extractor and the newly added layer will then process only the limited number of features generated by the first .
this alteration allows for the reduction in dimension of the first hidden layer , which reduces the total number of connections to the input layer .
however , this also increases the network depth and results in a cost penalty ( in terms of requiring additional memory accesses and arithmetic operations associated with the newly added layer ) .
the key problem demonstrated is that even for these two simple configurations , there is no systematic way to determine which design yields superior performance without having trained and evaluated both .
even when the designer has _ a priori _ knowledge of the application , determining the optimal hyper - parameters is non - intuitive , especially for deep networks .
a concrete example of the described problem would be the design of an embedded system to recognize handwritten numerical characters ( such as those contained in the dataset ) .
if the implementation goal is throughput of categorized digits , and a penalty is incurred when a character is misclassified ( perhaps requiring manual intervention ) , then a smaller network that requires fewer clock cycles to evaluate may still result in an overall throughput exceeding those of more accurate alternatives .
in such a scenario , the engineer would require knowledge of the cost and performance of all pareto - optimal solutions in order to meet all requirements with the lowest implementation costs .
even more complex structures are those of , which have demonstrated state - of - the - art results in image recognition @xcite .
the use of convolutional layers further complicates the design process as they introduce a separation between the memory and processing requirements .
illustrated in figure [ fig : cnn_graph ] , convolutional layers are composed of trainable filters ( five _ 3-by-3 _ kernels are shown ) .
have the advantage of reduced memory requirements due to each convolutional filter reusing the same kernel weights for all input values .
however , this is at the expense of increased processing demands , the result of each convolutional filter requiring @xmath13 multiplication operations for each input value ( where the convolutional kernels are sized _ n - by - n _ ) . in a , each filter produces a processed copy of the input data ( as illustrated in figure [ fig : cnn_graph ] ) ; and without any form of down - sampling , this greatly increases the computational complexity of the following layers .
an example of a possible efficient down - sampling method is the inclusion of max - pooling ( or mean - pooling ) after convolutional layers @xcite .
illustrated in figure [ fig : cnn_graph ] , max - pooling refers to partitioning the filtered images into non - overlapping _ k - by - k _ regions , with each outputting the maximum pixel value within ( alternatively , mean - pooling would output the mean value ) .
all of these additional parameters further increase the design space dimensionality , forcing a designer to not only choose how many filters to use in each layer , but also the kernel and pooling sizes .
this greatly affects both the performance and computational complexity of the resulting networks .
the key contributions of this work are presented as follows : * we introduce a method , which employs techniques to predict classification accuracy , to automate the design of hyper - parameters .
this method is then validated with and designs targeting the -10 and image recognition datasets @xcite .
* we demonstrate that multi - objective hyper - parameter optimization can successfully be used as a method to reduce implementation cost ( computational complexity ) .
this work exists at the intersection of two fields : automated hyper - parameter optimization , and reduction of computational complexity . to our knowledge
, this work is the first that has applied automated hyper - parameter optimization as a method of reducing computational complexity .
the design of neural network hyper - parameters has long been considered to be unwieldy , unintuitive , and as a consequence , ideal for automated hyper - parameter optimization techniques such as in @xcite , @xcite , @xcite , and @xcite .
however , these works have been developed with the sole purpose of optimizing performance , with little regard to the resulting computational resource requirements .
two types of sequential hyper - parameter optimization algorithms were presented in @xcite ; in both cases experimental results compared positively to human designed alternatives .
these findings were echoed in @xcite , where it was demonstrated that a random search of a large design space , which contains areas that may be considered less promising , can at times exceed the performance of manually tuned hyper - parameters .
positive results were also presented in @xcite where similar algorithms were extended using a method to extrapolate the shape of learning curves during training , so as to evaluate fewer epochs for unfit solutions and reduce design time . since parameters which perform favourably for
a large network may not be optimal for a smaller alternative , and the most important hyper - parameters ( with the greatest impact on resulting performance ) may vary for different data sets , the need for multi - objective optimization ( especially where low power platforms are targeted ) is clear @xcite .
additionally , in @xcite an automated method was applied to the multi - objective optimization problem of application - specific design , a field which also consists of high - dimensionality solution spaces .
it was demonstrated that through the presented method could efficiently identify pareto - optimal architectures .
there also exists a body of work attempting to reduce the computational complexity of models through weight quantization or the removal of extraneous node connections ( pruning ) .
the research in @xcite and @xcite are examples of two methods that construct networks which reduce the need for multiplications , or modify already trained networks in order to minimize the number of non - zero weights . in @xcite , the authors attempted to reduce the computational resources required when evaluating fully - connected , as well as convolutional , networks through representing all weights as binary values of @xmath14 or @xmath15 .
doing so reduces the number of multiplication operations performed each forward pass ; however the requirement to store floating - point weights during training remains . the work in @xcite also compared trained binary weighted networks to traditional equivalents with equal layer dimensions and similar performance was demonstrated .
however , @xcite only considered very large network dimensions ; further benefits may be obtained from smaller optimized architectures . instead of restricting weights to specific values , in @xcite a pruning method
was presented in which a network can be trained while reducing the number of non - zero weights .
the resulting compressed networks have lower bandwidth requirements and require fewer multiplications due to most weights being zero .
the results in @xcite demonstrated up to 70% reductions in the numbers of floating - point operations required for various networks , with little to no reduction in performance .
however , such pruning methods are not mutually exclusive to the use of tools and could very well be implemented in conjunction with the presented methodology in order to compress an already optimized network configuration .
this work presents a approach that searches for pareto - optimal hyper - parameter configurations and has been applied to both and topologies .
the design space is confined to : the numbers of and convolutional layers , the number of nodes or filters in each layer , the convolution kernel sizes , the max - pooling sizes , the type of activation function , and network training rate .
these degrees of freedom constitute vast design spaces and all strongly influence the resulting networks cost and performance
. for design spaces of such size , performing an exhaustive search is intractable ( designs with over @xmath16 to @xmath0 possible solutions are not uncommon ) , therefore we propose to model the response surface using an for regression where the set of explored solution points is used as a training set .
the presented meta - heuristic is then used to predict the performance of candidate networks ; only points which are expected not to be pareto - dominated are explored .
a flowchart describing the implementation is shown in figure [ alg : dse ] .
the general steps performed during the design space exploration can be broken down into : 1 .
sample the next candidate point from a gaussian distribution centred around the previously explored solution ( or sample a random point for the first iteration ) .
2 . predict the candidate solution performance using the neural network , and calculate the cost as : + @xmath17 3 . compare the predicted results to the current pareto - optimal front .
if the candidate is predicted to be pareto - dominated , it is accepted with probability @xmath18 , otherwise it is accepted with probability @xmath19 .
if rejected , the previously explored solution is rolled back and the algorithm returns to step 1 .
if accepted , the candidate model is trained , tested , and the evaluated results are added to the training set of the ( which is then retrained ) .
finally , if the training set size exceeds the maximum number of desired iterations , the process ends .
otherwise , the algorithm returns to step 1 and a new solution is sampled .
the sampling strategy proposed is an adaptation of the metropolis - hastings algorithm . in each iteration
a new candidate is sampled from a gaussian distribution centred around the previously explored solution point . performing this
random walk limits the number of samples chosen from areas of the design space that are known to contain unfit solutions , thereby reducing wasted exploration effort .
however , exploring an inferior solution may eventually lead to those of superior performance , therefore the probability of accepting such a solution ( @xmath18 ) must remain greater than zero ; this also ensures that the training set for the remains varied .
all experimental results in section [ sec : experimental_results ] were obtained with @xmath20 .
we choose to model the response surface using a model with an input set representative of network hyper - parameters and a single output trained to predict the error of corresponding networks .
this is composed of two hidden layers and a linear output layer .
experimental results demonstrated that sizing the hidden layers with @xmath21 to @xmath22 the number of input nodes provided the best performance .
the network inputs are formed as arrays characterizing all explored dimensions .
integer input parameters ( such as number of nodes in a hidden layer , or size of the convolutional kernels ) are scaled by the maximum possible value of the respective parameter , resulting in normalized variables between 0 and 1 . for each parameter that represents a choice where the options have no numerical relation to each other ( such as whether or sigmoid functions are used ) an input mode is added and the node that represents the chosen option is given an input value of 1 , all others @xmath15 .
for example , a solution with two hidden layers with 20 nodes each ( assuming a maximum of 100 ) , using ( with the other option being sigmoid functions ) and with a learning rate of 0.5 would be presented as input values : @xmath23 $ ] .
the model was trained using , where 100 training epochs were performed on the set of explored solutions each time the next is evaluated ( and in turn , added to the training set ) .
the learning rate was kept constant , with a value of 0.1 , in order to train the network quickly during early exploration , when the set of evaluated solutions is limited .
we evaluated our strategy on applications targeting the standard and -10 image recognition datasets . in the case of designing models targeting the simpler problem ,
the results generated by the algorithm were compared to the true pareto - optimal front obtained from an exhaustive search performed on a constrained solution space .
such a limitation of the design space was required in order to render the exhaustive search tractable .
the algorithm was also evaluated on much larger design spaces for both and models targeting as well as an even larger one for the design of targeting -10 . in all cases only a few iterations were required in order to converge to approximated pareto - optimal fronts . in order to perform the ,
all models were trained and tested using the _ theano _ framework in order to take advantage of optimizations @xcite .
this allowed the entire and evaluation to be performed using a nvidia _
tesla k20c _ . upon completion ,
all explored network data and simulation results were stored to disk , allowing future design re - use .
to model cost , we assumed normalized memory access and multiply - accumulate operation costs .
these can be quantified at either the software or hardware levels ( depending on the target application ) . in both cases
the costs ( such as power , area , or time ) are implementation specific and the exact values used are transparent to the algorithm .
normalized costs assumed for all the experimental results are shown in table [ table : costs ] , which are based on the energy costs for 32-bit floating - point operations from @xcite .
m0.275 < m0.275 < m0.275 < operation & energy cost from @xcite & normalized cost + addition & @xmath24 & n / a + multiplication & @xmath25 & n / a + multiply - accumulate & @xmath26 & @xmath27 + dram memory access & @xmath28 & @xmath29 + [ table : costs ] we experimentally validated the algorithm on two separate image recognition databases ; future work is planned to address a broader range of benchmarks .
the first , , contains @xmath30 pixel grayscale images of handwritten numeric characters ( from 0 to 9 ) .
the second , -10 , is a much more complicated dataset composed of @xmath31 rgb colour images , each belonging to one of ten object categories .
the network hyper - parameters composing the design spaces explored are outlined in table [ table : dse_parameters ] .
in addition , several parameters were kept constant for all experiments : output layers were composed of _ softmax _ units ( with _ jth _ output defined as @xmath32 for a layer with @xmath33 nodes ) and all network training was performed with categorical cross - entropy @xmath34\right)$ ] as loss function @xcite .
finally , for all except the reduced design problem , batch normalization was included after each network layer in order to smooth the response surfaces @xcite .
[ table : dse_parameters ] in order to evaluate the efficiency with which the method approximates the true pareto - optimal front , we first compare experimental results to those of an exhaustive search targeting the design of models for the dataset . in order to make an exhaustive search tractable
, we limited the design space to only the values outlined in the first section of table [ table : dse_parameters ] .
this resulted in a moderate design space of @xmath35 solutions , all of which were trained and tested .
the results of executing the algorithm for 200 iterations ( each iteration represents a design training , evaluation , and model update pass ) are plotted alongside those of the exhaustive search in figure [ fig : mnist_exhaustive ] .
these results demonstrate that the true pareto - optimal front is very closely approximated by the presented method , while requiring very few solutions to be evaluated .
however , it should be noted that is by nature non - deterministic and training the same network twice may yield different performances .
consequently , the generated results ( shown in figure [ fig : mnist_exhaustive ] ) dominate those of the exhaustive search at several points .
this figure also demonstrates that the majority of solutions evaluated by the algorithm remain within a close vicinity of the true pareto - front , successfully avoiding pareto - dominated areas .
since the true pareto - optimal front is known for this restricted case , we use as the metric of evaluation ; quantifying how closely the approximated set differs from the exact @xcite .
is defined in eq . , where @xmath36 and @xmath37 are the approximate and exact pareto - optimal fronts , and the @xmath38 function represents the normalized distance between solutions @xmath39 and @xmath40 .
@xmath41 the evolution of the approximated pareto - optimal set discovered by the algorithm , as a function of iterations completed is plotted in figure [ fig : mnist_mlp_reduced_results ] .
evident from this figure is that the optimal front obtained from the algorithm progressively approaches the true pareto - optimal , while evaluating only a comparatively small number of iterations .
the proposed method identifies the optimal solutions with high accuracy , achieving low values of 7.1% after completing only 50 iterations , 5.0% after 100 , and 3.6% after 200 .
the results in table [ table : adrs ] also demonstrate fast convergence , with the largest changes occurring over the first 30 iterations .
further execution yields more gradual changes as the approximated pareto - front asymptotically approaches the exact . decreasing design time ( both in terms of computation and man - hours )
, these results also demonstrate that by exploring less than 1% of the design space , the method closely predict which hyper - parameters are required for optimal solutions . in addition , the generated set of pareto - optimal configurations also exposes the trade - offs application designers may want to make for cost - constrained systems , allowing for more informed design choices to be made .
< m0.225 < m0.2 < iterations & of explored set & execution time + 10 & 30% & 2.0 min + 20 & 21% & 4.1 min + 30 & 8.4% & 6.4 min + 50 & 7.1% & 11.2 min + 70 & 6.1% & 16.4 min + 100 & 5.0% & 25.4 min + 150 & 4.5% & 45.2 min + 200 & 3.6% & 70.1 min + [ table : adrs ] in order to evaluate performance of the heuristic method for much larger designs , the algorithm was run on the remaining spaces described in table [ table : dse_parameters ] for both and design problems . the total execution times ( on an intel _
xeon e5 - 1603 _ with 8 gb of and a nvidia _ tesla k20c _ ) for the design examples are listed in table [ table : runtime ] and the corresponding pareto - optimal results ( plotted as functions of the number of iterations completed ) are shown in figure [ fig : experimental_results ] . in these plots , the colour scheme presents solutions with low error in blue , and low cost in red .
even though the outputs can not be directly compared to the true results from an exhaustive search ( the -10 example design space exceeds @xmath16 solutions ) , the trends discussed in section [ sec : exhaustive ] are mirrored by all plots in figure [ fig : experimental_results ] .
m0.27 < m0.29 < m0.29 < target & algorithm execution time & mean solution evaluation time + ( ) & 3.6 h & 30 s + ( ) & 18 h & 2 min + -10 ( ) & 66 h & 8 min + [ table : runtime ] because of the intractable nature of such exhaustive searches , the generated results are not expected to always predict the true pareto - optimal fronts . instead
, automated exploration must only match , or exceed , the fitness of manually designed networks in order to provide cost reductions . in comparison with manually designed architectures in literature ,
the pareto - optimal results in figure [ fig : mnist_cnn_results ] include points that offer equivalent performance to the designs in @xcite and @xcite , with implementation costs as low as 25% of their manually designed counterparts ( when weighted with the same cost model detailed in table [ table : costs ] ) .
this demonstrates that the proposed method can design with vastly reduced cost requirements and same levels of performance when compared to manual design approaches .
furthermore , the pareto - optimal results can also find data points with substantial cost savings penalized only by a small decrease in performance . given that increasing accuracy by only 0.01%
requires a doubling in implementation cost ( for designs shown in figure [ fig : mnist_cnn_results ] ) , these additional points are invaluable for application platforms with extremely stringent cost budgets . in order to validate the assumption that a neural network can be trained through regression to model the response surface with sufficient accuracy , the prediction error is plotted in figure [ fig : rsm_error ]
this graph plots the absolute value of the error ( % difference between the predicted and the evaluated performance of each explored solution ) for each of the 500 algorithm iterations performed during the design example ( with results in figure [ fig : mnist_cnn_results ] ) .
the narrow error spikes , which are expected , occur at points where the algorithm encounters previously unexplored areas . as these solutions are added to the training set , the prediction error decreases as the response surface model
is updated , and the spikes begin to occur less frequently . outside of these sparse peaks ,
the prediction accuracy is exceptionally high ; the mean error over the last 95 iterations ( all points after the last spike ) is only 0.35% .
a method to automate the multi - objective optimization of neural network hyper - parameters , reducing both algorithm error and computational complexity , was presented . when compared to the results of an exhaustive search on a restricted design space targeting the dataset , the presented method was shown to dramatically reduce computation time required for convergence to a pareto - optimal solution set .
a low of 5% was achieved after only 100 iterations ; in practice , fewer solution evaluations may be required , with corresponding execution times less than those in table [ table : runtime ] .
furthermore , scalability of the method was demonstrated on larger design spaces for both and models targeting the -10 dataset as well .
even when evaluated on massive design spaces , the presented method was found to still efficiently converge to a diverse pareto - optimal front .
not only was the automation of hyper - parameter design process demonstrated to be both feasible and time efficient , but when compared to manually designed networks from literature , the automated technique produced results with near identical performance while reducing the associated costs by a factor of @xmath42 .
as applications for make further inroads in mobile and embedded market segments , the need to reduce time - to - market and costs will necessitate the use of such automated design methods .
this work was supported in part by a scholarship from the , as well as equipment donations from nvidia corporation .
m. courbariaux , y. bengio , and j .-
binaryconnect : training deep neural networks with binary weights during propagations . in _ advances in neural information processing systems 28
_ , pages 31233131 .
curran associates , inc . , 2015 .
t. domhan , j. t. springenberg , and f. hutter .
speeding up automatic hyperparameter optimization of deep neural networks by extrapolation of learning curves . in _ proc .
of the twenty - fourth int .
joint conf . on artificial intelligence ,
ijcai _ , pages 34603468 , july 2015 .
s. ishibushi et al .
statistical localization exploiting convolutional neural network for an autonomous vehicle . in _ industrial electronics society , iecon 2015 - 41st annual conf . of the ieee _ , pages 13691375 , nov 2015 .
a. krizhevsky , i. sutskever , and g. e. hinton .
classification with deep convolutional neural networks . in _ advances in neural information processing systems 25 _ , pages 10971105 .
curran associates , inc . , 2012 .
t. okabe , y. jin , and b. sendhoff .
a critical survey of performance indices for multi - objective optimisation . in _ evolutionary computation , 2003 .
the 2003 congress on _ , volume 2 , pages 878885 , dec 2003 .
s. r. young et al . optimizing deep learning hyper - parameters through an evolutionary algorithm . in _ proc . of the workshop on machine learning in high - performance computing environments
_ , mlhpc 15 , pages 4:14:5 , new york , ny , usa , 2015 . | artificial neural networks have gone through a recent rise in popularity , achieving state - of - the - art results in various fields , including image classification , speech recognition , and automated control .
both the performance and computational complexity of such models are heavily dependant on the design of characteristic hyper - parameters ( e.g. , number of hidden layers , nodes per layer , or choice of activation functions ) , which have traditionally been optimized manually . with machine learning penetrating low - power mobile and embedded areas ,
the need to optimize not only for performance ( accuracy ) , but also for implementation complexity , becomes paramount . in this work
, we present a multi - objective design space exploration method that reduces the number of solution networks trained and evaluated through response surface modelling .
given spaces which can easily exceed @xmath0 solutions , manually designing a near - optimal architecture is unlikely as opportunities to reduce network complexity , while maintaining performance , may be overlooked .
this problem is exacerbated by the fact that hyper - parameters which perform well on specific datasets may yield sub - par results on others , and must therefore be designed on a per - application basis . in our work
, machine learning is leveraged by training an artificial neural network to predict the performance of future candidate networks .
the method is evaluated on the and -10 image datasets , optimizing for both recognition accuracy and computational complexity .
experimental results demonstrate that the proposed method can closely approximate the pareto - optimal front , while only exploring a small fraction of the design space . |
the radio and fir luminosities of star - forming galaxies are linearly correlated over three decades in luminosity @xcite . this fir - radio correlation ( frc )
is driven by star - formation .
starlight emitted by young , massive stars in the optical and uv is reprocessed by dust into far - infrared radiation . at the same time , the supernova remnants of massive stars accelerate cosmic rays ( crs ) , which produce synchrotron radio emission in the galaxies magnetic fields .
the relation at @xmath2 is linear except in the very lowest luminosity galaxies @xcite .
since the frc has less than a factor of 2 scatter in the local universe , radio luminosity is often used as a tracer of star - formation at both low and high redshift ( e.g. , * ? ? ?
* ; * ? ? ?
although the frc has been mainly studied in the low-@xmath1 universe , there has been recent interest in understanding how it applies to star - forming galaxies at high @xmath1 .
there have been several conflicting results observationally .
a number of studies have found that the frc holds unchanged or with little evolution at high redshifts ( e.g. , @xcite ; @xcite suggest radio - dim ulirgs at high @xmath1 ) .
submillimeter galaxies , however , seem to be radio - bright , by a factor of @xmath5 ( @xcite ) .
in particular , @xcite and @xcite argue that their samples of submillimeter galaxies are intrinsically radio bright with respect to the local frc , with no significant contamination from radio bright agn .
in addition to the work on individual star - forming galaxies , there have also been new investigations into both the recently - resolved fir and the unresolved ghz radio backgrounds .
the fir background was detected by cobe and has long been attributed to star - formation ( see the reviews by * ? ? ?
* ; * ? ? ?
the star - formation origin of the fir background was recently confirmed by blast @xcite .
a detection of the extragalactic ghz radio background has also been recently reported by arcade2 @xcite .
predictions for the strength of the diffuse radio background from star formation rely on the frc holding out to @xmath6 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , where most star - formation occurs . in this paper
, we describe how and why the linearity and normalization of the frc are broken ( or preserved ) for different types of galaxies at high @xmath1 .
we describe the basic physics of our model and the elements of the high-@xmath1 universe that physically allow the frc to break down in [
sec : theory ] . in
[ sec : procedure ] , we briefly describe our underlying procedure . in [ sec : results ] , we discuss our results , providing predictions for when star - forming galaxies should be more or less radio luminous than the @xmath2 frc suggests .
we also summarize the implications of our predictions for radio emission as a star formation tracer in [ sec : sftracer ] and estimate the effects of radio - bright smgs on the diffuse radio background in [ sec : radiobackground ] . throughout this paper ,
primed quantities denote the source s rest - frame and unprimed quantities are for the observer - frame .
@xcite ( ltq ) recently presented a theory of the frc over its entire span at low redshift , from normal galaxies like the milky way to the densest ulirgs like arp 220 .
the linearity and normalization of the frc provide strong constraints on the magnetic energy density in galaxies , which governs the strength of synchrotron losses . indeed , the energy density of starlight in dense galaxies is so large that inverse compton losses alone shorten the cr electron and positron lifetime to @xmath7 in galaxies like arp 220 @xcite .
the magnetic field in arp 220-like starbursts must exceed mg for the cosmic ray electrons and positrons to radiate a large enough fraction of their energy in synchrotron emission to account for the observed ghz radio emission .
such galaxies are likewise sufficiently dense that ionization and bremsstrahlung losses are also rapid .
these considerations imply that the magnetic energy density must increase dramatically from normal spirals to dense bright starbursts .
the observed schmidt law allows us to quantify these dependences since it connects the star - formation rate surface density ( @xmath8 ) , which is proportional to the energy density in starlight ( @xmath9 ) , to the gas surface density : @xmath10 according to @xcite , or @xmath11 according to @xcite . using the @xcite schmidt law
, ltq found that the magnetic field strength @xmath12 must increase as @xmath13 or @xmath14 , where @xmath15 , implying almost a thousandfold increase in @xmath12 from milky way - like galaxies to arp 220-like ulirgs .
this strong dependence of @xmath12 on the global properties of the galaxy is largely responsible for the radio emission from starbursts and is required by the existence of the frc .
it is also crucial to the physics of how the frc is maintained or broken at high z. ltq was unable to distinguish between the possibilities of @xmath12 scaling with @xmath16 and @xmath12 scaling with @xmath17 .
ltq found that over most of the range of the frc , normal galaxies and starbursts are calorimetric .
they are electron calorimeters , meaning cr electrons lose most of their energy before escaping .
they are also uv calorimeters , meaning uv light is efficiently absorbed by dust and converted to fir light .
if cr electrons cool mainly by synchrotron losses , then the ratio of fir emission to radio emission from primary cr electrons should be constant , since each is some fraction of the power released by star - formation .
this is the essence of calorimeter theory , first proposed by @xcite .
however , ltq found that even in this strict calorimeter limit , @xmath18 should decrease systematically by @xmath19 from low surface brightness galaxies to dense ulirgs .
the critical synchrotron frequency causes this `` @xmath20 effect '' : at fixed frequency , as the magnetic field strength increases with increasing surface density , lower energy electrons dominate the synchrotron emission because the synchrotron frequency of electrons , @xmath20 , is proportional to @xmath21 , where @xmath22 is the electron energy . since the injected cr spectrum is likely steeper than @xmath23 , there is more power radiated at low energies .
thus , radio emission increases with surface density of gas and star formation , even including only synchrotron losses in the pure electron and uv calorimeter limit . on top of calorimetry
, ltq proposed that two different conspiracies combine to produce the linear frc , each working in a different surface density regime .
the `` low-@xmath16 conspiracy '' arises as both uv calorimetry and electron calorimetry gradually fail at relatively low surface densities . ]
( @xmath24 ; @xmath25 ) .
cosmic ray electrons escape diffusively without radiating all of their energy , decreasing the radio emission from the calorimetric expectation .
however , some uv photons also escape without being reprocessed into fir by absorption .
quantitatively , each effect alters the luminosity by a factor of @xmath26 in the very lowest surface density galaxies ( @xmath27 ; @xmath28 ) .
therefore , both the fir and radio emission decrease roughly equally , to maintain a constant @xmath18 ( see also * ? ? ?
* ) . at the other extreme ,
high surface density compact starbursts have two completely different opposing processes that affect the radio emission .
compact starbursts are expected to be proton calorimeters : cr protons lose almost all of their energy through pion interactions with the dense ism of starbursts , emitting @xmath29-rays , neutrinos , and secondary electrons and positrons ( * ? ? ? * ; * ? ? ?
* ; * ? ? ?
the secondary electrons and positrons themselves emit synchrotron radio emission , and outnumber the primary cr electrons .
proton calorimetry therefore greatly enhances the radio emission , all else being equal .
although escape losses are negligible in starbursts , synchrotron does compete with the other important cooling processes : inverse compton , bremsstrahlung , and ionization ( * ? ? ?
these are particularly important in bright , dense starbursts , because they have very high gas densities and photon energy densities .
these other losses reduce the share of cr electron and positron power going into radio by a factor of @xmath30 . between the secondaries and the @xmath20 effect increasing @xmath31 , and the non - synchrotron losses suppressing @xmath31 ,
a linear frc correlation is produced .
this is the `` high-@xmath16 conspiracy '' of ltq , and it operates to produce a linear frc even though these starbursts are completely calorimetric ( for protons , electrons , and positrons ) and escape losses are negligible . galaxies at high redshift can be used to test the theory of the frc presented in ltq . at high @xmath1
, we expect that both the low- and high-@xmath16 conspiracies are unbalanced for some galaxies . at low @xmath16 ,
the energy density of the cmb ( @xmath32 ) competes with the energy density in both starlight and magnetic fields .
crs in galaxies face increasing losses from inverse compton since the cmb is stronger .
normal galaxies therefore should be radio dim , as inverse compton losses divert power from synchrotron .
starbursts , however , where crs already face intense ic losses from starlight , should not be affected except at the highest redshifts ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in this paper
, we quantify how much the radio emission is dimmed by increased ic losses from the cmb as a function of @xmath1 . on the other hand , high-@xmath1
starbursts often have different morphologies than those observed of low-@xmath1 ulirgs .
local starbursts including ulirgs are typically compact , with radii of a few hundred parsecs and vertical scales heights of less than a hundred parsecs @xcite .
these properties do not hold for submillimeter galaxies ( smgs ) , very bright ( @xmath33 ) starbursts at @xmath3 mostly powered by star - formation and not agns ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
smgs are usually several kpc in diameter ( several times larger than low-@xmath1 ulirgs ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and have much lower surface densities than their low-@xmath1 counterparts of the same luminosity ( @xcite , but see @xcite ) .
submillimeter galaxies have large random velocities compared to their rotation speeds , implying scale heights of @xmath34 @xcite . at high redshift
, we therefore must consider a class of _ _ puffy starbursts _ _ , with @xmath35 ( @xmath36 ) but scale heights of @xmath37 , in addition to the @xmath38 `` compact starbursts '' typical of local ulirgs considered in ltq .
importantly , puffy starbursts will have a smaller volume density for a given @xmath16 and @xmath8 .
since bremsstrahlung , ionization , pion , and possibly synchrotron losses all depend on the volume density instead of surface density , these loss processes may all be weaker in puffy starbursts like smgs .
with these losses suppressed , the high-@xmath16 conspiracy will become unbalanced , and the frc can be broken .
these starbursts can either be radio - bright or radio - dim , depending on their magnetic field strengths .
since the scale height is so large and the volume density is relatively low for smgs , we can determine whether @xmath12 scales with @xmath16 or @xmath17 with these galaxies , breaking the degeneracy in ltq . because @xmath15 , if @xmath12 increases with @xmath17 , then the magnetic field in puffy starbursts will be weak and synchrotron radio emission will be dim compared to compact starbursts with the same @xmath16 .
however , if @xmath12 increases with @xmath16 , the magnetic field will be strong and synchrotron radio emission will be bright in puffy starbursts compared to compact starbursts with the same @xmath16 .
we model galaxies and starbursts as uniform disks of gas , with scale height @xmath39 , star formation rate surface density is directly proportional to @xmath8 .
the conversion factor is @xmath40 , where @xmath41 is the star formation rate surface density in units of @xmath42 . ]
@xmath8 , and gas surface density @xmath16 .
we solve the diffusion - loss equation to find the steady - state equilibrium cr spectra in galaxies and starbursts .
injection , escape , and cooling losses all compete at each energy to determine the final cr spectrum ( see ltq for details ) .
the relevant scale height is the height of the volume in which crs are confined and produce synchrotron radiation .
normal galaxies ( @xmath43 or @xmath44 ) have large radio halos with @xmath34 , which we adopt as the cr scale height , even though the gas disk is much thinner .
compact starbursts are much smaller , with @xmath45 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; we use this as their scale height because they are probably good enough calorimeters to prevent most of the cr electrons / positrons from escaping out into the galactic haloes at high enough surface densities ( @xmath46 or @xmath36 ) . the cr disk scale height for our prototypical puffy starbursts , the smgs ,
is not yet directly measured , but it must be at least the gas scale height , if the star formation is distributed throughout the gas .
the gas scale height for smgs is kinematically inferred to be @xmath34 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which we adopt as the cr scale height of puffy starbursts .
our models span the entire observed range in @xmath16 for galaxies and starbursts , from @xmath47 to @xmath48 . combined with the scale height , @xmath16 determines the rate of injection , escape , and cooling at all energies .
the total rate of injection per unit volume is proportional to the volumetric star - formation rate , @xmath49 .
the observed schmidt law directly connects @xmath16 and @xmath8 @xcite .
we consider both @xmath10 ( * ? ? ?
* hereafter k98 ) and @xmath11 ( * ? ? ?
* hereafter b07 ) , using the normalizations given by k98 and b07 respectively .
the b07 relation was explicitly derived for high-@xmath1 galaxies , including smgs . protons and electrons
are injected into our model galaxies with an energy spectrum @xmath50 ; in this paper , we use @xmath51 in almost all cases .. the slightly different @xmath52 makes small quantitative differences , but does not change our conclusions . ]
escape , synchrotron , inverse compton , bremsstrahlung , and ionization losses are included for electrons and positrons , along with escape , ionization , and pion losses for protons .
diffusive escape times are normalized to the milky way value , with a loss time of @xmath53 .
we also consider variants with advective escape , as in starburst super - winds .
we assume that cosmic rays travel through gas with density @xmath54 , where @xmath55 is the mean ism number density and @xmath56 .
the magnetic field is parametrized as @xmath57 or @xmath58 , normalized by the milky way magnetic field strength .
we searched for models that satisfy the local @xmath2 frc for normal galaxies and compact starbursts .
the cr proton spectrum is then normalized by milky way cr proton constraints .
the parameters chosen for each variant are used to predict the frc for puffy starbursts and at high @xmath1 . in this paper , we consider several variants to get a sense of how the frc varies with redshift : * the ltq standard model , with no winds , @xmath57 , and the k98 star - formation relation . * the ltq model with winds of @xmath59 in starbursts ( @xmath35 ) , @xmath58 , and the k98 star - formation relation . * a variant with no winds , @xmath57 , and the b07 star - formation relation . * a variant with winds of @xmath59 in starbursts ( @xmath35 ) , @xmath57 , and the b07 star - formation relation . * a variant with winds of @xmath59 in starbursts ( @xmath35 ) , @xmath58 , and the b07 star - formation relation .
we show in table [ table : models ] that we can reproduce an acceptably linear local frc for each variant .
we are able to adopt parameters consistent with milky way - derived cr proton constraints such that @xmath60 varies by a factor of 1.7 - 2.2 from normal galaxies to compact starbursts .
this variation is consistent with the factor of @xmath19 scatter in the frc and winds , in which the winds invariably make electron escape too efficient when @xmath61 ( @xmath62 with b07 ) , suppressing its radio luminosity ( see the detailed discussion in ltq , appendix a.2 ) .
we are unable to find any parameters where the variation in @xmath63 is less than 2.0 , but we consider a variation of 2.2 adequate for this paper s purposes .
[ ftnt : windvariant ] ] ( e.g. , * ? ? ?
* ) . as in ltq ,
the need for a linear local frc strongly constrains the magnetic field in galaxies to scale as either @xmath64 or @xmath65 ( see [ sec : theory ] ) .
the dependence of the fir emission at a given wavelength is beyond the scope of this paper , since it depends on the exact sed of the fir emission in the galaxy .
instead , we simply use the total infrared ( tir ) emission , assuming it is all of the uv light reprocessed by the dust , and use @xmath66 as a proxy for @xmath67 .
@xcite estimate that @xmath68 , a correction we apply to the local value of @xmath69 from @xcite to find the normalization of @xmath63 at @xmath2 ( see ltq ) .
observable defined in @xcite , we then divide our calculated @xmath67 by @xmath70 to get the true fir luminosity .
however , we do not account for different @xmath70 values for normal galaxies or smgs . ]
we assume that the total tir emissivity ( defined here as luminosity per volume ) has been measured and corrected for redshift to its rest - frame value , @xmath71 .
we can also calculate the observable quantity @xmath72 as @xmath73 @xcite . for the radio emission ,
we calculate both the rest - frame radio emissivity @xmath74 where @xmath75 , and the rest - frame radio emissivity as estimated by an observer using a k - correction .
note that we are _ not _ actually trying to calculate the specific flux in the observer frame , but instead what an observer will infer for the _ rest - frame _ flux after using a k - correction galaxy at 2 ghz and trying to infer the 1 ghz flux by using the spectral slope .
the observer has an observed radio flux at 1.4 ghz , and can easily calculate the rest - frame flux at 1.4 ( 1 + z ) ghz flux , but wants the rest - frame flux at 1.4 ghz . ] : the rest - frame flux is what matters when we are considering the true , intrinsic evolution of the frc .
hence , although the specific flux in the observer - frame will have a bandwidth compression factor , we assume the observer will take it back out to get the inferred rest - frame specific flux . to calculate the observer - inferred rest - frame emissivity at rest - frame frequency @xmath75 , we start with @xmath76 , where @xmath77 and @xmath78 .
since the synchrotron spectra are typically expected to fall off as @xmath79 from observations of local star - forming galaxies , the radio luminosity can be k - corrected by multiplying by @xmath80 .
we therefore calculate @xmath81 , which we will refer to as the `` inferred '' radio emissivity .
the ratio of the tir and radio luminosities @xmath82 is then @xmath83 in the rest frame , and would be inferred to be @xmath84 in the rest - frame by observers .
[ sec : discussion ] we show the rest - frame frc for our model with @xmath85 , the b07 star - formation law , and no winds , in figure [ fig : lfirradiorest ] ( _ left panel _ ) as an example .
the solid dark red line is the @xmath86 frc .
all solid lines assume that starbursts are compact , with @xmath38 .
it is clear from figure [ fig : lfirradiorest ] that compact starbursts show little evolution in the frc , while low surface density galaxies have lower radio luminosities at high redshift .
this behavior is robust in all of the variants .
the cause of evolution in the frc is inverse compton losses off the cmb for cr electrons and positrons .
figure [ fig : frcevolution ] ( _ left panel _ ) shows that the frc should display relatively little evolution out to @xmath87 , except for the lowest surface brightness galaxies . however , normal galaxies have suppressed synchrotron radio emission at @xmath3 , a factor of @xmath19 for @xmath88 ( @xmath89 ) and of order 10 for @xmath90 ( @xmath91 ) .
the radio luminosities continue to fall with redshift . at @xmath92 , ic off
the cmb starts to matter even for the weaker starbursts ( @xmath61 ; @xmath93 ) .
dense starbursts ( @xmath94 ; @xmath95 for the k98 law or @xmath96 for the b07 law ) remain on a linear frc even at @xmath97 .
the strong cooling from bremsstrahlung , ionization , and ic off starlight implied by the high-@xmath16 conspiracy ( [ sec : theory ] ) acts as a buffer against ic losses off the cmb .
similarly , diffusive escape provides a similar buffering effect in low-@xmath16 galaxies , so that the increased ic losses off the cmb does not suppress the radio emission quickly .
in other words , most of the power from ic scattering of cmb photons does not come at the expense of synchrotron , but of other losses .
the radio dimming can not be described by a simple competition between magnetic field energy density and the cmb energy density ; the cmb energy density must also compete with _ every other loss process_. the buffering actually serves as an important test for both conspiracies described in
[ sec : theory ] . in high-@xmath16 galaxies and starbursts
, we predict that bremsstrahlung , ionization , and inverse compton of starlight already take a large portion of a electron s energy budget ; hence , the high-@xmath16 conspiracy works to hold @xmath63 constant out to quite high redshift .
the buffering essentially doubles the redshift that compact starbursts remain on the frc ; the weakest starbursts ( @xmath61 ; @xmath93 ) are radio - dim by a factor of @xmath98 at @xmath99 with buffering , instead of @xmath100 without the buffering . in low-@xmath16 galaxies , electrons and positrons
can easily escape before they lose energy to inverse compton off the cmb at low enough @xmath1 ; thus , the low-@xmath16 conspiracy also works to hold @xmath63 constant .
the suppression due to ic losses off the cmb seen in figure [ fig : lfirradiorest ] can be estimated by examining the ratios of the synchrotron cooling time to the total loss time , including both escape and cooling losses . in appendix
[ sec : cmbradiodim ] , we derive the ratios of loss times and we show that for a given schmidt law , a critical redshift @xmath101 can be defined for each @xmath16 and @xmath8 at which radio emission is suppressed .
we define @xmath101 to be the redshift when the rest - frame radio luminosity at @xmath75 is suppressed by a factor of 3 compared to @xmath86 . for our model with the b07 star - formation law , no winds , and @xmath85 , we find the critical redshifts are approximated as @xmath102 where @xmath8 is in units of @xmath42 .
we refer the reader to appendix [ sec : cmbradiodim ] , where we present similar relations for our other models of the frc .
we note that the radio suppression could , conceivably , be used to measure the temperature of the cmb at high redshift . in principle
, this method could apply to any galaxy with a radio and fir detection .
however , the conspiracies would have to be accounted for , and they are affected significantly by both the gas surface density @xmath16 and scale height @xmath39 ( see [ sec : puffy ] ) .
any measurement of the cmb temperature would depend on assumptions of the galaxy properties and would be model - dependent .
as figure [ fig : frcevolution ] ( _ right panel _ ) shows , the inferred rest - frame values of @xmath103 show additional apparent evolution , simply because the spectral slopes of the galaxies are not exactly @xmath104 .
compact starbursts have flatter spectra and @xmath105 are instantaneous spectral slopes , not the measured spectral slopes between two observed frequencies , unless otherwise noted .
[ ftnt : alphanotation ] ] with @xmath106 at 1.4 ghz , so by adopting @xmath107 they _ appear _ to become slightly radio brighter until @xmath87 , after which at higher frequencies their spectra steepen , and their radio emission appears to dim again at higher @xmath1 .
normal galaxies have steeper spectra with @xmath108 at 1.4 ghz , so their apparent radio luminosity sinks below their true radio luminosity at high redshift . at higher redshifts , our models predict that galaxies have intrinsically steeper radio spectra , because of increased ic losses from the cmb . the rest - frame spectral slope @xmath109 of normal galaxies at 1.4 ghz and asymptotes at @xmath110 by @xmath111 , when losses are dominated by ic for our injection spectrum of crs with @xmath51 .
there is much less intrinsic rest - frame evolution of starburst radio spectra ; even at @xmath92 , @xmath109 increases only by 0.1 for the weakest compact starbursts .
the observable @xmath112 shows much more pronounced evolution with redshift for starbursts , increasing by @xmath113 to @xmath6 , and @xmath114 to @xmath115 .
this effect arises simply because we predict the cr electron and positron spectra steepen with rest - frame frequency as synchrotron and ic losses become stronger ( e.g. , * ? ? ?
* ltq ) : at higher redshift and fixed observing frequency , we are seeing higher energy electrons and positrons . an important effect that we do not consider is the thermal free - free radio emission .
this will set a minimum total observed radio emission that is directly proportional to the star - formation luminosity .
thus , the true total radio deficit at ghz will not be as big as the synchrotron - only deficits plotted in figures [ fig : lfirradiorest ] and [ fig : frcevolution ] .
thermal free - free emission will also flatten the spectrum , especially at high frequencies . however , free - free emission is much fainter than the synchrotron luminosity at ghz frequencies , except in the faintest star - forming galaxies @xcite . [
sec : puffy ] as shown in figure [ fig : lfirradiorest ] , puffy starbursts fall on a linear fir - radio correlation of their own ( dotted lines ) , in line with observations of smgs @xcite .
the radio luminosity of these galaxies is nonetheless the result of a conspiracy , between ic losses on starlight , which decrease the radio luminosity , and the enhanced radio emission from secondary electrons and positrons , and the @xmath20 effect ( [ sec : theory ] ) .
the variation in @xmath63 for puffy starbursts alone is usually less than a factor of @xmath116 over the range @xmath117 ( see table [ table : models ] ; in most variants the variation is @xmath118 ) . like compact starbursts ,
escape plays essentially no role in most of the models , except that winds can slightly decrease the radio emission in relatively tenuous @xmath61 ( @xmath93 ) starbursts ( footnote [ ftnt : windvariant ] ) .
it is plain from figure [ fig : lfirradiorest ] ( _ left panel _ ) that the normalization of the puffy starburst frc ( dotted lines ) is different than the frc of the compact starbursts and normal galaxies ( solid lines ) when @xmath85 .
we show in table [ table : models ] that models with @xmath119 have radio - bright puffy starbursts compared to the observed local frc , by a factor of @xmath120 at @xmath86 . like the compact starbursts
, puffy starbursts show little rest - frame evolution in @xmath63 , except at relatively low surface densities ( @xmath61 ; @xmath93 ) , where they become radio dim at high redshifts because of ic losses on cmb photons .
therefore , we predict that puffy starbursts , which are mainly observed at high @xmath1 , have _ intrinsically _ different radio properties not caused by their redshift .
we propose a natural explanation of this radio excess in the framework of ltq ( [ sec : theory ] ) . in dense starbursts , protons are efficiently converted into secondary electrons and positrons through inelastic proton - proton scattering , which contribute to the synchrotron emission .
furthermore , the @xmath20 effect increases @xmath63 for starbursts which have larger @xmath12 .
compact starbursts lying on the @xmath86 frc balance these effects with increased bremsstrahlung , ionization , and ic losses , which compete with the synchrotron losses and suppress the excess radio luminosity ( [ sec : theory ] ) . in puffy starbursts with relatively low volume densities compared to their compact cousins at fixed @xmath16 ,
however , bremsstrahlung and ionization are not strong enough to compensate for these effects .
only synchrotron and ic losses remain , upsetting the conspiracy .
the radio excess predicted by this picture is systematically greater when using the k98 star - formation relation , because the ic loss rate on starlight is smaller at fixed surface density by a factor of 2.5 to 11 from weak starbursts to the densest starbursts .
this freedom to vary the radio - excess with @xmath12 does not exist in the standard calorimeter model , where all cr electron / positron energy goes into synchrotron emission , and the radio emission saturates .
the balance between synchrotron and the other forms of cooling can be changed in starbursts to alter the normalization of the frc , even though escape is negligible .
there is a reason to expect our allowed @xmath119 specifically : radiation pressure may drive turbulence and enhance the magnetic field until its energy density is comparable to radiation ( e.g. , * ? ? ?
if the k98 relation holds , then since the magnetic energy density scales as @xmath121 , @xmath85 , and if the b08 relation holds , then @xmath122 .
this explanation is more problematic for the b07 relation and @xmath85 . to the _
total _ milky way magnetic field strength of @xmath123 near the solar circle . if most of the magnetic field strength in starbursts is driven by turbulence , perhaps scaling to the disordered galactic magnetic field strength ( @xmath124 ; * ? ? ?
* ) near the solar circle would make more sense , because the ordered magnetic field arises from a different process .
the lower normalization for @xmath12 then would partly compensate for the steeper dependence on @xmath16 .
] however , while we assume that there is a parametrization for @xmath12 that applied to both compact starbursts and puffy starbursts , the radio excess should arise more generally .
the radio excess arises simply because synchrotron cooling time is shorter than the bremsstrahlung and ionization cooling times in puffy starbursts , but longer in compact starbursts at fixed @xmath16 : at fixed @xmath125 , @xmath126 and @xmath127 .
we therefore expect there to be a radio excess with respect to the frc in any puffy starburst with a strong enough magnetic field , with the exact enhancement depending on magnetic field strength because of the remaining competition , after ionization and bremsstrahlung are sub - dominant , from the ic losses on starlight .
, however , there will be more scatter in @xmath103 for puffy starbursts at fixed @xmath39 and @xmath1 .
smgs would not form their own frc if @xmath12 was the same for all smgs regardless of @xmath16 , or if @xmath12 increased _ very _ steeply with @xmath16 , because there still must be a conspiracy with ic losses off starlight . since smgs
do appear to form their own frc ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , this could be evidence specifically that their magnetic field strengths increase with @xmath16 ( or @xmath17 ) . ]
our results imply that a moderate radio excess at the factor of @xmath5 level alone is not a safe indicator of the presence of a radio - loud agn , especially at high redshifts where smgs are observed . while radio excess with respect to the local frc has been suggested as a selection criterion for radio - loud agns ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , our models with @xmath128 imply that @xmath129 for puffy starbursts powered by star - formation alone .
a radio excess is inexplicable in our models only when the source is an order of magnitude brighter ( @xmath130 ) in the radio than predicted from the fir emission .
while smgs are relatively rare and may not be a problem in small samples , we recommend that other means be used to be sure that the radio - excess is caused by an agn , such as a flat radio spectrum , radio morphology , mid - ir colors , or the presence of strong x - ray emission ( see also * ? ? ?
a magnetic field dependence of @xmath131 appears to hold for galactic molecular clouds ( e.g. , * ? ? ?
