article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
applications of perturbation theory in quantum mechanics require summations over a complete set of states of the lowest - order hamiltonian .
usually , the relevant spectrum is innumerable . in practical applications such eigenspectra
are often modeled using finite basis sets , chosen to be numerically complete . since the sets are finite , the otherwise infinite summations become amendable to numerical evaluations .
the use of a finite basis set composed of piecewise polynomials , so - called b - splines @xcite , has proven to be particularly advantageous in atomic physics and quantum chemistry applications @xcite . in this approach ,
an atom is placed in a sufficiently large cavity and the atomic wavefunctions are expanded in terms of the underlying b - spline set .
further , the variational galerkin method is invoked and the solution of the resulting matrix eigenvalue problem produces a quasi - spectrum for the atom . in non - relativistic calculations ,
the lowest - energy orbitals of the resulting basis set closely agree with those of the unperturbed atom , and calculations of various properties of the low - lying states can be carried out .
in particular , one could generate single - particle orbitals in some suitable lowest - order approximation , and use the resulting basis set in applications of many - body perturbation theory ( mbpt ) .
application of the outlined approach to the _ relativistic _ problems brings in a complication the appearance in the atomic quasi - spectrum of non - physical `` spurious '' states .
these states appear in the solution of the single - particle radial dirac equation for @xmath1 angular symmetry , @xmath2 ( @xmath3 orbitals ) .
the spurious states rapidly oscillate and , moreover , spoil the mapping of the generated quasi - spectrum onto the low - energy orbitals of the atom . at the same time
they are required for keeping the set complete .
the problem of spurious states was discussed in the literature in details , see e.g. , ref .
@xcite , and several solutions were proposed . in the pioneering applications of the b - splines in relativistic many - body problem by the notre dame group , @xcite added an artificial potential spike centered at the origin to the hamiltonian matrix .
the overall effect was to shift the spurious states to higher energies thus restoring the low - energy mapping to the physical states .
we will refer to the sets generated using this prescription as the notre dame ( nd ) sets .
another solution was to use `` kinetically - balanced '' sets @xcite , which related the small and the large components of the basis set functions via the pauli approximation .
recently , an extension of this method was proposed in ref .
@xcite . here , due to additional relations between the small and large components , the negative ( dirac sea ) and positive
energy spectra are treated in a symmetric fashion . to emphasize this built - in symmetry
, the authors refer to their method as the `` dual - kinetic - balance '' ( dkb ) approach . in both methods ( by contrast to the nd prescription )
, the spurious states were shown to be completely eliminated from the quasi - spectrum .
motivated by the success of the dkb method in computing properties of hydrogen - like ions@xcite , here we investigate the suitability of the dkb method in modeling the spectrum of the ( non - local ) dirac - hartree - fock ( dhf ) potential .
we compare the performance of the nd and dkb methods by computing various properties of cs atom : single - particle energies , hyperfine integrals , and the parity - non - conserving amplitude of the @xmath0 transition .
we find that for properties involving matrix elements accumulated near the nucleus , the dkb method outperforms the nd approach . otherwise ,
if the electronic integrals are accumulated far from the nucleus , both methods produce results of a similar quality .
in addition , we investigate a possibility of using varying number of basis set functions for different angular symmetries .
summations over intermediate states in expressions of perturbation theory are carried out both over angular quantum numbers @xmath4 and for fixed @xmath4 over radial quantum numbers .
usually , as @xmath5 ( and @xmath6 ) increases , the correlation corrections due to higher partial waves become progressively smaller .
intuitively , one expects that for higher partial waves it would be sufficient to use smaller radial basis sets of lesser quality .
this would reduce storage requirements for many - body calculations ( for example , in implementing coupled - cluster formalism ) and would speed up numerical evaluations .
while such an approach is common in quantum chemistry , see , e.g. , ref.@xcite , the question of building the optimal b - spline basis set was not addressed yet in relativistic many - body calculations .
we illustrate optimizing the basis sets by computing the second - order energy correction for several states of cs .
this paper is organized as follows .
first we recapitulate the galerkin - type approach to generating a finite - basis set quasi - spectrum for the dirac equation in section [ sec : setup ] .
the variational method is invoked for relativistic action and the problem is reduced to solving the generalized eigenvalue problem in section [ sec : matrix ] .
the dhf potential is specified in section [ sec : pot ] .
further we specify nd and dkb sets in section [ sec : ndanddkb ] and boundary conditions in section [ sec : spubnd ] .
a numerical analysis of cs atom is provided in section [ sec : numerics ] . in sections [ sec :
enhfi ] and [ sec : pnc ] we compare the performance of the nd and the dkb sets by computing single - particle energies , hyperfine integrals ( section [ sec : enhfi ] ) , and parity - nonconserving amplitudes ( section [ sec : pnc ] ) using both methods .
the spurious states arising from the nd method are examined in section [ sec : spur ] . in section [ sec : e2 ] we consider second - order energy corrections in the dkb method and discuss a strategy of optimizing the size of the basis set .
we are interested in solving the eigenvalue equation @xmath7 for the dirac hamiltonian , @xmath8 , where @xmath9 is the nuclear potential and @xmath10 is the mean - field ( dirac - hartree - fock ) potential .
@xmath10 is in general a non - local potential .
assuming that both potentials are central one may exploit the rotational invariance to parameterize the solutions as @xmath11{c}% ip_{n\kappa}(r)\ \omega_{\kappa m}(\mathbf{\hat{r}})\\ q_{n\kappa}(r)\ \omega_{-\kappa m}(\mathbf{\hat{r } } ) \end{array } \right ) , \label{eq : bispinorspher}%\ ] ] with @xmath12 being the spherical spinors .
the solutions depend on the radial quantum number @xmath13 and the angular quantum number @xmath14 .
the large , @xmath15 , and small , @xmath16 , radial components satisfy the conventional set of radial dirac equations @xmath17 these radial equations may be derived by seeking an extremum of the following action @xcite @xmath18{c}% p\left ( r\right ) \\ q\left ( r\right ) \end{array } \right ) dr \nonumber \\ & + & \frac{1}{2}\int v_{\mathrm{nuc}}\left ( r\right ) \left ( p\left ( r\right ) ^{2}+q\left ( r\right ) ^{2}\right ) dr - c^{2}\int q\left ( r\right ) ^{2}dr \label{eq : action } -\varepsilon\frac{1}{2}\int \left ( p\left ( r\right ) ^{2}+q\left ( r\right ) ^{2}\right ) dr+\delta s^{\mathrm{bnd}}+\delta s^{\mathrm{spur}},% \ , , \nonumber\end{aligned}\ ] ] where the @xmath4-dependent pauli operators are defined as@xmath19 radial integrals here and below have implicit limits from @xmath20 to @xmath21 , where @xmath22 is the radius of the confining cavity .
boundary conditions may be imposed by adding the term @xmath23 to the action .
the term @xmath24 controls the appearance of the spurious states in the quasi - spectrum .
we will specify these two terms below .
we employ two finite basis sets @xmath25 and @xmath26 @xmath27 , which , since we are interested in solving the ( angularly - decoupled ) radial equations , may depend on the angular quantum number @xmath4 .
we may expand the large and small components in terms of these bases @xmath28 the expansion coefficients being the same for both the large and the small components .
the above expansions are plugged into the action , eq .
( [ eq : action ] ) , and , further , its extremum is sought by varying the expansion coefficients .
as a result one arrives at the following generalized eigenvalue equation@xmath29 where @xmath30 and @xmath31 are @xmath32 matrixes and @xmath33 is the vector of expansion coefficients in eq.([eq : pexpansion ] ) .
the matrix elements @xmath30 are given by@xmath34 the matrixes entering the definition of @xmath30 correspond to various pieces of the radial dirac equations , @xmath35 dr% \nonumber\\&%\quad + \left ( v_{\mathrm{dhf}}\right ) _ { ij},\\ s_{ij } & = \int s_{i}\left ( r\right ) s_{j}\left ( r\right ) dr,\end{aligned}\ ] ] with the matrix elements of the dhf potential , @xmath36 , given below .
the terms @xmath37 and @xmath38 arise from the boundary and spurious state corrections in the action . finally , the matrix @xmath31 , @xmath39 dr,$ ] reflects the fact that the basis sets may be non - orthonormal .
the nuclear potential @xmath40 is generated for a nucleus of a finite size .
we employ the fermi distribution with the nuclear parameters taken from ref .
@xcite . as to the dirac - hartree - fock potential
, we employ the frozen - core approximation . in this method ,
the calculation is carried out in two stages .
first , the core orbitals are computed self - consistently .
second , based on the precomputed core orbitals , the dhf potential is assembled for the valence orbitals and the valence orbitals are determined . in the valence part of the problem ,
the core orbitals are no longer adjusted .
explicitly , for a set of the angular symmetry @xmath4,@xmath41{cc}% l_{i } & s_{i}% \end{array } \right ) \hat{v}_{\mathrm{dhf}}\left ( \begin{tabular } [ c]{l}% $ l_{j}$ \\ $ s_{j}$% \end{tabular } \right ) \nonumber\\ & = \left ( v_{\mathrm{dhf}}^{\mathrm{dir}}\right ) _ { ij}+\left ( v_{\mathrm{dhf}}^{\mathrm{exc}}\right ) _ { ij},\label{eq : vdhf}\\ \left ( v_{\mathrm{dhf}}^{\mathrm{dir}}\right ) _ { ij } & = \sum_{a\in \mathrm{core}}\left ( 2j_{a}+1\right )
\nonumber\\ & \times\int v_{0}\left ( a , a , r\right ) \left [ l_{i}\left ( r\right ) l_{j}\left ( r\right ) + s_{i}\left ( r\right ) s_{j}\left ( r\right ) \right ] dr,\\ \left ( v_{\mathrm{dhf}}^{\mathrm{exc}}\right ) _ { ij } & = -\sum_{a\in \mathrm{core}}\sum_{l}\left ( 2j_{a}+1\right ) \lambda_{\kappa l\kappa_{a}}\nonumber\\ & \times\int v_{l}\left ( a , j , r\right ) \left [ l_{i}\left ( r\right ) p_{a}\left ( r\right ) + s_{i}\left ( r\right ) q_{a}\left ( r\right ) \right]dr , \end{aligned}\ ] ] with the conventionally defined multipolar contributions @xmath42 dr^{\prime},\end{aligned}\ ] ] and the angular coefficient@xmath43{ccc}% j_{a } & j_{b } & l\\ -1/2 & 1/2 & 0 \end{array } \right ) ^{2}.\ ] ] now we specify the notre dame ( nd ) and the dual - kinetic - balance ( dkb ) basis sets . both operate in terms of b - spline functions and first we recapitulate the relevant properties of these splines . a set of @xmath13 b - splines of order @xmath44 is defined on a supporting grid @xmath45 . usually , the gridpoints are chosen as @xmath46 in our calculations the intermediate gridpoints are distributed exponentially .
b - spline number @xmath47 of order @xmath48 , @xmath49 , is a piecewise polynomial of degree @xmath50 inside @xmath51 .
it vanishes outside this interval .
this property simplifies the evaluation of matrix elements between the functions of the basis set .
in addition , we will make use of the fact that as @xmath52 , the first @xmath44 splines behave as ( all the remaining splines are zero)@xmath53 and at @xmath21 , all splines vanish except for the last spline , @xmath54 .
the notre dame set is defined as@xmath55{cc}% b_{i}^{\left ( k\right ) } \left ( r\right ) & 1\leq i\leq n\\ 0 & n < i\leq2n \end{array } \right . , \label{eq : ndset}\\ s_{i}^{\mathrm{nd}}\left ( r\right ) & = \left\ { \begin{array } [ c]{cc}% 0 & 1\leq i\leq n\\ b_{i - n}^{\left ( k\right ) } \left ( r\right ) & n < i\leq2n \end{array } \right . .\nonumber\end{aligned}\ ] ] it corresponds to an independent expansion of the large and small radial components into the b - spline basis .
the dkb set involves the pauli operators and enforces a kinetic balance between contributions to the components:@xmath56{cc}% -b_{i}^{\left ( k\right ) } \left ( r\right ) & 1\leq i\leq n\\ -\frac{1}{2c}\hat{o}_{-}^{\kappa}~b_{i - n}^{\left ( k\right ) } \left ( r\right ) & n < i\leq2n \end{array } \right . , \label{eq : dkbset}\\ s_{i}^{\mathrm{dkb}}\left ( r\right ) & = \left\ { \begin{array } [ c]{cc}% \frac{1}{2c}\hat{o}_{+}^{\kappa}~b_{i}^{\left ( k\right ) } \left ( r\right ) & 1\leq i\leq n\\ b_{i - n}^{\left ( k\right ) } \left ( r\right ) & n < i\leq2n \end{array } \right . .\nonumber\end{aligned}\ ] ] notice that , as discussed below , to satisfy the boundary conditions , we will use only a subset of the entire dkb basis . with the nd set ,
the spurious states are shifted away to the high - energy end of the quasi - spectrum by adding the following action @xcite @xmath57 for @xmath58 and @xmath59 for @xmath1 .
this correction may be seen as arising from an artificial @xmath60function - like potential centered at the origin .
unfortunately , as shown below in numerical examples , this additional spike perturbs the behavior of the orbitals near the nucleus .
( as discussed below @xmath61 also sets the boundary conditions at @xmath62 ) the dkb set does not have the spurious states at all , so that @xmath63 .
we need to specify boundary conditions at @xmath20 and at the cavity radius , @xmath21 .
we start by discussing the boundary conditions at @xmath20 . for a finite - size nucleus
the radial components behave as @xmath64 in the notre dame approach , the boundary conditions are imposed variationally by adding the boundary terms to the action integral .
varying @xmath24 , ref.@xcite , effectively reduces to @xmath65 . in practice , because of the variational nature of the nd constraint , the large component , although being small , does not vanish at the origin , and the limits , ( [ eq : pqsmallr ] ) , are not satisfied .
alternatively , @xcite proposed to impose @xmath66 by discarding the first b - spline of the set ( this is the only b - spline that does not vanish at @xmath20 ) .
this is a hard constraint , since the wavefunction , eq.([eq : pexpansion ] ) , would vanish identically at the origin . in our calculations , because of our motivation in building the smallest possible basis set , we extend this scheme further .
we exploit the power - law behavior of the b - splines , eq.([eq : bsplineatr0 ] ) , and match it to the small-@xmath67 limits ( [ eq : pqsmallr ] ) . to satisfy the matching , we need to include the b - splines starting with the sequential number ( @xmath68 must be smaller than the order of the splines @xmath44 ) .
@xmath69{cc}% \ell+2 , & \kappa<0\\ \ell+1 , & \kappa>0 \end{array } \right . .
\label{eq : imin}\ ] ] for @xmath70 and @xmath71 and this is equivalent to the boundary condition of ref . @xcite . for higher partial waves ,
however , an increasingly larger number of splines is discarded : e.g. , for @xmath72 .
one should notice that for a basis that includes partial waves @xmath73 , for a faithful representation of the small-@xmath67 behavior in all the partial waves , one needs to require the order of the splines to be at least of @xmath74 .
in particular , for @xmath75 , @xmath76 . when the first b - spline of the set , @xmath77 , is included in the basis ( as in the nd approach ) , there is another difficulty in the calculations : since it s value does not vanish at @xmath20 , the matrix elements @xmath78 and @xmath79 , which contain matrix elements of @xmath80 , are infinite in absolute value @xcite . in practical calculations ,
one uses gaussian quadratures to evaluate this integral , so the result of the integration is finite .
yet this introduces arbitrariness in the nd calculations and may be a reason for a relatively poor representation of the orbitals near the origin . at the cavity radius , to avoid overspecifying the boundary conditions for the dirac equation , the nd group used the boundary condition @xmath81 .
as with the conditions at the origin , this relation was encouragedvariationally .
in our calculations ( similarly to ref .
@xcite ) we use the hard condition @xmath82 by removing the last b - spline from the set .
examination of the resulting orbitals reveals that the wavefunctions acquire a non - physical inflection towards the end of the supporting grid , while the nd orbitals behave properly .
further numerical experimentation , however , shows that the inflection does not degrade the numerical quality at least for the atomic properties of the low - lying bound states of interest .
at the same time , throwing away the last b - spline reduces the number of basis functions and leads to a more compact set . to summarize
, we will use the dkb basis set that includes b - splines with sequential numbers @xmath83 , eq .
( [ eq : imin ] ) to @xmath84
. we will simply refer to this choice as the dkb basis .
when we refer to @xmath85 dkb functions for a given partial wave , it would imply larger underlying b - spline set , e.g. , for @xmath86-waves the total number of b - splines would be @xmath87 . for the nd basis @xmath88 , as all b - splines participate in the expansion . in the remainder of this paper we present results of numerical analysis for @xmath89cs atom .
it is an atom with a single valence electron outside a closed - shell core . as the first step we carry out finite - difference dirac - hartree - fock calculations for cs core .
the core orbitals are fed into the spline code where they are used to compute the matrix elements of the @xmath90 potential , eq.([eq : vdhf ] ) . as in ref .
@xcite the numerical accuracy is monitored by comparing the resulting quasi - spectrum with the dhf energies from the finite - difference code .
here we provide numerical examples involving both nd and dkb sets for cs atom . in the two sections immediately following
we compare the performance of the nd and the dkb sets .
we generate the quasi - spectrum using both nd and dkb sets and carry out comparisons for single - particle energies and hyperfine integrals in section [ sec : enhfi ] and parity - nonconserving amplitude in section [ sec : pnc ] .
section [ sec : spur ] contains an analysis of spurious states in the nd method . in section [ sec : e2 ] we analyze second - order energy corrections and discuss a strategy of optimizing the size of the basis .
we compare nd and dkb quasi - spectrums with energies obtained using a finite - difference dhf code for the low - lying valence states in table [ tab : enhfi ] .
the nd and dkb calculations were carried out using @xmath85 basis functions for b - splines of order @xmath75 .
we used a cavity of radius @xmath91 bohr . for the cavity of this size ,
only a few lowest - energy orbitals remain relatively unperturbed by the cavity . from examining table [ tab : enhfi ] , it is clear that both nd and dkb sets have a similar accuracy for energies .
in the second part of table [ tab : enhfi ] we compare values of the radial integrals entering matrix elements of the hyperfine interaction due to the electric ( ej ) and magnetic ( mj ) multipolar moments of a point - like nucleus @xmath92 the angular selection rules require @xmath93 .
we use identical integration grid for all three cases ( dhf , nd , dkb ) listed in the table .
the grid is sufficiently dense near the origin , it contains about 100 points inside the nucleus .
the numerical integration excludes the first interval of the grid . from examining the table
we find that the dkb set outperforms the nd basis .
while the nd set still recovers two - three significant figures for the magnetic - dipole coupling , it produces wrong results for electric - quadrupole and magnetic - octupole integrals .
certainly , the accuracy in the nd case improves for a larger basis set , but larger basis sets come at an additional computational cost .
we carried out similar comparisons for matrix elements of the electric - dipole operator .
as for the energies , we find that both the nd and dkb sets perform with a similar numerical accuracy .
as we have mentioned , @xmath94 variationally encourages the boundary condition @xmath95 . however , there is no such explicit encouragement for @xmath96 .
here we qualitatively discuss the observed properties of the radial components near the origin for the cs nd set .
we see that , though their general behavior is to approach zero , the large component wavefunctions often have small improper inflections or oscillations near the origin .
small component wavefunctions , on the other hand , often do not even approach zero .
these improper behaviors seem to be exemplified as we look at states higher in the spectrum . as we have seen here , such improper behavior of both the large and small component radial functions near the origin prove detrimental for properties accumulated near the nucleus .
we conclude that while producing the results of a similar quality for matrix elements accumulated far from @xmath20 , the dkb set provides a better approximation to the atomic orbitals near the nucleus .
lldddd & & & & & + @xmath97 & fd & -0.1273680 & 1.114751[-1 ] & & + & dkb & -0.1273674 & 1.114741[-1 ] & & + & nd & -0.1273682 & 1.121812[-1 ] & & + @xmath98 & fd & -0.5518735[-1 ] & 3.063077[-2 ] & & + & dkb & -0.5518714[-1 ] & 3.063069[-2 ] & & + & nd & -0.5518750[-1 ] & 3.084164[-2 ] & & + @xmath99 & fd & -0.8561589[-1 ] & -1.252026[-2 ] & & + & dkb & -0.8561576[-1 ] & -1.252018[-2 ] & & + & nd & -0.8561616[-1 ] & -1.218362[-2 ] & & + @xmath100 & fd & -0.8378548[-1 ] & 4.649107[-3 ] & 6.693978[-1 ] & 8.725496[0 ] + & dkb & -0.8378543[-1 ] & 4.649117[-3 ] & 6.694037[-1 ] & 8.744759[0 ] + & nd & -0.8378538[-1 ] & 4.649454[-3 ] & 1.591405[+2 ] & 1.924786[7 ] + @xmath101 & fd & -0.6441964[-1 ] & -3.543808[-3 ] & 1.702467[-1 ] & -8.950592[1 ] + & dkb & -0.6441961[-1 ] & -3.543799[-3 ] & 1.702736[-1 ] & -9.202786[1 ] + & nd & -0.6441970[-1 ] & -3.038702[-3 ] & 1.643100[+5 ] & 3.663198[10 ] + @xmath102 & fd & -0.6452977[-1 ] & 2.257618[-3 ] & 1.562954[-1 ] & 1.938370[1 ] + & dkb & -0.6452976[-1 ] & 2.257616[-3 ] & 1.562953[-1 ] & 1.938705[1 ] + & nd & -0.6452969[-1 ] & 2.257607[-3 ] & 7.144243[+0 ] & 7.235873[5 ] + so far we examined properties of the individual basis set orbitals with sufficiently low energies .
the real power of the finite basis set technique lies in approximating the entire innumerable spectrum by a finite size quasi - spectrum .
this is important , for example , in computing sums over intermediate states ( green s functions ) in perturbation theory . from the discussion of the preceding section it is clear that the difference in quality of the nd and dkb basis sets is expected to become most apparent for the properties accumulated near the nucleus . here , as an illustrative example , we consider the parity - nonconserving ( pnc ) amplitude for the @xmath103 transition in @xmath104 .
this amplitude appears in the second order of perturbation theory for the otherwise forbidden dipole transition and it can be represented as a sum over intermediate states @xmath105 @xmath106 here @xmath107 and @xmath108 are electric - dipole and weak interaction ( pseudo - scalar ) operators , and @xmath109 are atomic energy levels .
we will compute this expression in the single - particle approximation .
specifically , the index @xmath13 runs over the entire dhf quasi - spectrum for @xmath110 partial wave , including core orbitals .
the weak hamiltonian reads @xmath111 where @xmath112 is the fermi constant , @xmath113 is the weak charge , @xmath114 is the dirac matrix ( it mixes large and small components ) , and @xmath115 is the neutron density distribution .
for consistency with the previous calculations the @xmath115 is taken to be the proton fermi distribution of ref .
notice that the matrix elements of the weak interaction are accumulated entirely inside the nucleus .
we evaluate the sum ( [ eq : epnc ] ) using the dkb and the nd sets with @xmath85 basis functions of order @xmath75 .
the integration grids are the same in both cases and include large number of points ( @xmath116 ) inside the nucleus .
the pnc amplitude is conventionally expressed in units of @xmath117 , where @xmath118 is the number of neutrons in the nucleus of @xmath89cs . in these units ,
the results are @xmath119 the finite - difference value is taken from ref .
again we note that the dkb set offers an improved performance over the notre dame set . reaching the comparable accuracy in the nd approximation
requires a larger basis set .
for example , @xmath120 nd set reproduces the dkb result for the pnc amplitude .
further insights may be gained from examining individual contributions of the intermediate states in the pnc amplitude .
we plot individual contributions of the intermediate state @xmath121 to the pnc amplitude in fig .
[ fig : pncindividual ] . both terms in eq.([eq : epnc ] )
are included .
we computed the data using @xmath122 nd and dkb basis sets generated in a cavity of @xmath91 bohr . from the plot , we observe that the dominant contribution arises from the low - lying valence states . as @xmath13 increases ,
the contributions become quickly suppressed ( there is 10 orders of magnitude suppression for @xmath123 ) .
this is due to both increased energy denominators and decreased density at the nucleus for high @xmath13 .
comparison between the basis sets reveals that their contributions are identical until @xmath124 . for higher principal quantum numbers ,
the dkb contributions monotonically decrease , while nd contributions become irregular .
moreover , at @xmath125 the nd contributions start to flip signs with increasing @xmath13 .
generally , this oscillating pattern would lead to a deterioration of the numerical accuracy .
we believe that the described irregularity is again due to the aforementioned improper behavior of orbitals near the nucleus , as the matrix elements of the weak hamiltonian is accumulated largely in this regime . to summarize , the dkb set is numerically complete and is well suited for carrying out practical summations over intermediate states in perturbation theory .
the comparison with the nd set shows that , at least for the pnc amplitude , the nd convergence pattern becomes affected by the inaccurate representation of the orbitals near the nucleus ( this is exemplified for states higher in the spectrum ) , while the dkb set exhibits a monotonic convergence . additionally , the dkb set is devoid of spurious states , and therefore the incidental inclusion of spurious states in summations over intermediate states is of no concern .
intermediate states to the pnc amplitude , eq.([eq : epnc ] ) , as a function of the principal quantum number @xmath13 .
the units of the pnc amplitude are @xmath117 .
the results from the dkb basis are represented by squares and those from the nd basis by circles .
both sets contain @xmath126 basis orbitals and use identical integration grids . due to the employed logarithmic scale ,
we plot the absolute values of the contributions .
we use hollow symbols to indicate negative contributions and filled symbols for marking positive contributions . ] here we analyze the effect that the additional term @xmath127 has on spurious states which occur in the nd method .
we start by taking @xmath128{cc } x \frac{c}{2}p\left ( 0\right ) ^{2}-\frac{c}{2}p\left ( 0\right ) q\left ( 0\right ) , & \kappa<0\\ x c^{2}p\left ( 0\right ) ^{2}-\frac{c}{2}p\left ( 0\right ) q\left ( 0\right ) , & \kappa>0\\ \end{array } \right . , \ ] ] with @xmath129 an adjustable parameter . for the case of @xmath130
this is equivalent to @xmath94 .
we mentioned previously that @xmath131 variationally encourages the boundary condition @xmath95 ; this is also true for any choice of @xmath129 .
the specific choice of @xmath129 effectively alters the degree of such variational `` encouragement . ''
we find that setting @xmath132 results in a single spurious state which appears as the lowest energy eigenstate for each @xmath1 . by subsequently increasing the value of @xmath129 towards @xmath133 we
may then deduce the effect that @xmath134 has on these spurious states as well as the rest of the spectrum .
we find that small increases in @xmath129 from @xmath132 causes the energy of the spurious state to increase , while the other states remain essentially unaffected ( except for the case of near degeneracy with one of these states ; this situation is discussed below ) . as @xmath129
is increased from zero , we may watch as spurious states for each @xmath1 first appear as the lowest energy state , then move up to the second lowest energy state , then to the third lowest energy state , etcetera . fig .
[ fig : spurshift ] displays this effect for the case of cs . at various values of @xmath129 . here
@xmath86 corresponds to the location of the spurious state in the spectrum ( e.g. , @xmath135 corresponds to the spurious state appearing as the second lowest energy state for that @xmath4 ) .
we used a set with @xmath85 b - splines of order @xmath75 confined to a cavity of radius @xmath136 bohr . ]
it is also interesting to analyze the effect of the spurious state when its energy is nearly degenerate with another state in the spectrum ( we will refer to this other state as the `` genuine '' state ) . as the energies approach degeneracy by varying @xmath129 , we observe that the spurious state begins to mix in with the genuine state .
the first evidence for this is that the energy of the genuine state starts to become affected by the presence of the spurious state .
secondly , we see that the radial wavefunctions @xmath137 and @xmath138 of the genuine state begin to acquire non - physical `` bumps '' that oscillate in a way that corresponds to the rapid oscillations of the respective spurious state radial functions . as
the energy becomes nearly degenerate , the two states mix to such a degree that it is not even possible to define one as the `` genuine state '' and the other as the `` spurious state . ''
this effect is shown in fig .
[ fig : spur4d32 ] , where the genuine state is taken as the 4@xmath139 state of cs .
as @xmath129 is increased such that the spurious state becomes embedded in the quasi - continuum part of the spectrum , it mixes with multiple neighboring states , and at this point it becomes difficult to track or even define the spurious state . of the @xmath140 ( @xmath141 ) state of cs when the spurious state is ( @xmath142 ) just below it in the spectrum ( @xmath143 ) , ( @xmath144 )
nearly degenerate with it ( @xmath145 ) , and ( @xmath146 ) just above it in the spectrum ( @xmath147 ) . for the case ( @xmath144 ) , the notation @xmath148 is used to indicate that the states are some linear combination of the two states and not easily defined as one or the other . here
@xmath149 refers to the energy in a.u . with
rest energy subtracted off . for reference ,
the nd set ( @xmath133 ) gives the lowest three @xmath139 states to have energies : @xmath150 , @xmath151 ; @xmath140 , @xmath152 ; @xmath101 , @xmath153 .
we used a set with @xmath85 b - splines of order @xmath75 confined to a cavity of @xmath136 bohr ; note that these plots only extend to @xmath154 bohr , however . ] presumably increasing @xmath129 to @xmath133 ( the notre dame case ) shifts the spurious states all the way to the end of the spectrum .
evidence for this is seen , for example , in the @xmath139 set used for fig .
[ fig : spur4d32 ] ( @xmath85 , @xmath75 , @xmath136 bohr ) . here the last few levels ( excluding the very last level ) have an energy spacing on the order of @xmath155 a.u . , whereas the energy spacing to the final level is on the order of @xmath156 a.u .
furthermore , increasing @xmath129 past @xmath130 only has a substantial effect on the energy of the final level ( for @xmath157 the energy spacing to the final level is then on the order of @xmath158 a.u . ) . up to this point
we have been exclusively considering cases with @xmath1 , as these have been the only angular symmetries for which spurious states have previously been known to occur .
now we shall consider the effect that @xmath134 has on the cases of @xmath58 .
as with the @xmath1 cases , we see that increasing @xmath129 past @xmath130 only has a substantial effect on the energy of the final level .
this is an indication that a spurious state may actually lie at the end of the spectrum for @xmath58 angular symmetries as well .
in fact , we find that setting @xmath129 to a negative value significantly below @xmath159 results in a single spurious state which appears as the lowest energy eigenstate for each @xmath58 . by subsequently increasing @xmath129 from this point
, we may watch as each spurious state is shifted towards the higher energy end of the spectrum , similar to what is observed with @xmath1 spurious states .
now we return briefly to subject of the matrix elements @xmath78 and @xmath79 of the nd basis , which are suspected to contribute to poor representation of orbitals near the nucleus . from eqs .
( [ eq : dij ] ) and ( [ eq : ndset ] ) we see that these include the integral @xmath160 ^ 2 dr$ ] , which is infinite due to the non - vanishing property of the first b - spline at @xmath20 .
numerically this integral is evaluated by gaussian quadrature , producing finite values .
we observe that for the ( @xmath85 , @xmath75 , @xmath136 bohr ) cs set , increasing the numerical value of this integral to 20 times its gaussian quadrature value results in the reappearance of spurious states as the lowest energy eigenstates for each @xmath1 .
simultaneously , the energy of highest energy eigenstate for each @xmath58 is increased substantially .
evidently the capability of @xmath127 to shift the spurious state to the end of the spectrum for @xmath1 angular symmetries is reliant upon the inaccurate numerical evaluation of this ( theoretically ) infinite - value integral .
the claim made in this section of observing spurious states for @xmath58 angular symmetries may seem surprising at first . @xcite have proved that an independent expansion of large and small radial components with a finite set of basis functions leads to a single spurious state for @xmath1 angular symmetries .
this proof assumes the basis functions to vanish at the origin , and the result of this proof is consistent with experience when such basis functions are employed .
arguably , connection is lost immediately with this proof because the nd set includes the first b - spline , which does not vanish at the origin . as we have seen here
, the nd method depends on numerical inaccuracies in evaluating infinite - value integrals in order to manipulate spurious states arising in @xmath1 cases . because the nd method also includes the first b - spline for @xmath58 angular symmetries ( and hence similar numerical inaccuracies )
, we would have no reason to discount the possibility of spurious states from occurring in these cases as well . in the preceding sections we established that the dkb sets are more robust than the nd bases .
for a comparable size of the basis set the dkb basis exhibits better numerical accuracy for properties accumulated at small radii . or , we may say that for a fixed numerical accuracy , the dkb basis may contain smaller number of basis functions . in this section
we investigate a related question : what the smallest possible basis set is for a given numerical accuracy . keeping the set as small as possible
speeds up many - body calculations that usually require multiple summations over intermediate states . also , smaller basis sets reduce storage requirements for expansion coefficients in all - order techniques such as configuration interaction or coupled - cluster methods .
quantifying the numerical accuracy requires choosing some metric , which characterizes deviation of the selected property for a given basis from its exact value .
apparently , one should select the `` metric '' so that it can be easily computed and is related to the relevant atomic properties . as an example of an optimization measure , here we choose the second - order correction to the energy of a valence electron , @xmath161 . in the frozen - core dhf approximation
@xmath161 is the leading many - body correction to the energy .
it is given in terms of the coulomb integrals @xmath162 and single particle energies @xmath109 as ( see , e.g. , ref .
@xcite ) , @xmath163 where @xmath164 is the antisymmetrized coulomb integral .
the summations are carried out over core orbitals ( labels @xmath142 and @xmath144 ) and virtual ( non - core ) orbitals ( labels @xmath165 and @xmath13 ) .
each summation implies summing over principal quantum numbers , angular numbers @xmath4 , and magnetic quantum numbers .
the summation over magnetic quantum numbers may be carried out analytically and we are left with summations over radial functions .
we define a contribution of an individual partial wave @xmath6 , @xmath166 , as the difference @xmath167 , where @xmath168 stands for truncated eq .
( [ eq : e2v ] ) ; it includes summations over intermediate states ( both core and virtual ) with the orbital angular momentum up to @xmath6 .
since the calculations necessarily involve coulomb integrals between orbitals of different angular momenta , the numerical error in @xmath169 is affected by the accuracy of representation of all partial waves up to @xmath6 .
first , in table [ tab : ensecond ] , we present results for a large set of @xmath126 basis functions for each partial wave .
we use a sufficiently large cavity of @xmath91 bohr in this calculation .
these results will serve as a benchmark for comparisons with the less complete ( optimized ) sets .
we observe that the dominant ( 60% ) contribution comes from @xmath170-waves , @xmath171 .
qualitatively , the second - order energy correction may be described as core - polarization effect .
the outer @xmath172-orbitals of the core are relatively `` soft '' and are easily polarizable .
since the core does not contain @xmath173 and higher partial waves , after peaking at @xmath171 , the partial - wave contributions become suppressed as @xmath6 increases .
qualitatively , this suppression arises due to increased centrifugal repulsion of higher partial waves and associated smaller overlaps with core orbitals in the coulomb integrals of eq .