* ) , and ltq found that the fir - radio correlation was consistent with this magnetic field dependence .
the existence of galaxies with different scale heights allows us to distinguish the two possibilities for magnetic field scaling .
if @xmath57 , then the magnetic field strength will be the same for all galaxies with the same @xmath16 , regardless of scale height . in models with @xmath14 , by contrast , the magnetic field strength is weaker in puffy starbursts than in compact starbursts with the same @xmath16 .
as seen in figure [ fig : lfirradiorest ] ( _ right panel _ , dotted lines ) , puffy starbursts again form their own frc . in models with @xmath14 , they are radio dim compared to the @xmath2 frc .
we show in table [ table : models ] that the normalization of the frc is radio - dim by a factor of @xmath132 .
we can explain this in the ltq theory of the frc as well .
the magnetic field strength must increase more slowly with @xmath17 than with @xmath16 to reproduce the @xmath2 frc , because @xmath133 is 10 times smaller in compact starbursts than normal galaxies and puffy starbursts .
compact starbursts are highly compressed with respect to normal galaxies , so they have strong magnetic fields and synchrotron radio emission is strong enough to compete with the other losses .
puffy starbursts are not compressed , so that their magnetic fields are weak and synchrotron losses can not keep up with ic losses , nor with bremsstrahlung and ionization as 1.4 ghz emission traces ever lower electron energies at higher magnetic field strengths .
puffy starbursts therefore turn out to be radio dim compared to compact starbursts on the @xmath2 frc , if @xmath14 . as before ,
the b07 star - formation relation predicts greater ic losses and therefore weaker radio emission . because of the claims in the literature that smgs are radio bright
, we do not favor these models .
the suggested relative radio brightness of high-@xmath1 smgs therefore provides some evidence that , in fact , @xmath134 rather than @xmath14 .
however , the matter of whether high-@xmath1 smgs are in fact radio bright is not yet settled .
although ltq concluded that @xmath12 must increase dramatically from normal galaxies to dense starbursts ( [ sec : theory ] ) , they were unable to distinguish between these two possibilities with the @xmath2 frc alone .
for this reason , high-@xmath1 starbursts and their qualitatively different morphologies compared to those at @xmath2 can distinguish theories of the frc .
a prediction of all of our variants is that puffy starbursts like submillimeter galaxies should have steep non - thermal radio spectra , with @xmath135 ( see table [ table : models ] ) .
the steep spectra are caused by strong synchrotron cooling in the @xmath119 case and the relatively stronger ic cooling off starlight in the @xmath14 case .
in general , puffy starbursts should have roughly the same @xmath136 as normal galaxies in the local universe , which tends to be somewhat higher ( @xmath137 ) than in compact starbursts ( @xmath138 )
. the slope should hold even out to extremely high @xmath16 , as long as starbursts are puffy .
in contrast , we find that @xmath139 in compact starbursts , because of efficient ionization and bremsstrahlung losses , which flatten the equilibrium cr spectrum because of their energy dependence .
as we note in ltq , our predicted spectral index for normal galaxies is somewhat too high , and this difference in @xmath136 may carry over to the puffy starbursts .
however , the significant difference in @xmath136 between compact and puffy starbursts should remain as a general prediction of our model : compact starbursts should have flatter spectra than puffy starbursts .
the high spectral slopes can be observed either with direct measurements of multifrequency data of individual submillimeter galaxies , or with single frequency observations at a variety of redshifts .
there are relatively few measurements of @xmath136 for submillimeter galaxies specifically ; faint radio sources have @xmath140 @xcite , though that sample includes both compact starbursts and agns .
@xcite do find that @xmath141 for smgs , comparable to our predictions .
they also find that submillimeter galaxies have a radio - excess , in agreement with figure [ fig : lfirradiorest ] .
more recently , @xcite found an average @xmath142 , which is somewhat flatter than our models .
these spectral slopes are not different from normal star - forming galaxies , but are noticeably steeper than local ulirgs @xcite .
however , we do not account for free - free absorption , which probably flattens the spectra of local ulirgs like arp 220 at low frequency @xcite , and is not well understood in smgs .
since puffy starbursts have steeper spectra than compact starbursts , we expect their inferred @xmath63 will increase with redshift : if the true radio spectral slopes of smgs are greater than the assumed @xmath136 by @xmath143 , they will appear to become radio dimmer by a factor @xmath144 , or up to @xmath145 at @xmath146 . in figure
[ fig : starburstspectra ] , we show the expected radio synchrotron spectra of starburst galaxies , without correcting for thermal absorption or thermal emission . at a rest - frame frequency of 1 ghz , puffy starbursts ( dashed ) have steeper radio spectra than compact starbursts ( solid ) .
note that at high frequencies ( @xmath147 ) , the ratio of the radio luminosities per unit star formation of the compact and puffy starbursts asymptotes to a value set by the ratio of @xmath148 and @xmath149 in these starbursts . at these high frequencies , only synchrotron and ic cooling are effective , and ic cooling would be the same for puffy and compact starbursts because of the schmidt law ( [ sec : theory ] ) .
for @xmath57 , @xmath148 is the same for puffy and compact starbursts , but for @xmath58 , puffy starbursts have much smaller @xmath148 .
thus , measurements of the synchrotron radio emission of smgs at high @xmath125 could determine the magnetic field strength of smgs and determine which scenario applies .
of course , there are unlikely to be two perfectly distinct populations of compact starbursts and puffy starbursts . instead , there may be a continuum variation in scale heights from tens to thousands of parsecs .
we would then expect to see a larger scatter , both in @xmath63 and @xmath136 , in a full sample of both the most compact and the most puffy starbursts .
@xcite find that submillimeter galaxies do have a larger scatter in @xmath150 than other galaxies .
however , @xcite find a relatively small scatter of @xmath151 in smg radio spectral index .
importantly , for larger @xmath39 , both @xmath63 and @xmath109 asymptote as cr electron and positron losses are entirely determined by synchrotron and ic ; @xmath34 starbursts are already near this limit . in other words , for arbitrarily large @xmath39 , the radio excess with respect to the @xmath2 frc asymptotes to a value of @xmath152 , depending on the assumed schmidt law , and the radio spectral slope asymptotes to @xmath110 for @xmath51 . relying on the frc ,
a number of studies have used the ghz radio emission as a tracer of star - formation ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
radio emission has the advantage that it is unaffected by dust obscuration , making it potentially very useful in starbursts , and therefore for most of the star - formation at @xmath153 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our models let us evaluate the theoretical basis for radio as a sfr indicator at all redshifts .
we show the predicted radio emissivity as a function of star formation in figure [ fig : sfrradio ] . at low surface densities ( @xmath154 ; @xmath25 )
, synchrotron radio has a non - linear dependence on star - formation at all redshifts .
the weak radio emission is caused by electrons escaping their host galaxies , as inferred by @xcite and discussed in ltq .
that is , normal galaxies are not perfect electron calorimeters , so the radio emission is not a reliable star - formation tracer at low @xmath8 . at higher redshift ,
synchrotron emission is diminished by inverse compton off the cmb . for a milky way - like galaxy , we find that radio is a good star - formation tracer at @xmath155 , but underestimates it significantly by @xmath156 ( eq . [ eqn : b07zcrit ] ) .
already galaxies with star formation rates similar to galactic levels are beginning to be observed in the radio at high redshift @xcite , so that the ic suppression may soon be observed .
however , ic losses off the cmb should not be important even in the weakest starbursts until @xmath157 , as is also visible by the redshift evolution of the frc in figure [ fig : frcevolution ]
. radio emission does grow linearly with star - formation rate between normal galaxies with @xmath88 ( at @xmath158 ) and compact starbursts .
therefore , it serves as an acceptable star - formation indicator for these galaxies .
however , if @xmath119 , puffy starbursts like smgs have about @xmath120 times the radio emission at any given star - formation rate than compact starbursts .
therefore , we expect that if @xmath119 , the usual radio emission estimate based on the @xmath2 frc will _ overestimate _ their star formation rates by a factor of @xmath159 .
the excess is greatest at @xmath160 corresponding to @xmath161 , typical of observed smgs . among the puffy starbursts themselves
, the radio emission grows linearly with star - formation rate .
if instead @xmath14 , the radio emission will underestimate the star - formation rate .
however , because the smgs lie on their own frc , there is little real redshift evolution in the radio emissivity of puffy starbursts , because of the buffering provided by ic losses off starlight ( see [ sec : frcevolution ] , appendix [ sec : cmbradiodim ] ) . assuming a spectral slope @xmath107 will also underestimate the radio emissivity , since these galaxies can have steep radio spectra .
this explains the apparent evolution with @xmath1 of puffy starbursts in figure [ fig : sfrradio ] : we have applied a k - correction using a typically employed @xmath107 when , in fact , the true synchrotron spectra are steeper .
if the radio star - formation tracer could be calibrated to the special conditions in puffy starbursts like smgs , taking into account their different scale height , radio emissivity , and spectral slopes , we predict that radio would be a more accurate star - formation tracer for them .
we can also directly calculate the _ observed _ ghz radio flux density @xmath162 from synchrotron emission of star - forming galaxies .
we show the predicted flux density per unit star - formation at observer - frame 1.4 ghz in figure [ fig : ghzradiofluxes ] .
surveys with the expanded very large array ( evla ) will have a continuum sensitivity of @xmath163 at frequencies of 1 - 50 ghz , and it should be able to directly detect galaxies with @xmath164 and star - formation rates of @xmath165 out past @xmath166 . as stated in @xcite , a starburst like m82 with sfr @xmath167 will become undetectable past @xmath87 .
however , the buffering effect we emphasize in this paper preserves the radio emission of dense starbursts at high @xmath1 , so that bright starbursts will be detectable further : starbursts with sfr @xmath168 will be detectable with evla at 1.4 ghz out to @xmath169 and the most intense starbursts ( sfr @xmath170 and @xmath171 ) will be detectable out past @xmath97 in synchrotron emission .
@xcite predicts the square kilometer array ( ska ) will be sensitive to star - forming galaxies with flux densities of @xmath172 .
if this sensitivity is attained , then milky way - like galaxies ( @xmath173 ; sfr @xmath174 ) will be directly detectable at 1.4 ghz in synchrotron emission out to @xmath3 , and starbursts with sfr @xmath174 will be detectable at 1 ghz beyond @xmath175 . even at @xmath97 , the synchrotron emission of dense , compact starbursts ( @xmath176 ) with sfr greater than @xmath177 should be detectable at 1.4 ghz with ska ( models with @xmath178 have radio - dim puffy starbursts , and these are detectable at @xmath99 with the ska only for sfr greater than @xmath179 ) .
by contrast , @xcite found a sensitivity of @xmath180 , based on the free - free emission ; this limit will apply to normal galaxies and weak starbursts where the synchrotron emission is suppressed .
these sensitivities assume that natural confusion , in which radio sources overlap , will not hamper the ska ; estimates for the natural confusion limit vary from njy to @xmath181 levels ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
ska and evla will also have good spectral coverage , which may help measurements of the spectral index .
if a galaxy is detected at @xmath182 ( @xmath183 for evla and @xmath184 for ska ; @xcite ) at two different frequencies @xmath185 and @xmath186 with @xmath187 , then @xmath188 can be constrained to @xmath189 at the @xmath190 level .
evla will be better at high observer - frame frequencies ( 1 - 50 ghz )
. a problem for the evla will be the increasing fraction of thermal emission , which is expected to dominate the emission of starbursts at @xmath191 .
evla will therefore not easily measure the nonthermal spectral indices of galaxies at high z. on the other hand , the ska will face free - free absorption when observing low - z starbursts ; for example , arp 220 may be optically thick even at 1 ghz @xcite . at high redshift , however , the rest - frame frequencies ska will observe will be less affected by free - free absorption .
star - formation in galaxies over cosmic time produces a diffuse radio synchrotron background .
our models indicate that submillimeter galaxies and other puffy starbursts ought to have enhanced radio emission . in principle
, this means that the synchrotron radio background could be up to @xmath159 times higher than usually predicted from the cosmic infrared background ( cib ) and a naive application of the @xmath2 frc .
the magnitude of this implied radio excess is interesting , because arcade2 recently reported an excess radio background at 3 , about five times higher than expected from star formation @xcite .
@xcite found that constraints on inverse compton emission require the excess to come from regions with galactic - level ( @xmath192 ) magnetic fields , and suggest an evolution of the frc as the source of the reported excess .
however , while smgs are individually very bright and contribute much to the cosmic star - formation rate at @xmath193 , they are not typical starbursts @xcite .
instead , they seem to represent a transient phase that can survive about 100 before their gas is depleted @xcite .
we can estimate the total radio background enhancement by scaling to the contribution of smgs to the cib , which is largely reprocessed starlight from galaxies at @xmath153 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and adjusting by the smg radio excess .
submillimeter galaxies do provide the majority of light at @xmath194 , but this is only a small fraction of the total ir background . at the peak of the cib ( @xmath195 ) ,
submillimeter galaxies provide @xmath196 of the total power , and possibly only @xmath197 @xcite .
optimistically , the radio excess from smgs would be @xmath198 .
this is significant , but not enough to explain the very large arcade2 excess .
more conservatively , the excess is more likely @xmath199 , and could be as little as a few percent .
since the number of radio sources down to several @xmath181 is well known , we can estimate the fraction of the expected radio background comes from bright smgs .
@xcite find a tir background of about @xmath200 ; from the normalization of the fir - radio correlation we expect a 1.4 ghz background of @xmath201 .
@xcite found an average radio flux density of @xmath202 for bright smgs ( @xmath203 ) .
the number counts of @xmath204 sources imply that they have a density of @xmath205 , contributing roughly @xmath206 to the 1.4 ghz radio background ( e.g. , * ? ? ?
bright smgs have an approximate density of @xmath207 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , so they constitute about @xmath145 of the background from @xmath204 sources , or @xmath208 of the expected 1.4 ghz background from star - formation .
this is roughly in line with our estimate of @xmath209 , although the uncertainties are large enough that it could be consistent with smgs lying on the frc . in any case , it is fairly clear that bright smgs are not the source of the arcade excess
. the total excess could be greater if most starbursts at @xmath153 are puffy , and not just smgs .
we do not expect this simply because most current studies show that the local frc _ does _ hold out to high redshift for most observed galaxies and starbursts ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the spectral slope would also be a problem : arcade2 inferred @xmath210 , while we predict a spectral slope @xmath211 , steeper than local compact starbursts .
resolved radio sources in the range @xmath212 , which are expected to be star - forming galaxies @xcite , contribute about @xmath213 to the 1.4 ghz radio background @xcite , which is consistent with the total energetics expected from the frc .
@xcite also find consistency between the frc and the observed radio source counts at @xmath214 . stacking studies of fainter radio sources
have given somewhat ambiguous results on whether the frc applies @xcite , but @xcite find no large evolution in the frc down to @xmath215 .
any large radio excess would have to come either from a new population of low luminosity galaxies or very high-@xmath1 galaxies ( see figure [ fig : ghzradiofluxes ] ) ; extrapolations of the higher flux source populations do not predict a large radio excess ( e.g. * ? ? ? * ) .
nonetheless , our work indicates that a radio excess from smgs can be significant .
conversely , if the luminosity function of galaxies was very steep , most galaxies could be intrinsically radio - dim with respect to the frc , because of ic losses off the cmb ( see [ sec : frcevolution ] and figures [ fig : lfirradiorest ] and [ fig : frcevolution ] ) . then the frc would overestimate the strength of the radio background .
however , most star - formation at high @xmath1 is believed to have occurred in starbursts @xcite , so this possibility is unlikely .
we have applied the theory of ltq to predict the frc for redshifts @xmath0 .
we use one - zone models of galaxies and starbursts with cr injection , cooling , and escape to predict the equilibrium , steady - state radio spectra of galaxies and starbursts over the entire range of the frc .
our goals were to determine how and why the low- and high-@xmath16 conspiracies crucial to the @xmath2 frc ( [ sec : theory ] ) affect the frc at high redshift , and to provide a quantitative model for predicting the critical redshifts at which galaxies deviate from the @xmath2 frc .
we find the following : 1 . for compact starbursts ( @xmath45 ) , we find relatively little evolution in the fir - radio correlation out to @xmath216 ( figure [ fig : lfirradiorest ] ) .
this is partly because the magnetic energy density in galaxies is strong enough to dominate the cmb even at high redshifts .
however , the high-@xmath16 conspiracy ( [ sec : theory ] ) also acts as a buffer against ic losses off the cmb ; the increased ic losses must compete with the already present bremsstrahlung , ionization , and ic off starlight in addition to synchrotron losses .
the rest - frame radio spectral slope @xmath109 at fixed @xmath125 does not change with @xmath1 , but the observed @xmath136 at fixed @xmath217 increases because the non - thermal synchrotron radio spectrum steepens at higher rest - frame frequency .
2 . we derive in appendix [ sec : cmbradiodim ] the critical redshifts when inverse compton losses off the cmb suppress the radio luminosity of galaxies compared to the @xmath2 frc .
these relations are given for our standard model in equation [ eqn : b07zcrit ] .
the non - thermal radio luminosity is suppressed severely in milky way - like galaxies ( @xmath89 ) at @xmath3 and the weakest compact starbursts at @xmath218 .
the spectrum at ghz steepens to @xmath219 because of these enhanced ic losses .
nonetheless , the low-@xmath16 conspiracy ( [ sec : theory ] ) also acts to prevent the radio emission from steeply falling with redshift , since inverse compton losses off the cmb must be more efficient than diffusive escape , not just synchrotron losses ( see [ sec : frcevolution ] ) .
ltq found that the @xmath86 frc demands that @xmath12 scales with @xmath17 or @xmath16 in galaxies lying on the schmidt law . in models with @xmath119
, we find that puffy starbursts with @xmath37 such as smgs are radio bright compared to the @xmath86 frc by a factor of @xmath159 .
this follows from a breakdown of the high-@xmath16 conspiracy ( [ sec : theory ] ) : bremsstrahlung and ionization cooling are weak in puffy starbursts relative to the compact starbursts that predominate in the @xmath86 universe .
in contrast , in models with @xmath14 , we find that puffy starbursts are radio dim compared to the observed frc , because of weak synchrotron cooling relative to the ic losses .
since several studies have reported radio excesses for smgs , we favor the @xmath119 scaling ; however the issue of whether smgs are radio - bright is still not fully resolved . in either case ,
puffy starbursts show little true evolution with @xmath1 , though they may appear to have fainter rest - frame radio luminosities at high @xmath1 because of their steep spectra .
puffy starbursts inevitably have high @xmath136 ( @xmath220 ) , since bremsstrahlung and ionization losses are weak with respect to synchrotron and ic
. a key prediction of our scenario with @xmath119 is that the variations in @xmath63 will be correlated with scale height at fixed @xmath8 , since the radio - excess in our models is a direct consequence of the large cr scale height and the small bremsstrahlung and ionization losses it causes .
radio - excess ( low @xmath221 ) puffy starbursts will have steeper radio spectral slopes ( bigger @xmath136 ) , larger velocity dispersions @xmath222 compared to their rotation speeds @xmath223 , and possibly moderately cooler dust temperatures ( smaller @xmath224 ) .
4 . as previously expected , radio emission can be a poor tracer of star formation in low surface density galaxies , because of electron escape and ic losses off the cmb . for our preferred @xmath119
scaling , radio emission overestimates star - formation rate by a factor of @xmath120 in puffy starbursts .
star - formation rate is underestimated by synchrotron radio emission with the @xmath14 scaling .
while smgs may be individually radio bright compared to the local frc , they contribute a relatively small fraction of the cosmic infrared background and the total star - formation luminosity of the universe .
this means that they enhance the star - formation radio background by @xmath225 , and possibly around @xmath209 , with respect to a naive application of the @xmath86 frc . as in ltq
, we did not exactly match the observed radio spectral slopes of galaxies , with @xmath108 in normal galaxies and @xmath106 in compact starbursts .
this will have a slight effect on the k - correction .
an error of @xmath226 in @xmath136 should only affect inferred radio luminosity by 30% at @xmath146 and 60% at @xmath100 .
nevertheless , the prediction of steeper radio spectra in puffy starbursts with respect to compact starbursts at all relevant @xmath1 should be robust . our explanation for the small @xmath63 ratio in submillimeter galaxies as a breakdown of the high-@xmath16 conspiracy
is based purely on the steady - state spatially - averaged synchrotron emission , but the details of the fir emission may also matter . throughout this paper , we have simply assumed that the bolometric fir luminosity could be correctly inferred from observations , and have assumed the same uv opacity for all galaxies at all redshifts .
the total fir emission is also likely to depend on the metallicity , and may be lower at high @xmath1 for the lowest surface density galaxies .
the exact far - infrared sed is important in determining the fir emission when observations have only been made at only a few wavelengths .
the presence of agns , a different imf at high @xmath1 , and selection biases may also affect the inferred @xmath221 of smgs .
we did not include the effects of galaxy evolution on the cr spectrum in our models .
we argued in ltq that it should not matter for quiescent spirals or for extreme starbursts , because the cr lifetime is much shorter than the time dependence of stellar populations .
however , galaxy evolution may play a role in weaker starbursts @xcite and in post - starburst galaxies @xcite .
studies of merging normal galaxies and galaxies in clusters have indeed found that they are radio bright with respect to the frc , possibly because of compression of magnetic fields or shock acceleration @xcite .
we also assumed that the magnetic field strength at a given density does not depend on redshift .
it is not entirely clear how long normal galaxies take to build up their magnetic fields , or even what process is at work ( see the reviews in * ? ? ?
* ; * ? ? ?
* ) , though there are theoretical mechanisms that can rapidly generate strong magnetic fields .
studies at @xmath3 indicate that normal milky way - like galaxies had magnetic fields with similar strengths to the present @xcite . at the very highest redshifts ,
magnetic field strengths might be weaker , because the seed fields were essentially zero compared to the present strengths . starbursts also may build their magnetic fields up in much shorter times than normal galaxies , and through a different process than normal galaxies @xcite . finally , we have used one - zone models , which are appropriate if the crs sample all of the gas phases in each galaxy s ism .
however , the ism is known to be clumpy in the milky way , in compact starbursts like arp 220 ( e.g. , * ? ? ?
* ) , and even in the smgs themselves @xcite .
a full understanding of the frc will probably require models that take into account the inhomogeneity of star - forming galaxies .
we thank eric murphy and michal michaowski for critical readings of the text , eliot quataert for many stimulating conversations , and the galprop team for making their code and its subroutines freely available .
galprop is available at http://galprop.stanford.edu .
we would like to thank igor moskalenko for sharing his group s estimates of the milky way s total @xmath29-ray luminosity for normalization of our models of the frc .
t. a. t. is supported in part by an alfred p. sloan fellowship .
at high redshift , the nonthermal synchrotron radio luminosity of galaxies is suppressed by the cmb .
this is because the inverse compton process shortens the lifetime that crs have to radiate synchrotron ; equivalently , their energy goes into inverse compton photons instead of synchrotron radio . because of the shorter times the cr electrons and positrons have to radiate synchrotron , the redshift @xmath101 when the radio luminosity of a galaxy is quenched by a factor @xmath227 is given by @xmath228 the loss time includes all cooling and escape losses , so that @xmath229 in some regimes , some losses may be neglected .
for example , diffusive losses are unimportant in starbursts , as are ic losses off the cmb at @xmath86 .
advective losses may be unimportant in normal galaxies and also for the densest starbursts .
we can immediately see that the presence of losses besides synchrotron and ic off the cmb prevent the suppression of radio . in the case
when no losses other than synchrotron and ic off the cmb exist , equation [ eqn : radiosuppress ] reduces to @xmath230 the presence of other loss processes , whether escape or cooling , requires a higher @xmath101 for a fixed quench factor @xmath227 : @xmath231 essentially the ic losses off the cmb must compete with not only synchrotron losses , but every other cooling and escape process as well . with the high-@xmath16 conspiracy , @xmath232 can be of order @xmath233 , so the energy density of the cmb must be @xmath30 times greater to suppress the radio emission than would be naively expected .
we give the loss times for each process in ltq . from those lifetimes
, we can find the ratios of the synchrotron cooling timescale to the other loss timescales : @xmath234 where @xmath235 , @xmath236 , @xmath237 is the wind speed in units of 300 @xmath238 , and @xmath239 is the rest - frame frequency divided by 1.4 ghz .
our models parameterize @xmath12 in terms of the gas surface density @xmath16 : @xmath240 finally , the schmidt law allows us to convert between gas surface density @xmath16 and surface density of star - formation rate @xmath8 : @xmath241 we can approximately solve for the redshift @xmath101 when @xmath242 for galaxies and starbursts in each of the scenarios in table [ table : models ] . for the scenario with the b07 star - formation law , @xmath85 , and no winds , we find : @xmath243 where @xmath8 is in units of @xmath42 . for the scenario with the b07 star - formation law , @xmath244 , and winds in starbursts ,
we find : @xmath245 for the scenario with the b07 star - formation law , @xmath246 , and winds in starbursts , we find : @xmath247 for the standard model of ltq , with the k98 star - formation law , @xmath85 , and no winds , we find : @xmath248 finally , for the model of ltq with the k98 star - formation law , @xmath131 , and winds in starbursts , we find : @xmath249 condon , j. j. 2009 , sensitive continuum survey with the ska : goals and challenges ( manchester : ska program development office ) , ska memo # 114 ( http://www.skatelescope.org/pdf/memos/114_memo_condon.pdf ) lllccccc k98 & n & @xmath250 & 1.7 ( 1.5 ) & 3.9 ( -0.59 ) & 0.92 - 0.96 & 0.44 - 0.64 & 0.69 - 0.92 + k98 & y & @xmath251 & 1.7 ( 1.7 ) & 0.85 ( 0.07 ) & 0.88 - 0.94 & 0.43 - 0.61 & 0.78 - 0.85 + b07 & n & @xmath250 & 1.7 ( 1.5 ) & 2.2 ( -0.35 ) & 0.87 - 0.96 & 0.57 - 0.67 & 0.86 - 0.98 + b07 & y & @xmath252 & 2.2 ( 1.7 ) & 3.4 ( -0.53 ) & 0.85 - 0.91 & 0.45 - 0.59 & 0.75 - 0.87 + b07 & y & @xmath253 & 2.0 ( 2.6 ) & 0.49 ( 0.31 ) & 0.87 - 0.91 & 0.49 - 0.64 & 0.88 - 0.91 + [ table : models ] | we construct one - zone steady - state models of cosmic ray ( cr ) injection , cooling , and escape over the entire dynamic range of the fir - radio correlation ( frc ) , from normal galaxies to starbursts , over the redshift interval @xmath0 .
normal galaxies with low star - formation rates become radio - faint at high @xmath1 , because inverse compton ( ic ) losses off the cmb cool cr electrons and positrons rapidly , suppressing their nonthermal radio emission . however , we find that this effect occurs at higher redshifts than previously expected , because escape , bremsstrahlung , ionization , and starlight ic losses act to counter this effect and preserve the radio luminosity of galaxies .
the radio dimming of star - forming galaxies at high @xmath1 is not just a simple competition between magnetic field energy density and the cmb energy density ; the cmb must also compete with every other loss process .
we predict relations for the critical redshift when radio emission is significantly suppressed compared to the @xmath2 frc as a function of star - formation rate per unit area . for example
, a milky way - like spiral becomes radio - faint at @xmath3 , while an m82-like starburst does not become radio - faint until @xmath4 .
we show that the `` buffering '' effect of non - synchrotron losses improves the detectability of star - forming galaxies in synchrotron radio emission with evla and ska .
additionally , we provide a quantitative explanation for the relative radio brightness of some high-@xmath1 submillimeter galaxies .
we show that at fixed star formation rate surface density , galaxies with larger cr scale heights are radio bright with respect to the frc , because of weaker bremsstrahlung and ionization losses compared to compact starbursts .
we predict that these `` puffy starbursts '' should have steeper radio spectra than compact galaxies with the same star - formation rate surface density .
we find that radio bright submillimeter galaxies alone can not explain the excess radio emission reported by arcade2 , but they may significantly enhance the diffuse radio background with respect to a naive application of the @xmath2 frc . |
all graphs considered in this paper are finite , loopless , and without multiple edges .
a _ plane graph _ is a particular drawing of a planar graph in the euclidean plane . for a graph @xmath0
, we use @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 to denote its vertex set , edge set , order , size and minimum degree , respectively . for @xmath3 , @xmath23 denotes the set of neighbors of @xmath8 in @xmath0 .
if there is no confusion about the context , we write @xmath24 for @xmath23 . a @xmath25-coloring of @xmath0 is a mapping @xmath5 from @xmath18 to a color set @xmath26 such that @xmath27 for any adjacent vertices @xmath28 and @xmath29 . a graph is @xmath25-colorable if it has a @xmath25-coloring .
cowen , cowen , and woodall @xcite considered _ defective _ colorings of graphs .
a graph @xmath0 is said to be _
@xmath25-colorable _ , or simply , _
@xmath10-colorable _ , if the vertices of @xmath0 can be colored with @xmath25 colors in such a way that each vertex has at most @xmath7 neighbors receiving the same color as itself .
obviously , a @xmath30-coloring is an ordinary proper @xmath25-coloring .
a _ list assignment _ of @xmath0 is a function @xmath1 that assigns a list @xmath2 of colors to each vertex @xmath3 .
an @xmath1-coloring with impropriety of integer @xmath7 , or simply an _
@xmath4-coloring _ , of @xmath0 is a mapping @xmath5 that assigns a color @xmath6 to each vertex @xmath3 so that at most @xmath7 neighbors of @xmath8 receive color @xmath9 .
a graph is _
@xmath25-choosable _ with impropriety of integer @xmath7 , or simply _
@xmath10-choosable _ , if there exists an @xmath4-coloring for every list assignment @xmath1 with @xmath11 for all @xmath3 . clearly , a @xmath30-choosable is the ordinary @xmath25-choosability introduced by erds , rubin and taylor @xcite and independently by vizing @xcite .
the concept of list improper coloring was independently introduced by krekovski @xcite and eaton and hull @xcite .
they proved that every planar graph is @xmath31-choosable and every outerplanar graph is @xmath32-choosable .
these are both improvement of the results showed in @xcite which say that every planar graph is @xmath31-colorable and every outerplanar graph is @xmath32-colorable .
let @xmath33 denote the _ girth _ of a graph @xmath0 , i.e. , the length of a shortest cycle in @xmath0 .
the @xmath10-choosability of planar graph @xmath0 with given @xmath33 has been studied by krekovski in @xcite .
he proved that every planar graph @xmath0 is @xmath34-choosable if @xmath35 , @xmath32-choosable if @xmath36 , @xmath37-choosable if @xmath38 , and @xmath39-choosable if @xmath40 and @xmath41 .
recently , cushing and kierstead @xcite proved that every planar graph is @xmath42-choosable .
so it would be interesting to investigate the sufficient conditions of @xmath14-choosability of subfamilies of planar graphs where some families of cycles are forbidden .
krekovski proved in @xcite that every planar graph without @xmath17-cycles is @xmath14-choosable .
lih et al .
@xcite proved that planar graphs without @xmath12- and @xmath13-cycles are @xmath14-choosable , where @xmath15 .
later , dong and xu @xcite proved that planar graphs without @xmath12- and @xmath13-cycles are @xmath14-choosable , where @xmath16 .
moreover , xu and zhang @xcite asked the following question : [ q-1 ] is it true that every planar graph without adjacent triangles is @xmath14-choosable ?
recall that there is a planar graph containing @xmath12-cycles that is not @xmath14-colorable @xcite .
therefore , while describing @xmath14-choosability planar graphs , one must impose these or those restrictions on @xmath12-cycles .
note that in all previously known sufficient conditions for the @xmath14-choosability of planar graphs , the @xmath12-cycles are completely forbidden . in this paper
we allow @xmath12-cycles , but disallow them to have a common edge with relatively short cycles .
the purpose of this paper is to prove the following [ main01 ] every planar graph without @xmath12-cycles adjacent to @xmath17- and @xmath12-cycles is @xmath14-choosable .
clearly , theorem [ main01 ] implies corollary [ cor1 ] which is a common strengthening of the results in @xcite .
[ cor1 ] every planar graph without @xmath12-cycles is @xmath14-choosable
. moreover , theorem [ main01 ] partially answers question [ q-1 ] , since adjacent triangles can be regarded as a @xmath12-cycle adjacent to a @xmath17-cycle .
a vertex of degree @xmath25 ( resp .
at least @xmath25 , at most @xmath25 ) will be called a _ @xmath25-vertex _ ( resp . _ @xmath43-vertex _ , _ @xmath44-vertex _ ) .
a similar notation will be used for cycles and faces .
a _ triangle _ is synonymous with a 3-cycle . for @xmath45
, we use @xmath46 to denote the boundary walk of @xmath47 and write @xmath48 $ ] if @xmath49 are the boundary vertices of @xmath47 in cyclic order . for any @xmath3 , we let @xmath50 denote the neighbors of @xmath8 in a cyclic order .
let @xmath51 be the face with @xmath52 and @xmath53 as two boundary edges for @xmath54 , where indices are taken modulo @xmath55 .
moreover , we let @xmath56 denote the number of @xmath17-faces incident to @xmath8 and let @xmath57 denote the number of @xmath17-vertices adjacent to @xmath8 .
an @xmath58-face @xmath59 $ ] is called an @xmath60-_face _ if the degree of the vertex @xmath61 is @xmath62 for @xmath63 .
suppose @xmath8 is a @xmath12-vertex incident to a @xmath64-face @xmath47 and adjacent to two @xmath17-vertices not on @xmath46 .
if @xmath65 , then we call @xmath8 a _ light _ @xmath12-vertex . otherwise , we call @xmath8 a _ soft _ @xmath12-vertex if @xmath66 . a vertex @xmath8 is called an _ @xmath67-vertex _ if it is either a @xmath17-vertex or a light @xmath12-vertex . moreover
, we say a @xmath17-face @xmath68 $ ] is an @xmath69-face if @xmath70 for each @xmath71 and @xmath72 is an @xmath67-vertex .
suppose @xmath8 is a @xmath73-vertex incident to two 3-faces @xmath74 $ ] and @xmath75 $ ] .
let @xmath76 be the neighbour of @xmath8 not belonging to the @xmath17-faces .
if @xmath77 and @xmath78 is a @xmath79-face , then we call @xmath8 a _ bad _ @xmath73-vertex
. -vertex @xmath8 , a soft @xmath12-vertex @xmath80 and a bad @xmath73-vertex @xmath81.,title="fig:",width=384 ] + for all figures in the following section , a vertex is represented by a solid circle when all of its incident edges are drawn ; otherwise it is represented by a hollow circle .
moreover , we use a hollow square to denote an @xmath67-vertex .
the proof of theorem [ main01 ] is done by reducible configurations and discharging procedure .
suppose the theorem is not true .
let @xmath0 be a counterexample with the least number of vertices and edges embedded in the plane .
thus , @xmath0 is connected .
we will apply a discharging procedure to reach a contradiction .
we first define a weight function @xmath82 on the vertices and faces of @xmath0 by letting @xmath83 if @xmath3 and @xmath84 if @xmath45 .
it follows from euler s formula @xmath85 and the relation @xmath86 that the total sum of weights of the vertices and faces is equal to we then design appropriate discharging rules and redistribute weights accordingly .
once the discharging is finished , a new weight function @xmath88 is produced .
the total sum of weights is kept fixed when the discharging is in process . nevertheless , after the discharging is complete , the new weight function satisfies @xmath89 for all @xmath90 .
this leads to the following obvious contradiction , @xmath91 and hence demonstrates that no such counterexample can exist . in this section
, we will establish structural properties of @xmath0 .
more precisely , we prove that some configurations are reducible .
namely , they can not appear in @xmath0 because of the minimality of @xmath0 .
since @xmath0 does not contain a @xmath12-cycle adjacent to an @xmath92-cycle , where @xmath93 , by hypothesis , the following fact is easy to observe and will be frequently used throughout this paper without further notice . _ for @xmath96 , let @xmath97 $ ] denote the subgraph of @xmath0 induced by @xmath98 .
we simply write @xmath99 $ ] .
let @xmath1 be an arbitrary list assignment of @xmath0 , and @xmath5 be an @xmath100-coloring of @xmath101 . for each @xmath102
, let @xmath103 , and we call @xmath104 an _ induced assignment _ of @xmath97 $ ] from @xmath5 .
we also say that @xmath5 can be extended to @xmath0 if @xmath97 $ ] admits an @xmath105-coloring .
_ [ key-0 ] suppose that @xmath0 contains the configuration @xmath106 , depicted in figure [ fig3 ] .
let @xmath5 be an @xmath100-coloring of @xmath101 , where @xmath107 . denote by @xmath104 an induced list assignment of @xmath97 $ ] . if @xmath108 for each @xmath109 , then @xmath5 can be extended to the whole graph @xmath0 .
* proof . * since @xmath108 for each @xmath109 , we can color each @xmath61 with a color @xmath110 properly . note that @xmath111 .
if there exists a color in @xmath112 which appears at most once on the set @xmath113 , then we assign such a color to @xmath8 .
it is easy to check that the resulting coloring is an @xmath100-coloring and thus we are done .
otherwise , w.l.o.g .
, suppose @xmath114 , @xmath115 , and each color in @xmath116 appears exactly twice on the set @xmath117 . w.l.o.g .
, suppose @xmath118 .
( ii ) if @xmath119 is a light @xmath12-vertex , denote by @xmath124 the other two neighbors which are different from @xmath8 and @xmath72 .
erase the color of @xmath119 , color @xmath8 with @xmath122 , and recolor @xmath125 and @xmath126 with a color different from its neighbors .
we can do this since @xmath127 by definition .
next , we will show how to extend the resulting coloring , denoted by @xmath128 , to @xmath0 . if @xmath129 , then color @xmath119 with a color in @xmath130 .
otherwise , we color @xmath119 with a color in @xmath131 . in each case
, one can easily check that the obtained coloring of @xmath0 is an @xmath100-coloring . *
suppose that @xmath8 is adjacent to three @xmath17-vertices @xmath139 and @xmath140 .
denote @xmath141 . by the minimality of @xmath0
, @xmath142 admits an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
it is easy to deduce that @xmath144 and @xmath108 for each @xmath145 .
so for each @xmath61 , we assign the color @xmath110 to it .
now we observe that there exists a color in @xmath112 appearing at most once on the set @xmath146 .
we color @xmath8 with such a color .
the obtained coloring is an @xmath100-coloring of @xmath0 .
this contradicts the choice of @xmath0 . *
it suffices to prove that @xmath0 does not contain a @xmath147-face by ( a3 ) .
suppose @xmath68 $ ] is a @xmath17-face with @xmath148 .
for each @xmath145 , let @xmath149 denote the other two neighbors of @xmath61 not on @xmath46 . denote by @xmath142 the graph obtained from @xmath0 by deleting edge @xmath150 . by the minimality of @xmath0 , @xmath142 has an @xmath100-coloring @xmath5 . if @xmath151 , then @xmath0 itself is @xmath100-colorable and thus we are done
otherwise , suppose @xmath152 .
if @xmath5 is not an @xmath100-coloring of the whole graph @xmath0 , then without loss of generality , assume that @xmath153 and @xmath154 . moreover
, none of @xmath125 s neighbors except @xmath119 is colored with 1 .
first , we recolor each @xmath61 with a color @xmath155 in @xmath156 , where @xmath145
. we should point out that @xmath155 may be the same as @xmath157 , but it does not matter . note that if at most two of @xmath158 are equal then the resulting coloring is an @xmath100-coloring and thus we are done .
otherwise , suppose that @xmath159 . since @xmath160 and @xmath161 , we may further reassign color 1 to @xmath119 to obtain an @xmath162-coloring of @xmath0 .
this contradicts the choice of @xmath0 .
* suppose @xmath68 $ ] is a @xmath134-face incident to two light @xmath12-vertices @xmath72 and @xmath140 . by definition , we see that each @xmath61 ( @xmath163 ) is incident to two other @xmath17-vertices , denoted by @xmath164 and @xmath165 , which are not on @xmath46 .
let @xmath142 denote the graph obtained from @xmath0 by deleting edge @xmath166 .
obviously , @xmath142 has an @xmath100-coloring @xmath5 by the minimality of @xmath0 . similarly ,
if @xmath167 , then @xmath0 itself is @xmath100-colorable and thus we are done .
otherwise , suppose @xmath168 . if @xmath5 is not an @xmath100-coloring of @xmath0 , then w.l.o.g .
, assume that @xmath169 and @xmath118 .
erase the color of @xmath72 and recolor @xmath170 with a color @xmath171 different from its neighbors . if @xmath172 , then color @xmath72 with a color in @xmath173
otherwise , color @xmath72 with @xmath174 .
it is easy to verify that the resulting coloring is an @xmath100-coloring of @xmath0 , which is a contradiction .
* suppose that a @xmath73-vertex @xmath8 is incident to a @xmath79-face @xmath74 $ ] and adjacent to two @xmath17-vertices @xmath140 and @xmath175 .
let @xmath176 . by the minimality of @xmath0 , @xmath142 has an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
obviously , @xmath108 for each @xmath177 and @xmath111 . by lemma [ key-0 ]
, @xmath5 can be extended to @xmath0 , which is a contradiction .
* suppose that a @xmath135-vertex @xmath8 is incident to two @xmath178-faces @xmath179 and one @xmath137-face @xmath180 such that @xmath181 for each @xmath182 , @xmath183 and @xmath76 is an @xmath67-vertex .
namely , @xmath76 is either a @xmath17-vertex or a light @xmath12-vertex
. let @xmath184 .
by minimality , @xmath142 admits an @xmath100-coloring @xmath5 .
denote by @xmath104 an induced list assignment of @xmath143 .
it is easy to verify that @xmath108 for each @xmath185 and @xmath186 .
so we can color @xmath61 with @xmath110 for each @xmath187 .
if there exists a color @xmath188 appearing at most once on the set @xmath189 , then we further assign color @xmath174 to @xmath8 and thus obtain an @xmath100-coloring of @xmath0 .
otherwise , each color in @xmath112 appears exactly twice on the set @xmath189 .
since @xmath76 is an @xmath67-vertex , we can apply versions of arguments ( i ) and ( ii ) in the proof of lemma [ key-0 ] to obtain an @xmath100-coloring of @xmath0 .
@xmath132 * proof . * ( f1 ) suppose to the contrary that @xmath193 .
let @xmath194 . by the minimality of @xmath0 , @xmath142 admits an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
notice that @xmath195 , @xmath196 , @xmath197 and @xmath198 .
first , we color @xmath8 with @xmath188 and color @xmath29 with @xmath199 .
then color @xmath81 with @xmath200 and @xmath28 with @xmath201 .
one can easily check that the resulting coloring of @xmath0 is an @xmath100-coloring .
this contradicts the assumption of @xmath0 .
( f2 ) suppose to the contrary that @xmath28 is a soft @xmath12-vertex .
by definition , @xmath28 has other two neighbors whose degree are both 3 , say @xmath125 and @xmath202 .
observe that neither @xmath125 nor @xmath202 is on @xmath46 .
let @xmath203 .
obviously , @xmath142 admits an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
for each @xmath204 , we deduce that @xmath205 .
moreover , @xmath206 .
we first color @xmath80 with @xmath207 and color @xmath81 with a color in @xmath208 . if at least one of @xmath125 and @xmath202 has the same color as @xmath9 , we can color @xmath28 with a color different from that of @xmath8 and @xmath29 . otherwise , we can color @xmath28 with a color different from @xmath125 and @xmath29 .
therefore , we achieve an @xmath100-coloring of @xmath0 , which is a contradiction .
@xmath132 * proof .
* suppose to the contrary that @xmath81 and @xmath8 are adjacent soft @xmath12-vertices such that @xmath209 $ ] is a @xmath12-face and @xmath210 are @xmath17-vertices , which is depicted in figure [ fig2 - 1 ] . by observation [ three structures](b )
, @xmath211 can not be coincided with @xmath212 , where @xmath213 .
let @xmath214 .
for each @xmath215 , we color @xmath211 and @xmath61 with a color in @xmath216 and @xmath217 , respectively .
if @xmath218 , then color @xmath81 with @xmath219 .
it is easy to see that there exists at least one color in @xmath220 which appears at most once on the set @xmath221 .
so we may assign such a color to @xmath8 .
now suppose that @xmath222 . by symmetry
, we may suppose that @xmath223 .
this implies that @xmath151 .
thus , we can first color @xmath81 with @xmath224 and then assign a color in @xmath225 to @xmath8.@xmath132 [ lem-2 ] suppose @xmath8 is a @xmath73-vertex incident to two @xmath17-faces @xmath74 $ ] and @xmath75 $ ] . let @xmath76 be the neighbour of @xmath8 not belonging to @xmath78 and @xmath226
. then the following holds .
* we only need to show that @xmath231 since @xmath232 by ( a1 ) .
suppose that @xmath76 is a @xmath17-vertex .
let @xmath233 . by the minimality of @xmath0 , @xmath142 has an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
it is easy to deduce that @xmath108 for each @xmath234 and @xmath186 .
so we first color each @xmath61 with @xmath110 .
observe that there exists a color @xmath188 that appears at most once on the set @xmath235 .
therefore , we can color @xmath8 with @xmath174 to obtain an @xmath100-coloring of @xmath0 .
* suppose that @xmath236 , @xmath77 and @xmath119 and @xmath140 are both @xmath67-vertices . by definition
, we see that @xmath61 is either a @xmath17-vertex or a light @xmath12-vertex , where @xmath71 .
let @xmath237 . by the minimality of @xmath0 , @xmath142 has an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
the proof is split into two cases in light of the conditions of @xmath140 . * * assume @xmath140 is a @xmath17-vertex .
it is easy to calculate that @xmath108 for each @xmath238 and @xmath111 . by lemma [ key-0 ]
, @xmath5 can be extended to @xmath0 . *
* assume @xmath140 is a light @xmath12-vertex . by definition ,
let @xmath239 denote the other two neighbors of @xmath140 not on @xmath240 .
recolor @xmath241 and @xmath242 with a color different from its neighbors .
next , we will show how to extend the resulting coloring @xmath128 to @xmath0 .
denote @xmath243 be the induced assignment of @xmath143 .
notice that @xmath244 for each @xmath245 .
if @xmath246 , then by lemma [ key-0 ] , @xmath128 can be extended to @xmath0 . otherwise , we derive that @xmath247 .
first we assign a color in @xmath248 to each @xmath61 , where @xmath245 .
it is easy to see that there is at least one color , say @xmath174 , belonging to @xmath249 that appears at most once on the set @xmath250 .
we assign such a color @xmath174 to @xmath8 .
then color @xmath140 with a color in @xmath251 but different from @xmath174 . *
suppose that @xmath78 and @xmath226 are both @xmath79-faces such that @xmath252 and @xmath119 and @xmath140 are @xmath67-vertices .
let @xmath253 .
obviously , @xmath142 has an @xmath100-coloring @xmath5 by the minimality of @xmath0 .
let @xmath104 be an induced list assignment of @xmath143 .
we assert that @xmath61 satisfies that @xmath108 for each @xmath109 and @xmath111 . by lemma [ key-0 ] , we can extend @xmath5 to the whole graph @xmath0 successfully .
@xmath132 * proof .
* suppose to the contrary that there is a @xmath17-face @xmath254 $ ] incident to two bad @xmath73-vertices @xmath8 and @xmath80 , depicted in figure [ fig4 ] .
let @xmath255 . by the minimality of @xmath0 , @xmath142 has an @xmath100-coloring @xmath5 .
let @xmath104 be an induced list assignment of @xmath143 .
since each @xmath256 has at most two neighbors in @xmath142 , we deduce that @xmath257 for each @xmath145 .
so we first color each @xmath256 with a color @xmath258 .
if @xmath205 , namely @xmath259 , then by lemma [ key-0 ] we may easy extend @xmath5 to @xmath0 , since @xmath108 for each @xmath145 .
otherwise , we deduce that there exists a color @xmath174 in @xmath260 that is the same as @xmath261 for some fixed @xmath262 .
color @xmath80 with @xmath174 and @xmath61 with a color @xmath110 firstly , where @xmath145 .
for our simplicity , denote @xmath263 .
first , suppose that there is a color , say @xmath264 , appearing at most once on the set @xmath265 .
we assign such a color @xmath266 to @xmath8 .
if @xmath267 , the obtained coloring is obvious an @xmath100-coloring . otherwise , assume that @xmath268 .
now we erase the color @xmath174 from @xmath80 .
one may check that the resulting coloring , say @xmath128 , satisfies that each of @xmath269 has at least one possible color in @xmath143 . in other words , @xmath270 for each @xmath271 .
hence , by lemma [ key-0 ] , we can easily extend @xmath128 to @xmath0 .
now , w.l.o.g . , suppose that @xmath114 , @xmath272 , @xmath273 and each color in @xmath116 appears exactly twice on the set @xmath265 .
it implies that @xmath274 .
we apply versions of discussion ( i ) and ( ii ) in the proof of lemma [ key-0 ] . after doing that , one may check that now @xmath8 is colored with @xmath275 and @xmath119 is recolored with a new color , say @xmath276 .
there are two cases left to discuss : if @xmath277 , namely the new color of @xmath8 is @xmath17 , then the obtained coloring is an @xmath100-coloring and thus we are done ; otherwise , we uncolor @xmath80 . again , it is easy to see that the resulting coloring , say @xmath278 , satisfies that @xmath279 for each @xmath271 . therefore , we can easily extend @xmath278 to @xmath0 successfully by lemma [ key-0 ] .
@xmath132 we now apply a discharging procedure to reach a contradiction .
suppose that @xmath81 is adjacent to a @xmath17-vertex @xmath8 such that @xmath280 is not incident to any @xmath17-faces .
we call @xmath8 a _ free _ @xmath17-vertex if @xmath281 and a _ pendant
_ @xmath17-vertex if @xmath282 . for simplicity
, we use @xmath283 to denote the number of free @xmath17-vertices adjacent to @xmath81 and @xmath284 to denote the number of pendant @xmath17-vertices of @xmath81 .
suppose that @xmath8 is a soft @xmath12-vertex such that @xmath285 $ ] is a @xmath12-face and @xmath286 .
if the opposite face to @xmath78 via @xmath8 , i.e. , @xmath226 , is of degree at least @xmath73 , then we call @xmath8 a _ weak _ @xmath12-vertex .
we notice that every weak @xmath12-vertex is soft but not vice versa . for @xmath287 and @xmath288 ,
let @xmath289 denote the amount of weights transferred from @xmath28 to @xmath29 .
suppose that @xmath68 $ ] is a 3-face .
we use @xmath290 to denote @xmath291 for @xmath292 .
our discharging rules are defined as follows : * case * @xmath309 . initially @xmath310 .
let @xmath68 $ ] with @xmath311 . by ( a1 )
, @xmath312 .
if @xmath120 , then @xmath302 by ( a2 ) .
together with ( b2 ) , we deduce that @xmath47 is either a @xmath313-face , a @xmath314-face , a @xmath315-face , a @xmath316-face or a @xmath317-face .
it follows from ( b3 ) and lemma [ key ] that every possibility is indeed covered by rule ( r1 ) .
obviously , @xmath47 takes charge 4 in total from its incident vertices .
therefore , @xmath318 .
* case * @xmath319 . clearly , @xmath320 .
assume that @xmath321 $ ] is a @xmath12-face . by ( a2 ) , there are no adjacent @xmath17-vertices in @xmath0 .
it follows that @xmath47 is incident to at most two @xmath17-vertices . by symmetry
, we have to discuss three cases depending on the conditions of these @xmath17-vertices .
* @xmath322 . by ( f1 )
, we deduce that at least one of @xmath81 and @xmath8 is of degree at least @xmath73 .
moreover , if one of @xmath81 and @xmath8 is a @xmath12-vertex , say @xmath8 , we claim that @xmath8 can not be weak by definition and ( b1 )
. hence , @xmath323 by ( r2 ) and ( r3 ) . * @xmath324 and @xmath325 .
note that @xmath81 and @xmath8 are both @xmath306-vertices .
similarly , neither @xmath81 nor @xmath8 can be a weak @xmath12-vertex .
it follows from ( r3.3 ) and ( r2 ) that each of @xmath81 and @xmath8 sends charge at least @xmath326 to @xmath47 .
so if one of them is a @xmath298-vertex , say @xmath8 , then by ( r2 ) we have that @xmath303 and thus @xmath47 gets @xmath327 in total from incident vertices of @xmath47 .
otherwise , suppose @xmath328 .
now by ( f2 ) , @xmath29 can not be a soft @xmath12-vertex and thus not weak .
hence , @xmath329 by ( r3.2 ) . * @xmath330 and @xmath325 .
namely , @xmath47 is a @xmath331-face . if at most one of @xmath332 is a weak @xmath12-vertex , then @xmath329
otherwise , by lemma [ adjacent soft ] , assume that @xmath8 and @xmath81 are weak @xmath12-vertices and thus soft .
we see that @xmath333 by ( r3.2.1 ) and ( r2.1 ) which implies that @xmath334 .
now we consider the vertices .
let @xmath8 be a @xmath25-vertex with @xmath337 by ( a1 ) . for @xmath3 , we use @xmath338 to denote the number of @xmath12-faces incident to @xmath8 .
so by observation [ three structures ] ( a ) and ( b ) , we derive that @xmath339 and @xmath340
. furthermore , @xmath341 by observation [ three structures ] ( c ) .
* case * @xmath309 .
then @xmath348 .
clearly , @xmath349 .
if @xmath282 , then there exists a neighbor of @xmath8 , say @xmath81 , so that @xmath8 is a pendant @xmath17-vertex of @xmath81 . by ( a2 ) , @xmath350 .
thus , @xmath351 by ( r4 ) .
otherwise , we obtain that @xmath352 by ( r4 ) .
* case * @xmath319 .
then @xmath353 .
note that @xmath354 . if @xmath355 , then @xmath356 and @xmath357 .
so @xmath358 by observation [ charge ] ( a ) .
if @xmath281 , then @xmath359 by ( b1 ) and @xmath360 . we need to consider following cases . * @xmath361 . w.l.o.g .
, assume that @xmath285 $ ] and @xmath362 $ ] are incident @xmath12-faces
. obviously , @xmath357 by observation [ three structures ] ( b ) .
however , @xmath363 by ( b1 ) . by ( r3 )
, @xmath8 sends charge at most 1 to @xmath51 , where @xmath364 . if @xmath365 , then @xmath366 and thus @xmath358 .
if @xmath367 , say @xmath119 is a @xmath17-vertex , then @xmath368 by ( r3.3 ) and thus @xmath369 by ( r4 ) .
now suppose that @xmath370 . by symmetry
, we have two cases depending on the conditions of these two @xmath17-vertices . if @xmath304 , then @xmath371 by ( r3.1.2 ) . by ( b1 ) , @xmath140 and @xmath175 are both @xmath306-vertices .
moreover , neither @xmath140 nor @xmath175 is a soft @xmath12-vertex according to lemma [ adjacent soft ] .
so by ( r3.2.2 ) , @xmath372 .
hence @xmath373 .
otherwise , suppose that @xmath374 , where @xmath215 and @xmath375 .
we derive that @xmath376 by ( r3.3 ) .
* @xmath377 .
w.l.o.g , assume that @xmath378 .
this implies that @xmath379 .
again , @xmath380 by ( r3 ) .
if @xmath381 then we have that @xmath382 by ( r4 ) .
so in what follows , we assume that @xmath370 . if @xmath286 then @xmath8 is a weak @xmath12-vertex , implying that @xmath8 sends nothing to @xmath78 .
so @xmath358 by ( r4 ) .
if @xmath304 , then @xmath357 by observation [ three structures ] ( b ) .
we deduce that @xmath383 by ( r3.1.2 ) and ( r4 ) .
otherwise , suppose @xmath374 , where @xmath215 and @xmath375 .
it follows immediately from ( r3.3)and ( r4 ) that @xmath369 . *
. obviously , @xmath358 by ( r4 ) .
now , in the following , we consider the case @xmath282 .
assume that @xmath78 is a @xmath17-face . by ( a1 ) and
( b2 ) , @xmath78 is either a @xmath384-face , a @xmath385-face or a @xmath386-face .
observe that @xmath387 .
first assume that @xmath356 . if @xmath78 is a @xmath384-face , then @xmath388 by ( b1 ) and hence @xmath382 by observation [ charge ] ( a ) and ( r2 ) .
next suppose that @xmath78 is a @xmath385-face .
if @xmath370 , then @xmath8 is a light @xmath12-vertex . by ( r1.3 ) , we see that @xmath8 sends nothing to @xmath78 and therefore @xmath358 by ( r4 ) .
otherwise , at most one of @xmath389 is a @xmath17-vertex and hence @xmath382 by observation [ charge ] ( a ) and ( r4 ) .
finally , we suppose that @xmath78 is a @xmath386-face . if neither @xmath119 nor @xmath72 is a bad @xmath73-vertex , then @xmath8 sends nothing to @xmath78 by ( r1.4 ) and thus @xmath358 by ( r4 ) . otherwise , one of @xmath119 and @xmath72 is a bad @xmath73-vertex . if follows directly from ( c2 ) that @xmath381 .
therefore , @xmath382 by ( r2 ) .
now suppose that @xmath377 . by observation [ three structures ] ( c ) , we may assume that @xmath362 $ ] is a @xmath12-face . in this case , @xmath357 .
if @xmath286 , then @xmath390 by ( r3.1.1 ) . it follows from ( b1 ) and ( c2 ) that @xmath47 is neither a @xmath384-face nor a @xmath391-face such that @xmath72 is a bad @xmath73-vertex .
so we deduce that @xmath78 gets nothing from @xmath8 by ( r1.3 ) , which implies that @xmath392 . if exactly one of @xmath389 is a @xmath17-vertex , then @xmath372 by ( r3,3 ) .
thus , @xmath393 by observation [ charge ] ( a ) and ( r4 ) .
otherwise , we suppose that @xmath389 are both of degree at least 4 . in this case ,
@xmath366 and hence @xmath382 by ( r3.2 ) and observation [ charge ] ( a ) .
assume @xmath281 .
if @xmath356 , then @xmath396 by ( r4 ) . if @xmath377 , then @xmath397
. thus @xmath398 by ( r2 ) and ( r4 ) .
now suppose that @xmath361 . by observation [ three structures ] ( c ) , we assert that @xmath388 .
so @xmath399 .
next assume @xmath282 , say @xmath78 .
then @xmath400 by observation [ charge ] ( b ) .
moreover , equality holds iff @xmath78 is a @xmath79-face .
so if @xmath401 then at most one of @xmath402 is a @xmath17-vertex by ( b4 ) .
furthermore , @xmath387 .
when @xmath356 , we deduce that @xmath403 by ( r4 ) .
when @xmath377 , by symmetry , say @xmath226 is a @xmath12-face , we have two cases to discuss : if @xmath404 , namely , @xmath76 is a @xmath17-vertex , then @xmath405 by ( r2 ) and neither @xmath140 nor @xmath175 takes charge from @xmath8 .
thus @xmath406 ; otherwise , @xmath357 and we have @xmath407 .
now suppose that @xmath408 . by ( r2 ) and
( r4 ) , @xmath409 if @xmath356 and @xmath410 if @xmath377
. now assume @xmath355 . by symmetry ,
assume @xmath78 and @xmath226 are both @xmath17-faces .
observe that @xmath356 .
for simplicity , denote @xmath411 and @xmath412 .
let @xmath413 .
if @xmath414 , then @xmath415 by ( r2 ) .
now assume that @xmath416 , i.e. , @xmath78 gets charge @xmath17 from @xmath8 .
it means that @xmath78 is a @xmath79-face by observation [ charge ] . by ( c3 )
, @xmath226 can not be a @xmath79-face .
this implies that @xmath417 .
moreover , if @xmath76 is a @xmath17-vertex , then @xmath226 is neither a @xmath229-face by ( c2 ) nor a @xmath418-face by ( c1 ) .
it follows from ( r1.4 ) and ( r1.5 ) that @xmath419 , since @xmath8 is a bad @xmath73-vertex .
thus , @xmath406 by ( r2 ) .
otherwise , we easily obtain that @xmath420 .
* case * @xmath421 .
notice that @xmath422 .
if @xmath8 is incident to a @xmath12-face @xmath51 , then by ( r2 ) we inspect @xmath8 sends a charge at most @xmath423 to @xmath51 , while @xmath308 to each of @xmath61 and @xmath424 .
so we may consider @xmath8 as a vertex which sends charge at most @xmath425 to @xmath51 .
so by ( r4 ) and observation [ charge ] , we have @xmath426 if @xmath427 , then @xmath428 .
now suppose that @xmath429 .
if @xmath354 then @xmath430 .
so , in what follows , assume that @xmath431 and @xmath432 for @xmath433 . clearly , @xmath356 .
similarly , if there are at most two of @xmath17-faces get charge @xmath434 in total from @xmath8 , then @xmath435 .
otherwise , suppose @xmath436 for each @xmath437 . by observation [ charge ] ( b ) , we assert that @xmath51 is a @xmath438-face . noting that a @xmath438-face is also a @xmath136-face , we may regard @xmath8 as a @xmath135-vertex which is incident to two @xmath136-faces and one @xmath438-face .
however , it is impossible by ( b5 ) . | a list assignment of a graph @xmath0 is a function @xmath1 that assigns a list @xmath2 of colors to each vertex @xmath3 .
an @xmath4-coloring is a mapping @xmath5 that assigns a color @xmath6 to each vertex @xmath3 so that at most @xmath7 neighbors of @xmath8 receive color @xmath9 . a graph @xmath0 is said to be @xmath10-choosable if it admits an @xmath4-coloring for every list assignment @xmath1 with @xmath11 for all @xmath3 . in 2001 , lih et al .
@xcite proved that planar graphs without @xmath12- and @xmath13-cycles are @xmath14-choosable , where @xmath15 .
later , dong and xu @xcite proved that planar graphs without @xmath12- and @xmath13-cycles are @xmath14-choosable , where @xmath16 .
there exist planar graphs containing @xmath12-cycles that are not @xmath14-choosable ( crown , crown and woodall , 1986 @xcite ) .
this partly explains the fact that in all above known sufficient conditions for the @xmath14-choosability of planar graphs the @xmath12-cycles are completely forbidden . in this paper
we allow @xmath12-cycles nonadjacent to relatively short cycles .
more precisely , we prove that every planar graph without @xmath12-cycles adjacent to @xmath17- and @xmath12-cycles is @xmath14-choosable .
this is a common strengthening of all above mentioned results .
moreover as a consequence we give a partial answer to a question of xu and zhang @xcite and show that every planar graph without @xmath12-cycles is @xmath14-choosable .
_ keyword : _
planar graphs ; improper choosability ; cycle .
= 0.5 cm |
spintronics continues to attract interest because of potential applications to information technology , but also because it has greatly improved our understanding of the role of different spin - dependent interaction mechanisms , i.e. of the spin - orbit coupling , for many - electron systems.@xcite several spintronic devices have been proposed that manipulate the carrier spin via spin - orbit coupling ( soc).@xcite in recent years , progress in nanofabrication and growth techniques has made it possible to produce high - quality quantum wires ( qwrs ) and investigate physics in these semiconductor nanostructures.@xcite the energy spectrum of qwr systems with strong soc has been studied experimentally@xcite and theoretically.@xcite for @xmath1-type qwrs , most of the works concentrate on the energy spectrum,@xcite but for @xmath0-type quantum structures , many investigations have been performed with the aim of understanding the electron spin relaxation.@xcite for qwrs , the spin relaxation time ( srt ) was measured,@xcite and calculated in the framework of a single - particle model@xcite and monte - carlo simulations.@xcite recently , liu _ _ et al.__@xcite investigated the srt for qwrs in the ( 110 ) crystal direction , but systematic studies of spin relaxation in @xmath0-type qwrs are scarce . a general way to obtain the nonequilibrium spin dynamics and the spin relaxation time in semiconductors heterostructures has been developed by wu _
et al._@xcite in this approach , the momentum and spin dependent distribution functions are calculated using microscopic kinetic spin bloch equations ( ksbes ) , which include the momentum - dependent dresselhaus and/or rashba soc together with the effect of the relevant scattering mechanisms.@xcite cheng _ et al . _
applied this approach to study electron spin relaxation in qwr systems .
@xcite however , the influence of higher subbands was not included in this work .
subsequently , the effects of higher subbands and their coupling via coulomb scattering were shown to important for the spin - relaxation of _ holes _ in qwrs@xcite and for the case of electrons in quantum wells.@xcite in this paper , we study the influence of higher subbands on the spin relaxation in @xmath0-type inas qwrs .
this allows us to investigate qwrs with a wide range of sizes . especially for larger diameters of the wires , the spin precession and the spin relaxation are expected to show different characteristics from narrow ones , as electrons populate more than one subband and therefore experience different socs and undergo intra - subbband and inter - subband coulomb scattering .
we find that the inter - subband coulomb scattering can make an important contribution to the spin relaxation .
we also study the influence of the growth direction of the qwrs on the spin relaxation .
in particular , we consider qwrs with ( 001 ) , ( 110 ) , and ( 111 ) growth directions , and show that the srt depends sensitively on the growth direction , quantum - wire size and the direction of the initial spin polariazation . this paper is organized as follows : in sec .
ii we describe our model and the ksbes .
our numerical results are presented in sec .
we conclude in sec . iv .
we model the inas qwr by a rectangular confinement potential , i.e. , we assume infinitely high barriers at @xmath2 and @xmath3 , and no confinement on the @xmath4 direction .
the hamiltonian , which describes the electronic single - particle states in the qwr is then taken to include the confinement potential @xmath5 , the rashba term @xmath6 and the dresselhaus term @xmath7 @xmath8 the rashba and dresselhaus terms are the two contributions to the internal @xmath9-dependent effective field , which leads to the dyakonov - perel spin dephasing mechanism . the rashba term @xmath10\end{aligned}\ ] ] is due to the inversion asymmetry of the crystal structure .
the dresselhaus term is different for different growth directions .
for a ( 100 ) inas qwr , the @xmath11 , @xmath12 and @xmath4 axes correspond to the [ 100 ] , [ 010 ] and [ 001 ] crystallographic directions , respectively , and the dresselhaus term can be written as : @xcite @xmath13 + \sigma_y [ k_y ( k_z^2 - k_x^2 ) ] \nonumber \\ & & \mbox { } + \sigma_z [ k_z ( k_x^2 - k_y^2 ) ] \ } ~.\end{aligned}\ ] ] for a ( 110 ) qwr , the @xmath11 , @xmath12 and @xmath4 directions correspond to the [ @xmath1410 ] , [ 001 ] and [ 110 ] crystallographic directions , and we have @xmath15 + 2 \sigma_y k_x k_y k_z \nonumber \\ & & \mbox { } + \sigma_z [ \frac{1}{2 } k_x ( k_x^2 - k_z^2 - 2 k_y^2 ) ] \}~.\end{aligned}\ ] ] for a ( 111 ) qwr , the @xmath11 , @xmath12 and @xmath4 directions correspond to the [ 11@xmath16 , [ @xmath1410 ] , and [ 111 ] crystallographic directions , and we have @xmath17 + \sigma_y [ \frac{1}{2\sqrt{3 } } k_x^3 \nonumber \\ & & \mbox { } + \frac{1}{2\sqrt{3 } } k_x k_y^2 -\frac{1}{\sqrt{6 } } k_x^2 k_z - \frac{1}{\sqrt{6 } } k_z ( k_x^2 + k_y^2 ) ] \nonumber \\ & & \mbox { } + \sigma_z [ \frac{\sqrt{3}}{\sqrt{2 } } k_x^2 k_y - \frac{1}{\sqrt{6 } } k_y^3 - \frac{2}{3 } k_z k_y^2 ] \}~.\end{aligned}\ ] ] as input for the ksbes we use the basis @xmath18 of single - particle states , which are obtained from the eigenfunctions of @xmath19 . for a hard - wall confinement potential which constricts the electrons in the @xmath11 and @xmath12 directions on mesoscopic length scales
, we can employ the envelope function approximation.@xcite thus we write the single - particle states in the form @xmath20 with @xmath21 where @xmath22 denotes the basis vectors in spin space , i.e. , eigenstates of @xmath23 . in the envelope function approximation
, the effective hamiltonian acting on the single - particle states is obtained from eqs .
( [ rashba_100]-[dresselhaus_111 ] ) by the replacements @xmath24 and @xmath25 where @xmath26 .
a similar replacement is done for @xmath27 . for our choice of the confinement potential
, we have in particular @xmath28}(1-\delta_{n_x , n_x^{\prime}})\ , \nonumber \\ & & \langle k_x^2 \rangle
= \frac{\hbar^2 \pi^2 n_x^2}{a_x^2 } \delta_{n_x , n_x^{\prime } } \ .\end{aligned}\ ] ] with corresponding results for @xmath29 and @xmath30 . from eqs .
( [ dresselhaus_100 ] ) and ( [ kxkx2 ] ) one can see that when @xmath31 and @xmath32 are sufficiently small and only the lowest subband in qwr is important , only the third term of eq .
( [ dresselhaus_100 ] ) is not zero , which means the effective magnetic field contributed by the dresselhaus term is along the @xmath4-direction . on the other hand , the third term in eq .
( [ rashba_100 ] ) is zero but the terms which proportional to @xmath33 and @xmath34 are not zero .
this means that the effective magnetic field contributed by the rashba term is in the @xmath11-@xmath12 plane when the wire width is sufficiently small .
nevertheless for larger width of the wire , when higher subbands are needed , all the terms in eqs .
( [ rashba_100 ] ) and ( [ dresselhaus_100 ] ) contribute .
moreover , one can see that when the confinement in the @xmath11 and @xmath12 directions is symmetrical , the third term of eq .
( [ dresselhaus_100 ] ) is zero , so that the effective magnetic field contributed by the dresselhaus term does not contain any component along the @xmath4-direction .
as we show in detail below , these differences in the spin - dependent single - particle states lead to significantly different behavior of the srt .
the complete dynamical information about spin - dependent single - particle properties is contained in the spin - density matrix @xmath35 .
its matrix elements are , in general , defined with respect to the complete set of quantum numbers @xmath36 , but because of the translation symmetry in @xmath4 direction , @xmath35 is diagonal in @xmath9 , i.e. , @xmath37 with @xmath38 .
we construct the ksbes by the non - equilibrium green function method as follows:@xcite @xmath39 the coherent terms can be written as @xmath40 -i \big [ h_{e}(k ) , \rho_k \big ] , \end{aligned}\ ] ] where @xmath41 = ab - ba$ ] denotes the commutator , and @xmath42 .
@xmath43 is a matrix in @xmath44 space and can be considered as a form factor .
its definition reads @xmath45 where @xmath46 \left[\frac{1}{\pi^2(m_1 - m_2)^2 - a^2 q^2 } - \frac{1}{\pi^2(m_1 + m_2)^2 - a^2 q^2}\right].\ ] ] the first term in eq .
( [ coh ] ) is the coulomb hartree - fock term , and the second term is the contribution from the single - particle hamiltonian , i.e. , eq . in @xmath44 space , which includes the spin - orbit coupling terms . for small spin polarization ,
the contribution from the hartree - fock term in the coherent term is negligible@xcite and the coherent spin dynamics is essentially due to the spin precession around the effective internal fields described by eqs .
( [ rashba_100]-[dresselhaus_111 ] ) the scattering contributions to the dynamic equation of the spin - density matrix include scatterings between electrons and nonmagnetic impurities , electrons and phonon , and electron and electron scatterings : @xmath47 + h.c.\ , \nonumber \\
\frac{\partial \rho_{{k}}}{\partial t}\big|_{\rm ph } & = & \pi \sum_{{\mathbf{q}},s_1,s_2,\lambda } |m_{{\mathbf{q}},\lambda}|^2 i_{\mathbf{q } } \
{ \delta(e_{s_1,k - q } - e_{s_2,k } + \omega_{{\mathbf{q}},\lambda } ) [ ( n_{{\mathbf{q}},\lambda } + 1)(1- { \rho}_{k - q } ) t_{s_1}i_{-{\mathbf{q } } } t_{s_2 } { \rho}_k \nonumber \\ & & \mbox { } - n_{\mathbf { q},\lambda } \rho_{k - q } t_{s_1 } i_{-{\mathbf{q } } } t_{s_2}(1-\rho_k ) ] + \delta(e_{s_1,k - q } - e_{s_2,k } - \omega_{{\mathbf{q}},\lambda } ) [ n_{{\mathbf{q}},\lambda}(1- { \rho}_{k - q } ) t_{s_1}i_{-{\mathbf{q } } } t_{s_2 } { \rho}_k \nonumber \\ & & \mbox{}- ( n_{\mathbf { q},\lambda}+1 ) \rho_{k - q } t_{s_1}i_{-{\mathbf{q } } } t_{s_2}(1-\rho_k ) ] \}+ h.c .
\ , \nonumber \\
\frac{\partial \rho_{{k}}}{\partial t}\big|_{\rm ee } & = & \pi \sum_{{\mathbf{q}},k^{\prime } } \sum_{s_1,s_2,s_3,s_4 } v_{\mathbf{q}}^2 \delta(e_{s_1 , { k}-q } - e_{s_2,k } + e_{s_3,k^{\prime } } - e_{s_4 , k^{\prime } -q})i_{\mathbf{q } } \nonumber \\ & & \mbox{}\times \ { ( 1-{\rho}_{k - q})t_{s_1 } i_{-{\mathbf{q } } } t_{s_2 } { \rho}_k \mbox{tr}[(1-\rho_{k^{\prime}})t_{s_3 } i_{\mathbf{q } } t_{s_4}\rho_{k^{\prime } -q } i_{-\mathbf { q } } ] \nonumber \\ & & \mbox{}- \rho_{\mathbf { k}-q } t_{s_1 } i_{-\mathbf { q } } t_{s_2}(1-\rho_k ) \mbox{tr}[\rho_{k^{\prime}}t_{s_3 } i_{\mathbf{q } } t_{s_4 } ( 1-\rho _ { k^{\prime}-q } ) i_{-\mathbf { q } } ] \ } + h.c . \ \ , \label{scat } \end{aligned}\ ] ] in which @xmath48 . the statically screened coulomb potential in the random - phase approximation ( rpa ) reads @xcite @xmath49 with the bare coulomb potential @xmath50 and @xmath51 in eq .
( [ scat ] ) , @xmath52 is the density of impurities , and @xmath53 is the impurity potential .
further , @xmath54 and @xmath55^{-1}$ ] are the matrix element of the electron - phonon interaction and the bose distribution function , respectively . the phonon energy spectrum for phonon mode @xmath56 and wavevector @xmath57
is denoted by @xmath58 . for the electron - phonon scattering
, we include the electron lo - phonon and electron ac - phonon scattering , for which the explicit expressions can be found in refs . and .
note that , due to the weakness of the soc , the energy dispersions @xmath59 in the scattering terms , eq . ( [ scat ] ) , are taken from the diagonalization of @xmath60 and do not include the soc , as in ref . .
this is different from our previous work in @xmath1-type qwr systems .
@xcite for holes , the soc is stronger because holes are derived from @xmath1-orbitals , which experience the spin - orbit interaction directly , and it is therefore necessary to include the soc in the single - particle energy dispersions.@xcite
we numerically solve the ksbes for the spin density matrix @xmath35 , from which we obtain the dynamics of the average spin for electrons with momentum @xmath9 via @xmath61 where @xmath62 is the single - particle spin operator .
the srt @xmath63 can then be defined by an exponential fit to the envelope of the @xmath4 component of the average spin of the ensemble of electrons : @xcite @xmath64 in all the numerical results , we include the electron - phonon and electron - electron scattering . as initial condition
we assume a spin polarization along the @xmath4-direction with a small initial spin polarization @xmath65 where @xmath66 is the total electron density .
.material parameters used in the calculation ( from ref .
unless otherwise specified ) . [ cols= "
< , < , < , < , < , < , < , < , < , < , < " , ] [ parameter ] @xmath67 ref .
_ vs. _ the qwr width in @xmath12 direction , @xmath32 , for ( 100 ) qwrs at different @xmath31 .
the electron density is @xmath68 @xmath69 and @xmath70 k. the arrows mark the densities at which the electron populations in the second and higher subbands are approximately 30 % .,width=321 ] in fig . 1 we plot the srt as a function of the qwr width in @xmath12 direction , @xmath32 , for various @xmath31 .
we choose ( 100 ) qwrs at a lattice temperature of @xmath70 k and a total electron density of @xmath71 @xmath69 . for a small qwr of width @xmath72 nm , the srt increases monotonously with @xmath32 because for a small qwr the spacing of the subbands is large , so that for the present conditions only the lowest subband in qwr is appreciably populated .
therefore , as discussed in sec .
ii , the effective magnetic field due to the dresselhaus term contains a longitudinal component , @xmath73 , that keeps the electronic spins aligned and thus inhibits spin precession , which is mainly due to the rashba term , thereby effectively reducing the spin relaxation.@xcite an interesting effect arises because @xmath73 is proportional to @xmath74 , which disappears when the wire widths , and therefore in the present model also the confinement wave functions , in @xmath11 and @xmath12 directions are identical . for vanishing @xmath73 , i.e. , for @xmath75 , the srt reaches a minimum of several 10 ps for the qwrs considered here . changing the wire size , leads to increasing @xmath76 and therefore to increasing srt .
this is clearly illustrated for the @xmath77 nm case .
when @xmath78 nm , one can also find a minimum of srt at @xmath79 nm for the reason described above .
however , one finds that the srt decreases with @xmath32 again when @xmath80 nm .
this is because the electrons start to populate higher subbands when the wire width increases .
then the effective magnetic field formed by the dresselhaus term contains not only a longitudinal component but also a transverse component , @xmath81 , which couples different subbands .
this coupling has the same effect as a spin precession , and therefore tends to make the srt shorter .
moreover , the contribution of the dresseslhaus term becomes as important as the contribution of the rashba term .
when @xmath82 nm , @xmath81 dominates over @xmath83 due to the increasing @xmath32 , and a faster spin relaxation results .
the different minima of the srt at @xmath79 nm and @xmath84 nm are likely due to the different energy gaps between different subbands in the two cases . _ vs. _
@xmath85 for ( 110 ) qwrs at @xmath86 k and @xmath87 @xmath69 .
@xmath88 : @xmath89 nm with an initial spin polarization along the @xmath11-direction ; @xmath90 : @xmath89 nm with an initial spin polarization along the @xmath12-direction ; @xmath91 : @xmath92 nm with an initial spin polarization along the @xmath11-direction.,width=321 ] the results of the previous subsection showed that the geometry of the qwr has a pronounced effect on the srt because it influences the different contributions to the soc .
we now investigate qwrs with different growth directions , and start with the case of ( 110 ) qwrs .
the soc for ( 110 ) qwr is quite different from the soc for ( 100 ) qwr as shown in eq .
( [ dresselhaus_100 ] ) and eq .
( [ dresselhaus_110 ] ) .
first we only consider the case of a narrow wire with @xmath89 nm , for which the electronic population is mainly in the lowest subband . in the presence of an electric field of the form @xmath93 , the relevant contributions are @xmath94~ , \\
\label{dresselhaus_110_1band } & & h^{110}_{d } = -\frac{1}{2 } { b^{6c6c}_{41 } } \sigma_x k_z [ \langle k_x^2 \rangle - k_z^2 + 2\langle k_y^2 \rangle ] ~.\end{aligned}\ ] ] the effective magnetic field formed by the dresselhaus term is along the @xmath11-direction , which corresponds to the [ @xmath1410 ] crystallographic direction , and the effective magnetic field formed by the rashba term is in the @xmath11-@xmath12 plane . if the direction of the total effective magnetic field formed by the soc is tuned to be exactly the direction of the initial spin polarization , then one can expect an extremely long srt as pointed out in refs . and .
for the qwrs considered in this paper , we can study the physics that gives rise to this effect in the following way : we take the initial spin polarization to be along the @xmath11-direction . in fig . 2 we plot the srt as a function of @xmath95 for temperature @xmath86 k , electron density @xmath96 @xmath69 , and wire geometry @xmath89 nm . the srt decreases with @xmath85 , which is a measure of the effective magnetic field along @xmath12-direction . as expected , a very long srt results when @xmath85 is very small , even though the effective magnetic field along the @xmath11-direction is not zero .
in comparison , for an initial spin polarization along the @xmath12-direction a much shorter srt is obtained , which hardly changes with @xmath85 .
this is because the polarized electronic spins precess around the @xmath11-direction and this precession is not influenced strongly by the @xmath12-component of the effective magnetic field . in the inset of fig . 2
, we also show the dependence of the srt on the angle @xmath97 , which the electric field in the @xmath11-@xmath12 plane forms with the @xmath11 axis . assuming @xmath98 kv / cm and an initial spin polarization along the @xmath11-direction
, it is found that the srt decreases with @xmath97 . for small @xmath97 ,
i.e. , for effective magnetic fields close to the @xmath11-direction , the srt goes up , in agreement with the previous discussion .
the result for a larger wire size , @xmath92 nm are also plotted in fig . 2 . in this case
the srt is never longer than 1 ns , even when @xmath85 is very small because now electrons populate higher subbands , and the effective magnetic field formed by the dresselhaus term contains not only the longitudinal component but also a transverse component .
k , @xmath99 @xmath69 , @xmath100 @xmath69 , and @xmath101m.,width=321 ] we compare the srt obtained by our calculation with monte - carlo results from ref . in fig . 3 .
for a temperature of @xmath102 k. the total electron density and the impurity density are taken to be @xmath103 @xmath69 and @xmath104 @xmath69 , and the wire geometry is @xmath105 m .
the results without the coulomb scattering are actually very close to those obtained by monte - carlo simulations . however ,
when the coulomb scattering is included , much longer srts result .
this result underscores the importance of coulomb scattering for spin relaxation in qwrs . _ vs. _
@xmath85 for different wire sizes @xmath106 at @xmath86 k and @xmath87 @xmath69 .
the growth direction of the qwr is along the [ 111 ] crystallographic direction .
the solid curves are the results with the initial spin polarization along @xmath4-direction and the dashed curve is the result with the initial spin polarization along @xmath11-direction.,width=321 ] for the quantitative analysis of the srt for another growth direction we choose ( 111 ) qwrs . here
, we again consider the case of small wire width and low temperature first .
the electric field is again taken to be @xmath93 , so that the soc can be written as @xmath107~ , \\ \label{dresselhaus_111_1band } h^{111}_{d } & = & { b^{6c6c}_{41 } } \ { - \frac{\sqrt{2}}{3 } \sigma_x k_z \langle k_y^2 \rangle - \frac{1}{\sqrt{6}}\sigma_y k_z ( \langle k_x^2 \rangle + \langle k_y^2 \rangle ) \nonumber \\ & & \mbox { } -\frac{2}{3 } \sigma_z k_z \langle k_y^2 \rangle \}~.\end{aligned}\ ] ] similar to the case of ( 110 ) qwrs , we expect very long srts if the total effective magnetic field points into the direction of the initial spin polarization , at least for narrow wires . for a numerical example of this effect
, we choose @xmath108 such that @xmath109 for a small wire with @xmath110 nm , so that the @xmath11 component of the total effective magnetic field is zero . for an initial spin polarization along the @xmath4-direction , which corresponds to the [ 111 ] crystallographic direction
, we plot the srt as a function of @xmath85 in fig . 4 at @xmath86 k and @xmath111 @xmath69 .
4 shows that when @xmath110 nm , there is a pronounced maximum of the srt at @xmath112 kv / cm , which fulfils the relation @xmath113 .
consequently , for this field strength , the direction of the total effective magnetic field is exactly along the direction of the initial spin polarization and this leads to a very long srt .
however , this effect much reduced for larger wire cross sections . when @xmath114 nm , there is still a maximum of the srt , but the maximum is much less pronounced than for the smaller wire , because now @xmath115 remains finite for all field strengths as @xmath116 is changed .
this trend continues for wire sizes of @xmath89 nm and @xmath117 nm .
finally , we analyze the the case of @xmath118 nm with the initial spin polarization along @xmath11-direction : here one also does not obtain a maximum of the srt because the direction of the effective magnetic field is no longer identical to the direction of the initial spin polarization .
_ vs. _ the electron density for ( 100 ) qwrs at different wire sizes .
( a ) @xmath119 nm ; ( b ) @xmath120 nm ( b ) .
@xmath70 k.,width=321 ] since the population of higher subbands has been shown to play an extremely important role for the srt , we next analyze how the electronic population can be changed without using a different wire geometry or growth direction , namely by varying the doping density and/or temperature . in fig .
5(a ) we plot the srt as a function of @xmath66 for a ( 100 ) qwr of size @xmath92 nm , and @xmath121 k. for the full calculation , the srt decreases with @xmath66 because more electrons are present at higher momenta and in higher subbands , so that the effective magnetic fields experienced by these electrons are larger . by the dp mechanism , this results in a faster spin relaxation.@xcite to investigate how scattering affects the spin relaxation , we first switch off inter - subband _ electron - phonon _ scattering , with the result that the srt becomes shorter .
when we switch off the inter - subband _ electron - electron _ scattering , the srt also becomes shorter .
this somewhat counterintuitive effect of scattering on the spin dephasing results because the influence of scattering is different for different regimes , i.e. , for strong and weak scattering : if we define @xmath122 and @xmath123 to be the effective momentum relaxation time , then @xmath124 is typically much smaller than 1 for the densities considered here ( @xmath125 at @xmath126 @xmath127 if @xmath128 contains only the contribution from the inter - subband electron - electron scattering ) .
this means we are in the strong scattering regime where the motional narrowing picture _ qualitatively _ describes the dependence of the srt on the momentum scattering time as srt @xmath129 .
( microscopically this means that the scattering is strong enough to prevent significant deviations from isotropic electron distributions , which are caused by the anisotropic soc contributions . ) thus switching off scattering contributions effectively leads to a longer momentum relaxation time , and therefore to shorter srts , while still remaining in the strong scattering regime .
further , our calculation shows that the inter - subband electron - electron scattering is more important than the inter - subband electron - phonon scattering .
when we further exclude the intra - subband electron - electron scattering , it is found that the srt becomes shorter when @xmath130 @xmath69 , but becomes longer when @xmath131 @xmath69 . here
the most dominant scattering process is suppressed , so that with increasing @xmath66 the motional narrowing regime is left , because electrons occupy states at higher momenta and higher subbands where the soc contributions rapidly increase . for higher densities ,
the weak scattering regime is reached where the anisotropy of the soc contributions becomes dominant .
additional scattering leads to a more efficient dephasing , so that the srt increases when we switch off scattering contributions .
we also plot the case of larger wire width with @xmath132 nm in fig .
compared to the case of fig .
5(a ) , the electrons are populating higher subbands , so that the dresselhaus term becomes more important .
therefore , the band structure anisotropy will become important at smaller @xmath66 .
5(b ) shows that the result of the full calculation intersects the one without the electron - electron scattering at @xmath133 @xmath69 . for this wire geometry
, the result without the inter - subband electron - electron scattering also crosses the full calculation . to see in more detail how the strength of anisotropic soc contributions affects the spin relaxation
, we also plot the result without the dresselhaus term .
for this somewhat artificial case , the srt becomes much longer .
this indicates that the contribution of the dresseslhaus term is very important for this wire width in contrast to the case of smaller wire width in fig .
5(a ) , because the transverse component of the effective magnetic field contributed by the dresselhaus term is important only for the large wire width when higher subbands are populated .
_ vs. _ the temperature for ( 100 ) qwrs at different wire sizes @xmath106 at carrier density @xmath68 @xmath69.,width=321 ] finally we look at the temperature dependence of the srt at different wire widths in fig . 6 .
we choose ( 100 ) qwrs at a representative electron density of @xmath111 @xmath69 .
the temperature affects the srt in two ways : for the smaller wires @xmath134 nm the carriers are confined in the lowest subband without populating higher @xmath9-states where the anisotropic soc contributions to the bandstructure become much stronger . in this strong scattering regime , the behavior can qualitatively be explained by the motional narrowing : increasing temperature leads to enhanced scattering with phonons and electrons with higher kinetic energies , so that the effective carrier lifetime becomes shorter , and the srt increases .
around @xmath135 nm and @xmath136 k this behavior crosses over to the weak scattering limit , in which higher subbands are populated , the precession frequencies around the effective internal fields become higher , and additional scattering leads to a more efficient dephasing , so that the srt decreases with @xmath137 .
in conclusion , we have investigated the spin relaxation of electrons in @xmath0-type inas qwrs .
the srt is calculated by numerically solving the microscopic ksbes including multiple subbands .
the inclusion of higher subbands allows us to investigate qwrs larger qwrs than in ref . , and we find that the quantum - wire size influences the spin relaxation time via the soc : the dresselhaus term contains a longitudinal contribution to the internal effective magnetic field , which can effectively reduce the spin precession , and thereby the spin relaxation .
it also contains a transverse component , which provides a contribution to the spin precession involving different subbands and enhances spin relaxation . when the wire width is small and only the lowest subband in the qwr is important , the longitudinal term is dominant .
when the wire width is large and higher subbands are populated , the transverse contribution dominates over the longitudinal one , and a faster spin relaxation results .
we also studied different growth directions for qwrs .
we show that one can obtain long spin relaxation time by optimizing the growth direction , quantum - wire width and the direction of the initial spin polariazation .
further , we investigated how the details of the microscopic scattering mechanisms and the spin - orbit effects in the band structure affect the spin relaxation .
the population of higher subbands was found to have decisive influence on the behavior of the srt .
for instance , if the geometry and external conditions are such that higher subbands become populated , the dependence of the srt on temperature is reversed because the motional narrowing regime is left .
this work was supported by the natural science foundation of china under grant no .
10725417 , the national basic research program of china under grant no .
2006cb922005 and the knowledge innovation project of the chinese academy of sciences .
we have also benefitted from a german - chinese cooperation grant from the bosch foundation .
one of the authors ( c.l . )
thanks j. h. jiang for many discussions . , ed . by d.
d. awschalom , d. loss , and n. samarth ( springer - verlag , berlin , 2002 ) ; j. fabian , a. matos - abiague , c. ertler , p. stano , and i. uti , acta phys . slov . * 57 * , 565 ( 2007 ) ; _ spin physics in semiconductors _ , ed . by m.
i. dyakonov ( springer , berlin , 2008 ) , and references therein .
i. uti , j. fabian , and s. das sarma , rev .
phys . * 76 * , 323 ( 2004 ) .
s. datta and b. das , appl .
. lett . * 56 * , 665 ( 1990 ) .
j. schliemann , j. c. egues , and d. loss , phys .
90 * , 146801 ( 2003 ) .
k. shen and m. w. wu , phys .
b * 77 * , 193305 ( 2008 ) ; l. wang , k. shen , s. y. cho , and m. w. wu , j. appl . phys . *
104 * , 123709 ( 2008 ) ; j. wan , m. cahay , and s. bandyopadhyay , physica e ( amsterdam ) * 40 * , 2659 ( 2008 ) .
l. n. pfeiffer , r. de picciotto , k. w. west , k. w. baldwin , and c. h. l. quay , appl .
. lett . * 87 * , 073111 ( 2005 ) .
r. danneau , w. r. clarke , o. klochan , a. p. micolich , a. r. hamilton , m. y. simmons , m. pepper , and d. a. ritchie , appl .
lett . * 88 * , 012107 ( 2006 ) .
o. klochan , w. r. clarke , r. danneau , a. p. micolich , l. h. ho , a. r. hamilton , k. muraki , and y. hirayama , appl .
. lett . * 89 * , 092105 ( 2006 ) .
hof , c. rossler , s. manus , j. p. kotthaus , a. w. holleitner , d. schuh , and w. wegscheider , phys . rev .
b * 78 * , 115325 ( 2008 ) .
s. farhangfar , phys .
b * 76 * , 205437 ( 2007 ) .
p. lehnen , t. schpers , n. kaluza , n. thillosen , and h. hardtdegen , phys .
b * 76 * , 205307 ( 2007 ) . b. k. agrawal , v. singh , r. srivastava , and s. agrawal , phys .
b * 74 * , 245405 ( 2006 ) .
s. pramanik , s. bandyopadhyay , and m. cahay , phys .
b * 76 * , 155325 ( 2007 ) .
s. zhang , r. liang , e. zhang , l. zhang , and y. liu , phys .
b * 73 * , 155316 ( 2006 ) .
d. csontos and u. zlicke , phys .
b * 76 * , 073313 ( 2007 ) ; appl . phys . lett * 92 * , 023108 ( 2008 ) ; d. csontos , p. brusheim , u. zlicke , and h. q. xu , phys .
b * 79 * , 155323 ( 2009 )
. y. arakawa , t. yamauchi , and j. n. schulman , phys .
b * 43 * , 4732 ( 1991 ) .
d. s. citrin and y .- c .
chang , phys .
b * 40 * , 5507 ( 1989 ) . o. stier and d. bimberg , phys .
b * 55 * , 7726 ( 1997 ) .
n. shtinkov , p. desjardins , r. a. masut , and s. j. vlaev , phys .
b * 70 * , 155302 ( 2004 ) .
r. danneau , o. klochan , w. r. clarke , l. h. ho , a. p. micolich , m. y. simmons , a. r. hamilton , m. pepper , d. a. ritchie , and u. zlicke , phys . rev
97 * , 026403 ( 2006 ) .
g. goldoni and a. fasolino , phys .
b * 52 * , 14118 ( 1995 ) ; g. goldoni , f. rossi , e. molinari , a. fasolino , r. rinaldi , and r. cingolani , appl . phys .
lett * 69 * , 2965 ( 1996 ) .
m. scheid , m. kohda , y. kunihashi , k. richter , and j. nitta , phys .