( [ eq : e2v ] ) .
for example , @xmath174 contributes only 0.6% of the total value .
now we would like to minimize the size of the set by choosing a different number of radial basis functions @xmath175 for different angular symmetries @xmath4 . to preserve a numerical balance between the fine - structure components ( for example
, this may be important while recovering non - relativistic limit ) we keep the same number of functions for a given orbital angular momentum @xmath6 , e.g. , @xmath176 . in table
[ tab : ensecond ] , we present an example of an optimized basis ( marked as `` small set '' ) . we also list resulting numerical errors for each partial wave by comparing @xmath177 with the result from the `` large set '' calculations .
we see that while the basis - set error in higher partial waves is as large as 40% , this hardly makes any influence on the total value of the correlation energy , because of contributions of higher @xmath6 are suppressed .
the total value of the correlation correction differs by about 1% from it s saturated value . considering that the correlation contribution to the energy is about 10% for the @xmath178 state
, the less complete set would introduce only 0.1% error for the total ionization energy .
so far we discussed the correlation energy correction for the @xmath97 state .
the optimized set remains sufficiently robust for other low - lying states as well .
we have carried out a comparison similar to table [ tab : ensecond ] for @xmath99 , @xmath100 , @xmath101 , and @xmath102 states . in all these cases the difference between the total @xmath179 values computed with the``large '' and `` small '' sets is about 0.5% . in practical calculations one
is often required to reproduce a number of properties with the same set .
the atomic properties may be quite dissimilar like hyperfine - structure interactions accumulated near the nucleus and dipole matrix elements determined by the valence region .
apparently , one has to carry out a similar low - order mbpt analysis for the relevant quantities to verify the suitability of the optimized set . from the computational point of view , using the optimized sets speeds up the numerical evaluations . in our illustrative example , the `` large set '' contains @xmath180 orbitals , while the `` small set '' is about four times smaller ( @xmath181 orbitals ) .
the resulting speed - up is sizable : our computations of eq .
( [ eq : e2v ] ) for the @xmath97 state ( @xmath174 ) are about 14 times faster with the optimized set , as expected due to @xmath182 scaling of the number of contribution in the most computationally demanding second term of eq.([eq : e2v ] ) .
similar scaling should hold for storage of expansion coefficients in all - order methods , e.g. , for storing triple excitations @xcite one expects @xmath183 scaling of the storage size .
usually higher partial waves produce larger number of angular channels and the scaling should be even steeper than @xmath183 .
further speed - up in mbpt summations and reduction in storage size may be attained by skipping a few basis set functions at the upper end of the quasispectrum . such additional truncation of the spectrum becomes apparent from fig .
[ fig : pncindividual ] , where contributions to the pnc amplitude for the upper three - quarters ( @xmath184 out of @xmath126 ) of the quasi - spectrum affect the total value below @xmath185 level of accuracy . .
contribution of individual partial waves @xmath6 to the second - order energy correction for the ground @xmath186 state of cs atom .
we use the dkb basis set of @xmath187 basis functions constructed from a subset of b - splines of order @xmath44 ( label @xmath188 ) .
calculations are carried out in a cavity of @xmath91 bohr for two basis sets `` large '' and `` small '' .
numerical integration grids are identical for both sets .
the column marked `` error '' refers to a relative error in a given partial - wave contribution , @xmath189 , caused by switching from the `` large '' to the `` small '' basis set .
[ tab : ensecond ] [ cols="<,^,<,^ , < , > " , ]
calculations of certain atomic properties , such as parity - violating amplitudes , transition polarizabilities between hyperfine levels , and many - body corrections to hyperfine interactions require accurate representation of atomic orbitals at both small and intermediate electron - nucleus distances . solving the many - body problem in high orders of perturbation theory
additionally calls for an efficient representation in terms of the basis sets .
the dual - kinetic - basis set is shown here to adequately meet both these demands .
previously the dkb basis was applied to systems with a single electron in a coulomb field , i.e. , hydrogen - like ions@xcite .
here we extended the dkb method to many - electron systems by generating the single - particle quasi - spectrum of the dirac - hartree - fock potential .
several numerical example for cs atom were presented .
we showed that the dkb method outperforms the widely - employed notre dame b - spline method @xcite for problems involving matrix elements accumulated at small distances .
in addition , we presented a strategy for optimizing the size of the basis sets by choosing progressively smaller number of basis functions for increasingly higher partial waves .
this strategy exploits suppression of contributions of high partial waves to typical many - body correlation corrections .
we would like to thank walter johnson , charlotte froese - fischer , and vladimir shabaev for discussions .
the developed dual - kinetic - balance code for the dirac - hartree - fock problem was partially based on codes by the notre dame group .
this work was supported in part by the national science foundation .
|
the dual - kinetic - balance ( dkb ) finite basis set method for solving the dirac equation for hydrogen - like ions [ v. m. shabaev _ et al .
_ , phys .
rev .
lett . * 93 * , 130405 ( 2004 ) ] is extended to problems with a non - local spherically - symmetric dirac - hartree - fock potential .
we implement the dkb method using b - spline basis sets and compare its performance with the widely - employed approach of notre dame ( nd ) group [ w.r .
johnson and j. sapirstein , phys .
rev .
lett . * 57 * , 1126 ( 1986 ) ] .
we compare the performance of the nd and dkb methods by computing various properties of cs atom : energies , hyperfine integrals , the parity - non - conserving amplitude of the @xmath0 transition , and the second - order many - body correction to the removal energy of the valence electrons .
we find that for a comparable size of the basis set the accuracy of both methods is similar for matrix elements accumulated far from the nuclear region .
however , for atomic properties determined by small distances , the dkb method outperforms the nd approach .
in addition , we present a strategy for optimizing the size of the basis sets by choosing progressively smaller number of basis functions for increasingly higher partial waves .
this strategy exploits suppression of contributions of high partial waves to typical many - body correlation corrections .
|
in this paper we classify holomorphic selfmaps of the complex projective plane @xmath0 that are integrable in the quite specific sense that they preserve an algebraic web .
recall that an _ algebraic web _ is given by a reduced curve @xmath1 , where @xmath2 is the dual projective plane consisting of lines in @xmath0 .
we say that the web is irreducible if @xmath3 is an irreducible curve .
the web is _ invariant _ for a holomorphic mapping @xmath4 if every line in @xmath0 belonging to @xmath3 is mapped to another such line .
see sections [ sec : webs ] and [ sec : invariance ] for details .
we will assume @xmath5 is noninvertible .
if @xmath3 is irreducible , it is of one of the following types : * a line ; * a smooth conic ; * a smooth cubic ; * a nodal cubic .
the maps in ( iii ) and ( iv ) are always critically finite . if @xmath3 is reducible , it is of one of the following types : * the union of two lines ; * the union of three lines in general position ; * the union of a conic and a line in general position
. the maps appearing in theorem a and b are quite rare among all holomorphic selfmaps , but nevertheless interesting .
indeed , they provide concrete examples of critically finite mappings .
see @xcite for examples and dynamics of critically finite maps , and @xcite for a survey of iterations of rational maps on projective spaces .
there are several other notions of integrability for selfmaps of @xmath0 . in
@xcite we classified invariant pencils of curves .
much more generally , favre and pereira @xcite have classified invariant foliations for rational maps .
the case of birational maps was studied earlier by cantat and favre @xcite .
related work includes the classification of totally invariant curves for holomorphic endomorphisms of @xmath0 @xcite and for birational maps of surfaces @xcite ; see also @xcite .
this note is organized as follows .
after some background in sections [ sec : webs ] and [ sec : invariance ] we describe in section [ sec : examples ] the mappings appearing in theorem a. the proof takes place in section [ sec : proof ] and section [ sec : reducible ] treats the case of a reducible web .
we thank the referee for a careful reading of the paper and many useful suggestions .
the second author was supported by the nsf . the work was partially completed during a visit to the mittag - leffler institute .
we start by reviewing some elementary facts of plane geometry .
let @xmath0 denote the complex projective plane .
the support of a divisor @xmath6 on @xmath0 is denoted by @xmath7 .
let @xmath2 the dual projective plane , that is , the set of complex lines in @xmath0
. then @xmath2 is itself isomorphic to the projective plane .
let @xmath8 denote the involutive duality between lines ( resp .
points ) in @xmath0 and points ( resp .
lines ) in @xmath2 . given a point @xmath9 ,
@xmath10 is the line of lines passing through @xmath11 . given a line @xmath12 , @xmath13 .
consider a reduced , irreducible curve @xmath14 ( resp .
@xmath15 ) of degree @xmath16 .
the _ dual curve _ @xmath17 ( resp @xmath18 ) is the curve of tangents to the local branches to @xmath19 .
if @xmath20 is a normalization map , then @xmath21 defined by @xmath22 , is also a normalization map . here
@xmath23 ( resp .
@xmath24 ) is the tangent line to the irreducible curve germ @xmath25 at @xmath26 .
the double dual @xmath27 is isomorphic to @xmath19 .
we shall need to compare the degree and singularities of a curve with those of its dual . to this end , define the ramification divisor of @xmath20 to be @xmath28 . here
@xmath29 is the largest integer @xmath30 such that @xmath31 , where @xmath32 denote the maximal ideals .
we then have the following plcker - type formula : @xmath33 where @xmath34 is the topological euler characteristic of @xmath35 .
this is proved using the riemann - hurwitz formula as in @xcite ; see also @xcite . a _ web _ @xmath36 on @xmath0 of degree @xmath37 is locally defined by an unordered set of @xmath37 holomorphic ( possibly singular ) foliations . in particular , through a general point @xmath9 passes exactly @xmath37 leaves , and these intersect transversely at @xmath11 .
globally , the leaves may exhibit complicated behavior and even be dense in @xmath0 .
see e.g. @xcite for general facts on webs .
we shall only consider the particular case of an _ algebraic web _ on @xmath0 . by definition , this is a web @xmath38 given by a reduced curve @xmath1 of degree @xmath39 .
the leaves of @xmath40 are exactly the lines in @xmath0 corresponding to the points on @xmath3 . through a generic point @xmath9
passes exactly @xmath37 distinct lines of the web
. see figure [ f1 ] for a picture of the algebraic web associated to a conic @xmath3 .
assume that @xmath3 is irreducible and @xmath41 . the normalization map @xmath42 then induces a symmetric rational map @xmath43 defined by @xmath44 , where @xmath45 is the line passing through ( distinct ) points @xmath46 .
let us record some facts that are easily established by direct computation .
the indeterminacy locus @xmath47 of @xmath48 is exactly the set of pairs @xmath49 with @xmath50 and pairs @xmath51 with @xmath52 but @xmath53 . in particular
, @xmath3 is smooth if and only if @xmath48 is holomorphic .
we have @xmath54 , where @xmath55 denotes the diagonal and @xmath56 the dual curve .
more precisely , @xmath57 . note that @xmath48 has topological degree @xmath58 .
further , @xmath48 is biholomorphic outside @xmath59 .
consider a smooth algebraic curve @xmath35 of genus @xmath60 and a surjective holomorphic map @xmath61 of topological degree @xmath62 .
the canonical divisor class @xmath63 has degree @xmath64 .
the riemann - hurwitz formula asserts @xmath65 , where @xmath66 is the ramification divisor .
taking degrees , we find @xmath67 . as @xmath62 , we have @xmath68 or @xmath69 , that is , @xmath35 is a rational or elliptic curve .
a subset @xmath70 is _ totally invariant _ if @xmath71 . for finite subsets
@xmath72 , this in fact implies @xmath73 .
when @xmath74 is an elliptic curve , @xmath75 lifts to an affine map @xmath76 and one easily sees that there is no finite totally invariant set . when @xmath35 is rational , it is not hard to prove that a totally invariant set contains at most two points , see e.g. ( * ? ? ? * theorem 1.5 , p.56 ) .
a holomorphic map @xmath4 can be written in homogeneous coordinates as @xmath77=[p(x , y , z):q(x , y , z):r(x , y , z)]$ ] , where @xmath78 , @xmath79 and @xmath80 are homogeneous polynomials on @xmath81 of the same degree @xmath82 , and @xmath83 .
the number @xmath84 is the _ algebraic degree _ of @xmath5 ; we shall assume @xmath62 .
the topological degree of @xmath5 is @xmath85 .
the ramification divisor @xmath86 of @xmath5 has degree @xmath87 .
it is known @xcite that if @xmath88 is a ( reduced , but possibly reducible ) curve such that @xmath89 , then @xmath90 and @xmath3 is the union of at most three lines .
any such line @xmath91 occurs with multiplicity @xmath92 in @xmath86 .
consider the algebraic web @xmath40 associated to a curve @xmath1 of degree @xmath37 .
assume for now that @xmath3 is irreducible .
let @xmath4 be a holomorphic mapping of algebraic degree @xmath93 .
we say that the web is _ invariant _ under @xmath5 if the image under @xmath5 of any line in the web is again a line in the web .
there is then an induced selfmap @xmath94 defined by @xmath95 .
the web is _ totally invariant _ if the preimage of any line in the web is a union of lines in the web .
a web is invariant if and only if it is totally invariant . moreover , if the web is invariant , then : * the induced map @xmath94 is regular , of topological degree @xmath84 ; * for any @xmath96 , we have @xmath97 as divisors on @xmath0 . clearly , the web is invariant if it is totally invariant .
also , ( ii ) is clear , since @xmath98 multiplies the degree of any effective divisor by @xmath84 . for @xmath99 ,
let @xmath100 denote the space of effective divisors of degree @xmath30 on @xmath0 .
then @xmath101 . let @xmath102 be the veronese map given by multiplication by @xmath84 : @xmath103 and @xmath104 the pushforward map induced by @xmath5 .
then @xmath98 is holomorphic , of topological degree @xmath85 , and @xmath105 is a holomorphic embedding .
suppose the web associated to @xmath3 is invariant .
then @xmath106 , so @xmath60 is holomorphic . now @xmath107 and @xmath108 .
restricting to @xmath3 implies @xmath109 , so @xmath60 has topological degree @xmath84 . in particular
, the preimage of every line in the web is the union of ( at most ) @xmath84 lines in the web , so the web is totally invariant . the induced selfmap @xmath94 preserves collinearity : if @xmath110 are collinear in @xmath2 , then so are @xmath111 .
indeed , @xmath112 .
conversely , if @xmath94 is a surjective holomorphic map preserving collinearity , then there is a unique holomorphic mapping @xmath4 satisfying @xmath95 . clearly , @xmath94 lifts uniquely through the normalization map @xmath42 to a holomorphic selfmap @xmath61 of topological degree @xmath62 .
in particular , @xmath35 is rational or elliptic .
moreover , @xmath4 lifts through the rational map @xmath113 to the selfmap @xmath114 .
this implies @xmath115 . for the proof of theorem
a we need to compare the ramification divisors @xmath116 and @xmath117 of @xmath5 and @xmath75 , respectively . in general
, @xmath86 has degree @xmath87 . since @xmath5 preserves the algebraic web associated to @xmath1 , we can write @xmath118 , where @xmath119 is the part of @xmath86 supported on the lines of the web , and @xmath120 is the `` sectional '' part of @xmath86 .
[ l : fibercritset ] if @xmath121 and @xmath122 , then the multiplicity of the point @xmath123 in @xmath66 equals the multiplicity of the line @xmath124 in @xmath125 .
if @xmath126 , @xmath127 and @xmath128 , then @xmath113 is a local biholomorphism at @xmath51 . picking @xmath129 generic gives the result .
we now go through the examples appearing in theorem a and briefly discuss their dynamics . the case
when @xmath3 is a line corresponds to a pencil of lines through a point @xmath9 . in other words , @xmath130 .
we have @xmath131 and @xmath132 .
pick homogeneous coordinates @xmath133 $ ] on @xmath0 such that @xmath134 $ ] .
then @xmath5 preserves @xmath3 if and only if it takes the form @xmath77=[p(x , y):q(x , y):r(x , y , z)]$ ] .
holomorphic selfmaps of @xmath2 preserving a pencil of curves were classified in @xcite .
their dynamics is studied in @xcite .
suppose @xmath1 is a smooth conic so that @xmath135 .
pick any holomorphic selfmap @xmath136 of degree @xmath62 .
the map @xmath137 is then holomorphic , of topological degree 2 , and @xmath138 induces a holomorphic selfmap @xmath4 of algebraic degree @xmath84 .
any holomorphic selfmap @xmath5 of @xmath0 preserving a conic @xmath1 is of this form .
we have @xmath131 and @xmath132 .
the dual curve @xmath56 is an invariant smooth conic and @xmath139 , see lemma [ l : sectcritset ] .
selfmaps as above were first introduced in a dynamic setting by ueda @xcite .
they can be used to provide simple examples of critically finite maps , in particular maps whose julia set is all of @xmath0 ( * ? ? ?
* proposition 4.1 ) .
let @xmath1 be a smooth cubic .
then @xmath140 is an elliptic curve and admits a group law : three points in @xmath3 are collinear if and only if their sum is zero .
choose the origin of the group law @xmath141 at any flex of @xmath3 and consider any holomorphic selfmap @xmath94 .
then @xmath60 preserves collinearity if and only if its translation factor is a flex of @xmath3 . in this case
, @xmath60 induces a selfmap @xmath4 preserving the web associated to @xmath3 , and any such @xmath5 is of this form .
we have @xmath142 and @xmath143 .
the dual curve @xmath56 is an invariant sextic with nine cusps and @xmath139 , see lemma [ l : sectcritset ] . in particular
, @xmath5 is always critically finite , @xmath144 .
it follows from ( * ? ? ?
* theorem 5.9 ) that the julia set of @xmath5 is all of @xmath0 .
let @xmath1 be a nodal cubic , with the node at @xmath145 .
there is a geometrically defined multiplicative group law on @xmath146 , given as follows : @xmath147 in the group if and only if @xmath148 , @xmath149 and @xmath150 are collinear in @xmath2 . here
the unit element @xmath151 can be chosen as any of the three flexes of @xmath3 . concretely ,
the unique normalization map @xmath152 such that @xmath153 and @xmath154 restricts to a group homomorphism @xmath155 .
we then see that @xmath94 preserves collinearity if and only if @xmath61 takes the form @xmath156 , where @xmath157 .
in fact , by changing the flex representing @xmath151 , we obtain @xmath158 .
we have @xmath159 and @xmath160 .
the dual curve @xmath56 is an invariant quartic with three cusps and one bitangent .
we have @xmath139 , see lemma [ l : sectcritset ] .
in particular , @xmath5 is critically finite . in suitable coordinates on @xmath2 and @xmath0
, we have @xmath161 , @xmath162 $ ] , @xmath163 $ ] and @xmath164 $ ] .
the selfmap of @xmath0 associated to @xmath165 is then a polynomial mapping @xmath166 of the form @xmath167 , where @xmath168 .
for example , @xmath169 .
without loss of generality , assume that @xmath170 and that @xmath3 is not a smooth cubic .
we shall prove , in a self - contained way , that @xmath3 is a nodal cubic .
an alternative approach , suggested by the referee , is to use a result in web theory to first show that @xmath171 : see remark [ rmk : referee ] .
we need two results , the proofs of which are given below .
let @xmath172 denote the multiplicity of the curve @xmath3 at a point @xmath173 .
[ l : singtotinv ] the singular locus @xmath174 is totally invariant for @xmath94 . moreover ,
if @xmath175 , then @xmath176 . as a consequence ,
the set @xmath177 is totally invariant for @xmath61 . for the second result , recall the notation @xmath119 and @xmath120 for the fiber and sectional parts of the ramification locus of @xmath5 , respectively .
[ l : sectcritset ] we have @xmath178 with equality if and only if @xmath179 as divisors .
now let us prove theorem a. the map @xmath61 has topological degree @xmath62 , so @xmath35 has to be a rational or elliptic curve . since we have assumed @xmath180
, @xmath3 is singular .
lemma [ l : singtotinv ] implies that @xmath181 is a totally invariant set for @xmath61 .
this is impossible if @xmath35 is elliptic , so @xmath3 must be rational .
moreover , @xmath182 consists of one or two points .
each such point corresponds to a line in @xmath0 that is totally invariant for @xmath5 , and hence contributes to @xmath119 as a divisor of degree @xmath92 .
* case 1 * : @xmath183 .
the point in @xmath182 contributes a line of multiplicity @xmath92 to @xmath119 . by lemma [ l : fibercritset ]
, the critical points of @xmath75 in @xmath184 contribute lines of total degree @xmath92 in @xmath119 .
thus @xmath185 .
lemma [ l : sectcritset ] gives @xmath186 , so that @xmath187 , and hence @xmath3 , is a conic .
this contradicts the assumption @xmath180 . * case 2 * : @xmath188 .
then @xmath61 has no critical points outside @xmath182 , so @xmath185 or @xmath189 , depending on whether @xmath190 or @xmath191 . if @xmath185 , lemma [ l : sectcritset ] implies @xmath186 , so that @xmath187 , and hence @xmath3 is a smooth conic , contradicting @xmath192 .
hence suppose @xmath191 and @xmath160 .
. then @xmath194 . by lemma [ l : singtotinv ] , @xmath195 .
it follows from applied to @xmath196 that @xmath197 .
now lemma [ l : sectcritset ] shows that @xmath198 .
we have assumed @xmath192 , hence @xmath199 . since @xmath191 and @xmath200
, @xmath3 is a nodal cubic .
we may assume that @xmath180 or else the irreducible curve @xmath3 is smooth , and there is nothing to prove .
note that for any @xmath96 , a generic line @xmath201 through @xmath173 intersects @xmath3 in exactly @xmath202 points .
dually , through a generic point on @xmath203 passes exactly @xmath202 distinct lines of the web .
in particular , @xmath204 if and only if a generic point on the line @xmath203 belongs to @xmath37 distinct lines of the web . now consider @xmath96 and @xmath205 .
let @xmath206 be a generic point and write @xmath207 .
assume that @xmath208 .
we will show that @xmath175 .
this will prove that @xmath174 is totally invariant .
first assume @xmath209 .
then @xmath5 is a local biholomorphism at @xmath11 , so @xmath11 and @xmath210 belong to the same number of lines of the web .
thus @xmath175 .
now assume @xmath211 .
then the kernel of the differential @xmath212 is one - dimensional .
thus there is at most one line @xmath213 in @xmath0 through @xmath11 such that the curve germ @xmath214 is transverse to the line @xmath215 at @xmath210 .
this implies that @xmath216 , that is , @xmath217 .
the reverse inequality always holds ( since @xmath3 is irreducible and not a line ) , and so @xmath176 .
in particular @xmath175 .
thus @xmath174 is totally invariant .
this implies that @xmath211 for every @xmath218 .
the above argument then shows @xmath176 for every @xmath175 .
the dual curve @xmath187 is the set of points in @xmath0 belonging to @xmath219 lines of the web .
it is therefore clear that @xmath220 .
in fact , we have @xmath221 since no line in the web is mapped into the irreducible curve @xmath187 .
write @xmath222 where @xmath223 , @xmath224 and @xmath225 are irreducible components of @xmath226 .
. then @xmath228 and @xmath229 equality holds if and only if @xmath230 , @xmath231 for all @xmath232 and @xmath233 .
we now prove theorem b. let @xmath234 be the irreducible components of @xmath3 and write @xmath235 . replacing @xmath5 by an iterate , we may assume that the web associated to each @xmath234 is ( totally ) invariant for @xmath5 .
write @xmath236 for any @xmath232 .
first assume @xmath237 for some @xmath232 , say @xmath238 .
from the analysis above we have @xmath239 . for @xmath240 ,
this implies @xmath241 and @xmath242 , hence also @xmath243 .
taking degrees and consulting section [ sec : examples ] we see that @xmath244 must be a conic .
it has to intersect each line @xmath245 , @xmath240 , in two distinct points @xmath148 , @xmath149
. then @xmath246 . in particular
, the line @xmath245 is unique .
thus @xmath247 , where @xmath244 is a conic , @xmath248 is a line and @xmath249 .
as @xmath5 preserves the web @xmath250 , it comes from a selfmap of @xmath251 for which @xmath148 and @xmath149 are totally invariant .
conversely , if @xmath60 is such a selfmap , the associated map @xmath4 leaves the lines @xmath252 and @xmath253 totally invariant .
one can then check that @xmath5 preserves the linear pencil of lines through @xmath254 , that is , the web @xmath250 .
now suppose all the irreducible components of @xmath3 are lines .
we can not have three concurrent lines , as then @xmath5 would admit a totally invariant line @xmath255 such that the restriction of @xmath5 to @xmath255 would have three totally invariant points .
we also can not have four lines , as @xmath5 then would admit five totally invariant lines .
[ rmk : referee ] as the referee points out , one can use known results in web geometry to directly show that @xmath171 in theorems a and b. indeed , it is known that if @xmath259 and @xmath260 are germs of linear webs of degree @xmath261 on @xmath262 ( i.e. the leaves are lines and through a general point passes at least four leaves ) and @xmath263 is a local biholomorphism mapping leaves to leaves , then @xmath264 extends to an automorphism of @xmath265 . if @xmath266 , one can get a contradiction by taking @xmath264 as the germ of @xmath4 at a generic point .
the result above goes back to the hamburg school of web geometry : see ( * ? ? ?
* section 42 ) or ( * ? ? ?
* corollaire 2 , p.535 ) for a precise modern reference .
see also @xcite for related results .
a. m. bonifant and m. dabija .
_ self - maps of @xmath267 with invariant elliptic curves_. in _ complex manifolds and hyperbolic geometry ( guanajuato , 2001 ) _ , contemp .
, 311 , 125 .
soc . , providence , ri , 2002 .
fornss and n. sibony .
_ critically finite maps on @xmath269_. in _ the madison symposium on complex analysis ( madison , wi,1991 ) _ , contemp .
, 137 , 245260 .
soc . , providence , ri , 1992
. t. ueda .
_ complex dynamical systems on projective spaces_. in _ chaotic dynamical systems_. advanced studies in dynamical systems * 13 * ( 1993 ) , 120138 .
isbn 978 - 9 - 81 - 021276 - 6 .
world sci .
river edge , nj .
|
we classify noninvertible , holomorphic selfmaps of the projective plane that preserve an algebraic web . in doing so
, we obtain interesting examples of critically finite maps .
|
elemental and isotopic abundances are the fossils of galactic archaeology . different elements are produced from stars on different timescales , therefore elemental and isotopic abundance ratios provide independent information on the `` age '' of a system and can be used as a form of `` cosmic clock '' .
the formation and evolutionary history of galaxies can be constrained in theoretical models by using the observed abundances of stars .
the space astrometry missions ( e.g. , gaia ) and large - scale surveys ( e.g. , hermes on the aat ) will produce unprecedented information on the chemodynamical structure of the milky way galaxy . theoretically @xcite succeeded in reproducing the average evolution of major elements ( except for ti ) in the solar neighbourhood including normal type ii supernovae ( sne ii ) and hypernovae ( hne ) . in this paper
we extend the discussion to isotope ratios ( 2 ) and more realistic chemodynamical simulations ( 3 ) . from the observed abundances ,
we discuss the possibility of non - standard chemical enrichment in the very metal - poor damped lyman @xmath3 system ( dla , 4 ) , globular clusters ( gcs , 5 ) , and dwarf spheroidal galaxies ( dsphs , 6 ) .
we include the latest chemical enrichment input into our chemical evolution models , which are summarised in @xcite .
the basic equations of galactic chemical evolution are described in @xcite .
figure [ fig : iso ] shows the evolution of the isotope ratios against [ fe / h ] for the solar neighbourhood
. in general , core - collapse supernovae are the main producers of the major isotopes with more minor isotopes synthesized at higher metallicity .
for this reason , the evolution of major to minor isotope ratios continuously decreases toward higher metallicity .
the slope changes at [ fe / h ] @xmath5 and @xmath6 are due to the onset of intermediate- and low - mass asymptotic giant branch ( agb ) stars , respectively .
the rapid change in the slope at [ fe / h ] @xmath7 is caused by type ia supernovae ( sne ia ) .
the time evolution of isotope ratios depends on the star formation and chemical enrichment histories of the system .
see @xcite for the models of bulge , thick disk , and halo .
in general , the ratios between the major and minor isotopes such as @xmath8mg/@xmath2 mg are smaller in the bulge and thick disk , and are larger in the halo because of the metallicity effect of supernovae
. however , the @xmath9o/@xmath10o ratio in the halo is low due to the production of @xmath10o from low - mass agb stars , as also seen in the high [ ( c , f)/fe ] abundances .
therefore , the isotopic ratios can be used as a tool to pick out the stars that form in a system with a low chemical enrichment efficiency .
this may be possible in our halo , but more likely in small satellite galaxies that were accreted onto our milky way galaxy .
evolution of isotope ratios against [ fe / h ] for the solar neighbourhood with agb yields .
see @xcite for the observational data sources for c , mg , and ti ( dots ) .
the solar ratios are shown with the solar symbols at [ fe / h ] @xmath11 in the upper panels .
, width=359 ]
in a real galaxy , the star formation history is not so simple and the interstellar medium ( ism ) is not homogeneous . in chemodynamical simulation @xcite ,
hydrodynamics and chemical enrichment are solved self - consistently through the galaxy formation .
star formation and chemical enrichment depend on the local density and mixing due to dynamical effects such as merging and migration are also naturally included .
figure [ fig : xfe ] shows the frequency distribution of stars in the solar neighborhood at present .
see @xcite for the bulge and the thick disk . in our simulations ,
the metallicity of the first enriched stars reaches [ fe / h ] @xmath12 . at later times ,
the star forming region becomes denser , and both metal richer and poorer stars than [ fe / h ] @xmath12 appear .
different from one - zone chemical evolution models , the following phenomena occur in the case of inhomogeneous enrichment : i ) the age - metallicity relation is weak . in other words ,
the most metal - poor stars are not always the oldest stars .
ii ) sne ia can affect the elemental abundance ratios at [ fe / h ] @xmath13 even with the metallicity inhibition of sne ia .
the sn ia contribution is characterised by low [ @xmath3/fe ] and high [ mn / fe ] .
iii ) the scatter of elemental abundance ratios becomes large if the supernova yield depends on progenitor metallicity such as na and iv ) some of cemp / nemp stars can be explained with the local enrichment from agb stars even without the binary effect .
in fact , the observed [ n / o]-[o / h ] trend can be reproduced with our simulation without including the effect of rotating massive stars .
the observations of very metal - poor dlas have opened a new window to study the chemical enrichment of the universe by the first generations of stars .
figure [ fig : dla ] shows the elemental abundance ratios from c to zn relative to fe .
we implement two cases with different explosion energies : @xmath14 erg ( sn , solid line ) and @xmath15 erg ( hn , short - dashed line ) for an initial mass of @xmath16 and metallicity @xmath17 ( see @xcite for the details ) .
an efficient mixing - fallback is adopted in both models , black hole masses are as large as @xmath18 , and thus the ejected iron mass is much smaller than `` normal '' sne / hne in @xcite that are responsible for the chemical enrichment in the galaxy . since c is synthesized in the outermost region of the ejecta , [ c / fe ] ratio is higher for faint sne / hne than `` normal '' sne / hne .
we then compare these model calculations with the available abundance measurements in the extremely metal - poor dla reported by ( * ? ? ?
* filled circles ) .
this dla was originally identified in the sdss spectrum of the qso j0035@xmath190918 .
follow - up high resolution spectroscopy indicated @xmath20 , @xmath21 @xmath22 , and [ fe / h ] @xmath23 .
1 ) a pronounced carbon enhancement [ c / fe]@xmath24 is found , which is well reproduced with our faint sn / hn models .
low - mass agb stars ( @xmath25 ) can also provide such high [ c / fe ] . however , such low - mass stars are unlikely to contribute at the redshift of the c - rich dla . with the star formation history in the solar neighborhood
, [ c / fe ] reaches the maximum value at @xmath26 due to the agb contribution @xcite .
2 ) the [ o / fe ] ratio is higher than for non - faint supernovae ( [ o / fe ] @xmath27 for @xmath28 ) and is consistent with the faint sn model because of the smaller ejected fe mass .
3 ) the [ ( si , s)/fe ] ratios are similar to those of non - faint supernovae , and also consistent with the faint sn / hn models .
4 ) the low al abundance strongly suggests that the enrichment source is not pop ii supernovae but primordial supernovae .
5 ) the low n abundance is also consistent with the faint sn / hn models , and the high [ c / n ] ratio can not be explained with mass loss from rotating massive stars or intermediate - mass agb stars ( @xmath29 ) . the elemental abundance pattern of the metal - poor c - rich dla ( filled circles ) and peculiar dla ( open circles ) .
the solid and short - dashed lines show the nucleosynthesis yields of faint core - collapse supernovae from @xmath28 stars with mixing - fallback .
the dotted line is for pisne from @xmath30 stars .
, width=283 ] in order to discuss the detailed explosion mechanism of the supernova , it is necessary to obtain the elemental abundances of iron - peak elements . for the c - rich dla ,
because of the low metallicity , it is impossible to detect heavier elements than s except for fe . in fig .
[ fig : dla ] we overplot the element abundances recently reported by ( * ? ? ?
* open circles ) for the @xmath31 dla in front of the gravitationally lensed quasar um637a .
this abundance pattern can also be explained with faint supernovae .
the dotted line is for the nucleosynthesis yields of the pair instability supernova ( pisn ) of a @xmath32 star @xcite .
compared with core - collapse supernovae , the abundance pattern of pisne can be summarized as follow : 1 ) the odd - z effect is much larger than @xmath33 dex .
2 ) @xmath34}$ ] are much larger than [ ( o , mg)/fe ] because of more extensive explosive oxygen burning .
3 ) @xmath35}$ ] is much larger because of the larger incomplete si - burning region .
4 ) @xmath36}$ ] are much smaller because of the much larger ratio between the complete and incomplete si - burning regions .
all of these characteristics disagree with the observed elemental abundances of these metal - poor dlas . even at high - redshift
, there is no signature of the existence of pisne .
the [ si / c ] for pisne is as large as @xmath37 , which is also inconsistent with the observational estimate in the intergalactic medium ( igm , [ si / c ] @xmath38 ) .
the igm abundance looks more consistent with normal ( non - faint ) core - collapse supernovae with [ c / fe ] @xmath39 and [ si / fe ] @xmath40 @xcite . in the early stages of chemical enrichment
, the ism is supposed to be highly inhomogeneous , so that the properties of the first objects can be directly extracted from the comparison between the observed elemental abundances and nucleosynthesis yields .
since the dla abundances reflect the chemical enrichment in gas - phase , the binary or accretion scenarios of the emp stars do not work .
to explain the abundance pattern of the c - rich dla , enrichment by primordial supernovae is the best solution .
it is interesting that the observed dla abundance is very similar to those of emp stars in the solar neighborhood including the ultra metal - poor star he0557 - 4840 ( [ fe / h ] @xmath41 , [ c / fe ] @xmath42 ) and bd+44@xmath43493 .
some of emp stars in dsphs and the galactic outer halo also show similar carbon enhancement at [ fe / h ] @xmath44 .
chemical enrichment by the first stars in the first galaxies is likely to be driven by core - collapse supernovae from @xmath45 stars , which may be consistent with the latest simulations of primordial star formation ( e.g. , * ? ? ?
fluorine is an intriguing , though currently poorly studied element .