* 101 * , 266401 ( 2008 ) .
m. ohno and k. yoh , phys .
b * 75 * , 241308(r ) ( 2007 ) .
s. pramanik , s. bandyopadhyay , and m. cahay , phys .
b * 68 * , 075313 ( 2003 ) ; _ ibid . _ * 73 * , 125309 ( 2006 ) ; p. upadhyaya , s. pramanik , s. bandyopadhyay , and m. cahay , _ ibid . _ * 77 * , 045306 ( 2008 ) .
j. liu , t. last , e. j. koop , s. denega , b. j. van wees , and c. h. van der wal , arxiv:0810.1413v1 .
a. a. kiselev and k. w. kim , phys .
b * 61 * , 13115 ( 2000 ) . a. dyson and b. k. ridley , phys . rev .
b * 72 * , 045326 ( 2005 ) .
p. schwab , m. dzierzawa , c. gorini , and r. raimondi , phys .
b * 74 * , 155316 ( 2006 ) .
t. sogawa , h. ando , s. ando , and h. kanbe , phy .
b * 58 * , 15652 ( 1998 ) .
a. w. holleitner , v. sih , r. c. myers , a. c. gossard , and d. d. awschalom , phy .
lett . * 97 * , 036805 ( 2006 ) ; new j. of phys .
* 9 * , 342 ( 2007 ) .
j. l. cheng , m. q. weng , and m. w. wu , solid state commun . * 128 * , 365 ( 2003 ) . m. w. wu and h. metiu , phys .
b * 61 * , 2945 ( 2000 ) ; m. w. wu and c. z. ning , eur .
j. b * 18 * , 373 ( 2000 ) ; m. w. wu , j. phys .
* 70 * , 2195 ( 2001 ) . m. q. weng and m. w. wu , phys .
b * 68 * , 075312 ( 2003 ) ; _ ibid . _ * 70 * , 195318 ( 2004 ) .
m. q. weng , m. w. wu , and l. jiang , phys .
b * 69 * , 245320 ( 2004 ) .
j. l. cheng and m. w. wu , j. appl .
* 99 * , 083704 ( 2006 ) . c. l , u. zlicke , and m. w. wu , phys .
b * 78 * , 165321 ( 2008 ) .
s. dhrmann , d. hgele , j. rudolph , m. bichler , d. schuh , and m. oestreich , phys .
* 93 * , 147405 ( 2004 ) .
r. winkler , _ spin - orbit coupling effects in two - dimensional electron and hole systems _
( springer , berlin , 2003 ) .
h. haug and s. w. koch , _ quantum theory of the optical and electronic properties of semiconductors _
( world scientific , singapore , 2004 ) .
h. haug and a. p. jauho , _ quantum kinetics in transport and optics of semiconductor _
( springer - verlag , berlin , 1996 ) .
d. stich , j. zhou , t. korn , r. schulz , d. schuh , w. wegscheider , m. w. wu , and c. schller , phys .
lett . * 98 * , 176401 ( 2007 ) ; phys .
b * 76 * , 205310 ( 2007 ) .
j. zhou , j. l. cheng , and m. w. wu , phys .
b * 75 * , 045305 ( 2007 ) ; j. zhou and m. w. wu , _ ibid . _
* 77 * , 075318 ( 2008 ) .
m. krau , m. aeschlimann , and h. c. schneider , phys .
lett . * 100 * , 256601 ( 2008 ) . | we investigate the electron spin relaxation of @xmath0-type inas quantum wires by numerically solving the fully microscopic kinetic spin bloch equations with the relevant scattering explicitly included .
we find that the quantum - wire size and the growth direction influence the spin relaxation time by modulating the spin - orbit coupling .
due to inter - subband scattering in connection with the spin - orbit interaction , spin - relaxation in quantum wires can show different characteristics from those in bulk or quantum wells and can be effectively manipulated by various means . |
the current unification scheme for active galactic nuclei ( herefater agn ) postulates the existence of a thick molecular / dusty torus surrounding a black - hole / accretion - disc system ( see , e.g. krolik 1999 ) .
attempts to unveil this torus in a nearby agn like ngc1068 have drawn considerable effort on the side of high spatial resolution imaging , in particular in the near - ir ( chelli et al .
1987 , gallais , 1991 , young et al .
1996 , marco et al .
1997 , thatte et al . 1998 ,
rouan et al .
1998 , alloin et al .
1998 , wittkowski et al .
1998 , weinberger , neugebauer & matthews 1999 , marco & alloin , 2000 ) and mid - ir ( braatz et al .
1993 , cameron et al .
1993 , bock et al .
1998 , lumsden et al 1999 ) .
with respect to the presence of such a torus , the most compelling results are : ( a ) vlba 8 ghz data ( gallimore et al .
1997 ) revealing a 1 pc elongated distribution of ionized gas at pa @xmath4 110 ( the inner `` hot zone '' of the torus ? ) , ( b ) adaptive optics ( ao ) observations at 2.2 @xmath0 m ( rouan et al , 1998 ) showing extended emission along pa @xmath6 102 and up to a 15 pc radius ( in addition to a prominent unresolved core with size less that 9 pc and in addition to another 20 pc radius extended emission along pa @xmath6 15 which follows the narrow line region ) and at 3.5 and 4.8 @xmath0 m ( marco & alloin , 2000 ) revealing structures similar in shape / orientation and on comparable scales , ( c ) near- and mid - ir imaging polarimetry ( lumsden et al . 1999 ) calling for the presence of hot and warm dust components fully consistent with those detected with ao .
the structure elongated at pa @xmath4 102 could trace the equatorial plane of an inclined dusty / molecular torus , from its inner walls of ionized material seen in radio , to its outer parts of thermally emitting dust with a temperature range from t @xmath4 1500 k ( dust sublimation temperature ) to a few 100 k. models by krolik & begelman ( 1986 ) , pier & krolik ( 1993 ) , efstathiou & rowan - robinson ( 1994 ) , granato & danese ( 1994 ) , and granato , danese & franceschini ( 1997 ) have explored torus sizes from 1 to 100 pc .
they predict emission in the near - ir and mid - ir and provide high resolution model - maps at various wavelengths .
therefore , it is particularly timely to obtain high resolution mid - ir images to compare with model - maps . here
, we present and discuss new results from diffraction - limited images obtained with the cfht / hawaii and eso / la silla 3.6 m telescopes , at 11.2 and 20.5 @xmath0 m respectively .
hereafter , the distance used for ngc1068 is 14.4 mpc , leading to the equivalence 1@xmath4 70 pc in the object .
the observations were performed using the cea / sap mid - ir camera camiras ( lagage et al 1993 ) , equipped with a boeing 128x128 pixels bib detector which is sensitive up to 28 @xmath0 m .
the agn in ngc1068 was observed with camiras attached to the eso / la silla 3.6 m telescope on 1998 november 8 - 10 and to the cfht / hawaii 3.6 m telescope on 1999 august 1 . during these two runs seeing and weather conditions humidity and amount of atmospheric precipitable water were particularly favorable to observing in the mid - ir and extremely stable .
the orientation of the array on the sky was carefully determined at the start of each observing run .
the pixel size was @xmath7 ( 20 @xmath0 m window , eso / la silla ) and @xmath8 ( 11 @xmath0 m window , cfht / hawaii ) . in the 20 @xmath0 m window
, we used a filter centered at 20.5 @xmath0 m and with a bandpass ( fwhm ) @xmath9 = 1.11 @xmath0 m .
this filter is free of any important line contribution . in the 11 @xmath0 m window , we used a filter centered at 11.2 @xmath0 m and with a bandpass @xmath9 = 0.44 @xmath0 m : although the 11.3 @xmath0 m pah line emission falls within this filter , one should notice that the spectroscopic results from the iso satellite do not show its presence in the agn of ngc1068 ( aperture of @xmath10 ) .
therefore we are confident that these two filters reflect mostly the mid - ir continuum emission of the agn . to avoid saturation of the detector by the high ambient photon background ,
the image elementary integration times were chosen to be 9.1 msec and 19.1 msec , respectively at 20.5 and 11.2 @xmath0 m .
standard chopping and nodding techniques were applied with a chopping throw of @xmath11 .
elementary images were coadded in real time during 3 sec chopping cycles . at 20.5
@xmath0 m a total integration time of 22.5 min was spent on the source observed at an airmass less than 1.3 . at 11.2 @xmath0 m ,
the total integration on source was of 65 min and the mean airmass of 1.04 .
a shift - and - add procedure was applied to the final images through each filter .
the final signal - to - noise ratio at the peak emission on the images is 265 at 20.5 @xmath0 m and 980 at 11.2 @xmath0 m .
lcccc component & extension & distance & flux 20.5@xmath0 m & flux 11.2@xmath0 m & ( @xmath12 ) & ( @xmath13 ) & ( 1@xmath14 error , ( jy ) ) & ( 1@xmath14 error,(jy ) ) core & @xmath15 & 0 & 57.5 ( 6 ) & 37.5 ( 2 ) \(a ) n - e & @xmath16 & 3.7 & 1.4 ( 0.1 ) & 0.15 ( 0.01 ) \(b ) n - e & @xmath17 & 1.8 & 1.1 ( 0.1 ) & 2.6 ( 0.15 ) \(c ) s - w & @xmath18 & 2.1 & 1.0 ( 0.1 ) & 0.3 ( 0.02 ) \(d ) s - w & @xmath17 & 3.5 & 0.15(0.01 ) & undetected [ parameters ] the nearby reference star @xmath19 4 eri was frequently monitored to serve as a psf and flux standard at 20.5 @xmath0 m , while @xmath20 ceti was used at 11.2 @xmath0 m ( van der bliek et al 1996 ) .
the precision achieved in flux measurements is + / 5% and + / 10% , respectively at 11.2 @xmath0 m and 20.5 @xmath0 m .
because final images are limited by the seeing and above all by the 3.6 m telescope diffraction - pattern ( fwhm of @xmath21 and @xmath22 at 20.5 @xmath0 m and 11.2 @xmath0 m respectively ) , applying a deconvolution procedure is mandatory .
we use the entropy method developed by pantin & starck ( 1996 ) , based upon wavelet decomposition . the shannon theorem ( 1948 )
prescription allows to persue the deconvolution down to twice the pixel size , in this case down to @xmath2 , and iterations are stopped according to the residual map , whose properties should be consistent with data noise characteristics .
we provide in figure 1 a set of the images at 20.5 @xmath0 m : the ngc1068 image before deconvolution , the psf stellar image , and the ngc1068 image after deconvolution . a similar set of images at 11.2
@xmath0 m is displayed in figure 2 .
the general orientation of the extended emission runs through the north - east to south - west quadrants , with a noticeable north - south elongation of the inner isophotes .
the deconvolved image and contours at 11.2 @xmath0 m are fully consistent with the map at 12.4 @xmath0 m obtained at a comparable spatial resolution by braatz et al ( 1993 ) . in the 10 @xmath0 m window ,
the speckle data ( over a @xmath23 region , with a resolution down to @xmath24 , but with a smaller dynamical range ) analyzed by bock et al ( 1998 ) indicate however a different orientation with the extended emission in the inner @xmath23 region running through the north - west to south - east quadrants : this does not appear to be consistent with either braatz et al ( 1993 ) or the new data set presented in this paper .
this discrepancy remains to be elucidated .
the deconvolved images , both at 11.2 and 20.5 @xmath0 m show the presence of a prominent central core .
cuts through the core along the north - south and east - west directions provide fwhm values of @xmath3 and @xmath2 respectively .
therefore , the core remains unresolved in the east - west direction , while it appears to be extended north - south with intrinsic fwhm size of around 50 pc .
again this result is in very good agreement with braatz et al ( 1993 ) .
regarding the extended emission , one notices that conspicuous features can be seen further away from the core at 20.5 @xmath0 m than at 11.2 @xmath0 m . in the north - east quadrant
we have identified two clouds , ( a ) and ( b ) , along pa @xmath6 35 , at mean distances from the core of @xmath25 and @xmath26 , respectively .
cloud(a ) , which is bright both at 11.2 and 20.5 @xmath0 m , appears to widen perpendicularly to the main direction of the extended emission and runs from pa @xmath6 15 to pa @xmath6 55 . in the south - west quadrant , the emission closest to the core extends along pa @xmath6 210 and then aligns north - south : two clouds , ( c ) and ( d ) , can be singled out at mean distances from the core of @xmath27 and @xmath28 .
the results are summarized in table 1 .
flux measurements at 11.2 and 20.5 @xmath0 m have been performed using a mask built from the 20.5 @xmath0 m image and delineating the extent of the four clouds and of the core , so that the 11.2 @xmath0 m to 20.5 @xmath0 m flux ratio can be derived in a consistent fashion .
the mask at the position of the core takes into account the full extent ( @xmath29 ) of the core ( rather than its fwhms values given in table 1 ) .
the resultant fluxes are given in table 1 : one notices that the core contributes a very large fraction of the mid - ir emission , 95.6% at 11.2 @xmath0 m and 94% at 20.5 @xmath0 m .
we have just seen that the mid - ir emission extends away from the core up to a radius of about 300 pc , both in the north - east and south - west quadrants . in order to compare this extension / structure to those observed at other wavelengths , in the radio range with the vla or in the uv / optical with hst ,
it is mandatory to register precisely the mid - ir map with respect to the near - ir , visible and radio ones . as we did not have the possibility to perform astrometric measurements through simultaneous observations in the mid - ir and near - ir or visible ( as performed in marco et al 1997 )
, we are assuming firstly that the cores at 11.2 and 20.5 @xmath0 m are coincident and secondly that their location also coincides with that of the cores observed at 4.8 , 3.5 and 2.2 @xmath0 m .
regarding the first assumption , one can check from the model of ngc1068 developed by granato et al ( 1997 ) which predicts the offsets between emission peaks at various wavelengths in the near - ir to mid - ir , that these offsets remain in the range of a few hundredths of an arc second and can be ignored . regarding the second assumption , the cores observed at 4.8 , 3.5 and 2.2 @xmath0 m
have been shown to be positionally coincident indeed ( rouan et al , 1998 , marco & alloin 2000 ) . and
finally , marco et al ( 1997 ) have obtained the astrometric positioning of the core at 2.2 @xmath0 m , with respect to the hst map in the visible of the central area in ngc1068 : it was found to be located @xmath30 south and @xmath31 west of the so - called optical continuum peak seen with hst ( lynds et al 1991 ) .
the study by marco et al ( 1997 ) also allowed a precise registration of the 2.2 @xmath0 m map with respect to the 12.4 @xmath0 m map from braatz et al ( 1993 ) and to available radio , optical and uv maps .
the conclusion is that the core at 4.8 , 3.5 and 2.2 @xmath0 m appears to be coincident with the 12.4 @xmath0 m peak position found by braatz et al , the radio source s1 , the symmetry center of polarization both in the uv / optical ( capetti et al 1995 ) and in the mid - ir ( lumsden et al 1999 ) .
we consider this core to feature the central engine ( hidden in uv / optical maps ) and from all the above quoted studies we can ascertain the consistency of the assumptions made earlier .
we have compared the mid - ir maps at 11.2 and 20.5 @xmath0 m to the radio maps and to the hst [ oiii]emitting cloud map , provided on a suitable scale in bland - hawthorn et al ( 1997 ) and with a registration as above .
we have reached the following conclusions : + compared to the radio maps from gallimore et al ( 1996a , b ) and muxlow et al ( 1996 ) , the observed north - south elongation of the mid - ir core is reminiscent of the north - south alignment of the radio sources s2 , s1 and c ( see figure 2 in bland
hawthorn et al 1997 for a synthesis of the radio source nomenclature after wilson & ulvestad 1987 ) . at the location of radio source c ,
the orientation of the radio jet - like structure changes abruptly and becomes close to pa @xmath6 35 , matching then the general direction of the mid - ir extended emission in the north - east quadrant .
the mid - ir cloud(a ) coincides perfectly with the wide base of the so - called northeast radio lobe . in the south - west quadrant ,
the radio jet - like structure is along pa @xmath6 210 again very well aligned with the mid - ir extended emission .
the so - called southwest radio hotspot is situated about @xmath5 away from the central engine , s1 , and can be put in correspondence with the mid - ir cloud(d ) .
+ compared with the hst [ oiii]emitting cloud map , the north - south elongation of the mid - ir core is found to overlap to the north with hst cloud - a and cloud - b , two nlr clouds distributed along the north - south direction .
the extended mid - ir emission in the north - east quadrant is along pa @xmath6 35 and arises to the eastern edge of the ionizing cone ( which , on a large scale extends between pa @xmath6 15 and pa @xmath6 35 ) defined by the [ oiii]emitting complexes , beyond cloud - f and to the east of cloud - g which is itself located about @xmath27 away from the central engine .
it should be noted as well that the mid - ir cloud(a ) is located very close to an optical emission knot early identified in the literature as the `` northeast knot '' in the direction pa @xmath6 35 and at a distance about @xmath5 from the central engine ( elvius 1978 for the first identification ) .
the extended mid - ir emission in the south - west quadrant has no counterpart in the hst [ oiii]emitting cloud map , as expected from the heavy obscuration on that side of the agn .
therefore , it is conspicuous that the extended mid - ir emission in the north - east quadrant arises aside the nlr ( as featured by the [ oiii]emitting clouds ) and from material which is not directly exposed to the central ionizing source .
however , the tight correlation existing between the mid - ir extended emission and the 4.9 ghz emission is a result which signals the importance of the radio jet - like emission impacting on interstellar material located above and below the equatorial plane of the dusty / molecular torus .
the emission core observed at 11.2 and 20.5 @xmath0 m shows a noticeable north - south extension , about 100 pc , while it remains unresolved ( @xmath32 50 pc ) along the east - west direction . in the model devised by granato et al ( 1997 ) of the agn in ngc1068 ,
a 100 pc torus and a viewing angle of 65 were the torus parameters finally selected to match the sed of the agn .
this model predicts indeed that maps of the torus emission in the mid - ir should be elongated perpendicularly to the torus plane , in this case roughly along the north - south direction , and on a scale of +
/ @xmath33 .
one should notice however that the model maps have been obtained after convolution with a psf fwhm @xmath34 , too narrow compared with the effective psf of the data presented in this paper .
yet , the predicted direction of the elongation is consistent with the observed one at 11.2 and 20.5 @xmath0 m and its predicted size has the right order of magnitude , if compared to the measured one .
+ comparing the core at 11.2 and 20.5 @xmath0 m to high resolution maps at 4.8 , 3.5 and 2.2 @xmath0 m ( marco & alloin 2000 , rouan et al 1998 ) is more difficult because of the difference in spatial resolution . indeed , the extensions ( both polar and equatorial ) seen at 4.8 , 3.5 and 2.2 @xmath0 m are entirely enclosed within the 11.2 and 20.5 @xmath0 m core size . yet , if the pa @xmath6 102 equatorial extension detected at 4.8 , 3.5 and 2.2 @xmath0 m features the torus itself , one might expect to see emission in the mid - ir from cool dust located further away than the warm dust emitting at 4.8 @xmath0 m , still along ( pa @xmath6 102 ) . such emission is not detected .
one possible interpretation is that the equatorial extension at 4.8 , 3.5 and 2.2 @xmath0 m does not directly feature the torus .
this indeed would be surprising , because the covering factor of the equatorial extension to the central source is smaller than expected on the basis of several arguments ( opening angle of the ionizing cone , ratio between ir and primary uv radiation ... ) .
the extended equatorial emission at 4.8 , 3.5 and 2.2 @xmath0 m could instead trace only the equatorial plane of the torus rather than the torus itself ( possibly smaller ) and outline the merging of the torus with the host galaxy disc , in regions of high density where star formation is occurring and provides an additional and local source of heating .
this would explain the presence of dust at a temperature much higher than that predicted by torus models which take only into account the heating from the central engine .
the dust emission in these regions located in the torus equatorial plane but `` around - the - torus '' could then be more prominent at 4.8 than at 11.2 and 20.5 @xmath0 m .
in fact there is a whole new area to be explored at the transition between the molecular / dusty torus and its environment .
the comparison of maps in the radio , mid - ir , near - ir and optical / uv can potentially bring clues about the origin of the mid - ir extended emission beyond the torus itself : extra dust components , location , source of heating , etc .
unfortunately , one can not yet confront in detail the observed maps with predicted model - maps for the following reason : existing agn models are targeted at representing the molecular / dusty torus itself rather than the full molecular / dusty environment of an agn and therefore do not take into account the presence of the nlr region or more generally the distribution of matter away from the torus itself .
for example , thermal emission from warm dust possibly surviving on the back and uv protected side of nlr clouds is not considered , the impact of the radio jets on intervening material such as the nlr clouds or massive molecular clouds away from the equatorial plane of the torus are not taken into account either .
on the contrary , current results on the extended mid - ir emission in ngc1068 are telling us that material is present and heated around and away from the torus itself , up to a distance of @xmath4 300 pc .
assuming silicate grains with a power law ( -3.5 exponent ) size distribution over the size range 0.01 to 0.1 @xmath0 m , the 11.2 to 20.5 @xmath0 m flux ratio observed in cloud(a ) and cloud(c ) implies temperatures of around 150 k. cloud(a ) is already more than 200 pc away from the central engine and on the edge of the ionizing cone : how are the grains heated up to this temperature ?
what is the role played by single photon transient heating of small grains ? by the radio jet - like structure which appears to coincide so closely with the mid - ir extended emission ?
these questions deserve a more complete analysis and a quantitative modeling which is deferred to a specific and later work .
mid - ir imaging at high angular resolution offers potential advantages in the study of agn environment because this wavelength range is specific of warm / cool dust emission ( and possibly synchrotron emission from electrons ) and because extinction is reduced .
the diffraction - limited images ( resolution ( @xmath2 ) ) presented in this work highlight the presence of a prominent core emitting about 95% of the total flux in the mid - ir , as well as of extended emission , up to @xmath5 to the south - west ( pa = 210 ) and @xmath5 to the north - east ( pa = 35 ) , broken into patchy components which are particularly conspicuous at 20.5 @xmath0 m and can be singled - out as individual clouds . the central core shows an unresolved east - west fwhm of @xmath2 and a north - south fwhm of @xmath3 corresponding to a resolved full size extension of @xmath4 100 pc .
the north - south elongation of the emission core agrees with predicted maps of the mid - ir emission from a 100 pc dusty / molecular torus surrounding the central engine in ngc1068 and observed under an inclination angle of around 65 . as a result of smaller optical depth
, the extended emission in the north - east and south - west quadrants is more prominent at 20.5 than at 11.2 @xmath0 m .
the extended emission follows roughly the direction of the radio - jet and radio - lobe structures . in the north - east quadrant
, the mid - ir emission is located at the eastern edge of the ionizing cone outlined by the hst [ oiii]emitting clouds . interpreting the complete molecular / dusty environment of the agn , both in the torus and away from it , pleas for the development of three - dimensional complex modeling .
high resolution imaging is the first step in disentangling the various components : the new generation of 8m10 m class telescopes provides a resolution of @xmath35 in the mid - ir .
subsequent integral - field spectroscopy with such a spatial resolution and interferometry will also constitute invaluable tools to resolve this type of problem .
we are gratefully indebted to r.jouan and m.lortholary for their efficient assistance with the camiras instrument , as well as to the staff at both eso / la silla and cfht / hawaii for their support during the observing runs .
we acknowledge careful comments from the referee r.antonucci . | we present diffraction - limited ir images at 11.2 and 20.5 @xmath0 m of the central @xmath1 region in ngc1068 , collected with the camiras instrument mounted at the f/36 ir focus of the cfht / hawaii 3.6 m telescope and at the f/35 ir focus of the eso / la silla 3.6 m telescope , respectively .
after deconvolution , the achieved resolution ( @xmath2 ) reveals a prominent central core emitting about 95% of the total flux at these wavelengths , as well as extended emission , to the south - west ( pa = 210 ) and to the north - east ( pa = 35 ) , broken into patchy components which are particularly conspicuous at 20.5 @xmath0 m and can be isolated as individual clouds .
the central core shows an east - west fwhm of @xmath2 ( hence unresolved ) and a north - south fwhm of @xmath3 corresponding to a resolved full size extension of @xmath4 100 pc .
such an elongated shape is in agreement with model predictions of a dusty / molecular torus surrounding the central engine in ngc1068 , observed under an inclination angle of around 65 . considering that the core at 11.2 and 20.5 @xmath0 m is coincident with the core also seen at 2.2 , 3.5 and 4.8 @xmath0 m and that this features the location of the central engine ( also the radio source s1 ) , the extended mid - ir emission is found to follow closely the radio jet like structure in both the north - east and south - west quadrants . in the north - east quadrant ,
we observe that the mid - ir emission arises predominantly on the eastern side of the ionizing cone defined by the hst nlr [ oiii]emitting clouds and is still quite prominent at the wide base of the so - called northeast radio lobe , @xmath5 away from the central engine .
a detailed comparison of the extended mid - ir emission with model predictions requires that future agn modeling includes both a molecular / dusty torus and a distribution of material away from the equatorial plane of the torus , i.e. in and around the nlr . |
the high - statistics neutrino observation from the next nearby supernova ( sn ) will provide a bonanza of information about the astrophysics of core - collapse phenomena and neutrino properties . in a recent physical review letter
a strong case was made for the importance of coincidence measurements between gravitational wave and neutrino signals from sn core bounce @xcite .
the largest existing sn neutrino detectors are super - kamiokande and icecube , reaching to a distance of about 100 kpc .
the expected distribution of galactic sne drops off quickly beyond about 20 kpc @xcite . at this `` pessimistic '' distance , super - kamiokande
can time the bounce to within a few tens of milliseconds , an interval comparable to the expected duration of the gravitational wave burst @xcite .
we here extend this study to icecube , a high - statistics sn neutrino detector that would have seen the sn 1987a neutrino signal with @xmath1 significance .
for galactic sne , the large rate of uncorrelated cherenkov photons provides excellent time - structure information . following ref .
@xcite we note that neutrino masses are small enough to neglect time - of - flight effects . recalling that the diameter of the earth is 42 ms , millisecond - coincidence measurements between detectors at different geographic locations requires determining the sn direction either by astronomical observations or by the electron - scattering signal in super - kamiokande .
when sn neutrinos stream through water or ice , cherenkov light is generated , primarily by the secondary positrons from inverse beta decay @xmath2 . while fine - grained detectors reconstruct individual neutrinos on an event - by - event basis , icecube only picks up the average cherenkov glow of the ice . to estimate the detection rate we follow ref .
@xcite , augmented with the latest icecube efficiencies @xcite .
the complete detector will have 4800 optical modules ( oms ) and the data are read out in 1.6384 ms bins , implying a total event rate of @xmath3 where @xmath4 , @xmath5 and @xmath6 .
the peak luminosity reaches @xmath75 and at that time @xmath8 mev . for a thermal or slightly pinched spectrum , @xmath9
altogether , we expect @xmath10 as a typical peak event rate for a sn at 10 kpc .
this signal is to be compared with a background of 280 s@xmath11 in each om @xcite , corresponding for 4800 oms to @xmath12 with an rms fluctuation of @xmath13 .
therefore , the sn - induced `` correlated noise '' in the entire detector is highly significant @xcite . from any one neutrino interacting in the ice , at most one cherenkov photon
is picked up : the signal counts are entirely uncorrelated .
the prompt @xmath14 burst immediately after bounce produces a peak rate of about @xmath15 , lasting for several ms .
its impact on the early count rate depends sensitively on the flavor oscillation scenario ( see below ) .
sn models suggest that neutrino emission for the first 20 ms after core bounce depends little on model assumptions or input physics @xcite , although beyond this early phase the accretion rate and therefore neutrino emission depends strongly , for example , on the progenitor mass profile .
the prompt @xmath14 burst reaches its peak at 57 ms post bounce , which also marks the onset of @xmath16 emission that is initially suppressed by the large @xmath14 chemical potential .
up to about 20 ms post bounce @xmath17 rises roughly linearly .
we thus represent the early icecube signal as @xcite @xmath18 with @xmath19 ms , @xmath20 ms and @xmath21 .
these parameters also provide an excellent fit to the first 100 ms of a numerical model from the garching group @xcite that is available to us .
we may compare these assumptions with the early - phase models of ref .
@xmath17 rises nearly linearly to @xmath222 within 10 ms .
the evolution of @xmath23 is also shown , a common quantity in sn physics that characterizes , for example , the efficiency of energy deposition ; the icecube rate is proportional to @xmath24 . at 10 ms after onset , @xmath25 reaches 15 mev , implying @xmath26 .
we thus estimate 10 ms after onset a rate of 280@xmath27 , to be compared with @xmath28 from eq .
( [ eq : risemodel ] ) . therefore , our assumed signal rise is on the conservative side .
of course , the early models do not fix @xmath29 and @xmath30 separately ; the crucial parameters are @xmath31 and @xmath32 . the maximum rate that is reached long after bounce is not relevant for determining the onset of the signal .
if flavor oscillations swap the @xmath16 flux with @xmath33 ( some combination of @xmath34 and @xmath35 ) , the rise begins earlier because the large @xmath14 chemical potential during the prompt @xmath14 burst does not suppress the early emission of @xmath33 @xcite .
moreover , the rise time is faster , @xmath36 larger , and the maximum luminosity smaller .
we use eq .
( [ eq : risemodel ] ) also for @xmath37 with @xmath38 , @xmath39 ms , and @xmath40 .
flavor oscillations are unavoidable and have been studied , for early neutrino emission , in ref .
@xcite . assuming the normal mass hierarchy , @xmath41 , no collective oscillations , and a direct observation without earth effects , table i of ref .
@xcite reveals that the @xmath14 burst would be completely swapped and thus nearly invisible because the @xmath42 elastic scattering cross section is much smaller than that of @xmath14 .
the survival probability of @xmath16 would be @xmath43 with @xmath44 the `` solar '' mixing angle . therefore
, the effective detection rate would be @xmath45 .
we use this case as our main example .
a typical monte carlo realization of the icecube signal for our example is shown in fig .
[ fig : simulation ] . one can determine the signal onset @xmath46 within a few ms by naked eye .
for a sn closer than our standard distance of 10 kpc , one can follow details of the neutrino light curve without any fit . ]
one can not separate the @xmath16 and @xmath33 components for the example of fig . [
fig : simulation ] .
therefore , we reconstruct a fit with a single component of the form eq .
( [ eq : risemodel ] ) , assuming the zero - signal background is well known and not fitted here . using a time interval until 100 ms post bounce ,
we reconstruct @xmath47 ms ( @xmath48 ) .
if we use only data until 33 ms post bounce we find @xmath49 ms . indeed , if one fits eq .
( [ eq : risemodel ] ) on an interval that ends long before the plateau is reached , we effectively fit a second order polynomial with a positive slope and negative second derivative at @xmath31 , whereas the plateau itself is poorly fitted and its assumed value plays little role . depending on the distance of the sn one will fit more or fewer details of the overall neutrino light curve and there may be more efficient estimators for @xmath31 .
our example only provides a rough impression of what icecube can do .
the reconstruction uncertainty of @xmath46 scales approximately with neutrino flux , i.e. , with sn distance squared .
the number of excess events above background marking the onset of the signal has to be compared with the background fluctuations .
therefore , a significant number of excess events above background requires a longer integration period if the flux is smaller , explaining this scaling behavior .
the interpretation of @xmath46 relative to the true bounce time depends on the flavor oscillation scenario realized in nature .
this is influenced by many factors : the value of @xmath50 , the mass ordering , the role of collective oscillation effects , and the distance traveled in the earth .
combining the signal from different detectors , using future laboratory information on neutrino parameters , and perhaps the very coincidence with a gravitational - wave signal may allow one to disentangle some of these features .
however , as a first rough estimate it is sufficient to say that the reconstructed @xmath46 tends to be systematically delayed relative to the bounce time by no more than a few ms .
the statistical uncertainty of the @xmath46 reconstruction does not depend strongly on the oscillation scenario .
the authors of ref . @xcite used a sn emission model based on a two - component fit of the sparse sn 1987a data and described the overall neutrino signal in terms of several parameters which they say , after eq .
( 5 ) , are at odds with theoretical expectations . however ,
their parameter @xmath51 mev , describing the temperature of the neutrino - emitting gas during the accretion phase , gives @xmath52 mev ( see paragraph after eq . 15 in the published version of ref .
@xcite ) and thus is virtually identical to the corresponding @xmath25 from ref .
@xcite at 20 ms post bounce .
in other words , while the sn 1987a implied @xmath16 energies of ref .
@xcite are lower than theoretical expectations for the overall accretion phase , in the absence of flavor oscillations they agree nicely with the models of ref .
@xcite for the crucial first 20 ms . however , the chosen rise time @xmath53 ms is very long compared with the early - time models of ref .
@xcite . in ref .
@xcite the early neutrino signal after bounce was systematically studied for different input assumptions ( progenitor mass , equation of state , neutrino opacities ) , leading to very similar results . on the other hand
, one finds significantly different numerical examples in the literature . in ref .
@xcite the @xmath16 luminosity rises to @xmath54 after as much as 50 ms . in ref .
@xcite , a peak value of only @xmath55 is reached and the rise within 1020 ms after bounce is small . in ref .
@xcite the @xmath16 signal begins rising as late as 12 ms after the maximum of the @xmath14 burst .
there are many differences between these and other models in terms of numerical approach and input physics .
it would be extremely useful if another group would investigate the early neutrino signal in the spirit of ref .
@xcite for a range of physical assumptions and with attention to numerical details that may influence the early - time behavior . in view of the large range of possible distances to the next nearby sn and concomitant flux differences
, these uncertainties do not change our overall conclusions . in our fiducial example
icecube can reconstruct the signal onset within @xmath567 ms at @xmath48 cl for a sn at 20 kpc , comparable to what ref .
@xcite found for super - kamiokande .
ideally , of course , one would combine the measurements from several detectors .
a gravitational wave measurement of the core bounce in coincidence with neutrino onset would be of obvious astrophysical importance .
in addition one could test the weak equivalence principle .
both neutrinos and gravitational waves should suffer the same shapiro time delay in the gravitational potential of the galaxy .
for sn 1987a in the large magellanic cloud , this delay was a few months .
the coincidence of the neutrino burst with the rise of the light curve within a few hours proved an equal shapiro delay for photons and neutrinos to within about @xmath57 @xcite . a millisecond - scale coincidence between neutrinos and gravitational waves would extend and refine this test , in detail depending on the location of the sn . if the sn can not be located by astronomical means because of obscuration
, the electron - recoil signal in super - kamiokande or a future megatonne water cherenkov detector is the method of choice @xcite , whereas arrival - time triangulation was dismissed . with icecube almost complete , the situation has changed and triangulation can play a useful role after all , at least for a not - too - distant sn .
moreover , in europe a big detector of the laguna - class @xcite may become available ( representing one of three possible large - scale detector types ) and a megatonne water - cherenkov detector may be built in north america .
arrival - time triangulation between large - scale neutrino detectors on different continents could become viable .
it would be a worthwhile exercise to study the possibilities of bounce timing , based on common astrophysical assumptions , for different detector types and what can be learnt from a combined analysis . to this end
it would be worthwhile if groups other than the garching one would systematically study the numerical early - time neutrino signal to judge if indeed the dependence on physical assumptions is as small as found in ref .
@xcite .
we thank b. dasgupta , h .-
janka and a. mirizzi for discussions .
this research was supported in part by the u.s .
national science foundation under grants no .
opp-0236449 and phy-0354776 and by the alexander von humboldt foundation in germany ( f.h . ) and by the deutsche forschungsgemeinschaft under grant tr-27 `` neutrinos and beyond '' and the cluster of excellence `` origin and structure of the universe '' ( g.r . ) .
g. pagliaroli , f. vissani , e. coccia and w. fulgione , `` neutrinos from supernovae as a trigger for gravitational wave search , '' phys .
lett . * 103 * , 031102 ( 2009 ) [ arxiv:0903.1191 ] .
a. mirizzi , g. g. raffelt and p. d. serpico , `` earth matter effects in supernova neutrinos : optimal detector locations , '' jcap * 0605 * , 012 ( 2006 ) [ astro - ph/0604300 ] .
t. kowarik , t. griesel and a. pigsa ( for the icecube collaboration ) `` supernova search with the amanda / icecube detectors , '' proc .
31st icrc , d , poland ( 2009 ) [ arxiv:0908.0441 ] . c. pryor , c. e. roos and m. s. webster , `` detecting thermal neutrinos from supernovae with dumand , '' astrophys . j. * 329 * , 335 ( 1988 ) .
f. halzen , j. e. jacobsen and e. zas , `` ultra - transparent antarctic ice as a supernova detector , '' phys .
d * 53 * , 7359 ( 1996 ) [ astro - ph/9512080 ] .
a. s. dighe , m. t. keil and g. g. raffelt , `` detecting the neutrino mass hierarchy with a supernova at icecube , '' jcap * 0306 * , 005 ( 2003 ) [ hep - ph/0303210 ] .
m. kachelriess , r. toms , r. buras , h .-
janka , a. marek and m. rampp , `` exploiting the neutronization burst of a galactic supernova , '' phys .
d * 71 * , 063003 ( 2005 ) [ astro - ph/0412082 ] .
a. marek , h. t. janka and e. mller , `` equation - of - state dependent features in shock - oscillation modulated neutrino and gravitational - wave signals from supernovae , '' astron .
astrophys . * 496 * , 475 ( 2009 ) [ arxiv:0808.4136 ] .
b. dasgupta , a. dighe , g. g. raffelt and a. y. smirnov , `` multiple spectral splits of supernova neutrinos , '' phys .
rev . lett .
* 103 * , 051105 ( 2009 ) [ arxiv:0904.3542 ] .
h. duan , g. m. fuller , j. carlson and y. z. qian , `` flavor evolution of the neutronization neutrino burst from an o - ne - mg core - collapse supernova , '' phys .
lett . * 100 * , 021101 ( 2008 ) [ arxiv:0710.1271 ] .
b. dasgupta , a. dighe , a. mirizzi and g. g. raffelt , `` spectral split in prompt supernova neutrino burst : analytic three - flavor treatment , '' phys .
d * 77 * , 113007 ( 2008 ) [ arxiv:0801.1660 ] .
g. pagliaroli , f. vissani , m. l. costantini and a. ianni , `` improved analysis of sn1987a antineutrino events , '' astropart .
* 31 * , 163 ( 2009 ) [ arxiv:0810.0466 ] .
a. mezzacappa , m. liebendrfer , o. e. b. messer , w. r. hix , f. k. thielemann and s. w. bruenn , `` simulation of the spherically symmetric stellar core collapse , bounce , and postbounce evolution of a star of 13 solar masses with boltzmann neutrino transport , and its implications for the supernova mechanism , '' phys .
* 86 * , 1935 ( 2001 ) [ astro - ph/0005366 ] . c. l. fryer and p. a. young , `` late - time convection in the collapse of a 23 solar mass star , '' astrophys . j. * 659 * , 1438 ( 2007 ) [ astro - ph/0612154 ] .
m. herant , w. benz , w. r. hix , c. l. fryer and s. a. colgate , `` inside the supernova : a powerful convective engine , '' astrophys . j. * 435 * , 339 ( 1994 ) [ astro - ph/9404024 ] .
m. j. longo , `` new precision tests of the einstein equivalence principle from sn 1987a '' phys .
* 60 * , 173 ( 1988 ) . l. m. krauss and s. tremaine , `` test of the weak equivalence principle for neutrinos and photons , '' phys .
lett . * 60 * , 176 ( 1988 ) .
j. f. beacom and p. vogel , `` can a supernova be located by its neutrinos ?
, '' phys .
d * 60 * , 033007 ( 1999 ) [ astro - ph/9811350 ] .
r. toms , d. semikoz , g. g. raffelt , m. kachelriess and a. s. dighe , `` supernova pointing with low- and high - energy neutrino detectors , '' phys .
d * 68 * , 093013 ( 2003 ) [ hep - ph/0307050 ] .
d. autiero _ et al .
_ , `` large underground , liquid based detectors for astro - particle physics in europe : scientific case and prospects , '' jcap * 0711 * , 011 ( 2007 ) [ arxiv:0705.0116 ] . | generic model predictions for the early neutrino signal of a core - collapse supernova ( sn ) imply that icecube can reconstruct the bounce to within about @xmath0 ms at 95% cl ( assumed sn distance 10 kpc ) , relevant for coincidence with gravitational - wave detectors .
the timing uncertainty scales approximately with distance - squared .
the offset between true and reconstructed bounce time of up to several ms depends on the neutrino flavor oscillation scenario .
our work extends the recent study of pagliaroli et al .
[ phys .
rev .
lett .
103 , 031102 ( 2009 ) ] and demonstrates icecube s superb timing capabilities for neutrinos from the next nearby sn . |
the launch of space photometric missions , such as corot , most , or _ kepler _ , has revolutionized many branches of stellar astrophysics . extremely precise and nearly continuous photometric observations led to many discoveries and improvements in studies of various kinds of variable stars .
one of the fields that benefited the most are detached eclipsing binaries ( debs ) .
these are one of the most important objects for the astrophysics , as they allow to directly determine a number of fundamental stellar parameters , that are difficult or sometimes impossible to obtain otherwise .
the unprecedented photometric quality of _ kepler _ data has opened new possibilities in studies of debs , as it allows for better than before precision in derivation of light - curve - based parameters , like fractional radii or orbital inclination , and observations of various other phenomena , for example : differential rotation , activity cycles , pulsations , including tidally induced , solar - like oscillations , etc .
the _ kepler _ mission has revealed a zoo of objects , showing variety of characteristics .
the photometry alone , however , is not enough to accurately characterize a given deb .
one has to combine it with information obtained from spectra , optimally taken with high resolution .
this information includes for example : precise radial velocities ( rvs ) , monitored in time in order to observe their variability , or atmospheric parameters obtained from spectral analysis .
only such a complete set of data is useful for the purposes of modern stellar astrophysics .
studies of debs are important not only because of the information one can obtain for a particular object .
characterisation of a large number of targets is beneficial for statistical studies , from which one can draw conclusions about , e.g. , the amount of binaries and higher - order multiples , or mechanisms of their formation and evolution . for this reason , we have started a spectroscopic monitoring of the brightest _ kepler _ debs with the 1.88-m telescope of the okayama astrophysical observatory . for the purpose of precise characterization of stellar components , one should target those debs that are also double - lined spectroscopic binaries ( sb2s ) .
however , it is also possible derive many important information from single - lined ( sb1 ) objects . in this study
we present a sample of such systems , showing the variety they represent from primary components of exotic binaries , to multiples of different orders .
this work is organized as follows . in section [ sec_targets ]
we briefly present the target selection criteria for our whole program , and basic information about the systems presented here .
section [ sec_data ] presents the data used for this study .
section [ sec_analysis ] presents all the methods used for this work , including spectra reduction , derivation of rvs and eclipse timing measurements , modelling rv and light curves , and comparison with isochrones .
the results , are presented in section [ sec_results ] , and the last section contains a short summary .
[ cols="<,<,<,<,<,<,^,<,<",options="header " , ] | we present results of our spectroscopic observations of nine detached eclipsing binaries ( debs ) , selected from the _ kepler _ eclipsing binary catalog , that only show one set of spectral lines .
radial velocities ( rvs ) were calculated from the high resolution spectra obtained with the hides instrument , attached to the 1.88-m telescope at the okayama astrophysical observatory , and from the public apogee archive .
in our sample we found five single - lined binaries , with one component dominating the spectrum .
the orbital and light curve solutions were found for four of them , and compared with isochrones , in order to estimate absolute physical parameters and evolutionary status of the components .
for the fifth case we only update the orbital parameters , and estimate the properties of the unseen star .
two other systems show orbital motion with a period known from the eclipse timing variations ( etvs ) . for these we obtained parameters of outer orbits , by translating the etvs to rvs of the centre of mass of the eclipsing binary , and combining with the rvs of the outer star .
of the two remaining ones , one is most likely a blend of a faint background deb with a bright foreground star , which lines we see in the spectra , and the last case is possibly a quadruple bearing a sub - stellar mass object . where possible , we compare our results with literature , especially with results from asteroseismology .
we also report possible detections of solar - like oscillations in our rvs .
[ firstpage ] binaries : spectroscopic binaries : eclipsing stars : evolution stars : fundamental parameters stars : late - type stars : individual : kic 03120320 , kic 04758368 , kic 05598639 , kic 08430105 , kic 08718273 , kic 10001167 , kic 10015516 , kic 10614012 , kic 10991989 |
it has long been recognized that coalescences of binary systems of two black holes could be an important source of gravitational waves @xcite , both for the ground based interferometric detectors ligo @xcite and virgo @xcite currently under construction , and also for the possible future space - based interferometer lisa @xcite .
the orbits of binary black holes ( bbhs ) gradually decay from energy and angular momentum loss to gravitational radiation .
eventually , the holes coalesce to form a final black hole . for gravitational radiation reaction to successfully drive the binary to merge in less than a hubble time
, the initial orbital period must be @xmath24 , where @xmath25 is the total mass of the binary ; thus the critical orbital period is of order days for solar mass black holes , and of order years to hundreds of years for supermassive black holes ( @xmath26 ) .
the process of coalescence can be divided up into three more or less distinct phases : * an adiabatic _ inspiral _ , during which the gravitational radiation reaction timescale is much longer than the orbital period .
the inspiral ends when the binary orbit becomes relativistically dynamically unstable at an orbital separation of @xmath27 ( in units where @xmath28 ) @xcite .
the gravitational waves from the inspiral carry encoded within them the masses and spins of the two black holes , some of the orbital elements of the binary and the distance to the binary @xcite .
* towards the end of the inspiral , the black holes encounter the dynamical instability and make a gradual transition from a radiation - reaction driven inspiral to a freely - falling plunge @xcite , after which , even if the radiation reaction could be turned off , the black holes would still merge .
we will call the subsequent plunge and violent collision the _ merger _ phase .
gravitational waves from the merger could be rich with information about the dynamics of relativistic gravity in a highly nonlinear , highly dynamic regime about which we have a poor theoretical understanding today .
* as the final black hole settles down to a stationary kerr state , the nonlinear dynamics of the merger gradually become more and more describable as oscillations of this final black hole s
quasinormal modes @xcite .
the corresponding emitted gravitational waves consist of a superposition of exponentially damped sinusoids .
we will call the phase of the coalescence for which the emitted gravitational - wave signal is dominated by the strongest @xmath29 quasinormal mode signal the _ ringdown _ phase .
the waves from the ringdown carry information about the mass and spin of the final black hole @xcite .
( note that , for want of a better terminology , throughout this paper we consistently use _ coalescence _ to refer to the entire process of inspiral , merger and ringdown , and reserve the word merger for the phase intermediate between inspiral and ringdown . ) in this paper we focus primarily on bbhs in which the masses of the two black holes are approximately the same , although we do also consider sources where one bh is much smaller than the other .
we consider three different classes of bbhs : \(i ) _ solar mass _ black hole binaries , _
i.e. _ , binaries that are formed from massive main - sequence progenitor binary stellar systems .
such bbhs are expected to have total masses in the range @xmath30 , but not much larger than this .
the rate of coalescence of solar - mass bbhs in the universe is not very well known .
however , theory suggests that most of the bbh progenitor systems may not disrupt during the stellar collapses that produce the black holes , so that their coalescence rate could be about the same as the birth rate for their progenitors , about @xmath31 years in our galaxy , or several per year within a distance of @xmath32 @xcite .
note that this coalescence rate is roughly the same as the expected event rate for what has traditionally been regarded as the most promising source for ground based interferometers , coalescences of neutron star
neutron star ( ns - ns ) binaries @xcite
. the expected rate of ns - ns coalescences is more firmly known , however , since it is based on extrapolations from detected progenitor ns - ns systems @xcite .
\(ii ) _ intermediate mass _ black hole binaries , with total masses in the range @xmath33 .
in contrast to the cases of solar mass black holes and supermassive black holes ( discussed below ) , there is little direct observational evidence for the existence of black holes in this mass range . although there have been suggestions that the globular cluster m15 harbors a black hole of mass @xmath34 @xcite , theoretical modeling combined with the most recent hst observations neither confirm nor rule out this possibility @xcite . despite the lack of observational evidence , it is plausible that black holes in this mass range are formed in the cores of globular clusters , or in galactic nuclei in the process of formation of a supermassive black hole @xcite .
simulations by quinlan and shapiro suggest that black holes with @xmath35 could be formed in the evolution of dense stellar clusters of main sequence stars in galactic nuclei @xcite , and that coalescences of binaries of such black holes could be possible en route to the formation of a supermassive black hole .
we are prompted to consider these intermediate mass bbhs by the following consideration : even if the coalescence rate of intermediate mass bbhs is @xmath36 that of ns - ns binaries ( which is thought to be @xmath37 in our galaxy as discussed above ) , these sources would still be seen more often than ns - ns binaries by ligo s initial and advanced interferometers , and thus could be the first detected type of source .
( see sec .
[ results ] for further details . )
\(iii ) _ supermassive _ black holes binaries : there is a variety of strong circumstantial evidence that supermassive black holes ( smbhs ) in the mass range @xmath38 are present in quasars and active galactic nuclei , and also that @xmath39
@xmath40 of nearby massive spiral and elliptical galaxies harbor quiescent smbhs @xcite .
see ref .
@xcite for a review of this evidence .
one of the main scientific goals of the lisa project is to detect and monitor various processes involving smbhs , such as the capture of compact stars @xcite , and their formation @xcite . in particular , the coalescences of smbh binaries that are formed in galaxy mergers , in which the individual smbhs are driven together by dynamical friction and gas accretion until gravitational radiation reaction takes over @xcite ,
have often been suggested as a promising source for space - based interferometers @xcite .
such coalescences would be detectable throughout the observable universe with large signal to noise ratios @xcite .
there is also observational evidence for smbh binaries : wiggles in the radio jet of qso 1928 + 738 have been attributed to the orbital motion of a smbh binary @xcite , as have time variations in quasar luminosities @xcite and in emission line redshifts @xcite .
the overall event rate is uncertain , but could be large ( @xmath41 ) , especially if the hierarchical scenario for structure formation is correct @xcite .
detailed theoretical understanding and predictions of the gravitational waveforms @xmath42 and @xmath43 produced in bbh coalescences will facilitate both the detection of the gravitational - wave signal , and the extraction of its useful information . in situations where a complete family of theoretical template waveforms is available
, it will be possible to use the procedure of wiener optimal filtering to search the interferometer data streams and to detect the gravitational - wave signal @xcite .
the resulting signal - to - noise ratios ( snrs ) can be larger than those obtainable without theoretical templates by a substantial factor ; see sec .
[ derivesnrformulagen ] .
thus , while it is possible to detect the various phases of bbh coalescences without theoretical templates , such templates can greatly increase the effective range of the interferometers and the event detection rate .
( accurate theoretical templates are also essential for extracting the maximum amount of information from detected signals @xcite . )
such theoretical template waveforms are available for the inspiral and ringdown phases of the coalescence , but not yet for the merger phase , as we now discuss . for the inspiral phase of the coalescence
, the gravitational waves and orbital evolution can be described reasonably well using the post - newtonian approximation to general relativity . to date
, inspiral waveforms have been calculated up to post-2.5-newtonian order @xcite , and the prospects look good for obtaining waveforms up to post-3.5-newtonian order @xcite .
post - newtonian templates will be fairly accurate over most of the inspiral , the most important error being a cumulative phase lag @xcite .
this cumulative phase lag will not be important for searches for inspiral waves ; template phasing error will be largely compensated for by systematic errors in best - fit values of the binary s parameters , and the signals will still be found @xcite .
by contrast , template inaccuracies will be significant when one attempts to extract from the data the binary s parameters .
in particular , post - newtonian templates errors start to become very significant around an orbital separation of @xmath44 @xcite , well before the end of the inspiral at the dynamical orbital instability ( @xmath27 ) .
templates for the phase of the inspiral between roughly @xmath45 and @xmath46 will most likely have to be calculated using methods other than the post - newtonian approximation .
moreover , the methods of full blown numerical relativity can not be applied to this `` intermediate binary black hole '' ( ibbh ) phase , since the total time taken to evolve from @xmath45 to @xmath46 is about @xmath47 , too long for supercomputer simulations to even contemplate evolving .
alternative analytic and numerical methods for calculating gravitational waveforms from the ibbh portion of bbh inspirals , based on the adiabatic approximation , are under development @xcite ; it is likely that such alternative methods will be successfully developed and implemented before the gravitational - wave detectors begin their measurements @xcite .
waveforms from the dynamic , complicated merger phase of the coalescence can only be obtained from numerical relativity . unlike mergers of neutron star binaries ,
bbh mergers are particularly clean in the sense that there is no microphysics or hydrodynamics to complicate simulations of the evolution , and external perturbations are negligible : the entire merger can be described as a solution to the vacuum einstein equation @xcite .
finding that solution is not a particularly easy task : a major computational effort to evolve the vacuum einstein equation for bbh mergers using massive computational resources is currently underway , funded by the national science foundation s grand challenge program @xcite .
the ringdown phase of the coalescence can be accurately described using perturbation theory on the kerr spacetime background @xcite .
the gravitational waveforms from this phase are well understood , being just exponentially damped sinusoids .
thus , wiener optimal filtering is feasible for searches for ringdown waves .
the principal purpose of this paper is to estimate , in more detail than has been done previously , the prospects for measuring gravitational waves from the three different phases of coalescence events ( inspiral , merger and ringdown ) , for various different detectors , and for a wide range of bbh masses .
we estimate in each case the distances to which the different types of source can be seen by calculating expected snrs . in particular , we determine for each bbh mass and each detector whether a coalescence event is most effectively detected by searching for the inspiral portion of the signal , or the merger portion , or the ringdown portion .
we also determine how much the availability of theoretical template waveforms for the merger phase could increase the event detection rate .
previous estimates of snrs for ground - based interferometers have focused on the inspiral @xcite and ringdown @xcite phases , and also focused on solar - mass bbhs . for space - based interferometers , previous estimates of snrs from the merger phase @xcite
were restricted to specific mass values and did not consider the ringdown portion of the signal . in a companion paper
, we discuss in detail the useful information carried by the three phases of the gravitational - wave signal , and methods and prospects for extracting this information both with and without templates for the merger phase @xcite . in the following subsection
we describe our calculations and summarize the assumptions underlying our estimated signal - to - noise ratios . in sec .
[ results ] we summarize our main results and conclusions , and an outline of the paper is given in sec .
[ map ] .
we calculate snrs for three different types of interferometer : initial and advanced ground - based interferometers ( ligo / virgo ) , and the proposed space - based interferometer lisa .
the noise spectra of the initial and advanced ground - based interferometers we took from ref .
@xcite , and that for lisa we obtained from ref .
our approximate versions of these noise spectra are given in eqs .
( [ noisespec ] ) ( [ noisespeclisa ] ) , and are illustrated in figs .
[ hcharfig ]
[ hcharfig2 ] in sec .
[ egs ] .
we consider the following three different signal - detection methods : \(i ) _ matched filtering searches : _ for those phases of the coalescence for which a complete set of theoretical templates will be available ( the inspiral , the ringdown , and possibly the merger ) , the method of matched filtering or optimal filtering can be used to search for the waves @xcite .
the theory of matched filtering is briefly sketched in sec .
[ derivesnrformula1 ] . for any source of waves , the snr , @xmath48 , obtained from matched filtering
is related to the gravitational waveform @xmath49 measured by the interferometer and to the spectral density @xmath50 of the strain noise in the interferometer via @xcite @xmath51 where @xmath52 is the fourier transform of @xmath49 defined by eq .
( [ fourier ] ) .
the snr ( [ snr0 ] ) depends , through the waveform @xmath49 , on the orientation and position of the source relative to the interferometer . in sec .
[ derivesnrformula ] we show that if we perform an rms average over source orientations and positions ( at a fixed distance ) , the rms snr thus obtained depends only on the energy spectrum @xmath53 carried off from the source by the gravitational waves .
the resulting relationship between the waves energy spectrum and the rms angle - averaged snr forms the basis for most of our calculations .
it is given by [ _
cf_. eq . ( [ snraveraged ] ) ]
@xmath54 , \label{snraveraged0}\ ] ] where @xmath4 is the source s cosmological redshift and @xmath55 is the luminosity distance from the source . in order for a signal to be detected
, the waves measured snr must be larger than the detection threshold @xmath56 see , for example , ref .
@xcite and also sec .
[ nonlinear ] . here
@xmath57 is the number of independent starting times of the gravitational wave signal that are searched for in the data set , determined by the total duration of the data set ( of order one year ) and the sampling time .
the quantity @xmath58 is the total number of statistically independent waveform shapes in the set of signals that one is searching for @xcite .
\(ii ) _ band - pass filtering searches : _ for the merger phase , a complete set of theoretical templates may not be available .
it therefore may not be possible to use matched filtering , and other search methods will need to be employed .
band - pass filtering , followed by setting a detection threshold in the time domain , is a simple method of searching an interferometer data stream for bursts of unknown form @xcite . in sec .
[ derivesnrformula1 ] we derive an approximate relation between the snr obtainable from band - pass filtering , and the snr ( [ snr0 ] ) obtainable from matched filtering , for any burst of waves , namely @xmath59 here @xmath60 is the duration of the burst and @xmath61 is the bandwidth of the band - pass filter [ _ cf_. eq .
( [ simp1 ] ) ] .
the quantity @xmath62 is the dimension of the linear space of signals being searched for , which is roughly the same as the `` number of cycles '' of the gravitational waveform .
we use this formula in sec .
[ detectmerger ] to estimate band - pass - filter search snrs for the merger waves from bbh coalescences , by inserting on the right hand side the rms angle - averaged matched - filter search snr ( [ snraveraged0 ] ) , and by making estimates of @xmath60 and @xmath61 .
noise - monitoring , nonlinear filtering searches : _ the traditional view has been that the snr ( [ bpa ] ) is about the best that can be achieved in the absence of theoretical templates ; that is , that the gain in snr obtainable from matched filtering is approximately the square root of the number of cycles in the gravitational wave signal . this view is based on the the assumption that the search method used in the absence of templates is band - pass filtering or something very similar .
however , we suggest in sec .
[ nonlinear ] an alternative search method , motivated by bayesian analyses and incorporating nonlinear filtering , which performs much better than band - pass filtering and in some cases almost as well as matched filtering .
in essence , one monitors the noise level in the interferometer in a certain frequency band , over short timescales , and looks for statistically significant changes .
the noise level is estimated by calculating the quantity @xmath63 where @xmath60 is the maximum expected duration of the signal , and @xmath64 is a suitably pre - filtered version of the interferometer data stream .
the efficiency of this noise - monitoring search method can not usefully be described in terms of a signal - to - noise ratio , since the detection statistic is very non - gaussian . instead
, its efficiency can be described in the following way .
let @xmath48 denote the snr that would be obtained if matched filtering were possible [ eq . ( [ snr0 ] ) ] .
we use @xmath48 as a convenient parameterization of the signal strength ; as such , it is meaningful even in situations where matched filtering can not be carried out .
then , a signal will be detected with high confidence using the noise - monitoring technique whenever @xmath48 is larger than the threshold @xmath65 , where to a good approximation @xmath65 satisfies @xmath66 here @xmath67 is the dimension of the signal space discussed above ( or the number of independent frequency bins in the fourier domain ) .
the derivation of eq .
( [ nlthreshold ] ) is given in sec .
[ nonlinear ] .
the relation ( [ snraveraged0 ] ) forms the basis of our snr calculations .
we use the snr thresholds ( [ mfthreshold ] ) and ( [ nlthreshold ] ) to deduce from the snr values the detectability of the various parts of the gravitational - wave signal . to calculate the snrs
, we also need to specify the waves energy spectra for the three different phases of the coalescence .
as we now outline , for the inspiral and ringdown phases the waves energy spectrum is essentially known , while for the merger phase we make an educated guess of @xmath53 .
[ signalassumptions ] gives more details .
: we use the leading order expression for @xmath53 obtained using newtonian gravity supplemented with the quadrupole formula @xcite [ eq . ( [ dedfinspiral ] ) ] . strictly speaking
, this spectrum describes the snr that would be achieved by searching for newtonian , quadrupole waves using newtonian , quadrupole templates .
the actual snr obtained when searching for a real , general - relativistic inspiral waveform using post - newtonian templates should deviate from this by only a few tens of percent @xcite .
we terminate the spectrum at the frequency @xmath68 which is ( roughly ) the frequency of quadrupole waves emitted at the orbital dynamical instability at @xmath69 @xcite . for lisa ,
we assume that the measurement process lasts at most one year , and choose the frequency at which the inspiral spectrum starts accordingly . :
the spectrum that we use [ eq . ( [ dedfqnr ] ) ] is determined ( up to its overall amplitude ) by the characteristics of the @xmath29 quasi - normal ringing ( qnr ) mode of the final kerr black hole .
this mode is the most slowly damped of all qnr modes , so we expect it to dominate the last stages of the gravitational - wave emission .
the qnr spectrum depends on three parameters : the quasi - normal modes frequency @xmath70 and damping time @xmath71 , and the overall amplitude of the quasinormal mode signal .
equivalently , the three parameters can be taken to be the mass @xmath25 and dimensionless spin parameter @xmath72 of the final black hole ( which determine @xmath70 and @xmath71 ) and the total amount of energy radiated in the ringdown ( which determines the overall amplitude ) .
the spectrum is peaked at @xmath73 with width @xmath74 . in our analyses
, we ( somewhat arbitrarily ) assume that @xmath75 .
it seems likely that in many coalescences the spin of the final black hole will be close to maximal , since the total angular momentum of the binary at the end of the inspiral is @xmath76 when the individual black holes are non - spinning @xcite , and the individual black hole spins can add to this .
exactly how close to maximally spinning the final black hole will be is a matter that probably will not be decided until supercomputer simulations or observations settle the issue . in any case , the ringdown snr values that we obtain depend only weakly on our assumed value of @xmath72 [ _ cf_. eq . ( [ qnrsnrii ] ) ] , for fixed total energy radiated in the ringdown .
the overall amplitude of the ringdown signal depends upon one s delineation of where `` merger '' ends and `` ringdown '' begins , which is somewhat arbitrary . for equal - mass bbhs , we assume a value of the overall amplitude that corresponds to a total radiated energy in the ringdown of @xmath77a radiation efficiency of @xmath78 .
this number is based on a back - of - the - envelope , quadrupole - formula - based estimate of the qnr mode s amplitude when the distortion of the horizon of the black hole is of order unity ( _ cf_. sec .
[ ringdownphase ] ) .
although this radiation efficiency seems rather high , there have been numerical simulations of the evolution of distorted , spinning black holes in which the ringdown waves carry away @xmath79 of the black hole s total mass energy @xcite . for non equal - mass black holes , we assume that the total energy radiated is @xmath80 , where @xmath81 and @xmath82 is the reduced mass of the binary .
this function gives the correct results for the equal - mass case and also gives the correct scaling law in the regime @xmath83 ; for general mass ratios the scaling law is probably a good approximation . :
realistic merger energy spectra will vary substantially from event to event ( depending on the initial spins of the inspiraling black holes ) .
currently , we have very little concrete information about such spectra , pending supercomputer simulations of bbh mergers . in sec . [ mergerphase ]
we describe various circumstantial pieces of evidence , culled from the literature , relevant to merger spectrum .
based on that evidence , we adopt the following crude model for equal - mass bbhs : we assume a flat spectrum @xmath84 extending from the frequency @xmath85 of quadrupole waves at the end of inspiral to the quasinormal ringing frequency @xmath86 , with amplitude such that the total radiated energy in the merger is @xmath12 of the total mass energy of the spacetime [ eq . ( [ dedfmerger ] ) ] .
we outline in sec .
[ mergerphase ] two different `` handwaving '' arguments which suggest that in favorable cases the merger radiation efficiency may be as high as our assumed value of @xmath87 .
one of these arguments , due originally to smarr @xcite and explored by detweiler @xcite , is based on extrapolation of perturbation theory results ; the other argument is based on angular momentum conservation . our assumed radiation efficiencies of @xmath78 and @xmath12 for the ringdown and merger phases should perhaps be interpreted as reasonable upper bounds that could be achieved in favorable cases , rather than as best - guess estimates .
we note that numerical simulations that have been performed to date ( which are restricted to axisymmetric situations ) generally yield lower radiation efficiencies than we have assumed @xcite ; moreover , these axisymmetric simulations generally find that ringdown waves carry most of the radiated energy . in sec .
[ mergerphase ] we argue that the radiated energy in the merger phase could be boosted by the lack of symmetry in generic black hole mergers , and especially by the individual black holes spins ( if these spins are large ) . for non - equal mass bbhs
, we again simply reduce the energy spectrum by the factor @xmath81 , while the upper and lower frequencies @xmath88 and @xmath70 are taken to be independent of @xmath82 . by inserting our assumed energy spectra ( [ dedfinspiral ] ) , ( [ dedfmerger ] ) and ( [ dedfqnr ] ) into eq . ( [ snraveraged0 ] )
, we obtain optimal - filtering snrs for the three different phases of bbh coalescences as a function of the redshifted total mass @xmath89 of the binary .
the results are shown in figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] and formulae summarizing the results are given in appendix [ appsnr ] . also in sec .
[ mergernofilters ] we estimate that , for the merger waves , the number of independent frequency bins @xmath90 which characterize the signal falls roughly in the range @xmath91 , and that a conservative upper - bound is @xmath92 .
we use this upper bound in sec .
[ detectmerger ] to estimate the snr threshold ( [ nlthreshold ] ) for merger waves using noise - monitoring searches when templates are unavailable .
we discuss the implications of these snrs and snr thresholds in sec .
[ detection ] ; here we summarize our main conclusions : * ground - based interferometers can study black - hole mergers in the mass range @xmath0 to @xmath1 ; lisa , by contrast can study mergers in the mass range @xmath93 .
* ground - based interferometers can do moderate snr ( a few tens ) , moderate accuracy studies of the dynamics of merging black holes .
lisa , by contrast , can do high snr ( a few @xmath94 ) , high - accuracy studies .
* coalescing black holes are fairly likely to be the first sources detected by the ligo / virgo network of ground - based interferometers : because of their larger masses , they can be seen out to much greater distances than coalescing neutron star binaries .
( with the initial ligo interferometers , binary black holes with @xmath6 can be seen out to @xmath5 , whereas binary neutron stars can be seen out to @xmath95 @xcite ) .
estimates of the number of bbh coalescences per ns - ns coalescence by experts in binary evolution range from @xmath96 @xcite to @xmath97 @xcite ; the distance gain factor for bbhs could easily compensate for the reduced birth rate . *
low - mass bbhs [ @xmath8 for the first ligo interferometers ; @xmath9 for the advanced ; @xmath10 for lisa ] are best searched for via their well - understood inspiral waves ; while more massive bbhs must be searched for via their more poorly understood merger waves and/or their well - understood ringdown waves .
* when the interferometers sensitivities has reached a level roughly half way between the first ligo interferometers and the advanced ligo interferometers @xcite , a search based on the inspiral waves should be capable of finding low - mass bbhs out to about @xmath98 . * a search for massive bbhs based on the ringdown waves
can be performed using matched filtering .
we show in sec .
[ qnrsearches ] that the number of independent templates needed for such a search is about @xmath13 or less , assuming that one wants the event rate reduction due to discreteness of the template family to be no more than @xmath12 . with the first ligo interferometers ,
such a search should be capable of finding equal - mass bbhs in the mass range @xmath99 to @xmath100 out to about @xmath32 ; with advanced ligo interferometers , bbhs with @xmath101 should be detectable out to @xmath19 ; and with lisa , bbhs with @xmath102 should be visible out to @xmath21 .
these distances are reduced by a factor @xmath103 for non - equal - mass bbhs . *
the effectiveness of a search based on the merger waves will depend on how much one has learned about the merger waveforms from numerical relativity simulations . with only a knowledge of the merger waves range of frequency bands and range of temporal durations , a search based on merger waves
can be performed using the noise - monitoring search algorithm discussed above .
such a search should increase the number of discovered bbhs by a factor @xmath104 over those found from the inspiral and ringdown waves . on the other hand , a full set of merger templates based on numerical relativity simulations
could further increase the number of discovered bbhs by an additional factor of up to @xmath105 .
the organization of this paper is as follows . in sec .
[ derivesnrformulagen ] we discuss in detail the three methods of searching for gravitational wave signals referred to above . in sec .
[ derivesnrformula ] we derive the general formula [ eq .
( [ snraveraged ] ) ] discussed above for the optimally - filtered snr of a gravitational - wave source at a cosmological distance , in terms of the source s emitted spectrum , redshift and luminosity distance , and the detector s noise spectrum .
this formula is one underpinning for the bbh snr computations of appendix [ appsnr ] and sec .
[ snrsection ] . in sec .
[ derivesnrformula1 ] we derive the approximate formula [ eq .
( [ simp1 ] ) ] discussed above for the snr obtainable by band - pass filtering in the absence of theoretical templates . in sec .
[ nonlinear ] we describe our suggested noise - monitoring search , and derive the relevant detection threshold [ eqs .
( [ e3 ] ) and ( [ e4 ] ) ] . in sec .
[ threelittlepigs ] we split the waveform from a coalescing equal - mass black hole binary into three successive epochs : inspiral , merger and coalescence .
then we estimate the frequency bands and spectra for each of these epochs : the merger phase in sec .
[ mergerphase ] [ eqs .
( [ mergefreq ] ) , ( [ fqnrdef ] ) , ( [ dedfmerger ] ) ] ; the inspiral phase in sec . [ inspiralphase ] [ eq . ( [ dedfinspiral ] ) ] ; and the ringdown phase in sec . [ ringdownphase ] [ eqs.([dedfqnr ] ) ( [ qnrenergytot ] ) ] .
we have already summarized , above , the assumptions that underlie our assumed merger and ringdown spectra , and the uncertainties in those assumptions ; we give a more detailed description in secs . [ mergerphase ] [ ringdownphase ] . a subtlety in the derivation of the ringdown energy spectrum is explained in appendix [ ringdown3 ] . in sec .
[ mergernofilters ] we make rough estimates of the range of the bandwidths and temporal durations of merger waveforms . in sec .
[ noisecurves ] we devise a simple piece - wise power - law analytic fit [ eq .
( [ noisespec ] ) ] to the noise spectra of an initial ligo interferometer , an advanced ligo interferometer , and a space - based lisa interferometer .
this single formula , by adjustment of its parameters , can describe all three interferometer types . in appendix [ appsnr ] , we insert the interferometer noise spectra from sec . [ noisecurves ] , and the estimated bbh inspiral , merger , and ringdown spectra from secs . [ mergerphase ] , [ inspiralphase ] and [ ringdownphase ] , into the general snr formula ( [ snraveraged ] ) .
that formula s frequency integral is then performed to compute the angle - averaged , optimal - filtering snr for each type of interferometer and each phase of bbh coalescence .
the resulting snrs for an initial ligo interferometer , an advanced ligo interferometer , and a lisa interferometer , are expressed in terms of the binary s total mass @xmath25 , cosmological redshift @xmath4 , and luminosity distance @xmath106 . in sec .
[ generalresults ] , we graph the snr formulae derived in appendix [ appsnr ] ( fig .
[ initialligosnr ] for an initial ligo interferometer ; fig .
[ advancedligosnr ] for an advanced ligo interferometer ; fig .
[ lisasnr ] for lisa ) .
we give intuitive insight into these snrs in sec .
[ egs ] by re - expressing the power snr for a source as @xmath107 ^ 2,\ ] ] where @xmath108 is the source s `` characteristic amplitude '' as a function of frequency , and @xmath109 is the detector s rms noise in a bandwidth equal to frequency for sources with random orientations .
we give plots of @xmath108 and @xmath109 for five specific examples of binaries a @xmath110 binary at distance @xmath111 and a @xmath112 binary at cosmological redshift @xmath113 ( fig .
[ hcharfig ] ) ; a @xmath114 binary at cosmological redshift @xmath115 ( fig .
[ hcharfig1 ] ) ; a @xmath116 binary at @xmath117 and a @xmath118 binary at @xmath113 observed by lisa ( fig .
[ hcharfig2 ] ) . for each example , we explain , in terms of the plots , why the inspiral , merger , and ringdown snrs have the values shown in figs .
[ initialligosnr ] , [ advancedligosnr ] or [ lisasnr ] . the snr graphs in figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] are the foundation for our conclusions , summarized above , about what features of which binaries should be observable with which interferometers . a detailed discussion of these conclusions is given in sec . [ detection ] . in sec .
[ qnrsearches ] , we estimate the number of templates required for a search for ringdown waves based on matched filtering , estimate the snr detection thresholds , and hence the range of the various interferometers for ringdown waves . in sec .
[ detectmerger ] we examine the prospects for searches for bbhs via their merger waves , both with and without templates .
in this section we discuss in detail the various signal - search methods which were briefly described in the introduction .
we start in sec .
[ derivesnrformula1 ] by deriving the approximate relation ( [ bpa ] ) between the snr achievable using matched filtering searches for signals and the snr obtainable via band - pass filtering searches .
this approximate relation is closely related to the standard lore that the gain factor due to matched filtering is the square root of the number of cycles in the waveform @xcite , which strictly speaking is applicable only to waveforms that are almost monochromatic . for completeness , we discuss in sec .
[ derivesnrformula1 ] both the standard lore relation [ eq .
( [ snrgain ] ) ] and the approximate more general relation [ eq .
( [ simp1 ] ) ] . in sec .
[ nonlinear ] we describe our proposed noise - monitoring search method , and derive the detection threshold ( [ nlthreshold ] ) discussed in sec .
[ results ] . finally , in sec .
[ derivesnrformula ] we derive the general formula [ eq .
( [ snraveraged ] ) ] for the angle - averaged , optimal - filtering snr for a gravitational - wave source , which was discussed in the introduction .
consider the situation where some arbitrary gravitational - wave burst @xmath49 is present in the data stream @xmath64 , so that @xmath119 where @xmath120 is the noise .
if one integrates a filter @xmath121 against the data stream @xmath64 to produce a number , @xmath122 , then the standard definition of the signal to noise ratio is @xcite @xmath123 \over \sqrt { 4 \int_0^\infty df \ , \left| { \tilde k}(f ) \right|^2 s_h(f ) } } ; \label{snrgeneral}\end{aligned}\ ] ] see , _ e.g. _ , refs .
@xcite . here
tildes denote fourier transforms according to the convention @xmath124 and @xmath50 is the power spectral density of strain noise in the detector @xcite .
now , consider searching for a signal @xmath49 when the only information about it that one has is its approximate bandwidth in the frequency domain .
perhaps the simplest search algorithm one could use to search for the signal is to choose for @xmath121 the following band - pass filter : @xmath125 here @xmath126 is the step function and @xmath127 is the starting time of the filter .
this filter chops out all the data in the frequency domain except that in a bandwidth @xmath61 about a characteristic central frequency @xmath128 @xcite .
suppose that the frequency interval has been chosen wisely , so that the signal @xmath49 has negligible power outside the interval
. then @xmath52 can be taken to vanish outside the chosen bandwidth , and eqs .
( [ snrgeneral ] ) and ( [ bpfilter ] ) yield @xmath129 where @xmath130 is the rms fluctuation in the noise at frequency @xmath131 in a bandwidth equal to @xmath131 .
the starting time of the filter @xmath127 is then varied to give the maximum filter output @xmath132 , which is achieved at some value @xmath133 of @xmath127 . at this maximum overlap time
, the snr is given by eq .
( [ snrbp ] ) with @xmath127 replaced by @xmath134 . in particular , for broadband signals for which @xmath135 ( as opposed to @xmath136 ) , eq .
( [ snrbp ] ) simplifies to the standard result @xcite @xmath137 by contrast , if the shape of the signal is known one can use the well known optimal filter @xmath138 , for which the snr squared is , from eq .
( [ snrgeneral ] ) @xcite @xmath139 a crucial element of both optimal filtering searches and most especially band - pass filtering searches with ground - based interferometers is the use of coincidencing between different interferometers to circumvent the effects of non - gaussian noise bursts @xcite .
coincidencing between the 4 interferometers in the ligo / virgo network ( the hanford 2 km , hanford 4 km , livingston 4 km and pisa 3 km interferometers ) should be sufficient to achieve this . to be conservative , our assumed detection thresholds for the snr values ( which we discuss in sec .
[ detection ] ) are based on combining just the two ligo 4 km interferometers , albeit with assumed gaussian statistics .
the standard lore approximate relationship between the optimal - filtering snr ( [ snr ] ) and the band - pass filtering snr ( [ snrbp1 ] ) can be obtained as follows .
consider the special case of a waveform that is quasi - monochromatic , _
i.e. _ , of the form @xmath140 , \label{qm}\ ] ] where the amplitude @xmath141 and instantaneous frequency [ given by @xmath142 are slowly evolving . using eq .
( [ snr ] ) and the stationary phase approximation to the fourier transform of the signal ( [ qm ] ) yields @xmath143 ^ 2 \over h_{\rm rms}(f)^2 } , \label{snrgain}\ ] ] where @xmath144 is the number of cycles spent within a bandwidth @xmath145 centered on @xmath131 , and @xmath146 is the time at which the gravitational - wave frequency is @xmath131 . by comparing eqs .
( [ snrbp1 ] ) and ( [ snrgain ] ) it can be seen that @xmath147 is the gain factor in snr squared for optimal filtering over band - pass filtering , per logarithmic interval in frequency @xcite . for general signals
@xmath49 which are not quasi - monochromatic , eq .
( [ qm ] ) does not apply .
however , we can derive an approximate formula for the snr ( [ snrbp1 ] ) for general signals as follows .
approximating @xmath50 to be constant in eq .
( [ snr ] ) gives @xcite @xmath148 ^ 2 \nonumber \\
& \approx & 2 f_{\rm char } t \ , { { \bar h}^2 \over h_{\rm rms}(f_{\rm char})^2 } \label{snropt2}\end{aligned}\ ] ] where @xmath149 is a rms average of @xmath49 and @xmath60 is the effective duration of the signal . comparing eqs .
( [ snropt2 ] ) and ( [ snrbp ] ) we find that @xmath150 where the `` number of _ a priori _ frequency bins '' is @xmath151 the reason for this terminology is as follows .
suppose that we are trying to detect a signal whose total duration we know to be less than or equal to @xmath60 , and whose spectrum we know to lie in some frequency interval of bandwidth @xmath61 .
when we discard the measured data outside of these time and frequency intervals , the remaining relevant data is described by @xmath67 real fourier coefficients ( or equivalently , @xmath152 frequency bins ) . the quantity @xmath152 is the number of independent real variables that parameterize the space of signals for which we will search .
this notion of number of _ a priori _ frequency bins is closely related to the notion of number of cycles in the waveform .
the number of waveform cycles will be approximately @xmath153 , and so for a broadband burst for which @xmath154 we have @xmath155 .
there is a distinction between the two concepts , however : the number of cycles is intrinsic to the signal , whereas the number of frequency bins in part characterizes our _ a priori _ information or assumptions about the signal and not the signal itself .
this is because for band - pass filtering we must choose at the start some bandwidth @xmath61 .
since the true bandwidth of the signal will vary from one signal to another and will not be known in advance , the true bandwidth of the signal will in general be somewhat less than @xmath61 if we have chosen @xmath61 wisely .
the quantity @xmath152 depends not on the true bandwidth of the signal but on our assumed bandwidth @xmath61 .
the first factor on the right hand side of eq .
( [ simp ] ) is the ratio between the peak strain amplitude @xmath156 in the time domain and an rms value @xmath149 of this strain amplitude . by defining the effective duration @xmath60 of the signal to be given by @xmath157 ^ 2 = t \ , h(t_{\rm best})^2 , \label{tdef}\ ] ]
this factor reduces to unity . with this interpretation of @xmath60 in eq .
( [ nbinsdef ] ) , eq .
( [ simp ] ) reduces to the formula @xmath158 discussed in the introduction .
we use this formula in sec .
[ detectmerger ] . the snr threshold @xmath159 appropriate for matched filtering
is approximately given by @xcite @xmath160 which to a good approximation reduces to @xmath161 [ _ cf .
_ eq . ( [ mfthreshold ] ) where the factor of @xmath162 , which does not have a large effect , was omitted ] . here
the number @xmath163 is the number of statistically independent waveforms with snr @xmath164 in the set of signals to be searched for @xcite ; eq .
( [ mfthreshold1 ] ) must be solved self - consistently to determine @xmath159 . in this section
we describe our suggested noise - monitoring method for searching for gravitational wave bursts of unknown form ; more details can be found in ref .
in essence , the method consists of monitoring over short timescales the total rms noise in the detector output in the frequency band in which the signal is expected , and waiting for statistically significant changes in one s estimate of the noise power .
suppose that the maximum expected signal duration is @xmath60 , and that the interferometer output is @xmath64 .
we construct a quantity @xmath165 in the following way .
first , focus attention on the data stream @xmath166 in the time interval @xmath167 . because the data stream is in fact discrete and not continuous , this interval of data can be represented by the numbers
@xmath168 for @xmath169 , where @xmath170 is the sampling time . from eq .
( [ decompos1 ] ) we have @xmath171 where @xmath172 is the gravitational wave signal and @xmath173 is the noise .
now because the interferometer noise is colored , the noise matrix @xmath174 will not be diagonal . here
, angle brackets denote ensemble averaging over realizations of the noise .
if one performs an fft just of this finite stretch of data , the noise matrix on the new basis will not be diagonal either because of aliasing effects .
however , it is possible to diagonalize the matrix ( [ noisem ] ) and change to a basis on which the noise is diagonal .
we will denote this new basis by capital roman letters @xmath175 .
the data points @xmath176 on this new basis can be chosen to correspond approximately to frequencies @xmath177 , @xmath178 @xcite .
equation ( [ noisem ] ) can now be replaced by @xmath179 the data @xmath176 extend up to some high frequency ( of order several khz ) determined by the sampling time .
we next discard all data above some upper cutoff frequency .
( note that we have effectively performed band - pass filtering of the data , since the restriction to a segment of length @xmath60 in the time domain removes frequency components at @xmath180 . )
let us denote by @xmath152 the total number of data points remaining , which will be approximately given by @xmath181 where @xmath61 is the bandwidth of our effective band - pass filter . in terms of this basis , matched filtering consists of calculating , for each trial waveform shape @xmath182 , the quantity @xmath183 ( we are assuming here that all the trial waveform shapes have duration less than @xmath60 and in the frequency domain have most of their power within the bandwidth @xmath61 ) .
we introduce the notation @xmath184 ; then the optimal filtering snr ( [ snr0 ] ) is given by @xmath185 thus , the quantity @xmath186 is the optimal filtering signal - to - noise squared per data bin .
note that throughout this subsection , we use @xmath48 as a convenient parameterization of the signal strength , which is meaningful even in situations where templates are not available and where optimal filtering can not be carried out .
band - pass filtering ( of a pre - whitened data stream ) approximately corresponds in this language to calculating the statistic @xmath187 this will have an expected value of roughly @xmath188 [ _ cf . _
( [ simp1 ] ) ] if the signal is not sharply peaked in frequency but instead is spread out over the bandwidth @xmath61 .
the statistic used to diagnose the presence of a signal is @xmath189 this statistic is essentially an estimate of the apparent rms noise power in the given bandwidth over the given time interval , up to an additive constant .
the additive constant , @xmath190 , is chosen so that when no signal is present , @xmath191 and thus @xmath165 random walks over positive and negative values .
when a signal is present , @xmath165 will with high probability be large and positive .
one monitors @xmath165 as a function of time and sets a detection threshold such that when @xmath165 exceeds the threshold , the probability that the data stream contains no signal is very low ( see below ) .
note that this search method constitutes a type of nonlinear filtering .
the technique of monitoring the statistic ( [ qdef ] ) is closely related to two commonly used techniques in radio astronomy . in the first such technique
, observers sum the power from frequency bins which are expected to contain harmonics of the signal they are trying to detect .
this procedure is not quite as good as coherently combining the signal from all the frequency bins but is much easier , computationally .
the second technique @xcite is useful when one is looking for periodic signals in a data train that is too long to fourier transform .
one splits the data train into some number @xmath192 of shorter data segments , takes the fft of each shorter segment , and adds the ffts incoherently ( _ i.e._. , adds the individual power spectra ) .
again , this is not the optimal search method , but it is often a useful thing to do given finite computational resources .
one difference between our suggested noise - monitoring search method and the radio astronomy techniques is the following : one adds the frequency bins incoherently in the statistic ( [ qdef ] ) because the phase relationships are unknown ; in the radio astronomy contexts , one adds the frequency bins incoherently to save on computation time .
but the techniques themselves are conceptually very similar .
we now turn to a derivation of the efficiency and performance of our suggested noise - monitoring search method .
it is straightforward to show that when a signal is present @xmath193 and that @xmath194 ^ 2 \right > = 4 \rho^2 + 2 { \cal n}_{\rm bins}. \label{e2}\ ] ] when no signal is present , eqs .
( [ e1 ] ) and ( [ e2 ] ) continue to hold with @xmath195 .
these equations show that a signal should be detectable in the regime @xmath196 as well as at larger @xmath48 , since in the regime @xmath197 the expected value ( [ e1 ] ) of @xmath198 is large compared to its rms value in the absence of a signal . by contrast , a signal is detectable using band - pass filtering only in the regime @xmath199 , from eq .
( [ hh ] ) and associated discussion .
the noise - monitoring method thus extends the domain of detectable signals to include the regime ( [ domain ] ) . the approximate snr threshold predicted by eq .
( [ e2 ] ) is correct in order of magnitude , but to obtain an accurate snr threshold it is necessary to calculate the full probability distribution for the statistic @xmath198 , since its statistical properties are very non - gaussian .
this probability distribution is given by , from eqs .
( [ basicc ] ) , ( [ noisem1 ] ) and ( [ qdef ] ) , @xmath200 = { \gamma[{\cal n}_{\rm bins}/2,(q_0 + { \cal n}_{\rm bins})/2 ] \over \gamma({\cal n}_{\rm bins}/2 ) } \label{probq}\ ] ] where @xmath201 is the incomplete gamma function and @xmath202 is the usual gamma function .
suppose now that we examine a number @xmath57 of different starting times @xmath203 , and that we wish to find the number @xmath204 such that the probability ( [ probq ] ) of @xmath165 exceeding @xmath204 for any @xmath203 , in the absence of a signal , is some small number @xmath205 .
this threshold @xmath204 is obtained by solving @xmath206 \over \gamma({\cal n}_{\rm bins}/2 ) } = { \epsilon \over { \cal n}_{\rm start - times}}. \label{e3}\ ] ] from eq .
( [ e1 ] ) , this threshold will be exceeded by a signal whenever the signal strength ( [ snr00 ] ) satisfies @xmath207 equations ( [ e3 ] ) and ( [ e4 ] ) determine the threshold @xmath65 as a function of the parameters @xmath162 , @xmath208 , and @xmath152 ; we use these formulae in sec .
[ detectmerger ] .
we now discuss an approximate formulae for the threshold @xmath65 which gives an insight into the nature of the nonlinear filtering algorithm .
it can be shown @xcite that for @xmath209 , @xmath65 can be obtained to a good approximation by solving the equation @xmath210 [ _ cf . _ eq . ( [ nlthreshold ] )
discussed in the introduction ] .
it is possible to understand the origin of eq .
( [ nlthreshold1 ] ) in the following simple way .
the total number @xmath211 of waveform shapes of duration less than or equal to @xmath60 , with bandwidth @xmath61 and with snr @xmath212 which are distinguishable in the interferometer noise is approximately given by @xcite @xmath213 let us denote by @xmath214 the manifold of all possible waveforms satisfying these criteria duration less than or equal to @xmath60 , bandwidth @xmath61 and snr @xmath212 .
not all of the waveforms in this manifold correspond to physically interesting bbh merger waveforms ; the set @xmath215 of bbh merger waveforms will form a sub - manifold of @xmath214 .
consider now the hypothetical limit in which the sub - manifold @xmath215 becomes so large ( perhaps curving back and intersecting itself ) that , when smeared out by the interferometer noise , it effectively fills the entire manifold @xmath214 . in this limit , the number @xmath58 of bbh merger waveform shapes for which we are searching approaches @xmath211 .
it is clear that in this limit , knowledge of the waveform shapes is not useful in searching for the waves , and that a search method which simply seeks the maximum overlap between the measured signal and all possible waveform shapes ( thus not requiring templates ) would perform just as well as matched filtering .
this is essentially what the nonlinear search method does . using this insight ,
the threshold ( [ nlthreshold1 ] ) can be obtained simply by combining eqs .
( [ mfthreshold1 ] ) and ( [ numshapes ] ) .
the above derivation is based on frequentist statistics . in ref .
@xcite a bayesian analysis is outlined of the detection of gravitational wave signals of unknown form which automatically identifies the statistic @xmath165 as optimal , and which also approximately reproduces the detection threshold @xmath65 . in practice , this search method would be combined with coincidencing between interferometers to achieve high detection reliability and to reduce the effects of non - gaussian noise , as is the case with band - pass filtering as discussed in sec .
[ derivesnrformula1 ] . matched filtering could be more efficient than the noise - monitoring method at combating non - gaussian noise via coincidencing , for the following reason .
when coincidencing with templates , one can demand that the snr in each interferometer be above the appropriate threshold , _ and _ that the signal - parameter values deduced in each interferometer be consistent with each other . for the noise - monitoring searches , one can only demand that the snr in each interferometer be above the appropriate threshold .
hence , matched filtering has more discriminating power against situations in which all the interferometers have moderately large non - gaussian noise spikes somewhere in the relevant time window .
non - gaussian noise may therefore make the less - discriminating noise - monitoring search perform somewhat worse in practice , relative to matched filtering searches , than is indicated by the threshold ( [ nlthreshold1 ] ) . in this section
we derive the relation [ eq .
( [ snraveraged ] ) ] between the expected value of the optimal - filtering snr ( [ snr ] ) , and the energy spectrum of gravitational waves emitted by the source . in general , the snr ( [ snr ] ) for a burst of waves depends on the details of the gravitational waveform , on the orientation of the source with respect to the interferometer , and on the direction to the source .
by contrast , the quantity @xmath216 , the average of the squared snr over all orientations of and directions to the source , depends only on the total energy per unit frequency @xmath217 carried off from the source by the waves .
specifically , consider a gravitational - wave source located at a cosmological redshift @xmath4 and corresponding luminosity distance @xmath55 .
let the locally measured frequency of the waves near the source be @xmath218 , related to the frequency @xmath131 measured at the interferometer by @xmath219 .
let the locally measured energy spectrum of the waves be @xmath220 .
then the orientation - averaged snr squared measured at the interferometer is given by @xmath221 .
\label{snraveraged}\ ] ] note that the relation ( [ snraveraged ] ) refers to an angle - averaged snr obtained from an _ rms average _ of signal amplitudes over different possible orientations of the source and interferometer .
this averaging convention differs from that adopted in refs .
@xcite . there
, the angle - averaged snr is defined to be a cube root of an average of cubed signal amplitudes , rather than an rms average . that `` cube root of a mean cube '' averaging method is the appropriate method for calculating the expected event detection rate @xcite . as a result , the snr formulae used in refs .
@xcite are a factor of @xmath222 larger than the formulae used in this paper , the factor of @xmath222 being an approximation to the effect of the different angle - averaging methods .
we now turn to the derivation of eq .
( [ snraveraged ] ) .
consider first the case where the source of gravitational waves is sufficiently close that cosmological effects can be neglected .
let the source be at a distance @xmath223 from the detector and at a location @xmath224 on the sky .
let @xmath225 denote the direction towards the detector ( spherical polar angles ) with respect to a set of cartesian axes centered at and determined by the source .
let the two independent polarizations of the strain amplitude at the interferometer be @xmath226 and @xmath227 , and let the polarization angle be @xmath228 .
then the response of the interferometer will be @xmath229 , where @xmath120 is the interferometer noise , and @xmath230 here @xmath231 and @xmath232 are the interferometer beam pattern functions , given in , _
e.g. _ , ref .
the dependence of the fourier transformed waveform @xmath233 on @xmath223 is clearly of the form @xmath234 for some function @xmath235 ; we define @xmath236 similarly . combining eqs .
( [ snr ] ) , ( [ decompos ] ) and ( [ rout ] ) gives @xmath237 and @xmath238 .
the average over polarizations and over the sky location gives @xmath239 , @xmath240 @xcite , where the meaning of the angular brackets is given by , for example , @xmath241 from eq .