@xmath46f is mainly produced by core and shell he - burning at @xmath47 @xmath48 k in low - mass and massive stars .
the production of f in agb stars is highly mass dependent , where f production peaks at @xmath49 at solar metallicity ; in higher mass models f is destroyed by @xmath3-captures caused by the higher temperatures reached during he - burning . in @xcite , we have calculated the nucleosynthesis yields of core - collapse supernovae ( sne and hne ) including the @xmath50-process .
most of the energy from core - collapse supernovae is released as neutrinos and anti - neutrinos ( @xmath51 erg ) .
the @xmath50-process does not affect the yields of major elements such as fe and @xmath3 elements , but it increases those of some elements such as b , f , k , sc , v , mn , and ti .
figure [ fig : fo ] shows the evolution of [ f / o ] against [ o / h ] . without the agb yields and the @xmath50-process ( short - dashed line ) , the predicted f abundance is too low to meet the observational data at all metallicities . with the agb yields ( long - dashed line ) ,
[ f / o ] shows a rapid increase from [ o / h ] @xmath52 toward higher metallicities , which corresponds to the timescale of @xmath53 stars in the solar neighborhood . at [ o / h ]
@xmath39 , [ f / o ] reaches @xmath54 , which is @xmath55 dex larger than the case without the agb yields . however , the present [ f / o ] ratio is still significantly lower than the observations at [ o / h ] @xmath39 .
note that the f yields from agb stars were increased with updated reaction rates .
evolution of the [ f / o ] ratio against [ o / h ] for the solar neighbourhood with sne ii , hne , and sne ia only ( short - dashed lines ) , with agb stars ( long - dashed lines ) , with the @xmath50-process of sne ii and hne ( solid line and dot - dashed line for @xmath56 and @xmath57 erg , respectively ) .
the dotted line is for the model for gcs .
see @xcite for the observational data sources .
, width=226 ] the timescale of supernovae is much shorter than agb stars , which means that the [ f / o ] ratio at low metallicities can be strongly enhanced by the @xmath50-process occurring in core - collapse supernovae . with the standard neutrino luminosity @xmath56 erg ( solid line )
, the [ f / o ] ratio shows a plateau of [ f / o ] @xmath58 at [ o / h ] @xmath59 , and reaches [ f / o ] @xmath60 at [ o / h ] @xmath61 .
this is consistent with the observational data of field stars at @xmath62 [ o / h ] @xmath63 .
if we adopt a larger @xmath64 erg ( dot - dashed line ) , [ f / o ] can be as large as @xmath65 at [ o / h ] @xmath39 . in the bulge
the star formation timescale is shorter and the average metallicity is higher than the solar neighborhood , but the [ f / o ] ratio is not so different at [ o / h ] @xmath39 .
the observations for the bulge stars ( filled circles ) might suggest that the initial mass function is also different , although the number of observations is too small to make a conclusion . at @xmath66
[ f / o ] @xmath67 the observational data are for stars in gcs , which seem to be more consistent with the models with the agb yields only than with the @xmath50-process . however , it is unlikely that the existence of the @xmath50-process depends on the environment or metallicity .
the neutrino luminosity may be small in the case of faint supernovae with a large black hole , which give high [ @xmath3/fe ] , but there is no significant difference seen in the [ @xmath3/fe ] ratio between field halo stars and gc stars .
one possible scenario is as follows : in gcs , the contribution from low - mass supernovae is smaller than in the filed . since the star formation occurs in a baryon dominated cloud with very high density , the initial star burst can be very intense .
after the initial star burst , because of the small gravitational potential , galactic winds are generated immediately after the explosion of massive supernovae , which may remove the contribution from low - mass supernovae .
the small production of @xmath3 elements from low mass supernovae means that the [ f / o ] ratio can reach values as large as @xmath39 .
in contrast , massive supernovae produce more @xmath3 elements which results in [ f / o ] ratios of @xmath68 , consistent with the observational data .
the dotted line shows an example of such a gc model ( see @xcite for the details ) . at low metallicities
( [ o / h ] @xmath59 ) f is produced only from supernovae , and thus the observations of field stars at low - metallicities are important for constraining the neutrino luminosity released from a core - collapse supernova . if the neutrino luminosity is specified , the f abundance along with c could be a good clock in the study of galactic archaeology to distinguish the contribution from agb stars and supernovae .
the f observations of stars in gcs suggest that the star formation and chemical enrichment histories of gcs are different from those of field stars and that low - mass supernova played a smaller role in shaping the chemical evolution of these systems .
this is the opposite to the situation for dsphs .
dsphs show low [ @xmath3/fe ] and low [ mn / fe ] @xcite , of which elemental abundance pattern is more consistent with the enrichment from low - mass sne ii than that from the sn ia enrichment ( @xcite , see also figure 20 of @xcite ) .
there is no strong metallicity dependence on mn yields of sne ia ( t. ohlubo , k. nomoto et al , in prep . ) . in dsphs , the dark matter component is large , the gas density is low , the star formation rate is low , and thus the contribution from massive supernovae is expected to be smaller than in the milky way halo .
the elemental abundance pattern ( low [ ( cu , zn)/fe ] ) of anomaly stars in the galactic halo @xcite is also consistent with low - mass sne ii .
|
elemental and isotopic abundances are the fossils of galactic archaeology . the observed [ x / fe]-[fe / h ] relations in the galactic bulge and disk and the mass - metallicity relation of galaxies are roughly reproduced with chemodynamical simulations of galaxies under the standard @xmath0-cdm picture and standard stellar physics .
the isotopic ratios such as @xmath1o and @xmath2 mg may require a refinement of modelling of supernova and asymptotic giant branch stars .
the recent observation of the carbon - rich damped lyman @xmath3 system can be reproduced only with faint core - collapse supernovae .
this suggests that chemical enrichment by the first stars in the first galaxies is driven not by pair - instability supernovae but by core - collapse supernovae ( @xmath4 ) .
the observed f abundances can be reproduced with the neutrino processes of core - collapse supernovae . as in f ,
the observations of elemental abundances in small systems may requires further complications of chemical enrichment . in globular clusters
the relative contribution from low - mass supernovae is likely to be smaller than in the field , while the contribution from massive supernovae seems smaller in dwarf spheroidal galaxies than in the solar neighborhood .
|
bhabha scattering is an important process for extracting physics from experiments at electron - positron colliders primarily because it provides an effective means for determining luminosity .
these measurements depend on having precise theoretical predictions for the bhabha scattering cross sections . as yet , the complete next - to - next - to - leading order ( nnlo ) qed corrections needed for reducing theoretical uncertainties have not been computed . in this paper
we present the complete two - loop matrix elements that would enter into such a computation .
this calculation also provides a means for validating techniques that can be applied to physically important but more intricate qcd calculations .
it also provides an additional explicit verification of a general formula due to catani @xcite for the structure of two - loop infrared divergences , and allows us to determine the process - dependent terms for the processes at hand .
in bhabha scattering there are two distinct kinematic regions : small angle bhabha scattering ( sabs ) , and large angle ( labs ) . in the lep / slc energy range ,
sabs is used to measure the machine luminosity via a dedicated small angle luminosity detector .
sabs has a large cross section about four times larger than @xmath6 decay in the @xmath7 window making it particularly effective as a luminosity monitor . at the same time
, sabs is calculable theoretically with high accuracy from known physics ( mainly qed ) , apart from hadronic vacuum polarization corrections that rely upon the experimental data for @xmath8 annihilation into hadrons at low energy @xcite .
therefore , sabs is an important ingredient in measuring any absolute cross section .
for instance , the measurement of the hadronic cross section at the @xmath6 peak , @xmath9 , which enters several precision observables , is especially dependent on an accurate theoretical understanding of bhabha scattering . at lep / slc , large angle bhabha scattering interferes with @xmath10 and
so it is needed to disentangle important parameters such as the electroweak mixing angle .
it is also useful for measuring the luminosity at flavor factories such as babar , belle , da@xmath11ne , vepp-2 m , and bepc / bes @xcite .
a peculiarity of future electron linear colliders is that the luminosity spectrum is not monochromatic because of the beam - beam effect . because of this , measuring the total small angle cross section of bhabha scattering alone is not sufficient , and therefore the angular distribution of labs was proposed for disentangling the luminosity spectrum @xcite . due to the experimental importance of this process
, significant effort has been devoted to developing monte carlo event generators see for instance ref .
@xcite for an overview . in order to match the impressive experimental precision , a complete inclusion of nnlo qed quantum effects
has become necessary . on the theoretical side , however , the calculation of two - loop four - point amplitudes has been a roadblock to further progress . in this article
we present the two - loop virtual qed corrections to the differential cross section for bhabha scattering , i.e. , the two - loop amplitude interfered with the tree amplitude and summed over all spins .
we neglect the small electron mass in comparison to all other kinematic invariants , and use dimensional regularization to handle the ensuing infrared divergences . besides these contributions ,
a number of other virtual and real emission contributions ( discussed in the conclusions ) still need to be obtained before a full monte carlo program for the bhabha scattering cross section can be constructed .
the two - loop qed four - fermion amplitudes are also a useful testing ground for two - loop qcd calculations containing more than one kinematic invariant , which are required for higher - order jet cross sections and other aspects of collider physics . for processes that depend on a single momentum invariant , a number of important quantities have been calculated up to four loops , such as the total cross section for @xmath8 annihilation into hadrons and the qcd @xmath12-function @xcite .
in contrast , the only complete two - loop four - point scattering amplitudes presently known for generic kinematics in massless gauge theory are the @xmath13 super - yang - mills amplitudes @xcite , and @xmath14 in a single helicity configuration in pure gauge theory @xcite .
the two - loop amplitudes required for nnlo computations of jet production in hadron colliders , or for nnlo three - jet rates and other event shape variables at @xmath8 colliders , remain uncalculated .
we note in passing that partial results for the leading - color part of two - loop contributions to quark - quark scattering have very recently appeared @xcite .
two important technical breakthroughs are the calculations of the dimensionally regularized scalar double box integrals with planar @xcite and non - planar @xcite topologies and all external legs massless , and the development of reduction algorithms for the same types of integrals with loop momenta in the numerator ( tensor integrals ) @xcite . related integrals , which also arise in the reduction procedure , have been computed in refs .
@xcite . taken together
, these results are sufficient to compute all loop integrals required for @xmath15 massless scattering amplitudes at two loops , thus removing a major obstacle to several types of nnlo calculations .
in this paper we use these techniques to evaluate the integrals encountered in the bhabha calculation .
an even more recent result concerning two - loop planar double box integrals with one massive external leg @xcite holds promise for the nnlo computation of three - jet rates at @xmath8 colliders
. there has also been significant progress in developing general formalisms for other aspects of nnlo computations involving massless particles .
the motivation has typically been infrared - safe observables in qcd , but many of the developments can be applied to the bhabha process as well .
the developments include an understanding of the intricate structure of the infrared singularities that arise when more than one particle is unresolved ( i.e. , is soft or collinear with another particle ) @xcite . improved approximations to the nnlo correction to splitting functions
have been constructed recently as well @xcite .
infrared divergences are a significant complication in all the qcd and qed computations mentioned above . in any suitably `` infrared - safe ''
observable all final - state divergences will cancel @xcite .
however , divergences occur in individual amplitudes for fixed particle number , and it is very useful to have a general description of such divergences .
catani has presented a general formula for the infrared divergence appearing in any two - loop qcd amplitude @xcite . by appropriately adjusting group theory factors ,
it is straightforward to convert catani s qcd formula to a qed formula , allowing us to directly verify it .
moreover , we extract the exact form of a process - dependent term in the formula , for the case of qed scattering of four charged fermions . previously , the only process for which this term had been extracted @xcite was the quark form factor which enters drell - yan production @xcite .
( it should also now be possible to extract it for @xmath16 higgs using the recent two - loop computation @xcite . )
interestingly , a simple generalization of the quark form factor term ( converted to qed ) correctly predicts the process - dependent term for the @xmath17 and bhabha amplitudes .
we also use catani s formula to conveniently organize the infrared divergences and to absorb some of the finite terms .
the previously computed non - abelian gauge theory amplitudes @xcite were obtained via cutting methods .
the low multiplicity and relative simplicity of the @xmath17 and bhabha scattering feynman diagrams makes it relatively easy to directly compute the diagrams , as we do here .
we include here only the pure qed diagrams , neglecting for example the contributions of @xmath6 exchange , and hadronic vacuum polarization effects .
the former are negligible at this order in sabs and in labs at flavor factories .
the hadronic contributions are important , but much of their effect is straightforward to include by introducing a running coupling .
we perform the calculation in dimensional regularization @xcite with @xmath18 and set the small electron mass to zero , since it is the only form in which the required two - loop momentum integrals are known .
moreover , it provides a powerful method for simultaneously dealing with both the infrared and ultraviolet divergences encountered in gauge theories .
traditionally , dimensional regularization is not used for qed , in part because the infrared divergences are relatively tame compared to non - abelian gauge theories , so photon and electron masses are sufficient for cutting off the theory .
another important reason for using dimensional regularization is to validate techniques that can also be applied to the more complicated case of qcd . in qcd ,
dimensional regularization is the universally utilized method for dealing with divergences . in the high - energy bhabha process ,
even with an `` infrared - safe '' ( calorimetric ) final - state definition , the electron mass will still appear in large logarithms of the form @xmath19 due to initial - state radiation .
however , in the dimensionally regulated amplitudes these singularities ( like all others ) appear as poles in @xmath20 .
it may therefore be most convenient to handle the initial - state singularities using an electron structure function method @xcite implemented in the @xmath21 collinear factorization scheme .
in the next section we briefly describe our method for computing the two - loop amplitudes
. then we describe catani s formula for the divergence structure of the amplitudes , followed by a presentation of the finite ( @xmath22 ) terms for both @xmath23 and bhabha scattering . in the final section
we give our conclusions , including some discussion of the remaining ingredients still required for construction of a numerical program for bhabha scattering at this order .
the 16 independent feynman diagram topologies describing the two - loop qed corrections to @xmath17 and bhabha scattering are enumerated in . in this figure
we have suppressed the fermion arrows . after including the fermion arrows and distinct labels for the external legs ,
there are a total of 47 feynman diagrams ; however , many of these diagrams generate identical results . of the 47 diagrams ,
35 contain no fermion loop , 11 contain one fermion loop , and 1 contains two fermion loops
. the bhabha amplitude may be obtained from the @xmath17 amplitude by adding to it the same set of diagrams , but with an exchange of one pair of external legs .
the @xmath24 and @xmath25 amplitudes may , of course , be obtained by crossing .
4.2 truein -.2 cm we have evaluated these diagrams interfered with the tree amplitudes and summed over spins in the conventional dimensional regularization ( cdr ) scheme .
this interference gives directly the two - loop virtual correction to the @xmath26 differential cross section .
the rules for implementing cdr are straightforward because all particle are treated uniformly in all parts of the calculation . in this scheme ,
all momenta and all lorentz indices are taken to be @xmath27 dimensional vectors .
( the @xmath28-matrices remain as @xmath29 matrices ; i.e. , @xmath30 = 4 $ ] . ) after performing all @xmath28-matrix algebra present in the two - loop feynman diagrams , we use the conservation of momenta flowing on the internal lines to express the tensor structure of the diagrams in terms of inverse scalar propagators and a small number of additional scalar invariants containing loop momenta .
the inverse scalar propagators cancel propagators in the denominator to generate simpler `` boundary '' integrals . to handle the integrals containing scalar invariants
, we introduce feynman parameters and interpret the resulting integrals in terms of scalar integrals with multiple propagators , which are then reduced to a set of master integrals with the help of equations in refs .
@xcite . proceeding in this way
, we obtain an expression for the amplitude in terms of master integrals ( of the type listed in ref .
@xcite , plus a few more for the planar double box topology ) multiplied by coefficient functions .
this expression is in principle valid to an arbitrary order in @xmath20 , assuming that the master integrals could be evaluated to such an order .
however , it is a bit too lengthy to present here , and for nnlo computations only the series expansion in @xmath20 through @xmath22 is required . to carry out this expansion
, we use expansions of the master integrals presented in refs . @xcite . as noted in ref .
@xcite , there is a slight problem with the original choice of basis @xcite for the two master planar double box integrals .
in that basis , the coefficients for generic tensor integrals contain @xmath31 poles , necessitating an @xmath32 evaluation of the master integrals .
several solutions to this problem have been presented @xcite .
we have used a slightly different solution , which is simply to use the original pair of master integrals defined in ref .
@xcite , except evaluated in @xmath33 instead of @xmath18 . in @xmath33
the integrals have neither ultraviolet nor infrared divergences , making them simpler to evaluate through @xmath22 than the @xmath18 integrals .
many of the master integral expansions quoted in refs .
@xcite are in terms of nielsen functions @xcite , @xmath34 with @xmath35 .
we have found it useful to express the results instead in terms of a minimal set of polylogarithms @xcite , @xmath36 with @xmath37 , using relations such as @xmath38 here @xmath39 the analytic properties of the non - planar double box integrals are somewhat intricate @xcite , since they are not real in any of the three kinematic channels for the @xmath26 process , @xmath40 where @xmath41 , @xmath42 , and @xmath43 .
therefore we shall present explicit formulae for the finite terms in the amplitude in both the @xmath44- and @xmath45-channels ; those in the @xmath46-channel will be related by symmetries .
dimensionally regulated two - loop amplitudes for four massless fermions contain poles in @xmath47 up to @xmath48 .
the structure of most of these singularities has already been exposed by catani @xcite , who described the infrared behavior of general two - loop qcd processes .
we shall therefore adopt his notation in presenting our results .
we work with ultraviolet renormalized amplitudes , and employ the @xmath21 running coupling for qed , @xmath49 .
of course this scheme can always be converted to another one , for example @xmath50 defined via the photon propagator at momentum transfer @xmath51 , by a finite renormalization .
the relation between the bare coupling @xmath52 and @xmath49 through two - loop order can be expressed as @xcite @xmath53 \ , , \equn\label{twoloopcoupling}\ ] ] where @xmath54 $ ] and @xmath55 is euler s constant .
the first two coefficients of the qed beta function are @xmath56 where @xmath57 is the number of light ( massless ) charge 1 fermions .
the renormalized four - fermion amplitude is expanded as @xmath58 \ , .
\cr } \equn\label{renexpand}\ ] ] the infrared divergences of a renormalized two - loop amplitude in qcd or qed are @xcite , @xmath59 where @xmath60 is a color space vector representing the renormalized @xmath61 loop amplitude .
the subscript @xmath62 stands for the choice of renormalization scheme , and @xmath63 is the renormalization scale .
these color space vectors give the amplitudes via , @xmath64 where the @xmath65 are color indices .
the divergences of @xmath66 are encoded in the color operators @xmath67 and @xmath68 . in the qed case ,
the color space language is clearly unnecessary ; @xmath69 and @xmath70 are just numbers . in qcd , the operator @xmath67 is given by @xmath71 \biggl ( \frac{\mu^2 e^{-i\lambda_{ij } \pi}}{2 p_i\cdot p_j } \biggr)^{\e } \ , , \equn\label{catanigeneral}\ ] ] where @xmath72 if @xmath73 and @xmath74 are both incoming or outgoing partons and @xmath75 otherwise .
the color charge @xmath76 is a vector with respect to the generator label @xmath77 , and an @xmath78 matrix with respect to the color indices of the outgoing parton @xmath73 .
the values required for qcd are @xmath79 for qed we let @xmath80 , @xmath81 , @xmath82 and @xmath83 , where the @xmath84 are the electric charges , to obtain @xmath85 \ , , \equn\label{catani1qed}\ ] ] for the four - fermion amplitude @xmath86 ( note that the charges of incoming states should be reversed in computing @xmath87 . )
the operator @xmath88 is given in qcd by @xcite @xmath89 where the coefficient @xmath90 is : @xmath91 for the qed process ( [ eemumukin ] ) , we insert @xmath92 from , take @xmath93 and @xmath94 from , and let @xmath95 .
the function @xmath96 is process - dependent but has only _ single _ poles : @xmath97 ref .
@xcite does not give an expression for @xmath96 for a general amplitude , but only for the case of a @xmath98 pair , i.e. a single charged fermion pair .
the result , which is extracted from the two - loop qcd computation of the electromagnetic form factor of the quark @xcite , is @xmath99 \ , \cr } \equn\label{catanihqq}\ ] ] where @xmath100 performing the usual conversion to qed yields a result applicable to the electromagnetic form factor of the electron , @xmath101 \ , .
\equn\label{catanihee}\ ] ] using our two - loop computation , and an all - orders - in-@xmath20 computation of the one - loop amplitude for @xmath23 ( see sect . [ eemumuoneloopsubsection ] ) , we have verified that the singular behavior of the @xmath17 amplitude in cdr agrees precisely with that predicted by in all three kinematic channels .
in addition , we have extracted the function @xmath102 controlling the @xmath31 poles in .
we obtain @xmath103 \cr & \hskip2.7 cm \times \biggl [ -{3\over4 } + 6 \zeta_2 - 12 \zeta_3 + \biggl ( - { 25\over27 } + \zeta_2 \biggr ) n_{\ ! f } \biggr ] \ , .
\cr } \equn\label{usheemumu}\ ] ] this result agrees with a `` naive '' generalization from the form factor case , in which one sums over the six pairs of charged legs in the four - point amplitude , weighted by the sign of the charge product @xmath104 .
( note that the factors of @xmath105 are purely conventional here , since their deviation from unity only contributes at the level of finite parts , @xmath22 .
however , the overall normalization is predicted correctly by the sum over the six pairs . ) in order to verify the structure of the infrared singularities , and to extract the finite remainder of the two - loop amplitude presented below , we computed the one - loop @xmath17 amplitude ( interfered with the tree amplitude ) to all orders in @xmath20 .
the result is @xmath106 \biggr ] \bigg\vert_{\xi=1 } \
, , \equn\label{oneloopallorders1}\ ] ] where the first term is the @xmath21 counterterm , expressed in terms of the tree - level interference @xmath107 \ , \equn\label{cdrtree}\ ] ] and @xmath108 \boxsixint(s , t ) \cr & \hskip1 cm - 4 { \xi \over 1 - 2\e } { t \over s^2 } \bigl [ ( 4 - 12\e+7\e^2 ) t^2 - 6\e ( 1 - 2\e ) t u + ( 4 - 10\e+5\e^2 ) u^2 \bigr ] \triint(t ) \cr & \hskip1 cm - { 8 \over ( 1 - 2\e)(3 - 2\e ) } { 1\over s } \biggl [ 2\e ( 1-\e ) \ , t
( ( 1-\e ) t - \e u ) \ , n_{\ !
f } - \e ( 3 - 2\e ) ( 2-\e+2\e^2 ) t u \cr & \hskip5 cm + ( 1-\e ) ( 3 - 2\e ) ( 2 - ( 1-\xi ) \e + 2 \e^2 ) t^2 \biggr ] \triint(s)\ , .
\cr } \equn\label{oneloopallorders2}\ ] ] the symmetry operation @xmath109 acts as @xmath110 after carrying out the operation of @xmath109 , one should then set @xmath111 .
( basically , @xmath112 allows us to separate diagrams based on whether they have an even or odd number of photons attached to the muon line . because photons have @xmath113 , this criterion governs the @xmath114 symmetry properties . ) in , @xmath115 and @xmath116 are one - loop box and triangle integrals , the former evaluated in an expansion around @xmath33 . for the divergence formula ( [ twoloopcatani ] ) ,
we need their series expansions in @xmath20 through @xmath117 . in the @xmath45-channel , where the functions are manifestly real ,
their expansions are given by @xmath118\ + \
\ord(\e^3 ) , \cr \triint(s ) & = - { ( -s)^{-1-\e } \over \e^2 } \biggl [ 1 - { \pi^2\over12 } \ , \e^2 - { 7\over3 } \zeta_3 \ , \e^3 - { 47\over16 } \zeta_4 \ , \e^4 \biggr]\ + \ord(\e^3)\ , , \cr } \equn\label{oneloopintegralexp}\ ] ] where @xmath119 the expansions in the @xmath44- and @xmath46-channels can be found using analytic continuation formulae such as @xcite @xmath120 where @xmath121 is an imaginary infinitesimal added to @xmath122 or @xmath45 before continuing .
we have verified that through @xmath123 our result for the one - loop amplitude agrees with a previous calculation @xcite , up to terms which can be identified as being due to the conversion between dimensional regularization and a photon mass regularization . in comparison with the process
@xmath17 described above , the bhabha scattering process @xmath124 has additional exchange diagrams . in general , the interference required for bhabha scattering
is given by @xmath125 \ , \cr } \equn\label{bhabhapermsum}\ ] ] where the symmetry @xmath126 acts as @xmath127 @xmath128 is the @xmath61-loop amplitude for @xmath17 , and @xmath129 is the same @xmath61-loop amplitude but with legs 1 and 3 interchanged ( taking into account the fermi statistics minus sign ) . in the cdr scheme , the tree - level exchange contribution required for bhabha scattering in is @xmath130 \ , .
\equn\label{cdrcrosstree}\ ] ] the one - loop exchange contribution , evaluated to all orders in @xmath20 , is given by @xmath131 where @xmath132 \triint(s ) \cr & \hskip0.3 cm + { 8 \over 1 - 2 \e } { 1\over s } \bigl ( \e ( 2 - 5\e+2\e^2-\e^3 ) t^2 + \e ( 1 - 3\e+\e^2-\e^3 ) t u - ( 1-\e ) ( 2 - 3\e-\e^2 ) u^2 \bigr ) \triint(t ) \cr & \hskip0.3 cm - { 8 \over 1 - 2 \e } { u\over s t } \bigl ( \e ( 2 - 4\e+\e^2-\e^3 ) t^2 + \e ( 2 - 3 \e-\e^3 ) t u - ( 1-\e ) ( 2 - 4\e-\e^2 ) u^2 \bigr ) \triint(u ) \ , . \cr } \equn\label{oneloopexchangeallorders2}\ ] ] using these results , and the computation of the two - loop exchange terms , we again find that the additional singular terms in bhabha scattering are described by , where ( not surprisingly ) @xmath96 is given by precisely the same expression ( [ usheemumu ] ) that we found for @xmath17 .
finally we give the real ( dispersive ) part of the finite remainder in , interfered with the tree amplitude in the cdr scheme .
first we treat the @xmath23 process ( [ eemumukin ] ) .
it is convenient to decompose the finite part according to the number of light flavors , @xmath57 , @xmath133\ = \ 8 \bigl [ f^{(0 ) } + n_{\ !
f } \ , f^{(1 ) } + n_{\ !
f^{(2 ) } \bigr]\ , . \equn\label{twoloopremainder}\ ] ] in the @xmath44-channel , the functions @xmath134 are given by @xmath135 \biggr ] \bigg\vert_{\xi=1 } \ , \equn\label{ffromfs}\ ] ] where @xmath136 \cr & \hskip0.5 cm + ( x - y ) \biggl [ 8
\li4(-x / y ) + 6 ( 2+\xi ) \li4(-x ) + ( 4 ( 1-\xi ) x - 12 y ) \li3(-x ) \cr & \hskip2.0 cm - \biggl ( ( 2-\xi
) x^2 - 4 x y + { 4\over3 } \pi^2 \biggr ) \li2(-x ) - ( 2+\xi ) \bigl ( \li3(-x ) - x \li2(-x ) \bigr ) \cr & \hskip2.0 cm + \biggl ( - { 1\over12 } ( 6+\xi ) x^3 + { 2\over3 } x^2 y + { 1\over6 } ( 1 + 4 \xi ) \pi^2 x + { 1\over6 } ( 10+\xi ) x^2 - { 1\over2 } ( 2-\xi ) x y \cr & \hskip2.5 cm + { 1\over2 } ( 1 + 6\xi ) x - 4 ( 2-\xi ) \zeta_3 - ( 1 + 4\xi ) { \pi^2\over6 } - 6 \xi \biggr ) x + \xi \biggl ( - 6 \zeta_4 - 2 \zeta_3 + 2 { \pi^2\over3 } \biggr ) \biggr ] \cr & \hskip0.5 cm + ( 2-\xi ) \bigl ( \li3(-x ) - x \li2(-x ) \bigr ) + \biggl ( { 1\over6 } ( 5 - 3 \xi ) x^2 - { 1\over2 } ( 3-\xi ) x y - ( 1 - 4\xi ) { \pi^2\over6 } \biggr ) x \cr & \hskip0.5 cm + \biggl ( - { 1\over2 } ( 1 + 6 \xi ) x - { 1\over2 } y - 6\xi \biggr ) x - 4 \zeta_3 + { \pi^2\over3 } \ , , \cr } \equn\label{fs0}\ ] ] @xmath137 + { 87\over2 } \ln(\mu^2/s ) + { 35\over2 } \zeta_3 + { 7\over4 } \pi^2 + { 685\over9 } \biggr ) \cr & \hskip1 cm - 2 \xi ( x - y ) \biggl ( \bigl ( x^2 - 3 ( x - 1 ) \ln(\mu^2/s ) - { 13\over2 } x + 4 \pi^2 + 8 \bigr ) x - 2 \pi^2 \biggr ) \cr & \hskip1 cm - \xi ( 3 x + 6 \ln(\mu^2/s ) + 16 ) x \biggr\ } \ , , \cr } \equn\label{fs1}\ ] ] @xmath138\ , , \cr } \equn\label{fs2}\ ] ] with @xmath139 and the symmetry operation @xmath109 is given in . in the @xmath45-channel , the @xmath134
are given by @xmath140 where @xmath141 + 2 ( 6+\xi ) { \pi^2\over3 } \li2(-v ) \cr & \hskip2.5 cm - { 4\over3 } \xi v^3 w + ( 4-\xi ) v^2 w^2 - { 2\over3 } ( 7 - 2\xi ) v w^3 - { 1\over6 } ( 1 - 2\xi )
w^4 \cr & \hskip2.5 cm + 6 v w^2 - 3 w^3 + 16 \xi v w + { 1\over2 } ( 9 - 16 \xi ) w^2 + { 93\over4 } \xi w \cr & \hskip2.5 cm + \pi^2 \biggl ( - 2 v^2 + { 2\over3 } ( 3+\xi ) v w - { 1\over3 } ( 3-\xi ) w^2 - 6 v - ( 3-\xi ) w \biggr ) \cr & \hskip2.5 cm + 4 \bigl ( ( 2-\xi ) v - 2 ( 1+\xi ) w \bigr ) \zeta_3 + 34 \zeta_4 - 15 \zeta_3 - ( 25 + 96 \xi ) { \pi^2\over12 } + { 511\over16 } \biggr ] \cr & \hskip0.5 cm - ( x - y ) \biggl [ 6 \bigl ( ( 2+\xi ) \li4(-v / w ) - ( 2-\xi ) \li4(-v ) \bigr ) - 16 \li4(-w ) \cr & \hskip2.2 cm + 4 \bigl ( ( 1-\xi ) w + ( 2+\xi ) v \bigr )
\li3(-w ) + 4 \bigl ( 4 v - ( 2+\xi ) w \bigr ) \li3(-v ) \cr & \hskip2.2 cm - ( 2+\xi ) \bigl ( \li3(-w ) + w \li2(-v ) \bigr ) - 4 \bigl ( \li3(-v ) - v \li2(-v ) \bigr ) \cr & \hskip2.2 cm + \biggl ( - 4 v^2 + 2 ( 2+\xi ) v w + ( 2-\xi ) w^2 - ( 10 + 3 \xi ) { \pi^2\over3 } \biggr ) \li2(-v ) \cr & \hskip2.2 cm + { 1\over3 } \xi v^4 - { 2\over3 } ( 2+\xi ) v^3 w + { 5\over2 } ( 2+\xi ) v^2 w^2 - { 1\over3 } ( 4 + 7\xi ) v w^3 + { 1\over3 } ( 2+\xi )
w^4 \cr & \hskip2.2 cm + { 2\over3 } \xi v^3 - ( 2+\xi ) v^2 w + 2 v w^2 - { 1\over6 } ( 10+\xi ) w^3 + ( 5 + 6 \xi ) v w - { 1\over2 } ( 5 + 6 \xi ) w^2 \cr & \hskip2.2 cm + { \pi^2\over6 } \bigl ( 2 ( 6-\xi ) v^2 - 2 ( 13 + 4\xi ) v w + ( 5 + 3\xi ) w^2 + 2 ( 12-\xi ) v - ( 3 - 2\xi ) w \bigr ) \cr & \hskip2.2 cm - 4 \bigl ( ( 2-\xi ) v - w \bigr ) \zeta_3 - 6 \xi ( 2 v - w ) + ( 121 + 3\xi ) \zeta_4 + ( 2 + 5\xi ) \zeta_3 - ( 15 + 26\xi ) { \pi^2\over6 } \biggr ] \cr & \hskip0.5 cm + 2 \xi \bigl ( \li3(-v ) - v \li2(-v ) - v^2 w \bigr ) - ( 2-\xi ) \bigl ( \li3(-w ) - w \li2(-w ) \bigr ) \cr & \hskip0.5 cm - ( 1-\xi ) v w^2 + { 1\over6 } ( 5 - 3 \xi ) w^3 + 6 \xi v w + { 1\over2 } ( 1 - 6 \xi ) w^2 - 6 \xi w \cr & \hskip0.5 cm + { \pi^2\over6 } \bigl ( 2 ( 4 + 3 \xi ) v + ( 8 - 5 \xi ) w \bigr ) - ( 6+\xi ) \zeta_3 + ( 1 - 18 \xi ) { \pi^2\over6 } \ , , \cr } \equn\label{fu0}\ ] ] @xmath142 \cr & \hskip2.7 cm + 87 \ln(\mu^2/(-s ) ) + ( 35 + 12\xi ) \zeta_3 + ( 7 - 58\xi ) { \pi^2\over2 } + { 1370\over9 } \biggr ] \cr & \hskip0.5 cm + \xi ( x - y ) \biggl [ 2 v^3 + 2 ( v - w)^3 + 13 ( 2 ( v - w ) v + w^2 ) \cr & \hskip2.5 cm + 6 \bigl ( 2 ( v - w ) v + w^2 + 2 v - w + \pi^2 \bigr ) \ln(\mu^2/(-s ) ) + 16 ( 2 v - w ) \cr & \hskip2.5 cm - 2 \pi^2 ( v+w ) + 9 \pi^2 \biggr ] \cr & \hskip0.5 cm - \xi \biggl [ \bigl ( 3 w - 6 v + 6 \ln(\mu^2/(-s ) ) + 16 \bigr ) w + 3 \pi^2 \biggr ] \biggr\ } \ , , \cr } \equn\label{fu1}\ ] ] @xmath143 \ , .
\equn\label{fu2}\ ] ] here @xmath144 , @xmath145 are defined in , whereas @xmath146 , @xmath147 , @xmath148 , @xmath149 are defined in . in the @xmath46-channel ,
the functions @xmath134 are given by the action of the symmetry @xmath109 of on the @xmath45-channel results , @xmath150 \big\vert_{\xi=1 } \ , .
\equn\label{ftfromfu}\ ] ] the two - loop virtual contribution to the @xmath17 unpolarized cross section , restoring overall factors and averaging over initial spins , is given by @xmath151 \ , .
\equn\label{fullcrosssection}\ ] ] for the finite two - loop remainder for the bhabha scattering process ( [ bhabhakin ] ) , we quote only the @xmath152 symmetric sum of the two exchange terms required by .
again we decompose the answer according to @xmath57 , @xmath153 + u \biggl [ \re \bigl [ \cm_4^{(2){\rm fin } } \cmt_4^{(0)\,\dagger } \bigr ] \biggr]\biggr\ } \ = \ 8 \bigl [ \ft^{(0 ) } + n_{\ !
f } \ , \ft^{(1 ) } + n_{\ ! f}^2 \ , \ft^{(2 ) } \bigr]\ , .
\equn\label{bhabhatwoloopremainder}\ ] ] in the @xmath44-channel , the functions @xmath154 are given by @xmath155 where @xmath156 \cr & \hskip0.5 cm + { y(1-x ) \over x } \biggl [ - 10 \li4(-x ) + 6 x \li3(-x ) - \biggl ( x^2 + 2 { \pi^2\over3 } \biggr ) \li2(-x ) + { 1\over24 } x^4 - { 13\over12 } x^3 \cr & \hskip2.5 cm + { \pi^2\over3 } x^2 - { 5\over2 } ( ( x - y)^2 + \pi^2 ) + { 1\over2 } y^2 + \biggl ( - 6 \zeta_3 + { 5\over2 } \pi^2 + 12 \biggr ) x + 20 \zeta_4 \biggr ] \cr & \hskip0.5 cm + 16 \bigl ( \li4(-x / y ) - \li4(-y ) +
x \li3(-y ) \bigr ) + 4 ( 3 x - 2 y - 2 ) \li3(-x ) \cr & \hskip0.5 cm - 4 ( x^2 - 2 x y - 2 x + \pi^2 ) \li2(-x ) - { 5\over12 } x^4 - { 4\over3 } x^3 y + 8 x^2 y^2 - { 8\over3 } x y^3 + { 2\over3 } y^4 \cr & \hskip0.5 cm + { 5\over6 }
x^3 + x^2 y + 18 x^2 - 4 x y + 2 y^2 + { \pi^2\over3 } ( 11 x^2 - 20 x y + 4 y^2 + 9 x ) \cr & \hskip0.5 cm - 4 \zeta_3 ( 3 x - 2 y ) + 88 \zeta_4 + 8 \zeta_3 + 2 \pi^2 \ , \cr } \equn\label{ftildes0}\ ] ] @xmath157 \cr & \hskip0.5 cm - 2 x \biggl [ \biggl ( 3 \bigl ( \ln(\mu^2/s ) + \ln(\mu^2/(-t ) ) \bigr ) + 16 \biggr ) x^2 - 6 \pi^2 x \biggr ] \cr & \hskip0.5 cm - 4 y \biggl [ x^3 + \bigl ( 3 \ln(\mu^2/(-t ) ) + 8 \bigr ) x^2 + \biggl ( 3 \bigl ( \ln(\mu^2/s ) + \ln(\mu^2/(-t ) ) \bigr ) - 2 \pi^2 + 16 \biggr ) x - 3 \pi^2 \biggr ] \biggr\ } \ , , \cr } \equn\label{ftildes1}\ ] ] @xmath158 \ , , \equn\label{ftildes2}\ ] ] and @xmath144 , @xmath145 , @xmath159 , @xmath160 are defined in . in the @xmath46-channel ,
the functions @xmath154 are given by the action of the symmetry @xmath126 of on the @xmath44-channel results , @xmath161 \ , .
\equn\label{ftildetfromftildes}\ ] ] in the @xmath45-channel , the functions @xmath154 are given by @xmath162 \ , \equn\label{ftildefromftildeu}\ ] ] where @xmath163 \cr & \hskip0.5 cm + { y(1-x ) \over x } \biggl [ 5 \li4(-v / w ) + 6 v \bigl ( \li3(-v ) + \li3(-w ) \bigr ) \cr & \hskip2.5 cm - { 1\over2 } \biggl ( ( 3 v+w-2 ) ( v - w ) + 5 { \pi^2\over3 } \biggr ) \li2(-v ) \cr & \hskip2.5 cm + \biggl ( - { 1\over4 } v^3 + { 1\over2 } v^2 w + { 13\over12 } v^2 - { 9\over4 } v w + { 5\over2 } v - 12 + { \pi^2\over12 } ( v + 7 ) \biggr ) v \biggr ] \cr & \hskip0.5 cm + 16 \bigl ( \li4(-v ) + w \li3(-v ) \bigr ) - 12 v \bigl ( \li3(-v ) + \li3(-w ) \bigr ) + 8 \li3(-v ) \cr & \hskip0.5 cm + \biggl ( ( 3 v+w-4 ) ( v - w ) + 5 { \pi^2\over3 } \biggr ) \li2(-v ) + \biggl ( - { 5\over12 } v^3 + 2 v^2 w - { 9\over4 } v w^2 + { 5\over6 } v^2 + { 1\over2 } v w \cr & \hskip1.5 cm + 18 v - 16 w + { \pi^2\over6 } ( - 4 v + 9 w - 7 ) \biggr ) v - { 253\over4 } \zeta_4 + 8 \pi^2 \ , , \cr } \equn\label{ftildeu0}\ ] ] @xmath164 \cr & \hskip0.5 cm - x \biggl [ 6 \bigl ( ( v - w)^2 + \pi^2 + 2 ( v - w ) \bigr ) \ln(\mu^2/(-s ) ) \cr & \hskip1.5 cm + ( 13 v - 31 w + 10 \pi^2 + 65 ) ( v - w ) + 16 \pi^2 \biggr ] \cr & \hskip0.5 cm - 2 y \biggl [ - 2 v^3 + 18 v^2 w + 12 v w \ln(\mu^2/(-s ) ) + 9 v^2 + 20 v w + 18 ( v+w ) \ln(\mu^2/(-s ) ) \cr & \hskip1.5 cm + 33 v + { \pi^2\over2 } \bigl ( 4 v - 12 \ln(\mu^2/(-s ) ) - 29 \bigr ) \biggr ] \biggr\}\ , , \cr } \equn\label{ftildeu1}\ ] ] @xmath165 \ , .
\equn\label{ftildeu2}\ ] ] here @xmath144 , @xmath145 are defined in , whereas @xmath146 , @xmath147 , @xmath148 , @xmath149 are defined in .
we performed several checks on our calculation .
the calculation was performed with the computer algebra programs maple , mathematica , and form . to check the code ,
large parts of the calculation were performed independently with alternative programs written in different languages .
various checks were applied to the integral reduction procedures described in refs .
@xcite and our implementation of them .
for example , we reproduced the double box ultraviolet divergences in @xmath166 and @xmath167 reported in ref .
@xcite , and several other previously calculated double box integrals @xcite .
an additional check on the non - planar tensor integrals is that unphysical @xmath168 poles occur in the representation of these integrals in terms of the master integral basis we used @xcite ; however , in the series expansion in @xmath20 such poles drop out after delicate cancellations between the various terms .
we checked the gauge invariance of the scattering amplitude by explicitly calculating the feynman diagrams in a general @xmath112 gauge and observing that the gauge dependence drops out in the final result .
this provides a non - trivial check of the diagrams and parts of the integral reduction procedure .
a strong check on the final result is provided by the matching of the ir divergence structure of the two - loop scattering amplitude with catani s formula ( [ twoloopcatani ] ) , as discussed in sect .
[ divergencesubsection ] .
a given integral will contribute to both infrared divergences and to finite terms .
thus a check of the divergent terms provides an indirect check that the finite terms have been correctly assembled . finally , we observed for small scattering angles a suppression of the leading logarithms , @xmath169 , e.g. in the limit @xmath170 in the @xmath46-channel for process ( [ eemumukin ] ) .
in other small - angle limits ( those not enhanced by the photon propagator pole ) the leading power - law behavior is of course less singular , but it is dressed by large logarithms of the type @xmath171 and @xmath172 .
but in the @xmath46-channel @xmath170 limit it cancels down to @xmath173 and @xmath174 .
this behavior is in accord with a generalized eikonal representation for small - angle scattering @xcite .
in this paper we presented the two - loop qed corrections to @xmath175 and to bhabha scattering .
we presented the results in terms of two - loop amplitudes interfered with tree amplitudes and summed over spins in the context of conventional dimensional regularization . in these results
we have set the small electron and muon masses to vanish .
( this is an excellent approximation for the highest energy current and future electron - positron colliders . )
the two - loop amplitudes presented in this paper are infrared divergent . to make use of them in a monte carlo program for the nnlo terms in the cross section
, they must be combined with lower - loop matrix elements including photon emission , which should be computed using conventional dimensional regularization , at least in the singular regions of phase space . in particular , the pieces that need to be computed ( for the bhabha case ) are the interference of the dimensionally regularized one - loop four - point amplitude with itself does not appear to be in the literature .
nevertheless , it should be relatively straightforward to obtain , given that it involves only one - loop amplitudes with four - point kinematics .
the required integrals are given to sufficiently high order in @xmath20 in .
the qed one - loop five - point amplitude interfered with the five - point tree is a rather involved object to compute from scratch .
however , the closely related one - loop helicity amplitudes for one photon and two quark pairs are known @xcite , and it is a relatively simple matter to modify the color factors to obtain the corresponding qed amplitudes .
the one - loop helicity amplitudes are in the t hooft - veltman scheme .
they can be converted to conventional dimensional regularization by altering the tree amplitude appearing in the coefficient of their singular terms @xcite .
thus the @xmath179 and @xmath177 one - loop amplitudes may be extracted from the known literature through @xmath180 .
because of the @xmath181 infrared divergences that are encountered in the phase - space integral , in regions where the photon is soft or collinear , one might seem to require the one - loop five - point amplitude through @xmath117 . however , this is not necessary @xcite .
instead , one can replace the five - point amplitudes in singular phase - space regions by a combination of four - point amplitudes ( which are given in this paper to the required order in the dimensional regularization parameter ) and splitting amplitudes @xcite .
the one - loop splitting amplitudes for qcd are enumerated to the required order in refs .
@xcite ; the case of qed follows as usual by an appropriate conversion of color factors .
the tree - level helicity amplitudes for @xmath182 and @xmath178 have been known for a while @xcite .
( they also can be converted from the four - quark two photon amplitudes in ref .
@xcite , for example . ) in infrared - divergent regions of phase space one must include higher order in @xmath183 contributions from the matrix elements .
systematic discussion of these regions , where two particles can be soft or three collinear , has been presented in refs .
@xcite for the case of qcd .
once again the results for qed can be obtained by a conversion of the color factors .
even with all of these matrix element ingredients assembled , it is a nontrivial task to devise a numerically stable method for carrying out the singular phase - space integrations .
nevertheless , this task is very analogous to that required to obtain qcd jet predictions at next - to - next - to - leading order , so it is likely that it will be attacked soon . besides the obvious application of the present paper to refined theoretical predictions for bhabha scattering and for electron - positron annihilation into muons , it also serves as a further test of methods that can be applied to analogous qcd processes .
we are confident that many more multi - particle two - loop amplitudes will be calculated before long .
s. catani , phys .
* b427 * , 161 ( 1998 ) [ hep - ph/9802439 ] .
arbuzov , v.s .
fadin , e.a .
kuraev , l.n .
lipatov , n.p .
merenkov and l. trentadue , nucl .
* b485 * , 457 ( 1997 ) [ hep - ph/9512344 ] . g. montagna , m. moretti , o. nicrosini , a. pallavicini and f. piccinini , nucl .
* b547 * , 39 ( 1999 ) [ hep - ph/9811436 ] ; + b. f. ward , s. jadach , m. melles and s. a. yost , phys . lett .
* b450 * , 262 ( 1999 ) [ hep - ph/9811245 ] .
carloni calame , c. lunardini , g. montagna , o. nicrosini and f. piccinini , nucl .
* b584 * , 459 ( 2000 ) [ hep - ph/0003268 ] .
frary and d.j .
miller , in _ proceedings , @xmath8 collisions at 500-gev , pt .
a ( munich / annecy / hamburg 1991 ) _ , report desy 92 - 123a , p , 379 ; + n. toomi , j. fujimoto , s. kawabata , y. kurihara and t. watanabe , phys .
* b429 * , 162 ( 1998 ) .
reports of the working group on precision calculations for the @xmath6 resonance , cern yellow report 95 - 03 ( 1995 ) , part 3 .
gorishnii , a.l .
kataev and s.a .
larin , phys .
lett . * b259 * , 144 ( 1991 ) ; + l.r .
surguladze and m.a .
samuel , phys .
* 66 * , 560 ( 1991 ) , err .
_ * 66 * , 2416 ( 1991 ) ; + t. van ritbergen , j.a . vermaseren and s.a .
larin , phys .
lett . * b400 * , 379 ( 1997 ) [ hep - ph/9701390 ] .
z. bern , j.s .
rozowsky and b. yan , phys .
* b401 * , 273 ( 1997 ) [ hep - ph/9702424 ] .
z. bern , l. dixon , d.c .
dunbar , m. perelstein and j.s .
rozowsky , nucl .
phys . * b530 * , 401 ( 1998 ) [ hep - th/9802162 ] .
z. bern , l. dixon and d.a .
kosower , jhep * 0001 * , 027 ( 2000 ) [ hep - ph/0001001 ] .
e.w.n . glover and m.e .
tejeda - yeomans , hep - ph/0010031 .
smirnov , phys .
* b460 * , 397 ( 1999 ) [ hep - ph/9905323 ] .
tausk , phys .
* b469 * , 225 ( 1999 ) [ hep - ph/9909506 ] .
v.a . smirnov and o.l .
veretin , nucl .
* b566 * , 469 ( 2000 ) [ hep - ph/9907385 ] . c. anastasiou , t. gehrmann , c. oleari , e. remiddi and j.b .
tausk , nucl .
phys . *
b580 * , 577 ( 2000 ) [ hep - ph/0003261 ] .
t. gehrmann and e. remiddi , nucl .
* b580 * , 485 ( 2000 ) [ hep - ph/9912329 ] . t. gehrmann and e. remiddi , nucl .
phys . proc .
* 89 * , 251 ( 2000 ) [ hep - ph/0005232 ] ; + c. anastasiou , j. b. tausk and m. e. tejeda - yeomans , nucl .
phys . proc .
* 89 * , 262 ( 2000 ) [ hep - ph/0005328 ] . c. anastasiou , e.w.n .
glover and c. oleari , nucl .
* b565 * , 445 ( 2000 ) [ hep - ph/9907523 ] . c. anastasiou , e.w.n . glover and c. oleari , nucl .
* b575 * , 416 ( 2000 ) [ hep - ph/9912251 ] . v.a .
smirnov , hep - ph/0007032 ; + t. gehrmann and e. remiddi , hep - ph/0008287 .
z. bern and g. chalmers , nucl . phys .
* b447 * , 465 ( 1995 ) [ hep - ph/9503236 ] ; + z. bern , v. del duca and c.r .
schmidt , phys .
* b445 * , 168 ( 1998 ) [ hep - ph/9810409 ] ; + d.a .
kosower and p. uwer , nucl .
* b563 * , 477 ( 1999 ) [ hep - ph/9903515 ] ; + z. bern , v. del duca , w.b .
kilgore and c.r .
schmidt , phys .
* d60 * , 116001 ( 1999 ) [ hep - ph/9903516 ] ; + s. catani and m. grazzini , hep - ph/0007142 .
campbell and e.w.n .
glover , nucl .
* b527 * , 264 ( 1998 ) [ hep - ph/9710255 ] ; + s. catani and m. grazzini , phys .
* b446 * , 143 ( 1999 ) [ hep - ph/9810389 ] ; nucl .
* b570 * , 287 ( 2000 ) [ hep - ph/9908523 ] .
a. gehrmann - de ridder and e.w.n .
glover , nucl .
* b517 * , 269 ( 1998 ) [ hep - ph/9707224 ] .
van neerven and a. vogt , phys .
* b490 * , 111 ( 2000 ) [ hep - ph/0007362 ] ; + a. retey and j.a.m .
vermaseren , hep - ph/0007294 .
t. kinoshita , j. math .
phys . * 3 * , 650 ( 1962 ) ; + t.d .
lee and m. nauenberg , phys .
rev . * 133 * , b1549 ( 1964 ) ; + j. collins , d. soper and g. sterman , in _ perturbative qcd _ , edited by a.h .
mueller ( world scientific , singapore , 1989 ) , and references therein . r.j .
gonsalves , phys .
rev . * d28 * , 1542 ( 1983 ) ; + g. kramer and b. lampe , z. phys . * c34 * , 497 ( 1987 ) , err .
_ * c42 * , 504 ( 1989 ) ; + t. matsuura and w.l .
van neerven , z. phys .
* c38 * , 623 ( 1988 ) ; + t. matsuura , s.c .
van der marck and w.l .
van neerven , nucl .
* b319 * , 570 ( 1989 ) .
harlander , hep - ph/0007289 .
g. t hooft and m. veltman , nucl .
* b44 * , 189 ( 1972 ) . v.n .
baier , v.s .
fadin and v.a .
khoze , nucl .
* b65 * , 381 ( 1973 ) ; + g. montagna , f. piccinini and o. nicrosini , phys .
rev . * d48 * , 1021 ( 1993 ) .
see e.g. k.s .
klbig , siam j. math .
* 17 * , 1232 ( 1986 ) .
l. lewin , _ dilogarithms and associated functions _
( macdonald , 1958 ) .
w. beenakker , f.a .
berends and s.c .
van der marck , nucl .
* b349 * , 323 ( 1991 ) . z. kunszt , a. signer and z. trcsnyi , phys . lett .
* b336 * , 529 ( 1994 ) [ hep - ph/9405386 ] .
a. signer , phys .
* b357 * , 204 ( 1995 ) [ hep - ph/9507442 ] . v. del duca , w.b .
kilgore and f. maltoni , nucl .
* b566 * , 252 ( 2000 ) [ hep - ph/9910253 ] . z. kunszt , a. signer and z. trcsnyi , nucl .
* b411 * , 397 ( 1994 ) [ hep - ph/9305239 ] ; + s. catani , m.h .
seymour and z. trcsnyi , phys .
rev . * d55 * , 6819 ( 1997 ) [ hep - ph/9610553 ] .
z. bern , l. dixon , d.c .
dunbar and d.a .
kosower , nucl .
phys . *
b425 * , 217 ( 1994 ) [ hep - ph/9403226 ] . z. bern , l. dixon and d.a .
kosower , nucl .
* b437 * , 259 ( 1995 ) [ hep - ph/9409393 ] . f. a. berends , p. de causmaecker , r. gastmans , r. kleiss , w. troost and t.t .
wu , nucl .
* b264 * , 265 ( 1986 ) ; + j. f. gunion and z. kunszt , phys .
* b176 * , 477 ( 1986 ) .
|
we present the two - loop virtual qed corrections to @xmath0 and bhabha scattering in dimensional regularization .
the results are expressed in terms of polylogarithms .
the form of the infrared divergences agrees with previous expectations .
these results are a crucial ingredient in the complete next - to - next - to - leading order qed corrections to these processes
. a future application will be to reduce theoretical uncertainties associated with luminosity measurements at @xmath1 colliders .
the calculation also tests methods that may be applied to analogous qcd processes .
hep - ph/0010075 slac pub8655 + ucla/00/tep/26 + october , 2000 * two - loop correction to bhabha scattering * 1.5 cm z. bern@xmath2 + _ department of physics and astronomy + ucla , los angeles , ca 90095 - 1547 _ 0.7 cm l. dixon@xmath3 + _ stanford linear accelerator center + stanford university + stanford , ca 94309 _ 0.6 cm and 0.4 cm a. ghinculov@xmath2 + _ department of physics and astronomy + ucla , los angeles , ca 90095 - 1547 _ 2 cm _ submitted to physical review d _ ' '' '' width 3.6 in @xmath4research supported by the us department of energy under grant de - fg03 - 91er40662.@xmath5research supported by the us department of energy under grant de - ac03 - 76sf00515 .
|
the study of damped lyman @xmath0 systems ( dlas ) is of great interest for numerous fields in astrophysics ( cf .
petitjean , these proceedings ) , including nucleosynthesis and the chemical evolution of galaxies and the universe at large .
in addition to investigations on the metallicity evolution with redshift ( e.g. pettini et al .
1994 , 1997 ) , spectroscopic studies with 10 m class telescopes have opened the opportunity to study ly@xmath0 absorption lines for a large number of elements including n , o , mg , al , si , s , cr , mn , fe , ni and zn . two main interpretations of the abundance ratios in dlas exist currently in the literature both suffering from several inconsistencies .
* 1 ) * lu et al . ( 1996 ) , prochaska & wolfe ( 1999 ) , outram et al . ( 1999 ) , and pettini et al . ( 2000 ) found that the relative abundance patterns indicate that the bulk of heavy elements in these high - redshift objects were produced by type ii supernovae .
however , when combined with the effects of dust depletion ( as indicated by the relative overabundance of zn relative to cr or fe ) this interpretation is inconsistent with other elemental abundance ratios such as n / o , s / fe , mn / fe and ti / fe .
this led lu et al .
( 1996 ) and prochaska & wolfe ( 1999 ) to conclude that the overabundance of zn / cr in dlas may be intrinsic to stellar nucleosynthesis in such systems ; a new physical process is then required to explain the damped ly@xmath0 abundance patterns . *
2 ) * pettini et al .
( 1997 , 1999 ) , vladilo ( 1998 ) and others argue that the relative abundances of cr , fe , and ni are consistent with a moderate degree of dust depletion that , once accounted for , leaves no room for the enhancement of @xmath0 elements over iron seen in metal - poor stars in the milky way .
this is in contradiction with a snii enrichment pattern , but it could be understood if star formation in these systems proceeded at a lower rate than in the early history of our galaxy ( e.g. pettini et al .
1999 ; centurin et al .
a similar conclusion is also drawn from the fe / h vs. redshift distribution of dlas when compared with our galaxy ( e.g. lu et al .
1996 ; prochaska & wolfe 2000 ) .
it is generally believed that the gas in dlas is mostly in the neutral stage due to the very high optical depth beyond the hydrogen ionisation limit .
therefore , it is assumed that all species with ionisation potentials lower than hydrogen are in a singly ionised stage , while other species are neutral .
however , lu et al . ( 1996 ) and prochaska & wolfe ( 1999 ) find a good correlation between the velocity structure of al iii and singly ionised species .
the ionisation potential of al@xmath1 is 18.8 ev , i.e. greater than that of hydrogen .
therefore , al@xmath2 is likely present in ionised , not in neutral gas . to explain the similarity of al iii and other low ionisation species line profiles , howk & sembach ( 1999 ) and izotov & thuan ( 1999 ) proposed that these lines originate in the same ionised region or in a mix of neutral and ionised clouds , and stressed the importance of abundance correction for ionisation effects . in the present work
we adopt a simple partially ionised two - region model for dlas and explore the possible implications on abundance ratio determinations within this framework .
full account of this work is given in izotov et al .
( 2000 , hereafter isc ) .
in the present paper we explore the following `` multi - component '' picture for dla systems : _ 1 ) _ an ionisation bounded region illuminated by an internal radiation field complemented by _ 2 ) _ a neutral region with a lower ( negligible ) metal content .
region 1 is modeled with the ionisation equilibrium code cloudy assuming internal stellar radiation and plane - parallel geometry .
the outer boundary of the slab is determined by the condition that a low electron temperature of 2000 k is reached .
we adopt stellar ionising spectra with various temperatures , and a metal abundance of 0.1 solar with relative element abundances equivalent to those observed in low - metallicity bcds ( e.g. , izotov & thuan 1999 ) . the ionisation parameter @xmath3 is varied over a wide range .
constraints on @xmath3 are derived from the observed al@xmath4/si@xmath1 column densities . to explore the average effects ,
a single `` typical '' value ( @xmath5 -3 ) is derived from the typical ratio @xmath6 [ @xmath7(al iii)/@xmath7(si ii ) ] @xmath8 -2 to -2.5 in dlas .
the corresponding ionisation correction factors , which are found to be non - negligible , are applied to all observed abundance ratios to be discussed below ( fig . 2 ) .
a similar model was recently considered by howk & sembach ( 1999 ) .
the main difference with their approach is that they consider ionised density - bounded regions surrounding the neutral , h i bearing clouds .
both qualitatively and quantitatively the resulting ionisation corrections are quite different from our model . for more details
see isc .
an important feature of both models is that the heavy element abundances in the ionised and neutral regions are a priori unrelated .
this implies in particular that relative abundance determinations with respect to hydrogen ( e.g. [ fe / h ] , [ zn / h ] ) , as derived from the observed column densities , may not be meaningful or are at least uncertain .
figure 1 shows that the observed correlation between [ si / fe ] and [ zn / fe ] which is attributed to progressive dust depletion ( prochaska & wolfe 1999 ) can also be explained by ionisation effects .
the best consistency is obtained for the radiation field corresponding to = 40 kk and @xmath3 values compatible with the observed al iii / si ii column density ratios . in this case one
finds also that after the ionisation correction is applied ( and assuming a final value [ zn / fe ] = 0 ) , the resulting [ si / fe ] values remain positive ( reminiscent of @xmath0-element enrichment ) .
no additional correction for dust depletion is required , but little depletion is not excluded .
( uncorrected ) . *
right : * observed n / o and n / si ratios in bcds ( filled circles : izotov & thuan 1999 ) and dlas ( open circles : lu et al .
1998 , outram et al . 1999 ) as a function of metallicity .
the typical ionisation correction is indicated by the arrow.,title="fig:",width=283 ] ( uncorrected ) . *
right : * observed n / o and n / si ratios in bcds ( filled circles : izotov & thuan 1999 ) and dlas ( open circles : lu et al .
1998 , outram et al . 1999 ) as a function of metallicity .
the typical ionisation correction is indicated by the arrow.,title="fig:",width=226 ] the corrected and uncorrected abundance ratios of various elements are shown in figure 2 ( left ) . except for cr
/ fe , all element ratios with respect to fe have downward corrections with typical values between @xmath8 0 ( mn / fe ) , -0.4 ( zn / fe ) and -0.6 dex ( al / fe ) . [ with respect to zn all ratios except for al / zn have to be corrected upwards . ] after correction we find that s and si ( overabundant in [ x / fe ] ) , mn and al ( underabundant ) , ti ( small overabundance ) , and zn ( @xmath8 solar ) follow quite well their respective snii pattern observed in metal - poor halo stars .
cr and ni ( including the oscillator strengths of fedchak & lawler 1999 ) show some disagreement whose origin is not understood yet ( atomic data ? other ? ) .
see isc for a more detailed discussion . the right panel in fig .
2 shows observed n / o and n / si ratios from metal - poor hii regions and measurements in dlas respectively . from our model
we find a typical upward correction of n i / si ii by 0.7 dex , as shown by the arrow .
once accounted for , it appears that the majority of measurements for dlas are at or above the n / o value from the bcds .
this suggests that a significant part of the difference previously found in [ n / si ] between hii regions and dlas ( lu et al .
1998 ) can be explained by ionisation effects .
interestingly , the recent direct measurement of n / o in a dla from unsaturated oi and ni lines ( open square ; molaro et al .
2000 ) , which are also little dependent on ionisation corrections , agrees as well with the metal - poor bcds . from our analysis
we conclude that , if our typical ionisation corrections apply , the present data shows no significant difference in n / o abundance ratios between low - metallicity bcds and dla systems implying a similar origin of these two elements . a detailed discussion of our assumptions and support from other data are presented in isc .
these working hypothesis have been adopted to study the main effects intrinsic stellar ionising radiation may have on observed heavy element patterns in such conditions .
future investigations should hopefully be able to improve on these issues and assess the validity of our picture .
it is the hope that the proposed simplified model reflects the main trends due to ionisation effects in dlas .
if correct , our picture offers a clear simplification in the understanding of heavy element abundance ratios in dlas and their comparison with the local universe .
99 centurin , m. , bonifacio , p. , molaro , p. , & vladilo , g. 2000 , , 536 , 540 fedchak , j.a . , lawler , j.e .
1999 , , 523 , 734 howk , j. c. , & sembach , k. r. 1999 , , 523 , l141 izotov , y.i . , schaerer , d. , charbonnel , c. , 2000 , , submitted ( isc ) izotov , y.i . , thuan , t.x .
1999 , , 511 , 639 lu , l. , sargent , w. l. w. , & barlow , t. a. 1998 , , 115 , 55 lu , l. , sargent , w. l. w. , barlow , t. a. , et al .
1996 , , 107 , 475 molaro , p. , bonifacio , p. , et al .
2000 , , in press ( astro - ph/0005098 ) outram , p. j. , chaffee , f. h. , & carswell , r. f. 1999 , , 310 , 289 pettini , m. , ellison , s. l. , steidel , c. c. , & bowen , d. v. 1999 , , 510 , 576 pettini , m. , ellison , s. l. , steidel , c. c. , et al . 2000 , , 532 , 65 pettini , m. , king , d. l. , smith , l. j. , & hunstead , r. w. 1997 , , 478 , 536 pettini , m. , smith , l. j. , hunstead , r. w. , & king , d. l. 1994 , , 426 , 79 prochaska , j. , & wolfe , a. m. 1999 , , 121 , 369 prochaska , j. , & wolfe , a. m. 2000 , , 533 , l5 vladilo , g. 1998 , , 493 , 583
|
the similarity between observed velocity structures of al iii and singly ionised species in damped ly@xmath0 systems ( dlas ) suggests the presence of ionised gas in the regions where most metal absorption lines are formed . to explore the possible implications of ionisation effects we construct a simplified two - region model for dlas consisting of an ionisation bounded region with an internal radiation field and a neutral region with a lower metal content . within this framework
we find that ionisation effects are important . if taken into account , the element abundance ratios in dlas are quite consistent with those observed in milky way stars and in metal - poor h ii regions in blue compact dwarf galaxies
. in particular we can not exclude the same primary n origin in both dlas and metal - poor galaxies . from our models no dust depletion of heavy elements needs to be invoked ; little depletion is however not excluded .
|
the relative abundance of exactly solvable models makes the one - dimensional ( 1d ) systems a unique testing ground for new concepts in condensed matter physics . however , during the last two decades the purely academical purpose of 1d studies transformed into an applied science as such extremely reliable 1d metallic materials as single wall carbon nanotubes ( swnts ) became widespread in laboratories @xcite .
the surging interest in low - dimensional materials is motivated by the rapidly shrinking lateral sizes of the microelectronic circuitry , which is soon expected to arrive at device dimensions at which any wire is truly one - dimensional from the point of view of current carrying electrons .
the most fundamental difference between the 1d metals and their higher - dimensional counterparts is the role of interactions .
while ( at least in clean systems ) the full perturbative expansion in correlation strength is controllable and even convergent in 3d and leads only to insignificant renormalization of parameters ( landau s fermi liquid theory ) , it diverges for one - dimensional systems . this is a clear indication of a new universality class the so - called tomonaga - luttinger liquid ( tll ) . in its simplest form it has first been discussed by tomonaga in @xcite .
a slight generalization was offered by luttinger @xcite . finally , an extremely convenient field theoretical reformulation has been presented by haldane @xcite .
it turns out , that whatever the precise form of the interaction potential , the generic correlation term in the tll universality class is the short ranged @xmath0-shaped interaction .
this and the reduced dimensionality allows one to diagonalize the corresponding hamiltonian with minimal effort just by a canonical transformation .
the physical properties of such systems are very interesting .
for instance , there is an algebraic singularity in the single particle density of states around the fermi energy @xmath1 due to a formation of highly correlated collective state .
the exponent of the corresponding power law is related to the interaction constant .
this effect is reflected in the transport properties of such 1d systems contacted by bulk 3d electrodes - the differential conductance is a power law of the applied voltage or temperature whatever is larger @xcite .
this has been seen in a number of experiments as a zero - bias anomaly @xcite .
the existence of exact solutions makes the correlated 1d metals extremely interesting for the emerging field of full counting statistics ( fcs ) @xcite .
the number of non - interacting electrons which are transferred through a structureless scatterer during a certain waiting time is expected to obey the binomial distribution @xcite .
it is unlikely that it is still valid as soon as the interaction potential among electrons becomes non - zero .
in fact , it is not even clear that under such conditions genuine physical electrons are involved in the transport processes .
for instance , in fractional quantum hall ( fqh ) samples the currents are believed to be carried by fractionally charged laughlin quasiparticles which have been observed in noise experiments @xcite .
therefore the study of such systems would give important insights into the non - equilibrium properties of strongly correlated electronic systems .
here we would like to review the developments in the field of fcs of such strongly correlated 1d systems .
we start with a brief description of the generic setup , whose transport characteristics one is usually interested in in section [ sect2 ] .
after that , in section [ sect3 ] , we present an exact analytic solution for the charge transfer statistics , which is possible at one special interaction strength . a generic way to solve the problem at arbitrary interaction strength
is discussed in the subsequent section [ sect4 ] . finally , as an application of the developed formalism we present some new results for the fcs of a contact between two tlls in section [ sect5 ] .
the starting point are the free 1d electrons with a linearized dispersion relation . as a result ,
the kinetic part describes two different fermion species : the right / left - movers , @xmath2 \ , , \ ] ] where @xmath3 and @xmath4 are the corresponding field operators , @xmath5 , and @xmath6 is the fermi velocity .
as one is mostly interested in the system s properties in the universality region which is essentially a low - energy regime , the linearization of the dispersion is unproblematic .
such electron segregation is even natural in fqh systems , where the r / l moving branches are spatially separated from each other and are not even allowed to interact in any way .
the generic interaction of the luttinger liquid universality class turns out to be extremely short ranged and effectively given by a @xmath0-shaped potential , the corresponding hamiltonian contribution being @xmath7 where @xmath8 is the interaction strength and @xmath9 is the full particle density operator .
it turns out that the fastest way to a diagonal hamiltonian leads through the bosonization representation , where both the kinetic as well as the interaction terms have similar shapes and can be merged by a mere rotation . in this way one obtains @xmath10 \ , ,
\end{aligned}\ ] ] where @xmath11 are the bosonic fields which can be directly related to the current density operators @xmath12 .
remarkably , the interaction strength is now reworked into the dimensionless parameter @xmath13 , which in the case of an fqh bar is obtained differently to be equal to the filling factor . to come that far it is usually not necessary to know the precise form of the relation between the phase fields and the original physical fermions the bosonization identity .
however , it is necessary as soon as one needs to generate the bosonic form of the interedge tunnelling operator @xmath14 .
fortunately , in the continuum limit this relation reduces up to an overall numerical prefactor to a rather simple prescription @xmath15 $ ] @xcite . as at least in the fqhe set - up the forward scattering
can clearly be dropped from the outset , the full hamiltonian then reads @xmath16 , where the back scattering ( in the case of chiral tll ) and interedge tunnelling ( in the case of a fqh device ) contribution is given by @xmath17 \ , .\ ] ] @xmath18 is the ( bare ) back scattering amplitude .
@xmath19 is equivalent to the so - called boundary sine - gordon ( bsg ) hamiltonian . from the renormalization group point of view
it is known to possess two fixed points : an infrared ( low energy ) and ultraviolet ( high energy ) ones . in the infrared limit , which is the one we are interested in
, the system becomes strongly coupled for @xmath20 ( repulsive interactions ) and @xmath18 flows to infinity . in the opposite situation @xmath21 ( attractive interactions ) @xmath18 is irrelevant , we do not consider this situation in the rest of the paper .
as has been realized by guinea in another context , the sine - gordon hamiltonian is simply diagonalizable not only at the trivial point @xmath22 , but also at @xmath23 as well @xcite .
indeed , @xmath24 $ ] trivially satisfies the original fermionic anticommutation relation at @xmath22 . on the other hand , setting @xmath23 one finds that the construction @xmath25 \right\}$ ] supplemented by appropriate prefactors is subject to fermionic statistics as well . in this section ,
we review the results of the fcs of a tll with a single impurity at the specific interaction parameter @xmath23 .
the results apply as well to the fcs of the fqh interedge tunnelling as the problems are isomorph . in the theory of fcs ,
the quantity of interest is the generating function @xmath26 , which is the fourier transform of the probability distribution @xmath27 of the charge @xmath28 crossing the impurity during time @xmath29 , @xmath30 .
the function @xmath26 generates moments of the charge @xmath31 transferred during time @xmath29 , @xmath32 , \ ] ] where @xmath33 is the @xmath34th cumulant of the distribution and @xmath35 is the time - dependent current through the scattering region .
the most powerful tool to calculate the resulting @xmath36 is the keldysh formalism .
it has been shown in ref .