( [ snr1 ] ) this gives @xmath242 where @xmath243 we now express the energy spectrum @xmath244 of the waves in terms of the quantity @xmath245 .
the local energy flux is @xmath246}},\ ] ] where the overbar means an average over several cycles of the wave .
switching to the frequency domain using parseval s theorem , inserting a factor of two to account for the folding of negative frequencies into positive , and using @xmath247 gives @xmath248 combining eqs .
( [ averagesnr ] ) , ( [ h2def ] ) and ( [ dedomegadf ] ) now yields @xmath249 .
\label{snraveraged1}\ ] ] this is essentially eq .
( [ snraveraged ] ) with @xmath250 and @xmath251 , the limiting form that applies when cosmological effects are neglected .
consider now sources at cosmological distances .
we can generalize eq .
( [ snraveraged1 ] ) to incorporate redshift effects as follows : ( i ) observe that eq .
( [ snraveraged1 ] ) is valid for arbitrary bursts of gravitational waves provided that we interpret the quantity @xmath252 as the locally measured energy flux @xmath253 .
( ii ) use the fact that the number of gravitons per unit solid angle per unit frequency is conserved for propagation in a friedmann - robertson - walker background in the geometric optics limit : @xmath254.\ ] ] here @xmath218 is the frequency at the source and @xmath219 is the frequency at the detector .
( iii ) the conversion factor at the detector from energy per unit solid angle to energy per unit area is just @xmath255 , where @xmath55 is the luminosity distance @xcite .
hence @xmath256.\ ] ] combining this with eq .
( [ snraveraged1 ] ) now yields eq .
( [ snraveraged ] ) .
an alternative derivation of the cosmological modifications to eq .
( [ snraveraged1 ] ) , which yields a useful rule of thumb for understanding redshift effects , is the following .
suppose that an arbitrary source of waves is specified by parameters @xmath257 for @xmath258 .
in the local wave zone the gravitational waveform is of the form @xmath259 for some function @xmath260 , where @xmath223 is the distance to the source . after propagating through a frw background
the waveform becomes @xmath261 where if @xmath257 has dimension @xmath262 or @xmath263 , then @xmath264 is just @xmath265 . equation ( [ generalredshift ] ) can be derived from the general solution for wave propagation in frw backgrounds in the geometric optics limit given in ref .
the fact that the waveform must depend on and reveal the `` redshifted '' source parameters is clear : all source parameters with dimension that we can measure from the waves must essentially come from timescales in the waveform , since the wave amplitude is dimensionless .
all these timescales will be redshifted .
there is a corresponding simple rule for how the snr is affected by cosmological effects .
suppose that when cosmological effects are neglected , the snr is of the form @xmath266 for some function @xmath267 . then the correct snr is just @xmath268 applying this rule to eq .
( [ snraveraged1 ] ) yields eq .
( [ snraveraged ] ) .
in this section we describe our assumptions concerning the gravitational - wave signal from bbh mergers , and the evidence that underlies those assumptions .
we start in subsection [ threelittlepigs ] by discussing the approximate splitting of the total waveform into the inspiral , merger , and ringdown epochs . in sec .
[ mergerphase ] we discuss the various circumstantial pieces of evidence , culled from the literature , about the energy spectrum and the total energy radiated during the nonlinear merger phase .
our assumed merger energy spectrum is given by eq .
( [ dedfmerger ] ) , with parameters given by eqs .
( [ mergefreq ] ) , ( [ fqnrdef ] ) and ( [ epsilonmergerdef ] ) . the inspiral energy spectrum is discussed in sec .
[ inspiralphase ] and specified in eq .
( [ dedfinspiral ] ) .
section [ ringdownphase ] discusses the spectrum and likely overall amplitude of the ringdown signal ; the spectral shape is given in eq .
( [ dedfqnr ] ) and our assumption about the overall energy radiated in eq .
( [ qnrenergy ] ) . in sec .
[ mergernofilters ] we estimate the snr obtainable in band - pass filtering searches for waves from the merger phase . as discussed in the introduction
, the coalescence and its associated gravitational - wave signal can be divided into three successive epochs in the time domain : inspiral , merger , and ringdown . physically , the inspiral consists of the portion of the coalescence in which the black holes are separated bodies that gradually lose energy and angular momentum , slowly spiraling towards one another .
by contrast , the merger consists of that portion of the coalescence in which the dynamics are highly nonlinear and must be treated by numerical relativity . with this in mind , it is useful to define the end of inspiral as the time and frequency @xcite at which numerically generated templates start to be needed . up to this time , post - newtonian templates , possibly supplemented with ibbh templates , will be used ( _ cf_. sec . [ calcstatus ] ) .
after inspiral , we expect there to be a lot of complicated merger dynamics that at present are not well understood .
the system will gradually settle down to a kerr black hole ; the last gravitational waves we expect to see are those produced by the quasi - normal ringing modes of this merged black hole .
it is clear that there will be a smooth transition in the gravitational waveform from the merger portion to the ringdown portion , as the effects of nonlinearities become less and less important with time .
as this happens , the signal should become increasingly well approximated by a linear combination of exponentially decaying sine waves .
this is the behavior that has been seen in numerical simulations of , for example , head - on collisions @xcite . at late times
, the @xmath29 mode will dominate over other quasi - normal modes , for two reasons which are of comparable importance : ( i ) the @xmath29 mode is the most slowly damped of all the qnr modes @xcite , and ( ii ) during coalescence , the binary will have a rotating shape roughly corresponding to spheroidal harmonic indices @xmath29 , and thus this mode will be preferentially excited @xcite .
we define the ringdown as beginning when the waveform becomes dominated by the @xmath29 qnr mode ; the merger thus contains those portions of the waveform where other modes and/or non - linear mode - mode couplings are important .
clearly there is some arbitrariness in the exact time at which the ringdown starts , related to the accuracy we require of the fit of the waveform to the ringdown signal . note that , by definition , the three phases of the signal are disjoint in the time domain
. it does not follow , though , that they should be disjoint in the frequency domain : their energy spectra might overlap .
however , it is at least approximately true that the inspiral and merger and disjoint in both time and frequency .
let @xmath269 be the time at which inspiral is said to end and merger to begin . then , since the inspiral waves chirp monotonically and since the adiabatic approximation is just beginning to break down at the end of the inspiral , the inspiral spectrum is confined , to good approximation , to the frequency region @xmath270 . here , @xmath88 is the frequency of the inspiral waves at @xmath271 .
we shall further assume that the merger waves spectrum is confined to the frequency regime @xmath272 ; this assumption should be valid to a moderately good approximation .
( we discuss below estimates of the value of @xmath88 . ) one particular component of the gravitational - wave signal , the christodoulou memory @xcite , _ will _ violate this assumption .
the memory component of the gravitational - wave signal has most of its power below @xmath88 in the frequency domain , but accumulates gradually during the inspiral , merger and ringdown in the time domain .
the memory waves from bbh coalescences will probably not be detectable with ground based interferometers , but very probably will be detectable with lisa @xcite .
we will neglect the memory component of the waves in our analysis , since it will not be as easy to detect as the components we do discuss .
the total amount of energy radiated in bbh mergers , and its distribution in frequency , is highly uncertain because detailed numerical calculations of these mergers have not yet been made . in this subsection
, we discuss what little evidence there is about the energy radiated , and describe our crude model of the spectrum .
we assume a uniform distribution in frequency from the lower frequency @xmath273 mentioned above and some upper frequency @xmath274 .
the total amount of energy radiated during a bbh coalescence will be some fraction @xmath162 of the total mass @xmath275 of the system : @xmath276 the fraction @xmath162 will depend only on the mass ratio @xmath277 , on the initial spins @xmath278 and @xmath279 of the two black holes , and on the initial direction @xmath280 of the orbital angular momentum @xcite : @xmath281 we can very roughly divide up this fraction as @xmath282 according to the amounts of energy radiated in the three different epochs of the waveform .
there is clearly some arbitrariness in this division of the radiated energy into three pieces , which is related to the choice of the frequency @xmath273 at which the inspiral is said to end and the merger to begin @xcite , and also to the choice of the time @xmath283 at which the merger is said to end and the ringdown to begin .
we now discuss methods of estimating @xmath88 , @xmath284 , and @xmath285 . roughly speaking
, we want to choose the interface between `` inspiral '' and `` merger '' to be at that point where post - newtonian templates cease to be useful and where numerically generated templates will need to be used . numerically generated initial data sets of two orbiting black holes provide some guidance about this interface .
cook @xcite compares the properties of such initial data sets to the predictions of post - newtonian theory at second post - newtonian order .
he finds , for example , that the discrepancy in the binding energy will be @xmath286 when the gravitational - wave frequency ( twice the orbital frequency ) is @xmath287 , where @xmath25 is the total mass of the system , and will be @xmath288 at @xmath289 @xcite . towards the end of the inspiral , the black holes make a gradual transition from a radiation - reaction driven inspiral to a freely - falling plunge @xcite , after which , even if the radiation reaction could be turned off , the black holes would still merge
. this will occur roughly at the last stable circular orbit ( lsco ) .
it would seem to make sense to choose @xmath290 , the frequency at which this occurs . strictly speaking
, the concept of an lsco makes sense only in the test particle limit ( @xmath291 ) .
various somewhat _ ad hoc _ methods have been used to estimate @xmath292 outside this regime , but the results differ by factors @xmath293 .
cook s initial data set analysis together with the calculation of an `` effective potential '' yields the estimate @xmath294 for equal mass black holes @xcite . in post - newtonian theory ,
the lsco can be defined by artificially turning off the radiation reaction terms in the equations of motion . using this method , kidder , will and wiseman estimate that @xmath295 @xcite .
they use hybrid equations of motion which are accurate to post-2-newtonian order for comparable mass ratios , and which are exact in the test particle limit .
the value they obtain varies by less than @xmath296 as the mass ratio is varied .
note that the frequency @xmath297 is close to the value @xmath298 obtained from the usual lsco at @xmath299 in the test particle limit ( here @xmath223 is the usual schwarzschild radial coordinate ) .
finally , earlier analyses by blackburn and detweiler used a variational principle together with the assumption of periodic solutions to einstein s equations to obtain the approximate lower bound @xmath300 @xcite . note that all of these estimates are for equal mass , non spinning black holes ; the value of the frequency @xmath292 can also vary by factors @xmath293 if the black holes are spinning or have different masses . in light of this uncertainty
, we somewhat arbitrarily adopt the conservative value of @xmath301 as the lower cutoff frequency for the energy spectrum .
this ( low ) value of @xmath273 is conservative in the sense that we can be reasonably sure that numerically generated templates will not be needed before @xmath302 . on the other hand
, it may optimistically overestimate the merger snr by increasing the number of cycles in what we define as our merger waveform at the expense of the number of cycles in the inspiral .
we next discuss our choice of upper frequency shutoff for the merger energy spectrum .
as discussed above , we define the end of the merger to occur at that time @xmath283 after which the waveform can be fit fairly accurately by the @xmath29 qnr mode signal .
the merger and the ringdown will therefore be disjoint in the time domain , but not necessarily in the frequency domain .
it seems likely , however , that an approximate upper bound for the frequencies carrying appreciable power during the merger is the quasinormal ringing frequency itself .
this is supported to some extent by calculations of the energy spectrum of waves produced in the test particle limit
2 of ref .
the energy spectrum @xmath53 in that case is peaked near the quasinormal mode frequency @xmath303 of a non - rotating black hole , and has fallen by 2 orders of magnitude at @xmath304 .
although this analysis does not incorporate any nonlinear effects , there is some evidence that the energy spectrum in the equal mass case will be qualitatively similar namely , the head - on collision of black holes in full numerical relativity _ has _ been computed , and the resulting waveforms are qualitatively similar to those predicted from perturbation theory @xcite .
it is not clear how relevant the head - on collision case is to the gradual inspiral ; but on the other hand , there is no other guidance available at the present time .
we will therefore use as our upper shutoff frequency for the merger spectrum the frequency @xmath70 of the @xmath29 quasinormal mode .
this frequency will depend on the unknown dimensionless spin parameter @xmath72 of the final kerr black hole ( its spin angular momentum divided by @xmath305 ) .
we choose to use the value of @xmath72 for which the frequency is highest : in the limit @xmath306 , the qnr frequency becomes @xmath307 ( more than twice the value for a schwarzschild black hole ) @xcite .
thus , our upper cutoff frequency is @xmath308 our reasons for assuming a high value of @xmath72 are discussed in sec .
[ ringdownphase ] below . finally , consider the total amount of energy @xmath309 radiated during the final merger and ringdown . to estimate this we consider first an extrapolation due to smarr @xcite
the idea is to calculate the energy radiated in the test particle approximation using black - hole perturbation theory .
the result will be of the form @xmath310 , where @xmath311 is a dimensionless constant , @xmath82 is the mass of the test particle , and @xmath25 is the total mass . now simply replace @xmath82 by the reduced mass @xmath312 of the two black holes .
this extrapolation works exactly in newtonian gravity supplemented by the quadrupole formula , in the sense that the dynamics of two bodies of comparable masses can be deduced from the corresponding dynamics in the extreme mass ratio limit .
surprisingly , it also works to within @xmath296 to predict the total energy radiated in the head - on collision of two equal - mass black holes @xcite .
thus , one might consider applying this extrapolation to inspiral - preceded mergers .
such a calculation would require a determination of the radiated energy from the final plunge of a test particle , _ i.e. _ , the solution of a coupled radiation reaction / radiation generation problem outside of the adiabatic regime ( although the coupling is important only in a brief transitional phase and essentially sets the initial conditions for an approximately geodesic plunge @xcite ) . this calculation has not yet been done .
however , there is an alternative approximate method to estimate the amount of energy radiated in the test particle limit .
the innermost , marginally bound ( unstable ) circular orbits should be a fair approximation to the motion during the final plunge .
detweiler @xcite showed that the amount of energy radiated per orbit by a test particle in such an orbit is of the form @xmath310 , where @xmath311 varies from @xmath313 at @xmath314 to @xmath315 at @xmath316 , where @xmath72 is the spin parameter of the black hole . assuming that there will be @xmath317 effective orbit during the final plunge
, detweiler estimated that @xcite @xmath318 where @xmath319 which is unity in the equal - mass case @xmath320 .
arguments based on conservation of angular momentum also suggest an approximate lower bound on the radiated energy of about @xmath321 for equal - mass bbhs in the most favorable cases , as we now outline . roughly speaking
, the system s angular momentum divides up as @xmath322 where @xmath278 and @xmath279 are the spins of the two black holes just before the final plunge , @xmath323 is the orbital angular momentum just before the plunge , @xmath324 is the angular momentum carried off by the gravitational waves , and @xmath325 is the spin of the final kerr black hole .
this splitting of the total angular momentum of the spacetime into various pieces is really only well defined only in a post - newtonian type of limit ; however , the effects of this ambiguity are presumably not important for the purposes of our crude , approximate argument .
we now specialize to the most favorable case in which @xmath278 , @xmath279 and @xmath323 are all aligned .
we take the orbital angular momentum at the beginning of merger to be @xmath326 , the value predicted by cook s initial data sets at @xmath327 @xcite .
we also assume that both inspiraling black holes are rapidly spinning , so that @xmath328 . equation ( [ conserve ] ) then yields @xmath329 since @xmath330 .
next , we use that fact that the energy @xmath331 and the angular momentum @xmath332 carried off by gravitons of frequency @xmath131 and azimuthal multipole order @xmath333 are related by @xcite @xmath334 if we estimate @xmath131 as @xmath335 , and make the admittedly optimistic assumption that most of the radiation is quadrupolar , we obtain from eqs .
( [ bound1 ] ) and ( [ eradlrad ] ) the crude estimate for the total radiated energy @xcite @xmath336 this estimate includes both merger and ringdown radiated energies ; we need to subtract out the ringdown portion to obtain the energy contained in the merger portion of the waveform .
below we estimate @xmath337 to be an approximate upper bound for the ringdown energy , and hence most of the energy ( [ eest1 ] ) should be radiated in the merger waves .
there is an additional , separate argument one can make which indicates that most of the energy ( [ eest1 ] ) should be radiated as merger waves and not as ringdown waves . as noted by eardley and hirschmann @xcite , any system with @xmath338 can not evolve to @xmath339 by radiating quadrupolar waves at the ringing frequency @xmath340 of a near - extremal kerr black hole .
this is because at this high frequency , too much mass - energy is radiated per unit angular momentum radiated ; eq . ( [ eradlrad ] ) with @xmath341 and with @xmath342 yields @xmath343 . hence , since the final black hole must have @xmath344 , a substantial amount of the radiation must be emitted at lower frequencies .
based on the estimates ( [ eest0 ] ) and ( [ eest1 ] ) , and on the estimated upper bound @xmath337 mentioned above for the ringdown radiated energy , we perhaps optimistically take @xmath321 to be the energy radiated during merger in the equal - mass case ( which corresponds to about @xmath345 total radiated energy in the merger and ringdown ) . for non equal - mass bbhs
we assume that the radiated energy is reduced by the factor ( [ reductionfactor ] ) , so that @xmath346 the rather high radiation efficiency of @xmath347 that we are assuming is probably most plausible in the context of rapidly spinning coalescing black holes .
in particular , if the spins and the orbital angular momentum are somewhat misaligned , the coalescence will be something like `` two tornados with orientations skewed to each other , embedded in a third , larger tornado with a third orientation '' @xcite .
intuitively one would expect that such systems have more `` settling down '' to do to get to the final kerr black hole , and that correspondingly the nonlinear , highly dynamical phase should last longer and/or produce more radiation . also , the potential barrier that surrounds the final black hole ( which normally tends to reflect back into the black hole the dominant waves of frequency @xmath348 ) presumably will effectively not be present during the violent phase of a merger in which the spins and orbital angular momentum are of comparable magnitude and are misaligned .
we note that coalescences which radiate as much energy as the estimate ( [ epsilonmergerdef ] ) will also likely radiate a substantial amount of linear momentum , and the consequent recoil of the final black hole could correspond to a kick velocity that is a moderate fraction of the speed of light
. finally , consider the shape of the energy spectrum @xmath349 between @xmath273 and @xmath350 .
for simplicity , and for lack of evidence in favor of anything more specific , we choose a flat spectrum : @xmath351 where @xmath352 and @xmath126 is the step function .
our assumed merger spectrum has the desirable feature that it is reasonably consistent with the inspiral energy spectrum at @xmath353 .
this can be seen as follows .
we do not expect a very large change in @xmath53 at the interface between inspiral and merger , for the following reason : throughout the inspiral , the energy spectrum can be written @xcite @xmath354 ^ 2,\ ] ] where @xmath147 is the number of cycles spent near frequency @xmath131 , and @xmath355 $ ] is the wave amplitude at the time @xmath146 when the waves frequency is @xmath131 [ see also eqs .
( [ snrgain ] ) and ( [ snropt ] ) ] . in the transition from inspiral to plunge ,
the amplitude @xmath355 $ ] should change only moderately , but @xmath147 could conceivably change by a large amount : from its pre - plunge value to @xmath356 .
it turns out that the number of cycles does not change very drastically : at the interface between inspiral and merger , @xmath357^{-5/3}. \label{ncycinspiral}\end{aligned}\ ] ] ( here , @xmath358 is the so - called chirp mass . ) hence , we do not expect @xmath53 to change too much .
we now show that our assumed values of @xmath53 from the end of the inspiral and from the beginning of the merger are reasonably close to one another , in agreement with the above expectation .
using the approximation that the motion is an adiabatic evolution from one circular orbit to another , the energy radiated during inspiral can be written @xmath359 , \label{adiabaticapprox}\ ] ] where @xmath360 is the binding energy of a particle in a circular orbit at orbital frequency @xmath361 ( recall that the gravitational - wave frequency is twice the orbital frequency ) . at our estimated plunge frequency @xmath362 , the hybrid equations of motion of kidder , will and wiseman @xcite predict in the equal mass case that @xmath363 while cook s initial data sets predict that @xcite @xmath364 the true value of @xmath365 is presumably somewhere between these two values .
we would expect the value of @xmath366 to be smaller than this , because the approximation ( [ adiabaticapprox ] ) is breaking down at this point some of the gravitational binding energy is being converted to inward radial kinetic energy instead of being radiated .
this picture is roughly consistent with our assumption of @xmath367 for @xmath368 , given that eqs .
( [ est1 ] ) and ( [ est2 ] ) apply to the non - spinning case , and we are interested in the rapidly spinning case for which @xmath349 could be a few times larger . the standard quadrupole formula prediction for the inspiral energy spectrum is ( see , _ e.g. _ , ref .
@xcite ) @xmath369 this approximate formula is adequate to estimate the snr obtained from optimal filtering of the inspiral waveform .
the formula will be accurate to within a few tens of percent up to @xmath370 @xcite .
using eq .
( [ dedfinspiral ] ) to estimate the snr effectively assumes that both the gravitational - wave signal and the theoretical templates used to cross - correlate with the data stream are given by the quadrupole approximation ( [ dedfinspiral ] ) .
the snr we calculate using this lowest order formula will be approximately the same as that found by cross - correlating real signals [ which incorporate higher order corrections to eq .
( [ dedfinspiral ] ) ] against sufficiently accurate theoretical templates . as outlined in sec .
[ calcstatus ] , the required template accuracy should be achievable by post - newtonian expansions @xcite , perhaps supplemented with alternative techniques for the latter , high frequency part of the signal at @xmath371 ( the ibbh regime ) .
we assume that the inspiral energy spectrum shuts off at @xmath372 , as discussed in sec .
[ mergerphase ] . as we have discussed , the ringdown portion of the gravitational - wave signal is that portion in the time domain that can be fit fairly accurately by an exponentially decaying sinusoid corresponding to the most slowly damped @xmath29 quasinormal mode of the final black hole .
the shape of the corresponding energy spectrum is well understood : it is just a resonance curve ( although see appendix b for discussion of a subtlety in the applicability of the concept of the waves energy spectrum for calculating ringdown snrs ) .
the overall amplitude of the energy spectrum , however , is not well understood . in this section
we discuss our estimate of the overall amplitude for the ringdown signal .
the qnr gravitational waveforms @xmath373 and @xmath374 are given by @xcite @xmath375 , \label{qnrwaveform}\end{aligned}\ ] ] for @xmath376 .
here we have chosen @xmath377 to be the time of the start of the ringdown , @xmath25 is the mass of the black hole , @xmath378 is its spin , and @xmath379 is a constant phase .
the quantities @xmath380 and @xmath238 are spherical polar coordinates centered on the black hole [ _ cf_. sec .
[ derivesnrformula ] ] , @xmath381 is a spin weighted spheroidal harmonic whose angle averaged rms value is @xmath382 and @xmath383 is a dimensionless coefficient that describes the magnitude of the perturbation when the ringdown begins .
the quantities @xmath350 and @xmath384 are the frequency and damping time , respectively , of the qnr mode of the black hole .
the quality factor @xmath198 of the mode is given by @xmath385 .
as mentioned in the introduction , there is a mapping between the values of @xmath25 and @xmath72 , and of @xmath70 and @xmath71 .
this mapping has been explored by leaver @xcite and echeverria @xcite .
leaver studied in depth the damping times and ringing frequencies of various qnr modes for various kerr black hole masses and spins . by numerically solving the teukolsky equation
, he produced extensive catalogs of the variation of @xmath71 and @xmath70 with black hole mass @xmath25 and dimensionless spin @xmath72 @xcite . from that data for the @xmath29 mode ,
echeverria @xcite produced the following analytic fits , which are good to within roughly 5% : @xmath386 \ { 1 \over 2 \pi m } \nonumber \\ \mbox { } \ , & = & \left[1 - 0.63(1-a)^{3/10}\right ] \
\left({37\,m_\odot\over m}\right)865\,{\rm hz}\nonumber\\ q & \approx & 2 \ , ( 1-a)^{-9/20}. \label{fqnrandq}\end{aligned}\ ] ] note that @xmath198 diverges as @xmath306 but that @xmath70 tends towards a constant .
the energy spectrum for the qnr waveform ( [ qnrwaveform ] ) is derived in appendix [ ringdown3 ] and is given by @xmath387 ^ 2 } \nonumber \\ \mbox { } & & + { 1 \over \left[(f+{f_{\rm qnr}})^2 + ( 2\pi{\tau})^{-2}\right]^2 } \bigg\ } \label{dedfqnr } \\ & \approx & { 1\over 8 } { \cal a}^2 q m^2 f_{\rm qnr } \ , \ , \delta(f - f_{\rm qnr } ) \left[1 + o(1/q)\right ] .
\label{dedfqnrapprox}\end{aligned}\ ] ] approximating the energy spectrum by a delta function as in eq .
( [ dedfqnrapprox ] ) will often ( but not always ) provide a fairly good approximation to the snr ; see appendix [ ringdown3 ] for more details .
the values of the spin @xmath72 of the final black hole ( and correspondingly , @xmath198 ) and also of the amplitude @xmath383 will depend on the initial parameters of the system as in eq .
( [ dependson ] ) .
this dependence is very poorly understood at present .
we expect the final black hole to be rapidly spinning since , as explained in sec .
[ mergerphase ] , the total angular momentum of the binary at the end of the inspiral is @xmath76 when even the individual black holes are non - spinning @xcite , and the individual black hole spins can add to this .
moreover , the individual black holes in the binary may typically have been spun up to near maximal rotation by an accretion disk @xcite . for definiteness , we somewhat arbitrarily take @xmath75 , which corresponds from eq .
( [ fqnrandq ] ) to @xmath388 and @xmath389 .
the final ringdown snr values we obtain vary only weakly with our assumed value of @xmath72 [ _ cf_. eq . ( [ qnrsnrii ] ) ] , for fixed total energy radiated in the ringdown .
although the value of the overall amplitude @xmath383 is uncertain , we can estimate a upper bound on @xmath383 for equal mass bbhs in the following way . when the @xmath29 quasinormal mode becomes the dominant mode present
, the merged body will be a distorted kerr black hole .
the horizon s cross section , seen looking down on this black hole from above , will be a rotating oval rather than a circle .
we can quantify the distortion by the ratio of the polar circumference about the long axis of this oval to the polar circumference about the short axis .
let @xmath390 denote the perturbation amplitude such that the ratio of circumferences is , say , @xmath391 .
it is fairly clear that nonlinear couplings between the @xmath29 mode and other modes will be important for @xmath392 ; thus , at this @xmath391 distortion ratio , the signal will not be well approximated by just the @xmath29 mode .
hence , @xmath390 is a reasonable upper bound for the true amplitude @xmath383 .
we could in principle calculate the upper bound @xmath390 by writing the spacetime metric as @xmath393 where @xmath394 is the kerr metric and @xmath395 is the @xmath29 quasinormal mode whose asymptotic form at large @xmath223 is given by eq .
( [ qnrwaveform ] ) , and by calculating from this metric the ratio of circumferences @xcite . for this paper , we used a much less sophisticated method to estimate @xmath390 . using the quadrupole formula
, we examined the radiation produced by a solid body that is distorted with the same @xmath391 circumference ratio as the merged black hole , and obtained the estimate @xmath396 @xcite . setting our waveform amplitude @xmath383 equal to this upper bound yields an rms angle - averaged waveform @xmath397 at the beginning of ringdown . from eq .
( [ dedfqnr ] ) , the corresponding radiated energy is @xmath398 as mentioned in the introduction , comparable ringdown radiation efficiencies @xmath399 have been seen in numerical simulations of the evolution of distorted , spinning black holes @xcite . to summarize , our assumed parameter values for the black hole dimensionless spin parameter @xmath72 and for the dimensionless amplitude parameter @xmath383 for equal - mass bbhs are @xmath400 these imply the following values for the quasi - normal ringing frequency @xmath70 , the quality factor @xmath198 and the fraction of total mass - energy radiated during ringdown @xmath401 : @xmath402 for non equal - mass bbhs , we assume that @xmath383 and @xmath403 are reduced by the factor ( [ reductionfactor ] ) . in sec .
[ derivesnrformula1 ] we showed that for any burst of gravitational waves , the band - pass filtering snr is smaller than the optimal filtering snr by a factor of approximately @xmath404 here @xmath60 is the duration of the merger and @xmath61 is the band - pass filter s bandwidth , which should be approximately the largest expected signal bandwidth .
the product @xmath152 , which is roughly the number of independent frequency bins ( or fourier coefficients ) needed to describe the signal , also determines how well matched filtering searches perform compared to noise - monitoring searches , as explained in sec .
[ nonlinear ] . in this section ,
we obtain a rough estimate of the parameter @xmath152 for the merger gravitational waves .
first consider the bandwidth @xmath61 .
our assumed bandwidth for the merger signal is @xmath405 [ since @xmath406 ; _
cf_. eqs .
( [ mergefreq ] ) and ( [ fqnrdef ] ) ] .
we can not , however , we completely confident that all signal power in the merger will lie at frequencies below @xmath70 , so a more appropriate choice might be @xmath407 .
also the quasinormal ringing frequency @xmath408 depends on the dimensionless spin parameter @xmath72 of the final black hole as given by eq .
( [ fqnrandq ] ) . choosing the highest possible value , @xmath409 , yields @xmath410 turn , now , to the effective duration @xmath60 of the signal , which is defined by eq .
( [ tdef ] ) .
clearly this parameter will vary considerably from event to event and its value is highly uncertain . to get a feeling for the range of possible values ,
let us consider the type of coalescence discussed in sec .
[ mergerphase ] , where both inspiraling black holes are nearly maximally spinning , with their spins and the orbital angular momentum of the binary all aligned . in this favorable case , as argued in sec .
[ mergerphase ] , at the end of inspiral near @xmath27 the binary has an excess @xmath411 of total angular momentum that it needs to shed before it can settle down to its final kerr state .
thus it seems reasonable to suppose that the two black holes will be somewhat centrifugally hung up and continue to orbit for several cycles before their event horizons merge .
in other words , the orbital dynamical instability may be suppressed by the aligned spins , so that what we call merger ( for data - analysis and computational purposes ) may dynamically resemble the inspiral .
the duration of the merger in this case may therefore be long .
by contrast , for inspiraling non - spinning black holes , there is probably no excess angular momentum that needs to be radiated after the orbital dynamical instability is reached , and so it is plausible that the two black holes merge rather quickly . in such a case , most of the emitted energy could come out in the ringdown waves rather than in merger waves , as has typically been seen in numerical simulations . to estimate the duration @xmath60 for the first type of scenario
, we assume that the gravitational - wave luminosity @xmath412 during the merger phase is roughly the same as the luminosity at the start of the ringdown , @xmath413 .
( here , @xmath414 is the ringdown radiation efficiency , and @xmath71 is the damping time of the qnr mode ) .
combining this with a total energy radiated during the merger of @xmath415 ( with @xmath416 ) yields the estimate @xmath417 clearly this estimate will become invalid for high values of @xmath71 ; in that limit , the high quality factor of the qnr mode causes a low quasi - normal ringing luminosity , whereas there is no reason for the merger luminosity to be comparably low .
nevertheless , we insert our assumed value of @xmath71 , which corresponds to a quality factor of @xmath418 [ eqs .
( [ fqnrandq ] ) and ( [ qnrvals ] ) ] into eq .
( [ testimate ] ) to obtain @xmath419 . combining this with eqs .
( [ simp1 ] ) and ( [ deltafestimate ] ) yields the estimate @xmath420 for inspiraling schwarzschild black holes , on the other hand , @xmath60 may not be much larger than @xmath421 ( assuming @xmath422 say ) , giving @xmath423 .
the factor @xmath424 is thus likely to lie in the range @xmath425 .
we adopt the estimate @xmath426 in sec .
[ detectmerger ] below to estimate the reduction in snr resulting from using band - pass filters instead of optimal templates .
we use the conservative , large value @xmath427 in sec .
[ detectmerger ] to estimate detection thresholds for noise - monitoring searches for signals .
in this section we describe our approximate , piecewise power law , analytic model for the noise curves for the initial ligo interferometers , for the advanced ligo interferometers , and for the lisa interferometer .
we express our model in terms of the dimensionless quantity @xmath130 , where @xmath50 is the one sided power spectral density of the interferometer noise @xcite .
the quantity @xmath428 is the rms fluctuation in the noise at frequency @xmath131 in a bandwidth @xmath429 @xcite .
our approximate model for the noise spectrum is @xmath430^{3/2 } & \mbox { $ \alpha f_m < f$.}\\ \end{array } \right.\ ] ] this noise curve scales like @xmath431 at high frequencies and like @xmath432 at low frequencies , and has a flat portion at intermediate frequencies .
the noise curve depends on four parameters : ( i ) a lower shutoff frequency @xmath433 below which the interferometer noise rapidly becomes very large and can be approximated to be infinite . for the ground
based interferometers , this low - frequency shutoff is due to seismic noise , while for lisa it is due to accelerometer noise ( ref .
@xcite , p. 23 ) .
( ii ) a frequency @xmath434 , which is the location of the center of the flat portion of the spectrum .
( iii ) a dimensionless parameter @xmath435 , which is the minimum value of @xmath436 .
( iv ) a dimensionless parameter @xmath437 which determines the width in frequency of the flat portion of the noise curve .
we approximate the noise curves by piecewise power laws in this way for calculational convenience . for the initial and advanced ligo interferometers , we determined best - fit values of the parameters @xmath433 , @xmath434 , @xmath435 and @xmath437 by fitting to the noise curves given in ref .
( note that fig .
7 of ref .
@xcite is a factor of 3 too small from @xmath438 to @xmath439 .
this error does not appear in fig .
10 of that reference @xcite . )
the resulting parameter values are : @xmath440 and @xmath441 for ground - based interferometers , the @xmath432 portion of our approximate formula ( [ noisespec ] ) is the thermal suspension noise and the @xmath431 portion models the laser shot noise . the intermediate portion smoothes the transition between these two pieces of the spectrum @xcite . for the space - based lisa interferometer , we determined best - fit values of the parameters @xmath434 , @xmath435 and @xmath437 by fitting to the noise curve given in ref .
@xcite , and obtained the lower cutoff frequency @xmath433 from ref .
the resulting parameter values are : @xmath442 our piecewise power - law model fits the true noise curve less accurately for lisa than for the ligo interferometers , but it is still a fairly good approximation .
the sensitivity of lisa at the lower end of its frequency window may be degraded somewhat by a background of gravitational waves from white dwarf binaries @xcite .
we neglect this issue here as this white dwarf noise level is fairly uncertain .
in this section we calculate the angle - averaged snrs for the three different stages of the coalescence ( inspiral , merger , and ringdown ) for initial ligo interferometers , for advanced ligo interferometers , and for lisa .
we do so for black hole binaries at various distances and of various masses .
these snr values are obtained by substituting the energy spectra ( [ dedfinspiral ] ) , ( [ dedfmerger ] ) and ( [ dedfqnrapprox ] ) estimated in sec . [ signalassumptions ] , together with the interferometer noise curves ( [ noisespecligoinitial ] ) , ( [ noisespecligoadvanced ] ) and ( [ noisespeclisa ] ) , into the snr formula ( [ snraveraged ] ) derived in sec . [ derivesnrformula ] .
the analytic calculations are carried out in appendix [ appsnr ] ; in this section we summarize the results in three graphs ( figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] ) , and discuss their implications .
we start in subsection [ egs ] by graphing the waves energy spectra and the interferometers noise spectra for several different examples of black hole binaries , in order to illustrate the factors determining the snr values .
the general results are presented in subsection [ generalresults ] .
we start by rewriting the general formula ( [ snraveraged ] ) for the snr in a more useful form . if we define the characteristic gravitational - wave amplitude @xmath443 , \label{hampdef}\ ] ] then from sec .
[ derivesnrformula ] it can be seen that the snr squared for an optimally oriented source is @xmath444 where @xmath445 .
however , eq . ( [ snraveraged ] ) shows that the angle - averaged snr squared is a factor of @xmath446 smaller than the optimal value ( [ snropt ] ) , so we can rewrite eq .
( [ snraveraged ] ) as @xmath447 where @xmath448 .
we interpret the quantity @xmath449 as the rms noise appropriate for waves from random directions with random orientations @xcite .
the usefulness of the quantities @xmath108 and @xmath109 is that plotting @xmath108 and @xmath109 for various sources illustrates [ from eq .
( [ snrsb ] ) ] the possible snr values and the distribution of snr squared with frequency ( see , _
e.g. _ , ref .
@xcite ) . in fig .
[ hcharfig ] , we show the rms noise amplitude @xmath109 for our model ( [ noisespec ] ) of the initial and advanced ligo interferometer noise curves , together with the characteristic amplitude @xmath108 for two different examples of bbh coalescences , namely a coalescence of total mass @xmath450 at a distance of @xmath451 , and a @xmath452 coalescence at a redshift of @xmath113 .
( we assume that the cosmological parameters are @xmath453 and @xmath454 . ) in each case , the sloped portion of the dashed @xmath455 line is the inspiral signal , the flat portion is our crude model of the merger , and the separate dotted portion is the ringdown .
note that the ringdown and merger overlap in the frequency domain since ( as we have defined them ) they are disjoint in the time domain .
by contrast , the inspiral and merger are approximately disjoint in both the frequency and time domains , since the adiabatic approximation is just breaking down at the end of the inspiral @xcite .
notice that in both cases , @xmath456 and @xmath457 , the waves characteristic amplitude @xmath108 is rather larger than @xmath109 for most of the merger spectrum for the advanced interferometers , indicating the detectability of the merger waveform when optimal filtering can be used . in particular , note that the waves should be quite visible to the advanced interferometers for the @xmath458 binary even though it is at a cosmological distance . even if such binaries are rare , they are visible to such great distances that measurements can be made of a huge volume of the universe , making them a potentially very important and interesting source .
cosmological binaries have an enhanced snr in part because the cosmological redshift moves their frequency spectrum down closer to ligo s optimal band ; this redshift enhancement effect can sometimes be so strong that for two identical mergers at different distances , the more distant merger can have the larger snr . figure [ hcharfig ] also shows that , of these two example bbh coalescences , only the nearby one at a distance of @xmath459 would be detectable by the initial interferometers .
such coalescences may yield an interesting event rate for the initial interferometers as discussed in the introduction , a coalescence rate of very roughly @xmath460 is expected for bbh coalescences in this mass range ( @xmath8 ) inside @xmath461 @xcite , although there is considerable uncertainty in this estimate .
a qualitatively different , possibly important type of source for the initial ligo interferometers ( and also for the advanced interferometers ) is the coalescence of black hole binaries with masses of order @xmath462 , as we have discussed in the introduction .
( we call such binaries intermediate mass bbhs , as distinct from solar mass or supermassive binaries . ) in fig . [ hcharfig1 ] we show the characteristic amplitude @xmath463 for a hypothetical bbh coalescence of total mass @xmath464 at a redshift of @xmath465 , corresponding to a luminosity distance of @xmath466 . note in particular that the initial ligo interferometer noise curve has best sensitivity at @xmath467 just where the ( redshifted ) ringdown frequency is located .
we discuss further in sec .
[ detection ] the range and possible detection rates for the initial ligo interferometers for this type of source .
turn , now , to the detection of supermassive bbh signals by the possible future space - based detector lisa @xcite .
lisa can study bbh mergers with far higher accuracy and resolution than the ground based interferometers , because the snr values are typically much higher ( @xmath468 ) .
our calculation of the inspiral snr values for lisa differs from the other snr calculations in the following way . in the frequency integral of eq .
( [ snraveraged ] ) , if one were to integrate over the whole frequency domain allowed by the noise curve ( as is done for the ligo initial and advanced interferometers ) , one would in some cases obtain the snr for a measurement which lasts several hundred years , which is unrealistic .
thus , it is necessary when calculating inspiral snrs for lisa to restrict the integral over frequency in eq .
( [ snraveraged ] ) to the domain that corresponds to , say , one year of observation ; see appendix [ appsnr ] .
figure [ hcharfig2 ] shows our approximate model [ eqs .
( [ noisespec ] ) and ( [ noisespeclisa ] ) ] of lisa s projected noise spectrum , together with the gravitational - wave amplitude @xmath463 for the inspiral , merger and ringdown stages of two different bbh coalescences : a bbh of total mass @xmath469 at a redshift of @xmath117 , and a bbh of total mass @xmath470 at a redshift of @xmath113 .
the @xmath469 bbh enters the lisa waveband at @xmath471 roughly one week before the final merger .
the snrs obtained in this case from the inspiral , merger and ringdown signals are approximately @xmath472 , @xmath473 and @xmath474 respectively .
the @xmath470 bbh enters the lisa waveband about twenty years before the final merger .
the snr obtained from the last year of the inspiral signal , from @xmath475 to @xmath476 is approximately @xmath477 , while the merger and ringdown snrs are about @xmath478 and @xmath23 respectively .
we now turn from these specific examples to the dependence of the snr values on the mass of and distance to the binary in general . in appendix [ appsnr ]
we obtain analytic formulae for the snr values for the three phases of bbh coalescences , and for the various interferometers . in this section
we plot the results for equal - mass bbhs , which are shown in figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] . the inspiral and merger curves in these figures ( except for the lisa inspiral curves ; see appendix [ appsnr ] ) are obtained from eqs .
( [ inspsnrformula2 ] ) and ( [ mergesnrformula2 ] ) of appendix [ appsnr ] , while the ringdown curves are obtained by numerically integrating eq .
( [ dedfqnr ] ) in eq .
( [ snraveraged ] ) .
the snr values for the initial ligo interferometers are shown in fig .
[ initialligosnr ] .
this figure shows that it is possible that an important source for the initial ligo interferometers may be the coalescences of binary black holes where the black holes have masses in the range of several hundred solar masses , which would be visible out to almost @xmath98 . for such sources ,
the inspiral portion of the signal would not be detectable , and one would need to search for the ringdown or merger portions of the signal in order to detect the waves . see sec .
[ detection ] for further discussion .
the event rate for such high mass bbhs is very uncertain ; see ref .
@xcite for a possible formation scenario .
intermediate mass bbhs with @xmath83 ( _ e.g. _ , @xmath479 , @xmath480 ) are presumably much more common in the universe than the intermediate mass bbhs with @xmath481 discussed above .
the snrs for such mixed binaries will be much lower , however .
as seen in appendix a , the merger and ringdown snr s scale as @xmath482 , while the inspiral ringdown scales as @xmath483 .
the difference in scaling occurs because the duration of the merger and ringdown signals are approximately independent of @xmath82 , but the duration of the inspiral signal scales as @xmath484 . for bbh coalescences of total mass less than or of order 20 @xmath485 , the results of fig .
[ initialligosnr ] can be summarized by the following approximate formulae for the initial ligo interferometers [ c.f .
( [ inspsnrformula2 ] ) , ( [ mergesnrformula2 ] ) , and ( [ qnrsnr2 ] ) of appendix [ appsnr ] ] : @xmath486 where @xmath487 is the fraction of the total mass - energy radiated in the merger signal and @xmath488 is the corresponding fraction for the ringdown signal
. figure [ initialligosnr ] also shows that the inspiral signals from low mass bbh mergers with @xmath8 should be visible to @xmath16 ( the snr detection threshold is @xmath489 @xcite ) .
the ground - based interferometers will , over a period of years , gradually be improved from the initial sensitivity levels to the advanced sensitivity levels @xcite .
roughly half way between the initial and advanced interferometers , the range of the detector system for @xmath8 bbhs will be @xmath490 . as explained in the introduction ,
it is thought to be plausible that most bbh progenitor systems do not disrupt during the stellar collapses that produce the black holes , so that their coalescence rate could be about the same as the birth rate for their progenitors , about @xmath31 years in our galaxy , or several per year within a distance of @xmath32 @xcite .
if this is the case , bbhs should be detected early in the gradual process of interferometer improvement .
it is also conceivable that bbhs could be seen closer than @xmath461 .
high mass stars may have been much more numerous among population iii stars due to the lower metallicity at high redshift , so there could be a large population of bbhs that are remnants of population iii stars living in galactic halos .
galactic nuclei could also produce a larger population of bbhs than normal stellar evolution in non - nuclear regions @xcite .
either of these scenarios could increase event detection rate and hasten the first detections of bbhs .
figure [ advancedligosnr ] shows the snr values for the advanced ligo interferometers . for bbh coalescences of total mass less than or on the order of a few tens of solar masses , the results in fig .
[ advancedligosnr ] can be summarized by the following approximate formulae for the advanced ligo interferometers [ c.f .
( [ inspsnrformula2 ] ) , ( [ mergesnrformula2 ] ) and ( [ qnrsnr2 ] ) of appendix [ appsnr ] ] : @xmath491 our estimate ( [ lowmassadv ] ) for the ringdown snr is roughly consistent with previous estimates by finn @xcite . from fig .
[ advancedligosnr ] it can be seen that for advanced ligo interferometers , equal - mass bbh inspirals will be visible out to @xmath492 for the entire range of masses @xmath493 .
thus , there is likely to be an interesting event rate .
indeed , the snrs will be high enough even for rather large distances that it should be possible to extract each binary s parameters with reasonable accuracy @xcite .
by contrast , the ringdown snr is fairly small except for the largest mass systems .
for very massive binaries or binaries that are closer than 1 gpc , advanced interferometers may measure some fairly good ringdown snrs , which would allow fairly good estimates of the mass and spin of the final black hole @xcite .
figure [ lisasnr ] shows the snr values obtainable from the three phases of bbh coalescences by lisa : the last year of inspiral , the merger and the ringdown .
we also show the snr value obtainable from one year of integration of the inspiral signal one hundred years before the merger , and a similar curve for one thousand years before the merger .
this figure shows that lisa will be able to perform very high accuracy measurements of bbh mergers ( snr values @xmath468 ) essentially throughout the observable universe ( @xmath494 ) in the mass range @xmath495 .
as discussed in the introduction , there is likely to be an interesting event detection rate .
the snr curves in fig .
[ lisasnr ] for measurements one hundred and one thousand years before merger show that many inspiraling bbhs that are far from merger should be detectable by lisa as well .
if the merger rate of smbh binaries turns out to be about one per year throughout the observable universe , then at any given time one would expect roughly one thousand smbh binaries to be a thousand years or less away from merger .
lisa will be able to monitor the inspiral of such binaries ( if they are of sufficiently low mass ) with moderate to large snr @xcite .
finally , it should be noted that the relative magnitude of the merger and ringdown snr values is somewhat uncertain . as we have discussed in sec .