@xcite that the generating function for the single impurity problem at @xmath23 may be written as @xmath37 with time - dependent hamiltonians @xmath38
\mp \frac{1}{4 } \lambda - \frac{1}{2 } e \tilde{v}t\right\ } , \ ] ] where @xmath39 orders operators along the keldysh contour and @xmath40\ ] ] is the generating function in the absence of backscattering .
furthermore , @xmath41 , where @xmath42 is the applied voltage ( that drops at the impurity site ) , and @xmath43 is the temperature .
if we follow matveev @xcite , we can transform the hamiltonian ( [ hpm ] ) in an elegant way and express it in terms of a new fermion .
this makes the calculation of the generating function a straightforward exercise with the resulting expression @xmath44\bigr\ } \nonumber\end{aligned}\ ] ] with @xmath45 + 1\}^{-1}$ ] , @xmath46 , and @xmath47 ( @xmath48 is the high - energy cutoff of the tll theory .
@xmath49 is of the order of the lattice spacing of the underlying lattice model . ) .
it has been shown in app .
c of ref .
@xcite that eq .
( [ stat ] ) can be brought into the more familiar form @xmath50\bigr\}\end{aligned}\ ] ] with @xmath51 + 1\}^{-1}$ ] and @xmath52 .
the two equivalent ways of writing the generating function eqs .
( [ stat ] ) and ( [ stat2 ] ) are dual to each other .
the reason is that eq .
( [ stat ] ) employs a formulation of the generating function using the energy - dependent reflection coefficient @xmath53 whereas eq .
( [ stat2 ] ) employs a formulation of the generating function using the energy - dependent transmission coefficient @xmath54 . in ref .
@xcite , the first three cumulants have been calculated in terms of special functions for arbitrary transmission , temperature , and applied bias . ] .
they read ( @xmath55 is the conductance quantum ) @xmath56 , \label{c1exakt } \\
\langle \delta^2 q \rangle & = & 2 t{d \langle \delta q \rangle \over dv } - \frac{e}{2 } t_b \coth\left({ev\over 2 t}\right ) { d \langle \delta q \rangle \over dt_b } + t
t_b{d^2 \langle \delta q \rangle \over dvdt_b } , \label{c2exakt } \\
\langle \delta^3 q \rangle & = & \frac{1}{2 } t t_b \frac{d^2 \langle \delta^2 q \rangle}{dv dt_b } - \frac{e}{2 } t_b \coth\left(\frac{ev}{2t}\right ) \frac{d \langle \delta^2 q \rangle}{dt_b } + 2 t \frac{d \langle \delta^2
q \rangle}{dv } \nonumber \\ & + & \frac{e}{2 } t t_b \coth\left(\frac{ev}{2 t } \right ) \frac{d^2 \langle \delta q \rangle}{dvdt_b } + \frac{e^2}{4 } t_b \sinh^{-2}\left(\frac{ev}{2 t } \right ) \frac{d \langle \delta q \rangle}{dt_b } , \label{c3exakt}\end{aligned}\ ] ] where @xmath57 is the digamma function . in the zero temperature limit ,
( [ c1exakt ] ) - ( [ c3exakt ] ) obey the relations ( [ cumrel ] ) between the higher order cumulants and @xmath58 derived earlier in ref .
as has already been mentioned above , the system can be mapped onto the sine - gordon model by a folding and a subsequent transformation to the even - odd basis @xcite .
after that procedure the excitations in the system are easily identifiable by means of the bethe ansatz technique : at integer @xmath59 there are current carrying charged kinks - antikinks , which we from now on denote by subscripts @xmath60 , and @xmath61 neutral ( charge zero ) breathers , which are bound states of kinks and antikinks . both excitation types are masseless and possess simple dispersion relations in terms of rapidity @xmath62 : @xmath63 and @xmath64 \
, e^\theta$ ] .
it is important to realise that these excitations are not related in any simple way to the original electrons .
nevertheless , only ( anti)kinks are scattered on the boundary and with a known quite simple scattering matrix @xmath65 given in ref . @xcite . for the latter we shall only need the resulting effective `` transmission coefficient '' of the ( anti)kinks through the boundary , @xmath66 where @xmath67 is the effective rapidity corresponding to the back scattering amplitude @xmath18 . in the non - interacting case
the knowledge of the impurity transmission and of the energy distribution function of the incoming particles ( essentially fermi distributions ) would be sufficient for construction of not only the non - linear @xmath68 but of the cgf as well @xcite .
this is different in the present case as not only are the quasiparticles distributed in a completely different way , the distribution functions of the ( anti)kinks [ @xmath69 , @xmath34 being the particle momenta ] at different energies ( or rapidities ) are correlated due to interactions in the bulk .
however , that does not matter in the expression for the current through the system .
just as in the non - interacting situation the corresponding expression only involves linear combinations of fermi distributions and it is natural that no higher correlations of population probabilities emerge even in the interacting case .
in fact , as is shown in @xcite , one obtains for the current @xmath70 \
, , \end{aligned}\ ] ] where @xmath71 is the expectation value of the density of the occupied states , @xmath72 is the temperature ( energy ) scale associated with @xmath18 and @xmath73 is the system length which is sent to infinity in the continuum limit ( s. below ) .
this situation changes completely as soon as one goes over to the noise properties of the system . here even in the non - interacting case products of fermi distribution functions emerge .
thus one has to expect some more involved bilinear products of @xmath74 at different momenta to enter the expression .
this indeed has been shown in @xcite .
the physical interpretation of the emerging picture is then very appealing : the transport is mediated by one by one elastic scattering of ( anti)kinks off the boundary with the known scattering matrix .
if only one channel with particles with momenta @xmath34 were involved the corresponding momentum generating function would be given by @xmath75
\right\}^{\tau / l } \ , .\end{aligned}\ ] ] @xmath29 is again the waiting time and @xmath76 corresponds to the spectral weight of the single @xmath34-channel .
it can easily be shown that for transport in multiple channels ( at different energies ) the corresponding momentum generating function is a product of single channel contributions @xmath77 .
the role of the breathers in the bulk however , is to entangle the single - particle energy distribution functions of the ( anti)kinks , hence the previous formula has to be averaged over all possible system configurations
. therefore we arrive at the following total cgf , @xmath78 \right\}^{\tau / l } \right\rangle\end{aligned}\ ] ] which was suggested in @xcite .
the exact results for @xmath23 follow from this ansatz quite naturally .
first of all under these conditions there are no breathers in the bulk and the ( anti)kinks are free fermions . that is why they fill the momentum space like ordinary electrons and are subject to fermi distributions @xmath79 . since the quasiparticles are now non - interacting , the averaging in ( [ startpoint ] )
is superfluous and must be dropped .
furthermore , the transmission coefficient ( [ transmcoeff ] ) is given by @xmath80 as @xmath81 and @xmath82 , @xmath83 being a cut - off which falls out of any observable . combining everything we immediately obtain the result ( [ stat2 ] ) .
another important limiting case is the zero temperature regime . here
all correlations among the distributions @xmath84 vanish and all irreducible momenta turn out to be given by @xcite @xmath85 \ , , \ ] ] where @xmath86 , @xmath87 $ ] , @xmath88[1 - 2 { \cal t}(\theta)]$ ] etc .
a following very convenient relation between different @xmath89-functions can be verified , @xmath90 with @xmath91 , which helps to reduce every single cumulant to a calculation of the current derivative with respect to impurity rapidity , @xmath92 summing up all cumulants into the generating functional one obtains @xmath93 although in general no closed analytic expressions for the current exist , the expansions for strong and weak back scattering turn out to be very useful .
there are clear physical pictures in these limits @xcite .
when the system is in vicinity of the weak coupling point , the true ground state is that of two completely disconnected systems .
it is evident that then only physical electrons with integer charge may tunnel between the subsystems . on the other hand , in the strongly coupled system a collective state between the edges
is formed , which in the case of a fqh bar has fractionally charged laughlin quasiparticles for elementary excitations .
for strong backscattering or weak tunneling , @xmath94 , where @xmath95 is defined by @xmath96\}/\gamma\ { g/[2 ( g-1)]\ } \
, , \ ] ] one obtains @xmath97 where @xmath98 with @xcite @xmath99 } \ , .\ ] ] observe that the counting field @xmath100 enters the expression ( [ wc1 ] ) with an integer prefactor @xmath101 . , where @xmath102 is the charge of current carrying excitations .
] the physical meaning of the expression ( [ wc1 ] ) is quite illuminating .
suppose that @xmath103 is the probability per unit time to transfer a particle of charge @xmath101 through the impurity barrier .
then the charge transferred in the time interval @xmath29 is the result of a poisson process for particles of charge one crossing the barrier , contributing a current @xmath104 , plus a poisson process for particles of charge two contributing a current @xmath105 , etc .
the logarithm of the fourier transform of the probability distribution of all these poisson processes would then be the expression ( [ wc1 ] ) .
the only subtle point now is that the signs of the rates @xmath103 are not quite right since a classical process would require that all rates are positive . certainly , the @xmath106term is indeed a poisson process for the tunneling of single electrons , but the joint tunneling of pairs of electrons ( and of multiples thereof ) @xmath107 comes with the different sign , which is an effect of quantum interference . in the opposite limit of weak backscattering
, the expression for @xmath108 reads @xmath109\ , , \end{aligned}\ ] ] where @xmath110 this form is quite similar to ( [ wc1 ] ) , but there are subtle differences . the first term in the expression ( [ wb1 ] )
represents the current of fractionally charged quasiparticles in the absence of the barrier .
the exponential factor @xmath111 indicates that now we have tunneling of quasiparticles of charge @xmath112 and of multiples thereof , and the minus sign in the exponent means that the tunneling diminishes the current instead of building it up as in the strong - backscattering regime .
the sign of the tunneling rate @xmath113 is now that of @xmath114 .
therefore , the perception of clusters of quasiparticles with fractional charge @xmath112 tunneling independently with a classical poisson process is quite appropriate when @xmath112 is small .
the classical limit is reached as @xmath112 goes to zero .
then all the rates @xmath113 are positive , but the quantum fluctuations @xmath115 with @xmath116 have faded away , @xmath117 \;=\ ; i \lambda\tau\ , g\ , g_0^ { } v \sqrt{1 - \big(\frac{t'_{bs}}{v}\big)^{2 } } \ , .\end{aligned}\ ] ] these results have been derived relying on the integrable approach .
alternatively , one may return to the bosonic representation , the bsg model ( [ hb ] ) with ( [ hb1 ] ) . upon integrating out the luttinger liquid modes away from the impurity located at @xmath118 ,
one arrives at the coulomb gas representation of the quantum impurity problem which at zero temperature involves long - range logarithmic interactions .
we see from this representation that the luttinger liquid modes act as an ohmic thermal reservoir @xcite .
reconsidering the cumulant generating function in the coulomb gas representation with the rigorous nonequilibrium keldysh or feynman - vernon method , one finds confirmation of the expressions ( [ wc1 ] ) ( [ wb2 ] ) and deeper understanding of the cumulant relation ( [ cumrel ] ) @xcite . for arbitrary impurity strength or finite temperatures one has to evaluate ( [ startpoint ] ) with the help of thermodynamic bethe ansatz technique .
the evaluation of @xmath119 products is rather demanding . in most of the experiments , however , still only the cumulants of lowest orders are measured . in @xcite a general procedure for the calculation of cumulants of any order is presented and the third irreducible moment is explicitly calculated .
interestingly , the leading low - temperature contribution to the zero temperature cumulant generating function can be calculated in analytic form . to this aim , we work in the coulomb gas representation and perform the low temperature expansion of the charge interaction factor , which is a power series in @xmath120 .
we find that the @xmath120contribution to the cumulant generating function can be universally written in terms of the second derivative of the zero temperature expression with respect to the voltage , whereas higher - order terms are non - universal . in the end
we find the cgf for strong backscattering in the form ( [ wc1 ] ) with the rate expressions @xmath103 given by @xmath121 k_m^{}(v ) = \big\ { 1 + \frac{\pi^2}{3 g } { \textstyle ( \frac{2}{g}-2 ) m \big[1+(\frac{2}{g}-2)m\big ] } \frac{t^2}{v^2}\big\ } \ , k_m^{}(v ) \,,\ ] ] where terms of order @xmath122 are disregarded . in the opposite weak - backscattering limit
we obtain in generalization of the expression ( [ wb1 ] ) @xmath123\ , , \end{aligned}\ ] ] where ( modulo terms of order @xmath122 ) @xmath124 \ , k_m^{}(v ) = \big\ { 1 + \frac{\pi^2 } { 3 g } ( 2g-2 ) m \big[1+(2g-2)m\big ] \frac{t^2}{v^2}\big\ } \ , \widetilde k_m^{}(v ) \;.\ ] ] the @xmath125-contribution to the cumulants is a distinctive signature of interacting 1d electrons . a related phenomenon is the universal @xmath120-behavior observed at low temperatures in open quantum systems with ohmic dissipation , e.g. @xmath120-enhancement in macroscopic quantum tunneling @xcite .
it is also the origin of the universal wilson ratio , observed e.g. in kondo systems .
it is expected that as the lateral dimension of the microelectronic circuitry becomes smaller , the ultimate nanoscale wiring would become truly one - dimensional from the point of view of current carrying electrons .
indeed , a metallic wire with a thickness of only several nm can only support electrons propagating as an @xmath126-wave since the states with higher angular momenta lie so high that even at room temperature they can not be excited .
thus , in the future nanoelectronics one has strongly interacting quantum wires as one of the basic circuitry elements .
the next to a simple wire complexity level is achieved by a point contact of two such conductors .
in fact , following the theoretical proposal of @xcite a number of successful experiments has been already conducted on swnt junctions measuring the corresponding non - linear @xmath68 @xcite .
we expect that noise or even fcs measurement will be conducted on such set - ups in the nearest future .
the most general way of coupling two conductors is not only by a particle exchange via tunnelling , but by an electrostatic interaction as well .
as has been discussed in @xcite , in the true fixed point the inter - wire tunnelling turns out to be completely irrelevant .
moreover , although the density - density coulomb coupling between the wires consists of a number of terms only one of them with the lowest scaling dimension survives in the fixed point .
the effective hamiltonian is then given by @xmath127 + \lambda_c \sin \left [ \phi_{l 1 } ( 0 ) - \phi_{r 1}(0 ) \right ] \ , \sin \left [ \phi_{l 2 } ( 0 ) - \phi_{r 2}(0 ) \right ] \ , , \end{aligned}\ ] ] where @xmath128 $ ] is given by ( [ hb ] ) for the wire @xmath129 and @xmath130 is the coupling constant
. a very convenient simplification of the problem is achieved by introduction of channel ( anti)symmetric variables according to the prescription is assumed , but this requirement is not restrictive in any way and a more general transformation can easily be conceived . ]
@xmath131 in this way the original problem separates into two commuting contributions , @xmath132 + p \
, \frac{\lambda_c}{2 } \cos \left\{\sqrt{2 } [ \phi_{l p } ( 0 ) - \phi_{r p}(0 ) ] \right\ } \ , .\end{aligned}\ ] ] it turns out that by the above rotation even the non - equilibrium transport problem separates .
not only the voltage applied to the new @xmath60 channels is given by the ( anti)symmetric combination of the original @xmath133 , but the physical currents in the @xmath134 channels are calculated via prescription @xmath135 .
in fact , this separation is even more fundamental : it is easily realised that even the fcs should factorize into two independent contributions with counting fields @xmath136 as @xmath137 \ , , \ ] ] where @xmath138 is given by ( [ startpoint ] ) .
of course , the first order cumulant with respect to either of the channels leads to the non - linear @xmath68 as discussed in @xcite , provided one employs the `` voltage drop '' way of coupling to the voltage sources @xcite .
now we want to analyse correlations of higher orders .
the direct ( within one of the channels @xmath129 ) noise contribution is trivially found by double derivative , @xmath139 where @xmath140 ^ 2 } { \chi^2(\lambda_\pm , v_\pm ) } \ , .\ ] ] the corresponding explicit formulas are given in @xcite .
as @xmath141 the noise locking found in @xcite follows immediately , @xmath142 another interesting quantity is the yet unstudied cross correlation @xmath143 remarkably , this correlation vanishes completely when there is no net transport in either of the wires .
even more surprising feature is that this kind of cancellation is temperature - independent .
thus the correlations ( [ finresult ] ) are unaffected by the johnson - nyquist noise .
this phenomenon along with the effect ( [ locking ] ) could be used in future experiments to identify the sophisticated multi - particle state the system is in .
in most cases the task of fcs calculation in a generic system requires application of nonequilibrium techniques .
usually this is a rather challenging endeavour for strongly correlated systems , where pronounced deviations from the conventional charge transfer statistics are expected .
one class of such systems covers interedge tunneling in the fractional quantum hall bars and impurities imbedded into chiral tomonaga - luttinger liquids . here
the charge of current carrying excitations is different from the elementary charge .
this results in a unique fcs , which differs considerably from the noninteracting setups .
despite the fact , that these systems are integrable with help of the thermodynamic bethe ansatz , there is no possibility to exploit it directly to address their fcs .
nevertheless , it still can be accessed using a number of shortcuts .
one of them is the possibility to map the corresponding boundary sine - gordon model to a noninteracting fermionic theory via refermionization procedure at some special nontrivial interaction strength .
the resulting hamiltonian turns out to be quadratic and easily diagonalizable with elementary methods making a genuine nonequilibrium evaluation of the fcs a rather simple task .
another way towards the fcs is offered by the equilibrium bethe ansatz solution , which is able to identify all possible excitations in the system as well as the corresponding scattering matrices .
it turns out that for the generic interaction strength there are only two excitations which are responsible for the transport : kinks and antikinks , whose reflections on the boundary correspond to the physical current carrying processes .
all other excitations are merely responsible for the rather involved ( non - fermi like ) energy distribution functions .
the values for the probabilities to occupy ( anti)kink states at different momenta @xmath84 are not mutually correlated as long as the system is at zero temperature .
thus the application of elementary statistical arguments immediately leads to the fcs .
the same result could be achieved by a straightforward coulomb gas expansion method .
interestingly , this approach is also able to supply the leading corrections around the @xmath144 result . at finite temperatures the distributions @xmath84
are correlated with each other prohibiting the fcs calculation along this path .
the cumulant generating function @xmath26 can be evaluated nonetheless by an ansatz which takes advantage of the fact , that the current transport is mediated by momentum conserving scattering of ( anti)kinks off the boundary .
indeed , the fcs can be expressed as a complicated average of a product of @xmath145s describing transport in individual channels corresponding to a designated momentum value .
all these techniques are not restricted to the single impurity problems only .
they can rather be applied to access the fcs of more complicated junctions and networks of quantum wires in the tll state .
we expect , that this kind of information is especially interesting in view of recent advances in the fields of nanoelectronics , spintronics and quantum computation .
ak acknowledges the support by the dfg grant ko 2235/2 .
bt was supported by the swiss nsf and the nccr nanoscience .
the authors would like to thank hubert saleur and markus kindermann for numerous fruitful discussions .
|
we review recent advances in the field of full counting statistics ( fcs ) of charge transfer through impurities imbedded into strongly correlated one - dimensional metallic systems , modelled by tomonaga - luttinger liquids ( tlls ) .
we concentrate on the exact analytic solutions for the cumulant generating function ( cgf ) , which became available recently and apply these methods in order to obtain the fcs of a non - trivial contact between two crossed tll .
|
this is the second paper in a sequence of papers devoted to the relations between higher categories and @xmath3-fold loop space theory . in the first paper
@xcite we developed the necessary categorical techniques which allow us to go back and forth between @xmath3-operads and classical symmetric operads .
the main goal of this paper is to clarify the geometric and homotopy theoretic aspects of this theory . to do this we restrict ourselves to a class of so called pruned @xmath9terminal @xmath10operads .
this is a slightly smaller category of @xmath3-operads than we considered in @xcite but it is big enough to include most applications we have in mind .
the reason is that the functor of desymmetrisation from symmetric operads to @xmath3-operads @xcite can be factorised through the category of pruned @xmath3-operads .
moreover , as in the unpruned case this desymmetrisation functor preserves endomorphism operads .
this allows us to construct a theory of symmetrisation very much in parallel to the unpruned case .
it turns out that pruned @xmath3-operads are easier to handle from the combinatorial point of view .
we have a conjecture , however , that the main results of this paper are also true in the case of @xmath4-terminal @xmath3-operads but so far we have been unable to prove it in this generality .
we apply the categorical methods of @xcite to the category of pruned @xmath4-terminal @xmath3-operads and to the even smaller class of reduced @xmath4-terminal @xmath3-operads .
these methods immediately imply the existence of some categorical symmetric operads @xmath11 and @xmath12 which represent the theories of internal @xmath4-terminal pruned @xmath3-operads and internal @xmath4-terminal reduced @xmath3-operads inside categorical symmetric operads , in full analogy with the categorical operad @xmath13 in @xcite .
our first significant result here is theorem [ phnrhnmn ] which asserts that the simplicial operads @xmath14 and @xmath15 where @xmath16 is the nerve functor , are @xmath17-operads in the category of simplicial symmetric operads and reduced simplicial symmetric operads respectively .
together with the theorems [ formulap ] and [ formular ] , these theorems show that once we have a space with an action of a contractible pruned @xmath4-terminal @xmath3-operad there is an action of an @xmath17-operad on this space .
an analogous result holds for the reduced operads in chain complexes .
as we conjectured in @xcite this should give a very natural proof of deligne s conjecture answering a question by kontsevich @xcite - operad acting naturally on a @xmath18-graph consisting of @xmath19-categories , @xmath19-functors and the complex of their derived natural transformations .
in the particular case of a @xmath19-category with one object and transformations of the identity functor , we obtain the hochschild complex of an associative algebra and then we apply our theorem [ enchain ] @xcite . ] .
there are , however , other results in the present paper which we believe are significant .
first we observe that the desymetrisation of the operad of fulton and macpherson @xmath20 of compactified real configuration spaces @xcite contains a contractible reduced @xmath4-terminal @xmath3-operad which we call the getzler - jones operad @xmath21 actually , this operad was discovered by getzler and jones in their remarkable preprint @xcite .
the apparatus of @xmath3-operads did not exist at that time and getzler and jones attempted to express the properties of @xmath22 in terms of a natural subdivision of the operad @xmath20 .
it turned out , however , that getzler - jones subdivision does not give a cellular structure compatible with the operadic structure of @xmath23 as was first observed by d.tamarkin ( see @xcite for an explanation of tamarkin s counterexample . )
this counterexample implied considerable technical difficulties in the proof of deligne s conjecture .
the second implicit appearance of @xmath22 was in the kontsevich and soibleman paper @xcite .
for @xmath24 they considered closed contractible subsets @xmath25 of @xmath26 where @xmath27 is a finite set with two complementary orders on it ( see definition [ complorder ] ) .
they used some properties of @xmath25 to prove deligne s conjecture for @xmath28-algebras .
we show that the space @xmath29 of arity @xmath27 is indeed equal to the kontsevich - soibelman @xmath25 for a pruned @xmath3-tree @xmath27 ( see section [ comporders ] for the explanation of the connection between trees and complementary orders , and section [ gjoperad ] for the definition of @xmath25 for any @xmath3 ) .
we show that @xmath22 can be considered as a natural analogue of @xmath20 in the category of reduced @xmath4-terminal @xmath3-operads .
in particular , set theoretically it is a free reduced @xmath4-terminal @xmath3-operad on the reduced @xmath3-collection of getzler - jones cells .
this last result leads to the theorem that the symmetrisation of @xmath22 is exactly @xmath20 which is the basis for the main results of our paper .
we also think that this gives an interesting new insight to the geometry of @xmath30 we are going to continue this study in the next paper of this series . the operad @xmath22 is still not cellular in the strong sense that it is not a geometric realisation of a poset of cells of a regular cw - complex .
counterexample works well in this case too .
nevertheless , its unbased version @xmath31 is a cellular object in the category of unbased reduced @xmath4-terminal @xmath3-operads in the model category theoretic sense , and in particular it is a cofibrant contractible operad .
the term unbased here means that we forget about nullary operations of our operads .
there is also an @xmath3-operad @xmath32 which is an unbased categorical reduced @xmath3-operad freely generated by its internal reduced @xmath3-operad .
the geometric realisation of the nerve of this operad is cofibrant and contractible and is strongly homotopy equivalent to the operad @xmath33 this implies a homotopy equivalence between the geometric realisation of @xmath34 and @xmath35 this operad @xmath36 ( the unbased categorical reduced symmetric operad freely generated by its internal reduced @xmath3-operad ) plays , therefore , the role of the nonexistent poset operad of cells of the getzler and jones decomposition . perhaps , the most intresting result of the above study of the combinatorics of the getzler - jones operad is the following generalisation of stasheff s classical theory of @xmath28-spaces .
the closure of a getzler - jones cell @xmath37 inside @xmath38 is a manifold with corners homeomorphic to the ball of dimension @xmath39 where @xmath40 is the number of edges of the @xmath3-tree @xmath41 the cellularity of @xmath31 means that an action of @xmath31 on a pointed space @xmath8 can be described as an inductive process of extension of higher homotopies from the boundary of @xmath42 to its interior in exact analogy with stasheff s description of @xmath28-spaces . and
our theorem [ symgj ] states that the collection @xmath43 where @xmath44 runs over the set of pruned @xmath3-trees , gives full coherence conditions for @xmath17-spaces .
if @xmath45 the collection @xmath46 is the sequence of associahedra @xcite and we get stasheff s theorem .
the difference between @xmath1-dimensional and higher dimensional cases appears in the existence of some cells ( tamarkin s cells ) in @xmath22 which are not completely on the boundary of @xmath47 this is not excessive information , however , just a defect of our language when we try to express the coherence laws in terms of an action of an @xmath4-terminal @xmath3-operad .
there is a way of avoiding this problem but we will have to pay a price by using more sophisticated @xmath3-operads which have full source and target operations . we will consider this subject in a forthcoming paper
. finally , in the last section of this paper we apply our techniques to the case of swiss cheese type operads @xcite .
the advantage of our categorical methods is that we have nothing to prove here once we put the right definitions of our main categories and functors in place .
we deduce immediately a symmetrisation formula for swiss cheese type @xmath3-operads - operadic point of view and for sending me a preliminary version of his papers concerning the action of swiss - cheese operads on associative algebras and their hochschild complexes @xcite .
] and other swiss - cheese analogues of the results for classical operads .
we hope that similar results can be obtained for symmetrisation of some other important coloured operads ; for example , operads for morphisms between @xmath17-algebras . this should lead to a better understanding of coherence conditions for such morphisms .
we will freely use the terminology from @xcite concerning @xmath3-trees , @xmath4-terminal @xmath3-operads and their algebras .
notice , however , that the main objects for us here are @xmath4-terminal @xmath3-algebras of our @xmath3-operads .
roughly speaking such an algebra is an object @xmath8 of our basic symmetric monoidal category @xmath48 together with an action @xmath49 where @xmath50 is the number of tips of @xmath27 satisfying some natural conditions .
an @xmath3-operad @xmath5 may have , however , more complicated types of algebras which involve source and target operations but we do not need them here .
so we will speak simply about category of algebras of @xmath5 having in mind the subcategory of its @xmath4-terminal algebras .
we refer the reader to @xcite for more discussion about this issue
. * acknowledgements . *
i would like to thank andr joyal for pointing out a mistake in the first version of this paper .
it is exactly the work on this mistake which led me to an understanding of the role of the getzler - jones operad in @xmath3-operad theory .
i am also very grateful to him for organising my visit to montreal in july 2003 and for our fruitful and inspiring discussions during that time .
i also wish to express my gratitude to c.berger , l.breen , e.getzler , s.lack , m.markl , j.stasheff , r.street , d.tamarkin and a.voronov for numerous illuminating conversations and correspondences .
finally , i gratefully acknowledge the financial support of the scott russell johnson memorial foundation and australian research council ( grant # dp0558372 ) .
here we consider a combinatorial techniques which will be used further to develop a theory of pruned @xmath3-operads .
the machinery of bar - codes from @xcite is an equivalent language but we prefer to work with the notion of complementary orders introduced by kontsevich and soibelman in @xcite . a partial order on a set @xmath8 is a transitive , antireflective relation on @xmath51 it is called linear if any two elements are comparable .
antireflective here means that the diagonal is always in the complement relation .
we choose this terminology just for technical reasons .
of course , we always can add the diagonal to our relations and work with reflective relations .
so if we are given a partial order @xmath52 we will often use the notation @xmath53 to mean that @xmath54 or @xmath55 this does not lead to any trouble .
observe , also that a partial order in our sense is always antisymmetric .
[ complorder ] let @xmath56 be a set .
suppose we have an @xmath3-tuple of partial orders @xmath57 on @xmath58 we call them complementary orders provided any two elements @xmath59 can be compared with respect to exactly one order @xmath60 a set with a given @xmath3-tuple of complementary orders on it will be called an @xmath3-ordered set .
let @xmath27 be a pruned @xmath3-tree @xmath61\stackrel{\rho_{n-1}}{\longrightarrow}[k_{n-1}]\stackrel{\rho_{n-2}}{\longrightarrow } ...
\stackrel{\rho_0}{\longrightarrow } [ 1 ] .\ ] ] then the relation : @xmath62 if and only if @xmath63 in @xmath64 $ ] and @xmath65 but @xmath66 defines @xmath3 complementary orders on the set of tips of @xmath41 * proof . *
the proof is obvious .
@xmath67 in fact we can characterise pruned @xmath3-trees in the following way : [ treeord ] suppose we are given @xmath3-complementary orders @xmath68 on a finite set @xmath8 such that * if @xmath69 and @xmath70 then @xmath71 then there exist a linear order on @xmath8 and a pruned @xmath3-tree @xmath27 such that the ordinal of its tips is @xmath72 $ ] and the complementary orders @xmath68 are determined by @xmath41 * proof . * the linear order @xmath52 on @xmath8 is defined by the requirement that @xmath63 if and only if there exists a @xmath73 such that @xmath74 let @xmath64 $ ] be the corresponding ordinal .
suppose that there exists a triple @xmath75 from @xmath64 $ ] such that @xmath76 but @xmath77 for some @xmath78 then @xmath77 so @xmath79 contradiction .
similarly for the other side .
so the order @xmath80 determines a subdivision on @xmath81 intervals of the ordinal @xmath64 .$ ] this can be considered as a surjection of ordinals @xmath64\rightarrow [ k_{n-1 } ] .$ ] obviously we have @xmath4 complementary orders on @xmath82 $ ] which satisfy the conditions of the lemma .
so we can proceed and construct a tree @xmath41 @xmath67 we will call an @xmath3-ordered set @xmath8 totally @xmath3-ordered if it satisfies the conditions of lemma [ treeord ] . if @xmath83 is totally @xmath3-ordered and the induced linear order makes it equal to the ordinal @xmath84 $ ] then we call @xmath8 an @xmath3-ordinal .
this includes the case of a terminal ordinal ( one element set and empty @xmath3 complementary orders ) and of an initial ordinal ( empty set ) .
we will often consider a special @xmath3-ordinal for which only one order @xmath85 is nonempty .
we will use the notation @xmath86 for such an ordinal on @xmath87 ( see @xcite for the picture of corresponding pruned tree with the same notation ) .
let @xmath8 and @xmath88 be two @xmath3-ordered sets .
an order preserving map ( or a map of @xmath3-ordered sets ) from @xmath8 to @xmath88 is a map @xmath89 such that @xmath69 in @xmath8 implies that @xmath90 for some @xmath91 or @xmath92 for @xmath93 suppose we are given two @xmath3-tuples of complementary orders @xmath94 and @xmath95 on the same set @xmath96 we will say that @xmath94 dominates @xmath95 ( notation @xmath97 ) if @xmath98 in @xmath94 implies @xmath99 for some @xmath91 or @xmath100 for @xmath101 in @xmath102 of course , @xmath94 dominates @xmath95 if and only if the identity map @xmath103 is a map of @xmath3-ordered sets .
let @xmath61\stackrel{\rho_{n-1}}{\longrightarrow}[k_{n-1}]\stackrel{\rho_{n-2}}{\longrightarrow } ...
\stackrel{\rho_0}{\longrightarrow } [ 1]\ ] ] and @xmath104\stackrel{\xi_{n-1}}{\longrightarrow}[s_{n-1}]\stackrel{\xi_{n-2}}{\longrightarrow } ...
\stackrel{\xi_0}{\longrightarrow } [ 1]\ ] ] be two pruned @xmath3-trees .
let @xmath105 \rightarrow [ s_n]$ ] be a map .
then there exists a map of trees @xmath106 such that @xmath107 if and only if @xmath108 is a map of @xmath3-ordered sets .
* proof .
* if such a @xmath109 exists the orderpreserving property of @xmath108 is obvious .
now we want to reconstruct a @xmath109 from @xmath108 .
we put @xmath110 \rightarrow [ s_n ] , $ ] of course .
now take a point @xmath111 from @xmath82 $ ] then its preimage under @xmath112 consists of an interval and we define @xmath113 where @xmath114 is an arbitrary element from the preimage of @xmath115 since @xmath108 preserves order , this definition is correct . indeed ,
if @xmath116 is another element from the preimage then @xmath117 and , hence , their images under @xmath118 are equal .
now we can check quite easily that the constructed @xmath119 preserves the @xmath4 complementary orders determined by @xmath120 and @xmath121 and we can proceed with our construction . @xmath67 we assume that the only degenerate pruned @xmath3-tree is @xmath122 .
we thus have [ ordn ] the category @xmath123 of @xmath3-ordinals and their order preserving maps is isomorphic to the category of pruned trees and their morphisms .
one can consider the poset @xmath124 of all total complementary @xmath3-orders on a fixed set @xmath125 with respect to the domination relation .
if @xmath126 we will denote this poset by @xmath127 the symmetric group @xmath128 acts naturally on @xmath127 let @xmath129 be the subcategory of @xmath130 whose objects are pruned @xmath3-trees with @xmath50 tips and whose morphisms are morphisms of trees which are bijections on tips .
we call such a morphism _ a quasibijection_. theorem [ ordn ] implies [ quasibijection ] there is a natural isomorphism of categories @xmath131 the set @xmath132 appeared many times in the literature @xcite .
it is isomorphic to the poset of cells of the classical fox - neuwirth stratification of configuration space @xmath133 ( see section [ fmandgj ] ) ; it is also isomorphic to the getzler and jones poset of bar - codes and to the _ milgram poset _ of @xcite .
we use the last name in this paper .
the category @xmath129 is isomorphic to berger s _ shuffle category _
@xcite , where it is also shown how to reconstruct @xmath132 as a grothendieck construction of a _ shuffle functor _
@xmath134 let @xmath135 be an order preserving map of @xmath3-ordinals .
for an element @xmath136 the preimage @xmath137 with its natural structure of an @xmath3-ordinal induced from @xmath27 will be called the fiber of @xmath108 over @xmath114 . following @xcite and @xcite we define _ a tree _ to be an isomorphism class of finite connected acyclic graphs with a marked vertex @xmath138 called the root .
a vertex which is not the root vertex and having valency more than @xmath18 is called _ an internal vertex_. the edges of this graph with one open end are called _ leaves . _
every edge of a tree has , therefore , a target vertex and a source vertex provided this edge is not a leaf .
every vertex also has a set of incoming edges and one outcoming edge if the vertex is not a root .
a monotone path in a tree is a sequence of edges such that the target of each edge in the sequence is equal to the source of the next edge .
the length of this path is the number of edges in the sequence .
the monotone paths are ordered by inclusion .
notice that there can be several maximal paths in the tree with respect to this order .
the _ length of a tree _ is the maximal length of the maximal monotone paths in the tree .
if we choose an incoming edge @xmath139 at a vertex @xmath140 we can construct a subtree by considering all vertices and edges which can be connected to @xmath140 by a monotone path .
we call this subtree _ a branch _ corresponding to @xmath141 a tree is called _ labelled _ by the set @xmath142 if there is given a bijection from the set of the leaves of the tree to @xmath143 a labelled @xmath3-planar tree is a labelled tree such that for every internal vertex @xmath140 the set of incoming edges is a totally @xmath3-ordered set .
since every totally @xmath3-ordered set has a canonical linear order then every @xmath3-planar tree has a canonical structure of a planar tree .
so we often will speak about @xmath3-planar trees as planar trees decorated by @xmath3-ordinals .
every such tree determines a structure of @xmath3-ordered set on the set of its labels . for a vertex @xmath140 of a planar tree @xmath144 and a label @xmath145
let us define @xmath146 to be the last incoming edge ( if it exists ) in the monotone path which connects the outcoming edge with label @xmath114 and @xmath147 then we put @xmath69 if @xmath148 for a vertex @xmath149 , which always exists and is unique . the domination relation
, therefore , induces a partial order on the set of @xmath3-planar trees labelled by the set @xmath150 we will use the same notation @xmath151 for this relation .
now we can easily give the definition of an @xmath3-operad based on @xmath3-ordinals and their morphisms .
we will , however , show that this is just a subcategory of the category of @xmath3-operads which we will call the category of pruned @xmath4-terminal @xmath3-operads @xmath152 where @xmath48 stands for a symmmetric monoidal category over which we consider the operads
. we will call a morphism of trees @xmath106 a _ full injection _ if it is injective , and bijective on tips .
obviously every fiber of @xmath109 is equal to @xmath153 if @xmath5 is an @xmath4-terminal @xmath3-operad in @xmath48 and @xmath109 is a full injection then we have the following composite morphism @xmath154 we call an @xmath3-operad _ pruned _ if this morphism is an identity for every full injection @xmath155 in the case of the empty set of fibers of @xmath109 ( i.e. @xmath156 ) , we require the morphism @xmath157 to be an identity .
this means , that for any degenerate tree @xmath158 there is an identification @xmath159 recall that the only degenerate pruned tree is @xmath160 a pruned ( @xmath4-terminal ) @xmath3-collection in @xmath48 is a family of objects @xmath161 , where @xmath27 runs over the set of pruned @xmath3-trees .
they form a category @xmath162 with respect to termwise morphisms of collections .
we will denote by @xmath163 the maximal pruned subtree of a tree @xmath41 then we can reformulate the definition of the pruned @xmath3-operad in the following way : a pruned @xmath4-terminal @xmath3-operad is given by a pruned @xmath4-terminal @xmath3-collection @xmath5 equipped with : * a morphism @xmath164 * a morphism @xmath165 for every morphism of trees @xmath166 between pruned trees in @xmath167 they must satisfy the usual identities .
this makes it obvious that the category of pruned operads is isomorphic to the category of @xmath3-operads based on @xmath3-ordinals .
[ pon ] if @xmath48 is a cocomplete symmetric monoidal category then the forgetful functor @xmath168 has a left adjoint @xmath169 and is monadic .
the free pruned @xmath3-operad monad on the category of pruned @xmath3-collections in @xmath170 is finitary and cartesian .
by slightly abusing notation we will denote the free pruned @xmath3-operad monad as well as its functor part by @xmath171 * proof .
* the description of the required monad on a pruned collection @xmath8 is analogous to the description of the free @xmath3-operad functor .
it is given by an obvious inductive process .
let us call a pruned tree @xmath27 an _ admissible _ expression of arity @xmath41 we also have an admissible expression @xmath139 of arity @xmath153 if @xmath172 is a morphism of pruned trees and the admissible expressions @xmath173 of arities
@xmath174 respectively are already constructed then the expression @xmath175 is also an admissible expression of arity @xmath41 we also introduce an obvious equivalence relation on the set of admissible expressions generated by pairs of composable morphisms of pruned trees and by two equivalences @xmath176 generated by the identity morphism of @xmath27 and the unique morphism @xmath177 notice however , that there are morphisms of trees all of whose fibers ( after the pruning operation ) are equal to @xmath153 we can form an admissible expression @xmath178 corresponding to such a morphism but it is not equivalent to @xmath179 unless @xmath109 is equal to the identity .
everything else is in complete analogy with the case of free @xmath4-terminal @xmath3-operads @xcite .
@xmath67 applying the general theory of internal algebras @xcite we get [ ph ] the @xmath18-functor of internal pruned @xmath3-operads is representable by a pruned @xmath3-operad @xmath180 the object @xmath181 of the canonical internal operad in @xmath182 is the terminal object in this category .
the nerve of @xmath183 is obtained by a bar - construction on the terminal pruned @xmath3-operad .
a more explicit description of @xmath183 will be given later .
the functor of desymmetrisation for general @xmath4-terminal @xmath3-operads factorises through the category of pruned @xmath4-terminal @xmath3-operads @xmath184 where @xmath185 is defined by the formulas identical to the formulas for the desymmetrisation functor from @xcite .
if @xmath48 is cocomplete the inclusion @xmath186 has a left adjoint @xmath187 it follows that the symmetrisation functor can also be factorised @xmath188 notice that we use the same notations for pruned versions of symmetrisation and desymetrisation functors as we used for the unpruned case in @xcite .
we believe that this does not lead to confusion since @xmath189 is a full subcategory of @xmath190 moreover , we will use the same notation in the reduced case in section [ reducedoperads ] . again , as in the upruned case @xcite , we get the following commutative square of adjunctions : ( 60,26 ) ( 10,20)(0,0 ) ( 9,6)(0,1)10 ( 11,16)(0,-1)10 ( 45,20)(0,0 ) ( 42,6)(0,1)10 ( 44,16)(0,-1)10 ( 18,21)(1,0)14 ( 32,19)(-1,0)14 ( 22,22 ) ( 22,16.7 ) ( 68,10 ) ( 35,10 ) ( 10,3)(0,0 ) ( 45,3)(0,0 ) ( 19.5,4.5)(1,0)14 ( 33.5,2.2)(-1,0)14 ( 24,5.4 ) ( 24,0.1 ) ( 4,10 ) ( 12,10 ) ( 35,10 ) ( 45,10 ) [ squareone ] so , again , by the general theory of internal algebras , we have a representable @xmath18-functor of internal pruned @xmath3-operads on the @xmath18-category of symmetric @xmath191-operads @xcite .
we will denote by @xmath11 the symmetric categorical operad which represents this @xmath18-functor .
the value of the composite @xmath192 on a pruned @xmath3-collection @xmath8 is easy to describe .
in arity @xmath50 it consists of all labelled planar trees with label @xmath87 decorated by pruned @xmath3-trees and elements of @xmath193 so theorem 2.3 from @xcite provides us with the following description of the operad @xmath194 the objects of @xmath11 are labelled planar trees decorated by pruned @xmath3-trees .
analogously to the categorical operad @xmath13 in @xcite , the morphisms in @xmath11 are generated by simultaneous contractions of the input edges of a vertex provided there exists a corresponding morphism in @xmath167 we also can grow an internal edge by introducing a decoration by the linear tree @xmath195 ( see @xcite for the analogous description of @xmath13 ) .
the description we gave for the @xmath3-operad @xmath183 in corollary [ ph ] is not very revealing .
we are going to make it more accessible by using some more structured planar trees we will call _
recall from @xcite that the square [ prunedoperads].[squareone ] induces a map
@xmath196 this is not an injective transformation but it shows that to every object of @xmath183 ( an admissible expression ) we can naturally associate a planar tree decorated by @xmath3-ordinals
. we will call it the underlying tree of the admissible expression .
we define a notion of composable tree by induction .
we call a corolla decorated by an @xmath3-tree @xmath27 ( or @xmath139 ) a decorated tree _
composable to @xmath27_. suppose we have already defined a notion of composable decorated tree for which their underlying trees have length less than or equal to @xmath197 let @xmath144 be an equivalence class of admisssible expressions whose underlying decorated tree has length @xmath198 suppose , that its root vertex @xmath138 is decorated by a pruned tree @xmath199 we will call the tree _
composable _ to an @xmath3-tree @xmath27 if it is equipped with a morphism of @xmath3-trees @xmath106 with fibers @xmath200 and for each @xmath114 the @xmath114-th branch at @xmath138 is composable to @xmath201 the following lemma is an @xmath3-operadic analogue of the trivial fact that , given a string of elements in a monoid , there is a canonical way to calculate its value by performing multiplication always starting from the most right pair of elements .
the value , however , does not depend on the method of multiplication .
we equally could choose as canonical , multiplication from the left end of the string . in the @xmath3-operad case , however , the situation is more subtle as example [ tamarkintree].2 shows .
[ composable ] for every equivalence class of admissible expressions of arity @xmath202 there is a unique decorated tree composable to @xmath27 representing this expression .
* proof .
* the proof is by a routine induction .
@xmath67 so one can think of the objects of @xmath203 as trees composable to @xmath41 the morphisms are generated by composition of some nodes but in contrast with symmetric operads this operation can give a tree which lies outside @xmath204 [ tamarkintree ] the following tree is composable but we can not produce the same composition if we start to compose it from its root to its leaves .
this is actually the combinatorial ` raison d'tre of tamarkin s counterexample to cellularity of the getzler - jones operad .
we also formulate some results about @xmath205 which can be proved following verbatim the proofs of the parallel results for @xmath13 in @xcite : there is a natural isomorphism @xmath206 this isomorphism induces an isomorphism of nerves
@xmath207 [ formulap ] let @xmath5 be a cocomplete symmetric @xmath191-operad and @xmath111 be an internal pruned @xmath3-operad in @xmath208 then @xmath209 where @xmath210 is the operadic functor generated by the operad @xmath111
. finally , coming back to the adjunctions ( [ padj1]),([padj2 ] ) we find one more interesting categorical pruned @xmath3-operad in this picture , namely @xmath211 it comes together with a canonical morphism @xmath212 obviously , @xmath213 is a categorical pruned @xmath3-operad freely generated by an internal ( not pruned ! ) @xmath3-operad . from this characterisation
we deduce a description of @xmath214 the objects are elements of the free pruned @xmath3-operad on the following pruned @xmath3-collection : for a pruned tree @xmath27 it consists of all full injections
@xmath215 so , for any nonpruned @xmath3-tree @xmath179 we have an object , which we will denote by @xmath158 as well , in the category @xmath216 and these objects form an internal @xmath3-operad .
the generators for morphisms correspond to the morphisms of nonpruned trees and have the form @xmath217 for a morphism of @xmath3-trees @xmath218 notice , however , that this morphism generates a morphism between admissible expressions which corresponds to @xmath219 the operad @xmath220 is a contractible simplicial operad .
if this conjecture is true then we would be able to prove that the operad @xmath221 of @xcite is equivalent to the little @xmath3-disks operad .
all our efforts to prove this conjecture have failed so far because some morphisms in @xmath213 are going in opposite directions , which creates a lot of combinatorial difficulties in analysing the homotopy type of @xmath211
we will call a symmetric operad @xmath5 _ reduced _ if @xmath222 the unit for tensor product , and the operadic unit is given by the identity . equivalently
a reduced symmetric operad can be described as a contravariant functor @xmath5 from the subcategory of nonempty ordinals and injective morphisms of the category @xmath223 @xcite such that @xmath224 ) = i , $ ] ( _ a reduced symmetric collection _ ) plus operadic composition for any surjection of finite sets .
operadic composition has to be natural with respect to the injections .
the maps of reduced symmetric operads ( reduced symmetric collections ) are the maps of operads ( natural transformations ) which induce identity morphisms in arity @xmath225 and @xmath226 we use the notation @xmath227 for the category of reduced symmetric operads and @xmath228 for the category of reduced symmetric collections .
observe that our category of reduced operads is just a subcategory of the category of reduced operads of berger and moerdijk @xcite since they do not require @xmath229 to be a unit of @xmath230 we also have a category of reduced nonsymmetric collections which we will denote by @xmath231 these are contravariant functors on injective maps of @xmath232 with the conditions @xmath224 ) = i .$ ] recall from @xcite , that @xmath232 is the full subcategory of @xmath233 with nonempty ordinals as objects .
the @xmath3-operadic counterpart of @xmath227 will be the category @xmath234 of reduced @xmath4-terminal @xmath3-operads .
a pruned @xmath4-terminal @xmath3-operad @xmath5 is called reduced if @xmath235 and its unit is given by the identity .
a morphism between two reduced @xmath3-operads is an @xmath3-operadic morphism which induces identity morphisms in arity @xmath122 and @xmath153 let @xmath236 be the category of pruned nondegenerate @xmath3-trees and their injective morphisms .
a reduced @xmath3-collection is a contravariant functor @xmath5 from @xmath236 to @xmath237 such that @xmath238 the definition of the category of _ reduced @xmath3-collections _
@xmath239 is obvious . as for the reduced symmetric operads we have an equivalent characterisation of the reduced @xmath4-terminal operads .
a reduced @xmath4-terminal @xmath3-operad is given by a reduced @xmath3-collection @xmath5 together with a multiplication @xmath240 for every surjection of @xmath3-trees @xmath241 satisfying associativity and unitary axioms and naturality with respect to injections of pruned trees .
[ ron ] if @xmath48 is a cocomplete symmetric monoidal category then the forgetful functor @xmath242 has a left adjoint @xmath243 and is monadic .
the free reduced @xmath3-operad monad on the category of reduced @xmath3-collections in @xmath170 is finitary and cartesian .
the analogous statements hold for symmetric operads and reduced nonsymmetric collections .
* proof .
* everything goes through in full analogy with the unpruned case .
the only difference is that we use only surjections of pruned trees and we require some more identifications : @xmath244 this identification leads to the effect that we do not have an underlying @xmath3-planar tree with a vertex decorated by the linear tree @xmath245
@xmath67 we also have a reduced desymmetrisation functor from the category @xmath227 to the category @xmath234 which we again will denote by @xmath246 and from general considerations of @xcite , we have a commutative square of adjunctions ( 60,26 ) ( 10,20)(0,0 ) ( 9,6)(0,1)10 ( 11,16)(0,-1)10 ( 45,20)(0,0 ) ( 42,6)(0,1)10 ( 44,16)(0,-1)10 ( 19,21)(1,0)14 ( 33,19)(-1,0)14 ( 22,22 ) ( 22,16.7 ) ( 35,10 ) ( 10,3)(0,0 ) ( 45,3)(0,0 ) ( 57,2 ) ( 20.5,4.5)(1,0)14 ( 34.5,2.2)(-1,0)14 ( 24,5.4 ) ( 24,0.1 ) ( 1,10 ) ( 12,10 ) ( 35,10 ) ( 45,10 ) hence , we can develop the theory of internal reduced @xmath3-operads .
we denote by @xmath247 the reduced categorical @xmath3-operad freely generated by an internal reduced @xmath3-operad and by @xmath248 the symmetric categorical operad freely generated by an internal reduced @xmath3-operad .
an @xmath3-tree is called reduced if it is pruned , nondegenerate and is not equal to @xmath153 a planar tree decorated by @xmath3-ordinals is called reduced if all the decorations are reduced . as in the pruned case
, we have a natural transformation @xmath249 and have a description of the free reduced @xmath3-operad functor in terms of composable reduced planar trees .
there are , however , some changes in the definition of composable trees due to the fact that we have an identity @xmath250 again we define the composable trees by induction on the length of the underlying tree , but this time for @xmath251 we call a tree composable to @xmath27 when it is a corolla decorated by an @xmath3-tree @xmath158 together with a quasibijection @xmath252 suppose we have already defined the notion of composable decorated tree for which the underlying trees have length less than or equal to @xmath197 let @xmath144 be an equivalence class of admisssible expressions whose underlying decorated tree has length @xmath198 suppose that its root vertex @xmath138 is decorated by a pruned tree @xmath199 we will call the tree _
composable _ to an @xmath3-tree @xmath27 if it is equipped with a surjection of @xmath3-trees @xmath106 with fibers @xmath200 and for each @xmath114 the @xmath114-th branch at @xmath138 is composable to @xmath201 notice that a branch can be empty if the corresponding fiber of @xmath109 is the linear tree @xmath153 [ composabler ] for every equivalence class of admissible expressions of arity @xmath27 there is a unique reduced tree composable to @xmath27 representing this expression .
a nice result is the following characterisation of the free reduced @xmath3-operad functor .
[ submonadr ] the natural transformation @xmath253 is injective .
the image of the inclusion @xmath254 consists of labelled reduced trees decorated by @xmath3-ordinals which are dominated by the @xmath3-ordinal @xmath41 * proof .
* it is obvious that a labelled decorated tree from the image of @xmath253 is dominated by @xmath41 we are going to describe a procedure which reconstructs a unique composable reduced decorated tree from a labelled reduced planar @xmath3-tree dominated by @xmath202 so our theorem will follow from lemma [ composabler ] .
we construct a composable tree by induction .
let a labelled decorated tree have length equal to @xmath226 so , this is a labelled corolla decorated by an @xmath3-ordinal @xmath199 suppose we know that this tree is dominated by an @xmath3-ordinal @xmath41 the last property means that the identity map of labels can be extended to a quasibijection @xmath255 by corollary [ quasibijection ] .
so we have a composable tree @xmath256 of length @xmath1 .
suppose we have already constructed a composable decorated tree of length @xmath257 for every labelled reduced @xmath3-planar tree of length @xmath257 dominated by @xmath41 let @xmath144 be a labelled @xmath3-planar tree of length @xmath258 such that at the root @xmath259 we have an @xmath3-ordinal @xmath199 let us define a surjection @xmath260 as follows : @xmath261 the condition @xmath262 implies that if @xmath69 in @xmath27 then either @xmath263 or if they are not equal then @xmath264 for @xmath91 or @xmath265 for @xmath266 so @xmath108 is a surjection of ordinals .
let @xmath267 be the list of fibers of @xmath268 it is not hard to see that the @xmath114-th branch of @xmath144 at the vertex @xmath138 is dominated by the @xmath3-ordinal @xmath269 so we can proceed by induction and finish the proof .
@xmath67 we now give a description of @xmath270 the objects of @xmath248 are labelled planar trees decorated by reduced @xmath3-trees .
the morphisms are generated by simultaneous contractions of the input edges of a vertex provided there exists a corresponding surjection in @xmath167 from this description we immediately get the following the categories @xmath271 and @xmath272 are finite .
there are no morphisms for growing internal edge like in the @xmath13 and @xmath11 cases .
for this reason the operad @xmath248 is even a finite poset operad but we do not need this property here .
this is quite an important property , however , and we are going to consider it in a separate paper . similarly to the pruned case we have the following . the category @xmath272 is isomorphic to the comma - category of @xmath248 over its internal operad object @xmath273 the @xmath3-operad structure is given by the following convolution product : given a surjection of pruned trees @xmath166 and objects in comma - categories @xmath274 we define an object of the comma - category over @xmath181 by the composite : @xmath275 where @xmath276 is multiplication in @xmath248 and the last morphism is the structure morphism of the internal operad @xmath111 in @xmath270 [ symnerve ] there is a natural isomorphism @xmath277 this isomorphism induces an isomorphism of nerves @xmath278 we also want to introduce an unbased version of reduced operads ( we follow the terminology of @xcite ) .
these are reduced symmetric operads without nullary operations and with @xmath279 notice , that we do not require @xmath280 to be @xmath281 but simply forget about the @xmath225-arity of our operads .
this is , of course , a linguistic difference but it helps to express the results nicely .
an unbased reduced @xmath3-operad in a symmetric monoidal category @xmath48 consists of a collection of objects @xmath161 , one for every reduced @xmath3-tree @xmath202 and @xmath282 together with a multiplication @xmath240 for every surjection of @xmath3-trees @xmath241 satisfying the usual associativity axiom and the unitarity axiom with respect to the identity morphism @xmath283 we will denote the category of unbased reduced symmetric operads by @xmath284 and the category of unbased reduced @xmath3-operads by @xmath285 all the previous results about reduced operads can be carried over to the unbased case . in particular
, we have a categorical unbased reduced operad @xmath286 representing internal unbased @xmath3-operads and a categorical unbased reduced symmetric operad @xmath287 it is not hard to see , however , that there are canonical operadic maps @xmath288 and @xmath289 where @xmath290 is obtained from @xmath286 by adding @xmath281 in the arity @xmath291 ( @xmath292 is obtained from @xmath36 by adding @xmath281 in the arity @xmath293 ) and these maps are isomorphisms in nondegenerate arities . finally , we want to produce a version of the symmetrisation formula .
this formula for reduced operads admits a nice enhancement .
the opposite of the milgram poset @xmath294 can be considered as a subcategory of @xmath271 which consists of labelled planar trees with only one decoration .
[ cofinal ] for every @xmath295 the comma - category @xmath296 is nonempty and connected .
so @xmath294 is a final subcategory of @xmath271 @xcite .
* proof . * for @xmath45
the lemma is obviously true so we can assume that @xmath297 it is not hard to see that every object from @xmath271 is dominated by one of the objects of @xmath298 an object @xmath299 is a labelled planar tree decorated by @xmath3-ordinals and , hence , determines a canonical linear order on the set of its labels @xmath300 without loss of generality we can assume that this ordered set is the ordinal @xmath84 .$ ] so it is dominated by the @xmath3-ordinal @xmath301 ( see section [ comporders ] for notation ) . let @xmath302 be dominated by @xmath303 and @xmath304 again without losing generality we can assume that @xmath305 and @xmath303 can be obtained from @xmath306 by permuting the labels .
let us construct a totally @xmath3-ordered set @xmath27 which dominates @xmath307 and is dominated by both @xmath308 and @xmath304 to do this we apply the following reconstruction algorithm to @xmath309 let @xmath114 be the first label in @xmath303 with respect to the linear order @xmath310 and @xmath311 be the second
. if @xmath63 in @xmath84 $ ] then we define @xmath312 in @xmath313 if , however , @xmath314 in @xmath84 $ ] then we put @xmath315 in @xmath313 we also put all the other labels in @xmath316 to be in the same order as in @xmath317 so we have constructed an object of @xmath294 which is dominated by @xmath303 and also dominates @xmath318 now we continue this process and do the same thing for the second and third consecutive labels in @xmath303 then for the third and fourth and so on . in this way we construct a sequence of @xmath3-ordered sets @xmath319 finally , we observe , that @xmath63 implies @xmath312 or @xmath315 or @xmath320 in @xmath27 by construction .
this means that @xmath321 and we have finished the proof .
@xmath67 [ formular ] let @xmath5 be a cocomplete reduced symmetric @xmath191-operad and @xmath111 be an internal reduced @xmath3-operad in @xmath208 then @xmath322 where @xmath323 is the operadic functor generated by @xmath111 .
the analogous formula holds in the unbased case . to be able to apply this theorem to the reduced operads in a symmetric monoidal category @xmath324 we have to exhibit @xmath48 as a categorical reduced symmetric operad similar to what was done in @xcite for the unreduced case .
our construction @xmath325 from @xcite produces only an unreduced operad .
in the reduced case we define the reduced categorical symmetric operad @xmath326 as follows @xmath327 with multiplication defined by iterated tensor product and trivial symmetric group action .
it is obvious that there is an isomorphism of the categories of internal reduced operads in @xmath326 and reduced operads in @xmath48 ( see @xcite for this property in the unreduced case ) .
a comment has to be made about endomorphism operads in the reduced situation .
first of all we have to consider only the pointed case ; i.e. we consider an object @xmath8 of our closed symmetric monoidal category @xmath48 together with a fixed morphism @xmath328 for every @xmath329 one can consider the following pullback ( 60,26 ) ( 8,20)(0,0 ) ( 8,16)(0,-1)10 ( 40,20)(0,0 ) ( 40,16)(0,-1)10 ( 17,20)(1,0)14 ( 22,22 ) ( 68,10 ) ( 35,10 ) ( 8,3)(0,0 ) ( 40,3)(0,0 ) ( 17,3)(1,0)14 ( 24,5.4 ) ( 13,10 ) ( 40,10 ) this symmetric collection has an obvious structure of an operad and we define _ the reduced endomorphism symmetric operad _ @xmath330 as follows : @xmath331 and @xmath332 an action of a reduced operad @xmath5 on a pointed object @xmath8 is then an operadic map @xmath333 usually an action of an operad on a pointed object is defined as an operad map from @xmath5 to its full endomorphism operad .
it is obvious , however , that such an action can be factorised through @xmath334 so our definition agrees with the usual one .
we give the analogous obvious definition for the reduced endomorphism @xmath3-operad @xmath335 now , we will have a canonical isomorphism @xmath336 as in the unreduced case . and ,
hence , similarly to the unreduced case we have for a reduced @xmath4-terminal @xmath3-operad @xmath337 the categories of @xmath6-algebras and @xmath5-algebras are isomorphic .
here we adapt the theory of @xcite to the case of @xmath3-operads .
recall that one of the main technical points of this theory is the existence of a cofibrantly generated model structure on the category of reduced symmetric operads in a monoidal model category @xmath230 this structure is transferred along the free symmetric operad functor from a model structure on symmetric collections in @xmath237 which in its turn is transferred along the free symmetric collection functor from the category of nonsymmetric collections .
the term transferred means that the weak equivalences ( fibrations ) are precisely the morphisms which become weak equivalences ( fibrations ) after application of the right adjoint functor .
[ bm1 ] there exists a transferred model structure on the category of reduced symmetric operads in a monoidal model category @xmath237 if * @xmath48 is cofibrantly generated and its unit @xmath56 is cofibrant ; * the comma category @xmath338 has a symmetric monoidal fibrant replacement functor ; * @xmath48 admits a commutative hopf interval .
[ bm2 ] if @xmath48 is a cartesian closed model category then there is a transferred model structure on the category of all symmetric operads in @xmath48 provided * @xmath48 is cofibrantly generated and the terminal object of @xmath48 is cofibrant ; * @xmath48 has a symmetric monoidal fibrant replacement functor .
we already pointed out that our notion of reduced symmetric operad is stronger than that of @xcite .
it is , however , not too hard to check that their proof works well in our situation .
we only have to take care of their construction of a cooperad @xmath339 from a commutative hopf object @xmath340 ( * ? ? ?
* proposition 1.1 ) .
if we put @xmath341 then @xmath342 is still a cooperad and it works for the reduced operads in our sense like @xmath339 works in @xcite .
so theorem [ bm1 ] holds in our case without any changes .
moreover , following the method of @xcite in the case of @xmath3-operads we can prove : [ bmn1 ] there exists a transferred model structure on the category of reduced @xmath4-terminal @xmath3-operads in a monoidal model category @xmath237 if * @xmath48 is cofibrantly generated and its unit @xmath56 is cofibrant ; * the comma category @xmath338 has a symmetric monoidal fibrant replacement functor ; * @xmath48 admits a commutative hopf interval .
moreover , the commutative square [ radjsq ] is a square of quillen adjunctions .
the analogous theorem holds in the unbased case .
[ bmn2 ] if @xmath48 is a cartesian closed model category then there is a transferred model structure on the categories of @xmath4-terminal @xmath3-operads in @xmath48 and pruned @xmath4-terminal @xmath3-operads in @xmath48 provided * @xmath48 is cofibrantly generated and the terminal object of @xmath48 is cofibrant ; * @xmath48 has a symmetric monoidal fibrant replacement functor .
the corresponding commutative squares of adjunctions are squares of quillen adjunctions .
examples of monoidal categories satisfying the conditions of these theorems are given in @xcite .
the most important for us are the categories of simplicial sets , the category @xmath343 of compactly generated topological spaces , and the category @xmath344 of chain complexes over a commutative ring @xmath345 with unit .
we note that the category of chain complexes satisfies the hypothesis of theorem [ bm1 ] and the categories of simplicial sets and topological spaces satisfy the asssumptions of theorem [ bm2 ] .
the weak equivalences ( fibrations ) between topological or simplicial @xmath3-operads are therefore operadic maps which are termwise weak equivalences ( fibrations ) in simplicial sets and @xmath346 the weak equivalences ( fibrations ) between chain @xmath3-operads are those operadic maps which are termwise quasi - isomorphisms ( epimorphisms ) .
since many of our categorical operads are given by a bar construction we would like to investigate now what are the homotopy theoretic properties of the bar - construction for @xmath3-operads in general .
we will do it for the general case of @xmath4-terminal @xmath3-operads .
the pruned and reduced cases are similar .
in addition , suppose @xmath48 is enriched in simplicial sets with simplicial @xmath347-functor @xmath348
satisfying @xmath349 where @xmath350 is the internal @xmath347-functor in @xmath230 then the categories of @xmath3-collections and @xmath3-operads become simplicially enriched . for two @xmath4-terminal @xmath3-collections
@xmath351 we define its simplicial set of morphisms as @xmath352 as in @xcite , let @xmath353 be the free @xmath4-terminal @xmath3-operad monad . then for two @xmath3-operads
@xmath354 we define their simplicial set of morphisms as the equalizer = 0,8 mm ( 90,10)(0,15 ) ( 78,20)(0,0 ) ( 38,20)(0,0 ) ( 53,20.8)(1,0)6 ( 53,19.2)(1,0)6 ( 2,20)(0,0 ) ( 17,20)(1,0)6 with the obvious horizontal morphisms generated by the operadic structures of @xmath5 and @xmath355 . if @xmath48 has tensors and cotensors with respect to simplicial sets then the categories @xmath356 and @xmath357 also have them . in particular , one can speak about the total object for a simplicial @xmath3-operad ( @xmath3-collection ) @xmath358 . by definition
this is the coend @xmath359\in\delta^{+ } } a^{n}\otimes \delta^{n } , \ ] ] where @xmath360,-)$ ] is the standard simplex of dimension @xmath3 .
let @xmath8 be an @xmath4-terminal @xmath3-operad in @xmath361 .
then the bar - construction @xmath362 of @xmath8 is the total object of the simplicial operad ( 60,12 ) ( 19,5)(0,0 ) ( 35,5.5)(-1,0)7 ( 35,4.5)(-1,0)7 ( 36,6.5 ) ( 45,5)(0,0 ) ( 71.5,5)(0,0 ) ( 61.5,5.8)(-1,0)7 ( 61.5,5)(-1,0)7 ( 61.5,4.2)(-1,0)7 ( 83,5 ) ( 54,7 ) let @xmath48 be a model category which satisfies the conditions of theorem [ bmn2 ] , has a simplicial enrichment @xmath363 satisfying [ simplenrich ] , and which is a simplicial model category with respect to these structures @xcite .
let @xmath8 be an @xmath4-terminal @xmath3-operad in @xmath48 with cofibrant underlying @xmath3-collection .
then the canonical operad morphism @xmath364 is a cofibrant replacement for @xmath8 in the model category of @xmath4-terminal @xmath3-operads .
the analogous theorem holds in the pruned case and in the reduced and unbased reduced case if @xmath48 satisfies the assumptions of theorem [ bmn1 ] . * proof . *
it follows from the general properties of bar - construction @xcite that the morphism @xmath365 is a deformation retraction in @xmath366 .
hence , @xmath365 is a trivial fibration of operads .
it remains to prove that @xmath367 is a cofibrant @xmath3-operad .
let @xmath368 be a trivial fibration of @xmath3-operads .
we have to show that any operadic map @xmath369 can be lifted to @xmath370 ( 60,30 ) ( 69,7)(0,0 ) ( 69,22)(0,-1)12 ( 69,25)(0,0 ) ( 55,7)(1,0)10 ( 71,15 ) ( 35,7)(0,0 ) ( 48,10)(3,2)5 ( 54,14)(3,2)5 ( 60,18)(3,2)5 by construction this amounts to the following lifting problem in the category of cosimplicial spaces ( 60,30 ) ( 72,7)(0,0 ) ( 72,22)(0,-1)12 ( 72,25)(0,0 ) ( 43,7)(1,0)12 ( 74,15 ) ( 40,7)(0,0 ) ( 43,10)(3,2)5 ( 49,14)(3,2)5 ( 55,18)(3,2)5 where @xmath371 is the cosimplicial simplicial set consisting of standard simplices . since @xmath371 is cofibrant in the reedy model structure @xcite it remains to show that @xmath372 is a trivial fibration .
we follow a method developed in @xcite .
we have to prove that in the diagram ( 100,45)(-10,0 ) ( 72,7)(0,0 ) ( 72,22)(0,-1)12 ( 72,25)(0,0 ) ( 41,7)(1,0)11 ( 74,15 ) ( 8,35)(1,-2)12.5 ( 22,35)(4,-1)30 ( 14,35)(1,-1)7 ( 25,7)(0,0 ) ( 12,38)(0,0 ) ( 25,25)(0,0 ) ( 25,22)(0,-1)12 ( 30,25)(1,0)22 ( 30.5,20)(-1,0)4 ( 30.5,20)(0,1)3.5 ( 22,31)(0,0 ) the canonical map @xmath373 to the pullback is a trivial fibration . in this diagram
@xmath374 is the @xmath114-th matching object of the corresponding cosimplicial object @xcite .
according to lemma 2.3 from @xcite the diagram above is isomorphic to the diagram ( 100,45)(-10,0 ) ( 71,7)(0,0 ) ( 71,22)(0,-1)12 ( 71,25)(0,0 ) ( 40,7)(1,0)10 ( 73,15 ) ( 8,35)(1,-2)12.5 ( 22,35)(4,-1)28 ( 14,35)(1,-1)7 ( 25,7)(0,0 ) ( 12,38)(0,0 ) ( 25,25)(0,0 ) ( 25,22)(0,-1)12 ( 28,25)(1,0)22 ( 30.5,20)(-1,0)4 ( 30.5,20)(0,1)3.5 ( 22,31)(0,0 ) ( 40,33)(0,0 ) ( 44,9)(0,0 ) here , @xmath375 is the latching object @xcite for the augmented cosimplicial object @xmath376 in @xmath377 and @xmath378 are generated by the canonical morphism @xmath379 if we show that this morphism is a cofibration , then @xmath373 will be a trivial fibration by the axioms for a simplicial model category .
we will actually prove that @xmath380 is an isomorphism onto a summand .
it was proved in @xcite[theorem 9.1 ] that the @xmath50-th iteration of the functor @xmath381 is given by the following formula : @xmath382 where @xmath383 are morphisms in @xmath384 the coface operators in @xmath376 are canonical inclusions on the summands corresponding to the operators of insertion of the identities to the chain @xmath385 the rest of the proof follows in complete analogy with lemma 4.1 of @xcite .
the proof in the reduced and unbased cases are analogous .
notice that , in our version of the category of reduced operads , any reduced collection is automatically well pointed in the sense of @xcite so we do not include this condition in the formulation of the reduced version of our theorem .
@xmath67 the simplicial @xmath3-operads @xmath386 are cofibrant @xmath3-operads in the categories of @xmath4-terminal simplicial @xmath3-operads , @xmath4-terminal simplicial pruned @xmath3-operads and unbased reduced @xmath4-terminal simplicial pruned @xmath3-operads respectively .
the simplicial symmetric operads @xmath387 are cofibrant simplicial symmetric operads .
the unbased reduced symmetric operad @xmath34 is cofibrant in the category of unbased reduced symmetric operads .
the same theorem is true for the geometric realisations of these operads and for the reduced operad of chain complexes of @xmath286 and @xmath287 * proof . *
the nerves of the above @xmath3-operads are bar - constructions on the terminal @xmath3-collection in the corresponding categories . in these categories
the terminal @xmath3-collection is obviously a cofibrant collection .
the cofibrantness of @xmath387 and @xmath34 follows from the corresponding quillen adjunctions between @xmath3-operads and symmetric operads and the fact that in these cases the symmetrisation functor commutes with nerves .
the topological and chain versions of the theorem follow from the general considerations of @xcite .