[ signalassumptions ] , we have assumed a total radiated energy of @xmath321 in the merger portion of the signal , and @xmath77 in the ringdown portion , a ratio of @xmath496 . in reality , it may turn out that in individual cases the ratio is as high as @xmath22 or as low as @xmath497 .
it may even turn out to be the case that for most coalescences , the ringdown portion of the waveform carries most of the radiated energy of the combined merger / ringdown regime ( depending possibly on the distribution of initial spins ) .
thus , the snr values shown in figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] should merely be taken as illustrative .
we now turn to a discussion of the implications of the above signal - to - noise ratios for the required search strategies for the various types of signals , and for the distances to which these types of signals can be detected .
one of the reasons that coalescences of compact objects are such good sources for gravitational - wave detectors is that the inspiral portion of the signal is very predictable , so that optimal filtering may be used in searching for the waves @xcite .
as we have discussed , optimal filtering enhances the achievable snr values by a factor of roughly @xmath498 , where @xmath499 is the number of cycles in the waveform in the frequency band of the detectors . for neutron star - neutron star ( ns - ns ) coalescences , @xmath499 will be on the order of several thousand , while for bbh coalescences it will be on the order of several hundred @xcite .
thus , for ns - ns coalescences , and also for bbh coalescences with total mass a few tens of solar masses or less , the inspiral portion of the gravitational - wave signal will be used to _ detect _ the entire waveform . in these cases
, it is not necessary to _ search _ for the merger and ringdown portions of the waveform , since it will be known roughly where in the interferometer data stream they are expected to lie .
however , there is no guarantee in these cases that the merger and ringdown portions will be visible in the noisy data stream .
a key issue is how visible these portions are and how much information can be extracted from them .
we discuss this issue in detail in paper ii @xcite . for larger mass black hole systems , however ,
our signal to noise ratio calculations show that the merger and ringdown snr values can be larger than the inspiral snr values . for equal - mass bbhs
, this will occur whenever @xmath500 for the initial ligo interferometers , and whenever @xmath501 for the advanced ligo interferometers .
moreover , the inspiral snr completely shuts off for large enough @xmath502 , as can be seen from figs .
[ initialligosnr ] and [ advancedligosnr ] .
the reason for this shutoff is that at large enough values of @xmath503 , the entire inspiral energy spectrum lies at frequencies below the `` seismic wall '' shutoff in the interferometer noise curve .
the mass cutoff occurs at @xmath504 for the initial interferometers , and at @xmath505 for the advanced interferometers .
admittedly , bbh binaries of total mass @xmath506 may well be very much more rare than bbh binaries of @xmath507 ; however , they will be visible to such great distances that there may be an interesting event rate .
moreover , for the initial ligo interferometers , the mass scale @xmath508 at which the inspiral snr becomes much smaller than the merger / ringdown snrs is not terribly high .
in such high mass cases for which the merger and inspiral snrs exceed the inspiral snr , it will be necessary to perform a search for the merger and/or ringdown portions of the signal , independently of any searches for inspiral signals , in order that all possible events be detected .
if one seeks to detect the waves merely by optimal filtering for the inspiral waveform , some fraction of the events will be missed which otherwise might have been detectable .
in fact , it may very well turn out that merger signals from bbh coalescences could be the dominant source for the initial ligo interferometers . therefore a search in real time should be performed in the data stream for ringdown and merger signals from high mass , bbh mergers .
one might imagine that the gravitational waves would generally be easier to detect by searching for the merger signal , since we have estimated that the snr values for the merger phase are typically a factor of a few larger than those for the ringdown ( _ cf . _ figs .
[ initialligosnr ] and [ advancedligosnr ] ) .
there are several factors that complicate this conclusion , however . on the one hand , the ringdown s waveform shape is better understood , which makes it easier to produce search templates and hence easier to detect the signal . on the other hand ,
the ratio between the merger and ringdown snrs is really quite uncertain , as discussed in sec .
[ generalresults ] , and so it is plausible that the merger snr will be larger than we have indicated relative to the ringdown snr . in any case , the ratio between merger and ringdown snrs will presumably vary a lot from event to event .
thus , it would seem that searches will be necessary for _ both _ types of signal in the data steam , at least for the mass range in which the ringdown snr is expected to exceed the inspiral snr .
[ from sec .
[ snrsection ] we estimate this mass range to be @xmath509 for the advanced interferometers , and @xmath510 for the initial interferometers . ] we summarize the discussion of this section by displaying the optimum search strategies for various mass ranges for the three different interferometers . in each case
below , the mass range marked merger refers to matched filtering searches for merger signals .
if merger templates are available , then in the indicated mass ranges merger searches will probably be more successful than inspiral or ringdown searches ; the question mark is a reminder that templates for the merger may not be available , and that consequently such searches might not be possible .
@xmath511 @xmath512 @xmath513 consider first the search for ringdown signals . in this case , since the shape of the signal is known up to several unknown parameters , it will be feasible to implement a search based on the method of matched filters , analogous to what is being planned for detecting mergers of ns - ns and low mass ns - bh and bbh binaries using the inspiral waveform @xcite . in other words , a bank of templates of the form ( [ qnrwaveform ] ) , for different values of the ringdown frequency and damping time ,
can be integrated against the ( suitably pre - filtered ) interferometer output .
this will be feasible for both the initial and advanced ligo interferometers , and also for lisa .
the number @xmath514 of required templates @xcite can be estimated by combining the formalism developed by owen @xcite and the results of echeverria and finn on the expected accuracy of measurement of of the ringdown frequency and damping time @xcite .
using eqs .
( 4.15 ) of ref .
@xcite and eqs .
( 2.23 ) and ( 2.28 ) of ref .
@xcite we find that the metric defined by owen on the space of parameters is given by @xcite @xmath515 where @xmath198 is the quality factor .
the formula ( [ metric ] ) for the owen metric is valid only in the high @xmath198 limit ; it has corrections of order @xmath516
. moreover , the formula is also only valid when the noise spectrum @xmath50 does not vary significantly within the resonance bandwidth @xmath517 . therefore estimates obtained from eq .
( [ metric ] ) for the number of template shapes required for ringdown searches will only be accurate to within factors of a few , but this is adequate for our purposes . using eq .
( 2.16 ) of ref .
@xcite we find that that the number of required templates is approximately @xmath518 , \label{ringntemplates}\ ] ] where @xmath519 , @xmath520 and @xmath521 are the extremal values of the quality factor and of the black hole mass that define the range of signal searches .
the quantity @xmath522 in the formula ( [ ringntemplates ] ) is the _ minimal match _
parameter introduced by owen ; its meaning is that the event detection rate obtainable by a lattice of templates that have a certain minimal match @xmath522 is smaller than the ideal event detection rate ( obtainable from an infinitely dense lattice of templates ) by the factor @xmath523 @xcite .
we assume @xmath524 as in ref .
@xcite , which corresponds to a @xmath525 loss in event rate due to the coarse gridding of the template space , and take @xmath526 [ which by eq .
( [ fqnrandq ] ) corresponds to @xmath527 . for the initial and advanced ligo interferometers , the mass range to be searched corresponds to roughly @xmath528 and @xmath529 ; this yields from eq .
( [ ringntemplates ] ) @xmath530 this is a rather small number of template shapes compared to the number expected to be necessary for inspiral searches@xcite , and therefore the ringdown search should be fairly easy to implement .
a similarly small number of required template shapes ( @xmath531 ) is obtained for lisa assuming @xmath532 and @xmath533 .
we next discuss the distance to which bbh mergers should be detectable via their ringdown signals . as explained in sec .
[ nonlinear ] , a rough estimate of the appropriate signal - to - noise ratio threshold for detection using one interferometer is @xcite @xmath534 } \label{thr}\ ] ] where @xmath60 is the observation time , @xmath170 is the sampling time and @xmath535 is as defined in section [ nonlinear ] .
in fact coincidencing between the 4 different interferometers in the ligo / virgo network will be carried out , in order to increase detection reliability and combat non - gaussian noise ( see sec .
[ derivesnrformula1 ] ) .
if the noise were exactly gaussian , the appropriate detection criterion would be to demand that @xmath536 where the sum is over the different snrs obtained in each interferometer . in order to combat non - gaussian noise ,
the detection criterion will be modified to require approximately equal snrs in each interferometer : @xmath537 we have chosen a factor of @xmath538 here to be conservative ; it corresponds to combining the outputs of just two interferometers ( say , the two ligo 4 km interferometers ) instead of four interferometers .
taking @xmath539 and @xmath540 yields the estimate @xmath541 for the initial and advanced ligo interferometers .
therefore , from fig . [ initialligosnr ] , we see that the initial ligo interferometers should be able to see ringdowns from equal - mass bbhs in the mass range @xmath542 out to @xmath543 , if the radiation efficiency @xmath488 is as large as we have estimated .
the advanced ligo interferometers , by contrast , should see ringdowns in the mass range @xmath544 out to @xmath19 ( from fig .
[ advancedligosnr ] ) . for non equal - mass bbhs
, these distances are reduced by the factor @xmath545 . for lisa ,
the detection threshold is given by eq .
( [ thr ] ) .
although lisa does incorporate several partially independent interferometers , we have used the noise spectrum ( [ noisespeclisa ] ) which is the effective noise spectrum that applies to the lisa detector as a whole @xcite .
thus it is consistent to treat lisa as one interferometer .
taking @xmath546 and @xmath547 , and using the value @xmath548 estimated above yields @xmath549 .
hence , from fig . [ lisasnr ] , lisa should see ringdowns in the mass range @xmath20 out to @xmath21 .
we next discuss the feasibility of searches for the merger signal . as we have explained , this will be most necessary when the merger snr is larger than both the inspiral and ringdown snrs by factors of a few ( since the fractional loss in event detection rate , if searches for the merger signal are not carried out , is the cube of the ratio of the snr values ) .
consider first the ideal situation in which theoretical template waveforms are available , so that matched filtering can be used in searches . from figs .
[ initialligosnr ] and [ advancedligosnr ] it can be seen that the merger snr values are larger than the inspiral / ringdown values by a factor of up to @xmath550 , in the mass ranges @xmath551 for initial ligo interferometers and @xmath552 for advanced ligo interferometers .
more precisely , in this mass range , @xmath553 } \ , \ , \alt 4 \left ( \epsilon_m \over 0.1 \right)^{1/2 } \ , \left ( \epsilon_r \over 0.03 \right)^{-1/2}. \label{larger}\ ] ] the detection threshold for merger searches should be approximately the same as that for inspiral and merger searches , if the number of template shapes @xmath58 is not too large ( see further discussion below ) . therefore , the gain in event rate over inspiral / ringdown searches , due to searching for merger waves , should vary between @xmath554 and about @xmath555 , depending on the mass of the system , if our estimates of @xmath487 and @xmath488 are approximately correct . clearly , the large possible gain in event rate shows the importance of searching for merger waves @xcite .
moreover , the ratio of @xmath556 of energies radiated in the merger and ringdown stages , which we have estimated to be roughly @xmath557 , is quite uncertain ; we think that in some cases it could be rather larger than this ( notwithstanding the fact that there has been at least one wagered bottle of wine in the community that @xmath558 generically @xcite ) . in such cases
the benefit of searching for the merger waves would be even greater .
it is not clear , however , how feasible it will be to produce a set of numerically generated templates that is complete enough to be used to successfully implement an optimal filtering search .
there may be a very large number of distinct waveform shapes , each of which will require extensive numerical computations .
if both black holes are spinning rapidly , the waveforms could depend in significant and nontrivial ways on 6 distinct angular parameters , suggesting that the number of distinct shapes could be very large .
turn next to the situation where templates are unavailable . consider first band - pass filtering searches . from the estimate @xmath559 of sec .
[ mergernofilters ] , combined with eq .
( [ simp1 ] ) , we obtain the relation @xmath560 therefore , from eq .
( [ larger ] ) it is apparent that the maximum achievable band - pass filtering snr for the merger is likely to be essentially no larger than the snrs obtained from the inspiral / ringdown signals .
however , as explained in secs .
[ assumptionsec ] and [ nonlinear ] , noise monitoring searches will be more efficient than band - pass filtering searches , and will be almost as efficient as matched filtering searches in most cases , when searching for merger waves .
[ for the inspiral waves , by contrast , noise - monitoring searches perform considerably worse than matched filtering searches , since the parameter @xmath152 is much larger for inspiral waves ( @xmath561 ) than it is for merger waves ( @xmath562 ) ] .
the event - detection rate from noise - monitoring is a factor @xmath563 lower than the event rate from matched filtering .
here @xmath65 is the noise - monitoring detection threshold , given by eqs .
( [ e3 ] ) and ( [ e4 ] ) as a function of the parameters @xmath162 , @xmath564 and @xmath152 , and @xmath565 is the matched filtering threshold , given by eq . ( [ mfthreshold0 ] ) as a function of the parameters @xmath58 and @xmath208 . equation ( [ eventrateloss ] ) is based on the assumption that the statistical properties of the noise are gaussian . in reality
the noise in the individual interferometers will have significant non - gaussian components .
coincidencing between several interferometers is expected to reduce the effect of these non - gaussian components , and the efficiency of this reduction may be higher for matched filtering than for noise - monitoring for the reason discussed in sec . [ nonlinear ] .
hence , if coincidencing does not completely efface non - gaussian effects , the loss factor in event rate @xmath566 may be somewhat higher than the estimate ( [ eventrateloss ] ) .
we now estimate the loss factor in event rate @xmath566 . to obtain the most pessimistic estimate possible of the loss in event rate , we use the following assumptions : ( i ) the number of template shapes in the matched filtering search is @xmath567 ; a more realistic larger number would yield a smaller @xmath566 .
( ii ) the number of frequency bins is @xmath427 , twice the upper limit estimated in sec .
[ mergernofilters ] .
( iii ) the number of starting times in the data stream is @xmath568 , corresponding to a sampling time of @xmath569 in a data set of one third of a year .
such a large sampling time would only be appropriate for the largest mass black holes ; more realistic sampling times will be smaller .
larger values of @xmath208 give smaller values of @xmath566 .
( iv ) the parameter @xmath162 in eqs .
( [ e3 ] ) and ( [ mfthreshold0 ] ) is @xmath535 . with these assumptions
we obtain @xmath570 , @xmath571 and the resulting loss in event rate factor is @xmath572 hence , the availability of templates could increase the event detection rate by a factor of at most @xmath23 , and more typically only @xmath573 over noise - monitoring searches .
this discussion assumed that the number of waveform shapes @xmath58 is small .
as this number increases , the performance of matched filtering searches gradually worsens until at @xmath574 , matched filtering performs about as well as noise monitoring ( see sec . [ nonlinear ] ) . from eqs .
( [ numshapes ] ) and ( [ e3 ] ) , the critical value of the number of shapes is @xmath575 for @xmath427 , and @xmath576 for @xmath577 , assuming @xmath578 .
the number of shapes will vary with the signal - to - noise level @xmath48 .
we can define an effective dimension @xmath579 of the manifold of signals by the equation @xmath580 = { 1 \over 2 } n_d(\rho ) \ln \left [ 1 + \rho^2 / n_d(\rho ) \right ] .
\label{ndimdef}\ ] ] from eq .
( [ numshapes ] ) , we see that @xmath581 is the dimension of the equivalent linear space of signals that has the same number of distinguishable waveform shapes with snr @xmath212 as the true ( nonlinear and curved ) manifold of merger signals . in fig .
[ eventgain : fig ] we show the gain factor @xmath566 in event rate as a function of @xmath90 , for the values @xmath582 , @xmath446 and @xmath22 .
the true value of @xmath579 is quite uncertain and will vary to some extent with @xmath48 ; at high signal - to - noise levels it could conceivably be as large as @xmath104 . by combining the gain factor of @xmath583 discussed above with the loss factor @xmath566
, it follows that noise - monitoring searches for merger waves should increase the number of discovered bbhs by a factor of up to @xmath104 over those found from inspiral and ringdown searches . for lisa ,
the expected snrs are so high that the availability of merger templates will likely have no impact on event detection rates . however , for lisa , as well as for ground - based interferometers , accurate merger templates will be vitally important for interpreting the detected waveforms , for extracting all the available information , and for testing general relativity @xcite .
it seems quite likely that the gravitational waves from merging bbh systems will be detected by the ground - based interferometers that are now under construction .
the initial ligo interferometers will be able to detect low mass ( @xmath584 ) coalescences of equal - mass bbhs out to about @xmath32 via their inspiral waves , and higher mass ( @xmath585 ) systems out to about @xmath461 via their ringdown waves .
the advanced ligo interferometers will be able to detect equal - mass bbh coalescences in the mass range @xmath586 to @xmath587 via their inspiral waves , and higher mass ( @xmath588 ) systems to @xmath19 via their ringdown waves . for non - equal mass bbhs
, these distances will be reduced by a factor of @xmath589 for inspiral signals and @xmath590 for ringdown signals .
searches for massive bbhs ( @xmath591 for ligo / virgo ) based on merger waves could increase the range of the interferometers by a additional factor of @xmath592 , without requiring detailed knowledge of the waveform shapes .
it seems likely that bbh coalescences will be detected early in the gradual process of improvement from the first interferometers to the advanced interferometers , rather than later , and there is a strong possibility that they will be the first sources of gravitational radiation to be detected .
theoretical template waveforms obtained from numerical relativity supercomputer simulations will be crucial for analyzing the measured merger waves .
a match of the detected waveform with a predicted waveform would be a triumph for the theory of general relativity and an absolutely unambiguous signature of the existence of black holes .
a complete set of such theoretical templates would also aid the search for bbhs , but not by a large amount .
the space - based interferometer lisa will be an extremely high precision instrument for studying the coalescences of supermassive bbhs .
coalescences with masses in the range @xmath593 should be detectable out to @xmath594 with very large snrs ( @xmath468 ) , via their merger and ringdown waves .
additionally , systems in the mass range @xmath595 should be detected to similar distances and with snrs @xmath596 via their inspiral waves .
we thank kip thorne for suggesting this project to us , for being a constant source of ideas and encouragement along the way , and for detailed comments on the paper s content and presentation .
we also thank patrick brady and david chernoff for some helpful conversations .
this research was supported by nsf grants phy9220644 , phy9408378 , phy9424337 , and phy9514726 , and by nasa grant nagw-4268 .
s.a.h.gratefully acknowledges the support of a national science foundation graduate fellowship , and .f . likewise acknowledges the support of an enrico fermi fellowship . in this appendix
, we give the details of our calculations of the signal - to - noise ratios .
the results of this appendix were used to make figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] in the text , and also the approximate snr formulae ( [ lowmassinitial ] ) and ( [ lowmassadv ] ) given in sec .
[ snrsection ] .
the following three subsections give the calculations for the inspiral waves , the merger waves and the ringdown waves respectively .
to calculate the angle - averaged snr squared for the inspiral , we insert the inspiral energy spectrum ( [ dedfinspiral ] ) and our parameterized model ( [ noisespec ] ) of an interferometer s noise spectrum into eq .
( [ snraveraged ] ) , and integrate from @xmath599 to @xmath600 .
the result is @xmath601 & \alpha f_m \le { f_{\rm merge}}/(1+z)\nonumber\\ { \cal f}_i(m , z , d)\left[9\alpha^{1/3}- 8\left({v\over\alpha}\right)^{1/3 } -\alpha^3\left({f_s\over f_m}\right)^{8/3}\right ] & f_m/\alpha \le { f_{\rm merge}}/(1+z ) < \alpha f_m \nonumber\\ { \cal f}_i(m , z , d ) \left[\alpha^{1/3}\left({\alpha^2\over v}\right ) -\alpha^3\left({f_s\over f_m}\right)^{8/3}\right ] & f_s \le { f_{\rm merge}}/(1+z ) < f_m/\alpha \nonumber\\ 0 & { f_{\rm merge}}/(1+z )
< f_s,\\ \end{array } \right.\ ] ] where @xmath602 and @xmath603^{5/3 } [ 4 \mu / m]}\over 80\pi^{4/3}d(z)^2 h_m^2 f_m^{1/3}}.\ ] ] here @xmath55 is the luminosity distance to the source , @xmath433 , @xmath437 , @xmath434 and @xmath435 are parameters characterizing the detector noise spectrum ( [ noisespec ] ) , and @xmath88 is given by eq .
( [ mergefreq ] ) . inserting the values of the noise spectrum parameters from eq .
( [ noisespecligoinitial ] ) for initial ligo interferometers , we obtain the following numerical values for the snr in the equal - mass case @xmath604 : @xmath605^{1/2 } & ( 1+z)m \le 18\,m_\odot \nonumber\\ 4.7\left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over18\,m_\odot}\right)^{5/6 } \left[1 - 0.71\left({{(1+z)m}\over18\,m_\odot}\right)^{1/3 } \right]^{1/2 } & 18\,m_\odot < ( 1+z)m \le 36\,m_\odot \nonumber\\ 2.7 \left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over36\,m_\odot}\right)^{-1/2 } \left[1 - 0.06 \left({{(1+z)m}\over36\,m_\odot}\right)^{8/3 } \right]^{1/2 } & 36\,m_\odot < ( 1+z)m \le 102\,m_\odot \nonumber\\ 0 & 102 \,m_\odot < ( 1+z)m .
\end{array } \right.\ ] ] for the noise curve parameters ( [ noisespecligoadvanced ] ) appropriate for advanced ligo interferometers we obtain @xmath606^{1/2 } & ( 1+z)m \le 37\,m_\odot \nonumber\\ 43\left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over37\,m_\odot}\right)^{5/6 } \left[1 - 0.65\left({{(1+z)m}\over37\,m_\odot}\right)^{1/3 } \right]^{1/2 } & 37\,m_\odot < ( 1+z)m \le 95\,m_\odot \nonumber\\ 31\left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over95\,m_\odot}\right)^{-1/2 } \left[1 - .021 \left({{(1+z)m}\over95\,m_\odot}\right)^{8/3 } \right]^{1/2 } & 95\,m_\odot < ( 1+z)m \le 410\,m_\odot \nonumber\\ 0 & 410 \,m_\odot < ( 1+z)m .
\end{array } \right.\ ] ] as explained in sec .
[ egs ] , the calculation of the inspiral snr for lisa differs from the other snr calculations in the following way .
if one were to integrate over the whole frequency domain in the interferometer waveband up to @xmath607 ( as is done for the initial and advanced interferometers in ligo ) , in some cases one would obtain the snr for a measurement of several hundred years duration , which is obviously irrelevant .
thus , it is necessary to restrict the integral over frequency in eq .
( [ snraveraged ] ) to the domain that corresponds to , say , one year of observation when calculating inspiral lisa snrs .
using the newtonian relationship for the rate of frequency sweep , we obtain for the frequency at time @xmath60 before merger in the equal - mass case @xmath608^{-3/8}.\ ] ] binaries of redshifted total mass @xmath503 larger than about @xmath609 enter the lisa waveband at @xmath610 less than one year before merger , while binaries of smaller redshifted mass spend more than one year in the lisa waveband . to calculate the snr , we insert eq .
( [ dedfinspiral ] ) into eq .
( [ snraveraged ] ) and integrate numerically from the larger of @xmath433 and @xmath611 to @xmath612 .
the resulting snr values are shown in fig .
[ lisasnr ]
. we also show in fig .
[ lisasnr ] the snr obtained from one year of observation one hundred years before the final merger , obtained by integrating from @xmath613 to @xmath614 , as well as a similar curve for one thousand years prior to merger . equation ( [ inspsnrformula1 ] ) applies to lisa only for @xmath615 . by combining eqs .
( [ inspsnrformula1 ] ) and ( [ noisespeclisa ] ) for @xmath616 together with an approximate fit to fig . [ lisasnr ] for @xmath617 we obtain for the snr from the last year of inspiral in the equal - mass case @xmath618^{1/2 } & 0.5\,{\rm m}m_\odot < ( 1+z)m \le 6.0\,{\rm m}m_\odot\nonumber\\
5.0\times10 ^ 4\left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over6\,{\rm m}m_\odot}\right)^{-1/2 } \left[1 - 0.006\left({{(1+z)m}\over6\,{\rm m}m_\odot } \right)^{8/3}\right]^{1/2 } & 6.0\,{\rm m}m_\odot < ( 1+z ) m \le 41 { \rm m}m_\odot \nonumber \\ 0 & 41 { \rm m}m_\odot < ( 1+z ) m.\end{array } \right.\ ] ] to calculate the merger snr we use the energy spectrum ( [ dedfmerger ] ) and follow the same procedure as above .
the result is @xmath619 & { f_{\rm merge}}/(1+z ) \ge
\alpha f_m \nonumber\\ { \cal f}_m(\epsilon_m , m , z , d ) \left[3\ln v-{{v^3-\kappa^3}\over\kappa^3}\right ] & f_m/\alpha\le{f_{\rm merge}}/(1+z)<\alpha f_m\le{f_{\rm qnr}}/(1+z ) \nonumber\\ ^{\rm i , a}{\cal f}_m(\epsilon_m , m , z , d ) \left[2 - { \alpha^6\over v^3 } - { v^3\over\kappa^3 } + 6 \ln \alpha\right ] & { f_{\rm merge}}/(1+z)\le f_m/\alpha<\alpha f_m \le { f_{\rm qnr}}/(1+z )
\nonumber\\ ^{\rm l}{\cal f}_m(\epsilon_m , m , z , d)\left[3\ , \ln \kappa \right ] & f_m/\alpha \le f_{\rm merge}/(1+z ) < f_{\rm qnr}/(1+z ) \le \alpha f_m \nonumber\\ ^{\rm i}{\cal f}_m(\epsilon_m , m , z , d ) \left[2 - \left({\alpha f_s\over f_m}\right)^3 - { v^3\over \kappa^3 } + 6 \ln \alpha\right ] & { f_{\rm merge}}/(1+z ) \le f_s < f_m/\alpha < \alpha f_m \le { f_{\rm qnr}}/(1+z ) \nonumber\\ ^{\rm a , l}{\cal f}_m(\epsilon_m , m , z , d ) \left[1 + 3\ln\left({\kappa\alpha^2\over v}\right ) -{\alpha^6\over v^3}\right ] & f_s \le { f_{\rm merge}}/(1+z ) < f_m/\alpha \le { f_{\rm
qnr}}/(1+z ) \nonumber\\ ^{\rm i , a}{\cal f}_m(\epsilon_m , m , z , d ) \left[1 - \left({\alpha f_s\over f_m}\right)^3 + 3\ln\left({\kappa\alpha^2\over v}\right)\right ] & { f_{\rm merge}}/(1+z)\le f_s <
f_m/\alpha\le{f_{\rm qnr}}/(1+z ) \nonumber\\ ^{\rm l}{\cal f}_m(\epsilon_m , m , z , d ) \left[\alpha^6 v^{-3 } ( \kappa^3 - 1)\right ] & f_s \le f_{\rm merge}/(1+z ) < f_{\rm
qnr}/(1+z ) \nonumber\\ { \cal f}_m(\epsilon_m , m , z , d ) \left[\left({\kappa\alpha^2\over v}\right)^3 -\left({\alpha f_s\over f_m}\right)^3 \right ] & { f_{\rm merge}}/(1+z)\le f_s<{f_{\rm qnr}}/(1+z ) \le f_m/\alpha \nonumber\\ 0 & { f_{\rm
qnr}}/(1+z ) < f_s . \end{array } \right.\ ] ] here @xmath620 is given by eq .
( [ vdef ] ) , @xmath487 is the fraction of total mass energy radiated during the merger ( which we have also denoted by @xmath621 in the body of the paper ) , @xmath622 , and @xmath623 ^ 2}\over 15\pi^2d(z)^2 h_m^2 { f_{\rm merge}}(\kappa-1)}.\ ] ] lines marked with the superscript `` i '' turn out to hold for the initial ligo interferometer parameters ; those with `` a '' hold for advanced ligo interferometer parameters ; and those with `` l '' hold for lisa . using the numerical values of the noise curve parameters ( [ noisespecligoinitial ] ) for initial ligo interferometers , and eqs .
( [ mergefreq ] ) , ( [ fqnrdef ] ) , ( [ vdef ] ) , and ( [ mergesnrformula1 ] ) we find for the initial ligo interferometers in the equal - mass case @xmath624^{1/2 } & \!\!18\,m_\odot < ( 1+z)m \le 36\,m_\odot\nonumber\\ 6.1\left({\epsilon_m\over 0.1}\right)^{1/2 } \left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over36\,m_\odot}\right ) \nonumber \\
\mbox{~~~~ } \times \left[1 + .23 \left({{(1+z)m}\over36\,m_\odot}\right)^{-3 } - 0.007\left({{(1+z)m}\over36\,m_\odot}\right)^3\right]^{1/2 } & \!\!36\,m_\odot < ( 1+z)m \le 102\,m_\odot\nonumber\\ 17.3\left({\epsilon_m\over 0.1}\right)^{1/2 } \left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over102\,m_\odot}\right ) \nonumber \\ \mbox{~~~~ } \times \left[1 - .17 \left({{(1+z)m}\over102\,m_\odot}\right)^3\right]^{1/2 } & \!\!102\,m_\odot < ( 1+z)m \le 118\,m_\odot\nonumber\\ 9.9 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over118\,m_\odot}\right ) \left[1 - 3.1\ln\left({{(1+z)m}\over230\,m_\odot}\right)\right]^{1/2 } & \!\!118\,m_\odot
< ( 1+z)m \le 230\,m_\odot\nonumber\\ 20 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({200\,{\rm mpc}\over d(z)}\right ) \left({{(1+z)m}\over230\,m_\odot}\right)^{-1/2 } \left[1 - 0.04\left({{(1+z)m}\over230\,m_\odot}\right)^3 \right]^{1/2 } & \!\!230\,m_\odot < ( 1+z)m \le 660\,m_\odot\nonumber\\ 0 & 660\,m_\odot < ( 1+z ) m. \end{array } \right.\ ] ] similarly using eq .
( [ noisespecligoadvanced ] ) we find for advanced ligo interferometers @xmath625^{1/2 } & \!\!37\,m_\odot < ( 1+z)m \le 95\,m_\odot\nonumber\\ 76 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over95\,m_\odot}\right ) \nonumber \\
\mbox{~~~~ } \times \left[1 - { 0.21}\left({{(1+z)m}\over95\,m_\odot}\right)^{-3 } -{0.013}\left({{(1+z)m}\over95\,m_\odot}\right)^3\right]^{1/2 } & \!\!95\,m_\odot < ( 1+z)m \le 240\,m_\odot\nonumber\\ 88 \left({\epsilon_m\over 0.1}\right)^{1/2 }
\left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over240\,m_\odot}\right ) \nonumber \\
\mbox{~~~~ } \times \left[1 - 3\ln\left({{(1+z)m}\over620\,m_\odot}\right ) -0.061\left({{(1+z)m}\over240\,m_\odot}\right)^{-3}\right]^{1/2 } & \!\!240\,m_\odot < ( 1+z)m \le 410\,m_\odot\nonumber\\ 150\left({\epsilon_m\over 0.1}\right)^{1/2 }
\left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over410\,m_\odot}\right ) \left[1 - 3.0\ln\left({{(1+z)m}\over620\,m_\odot}\right)\right]^{1/2 } & \!\!410\,m_\odot < ( 1+z)m \le 620\,m_\odot\nonumber\\ 220 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over
d(z)}\right ) \left({{(1+z)m}\over620\,m_\odot}\right)^{-1/2 } \left[1 - 0.013 \left({{(1+z)m}\over620\,m_\odot}\right)^3 \right]^{1/2 } & \!\!620\,m_\odot < ( 1+z)m \le 2600\,m_\odot\nonumber\\ 0 & \!\!2600\,m_\odot < ( 1+z ) m. \end{array } \right.\ ] ] finally , using the parameters ( [ noisespeclisa ] ) appropriate for lisa , we obtain @xmath626^{1/2 } & \!\!2.0\times10 ^ 5m_\odot < ( 1+z)m \le 1.3\,{\rm m}m_\odot\nonumber\\ 2.8\times10 ^ 4 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over1.3\,{\rm m}m_\odot}\right ) & \!\!1.3\,{\rm m}m_\odot < ( 1+z)m \le 6.0\,{\rm m}m_\odot \nonumber\\ 3.4\times10 ^ 4 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over6.0\,{\rm m}m_\odot}\right ) \nonumber \\ \mbox{~~~~ } \times \left[1 - 3\ln\left({{(1+z)m}\over39\,{\rm m}m_\odot}\right ) - \left({{(1+z)m}\over6.0\,{\rm m}m_\odot } \right)^{-3}\right]^{1/2 } & \!\!6.0\,{\rm m}m_\odot < ( 1+z)m \le 39\,{\rm m}m_\odot\nonumber\\ 3.6\times10 ^ 5 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over d(z)}\right ) \left({{(1+z)m}\over39\,{\rm
m}m_\odot}\right)^{-1/2 } & \!\!39\,{\rm m}m_\odot
< ( 1+z)m \le 41\,{\rm m}m_\odot\nonumber\\ 3.4\times10 ^ 5 \left({\epsilon_m\over 0.1}\right)^{1/2 } \left({1\,{\rm gpc}\over d(z)}\right ) \nonumber \\ \mbox{~~~~ } \times \left({{(1+z)m}\over41\,{\rm m}m_\odot } \right)^{-1/2 } \left[1 - 3.8 \times 10^{-3 } \left({(1+z)m\over 41\,{\rm m}m_\odot}\right)^{3 } \right]^{1/2 } & \!\!41\,{\rm m}m_\odot < ( 1+z)m \le 260\,{\rm m}m_\odot\nonumber\\ 0 & \!\!260 \,{\rm m } m_\odot < ( 1+z)m .
\end{array } \right.\ ] ] the ringdown snrs are calculated a little differently from the inspiral and merger snrs .
first , we use the effective energy spectrum ( [ dedfqnr ] ) which yields an estimate of the true snr obtainable from the model waveform ( [ qnrwaveform ] ) that is accurate to within a few tens of percent ( see appendix [ ringdown3 ] ) .
second , the integral over frequency in the snr formula ( [ snraveraged ] ) with the noise spectrum ( [ noisespec ] ) and the energy spectrum ( [ dedfqnr ] ) can not easily be evaluated analytically .
hence , we calculated this integral numerically to produce the plots of ringdown snr versus bbh mass shown in figs .
[ initialligosnr ] , [ advancedligosnr ] and [ lisasnr ] . in the remainder of this appendix
we derive approximate formulae for the ringdown snr as a function of mass , by approximating the ringdown energy spectrum as a delta function at the ringdown frequency [ _
( [ dedfqnrapprox ] ) ] .
this approximation yields ( see appendix [ ringdown3 ] and ref .
@xcite ) @xmath627 ^ 2 \over 20 \pi^2 d(z)^2 { f_{\rm qnr } } s_h[{f_{\rm qnr}}/(1+z)]}. \label{qnrsnr}\ ] ] using eq .
( [ fqnrandq ] ) and the relation ( [ qnrenergy ] ) between the dimensionless coefficient @xmath383 and the radiated energy we can rewrite formula ( [ qnrsnr ] ) as @xmath628 } \
, \left [ { ( 1+z ) m \over
d(z ) } \right]^2 \ , \left [ { 4 \mu \over m } \right]^2 , \label{qnrsnrii}\ ] ] where @xmath488 is the fraction of the total mass energy radiated in the ringdown , and @xmath629 an equivalent formula was previously obtained by finn @xcite .
we find the following numerical result when we insert our assumed values @xmath630 and @xmath75 for the ringdown signal together with the parameters for the initial ligo interferometer noise curve in the equal - mass case : @xmath631 the corresponding formulae for advanced ligo interferometers are @xmath632 finally , the corresponding formulae for lisa are @xmath633 by comparing eqs .
( [ qnrsnr2 ] ) ( [ qnrsnr4 ] ) with figs .
[ initialligosnr ] [ lisasnr ] it can be seen that the delta - function energy spectrum approximation is fairly good except for @xmath634 for advanced ligo interferometers and @xmath635 for lisa .
the approximation fails to capture the high mass tails of the snr curves .
there is a subtlety in calculating the snr for the ringdown waves , related to the fact that the snr squared does not accumulate locally in the time domain . in order to explain this subtlety ,
let us focus not on the angle - averaged snr squared which was our main concern in the body of the paper , but rather on the snr squared obtained in one interferometer from a specific source with specific relative angular orientations .
in this case the waveform @xmath49 seen in the interferometer , for @xmath376 , is of the form @xmath636 for some constants @xmath637 and @xmath638 , while @xmath49 is the ( unknown ) merger waveform for @xmath639 .
let us also focus first on the simple , idealized case of white noise , @xmath640 .
then , the snr squared ( [ snr ] ) accumulates locally in time : @xmath641 hence , for white noise , the snr squared from the ringdown is clearly unambiguously given by @xmath642 \nonumber \\ \mbox { } & \approx & { h_0 ^ 2 \tau \over 2 s_h } \left [ 1 + o(1/q)\right ] , \label{exactwhite}\end{aligned}\ ] ] where @xmath643 .
now consider the case when the noise is not exactly white .
naively , we expect that in the fourier domain the energy spectrum of the ringdown signal will be a resonance curve that peaks at @xmath644 with width @xmath645 .
thus , for large @xmath198 we would expect that most of the snr squared will be accumulated near @xmath644 , unless the noise spectrum varies very strongly with frequency .
moreover , if the noise spectrum @xmath50 does not vary much over the bandwidth @xmath645 of the resonance peak , then we would expect the formula ( [ exactwhite ] ) to be valid to a good approximation , with @xmath646 replaced by @xmath647 .
we show below that this is indeed the case : under such circumstances , eq .
( [ exactwhite ] ) is fairly accurate , and the resulting approximate ringdown snr is embodied in our approximate delta - function energy spectrum ( [ dedfqnrapprox ] ) and in eqs .
( [ qnrsnr ] ) - ( [ qnrsnr3 ] ) of appendix [ appsnr ] @xcite . in many cases of interest
, it will indeed be true that most of the snr squared for ringdown waves will be accumulated in the vicinity of the resonance peak , so that the snr will approximately be given by eq .
( [ exactwhite ] ) .
however , this will not always be the case .
for instance , suppose that we were lucky enough that two @xmath648 black holes were to merge at the center of our own galaxy .
would such an event be detectable by advanced ligo interferometers ?
clearly , most of the power in the ringdown waves in this case would be far below the ligo / virgo waveband .
however , given that the merger is only at @xmath649 , one might hope to be able to detect the tail of the ringdown waves that extends upwards in frequency into the ligo / virgo waveband . or ,
consider the detectability of a ringdown of a nearby @xmath650 black hole by lisa . in this case
most of the ringdown power is concentrated at frequencies above lisa s optimum waveband , and the detectability of the signal is determined by the amount of power in the low frequency tail of the ringdown . in such cases , it is clearly necessary to understand the power carried in the ringdown waves at frequencies far from the resonant frequency . normally , such an understanding is obtained simply by taking a fourier transform of the waveform @xmath49 . in the case of ringdown waves from bbh mergers , however , the waveform for @xmath639 is the unknown , merger waveform . in order to obtain
the snr squared from the ringdown signal alone , one might guess that the appropriate thing to do is to take @xmath651 for @xmath639 , and insert this together with eq .
( [ ringdownbasic ] ) into the standard formula ( [ snr ] ) for the signal to noise squared . however , the resulting energy spectrum has unrealistic high frequency behavior due to a discontinuity in @xmath49 at @xmath377 ( or a discontinuity in @xmath652 at @xmath377 in the case @xmath653 ) , and the resulting snrs can in some cases differ from the correct values ( see below ) by factors @xmath654 .
other choices for @xmath49 for @xmath639 [ for instance @xmath655 get around this problem but instead have unrealistic low frequency behavior . in any case , it is clear that these choices are somewhat _ ad hoc _ and that there should be some more fundamental , unique way to calculate the snr .
we now explain the correct method to calculate the snr .
the question that effectively is being asked is : what is the probability that there is a ringdown waveform present in the data stream , starting at ( say ) @xmath377 ? this probability is to be calculated given only the data from @xmath376 , without using the measured data from @xmath639 which is contaminated by the unknown merger waveform . to do this
one must effectively integrate over all possible realizations of the noise for @xmath639 .
the necessity for such an integration is illustrated by the following simple analogy .
suppose that one is measuring two real variables , @xmath656 ( `` waveform for positive @xmath203 '' ) and @xmath657 ( `` waveform for negative @xmath203 '' ) , and that the measured values of these variables are @xmath658 and @xmath659 .
suppose that because of the noise in the measurement process , the probability distribution for the true values of these parameters given their measured values is @xmath660 \bigg\}. \label{pbasic}\end{aligned}\ ] ] thus , @xmath656 and @xmath657 are gaussian distributed about their means @xmath658 and @xmath659 , and they are correlated .
if we assume that @xmath661 [ analogous to assuming @xmath662 for @xmath639 ] , we obtain for the probability distribution for @xmath656 @xmath664 where @xmath665 .
by contrast , if we instead calculate the probability distribution for @xmath656 alone by integrating over @xmath657 we find @xmath666 where @xmath667 .
it is clear in this simple example that one should use the reduced distribution ( [ pgood ] ) rather than the distribution ( [ pbad ] ) .
note also that the widths of the probability distributions ( [ pbad ] ) and ( [ pgood ] ) are different , and that the correct distribution ( [ pgood ] ) could not have been obtained from the joint distribution ( [ pbasic ] ) for any assumed choice of @xmath657 .
turn now to the analogous situation for random processes .
if @xmath120 is the interferometer noise , let @xmath668 denote the autocorrelation function .
define the inner product @xmath669 on the space of functions @xmath49 for @xmath376 , where the kernel @xmath670 is determined from @xmath671 for @xmath672 .
the quantity @xmath670 is analogous to the modified width @xmath673 in eq .
( [ pgood ] ) above . using the inner product
( [ innerhalfline ] ) , the usual theory of optimal filtering @xcite can be applied to random processes on the half line @xmath376 .
thus , if @xmath64 is the interferometer output and @xmath49 is the waveform ( [ ringdownbasic ] ) , the detection statistic is @xmath674 , and the snr squared for the measurement is @xmath675 ^ 2 \over e[y^2 ] - e[y]^2 } \nonumber \\ \mbox { } & = & \langle h | h \rangle \nonumber \\ \mbox { } & = & \int_0^\infty dt \int_0^\infty dt^\prime k(t , t^\prime ) h(t ) h(t^\prime ) , \label{fullans}\end{aligned}\ ] ] where @xmath676 $ ] means expectation value . if we define @xmath677 and @xmath678 the snr squared can be rewritten as @xmath679 note that the fourier transform @xmath680 of @xmath670 is not proportional to @xmath681 but instead is in general _ non - diagonal _ in frequency .
the formula ( [ finalans ] ) resolves the ambiguities discussed above in the method of calculating the ringdown snr ; the result does not require a choice of the waveform @xmath49 for @xmath639 .
unfortunately , the final answer ( [ finalans ] ) is complicated in the sense that it can not be expressed in the form ( [ snraveraged ] ) for any effective energy spectrum @xmath53 .
this is somewhat inconvenient for the purposes of this paper : the wave s energy spectrum is a useful and key tool for visualizing and understanding the snrs . clearly , an approximate , effective energy spectrum ( to the extent that one exists ) would be very useful .
we now turn to a derivation of such an approximate , effective energy spectrum , namely the spectrum ( [ dedfqnr ] ) which is used throughout the body of this paper .
we start our derivation by describing an alternative method of calculating the exact ringdown snr given by eqs .
( [ ringdownbasic ] ) and ( [ fullans ] ) .
it is straightforward to show that the quantity ( [ fullans ] ) can be obtained by ( i ) choosing _ any _ waveform @xmath49 for @xmath639 , ( ii ) calculating the snr from the usual formula ( [ snr ] ) , and ( iii ) minimizing over all choices of the function @xmath49 on the negative real axis .
we have experimented with several choices of @xmath682 for @xmath376 , namely @xmath683 , @xmath684 , @xmath685 .
we found that the snr obtained by minimizing over these choices is always ( for the entire black hole mass ranges discussed in sec .
[ snrsection ] ) within a few tens of percent of the snr obtained from the following prescription : ( i ) assume that @xmath49 for negative @xmath203 is identical to the waveform for positive @xmath203 except for the sign of @xmath686 ; _ i.e. _ , that @xmath687 for positive and negative @xmath203 .
( ii ) calculate the total snr using the standard formula ( [ snr ] ) .
( iii ) divide by a correction factor of @xmath538 in amplitude to compensate for the doubling up .
this prescription gives the correct , exact result ( [ fullans ] ) for white noise . for more realistic noise curves ,
the errors of a few tens of percent resulting from this prescription are unimportant compared to the uncertainty in the overall amplitude @xmath383 of the ringdown signal . moreover , the resulting snr values multiplied by @xmath538 are an upper bound for the true snr ( since if our _ ad hoc _ choice of @xmath49 for @xmath688 happened to be exactly right , then the prescription would underestimate the snr by @xmath538 ) .
we now explain how to obtain the energy spectrum ( [ dedfqnr ] ) from the above approximate prescription . from eqs .
( [ decompos ] ) and ( [ qnrwaveform ] ) it can be seen that the waveform as seen in one interferometer , before angle averaging , is given by eq .
( [ ringdownbasic ] ) with @xmath689 \ , { _ 2s^2_2(\iota,\beta , a ) } e^{i \varphi_0}. \label{ringdownconsts}\ ] ] here the angles @xmath690 , @xmath691 , @xmath228 , @xmath380 and @xmath238 have the meanings explained in sec . [ derivesnrformula ] .
let us now insert the waveform ( [ ringdownbasic ] ) into the formula ( [ fullans ] ) for the exact snr , and then average over the angles @xmath690 , @xmath691 , @xmath228 , @xmath380 and @xmath238 using eqs .
( [ angleaveragedefine ] ) and ( [ spheroidal ] ) .
this yields for the angle - averaged , exact snr squared @xmath692 ^ 2 \rangle = { 1 \over 20 \pi } \left [ \rho_{\rm exact}[h_{+,0}(t)]^2 + \rho_{\rm exact}[h_{\times,0}(t)]^2 \right ] , \label{angleaveragedexact}\ ] ] where @xmath693 $ ] denotes the exact snr functional ( [ fullans ] ) and @xmath694 for @xmath376 . now , for each of the two terms on the right hand side of eq .
( [ angleaveragedexact ] ) , we make the approximation discussed above consisting of using eqs .
( [ snr ] ) and ( [ ringdownansatz ] ) and dividing by 2 .
this yields @xmath692 ^ 2 \rangle \approx { 1 \over 10 \pi } \int_0^\infty df \ , { \left [ | { \tilde h}_{+,0}(f)|^2 + | { \tilde h}_{+,0}(f)|^2 \right ] \over s_h(f)},\ ] ] where it is understood that @xmath695 and @xmath696 have been extended to negative @xmath203 in the manner of eq .
( [ ringdownansatz ] ) . finally , evaluating the fourier transforms yields an angle averaged snr squared of the form ( [ snraveraged ] ) , with energy spectrum given by eq .