@xmath67 the nerves of @xmath247 and @xmath248 are not cofibrant .
the reason is that @xmath1 is not a cofibrant reduced @xmath3-collection .
the operadic structure on compactified moduli space of configurations of points in @xmath388 was first observed by getzler and jones in @xcite .
here we use an explicit approach of @xcite to describe this compactification .
let @xmath389}^{\ \ !
n}$}}$ ] be the quotient of the configuration space @xmath390 with respect to the obvious action of the @xmath391-dimensional lie group @xmath392 of affine transformations of the form @xmath393 where @xmath394 is a real number and @xmath140 is a vector from @xmath395 the functions @xmath396 @xmath397 embed @xmath389}^{\ \ !
n}$}}$ ] into a compact space @xmath398 = \prod_{\stackrel{\scriptstyle 1\le i , j \le
k}{i\ne j } } s^{n-1}\times \prod_{\stackrel{\scriptstyle 1\le i , j , l\le k } { i\ne j , j\ne l , i \ne l } } [ 0,\infty ] .\ ] ] the @xmath50-th space of the fulton - macpherson operad @xmath20 is obtained as the closure of the moduli space @xmath389}^{\ \ !
n}$}}$ ] inside @xmath399.$ ] notice , that we use here a reduced version of @xmath20 meaning that we put @xmath400 if we forget about nullary operations we will get the unbased version of the fulton - macpherson operad which we will denote by @xmath401 notice also that in @xcite the fulton - macpherson operad means @xmath401 in @xcite the following properties of @xmath20 were established * @xmath20 is a reduced symmetric operad weakly equivalent to the little @xmath3-cube operad ; * @xmath402 is an unbased reduced cofibrant operad weakly equivalent to the unbased little @xmath3-cube operad ; * set theoretically @xmath20 is the free reduced operad on the reduced symmetric collection @xmath403}^{\ \ !
n}$ } } ; $ ] * @xmath404 is a manifold with corners ; * @xmath405 is isomorphic to the stasheff operad of associahedra . later in this paper
we will use notations @xmath406 and @xmath407 for the coordinates of points in @xmath30 the following lemma describes the behaviour of these coordinates under operadic multiplication . [ multformula ] let @xmath408\rightarrow [ k]$ ] be a surjection in @xmath409 then @xmath410 where @xmath411 are images of @xmath412 in the fiber of @xmath109 over @xmath197 similarly @xmath413 all other values of @xmath414 can be deduced from the above table and the following relations between @xmath415 @xcite : @xmath416 * proof . *
the proof can be obtained using the explicit formulas for the operadic multiplication in @xmath20 from @xcite or techniques from @xcite .
@xmath67 following joyal @xcite we give a definition of a generalised @xmath3-tree .
a generalised @xmath3-tree @xmath8 is a chain of partially ordered sets and functions @xmath417 such that the induced order on @xmath418 is linear for all @xmath419 and @xmath420 a topological @xmath3-tree is an @xmath3-tree for which all @xmath421 are endowed with a topology and all @xmath422 are continuous functions .
the definition of a morphism of generalised @xmath3-trees ( also from @xcite ) coincides verbatim with the definition of morphism of finite @xmath3-trees in @xcite .
consider the following topological @xmath3-tree : @xmath423 @xmath424 where morphisms are projections on the first coordinates .
we introduce the @xmath3-tree structure on @xmath423 by ordering the fiber of each projection according to its natural order .
now , for every pruned @xmath3-tree @xmath202 one can consider the space of all injective @xmath3-tree morphisms from @xmath27 to @xmath425 this space is the classical _ fox - neuwirth cell _ @xmath426 corresponding to @xmath41 we will consider it as an open submanifold of configurations of points with the labelling prescribed by the order in @xmath427 @xcite .
the group @xmath392 obviously acts on @xmath426 and we call the corresponding quotient space _ a getzler - jones cell _ @xmath428 recall that the functor @xmath429 has a left adjoint @xmath430 } = \coprod_{|t|=[k ] } b_t .\ ] ] the configuration space @xmath133 admits the classical fox - neuwirth decomposition @xmath431 , \pi \in \sigma_k } \pi { \mbox{$\mathrm{fn}$}}_t.\ ] ]
this means that there is a bijective continuous map of collections @xmath432 where @xmath433 is the functor of symmetrisation of reduced nonsymmetric collections .
the inverse bijection is not continuous .
certainly @xmath434 is equivariant with respect to the action of @xmath392 and so we have a stratification of the moduli space of configurations @xmath435}^{\ \ !
n}$}}= \bigcup_{|t|=[k ] , \pi \in \sigma_k } \pi { \mbox{$\mathrm{mod}_{{t}}^{\ \ ! n}$}}\ ] ] and a corresponding continuous bijection which we will denote by the same letter @xmath436}^{\ \ !
n}$ } } .\ ] ] the free reduced symmetric operad functor @xmath437 is factorized as @xmath438 where @xmath439 is the free reduced symmetric operad functor on reduced symmetric collections .
so we have a bijective continuous map of operads @xmath440}^{\ \ !
n}$ } } ) \rightarrow { \mbox{$\bf fm$}}^n\ ] ] we would like to get a description of @xmath441 in terms of the functions @xmath442 let @xmath443 denote the open @xmath444-hemisphere in @xmath388 , @xmath445 : @xmath446{ll } x_1 ^ 2+\ldots+x_n^2 = 1 & \\
x_{p+1 } > 0 \ \mbox{and } \ x_i = 0 \
\mbox{if } \
1\le i \le p & \end{array } \right .
\right\}\ ] ] and let @xmath447{ll } x_1 ^ 2+\ldots+x_n^2 = 1 & \\
x_{p+1 } < 0 \ \mbox{and } \ x_i = 0 \ \mbox{if } \
1\le i \le p & \end{array } \right .
\right\ } \raisebox{-3 mm } { \ .}\ ] ] the closure of @xmath443 will be denoted @xmath448 and the closure of @xmath449 will be denoted @xmath450 observe , that @xmath451 [ gjcellt ] for a pruned @xmath3-tree @xmath452 the getzler - jones cell @xmath441 is equal to the set @xmath453}^{\ \ !
n}$ } } \left | \ \begin{array}[c]{ll } \
u_{ij}(x ) \in \stackrel{\scriptscriptstyle o \ \ \ \ \ \ \ } { s^{n - p-1}_{+ } } & \ \mbox{\rm if } \ i<_p j \ \mbox{\rm in } \
t \\ & \\ \
u_{ij}(x ) \in \stackrel{\scriptscriptstyle o \ \ \ \ \ \ \ } { s^{n - p-1}_{- } } & \ \mbox{\rm if } \ j<_p i \ \mbox{\rm in } \
t \end{array}\right .
\right\ } \raisebox{-7 mm } { \ .}\ ] ] it is a contractible open manifold of dimension @xmath454 where @xmath40 is the number of edges in the tree @xmath41 * proof .
* obvious from the definition of @xmath441 .
@xmath67 the dimension of an @xmath3-pruned tree @xmath455 is the integer @xmath456 we also put @xmath457 in virtue of ( [ gjdecomp ] ) the decomposition ( [ moddecomp ] ) can be extended to a decomposition of @xmath20 @xcite .
following @xcite we will call it _ the getzler - jones decomposition_. the cells of this decomposition are indexed by labelled reduced planar trees i.e. by the objects of @xmath458 with vertices decorated by points of getzler - jones cells .
we will call such a space indexed by an object @xmath459 _ a generalised getzler - jones cell _ and will denote it @xmath460 since the generalised getzler - jones cells do not intersect each other we have a correctly defined map @xmath461 [ gjtau ] let @xmath462 then @xmath463 if @xmath69 in @xmath464 or @xmath465 if @xmath466 in @xmath467 * proof . *
the proof is easily obtained by induction on the length of @xmath144 and lemmas [ multformula ] , [ gjcellt ] . @xmath67 the getzler - jones decomposition is not a cellular decomposition or stratification of @xmath23 since the boundary of the closure of a getzler - jones cell may not be equal to the union of low dimensional getzler - jones cells as was first observed by tamarkin ( see @xcite for a description of tamarkin s counterexample ) .
however , the following is true : [ fnclosure ] the closure of the getzler - jones cell @xmath468 is a manifold with corners homeomorphic to a ball of dimension @xmath469 * proof .
* we have to use the original description of @xmath470 in terms of the iterated blow - up of @xmath471 along its fat diagonal @xcite .
consider the closure of the getzler - jones cell @xmath441 in @xmath472 the intersection of this subspace with each of the diagonals is a manifold , so restricting the fulton - macpherson blow up procedure we get a manifold with corners homeomorphic to a ball since the blow up does not change the topological type of the manifold .
this manifold is homeomorphic to the closure of @xmath473 in @xmath30 @xmath67
let us take @xmath474 to be the free reduced @xmath3-operad generated by the reduced @xmath3-collection @xmath475 .
then we have a canonical inclusion of @xmath3-operads @xmath476 consider the composite @xmath477 @xmath478 @xmath479}^{\ \ !
n}$}}))\longrightarrow rdes_n({\mbox{$\bf fm$}}^n)\ ] ] which is an injective continuous map of @xmath3-operads .
the getzler - jones @xmath3-operad @xmath22 is the image of @xmath480 this is a reduced @xmath4-terminal @xmath3-operad .
if we forget about operations with degenerate arity , we obtain _ the unbased getzler - jones @xmath3-operad _
@xmath33 let @xmath455 be a nondegenerate pruned @xmath3-tree .
the following topological spaces are equal : * @xmath481 * @xmath482 * kontsevich - soibelman space @xcite @xmath483{ll } \
u_{ij}(x ) \in { s^{n - p-1}_{+ } } & \ \mbox{\rm if } \
i<_p j \ \mbox{\rm in } \
t \\ \ u_{ij}(x ) \in { s^{n - p-1}_{- } } & \
\mbox{\rm if } \ j<_p i \ \mbox{\rm in } \
t \end{array } \right .
\right\ } \raisebox{-3mm}{\ .}\ ] ] * proof .
* the equality @xmath484 readily follows from the definitions and theorem [ submonadr ] .
let us prove that @xmath485 the inclusion @xmath486 is obvious from lemma [ gjtau ] .
let us prove the inverse inclusion .
let @xmath487 we have to prove that @xmath488 let @xmath69 in @xmath467 then , by lemma [ gjtau ] , again @xmath489 since @xmath490 two possibilities exist . either @xmath77 in @xmath27 for some @xmath491 and then @xmath492 or @xmath493 in @xmath27 for some @xmath494 and @xmath495 in the first case we have @xmath496 which is possible only if @xmath497 in the second case @xmath498 and this is possible only if @xmath499 so @xmath144 is dominated by @xmath41 @xmath67 the space @xmath29 is a closed subspace of @xmath500 now we will study the properties of @xmath22 and its cousin @xmath33 [ gjcon ] the operad @xmath22 is contractible . * proof . * for @xmath24 kontsevich and soibelman prove contractibility of @xmath25 in @xcite[proposition 7 ] .
unfortunately the details are left to the reader .
so we will follow their idea and prove it for arbitrary @xmath3 ( but only for @xmath3-ordinals , not arbitrary complimentary orders as in @xcite ) . as in @xcite
we will use induction on the number of tips of a tree @xmath41 if @xmath27 has only two tips then obviously the space @xmath25 is @xmath448 , where @xmath73 is the index of the unique nonempty order on @xmath501 suppose we have proved this theorem for all @xmath27 with @xmath502 .$ ] let @xmath27 be a tree with @xmath503 , $ ] and let @xmath303 be an @xmath3-tree which is obtained from @xmath27 by cutting off the most right branch .
then we have an injection @xmath504 which induces a map @xmath505 let @xmath506 be a point .
let us suppose that @xmath111 belongs to the cell @xmath507 we will show that the fiber of @xmath508 is homeomorphic to a disk of dimension @xmath509 here , @xmath73 is such that @xmath510 in @xmath41 since the fiber obviously depends only on configurations of points the labels of which belong to the most right branch of the tree @xmath511 then without loss of generality we can assume that @xmath303 is a suspension of some @xmath4-tree and @xmath512 in @xmath27 .
the manifold @xmath513 is diffeomorphic to the intersection of the space of configurations of points @xmath514 which belong to the hyperplane @xmath515 with @xmath516 and @xmath517 with the fox - neuwirth cell @xmath518 let @xmath519 be the image of @xmath111 in this space . in its turn
@xmath441 is diffeomorphic to the space of configurations of points @xmath520 in @xmath388 such that @xmath521 belongs to the previous intersection and @xmath522 is in the open positive halfspace @xmath523 let @xmath491 be a sufficiently small number such that the closed balls @xmath524 of radius @xmath491 with the centre @xmath525 do not intersect each other . and
let @xmath345 be sufficiently big such that the union of all @xmath524 belongs to the interior of the ball @xmath526 let @xmath527 be the manifold with corners @xmath528 we claim , that there is a homeomorphism from @xmath527 to the fiber of @xmath508 over @xmath529 to construct such a homeomorphism we first choose an @xmath530 and @xmath531 such that the @xmath532 still do not interstect and their union is still in the interior of the ball @xmath533 for a fixed @xmath534 let @xmath535 for @xmath536 such that @xmath537 and let @xmath538 for @xmath536 such that @xmath539 we can construct a map @xmath540 from @xmath527 to the positive halfspace @xmath541 with the following properties : * the restriction of @xmath540 on @xmath542 is a homeomorphism of manifolds with corners ; * @xmath543 if @xmath544 * @xmath545 if @xmath539 now we construct a map @xmath546 from @xmath527 to @xmath547 to be equal at a point @xmath548 to @xmath549 where @xmath550 is an arbitrary sequence from the interior of @xmath527 which converges to @xmath551 obviously , @xmath546 is correctly defined and continuous .
it also maps injectively the interior of @xmath527 to the intersection of @xmath547 with @xmath552 points from the boundaries of @xmath553 are mapped injectively to the points of intersections of @xmath547 and @xmath554 where @xmath555 is the getzler - jones cell corresponding to a map of trees @xmath556 , @xmath557 this intersection is homeomorphic to the hemisphere @xmath558 points from the hyperplane @xmath516 are mapped under @xmath546 to the points of the getzler - jones cell corresponding to the configuration @xmath559 and , finally , points from the outer hemisphere boundary of @xmath527 are mapped to the getzler - jones cell corresponding to the map of trees @xmath560 , @xmath561 if @xmath562 and @xmath563 it is not difficult to check that @xmath546 is bijective and so it is a homeomorphism since both spaces are compact .
suppose now @xmath111 belongs to a generalized getzler - jones cell in @xmath564 then the fiber @xmath547 can be glued from the manifolds @xmath565 by the following inductive procedure .
first we construct @xmath566 where @xmath567 is the projection of @xmath111 to the configuration which decorates the root vertex @xmath138 of @xmath568 then we consider the vertices @xmath569 which can be connected to @xmath138 by exactly one edge .
we then construct @xmath570 where @xmath571 is the corresponding projection of @xmath529 by scaling up or down the configurations @xmath571 if necessary we always can make @xmath572 so we construct the next manifold by gluing the outer hemisphere @xmath573 to the inner hemisphere of @xmath566 in the place @xmath197 the following example illustrates the proof . here
[ 3 ] \rightarrow [ 1]\ ] ] @xmath575 and the getzler - jones cell corresponds to the map @xmath576 of @xmath18-trees @xmath577\rightarrow [ 1]\rightarrow [ 1 ] \ , \sigma(1 )
= \sigma(2 ) = 1 \ , \ \sigma(3 ) = \sigma(4 ) = 2 .\ ] ] obviously , this inductive procedure stops after a finite number of steps and produces a contractible manifold .
@xmath67 [ gjcof ] the operad @xmath578 is a cellular object in @xmath579 in particular , it is cofibrant .
* proof .
* first recall some terminology @xcite .
we will work in the category of unbased reduced @xmath3-operads and unbased reduced @xmath3-collections and we will call them simply @xmath3-operads and @xmath3-collections .
let @xmath580 .
let @xmath581 be the forgetful functor with left adjoint @xmath582 let @xmath583 be a cofibration of @xmath3-collections . _
the cellular extension @xmath584 $ ] generated by @xmath585 _ is the following pushout in @xmath586 ( 60,33 ) ( 43,25)(0,0 ) ( 43,21)(0,-1)10 ( 53,26 ) ( 77,25)(0,0 ) ( 77,21)(0,-1)10 ( 53,25)(1,0)14 ( 68,15 ) ( 41,15 ) ( 43,8)(0,0 ) ( 77,8)(0,0 ) ( 49,8)(1,0)22 ( 55,8 ) where the left vertical map is the counit of the adjunction .
we will prove that @xmath31 can be obtained as a sequential colimit of cellular extensions starting from the initial @xmath3-operad .
consider a filtration of the @xmath3-collection @xmath475 by the following subcollections @xmath587 @xmath588 we have an inclusion for every @xmath276 @xmath589 the image @xmath590 of this inclusion is closed and , hence , a compact suboperad of @xmath33 moreover , @xmath591 we want to show that @xmath592 is a cellular extension generated by a cofibration . indeed , let us consider the following @xmath3-collection @xmath593 @xmath594 for every @xmath3-tree @xmath202 @xmath595 is a closed subspace of @xmath596 moreover , if @xmath597 then proposition [ fnclosure ] implies that there exists a homeomorphism @xmath598 such that we have the following pushout in @xmath599 ( 60,33 ) ( 43,24)(0,0 ) ( 40,20)(0,-1)8 ( 53,26 ) ( 79,25)(0,0 ) ( 79,21)(0,-1)10 ( 56,25)(1,0)9 ( 68,15 ) ( 41,15 ) ( 43,6)(0,0 ) ( 79,7)(0,0 ) ( 56,7)(1,0)13 ( 55,8 ) hence , the inclusion @xmath600 is a cofibration in the category of unbased reduced @xmath3-collections .
besides that , we have a map of @xmath3-operads @xmath601 generated by the obvious map of collections @xmath602 we also can construct another @xmath3-collection @xmath603 with a bijective continuous map of @xmath3-collections @xmath604 a cofibration @xmath605 and a map @xmath606 the cofibration [ coprodl ] is actually a canonical coprojection into a coproduct .
all these maps of operads can be organised into a commutative cube .
( 60,55)(-20,-4 ) ( 0,25)(0,0 ) ( 10,22)(0,-1)14 ( 12,15 ) ( 24,25)(1,0)15 ( 23,26 ) ( 53,25)(0,0 ) ( 51,22)(0,-1)14 ( 57,21 ) ( 57,28 ) ( 5,5)(0,0 ) ( 19,5)(1,0)14 ( 23,6 ) ( 51,5)(0,0 ) ( 57,1 ) ( 57,8 ) ( -3,3 ) ( 50,30 ) ( 25,35)(0,0 ) ( 30,32)(0,-1)9.5 ( 30,21.2)(0,-1)3 ( 32,25 ) ( 42,35)(1,0)15 ( 43,36 ) ( 71,35)(0,0 ) ( 72,32)(0,-1)14 ( 77,31 ) ( 77,38 ) ( 30,15)(0,0 ) ( 55,15)(1,0)4.4 ( 40,15)(1,0)12.4 ( 43,16 ) ( 72,15)(0,0 ) ( 77,11 ) ( 77,18 ) ( 13,28)(1,1)7 ( 54,28)(1,1)7 ( 13,8)(1,1)7 ( 54,8)(1,1)7 the front square of this cube is a pushout in the category of @xmath3-operads because @xmath607 is a coprojection and the free operad functor preserves colimits .
all maps from the front square to the back square are continuous bijections .
so the induced morphism from @xmath608 to the pushout @xmath609 of the back square must be a continuous bijection as well , and , hence , the induced map @xmath610 is a continuous bijection .
moreover , the vertical map @xmath611 admits a section .
it follows that @xmath612 is epi .
since @xmath613 is a finite coproduct of compact spaces the space @xmath614 is compact .
therefore , @xmath615 is an isomorphism and we have proved our theorem .
@xmath67 by definition we have an inclusion @xmath616 and , by adjunction , a canonical morphism @xmath617 [ symgj ] this canonical map from @xmath618 to @xmath20 is an isomorphism .
the analogous result holds in the unbased case .
* proof .
* the canonical map @xmath619 is a continuous bijection .
hence , after application of @xmath620 to this map we have a composite of continuous bijections @xmath621 we also have the following commutative diagram ( 60,33 ) ( 15,25)(0,0 ) ( 15,21)(0,-1)10 ( 55,26 ) ( 80,25)(0,0 ) ( 80,21)(0,-1)10 ( 37,25)(1,0)20 ( 68,15 ) ( 42,15 ) ( 15,8)(0,0 ) ( 80,8)(0,0 ) ( 30,8)(1,0)10 ( 48,8)(0,0 ) ( 65,8)(-1,0)10 ( 55,8 ) where the composite of the left vertical and horizontal maps is a continuous bijection by [ gjdecomp ] .
hence , the canonical map from @xmath618 to @xmath20 is a continuous bijection , as well . by formula [ formular ]
the space @xmath622 is a finite colimit of compact spaces , hence , it is compact and therefore the bijection between @xmath618 and @xmath20 is a homeomorphism .
we finally are able to consider some applications of the results obtained .
[ fmcof ] the operad @xmath402 is a cellular object in the category of unbased reduced symmetric operads . in particular , it is cofibrant .
* proof .
* this is an easy consequence of the fact that @xmath620 is a left quillen functor .
@xmath67 the cofibrantness of @xmath402 was first claimed in @xcite without a proof . to the best of our knowledge
the proof first appeared in @xcite and uses a comparison between @xmath402 and its boardman - vogt @xmath623-construction . for a categorical operad @xmath5
we will denote by @xmath624 the geometric realisation of the nerve of @xmath208 [ rhnfmnstrong ] the operad @xmath625 is strongly homotopy equivalent to @xmath402 . * proof .
* both @xmath3-operads @xmath626 and @xmath31 are fibrant , cofibrant and contractible .
hence , they are strongly homotopy equivalent .
so are their images under @xmath627 @xmath67 the getzler - jones decomposition of @xmath402 is not a regular @xmath628-decomposition so we can not take the poset of its cells and form a categorical operad as was proposed in @xcite .
however , the previous corollary shows that @xmath36 is an appropriate substitute for this nonexistent poset operad .
[ rhnfmn ] there is a chain of weak operadic equivalences between @xmath629 and @xmath30 * proof . *
the operad @xmath630 is the bar construction on the terminal reduced @xmath3-operad .
so , we have a zig - zag @xmath631 which , after symmetrization , gives the following zig - zag of morphisms of symmetric operads @xmath632 by theorem [ symnerve ] . if we forget about nullary operations , we get the zig - zag of weak operadic equivalences @xmath633 since it can be obtained by symmerisation from the zig - zag of equivalences of fibrant cofibrant operads @xmath634 hence , the zig - zag [ zigzag ] consists of weak equivalences .
@xmath67 recall that the iterated monoidal category operad @xmath635 contains an internal @xmath3-operad @xcite .
it is easy to see that @xmath635 is a reduced categorical symmetric operad and its internal @xmath3-operad is also a reduced internal @xmath3-operad .
so we have a canonical map of operads @xmath636 recall also that the inclusion of the milgram poset @xmath637 induces a weak equivalence on the nerves @xcite .
the following theorem provides an alternative proof of the theorem [ rhnfmn ] .
[ nk ] the map of operads @xmath638 is a weak equivalence . *
* by theorems [ symnerve ] and [ formular ] , we have @xmath639 it is trivial to check that @xmath640 is a reedy cofibrant functor on @xmath641 because @xmath642 is a monomorphism . therefore , @xmath643 is a weak equivalence .
but @xmath644 is contractible , hence , we have a weak equivalence @xmath645 so in the commutative square ( 60,33 ) ( 35,23.5)(0,0 ) ( 37,21)(0,-1)10 ( 53,26 ) ( 72,25)(0,0 ) ( 72,21)(0,-1)10 ( 52,25)(1,0)12 ( 35,7.5)(0,0 ) ( 72,8.5)(0,0 ) ( 52,8.5)(1,0)12 both horizontal and left vertical arrows are weak equivalences and , therefore , the right vertical arrow is a weak equivalence .
finally , the functor @xmath646 is factorised as @xmath647 and the statement of our theorem follows .
@xmath67 the canonical map @xmath648 provides a cofibrant replacement for @xmath649 where @xmath650 is the unbased version of @xmath651 there is a canonical map @xmath652 since @xmath248 contains an internal pruned operad .
we actually can restrict @xmath653 to a subcategory of @xmath11 which consists of objects which do not contain degenerate decorations . obviously , the last subcategory is a deformation retract of @xmath194 the deformation retraction is given by the following composite ( dropping off of dead leaves ) : @xmath654 @xmath655 where @xmath303 is obtained from @xmath27 by dropping off a branch which corresponds to the degenerate fiber .
so we will consider this subcategory as the domain of @xmath653 but abusing notation will call it @xmath194 let @xmath295 be a labelled tree decorated by pruned @xmath3-trees .
obviously , an object from the fiber of @xmath653 over @xmath144 will be a labelled planar tree decorated by pruned @xmath3-trees such that the reduction of it ( i.e. the deletion of all vertices of valency @xmath18 together with their decorations ) is @xmath318 the morphisms are insertion of a vertex of valency @xmath18 decorated by @xmath656 and deletion of such decorations .
the following lemma is obvious from this description .
[ fiber ] the reduced decorated tree @xmath144 considered as an object of the fiber @xmath657 is a terminal object of this fiber .
[ section ] the operadic morphism @xmath653 has a ( nonoperadic ) section @xmath658 which maps an object @xmath144 to the terminal object @xmath144 in @xmath659 * proof . * to define @xmath660 on morphisms we have to define it on generators and then check correctness .
we define it on a generator corresponding to a surjection @xmath661 equal to the composite @xmath662 where we insert @xmath663 in the places where @xmath109 has one - tip fibers .
it is routine to check the relations .
@xmath67 from these two lemmas we have [ phnrhnmn ] the operadic functor @xmath653 induces a weak equivalence of simplicial operads @xmath664 so all three operads @xmath665 are @xmath17-operads . a pruned topological @xmath3-operad @xmath5 will be called _
contractible _ provided the unique map to the terminal @xmath3-operad is a weak equivalence i.e. every @xmath666 is a contractible topological space .
[ enspace ] let @xmath5 be a contractible pruned @xmath3-operad in the category of compactly generated hausdorff spaces such that every @xmath666 is a cofibrant topological space and let @xmath8 be an algebra of @xmath5 .
then @xmath8 has a structure of an @xmath17-space , so up to group completion @xmath8 is an @xmath3-fold loop space .
* proof . *
since @xmath8 is an algebra of @xmath5 it is also an algebra of @xmath667 and , by theorem [ formulap ] , @xmath668 it is not hard to check that the sequence @xmath669 has the structure of an operad and , moreover , the canonical map @xmath670 is operadic . but @xmath671 has the same homotopy type as @xmath672 because of contractibility of @xmath666 .
so @xmath8 is an algebra of a @xmath17-operad @xmath673 @xmath67 it is still possible for @xmath8 from the previous theorem to be an @xmath674-space for @xmath675 .
for example , if @xmath355 is any @xmath676-operad , then @xmath677 is contractible but @xmath678 .
a similar theorem holds for reduced @xmath3-operads in the category @xmath679 of chain complexes over a commutative ring with unit @xmath680 let @xmath681 be a reduced @xmath3-operad in @xmath679 which has @xmath682 for every pruned tree .
a reduced @xmath3-operad in @xmath679 equipped with an augmentation @xmath683 will be called _
contractible _ provided its augmentation is a weak equivalence . the method used in the proof of theorem [ enspace ]
can be used without change to prove the following .
[ enchain ] let @xmath5 be a contractible reduced @xmath3-operad in @xmath679 such that @xmath666 is a chain complex of projective @xmath345-modules for every @xmath41 let @xmath8 be an algebra of @xmath5 .
then @xmath8 admits an action of a symmetric reduced operad weakly equivalent to the operad of @xmath345-chains of the little disc operad .
finally , theorems [ gjcof ] and [ symgj ] imply that we can obtain the full solution of the coherence problem of @xmath3-fold loop spaces in the spirit of stasheff s original work @xcite using cells @xmath42 for higher @xmath3-trees instead of associahedra .
this was first claimed by getzler and jones but some doubts appeared since tamarkin came up with his counterexample .
our theorems [ gjcof ] and [ symgj ] do confirm that the @xmath402-algebra structure is equivalent to the existence of a sequence of inductive extensions of higher homotopies from the boundaries of @xmath42 to their interior even though @xmath38 is not always a pl - ball .
the exact combinatorics of @xmath42 will be discussed in a future paper .
here we just give a few examples of the manifolds @xmath42 which can be drawn on paper . if @xmath45 and @xmath684 $ ] then @xmath685 the stasheff associahedron as was said before . if @xmath24 we have two one - dimensional manifolds commonly known as associator and braiding
we also have three @xmath18-dimensional polytopes : a pentagon and two hexagons , and two @xmath686-dimensional polytopes , the associahedron @xmath687 and a @xmath686-dimensional polytope well known in the theory of yang - baxter operators .
finally we have three @xmath686-dimensional polytopes ( there should be a copy of the first one we do not show here ) which can be found in the paper of bar - natan @xcite .
it is recommended to the reader to try to draw up all the polytopes @xmath42 of dimension less than or equal to @xmath686 with @xmath688 ( there will be only three of them which are different from the above polytopes ) and @xmath689 ( only one new polytope ) .
this is a short section in which we define reduced ( both based and unbased ) swiss - cheese type @xmath3-operads and show that the techniques developed in the previous sections are powerful enough to easily carry out the main results for classical operads to the case of swiss - cheese type operads .
the swiss - cheese type operads ( _ sc - operads or sc type operads _ for short ) were introduced by voronov in @xcite with motivation to describe a finite dimensional model of the moduli space of genus - zero riemann surfaces from open - closed string theory .
the importance of this class of operads was also understood by kontsevich in @xcite who explained that the category of algebras of voronov s swiss - cheese operad is a natural place for developing a theory of deformation complexes and its higher dimensional generalisation .
recall first @xcite that a symmetric reduced sc - operad in a symmetric monoidal category @xmath48 is a special symmetric coloured operad with two colours and consists of a collection @xmath690 of objects of @xmath48 indexed by pairs of natural numbers with an action of the product of the symmetric groups
@xmath691 we also require that @xmath692 for a reduced operad .
the operadic multiplication is represented by a family of morphisms @xmath693 @xmath694 which must satisfy some natural associativity , equivariancy and unitarity conditions .
notice that we have two symmetric operads here : @xmath696 and @xmath697 an algebra of such an operad consists of a pair of objects @xmath698 such that @xmath696 acts on @xmath699 and @xmath700 acts on @xmath701 and all the other spaces @xmath690 provide the interplay between these two actions .
the main example of an unbased reduced symmetric sc - operad is voronov s swiss cheese operad @xmath702 whose @xmath703-space is the space of @xmath257 disjoint @xmath3-disks and @xmath50 disjoint @xmath3-semidisks in a big @xmath3-semidisk @xmath704 notice that our presentation of @xmath705 is different from voronov s picture in @xcite .
he defines @xmath706 by requiring @xmath707 it is more convenient for us to ask @xmath708 however , because it makes the relations between @xmath709 and the getzler - jones decomposition more evident .
a coloured pruned @xmath3-tree is a pruned @xmath3-tree @xmath27 equipped with a map of trees @xmath710 such that the induced @xmath3-ordinal structure on @xmath711 is a suspension over @xmath4-tree or a degenerate @xmath3-tree . in other words
@xmath715 sends the distinguished branch of @xmath27 to the distinguished branch of @xmath199 every coloured morphism between @xmath3-trees can be restricted to the distinguished branches of these trees . since these branches are suspensions of @xmath3-trees , one can consider the restriction of a morphism as a morphism between @xmath4-trees .
a coloured morphism between coloured @xmath3-trees is _ an injection _ if the underlying morphism of trees is an injection .
a coloured morphism between coloured @xmath3-trees is _ a surjection _ if the underlying morphism of trees is a surjection .
one can take a fiber of a coloured surjection between @xmath3-trees and then the fibers obtain canonical colouring .
a reduced sc type @xmath3-operad in a symmetric monoidal category @xmath48 is a collection @xmath716 where @xmath27 runs over all coloured pruned @xmath3-trees such that @xmath717 and @xmath718 where @xmath719 is a linear tree with its unique tip coloured by @xmath1 and @xmath720 is a linear tree with its tip coloured by @xmath721 this collection is equipped with a morphism @xmath165 for every coloured morphism of trees @xmath166 between coloured pruned @xmath3-trees .
they must satisfy the obvious associativity and unitarity conditions . for an unbased reduced sc - operad we use collections without degenerate trees and
define multiplication only with respect to surjections of coloured trees . immediately from the definition we see that as in the symmetric case an sc type @xmath3-operad @xmath5 gives rise to two operads @xmath722 and @xmath723 if we consider its restriction to the @xmath3-trees with no branches with colour @xmath1 or to the trees with no branches coloured @xmath721 but unlike in the symmetric case these two operads are of different types : @xmath722 is an @xmath3-operad but @xmath723 is an @xmath4-operad .
the reduced sc type @xmath3-operads form a category and we have a desymmetrisation functor from symmetric reduced sc - operads to the reduced sc type @xmath3-operads .
all the machinery of section [ reducedoperads ] is then applicable to the swiss - cheese case .
we denote by @xmath724 the corresponding functor of symmetrisation and by @xmath725 the free reduced symmetric sc - operad freely generated by an internal reduced sc type @xmath3-operad .
we have the following analogue of theorem [ formular ] [ formularsc ] let @xmath5 be a cocomplete reduced symmmetric categorical @xmath3-operad of sc type and @xmath726 be an internal reduced sc type @xmath3-operad in @xmath5 then @xmath727 where @xmath728 is the operadic functor representing the operad @xmath729 the unbased swiss - cheese analogue of the fulton - macpherson operad @xmath732 was defined by voronov in @xcite .
the definition of its based version @xmath733 is obvious .
it is also obvious now how to define the analogue of the getzler and jones operad @xmath734 a little gift for us is that , for a pruned coloured @xmath3-tree @xmath202 the spaces @xmath735 and @xmath736 coincide .
so we have the proof of our next theorem almost for free : * the operad @xmath737 is contractible and @xmath738 is a cellular and contractible unbased reduced sc type @xmath3-operad . *
the canonical map @xmath739 is an isomorphism .
the analogous result holds in the unbased case . *
the operad @xmath732 is cellular . *
the operad @xmath740 is strongly homotopy equivalent to @xmath741 * there is a chain of weak operadic equivalences between @xmath742 and @xmath743 it is not hard now to give the definitions of the endomorphism sc @xmath3-operad of a pair of objects @xmath744 and of the category of algebras of an sc @xmath3-operad .
the desymmetrisation functor preserves the endomorphism operad and , hence , symmetrisation preserves the category of algebras .
analogously to theorems [ enspace ] and [ enchain ] we have [ enspacesc ] let @xmath5 be a contractible reduced sc type @xmath3-operad in the category of compactly generated hausdorff spaces such that every @xmath666 is a cofibrant topological space and let @xmath698 be an algebra of @xmath5 .
then @xmath698 admits an action of a symmetric sc - operad weakly equivalent to voronov s reduced swiss - cheese operad .