( [ dedfqnr ] ) .
a. abramovici , w. e. althouse , r. w. p. drever , y. grsel , s. kawamura , f. j. raab , d. shoemaker , l. sievers , r. e. spero , k. s. thorne , r. e. vogt , r. weiss , s. e. whitcomb , and m. e. zucker , _ ligo : the laser interferometer gravitational - wave observatory _ ,
science * 256 * , 325 ( 1992 ) .
p. bender , a. brillet , i. ciufolini , k. danzmann , r. hellings , j. hough , a. lobo , m. sandford , b. schutz , and p. touboul .
_ lisa , laser interferometer space antenna for gravitational wave measurements : esa assessment study report_. r. reinhard , estec , 1994 .
p. bender , i. ciufolini , k. danzmann , w. folkner , j. hough , d. robertson , a. rdiger , m. sandford , r. schilling , b. schutz , r. stebbins , t. summer , p. touboul , s. vitale , h. ward , and w. winkler , _ lisa : laser interferometer space antenna for the detection and observation of gravitational waves _ ,
pre - phase a report , december 1995 ( unpublished ) .
l. s. finn , phys .
d * 46 * , 5236 ( 1992 ) .
see also f. echeverria , unpublished phd thesis , california institute of technology , addendum to ch . 2 ( 1993 ) .
this is an addendum to ref .
@xcite which shows that the methods used to calculate measurement errors in refs .
@xcite and @xcite are equivalent .
b. w. murphy , h. n. cohn , p. m. lugger and j. d. dull , bull .
* 185 * ( 1994 ) ; k. gephardt _ et al .
_ , _ fabry - perot observations of globular clusters iii : m15 _ , preprint 1996 ( astro - ph/9612116 ) . by post-@xmath697-newtonian order , we mean @xmath698 $ ] in secular terms and @xmath699 $ ] in dissipative , radiation reaction terms ( @xmath700 orders beyond the leading quadrupole radiation reaction term ) .
this is the standard counting method for post - newtonian waveforms .
families of template waveforms optimized for signal searches are in the early stages of development ; see b. s. sathyaprakash , phys .
d * 40 * , r7111 ( 1994 ) ; b. s sathyaprakash and s. v. dhurandhar , phys .
d * 44 * , 3819 ( 1991 ) ; * 49 * , 1707 ( 1994 ) ; a. krlak , k. d. kokkotas , and g. schfer , _ proceedings of the 17th texas symposium on relativistic astrophysics , ann . n. y. acad .
sci _ , 1995 ; t. a. apostolatos , phys .
d * 52 * , 605 ( 1995 ) .
even if such methods are not found , post - newtonian templates should be adequate for searching for inspiral waves even in the ibbh regime ; ibbh templates will likely be needed only as a foundation for extracting the waves information .
we also note that the method of pad approximants has recently been found to improve significantly the convergence of the post - newtonian series @xcite ; the implications of this for the ibbh regime are not yet clear .
one can convince oneself of the likelihood of this cleanliness by considering various possible sources of perturbations to the gravitational - wave signal .
accretion disks will have masses much smaller than the black hole masses ( for the equal mass coalescences that we are considering ) ; and moreover for solar mass bbhs any accretion disks from previous phases in the binary s evolution will be long gone before the black holes coalesce .
however , one case where a possible perturbing effect on the gravitational - wave signal may be observable is the stochastic influence of a surrounding stellar cluster on the orbits of members of a smbh binary during the early stages of the inspiral .
while strictly speaking this could make our assumption of the applicability of wiener optimal filtering invalid for some smbh inspirals , such inapplicability is not a cause for concern : the snrs are typically so high anyway that it should be possible to easily detect both the coherent portions of the inspiral radiation and any isolated changes in phase due to close - encounter perturbations .
the grand challenge alliance is an nsf funded collaboration among physicists and computer scientists at eight institutions : university of texas , austin ; ncsa / university of illinois ; university of north carolina , chapel hill ; cornell university ; syracuse university npac ; university of pittsburgh ; northwestern university ; and penn state university .
is richard matzner .
more information about the grand challenge alliance and their efforts to solve the binary black hole problem can be found at the www address http://jean-luc.ncsa.uiuc.edu/gc/gc.html ; see also ref .
@xcite for a review .
our convention for the normalization of the spectral noise density @xmath50 is such that @xmath701 where @xmath120 is the detector noise ( `` one - sided spectral noise density '' ) .
this determines the constant @xmath23 appearing in eq .
( [ snr ] ) .
note that the alternative convention used in refs .
@xcite results in a factor @xmath573 instead of @xmath23 in eq .
( [ snr ] ) . in eq .
( 29 ) of ref .
@xcite , the spectral noise density that is used is the one - sided spectral noise density ( as in this paper ) , and thus the overall factor in front of the integral should be @xmath23 , not @xmath573 .
note that we distinguish between the number of statistically independent waveform shapes @xmath58 in the manifold of waveforms that one is trying to detect using the matched filtering process , and the actual number of templates @xmath514 used in the computations .
the number @xmath58 is the volume of the subset of the manifold of signals with snr @xmath212 , in the natural metric induced by the statistical properties of the noise ; thus @xmath702 depends on @xmath48 . the number @xmath703 is defined in sec . [ qnrsearches ] . in general @xmath704 since the optimum computational method will have less than one template per statistically independent waveform shape . for the purposes of calculating snr thresholds , @xmath705 to a good approximation .
for instance , if the post - newtonian expansion is taken to order @xmath706 , we would find in the equal mass case ( see , _
e.g. _ , ref .
@xcite ) @xmath707.\ ] ] the correction to the newtonian formula is @xmath708 at @xmath709 ; when taken to higher order , there will be additional corrections . for the purposes of our discussion ,
these corrections are not important an estimate of the snr that is accurate to a few tens of percent is adequate . also , the loss in snr due the phase lag between the approximate , post - newtonian templates and the true , general - relativistic signal will be small @xcite . s. r. brandt and e. seidel , phys .
d * 52 * , 870 ( 1995 ) ( gr - qc/9412073 ) ; the run labeled r3 of a distorted kerr black hole with @xmath711 yields an energy radiated in ringdown waves of @xmath399 ( from table iv , and also from fig .
18 which shows that most of the radiated energy can be ascribed to the ringdown waves ) .
when we say that a source is visible out to a distance @xmath106 , we mean that the rate of detection of events is the roughly the same as the rate of occurrence of events within a sphere of radius @xmath106 .
note that some events within this sphere will be missed , and some outside this sphere will be detected , due to the beaming and orientation effects discussed in sec .
[ derivesnrformula ] .
rare , optimally oriented sources will be visible out to several times @xmath106
. it would be quite simple to improve on this filter .
for example , one could divide it by the noise spectral density in order to weight the filter according to the instrumental noise characteristics , _
i.e. _ , pre - whiten the data stream .
the main result of our analysis , the formula ( [ simp ] ) , would also approximately apply to such an improved filtering method . during the inspiral phase , the timescale over which the frequency of the waves changes is much longer than the instantaneous period , so it makes sense to talk about the frequency of the waves at a certain time .
in particular , we can ( roughly ) talk either about the time or the frequency at which the inspiral ends the notion of instantaneous frequency is just beginning to break down then . since @xmath162 is a scalar quantity , it can only depend on scalar products of @xmath278 , @xmath279 and @xmath280 and not on these vectorial quantities themselves .
therefore the number of independent variables that @xmath162 depends on is six . the inequality @xmath330 used in deriving eq .
( [ bound1 ] ) should really be @xmath713 , since it is the mass of the final black hole and not the initial total mass of the system that is relevant .
an argument parallel to that leading to eq .
( [ eest1 ] ) then yields the inequality @xmath714/4 $ ] or @xmath715 , where @xmath716 . hence ,
even if only @xmath717 of the emitted radiation is quadrupolar , the estimate @xmath718 used in sec .
[ mergerphase ] is probably still fairly realistic .
actually the qnr metric perturbation will vanish at the event horizon ; one could instead use the boundary of the ergo - region .
also the ratio @xmath566 of the circumferences will actually be gauge dependent . however , if we expand @xmath566 as @xmath719 , then the coefficient @xmath437 will be gauge invariant and we could take @xmath720 .
more specifically , an ellipsoidal solid body of mass @xmath25 of uniform density whose axes are given by @xmath721 gives rise to a luminosity @xmath722 ; comparing this to the luminosity @xmath723 for the waveform ( [ qnrwaveform ] ) at @xmath377 , and identifying @xmath724 and taking @xmath725 for @xmath726 say yields @xmath727 .
this crude estimation method clearly does not take into account the potential barrier through which the radiation must propagate ( among other things ) .
we imagine that a more precise calculation will find a somewhat smaller value of @xmath383 .
this error can be seen by inserting the parameter values @xmath728 , @xmath729 , and @xmath730 given in ref .
@xcite into the standard equation for suspension thermal noise due to viscous damping , as given in , _
e.g. _ , eq . ( 4.3 ) of ref . @xcite .
the resulting noise level is 3 times larger than the noise level shown in fig . 7 of ref .
@xcite for @xmath731 .
this error was not made in the corresponding figure v.4 of the 1989 ligo proposal @xcite .
[ k. s. thorne , private communication ] r.e .
al . , `` proposal to the national science foundation for the construction , operation , and supporting research and development of a laser interferometer gravitational - wave observatory '' , ( california institute of technology , 1989 ) , unpublished .
the @xmath732 piece in the noise curve happens to have the same frequency dependence that the test masses thermal noise would have if this noise were due to structural damping @xcite .
we emphasize that it is _ not _ our aim to alter the standard ligo advanced interferometer noise spectrum by assuming that structural internal thermal noise is dominant between @xmath733 and @xmath734 ; rather , the @xmath732 region is just a way to smooth the transition between the suspension thermal noise dominated and laser shot noise dominated pieces of the noise spectrum . see eq .
( 32 ) of ref .
note also that our @xmath109 is similar to the quantity @xmath735 defined in ref .
@xcite , which is the same as the quantity denoted @xmath736 in refs .
these quantities differ from @xmath109 by the factor @xmath737 , where @xmath738 is the number of independent starting times per third of a year that must be tried in the optimal filtering search process ; see eq .
( 34 ) of ref .
we omit this factor in our definition of @xmath449 since in paper ii @xcite we will be discussing snrs in situations when the waves have already been detected and where the factor @xmath739 is not relevant .
the factor instead appears in our snr thresholds discussed in sec .
[ detection ] .
note also that our signal - to - noise ratios differ from those in refs .
@xcite by an overall factor of @xmath222 , as discussed after eq .
( [ snraveraged ] ) above .
we obtained the metric ( [ metric ] ) in the following way : ( i ) we took the large @xmath198 limit of the matrix @xmath740 of ref .
@xcite , which was calculated assuming white noise , ( ii ) we inverted this matrix , took a @xmath741 sub - block and re - inverted , thus eliminating ( effectively integrating out ) the variables @xmath742 and @xmath60 .
this procedure for eliminating the time - of - arrival variable @xmath60 is equivalent to the procedure embodied in eq .
( 2.28 ) of ref .
( iii ) we divided by 2 in order to accord with the definition ( 2.12 ) of ref .
@xcite .
in the situation where merger templates are available , the optimum search method would in fact be a combination of inspiral , merger and ringdown searches , in which the entire inspiral merger
ringdown signal as a whole is searched for using matched filtering . to see that angle - averaging the approximate formula ( [ exactwhite ] ) yields the formula ( [ qnrsnr ] ) for the ringdown snr ,
corresponding to the approximate delta function energy spectrum ( [ dedfqnrapprox ] ) , proceed as follows .
starting from eq .
( [ ringdownconsts ] ) , average over the angles @xmath690,@xmath691,@xmath228,@xmath380 and @xmath238 using eqs .
( [ angleaveragedefine ] ) and ( [ spheroidal ] ) .
this yields @xmath745 substitute this into eq .
( [ exactwhite ] ) , make the replacement @xmath746 , and make the appropriate substitutions to account for cosmological effects as explained in sec .
[ derivesnrformula ] to obtain eq .
( [ qnrsnr ] ) . | we estimate the signal - to - noise ratios ( snrs ) that one would expect to measure from coalescing binary black hole ( bbh ) systems for the following three broadband gravitational - wave observatories : initial and advanced ground - based interferometers ( ligo / virgo ) and space - based interferometers ( lisa ) .
we focus particularly on the highly relativistic and nonlinear _ merger _ portion of the gravitational - wave signal , which comes after the adiabatic _ inspiral _ portion and before the _ ringdown _ portion due to the quasinormal ringing of the final kerr black hole .
ground - based interferometers can do moderate snr ( a few tens ) , moderate accuracy studies of the dynamics of merging black holes in the mass range @xmath0 to @xmath1 .
lisa , by contrast , can do high snr ( a few @xmath2 ) , high - accuracy studies of bbh systems in the mass range @xmath3 , where @xmath4 is the binaries cosmological redshift .
our estimated snrs suggest that coalescing black holes might well be the first sources detected by the ligo / virgo network of ground - based interferometers .
because of their larger masses , they can be seen out to much greater distances ( up to @xmath5 for @xmath6 for the initial ligo interferometers ) than coalescing neutron star binaries ( heretofore regarded as the `` bread and butter '' workhorse source for ligo / virgo , visible to @xmath7 by the initial ligo interferometers ) .
low - mass bbhs ( @xmath8 for the first ligo interferometers ; @xmath9 for the advanced ; @xmath10 for lisa ) are best searched for via their well - understood inspiral waves ; more massive bbhs must be searched for via their more poorly understood merger waves and/or their well - understood ringdown waves
. a search for low - mass bbhs based on the inspiral waves , at a sensitivity level roughly half way between the first ligo interferometers and the advanced ligo interferometers , should be capable of finding bbhs out to @xmath11 .
a search for massive bbhs based on the ringdown waves can be performed using the method of matched filters .
if one wants the reduction in the event rate due to the discreteness of the template family to be no more than @xmath12 , then the number of independent templates needed in the search will only be about @xmath13 or less .
such a search with the first ligo interferometers should be capable of finding bbhs in the mass range from about @xmath14 to about @xmath15 out to @xmath16 ; with advanced ligo interferometers , from about @xmath17 to about @xmath18 out to @xmath19 ; and with lisa , bbhs with @xmath20 should be visible out to @xmath21 .
the effectiveness of a search based on the merger waves will depend on how much one has learned about the merger waveforms from numerical relativity simulations . with only a knowledge of the merger waves range of frequency bands and range of temporal durations , a search based on merger waves
can be performed using a nonlinear filtering search algorithm .
such a search should increase the number of discovered bbhs by a factor of roughly @xmath22 over those found from the inspiral and ringdown waves . on the other hand , a full set of merger templates based on numerical relativity simulations
could further increase the number of discovered bbhs by a additional factor of up to about @xmath23 .
psfig.tex measuring gravitational waves from binary black hole coalescences : + i. signal to noise for inspiral , merger , and ringdown .
newman laboratory , cornell university , ithaca , ny 14853 - 5001 .
theoretical astrophysics , california institute of technology , pasadena , california 91125 . |
martnesitic transformations are first - order , displacive , solid - solid phase transformation taking place upon cooling below a characteristic temperature t@xmath7 from a high symmetry initial phase to a low - symmetry structure .
ni@xmath8mnga exhibits martensitic transformation upon cooling through 220k @xcite . moreover , it is ferromagnetic with a curie temperature , t@xmath9 370k making it a technologically important magnetic shape memory alloy .
alot of theoretical@xcite and experimental studies @xcite that reflect the fermi surface character of ni@xmath8mnga upon undergoing martensitic transformation have been carried out .
the early neutron diffraction study @xcite determined the structure of the stoichiometric ni@xmath8mnga to be cubic l2@xmath4 heusler type with a = 5.825 and a complete crystallographic structure determination of different martensitic phases has also been reported@xcite .
many other investigations have been carried out on the near - stoichiometric alloys confirming that the martensitic structure is a body centered tetragonal distortion of the initial cubic lattice .
the low temperature crystal structure of the non - stoichiometric ni - mn - ga alloys revealed that there exists different intermartensitic transformations as the lattice is subjected to periodic shuffling of the ( 110 ) planes along the [ 1@xmath100]@xmath11 direction of the initial cubic system @xcite with modulation period dependent on the composition as summarized in @xcite .
inelastic scattering measurements made on the single crystals showed the presence of precursor effect above t@xmath7 that gives rise to softening of [ @xmath12 @xmath12 0]ta@xmath8 phonon mode at wave vector @xmath13@xmath14 0.33 of the reciprocal lattice @xcite .
a large softening of certain elastic constants @xmath15 = @xmath16(@xmath17 ) takes place at this intermediate transition@xcite .
the dependence of @xmath15 on applied magnetic field proves the magnetoelastic origins of these interactions @xcite .
extensive studies in the past have attributed the structural transformations to the phonon anomalies occurring in the parent phase @xcite .
here , an incomplete softening of the [ @xmath12 @xmath12 0]ta@xmath8 phonon mode at a particular wave vector @xmath13(corresponding to the periodicity of the martensitic phase ) with displacement along the [ 110 ] direction takes place .
such a phonon softening is believed to be due to contribution from electron - lattice coupling and nesting of the fermi surface @xcite . in spite of intense efforts ,
the underlying mechanism giving rise to such a phase transformation is still not well understood .
the nature of modulations forming the super structures and the driving force for the martensitic transformation in these alloys is currently at debate .
recent calculations by @xcite indicate the importance of modulated structure and the shuffling of atomic planes in stabilizing the martensitic structure in stoichiometric and non - stoichiometric alloys . the stability of the structure is associated with a dip in the minority - spin density of states ( dos ) at the fermi level , related to the formation of hybrid states of ni d and ga p minority - spin orbitals @xcite .
the shuffles in these alloys are believed to be due to two different effects - modulations and tetrahedral distortions @xcite . also it is predicted that the amplitude of modulations are different for mn - ga and ni planes @xcite .
therefore a precise knowledge of the changes occurring in the local structure of constituent atoms is fundamental in understanding the mechanism involved in martensitic transformation .
exafs is an ideal tool to study such transformations by making a comparative study of the local structure in austinitic and martensitic phases .
it is with this objective that the present investigation was undertaken . in the present work ,
we report our study on mn and ga k - edge xafs carried out at room temperature and liquid nitrogen temperature in the ni - mn - ga system to explore the changes in local environment around these metal ions in the austenitic and martensitic phase .
we have carried out the measurements on the alloy compositions : ni@xmath0mn@xmath1ga with @xmath3 = 0 , 0.1 , 0.13 , and 0.16 .
the t@xmath7 in this series is known to increase systematically from 220k to 315k with increasing ni content@xcite .
essentially , samples with @xmath3 = 0 , and 0.10 are in the austinitic phase at room temperature , whereas @xmath3 = 0.13 undergoes a transition at @xmath14295k and @xmath3 = 0.16 represents the martensitic phase with t@xmath7 = 315k . at liquid nitrogen temperature
all the four samples are in the martensitic phase .
polycrystalline ingots of ni@xmath0mn@xmath1ga 0 @xmath2 @xmath3 @xmath2 0.16 were prepared by conventional arc - melting method in argon atmosphere .
the ingots were sealed in an evacuated quartz ampoule and annealed at 1000k for 48 hours followed by quenching in cold water to improve the homogeneity of the samples .
a piece of the ingot was crushed into fine powder and further annealed at 1000k for 24 hours in an evacuated quartz tube inorder to remove any internal stress .
room temperature powder x - ray diffraction patterns recorded on rigaku d - max iic diffractometer with cu k@xmath18 radiation indicated the samples to be phase pure with l2@xmath4 structure for @xmath3 = 0 , 0.1 and a modulated tetragonal structure for @xmath3 = 0.13 and 0.16 samples . energy dispersive x - ray ( edx ) analysis confirmed the compositions to be nominal . the temperature dependent magnetic susceptibility and four probe resistivity measurements yielded @xmath19 and @xmath20 values same as those obtained by @xcite .
absorbers for the exafs experiments were made by spreading very fine powder on a scotch tape avoiding any sort of sample inhomogeneity and pin holes .
small strips of the sample coated tape were cut and were held one on top of other .
enough number of such strips were adjusted to give absorption edge jump , @xmath21 .
exafs at mn and ga k - edges were recorded in the transmission mode at the exafs-1 beamline at elettra synchrotron source using si(111 ) as monochromator .
the measurements were carried out at room temperature and liquid nitrogen temperature ( henceforth called rt and lt respectively ) . the incident and transmitted photon energies
were simultaneously recorded using gas - ionization chambers filled with mixtures of he - n@xmath8 for mn edge and ar - n@xmath8 for ga edge .
measurements were carried out from 300ev below the edge energy to 1000ev above it with a 5ev step in the pre - edge region and 2.5ev step in the exafs region . at each edge , three scans were collected for each sample .
data analysis was carried out using ifeffit suite wherein theoretical fitting standards were computed with atoms and feff6 programs @xcite and fitting was done using feffit program @xcite . in the present series
although mn content changes from 1 to 0.84 , this change amounts to less than one atom per unit cell .
furthermore , the atomic number of mn ( z = 25 ) and that of ni ( z = 29 ) being similar , the x - ray scattering amplitudes and phase functions will not be drastically different and exafs would be insensitive to such a substitution .
therefore theoretical standards were calculated for stoichiometric ni@xmath8mnga and were fitted to exafs data of all the samples irrespective of mn content .
the @xmath22-weighted @xmath23 spectra at mn and ga k - edges in all the samples at rt and lt are shown in fig.[rt - all - k ] and fig.[lt - all - k ] respectively .
these spectra reflect the good quality of data up to 15@xmath24 .
the fourier transform ( ft ) magnitude in @xmath25 space of the @xmath22 weighted mn k - edge exafs at rt and lt are shown in fig.[mag - r - all ] .
at room temperature the samples with @xmath3 = 0 and 0.1 are in the austinitic phase .
therefore , the exafs spectra of these samples recorded at mn and ga k - edges was fitted using common set of variable parameters with fm3 m space group and lattice constant 5.825 . in this model , the correction to the path lengths was refined with a constraint , @xmath26 where @xmath27 is the nearest neighbour distance , kept fixed to 2.5223 obtained from the lattice constant , @xmath28 is the calculated bond length obtained from feff and @xmath29 is the change in first neighbour distance .
this approach reduces the number of variable parameters in the fit .
the thermal mean - square variation in the bond lengths , @xmath30 were varied independently for each path considered in the fit .
the fitting was carried out in @xmath25-space in the range 1 to 5 using four single scattering ( ss ) paths and one linear multiple scattering ( ms ) path along the body diagonal of the initial cubic cell . the magnitude and real component of ft of @xmath31 for mn and ga edge data
are shown in fig.[xafs - aus ] . as can be seen from the figure ,
the fits are quite satisfactory .
the bond distances and the final fitted parameters obtained are presented in the table [ tab - aus ] .
it is seen that there is a discrepancy in the mn - ga/ ga - mn bond distance for @xmath3 = 0.1 sample .
the @xmath30 values for this bond are also quite different .
the reason for this anomaly could be the proximity of its martensitic transformation temperature ( t@xmath7 = 285k ) to the temperature of exafs measurement ( 295k ) .
it is well known in literature that the martensitic transformations are preceeded by a pre - transformation effects like softening of phonon modes and anomalies in elastic constants .
inelastic neutron scattering studies have evidenced such anomalies in ni - mn - ga alloys @xcite .
the t@xmath7 for @xmath3 = 0.1 sample being only 10k below the room temperature , pre - transformational effects would be much intense here , giving rise to anamolies in @xmath30 and causing discrepancies in bond distance of the near - neighbour atoms .
fig.[mag - r - all](c ) and ( d ) demonstrates the low temperature exafs in @xmath25 space at mn edge for @xmath3 = 0 and 0.10 samples present in their martensitic phase . in the range r = 2.5 - 5.0
a difference in spectral signatures of the two alloys in the lt and rt data is quite evident .
this can be attributed to the lowering of symmetry from the parent cubic structure upon undergoing the martensitic transition .
consequently , the exafs analysis was carried out using a tetragonal structure with c / a @xmath32 1 @xcite .
the @xmath30 values obtained from the austinitic phase served as starting parameters and @xmath33 parameters were varied independently .
the co - ordination number for each path was kept fixed to its crystallographic value . as per the model , for mn as the absorbing atom , the first peak in the range r = 1.5 to 3.2 is due to the contribution from 8 ni atoms at 2.52 , 2 ga atoms at 2.78 and 4 ga atoms at 2.96 .
however , the @xmath30s for mn - ga paths obtained from this fitting differs vastly from each other with values of 0.003@xmath34 and 0.03@xmath34 respectively .
generally , at such close bond lengths , a large variation in @xmath30 especially for the bonds involving same type of atoms is not expected .
therefore it indicates that there is a large spread in the bond distance of longer ga neighbour and/or a different distribution of ga atoms around mn in the second and third shells .
thus the model was supplemented by carrying out fits with different combinations of 6 ga neighbours distributed in the two shells . in each of these fits
, @xmath30 was varied keeping the coordination number fixed to a particular distribution .
best fit was obtained for 4 and 2 ga atoms in the second and third shell respectively and the @xmath30 obtained are presented in table [ tab - mart ] .
thereafter , such re - grouping of bond lengths had to be incorporated for all the subsequent ss paths . here
, the important aspect brought out by the analysis is the change in atomic coordinations leading to distribution of the bond lengths .
this observation reflects the different arrangements of atoms in different crystal planes in the martensitic phase .
in other words , the constituent atoms have been displaced over varied distances giving rise to modulations in the crystal planes in trying to maintain volume conservation - an essential criteria for martensitic transition .
low temperature neutron diffraction studies on ni@xmath8mnga by brown _ et.al _ @xcite have reported a modulated structure for ni@xmath8mnga from which a similar grouping of bond length distribution of ga atoms around mn can is obtained .
refinement of ga k - edge data also presents a similar situation .
the parameters extracted from both the edges are presented in table [ tab - mart ] and ft fittings in @xmath25 space are shown in the fig.[lt-200-r ] .
the bond distribution with 4 and 2 coordination of mn atoms are present around the central ga atom at an averarge distance of 2.791 and 3.065 respectively .
it is seen from the table that the third bond distance with ga as central atom is larger by about 0.1 as compared to that with mn as the central atom .
also the @xmath30 values obtained from ga exafs for ga - mn bonds are higher than those obtained from mn exafs ( refer table [ tab - mart ] ) .
the physical significance of these observations is that the ga atoms have a smaller amplitude of displacement from its crystallographic position in comparison to mn . in other words ,
ga atoms are sluggish and do not get much displaced in undergoing a martensitic transition leading to a stronger hybridization between ga - ni in the martensitic phase .
another important observation here is the discrepancy in the bond distance of the of the ms path described in table [ tab - mart ] .
being a linear path along the body diagonal of the initial cubic cell , the length of this path should be the sum of mn - ni and ga - ni bond lengths .
apparently , this condition is not satisfied in the lt data which implies that the ms path is no longer linear .
it is this signature that once again brings in the prominence of movement of atoms in the martensitic phase .
the distortions in atomic positions further results in `` dimpling '' of crystal planes as evidenced by non - linearity of the ms path giving rise to modulated structures . in short ,
it is seen from our rt and lt exafs analysis of @xmath3 = 0 , 0.10 samples that the martensitic transformation causes the atoms to displace from its initial positions by varied amplitudes with least displacement of ga atoms causing local distortions .
these local distortions gives rise to modulations and may eventually lead to long - range ordering of unit cells over many atomic planes .
the samples with @xmath3 = 0.13 , 0.16 are martensitic at room temperature . as can be seen from fig.[mag - r - all](e ) and ( f ) , the rt spectral signatures of these samples show subtle differences in comparison to the lt spectra for @xmath3 = 0 , 0.1 .
thus a tetragonal structural model with @xmath35 1 was employed for interpretation of the spectra @xcite . in these samples ,
the rt and lt spectra are quite similar and hence the ft fittings in @xmath25 space are presented for the low temperature data alone in fig.[lt-213-r ] .
the bond distances obtained are presented in table [ tab - mart ] .
it is seen that the bond distances obtained from the two edges show notable difference , especially in the first shell . if one considers the difference between mn - ni and ga - ni bond distances alone , there is a change of about @xmath14 0.016 .
both the central atoms , mn and ga can be viewed to be at the body centered position of a reduced tetragonal structure formed by 8 ni atoms .
a non - uniformity in their bond distance with ni of the order of 10@xmath36 is unexpected and hints toward the microscopic changes influencing the formation of the macroscopic modulated phases .
also , the second and third shell distances obtained from the two edges are significantly different .
furthermore , as can be seen from table[tab - mart ] , ga - mn(@xmath14 3 ) bond distance is shorter that the mn - ga(@xmath14 3.2 ) bond distance .
this is exactly opposite to the trend observed in lt spectra of @xmath3 = 0 , 0.1 where ga - mn = 3.06 and mn - ga = 2.96 .
however , the @xmath30 of ga - mn bond is much larger ( 0.03 @xmath37 0.01 @xmath34 ) than that of mn - ga bond .
this typical behaviour of higher @xmath30 for ga - mn bonds is prevalent in all the samples in martensitic phase .
this result is critical because it is a direct indication of movement of constituent atoms from their crystallographic positions with ga having the least amplitude of displacement .
thus , when viewed from ga k - edge , the local structure seems to be much distorted with higher amplitudes of displacements for other constituent atoms and therefore a higher value of @xmath30 for the respective bond .
the local structural study of ni@xmath0mn@xmath1ga alloys in the austinitic and martensitic phases by exafs at mn and ga k - edge enable us to identify the microscopic changes influencing the formation of the macroscopic modulated structure .
based on our analysis the following results were obtained : * higher values of thermal mean square vibration , @xmath30 , for ga k - edge analysis in comparison to mn k - edge in the martensitic phase for @xmath3 = 0 , 0.10 samples .
* different values of bond distances for the same pair of atoms ( ga - mn ) in the martensitic phase of all samples . * a difference of 0.016 between mn - ni and ga - ni bond distance in @xmath3 = 0.13 , 0.16 .
firstly , the parameters obtained for the room temperature exafs data for @xmath3 = 0 , 0.1 are in line with those expected for the austenitic structure .
the low temperature exafs spectra , for these samples have features different to those of the room temperature and carry information about the martensitic phase .
the displacement of atoms occurring upon the structural change is reflected through higher values of @xmath30 .
this argument is further supported by change in coordination numbers to 4 + 2 and re - grouping of the two mn - ga / ga - mn bonds .
thus the local symmetry breaking upon structural phase transition leads to the movement of constituent atoms .
such a modulated structure of ni@xmath8mnga has been determined experimentally@xcite and effect of shuffling of atoms on the physics of martensitic transformation has also been studied theoretically@xcite . on inspection of the parameters extracted from our analysis
, it is seen that the @xmath30 values are higher in case of ga exafs than mn exafs for the same mn - ga bond in the martensitic phase .
this clearly indicates a spread or distribution of mn - ga bonds wherein mn atoms moves more freely in comparison to ga .
this observation is further substantiated by a higher average bond length obtained with respect to ga as central atom as compared to that obtained from mn .
therefore there is a larger spread in ga - mn distance than mn - ga . in other words ,
ga atoms are sluggish and have a smaller amplitude of displacement than other constituent atoms forming the alloy .
thermal and stress induced martensitic transitions in ni - mn - ga single crystals has been studied in the past @xcite wherein , it was shown that the sample in martensitic phase undergoes another stress induced transition from a structure with @xmath38 to that with @xmath39 .
a similar transition is seen here in ni@xmath0mn@xmath1ga when @xmath3 changes from 0.1 to 0.13 . a simplistic view , as obtained from our exafs analysis , of local structure from ga as central atom in the martensitic phase is as shown in fig . [ loc - str ] .
in the case of @xmath40 wherein @xmath41 is @xmath32 1 , the effect of modulations are shown to be about the @xmath42 axis .
the shorter mn - ga bond lengths are therefore depicted along the @xmath42 and @xmath43 axes while the longer one is shown along the @xmath44 axis . for @xmath45 , as described above , exafs can only be fitted to a structure with @xmath39 .
therefore in this case the longer mn - ga bond length is shown along the @xmath43 axis and the shorter ones along @xmath44 and @xmath42 axes .
a comparison of magnitudes of bond lengths clearly shows that in @xmath45 a rotation about @xmath42 axis transforms the structure similar to that of @xmath40 .
therefore the two martensitic structures are similar under rotation of cartesian axes . in order to further elucidate the local structure in ni - mn - ga system ,
we look at the nearest neighbour interaction of the absorbing atoms .
as mentioned in the previous section , there is a notable difference in the mn - ni and ga - ni bond lengths determined from the low temperature data for @xmath3 = 0.13 , 0.16 . comparing the estimates obtained
, it is seen that mn - ni bond distance is higher than the ga - ni distance .
the difference is greater than experimental resolution and demonstrates the locally distorted environment around mn and ga atoms leading to an electronic structure that is different in the austinitic and martensitic phases .
ni atoms reside inside the interlocking tetrahedral cages formed by mn and ga atoms .
a shorter ga - ni bond means that ga is more strongly bound to ni than mn .
thus the ga tetrahedra is distorted in a way that allows more space for the movement of mn atoms .
therefore mn atoms have higher amplitude of displacement from its crystallographic position than ga .
these tetrahedral distortions lead to re - distribution of electrons and is perhaps the root cause of band jahn - teller transitions observed in such alloys@xcite .
such a ni - ga hybridization has also been anticipated theoretically@xcite .
these calculations yeild energetically favorable hybrid states formed by ga and ni giving rise to a peak in the spin - down electronic density of states at the fermi level .
in this work , we have carried out a comparative analysis of changing local structures in ni@xmath0mn@xmath1ga alloys upon undergoing martensitic transition .
exafs measurements at mn k - edge and ga k - edge at room temperature and liquid nitrogen temperature were carried out .
the most significant feature of our analysis is the difference in the mn - ni and ga - ni bond length .
a shorter ga - ni bond implies an increased ga - ni hybridization in comparision to mn - ni in the martensitic phase . the present study is a direct experimental evidence for such hybridization which is seen more clearly in @xmath3 = 0.13 and 0.16 samples .
the differences in the mn - ni and ga - ni bond lengths leads to the distortion of the two tetrahedra formed by mn - ni and ga - ni .
the increased hybridization would probably lead to re - distribution of electrons causing a band jahn - teller effect .
it is also seen that the constituent atoms of ni - mn - ga system displace from their crystallographic positions by varying amplitudes in the martensitic phase .
the ga atoms seems to displace over very small amplitudes .
the uneven movement of the constituent atoms gives rise to dimpling of atomic planes and may eventually lead to modulated structures .
authors gratefully acknowledge financial assistance from department of science a nd technology , new delhi , india and ictp - elettra , trieste , italy for the proposal 2004646 and 2005743 .
thanks are also due to prof .
g. vlaic and dr .
luca olivi for help in exafs measurements and useful discussions .
p.a.b . would like to thank council for scientific and industrial research , new delhi for senior research fellowship .
50 p. j. webster , k. r. a. zeibeck , s. l. town and m. s. peak , philos .
mag * 49 * , 295 ( 1984 ) .
a. ayuela , j. enkovaara , k. ullakko and r. m. nieminen , j. phys . : condens
. matter . * 11 * , 2017 ( 1999 ) .
v. v. godlevsky and k. m. rabe , phys .
b * 63 * , 134407 ( 2001 ) .
a. ayuela , j. enkovaara and r. m. nieminen , j. phys . : condens
* 14 * , 5325 ( 2002 ) .
j. m. maclaren , j. appl .
* 91 * , 7801 ( 2002 ) .
y. lee , j. y. rhee and b. n. harmon , phys .
b * 66 * , 054424 ( 2002 ) . c. bungaro , k. m. rabe and a. dal corso , phys .
b * 68 * , 134104 ( 2003 ) .
a. t. zayak , p. entel , j. enkovaara , a. ayuela and r. m. nieminen j. phys.:condens . matter . *
15 * , 159 ( 2003 ) .
a. t. zayak and p. entel , mater .
a. * 378 * , 419 ( 2004 ) .
a. t. zayak , p. entel , k. m. rabe , w. a. adeagbo and m. acet , phys .
b * 72 * , 054113 ( 2005 ) .
s. r. barman , s. banik and aparna chakrabarti , phys .
b * 72 * , 184410 ( 2005 ) . v. v. martynov and v. v. kokorin , j. phys .
iii * 2 * , 739 ( 1992 ) .
k. ooiwa , k. endo and a. shinogi , j. magn .
magn . mater . *
104 * , 2011 ( 1992 ) .
s. wirth , a. leithe - jasper , a. n. vasilev , j. m. d. coey , j. magn .
magn . mater . *
167 * , l7 ( 1997 ) .
a. sozinov , a. a. likhachev , n. lanska and k. ullakko , appl .
. lett . * 80 * , 1746 ( 2002 ) .
a. n. vasilev , v. d. buchelnikov , t. takagi , v. v.kholvailo and e. i. estrin , _ physics - uspekhi _ , * 46 * , 559 ( 2003 ) and references therein . k. r. priolkar , p. a. bhobe , s. d. sapeco and r. paudel , phys . rev .
b * 70 * , 132408 ( 2004 ) . y. k. kuo , k. m. sivakumar , h. c. chen , j. h. su and c. s. lue , phys . rev .
b * 72 * , 054116 ( 2005 ) .
p. j. brown , j. crangle , t. kanomata , m. matsumoto , k - u .
neumann , b. ouladdiaf and k. r. a. ziebeck , j. phys.:condens . matter .
* 14 * , 10159 ( 2002 ) .
j. pons , v. a. chernenko , r. santamarta and e. cesari , acta . mater . * 48 * , 3027 ( 2000 ) and references therein .
a. zheludev , s. m. shapiro and p. wochner , phys .
b * 51 * , 11310 ( 1995 ) .
u. stuhr , p. vorderwisch , v. v. kokorin , p .- a .
lindgard , phys . rev .
b * 56 * , 14360 ( 1997 ) .
a. gonzlez - comas , e. obrado , l. manosa , a. planes , v. a. chernenko , b. j. hattink and a. labarta , phys .
b. * 60 * , 7085 ( 1999 ) .
l. manosa , a. gonzlez - comas , e. obrado , a. planes , v. a. chernenko , v. v. kokorin and e. cesari , phys .
b * 55 * , 11068 , ( 1997 ) .
a. planes , e. obrado , a. gonzlez - comas and l. manosa , phys .
* 79 * , 3926 , ( 1997 ) .
s. m. shapiro , y. noda , y. fujii and y. yamada , phys .
b * 30 * , 4314 ( 1984 ) .
s. m. shapiro , b. x. yang , y. noda , l. e. tanner and d. schryvers , phys .
b * 44 * , 9301 ( 1991 ) .
g. l. zhao and b. n. harmon , phys .
b * 45 * , 2818 ( 1992 ) .
g. l. zhao and b. n. harmon , phys .
b * 48 * , 2031 ( 1993 ) .
s. b. dugdale , r. j. watts , j. laverock , zs .
major , m. a. alam , m. samsel - czekala , g. kontrym - sznajd , y. sakurai , m. itou and d. fort , phys .
lett * 96 * , 046406 ( 2006 ) .
a. n. vasilev , a. d. bozhko , v. v. kholvailo , i. e. dikshetein , v. g. shavrov , v. d. buchelnikov , m. matsumoto , s. suzuki , t. takagi and j. tani , phys .
b * 59 * , 1113 ( 1999 ) .
b. ravel , j.synchrotron radiat . * 8 * , 314 ( 2001 ) .
s. i. zabinsky , j. j. rehr , a. ankudinov , r. c. albers and m. j. eller , phys .
b * 52 * , 2995 ( 1995 ) .
m. newville , b. ravel , d. haskel .
j. j. rehr , e. a. stern and y. yacoby , physica b * 208 * & * 209 * , 154 ( 1995 ) .
b. wedel , m. suzuki , y. murakami , c. wedel , t. suzuki , d. shindo and k. itagaki , j. alloys compd . *
290 * , 137 ( 1999 ) .
a. t. zayak , p. entel , j. enkovaara , a. ayuela and r. m. nieminen , phys .
b * 68 * , 132402 ( 2003 ) .
s. fujii , s. ishida and s. asano , j. phys .
japan * 58 * , 3657 , ( 1987 ) .
j. brown , a. y. bargawi , j. crangle , k - u .
neumann and k. r. a. ziebeck , j. phys.:condens . matter . * 11 * , 4715 ( 1999 ) .
rcccc atom and & & + coord .
& r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) + ni1 @xmath46 8 & 2.519(8 ) & 0.0081(3 ) & 2.518(2 ) & 0.0080(2 ) + ga1 @xmath46 6 & 2.909(3 ) & 0.03(1 ) & 2.795(8 ) & 0.014(1 ) + mn1 @xmath46 12 & 4.114(4 ) & 0.029(9 ) & 4.20(2 ) & 0.021(3 ) + ni2 @xmath46 24 & 4.824(5 ) & 0.019(3 ) & 4.85(1 ) & 0.019(2 ) + ga2 @xmath46 16 & 5.038(5 ) & 0.007(1 ) & 4.900(8 ) & 0.008(1 ) + msga3@xmath47ni1@xmath47mn ] @xmath46 8 & 5.038(5 ) & 0.0097(6 ) & 5.02(1 ) & 0.018(2 ) + & & & & + + atom and & & + coord .
& r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) + ni1 @xmath46 8 & 2.512(2 ) & 0.0077(2 ) & 2.516(3 ) & 0.0082(4 ) + mn1 @xmath46 6 & 2.901(2 ) & 0.030(6 ) & 2.88(5 ) & 0.026(7 ) + ga1 @xmath46 12 & 4.103(3 ) & 0.022(4 ) & 4.23(4 ) & 0.017(5 ) + ni2 @xmath46 24 & 4.811(4 ) & 0.015(1 ) & 4.79(1 ) & 0.016(2 ) + mn2 @xmath46 16 & 5.025(3 ) & 0.017(7 ) & 4.90(1 ) & 0.007(1 ) + msmn3@xmath47ni1@xmath47ga ] @xmath46 8 & 5.025(3 ) & 0.014(1 ) & 5.15(3 ) & 0.023(4 ) + lllllllll atom and & & & & + coord .
& r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) + ni1 @xmath46 8 & 2.518(3 ) & 0.0057(4 ) & 2.518(5 ) & 0.0056(5 ) & 2.528(2 ) & 0.0060(2 ) & 2.523(2 ) & 0.0051(1 ) + ga1 @xmath46 4 & 2.780(6 ) & 0.0043(5 ) & 2.768(8 ) & 0.0037(7 ) & 2.740(3 ) & 0.0045(3 ) & 2.739(5 ) & 0.0054(4 ) + ga2 @xmath46 2 & 2.96(3 ) & 0.008(3 ) & 2.95(2 ) & 0.009(2 ) & 3.12(2 ) & 0.010(3 ) & 3.23(4 ) & 0.012(5 ) + mn1 @xmath46 4 & 3.96(3 ) & 0.009(3 ) & 3.93(3 ) & 0.009(3 ) & 3.907(3 ) & 0.012(4 ) & 3.89(2 ) & 0.008(2 ) + mn2 @xmath46 8 & 4.19(1 ) & 0.009(2 ) & 4.19(1 ) & 0.009(2 ) & 4.22(1 ) & 0.009(1 ) & 4.23(2 ) & 0.011(2 ) + ni2 @xmath46 16 & 4.69(1 ) & 0.011(1 ) & 4.66(1 ) & 0.009(1 ) & 4.61(1 ) & 0.013(2 ) & 4.613(9 ) & 0.009(1 ) + ni3 @xmath46 8 & 4.90(1 ) & 0.006(1 ) & 4.905(8 ) & 0.0040(7 ) & 5.364(8 ) & 0.0044(8 ) & 5.327(8 ) & 0.0044(8 ) + msga3@xmath47ni1@xmath47mn ] @xmath46 16 & 5.068(6 ) & 0.0095(7 ) & 5.056(8 ) & 0.0097(8 ) & 5.088(4 ) & 0.0078(4 ) & 5.075(3 ) & 0.0068(3 ) + & & & & & & & & + + atom and & & & & + coord .
no . & r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) & r ( ) & @xmath30 ( @xmath34 ) + ni1 @xmath46 8 & 2.5111(8 ) & 0.00431(8 ) & 2.512(2 ) & 0.0044(2 ) & 2.512(1 ) & 0.0039(1 ) & 2.511(1 ) & 0.0042(1 ) + mn1 @xmath46 4 & 2.791(4 ) & 0.0078(5 ) & 2.776(8 ) & 0.0062(5 ) & 2.725(4 ) & 0.0063(5 ) & 2.722(5 ) & 0.0067(6 ) + mn2 @xmath46 2 & 3.065(2 ) & 0.012(2 ) & 3.06(2 ) & 0.013(4 ) & 3.0(2 ) & 0.04(3 ) & 3.0(1 ) & 0.03(2 ) + ga1 @xmath46 4 & 3.97(1 ) & 0.009(2 ) & 3.93(2 ) & 0.008(2 ) & 3.87(2 ) & 0.008(2 ) & 3.85(2 ) & 0.009(2 ) + ga2 @xmath46 8 & 4.215(8 ) & 0.0076(8 ) & 4.214(9 ) & 0.0079(9 ) & 4.25(1 ) & 0.009(1 ) & 4.248(7 ) & 0.009(1 ) + ni2 @xmath46 16 & 4.706(8 ) & 0.0048(6 ) & 4.676(6 ) & 0.0081(7 ) & 4.614(7 ) & 0.0078(6 ) & 4.619(7 ) & 0.0083(7 ) + ni3 @xmath46 8 & 4.889(3 ) & 0.0047(3 ) &
4.872(5 ) & 0.0033(4 ) & 5.313(8 ) & 0.0026(5 ) & 5.319(8 ) & 0.0025(5 ) + msmn3@xmath47ni1@xmath47ga ] @xmath46 16 & 5.069(7 ) & 0.0111(9 ) & 5.111(5 ) & 0.0089(6 ) & 5.102(4 ) & 0.0051(3 ) & 5.106(2 ) & 0.0047(2 ) + | the local atomic structure of ni@xmath0mn@xmath1ga with 0 @xmath2 @xmath3 @xmath2 0.16 alloys was explored using mn and ga k - edge extended x - ray absorption fine structure ( exafs ) measurement .
inorder to study the atomic re - arrangements that occur upon martensitic transformation , room temperature and low temperature exafs were recorded .
the changes occurring in the l2@xmath4 unit cell and the bond lengths obtained from the analysis enables us to determine the modulation amplitudes over which the constituent atoms move giving rise to shuffling of the atomic planes in the modulated structure .
the exafs analysis also suggests the changes in hybridization of ga-@xmath5 and ni-@xmath6 orbitals associated with the local symmetry breaking upon undergoing martensitic transition . |
Subsets and Splits