[ enchainsc ] let @xmath5 be a contractible reduced sc type @xmath3-operad in @xmath679 such that @xmath666 is a chain complex of projective @xmath345-modules for every @xmath41 let @xmath698 be an algebra of @xmath5 .
then @xmath698 admits an action of a symmetric reduced sc - operad weakly equivalent to the operad of @xmath345-chains of the voronov s reduced swiss - cheese operad .
voronov a. , homotopy gerstenhaber algebras , _ confrence mosh flato 1999 _ , * ii * ( dijon ) , 307331 , math .
phys . stud .
, 22 , kluwer acad .
dordrecht , ( 2000 ) .
voronov a. , the swiss - cheese operad , _ contemp .
, _ * 239 * ( 1998 ) 365373 .
|
it is well known that the forgetful functor from symmetric operads to nonsymmetric operads has a left adjoint @xmath0 given by product with the symmetric group operad .
it is also well known that this functor does not affect the category of algebras of the operad . from the point of view of the author s theory of higher operads , the nonsymmmetric operads are @xmath1-operads and
@xmath0 is the first term of the infinite series of left adjoint functors @xmath2 called symmetrisation functors , from @xmath3-operads to symmetric operads with the property that the category of one object , one arrow , . .
, one @xmath4-arrow algebras of an @xmath3-operad @xmath5 is isomorphic to the category of algebras of @xmath6 .
in this paper we consider some geometrical and homotopical aspects of the symmetrisation of @xmath3-operads .
we follow getzler and jones and consider their decomposition of the fulton - macpherson operad of compactified real configuration spaces .
we construct an @xmath3-operadic counterpart of this compactification which we call the getzler - jones operad .
we study the properties of getzler - jones operad and find that it is contractible and cofibrant in an appropriate model category .
the symmetrisation of the getzler - jones operad turns out to be exactly the operad of fulton and macpherson .
these results should be considered as an extension of stasheff s theory of @xmath1-fold loop spaces to @xmath3-fold loop spaces @xmath7 we also show that a space @xmath8 with an action of a contractible @xmath3-operad has a natural structure of an algebra over an operad weakly equivalent to the little @xmath3-disks operad .
a similar result holds for chain operads .
these results generalise the classical eckman - hilton argument to arbitrary dimension .
finally , we apply the techniques to the swiss cheese type operads introduced by voronov and prove analogous results in this case . 1991 math .
subj . class .
18d05 , 18d50 , 55p48
|
the control of the star formation efficiency ( sfe ) by turbulence is a central issue in our present understanding of star formation , and currently a topic of intense study ( see , e.g. , the reviews by * ? ? ?
* ; * ? ? ?
in recent years , several groups have studied the sfe of molecular clouds ( mcs ) using numerical simulations of isothermal turbulence , in which the entire numerical box represents the interior of a molecular cloud ( see , e.g. , the reviews by * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
one confusing issue is that simulations of _ driven _ turbulence seem to indicate that the sfe _ decreases _ as the turbulent rms mach number @xmath14 increases ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , while simulations of _ decaying _ turbulence suggest that the sfe _ increases _ with increasing @xmath14 @xcite .
this has prompted the question of how does turbulence actually originate and behave in real mcs . to answer this question ,
it has become necessary to investigate the entire evolutionary process of mcs .
the formation of mcs by collisions of warm neutral medium ( wnm ) streams has been intensely studied in recent years .
a vast body of numerical simulations has shown that moderate , transonic compressions in the wnm can nonlinearly trigger a phase transition to the cold neutral medium ( cnm ) ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , and that the dense gas produced by this mechanism is overpressured with respect to the mean wnm thermal pressure @xcite and turbulent , due to the combined action of kelvin - helmholz , thermal @xcite and nonlinear thin - shell @xcite instabilities @xcite . the turbulence produced by this mechanism
is continually driven for as long as the compression lasts .
the physical scenario of mc evolution was outlined by @xcite , who estimated the column densities necessary for the cloud to become self - gravitating , molecular , and magnetically supercritical , finding them to be comparable .
the one - dimensional physical conditions in the dense gas were calculated analytically by @xcite and @xcite .
a recent review of the subject has been presented by @xcite .
more recently , simulations including self - gravity and `` sink particles '' , which represent gravitationally collapsed objects ( stars or stellar clusters ) , and using finite - duration compressions in the wnm , although lacking stellar feedback and magnetic fields , have been used to study the evolution of the turbulent motions and of the sfe in a self - consistent manner ( * ? ? ?
* hereafter paper i ) . concerning the velocity dispersion in the clouds
, these authors found that the turbulence is intermediate between driven and decaying , since what decays is the driving rate of the turbulence as the inflows weaken with time .
however , they also found that the random motions are gradually replaced by global infall motions , as the cloud begins to contract gravitationally . concerning the sfe , paper i measured the masses of dense gas @xmath15 and of the collapsed stellar objects @xmath16 in the simulations , allowing a measurement of the sfe , defined as @xmath17 the resulting sfe was however too high , reaching @xmath18% roughly 6 myr after the time at which star formation ( sf ) had begun ( denoted @xmath19 ) , although this excessive sfe can possibly be attributed to the neglect of stellar feedback in that simulation . indeed , paper i estimated , using a prescription by @xcite and a standard imf @xcite , that by 3 myr after @xmath19 , enough massive stars would have formed as to be able to destroy the cloud by ionization . at that point ,
the sfe was @xmath20% , closer to the typical values @xmath21% reported observationally for full mc complexes @xcite .
one obvious possible reason for the relatively high values of the sfe obtained in this type of simulations is the neglect of stellar feedback and magnetic fields .
however , it is also of interest to perform a more systematic investigation of the degree of variability that the sfe could exhibit within the scenario of paper i , simply by varying the parameters of the wnm compressions triggering the formation of the cloud complex . in this paper
we undertake a first approach to such task , by varying three parameters of the wnm stream collisions modeled in the simulations .
first , we consider the inflow speed @xmath22 and the velocity dispersion of the background turbulence initially present in the medium , both measured by their respective mach numbers , @xmath0 and @xmath1 , with respect to the unperturbed wnm .
subsequently , we consider the mass in the colliding streams @xmath2 , as determined by their radius @xmath23 and length @xmath24 .
since the parameter space covered by these three parameters is already quite large , in this work we do not consider variations in the collision angle of the streams , instead having them collide head - on in all cases .
the paper is organized as follows . in [ sec : the_model ] we describe the numerical model and experiments . in [ sec : results ] we present our results , and in [ sec : conclusions ] we present a summary and our conclusions .
for the numerical simulations , we use the same numerical setup as that used in paper i , except that we now use the smoothed particle hydrodynamics ( sph ) + @xmath25-body code gadget-2 @xcite ( in paper i we used the previous version of the code , gadget ) , modified to include random turbulence driving and sink particles according to the prescription of @xcite , and including parameterized heating and cooling , as applied in paper i , using the fit of @xcite to a variety of atomic and molecular cooling processes .
this cooling function causes the gas to be thermally unstable , under the isobaric mode , in the density range @xmath26 .
we assume that the gas is all atomic , with a mean atomic weight of 1.27 .
the numerical box is periodic , with size @xmath27 . in all cases ,
we use @xmath28 particles , and set the mean number of particles within a smoothing volume to 40 . according to the criterion of @xcite , the mass resolution is twice the number of particles within a smoothing volume , or @xmath29 times the mass per particle .
the critical density for sink formation is set at @xmath30 , and the outer sink accretion radius is set at 0.04 pc .
an initial turbulent velocity field of one - dimensional velocity dispersion , characterized by its rms mach number @xmath1 and applied at scales between 1/4 and 1/8 of the box size , is added to the inflow velocity field , in order to trigger the instabilities that render the cloud turbulent .
note that this added turbulent velocity field is applied by turning on the random driver for the first few timesteps of the simulation s evolution , and that what we actually control is the energy injection rate parameter .
thus , simulations intended to have the same turbulence strength do so only approximately , as the flow s response is slightly different in every realization .
the initial conditions consist of a uniform medium at @xmath31 and @xmath32 k , in which two cylinders of length @xmath24 along the @xmath33 direction and radius @xmath23 are set to collide head - on at the @xmath34 plane of the simulation ( refer to fig . 1 of paper i ) .
note that the cylindrical inflows are entirely contained within the numerical box , since the boundaries are periodic .
the length @xmath24 is measured from the central collision plane , and is always shorter than the half - length of the box , implying that a small region between the edge of the inflows and the box boundaries is not given any velocity .
this region is partially evacuated during the subsequent evolution of the simulations , as the gas within it tends to fill the void left by the inflows .
table [ tab : run_params ] shows the various runs we performed for the present study , indicating the relevant parameters for each one .
in addition to the inflow length , radius , and velocity , defined above , table [ tab : run_params ] gives the mach number @xmath0 corresponding to @xmath22 at the initial temperature of the gas , for which the adiabatic sound speed is @xmath35 , the one - dimensional rms mach number of the initial turbulent motions , @xmath1 , the total mass contained in the simulation box , @xmath36 , the mass contained in the two inflows , @xmath2 , and the mass per sph particle , @xmath37 .
the runs are labeled mnemonically , with their names giving , in that orded , the inflow mach number @xmath0 , the background mach number @xmath1 , and the inflow mass , @xmath2 .
note that run mi1-mb.02-ma2e4 has the same parameters as run [email protected] from paper i , and we take these as the `` fiducial '' set of parameters .
note , however , that these two runs are not identical because they were performed with different codes .
gadget-2 differs in many ways from gadget , and in particular it takes longer timesteps , implying that the initial forcing ( used to trigger the instabilities in the dense layer ) is applied at different time intervals , and with different random seeds .
thus , the two runs are similar only in a statistical sense , but are not identical . for reference , in fig .
[ fig : run4_img ] we show a face - on view of mi1-mb.02-ma2e4 at the time when it is beginning to form stars , showing that the general morphology it develops is similar to that of run [email protected] from paper i ( compare to fig . 4 of that paper , noting that the linear scales shown are different in each figure ) .
we report the sfe as defined by eq .
( [ eq : sfe_def ] ) , with the `` dense '' gas defined as that with number density @xmath39 . however , because the sfe is actually a function of time , and sink formation begins at different times in different runs , in order to report a _ number _ for the sfe , we estimate it in two different ways .
one is to measure the `` absolute '' sfe 25 myr after the start of the simulation , which we denote as sfe@xmath4 .
the other is to measure the `` relative '' sfe , which we define as the sfe 5 myr after the onset of star formation in the simulation , and which we denote as sfe@xmath5 .
this is actually close to the `` star formation rate per free - fall time '' , sfr@xmath40 ( after sf has begun ) , as defined by @xcite , since the free - fall time for gas at @xmath41 is @xmath42 myr .
in this section , we consider the dependence of the sfe on the inflow velocity of the colliding streams , @xmath22 .
figure [ fig : mg_ms_of_t_vs_vinf ] shows the evolution of the dense gas mass ( _ top panel _ ) and the sink mass ( _ bottom panel _ ) for runs mi1-mb.02-ma2e4 , mi2-mb.02-ma2e4 , and mi3-mb.02-ma2e4 .
these runs have all parameters equal , except for the speed of the inflows ( cf .
table [ tab : run_params ] ) , which are varied from @xmath43 in mi1-mb.02-ma2e4 to @xmath44 in mi3-mb.02-ma2e4 .
note that the time at which the runs begin to form sinks , @xmath19 , is different in each case .
top panels _ of fig.[fig : sfe_of_t_vs_vinf ] then show the sfe , defined as in eq.([eq : sfe_def ] ) .
top left panel _ shows sfe@xmath4 , i.e. , starting from the beginning of the simulation , and up to a total time of 25 myr .
the _ top right panel _ shows the sfe starting from the time at which sink formation begins in each run , allowing one to read off sfe@xmath5 . , 2.5 and 3.5 respectively.,title="fig : " ] , 2.5 and 3.5 respectively.,title="fig : " ] ,
_ left _ ) , shown out to 25 myr after the start of the runs , and the `` relative '' sfe ( _ right _ ) , shown from the onset of sink formation , for runs mi1-mb.02-ma2e4 , mi2-mb.02-ma2e4 , and mi3-mb.02-ma2e4 .
sfe@xmath5 is the value of this curve at a relative time of 5 myr .
_ bottom panel : _ evolution of the one - dimensional velocity dispersion perpendicular to the direction of the inflows for the same three runs.,title="fig : " ] , _ left _ ) , shown out to 25 myr after the start of the runs , and the `` relative '' sfe ( _ right _ ) , shown from the onset of sink formation , for runs mi1-mb.02-ma2e4 , mi2-mb.02-ma2e4 , and mi3-mb.02-ma2e4 .
sfe@xmath5 is the value of this curve at a relative time of 5 myr .
_ bottom panel : _ evolution of the one - dimensional velocity dispersion perpendicular to the direction of the inflows for the same three runs.,title="fig : " ] , _ left _ ) , shown out to 25 myr after the start of the runs , and the `` relative '' sfe ( _ right _ ) , shown from the onset of sink formation , for runs mi1-mb.02-ma2e4 , mi2-mb.02-ma2e4 , and mi3-mb.02-ma2e4 .
sfe@xmath5 is the value of this curve at a relative time of 5 myr . _
bottom panel : _ evolution of the one - dimensional velocity dispersion perpendicular to the direction of the inflows for the same three runs.,title="fig : " ] from the _ top right panel _ of fig .
[ fig : sfe_of_t_vs_vinf ] we see that the mean slope of the curve sfe@xmath45 , denoted @xmath46 , decreases , although moderately , with increasing @xmath0 .
it is interesting , however , that all three runs form stars at roughly the same rate ( i.e. , the slopes of the curves in the _ right panel _ of fig .
[ fig : mg_ms_of_t_vs_vinf ] , which give the mass accretion rate onto the sinks , and which we identify with the star formation rate , sfr , are all similar ) . ) implies that @xmath47@xmath46@xmath48 sfr . ]
the decrease in @xmath46 is thus due to the larger mass growth rate of the cloud induced by the larger inflow velocities ( fig.[fig : mg_ms_of_t_vs_vinf ] , _ top _ ) , not to a smaller sfr .
the larger - mass clouds appear to be incapable of forming stars at a proportionally larger rate because the larger inflow velocity produces larger turbulent velocity dispersions @xmath49 in the clouds , . ] which have been shown to reduce the sfr in simulations of driven turbulence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this is illustrated in the _ bottom _ panel of fig .
[ fig : sfe_of_t_vs_vinf ] , which shows the evolution of the one - dimensional @xmath49 perpendicular to the direction of the inflows for the same three runs .. thus , the correct velocities in those figures are obtained by dividing by this factor . ]
the trends of sfe@xmath4 , sfe@xmath5 , and @xmath49 with the inflow speed are summarized in fig.[fig : sfe_vs_vinf ] . in this figure , @xmath49 is measured for run mi1-mb.02-ma2e4 at the time when it exhibits a minimum in the _ bottom _ panel of fig .
[ fig : sfe_of_t_vs_vinf ] , i.e. , @xmath50 myr .
this is equivalent to @xmath51 , where @xmath52 is the _ inflow crossing time _ , the time taken by the tail of the inflowing cylinders to reach the collision plane .
we do this in an attempt to capture the true velocity dispersion in the cloud after the initial transients have ended , but before global collapse sets in .
the latter is indicated by the smooth rise in @xmath49 for this run during the time interval @xmath53 myr .
then , for consistency , we measure @xmath49 for the other two runs also at 2/3 of their own @xmath54 . and sfe@xmath5 on the inflow speed for runs mi1-mb.02-ma2e4 , mi2-mb.02-ma2e4 , and mi3-mb.02-ma2e4 .
both sfes are seen to decrease with increasing @xmath22 , apparently due to an increase of the turbulent velocity dispersion @xmath49 as the inflow speed increases , which in turn reduces the sfr . ]
we now consider the response of the sfe to the amplitude of the initial turbulent velocity field .
note that this field was not originally intended to produce density condensations on its own , but just to sufficiently disorganize the inflow velocity field as to trigger the instabilities that render the cloud turbulent .
however , in the cases of stronger turbulence , we do observe clump formation everywhere in the box as a result of the initial background turbulence , and not just at the collision site of the inflows .
figure [ fig : sfet_vs_vrms ] shows the evolution of the sfe for runs mi1-mb.11-ma2e4 , mi1-mb.27-ma2e4 , mi1-mb.02-ma2e4 and mi0-mb.10-ma2e4 .
the first three runs differ only in the strength of the initial turbulence , measured by @xmath1 ( cf .
table [ tab : run_params ] ) .
the last run has nearly the same value of @xmath1 as mi1-mb.11-ma2e4 , but with no inflow velocity , in order to assess the amount of star formation induced solely by the turbulent field , in the absence of colliding streams .
, _ left panel _ ) and the `` relative '' sfe ( _ right panel _ ) for four runs , characterized by various values of the initial background turbulent mach number @xmath1 , indicated by the entry `` mb .
# # '' in the runs names .
three of the runs have the same inflow mach number , @xmath55 , while the fourth run has no inflows ( @xmath56 ) , in order to assess the sfe due exclusively to the initial background turbulence.,title="fig : " ] , _ left panel _ ) and the `` relative '' sfe ( _ right panel _ ) for four runs , characterized by various values of the initial background turbulent mach number @xmath1 , indicated by the entry `` mb .
# # '' in the runs names .
three of the runs have the same inflow mach number , @xmath55 , while the fourth run has no inflows ( @xmath56 ) , in order to assess the sfe due exclusively to the initial background turbulence.,title="fig : " ] from these figures , we see that @xmath19 becomes longer and @xmath46becomes smaller as larger values of @xmath1 are considered .
this appears to be due to the stronger fragmentation induced by the turbulent velocity field , which in the extreme case of mi1-mb.27-ma2e4 almost obliterates the inflows , and produces scattered clumps througout the simulation box , with very little remaining of the cloud formed by the inflows , as illustrated in fig.[fig : runs3_11_img ] , _ left panel_. it is also worth noting that run mi0-mb.10-ma2e4 , which has no inflows , has a much larger @xmath19 and a lower @xmath46 than run mi1-mb.11-ma2e4 , which differs from the former only in the presence of the inflows .
as shown in fig .
[ fig : runs3_11_img ] , this run also produces scattered clumps throughout the numerical box , although not as profusely as mi1-mb.27-ma2e4 .
so , we conclude that sink formation is still dominated by the colliding streams in mi1-mb.11-ma2e4 , although a small fraction of the sinks is contributed by the global turbulence .
figure [ fig : sfe_vs_mrms ] summarises the results of this section .
a clear trend of a decreasing sfe ( seen in both sfe@xmath4 and sfe@xmath5 ) with increasing @xmath1 is seen , which we interpret as a result of the reduction of the fragment mass with increasing turbulence strength ( at constant total mass ; * ? ? ?
* ) , and of the fact that smaller - mass fragments tend to have smaller sfes ( [ sec : sfe_vs_rinf ] ) . and sfe@xmath5 on the rms mach number , @xmath1 , of the initial ( background ) turbulent velocity perturbations .
both indicators are seen to decrease with increasing @xmath1 as a consequence of the progressively stronger fragmentation induced by the turbulence . ]
the last dependence of the sfe we analyse is on the mass content of the colliding inflows .
so , we consider inflows of various radii . however , since a very narrow inflow is necessarily more poorly resolved , we consider smaller simulation boxes in two of the cases , in order to better resolve the resulting clouds . specifically , as shown in table [ tab : run_params ] , @xmath23 in run mi1-mb.02-ma5e3 is half that in mi1-mb.02-ma2e4 .
run mi1-mb.06-ma2e3 has @xmath23 equal to that of mi1-mb.02-ma5e3 , but half the length , since the numerical box size of the former is half that of the latter .
finally , mi1-mb.06-ma6e2 has @xmath23 equal to half that of mi1-mb.06-ma2e3 .
so , the total mass contained in the inflows of runs mi1-mb.02-ma2e4 , mi1-mb.02-ma5e3 , mi1-mb.06-ma2e3 and mi1-mb.06-ma6e2 is , respectively , @xmath57 , @xmath58 , @xmath59 , and @xmath60 .
figure [ fig : sfet_vs_rcyl ] shows the evolution of sfe@xmath4 and sfe@xmath5 for these runs .
we see that there is a general trend for the sfe , in both its forms , to increase with the total mass involved in the stream collision .
there is only a reversal to this trend in the relative sfe between runs mi1-mb.02-ma2e4 and mi1-mb.02-ma5e3 because the latter has a large early maximum of sfe@xmath5 , although later it decreases , in a period of mass accumulation in the cloud at low sfr .
other than that , the trend is general , as shown in fig .
[ fig : sfe_vs_mcyl ] .
it is important to note that for all runs in this series we used the same energy injection rate of the turbulence driver .
however , runs mi1-mb.06-ma6e2 and mi1-mb.06-ma2e3 , performed in a smaller computational box , have an initial , background rms turbulent mach number that is roughly 2.5 times larger than that of runs mi1-mb.02-ma2e4 and mi1-mb.02-ma5e3 .
this may additionally reduce the sfe because of the additional fragmentation it produces , but we see that the trend of the sfe to decrease with decreasing inflow mass holds generally even at the same physical box size , so the result appears robust . ,
_ left panel _ ) and the `` relative '' sfe ( _ right panel _ ) for runs mi1-mb.02-ma2e4 , mi1-mb.02-ma5e3 , mi1-mb.06-ma2e3 , and mi1-mb.06-ma6e2 .
the radius of the cylindrical inflows for these runs is respectively 32 , 16 , 16 , and 8 pc .
runs mi1-mb.02-ma2e4 and mi1-mb.02-ma5e3 are performed in a 256-pc box , with inflow length 112 pc , while runs mi1-mb.06-ma6e2 and mi1-mb.06-ma2e3 are performed in a 128-pc box with the same number of sph particles ( thus being better resolved ) and a 48-pc inflow length.,title="fig : " ] , _ left panel _ ) and the `` relative '' sfe ( _ right panel _ ) for runs mi1-mb.02-ma2e4 , mi1-mb.02-ma5e3 , mi1-mb.06-ma2e3 , and mi1-mb.06-ma6e2 .
the radius of the cylindrical inflows for these runs is respectively 32 , 16 , 16 , and 8 pc .
runs mi1-mb.02-ma2e4 and mi1-mb.02-ma5e3 are performed in a 256-pc box , with inflow length 112 pc , while runs mi1-mb.06-ma6e2 and mi1-mb.06-ma2e3 are performed in a 128-pc box with the same number of sph particles ( thus being better resolved ) and a 48-pc inflow length.,title="fig : " ] and sfe@xmath5 on the inflows mass . a general trend for the sfe to increase with inflow mass is observed . ]
the trend discused above can be put in the context of the theory of ( * ? ? ? * hereafter km05 ) for the sfr@xmath40 .
this theory predicts a dependence of the sfr@xmath40 on the virial parameter @xmath11 and the rms mach number @xmath14 . here ,
@xmath61 where @xmath62 is the cloud s kinetic energy , @xmath63 is the cloud s mass , @xmath38 is its rms turbulent velocity dispersion , and @xmath64 is the cloud s gravitational energy . for our flattened clouds , we compute the gravitational energy assuming they can be approximated as infinitely thin , uniform disks of radius @xmath65 , and write @xmath66 where @xmath67 is the area of the disk , @xmath68 is the ( uniform ) surface density , and @xmath69 is the gravitational potential . in our case ,
the latter is given by @xcite @xmath70 where @xmath71 is the second complete elliptic integral .
thus , the gravitational energy is @xmath72 to compute the kinetic energy of the clouds , we note that the relevant velocity dispersion is the one produced in the clouds as a consequence of the inflow collision @xcite , rather than the initial background turbulent velocity , which is much smaller .
since all four simulations analysed in this section have the same @xmath22 , we use the value of @xmath49 measured for run mi1-mb.02-ma2e4 in [ sec : sfe_vs_vinf ] ( fig .
[ fig : sfe_vs_vinf ] ) for all of them , namely @xmath73 .
noting that this value is a one - dimensional velocity dispersion , we take @xmath74 .
we finally obtain , from equations ( [ eq : alpha_def ] ) and ( [ eq : eg_calc ] ) , @xmath75 figure [ fig : sfe_vs_alpha ] shows the results of this exercise .
the _ solid _ line shows the simulation data , while the straight _ dotted _ line shows a least squares fit to them .
the _ dashed _ line shows the result from @xcite , given by @xmath76 where we have taken @xmath77 .
we see that the prediction by km05 , although being numerically within the same range as the data , exhibits a significantly shallower slope .
specifically , the fit to our data has a slope @xmath78 , where the uncertainty is the @xmath79 error of the fit , while the slope of the km05 prediction , @xmath80 , lies beyond this error .
we discuss this result further in [ sec : conclusions ] . on the virial parameter @xmath11 .
the straight _ dotted _ line shows a least squares fit to the simulation data , with slope @xmath81 , while the _ dashed _ line shows the result from @xcite , with slope @xmath80 . ]
in this paper , we have considered the scenario of molecular cloud formation by wnm stream collisions , and investigated the dependence of the sfe on three parameters of this scenario , namely the inflow speed , the rms mach number of the background medium , and the total mass contained in the inflows . since the sfe , defined as in eq.([eq : sfe_def ] ) , is a time - dependent function because the cloud continues to accrete mass from the wnm while it forms stars , we have considered two estimators of its time integral , namely the absolute sfe after 25 myr from the start of the simulation , sfe@xmath4 , and the `` relative '' sfe , 5 myr after the onset of sf in the cloud , denoted sfe@xmath5 .
we have found a wide range of values of these estimators as we vary the parameters of the simulations . in general , the sfe decreases , although moderately , with increasing inflow velocity . in particular , sfe@xmath5decreases from @xmath7 to @xmath82 as @xmath0 increases from 1.25 to 3.5 .
note that the runs in this case have similar sfrs , and the decrease in the sfe and @xmath46 is due to the faster increase in cloud mass rather than to a decrease in the sfr . that the sfr is similar in all three runs , in spite of the larger gas mass is probably due to the larger turbulent velocity dispersion in the dense gas caused by the larger inflow speed , which tends to inhibit the sfr , thus compensating the tendency to have a larger sfr due to the larger cloud masses .
similarly , the sfe decreases with increasing background turbulence strength , as the latter progressively takes a dominant role in the production of the dense gas but , due to the relatively small scales at which the turbulence is excited , the clouds and clumps formed by it are significantly smaller than the cloud formed by the coherent stream collision . in this case ,
sfe@xmath5 decreases from @xmath7 to @xmath83 as the rms mach number of the background turbulence increases from @xmath84 to @xmath85 . finally , the sfe in general decreases with decreasing mass of the inflows ( at constant @xmath22 ) .
this may be a consequence of the fact that clouds formed by the collision of our inflows always have roughly the same density , temperature , and velocity dispersion , so smaller clouds are more weakly gravitationally bound , a condition known to decrease the sfe @xcite .
the end result is that sfe@xmath5 decreases from @xmath7 to @xmath86 as the mass in the colliding streams decreases from @xmath9 to @xmath10 .
it is important to stress that this result is not at odds with the well known fact that the sfe _ increases _ as the object mass decreases from the mass scale of a giant molecular cloud ( @xmath87 @xmath88 , sfe @xmath84 ; myers et al .
1986 ) to that of a cluster - forming core ( @xmath89 ; sfe @xmath85 0.5 , lada & lada 2003 ) , because in this case the cores mean densities are much larger than those of the gmcs , while in our case the mean densities of the various clouds are always comparable .
thus , our clouds do not conform to larson s ( 1981 ) density - size scaling .
the latter results , expressed in terms of the virial parameter @xmath11 , exhibit partial agreement with the prediction by km05 for the dependence of the sfe after a free - fall time ( what those authors called the star formation rate per free - fall time , or sfr@xmath40 ) , which is directly comparable to our sfe@xmath5 .
although their prediction , without any rescaling , lies in the same range of values as our observed efficiencies , it contains a much shallower dependence on @xmath11 than we observe .
this may be due to the fact that those authors assumed that the clouds were supported by turbulent pressure , possibly provided by stellar feedback , while our simulations lack such support .
however , the notion of turbulent support has been questioned recently by @xcite , and the low observed efficiency may be the result of the _ dispersal _ ( rather than support ) of the parent cloud by its stellar products @xcite .
more work is needed to determine the causes of the discrepancy .
however , it is noteworthy that run mi1-mb.06-ma2e3 , which has a value of @xmath11 closest to unity , has a reasonably realistic value of sfe@xmath5@xmath90% .
we conclude that the sfe , even in the absence of further agents such as magnetic fields and stellar feedback , is a highly sensitive function of the parameters of the cloud formation process , and may be responsible for significant intrinsic scatter in observational determinations of the sfe .
the numerical simulations were performed in the cluster at crya - unam acquired with conacyt grants 36571-e and 47366-f to e.v .- s .
g.c.g.acknowledges financial support from grants in106809 ( unam - papiit ) and j50402-f ( conacyt ) . a .- k.j .
acknowledges support by the human resources and mobility programme of the european community under the contract meif - ct-2006 - 039569 .
hennebelle , p. , mac low , m .-
vzquez - semadeni 2007 , in structure formation in the universe : galaxies , stars , planets , ed .
g. chabrier ( cambridge : cambridge university press ) , in press ( arxiv:0711.2417 )
|
we investigate the response of the star formation efficiency ( sfe ) to the main parameters of simulations of molecular cloud formation by the collision of warm diffuse medium ( wnm ) cylindrical streams , neglecting stellar feedback and magnetic fields .
the parameters we vary are the mach number of the inflow velocity of the streams , @xmath0 , the rms mach number , @xmath1 of the initial background turbulence in the wnm , and the total mass contained in the colliding gas streams , @xmath2 .
because the sfe is a function of time , we define two estimators for it , the `` absolute '' sfe , measured at @xmath3 myr into the simulation s evolution ( sfe@xmath4 ) , and the `` relative '' sfe , measured 5 myr after the onset of star formation in each simulation ( sfe@xmath5 ) . the latter is close to the `` star formation rate per free - fall time '' for gas at @xmath6 .
we find that both estimators decrease with increasing @xmath2 , although by no more than a factor of 2 as @xmath0 increases from 1.25 to 3.5 .
increasing levels of background turbulence ( injected at scales comparable to the streams transverse radius ) similarly reduce the sfe , because the turbulence disrupts the coherence of the colliding streams , fragmenting the cloud , and producing small - scale clumps scattered through the numerical box , which have low sfes .
finally , the sfe is very sensitive to the mass of the inflows ( at roughly constant density and temperature ) , with sfe@xmath5 decreasing from @xmath7 to @xmath8 as the mass in the colliding streams decreases from @xmath9 to @xmath10 or , equivalently , the virial parameter @xmath11 increases from @xmath12 to @xmath13 .
this trend is in partial agreement with the prediction by @xcite , since the latter lies within the same range as the observed efficiencies , but with a significantly shallower slope .
we conclude that the observed variability of the sfe is a highly sensitive function of the parameters of the cloud formation process , and may be the cause of significant scatter in observational determinations .
[ firstpage ] interstellar matter stars : formation turbulence
|
in the last part of the twentieth century , cosmology became a precision science getting observational data with an accuracy comparable with the data obtained through the standard model .
the last data arising from observations confirm that our universe is accelerating due to the existence of some unknown type of energy and also requires the existence of an unknown field that permeates the universe and dictates the formation and evolution of the structures at large scales .
both dark components of the universe are not predicted by the standard model or by the general theory of relativity either .
this opens the possibility to extend these theories to limits beyond the current ones and to formulate new paradigms that predict new physics , under the condition that in the appropriated limits these new theories reproduce the current standard observations .
so far the best accepted candidate to be the dark energy of the universe is the cosmological constant @xmath2 .
it is well - known that observations of the cosmic microwave background radiation ( cmb ) and galaxies surveys fit very well with a small value of @xmath2 .
the main problem is to relate @xmath2 with some physical phenomena .
there are many interpretations for doing so , for example , the most simple one is that the einstein equations have two free constants , the gravitational one @xmath3 and the cosmological constant @xmath2 . here
the cosmological constant contains only a geometrical interpretation .
we fix @xmath2 in the same way as we have fixed @xmath3 , namely , using cosmological observations .
the problem here is that we have now a theory with two coupling constants , giving rise to the interpretation that the einstein theory could not be the most fundamental theory for gravitation that can exist .
other example is to relate the cosmological constant with the vacuum energy of the universe . here
the problem is that the values derived from any particle physics models is many orders of magnitude too big ( or vanishes in some susy cases ) compared with the one obtained using astronomical observations @xcite . from this interpretation follows an incompatibility between the theory of general relativity and the particle physics models that we have to face on at this moment .
there are some other proposals where the space - time contains subspaces embedded in a higher dimensional one , for example the branes theory @xcite , the dgp model @xcite , ekpyrotic @xcite , @xcite or cyclic universe @xcite , @xcite in which it is proposed the existence of a four dimensional manifold embedded in a five dimensional bulk .
it is important to mention that in the last two models @xcite , @xcite and @xcite , @xcite the branes move through the bulk and collide giving origin to the big bang .
the main goal of this work is to explore the possibility that the dynamics of the universe follows from its initial global topology .
for doing so there are different ways , but the idea here is the following : 1 .
we start with two three - dimensional branes embedded in a 5-dimensional space - time , where the five dimensional einstein equations with cosmological constant are fulfilled . for facility
we will suppose that the branes are three dimensional concentric spheres @xmath0 ( fig .
[ fig:1 ] ) but this result can be generalized to the other two homogeneous topologies @xmath4 and @xmath5 .
the initial conditions are such that the interior brane has the scale factor @xmath6 and the exterior brane ( where we live ) has the scale factor @xmath7 , with @xmath8 ( fig .
[ fig:1 ] ) . for simplicity ,
each brane have a common center .
+ @xmath9 and the brane tensions @xmath10 . ] 2 .
we assume that matter is confined on the branes , the matter content of the interior brane is a self interacting scalar field ( spin zero field ) and of the exterior brane is the one of the standard model interactions ( spin one fields ) . with these hypothesis
we show that 1 . the scalar field on the interior brane acts as an inflaton , expanding this brane and provoking the collision with the exterior one , which was in thermal equilibrium before the collision .
it follows a big bang scenario like in the ekpyrotic models . 2 .
it is possible to choose the free parameters of the model such that after the collision , both branes expand together , ( see equations ( [ 34 ] ) and ( [ 35 ] ) ) .
the fluctuations induced by the collision in the scalar field provoke the potential wells that give rise to the structure formation in the exterior brane , acting as the seeds for the galaxies formation on the brane where we live , via gravitational interaction . after the collision
the physics of the exterior brane is close related with the ekpyrotic and cyclic models @xcite , @xcite .
the main difference of this toy model with previous ones is the hypothesis of the initial conditions and the matter content of the branes .
for example , after the collision the exterior brane behaves in a very similar way as in the model presented in @xcite .
the problem of this last model is the reheating period , where it was not very natural to reheat the universe . in the model presented here
the reheating epoch is not necessary because the heat of the universe is consequence of the collision of the branes , as in the ekpyrotic model @xcite , @xcite .
it is important to mention that we let some open questions and use the results of previous works ( see for example @xcite , @xcite , @xcite , @xcite , @xcite ) , the main goal of this work is to show that there exist the possibility that the initial topology of the universe could reproduce its observed dynamics . this work is organized as follows . in section [ sec_ii ]
we present the model in detail and write the main field equations in section [ sec_iii ] . in section [ sec_iv ] we present the different possible scenarios and regimes of the model . in section [ sec_v ]
we present the dynamics of the universe with this model and in section [ sec_vi ] we give some conclusions and remarks .
in this section we write the most important feature of the model in a general way .
suppose two concentric 3-dimensional branes embedded in a 5-dimensional bulk , ( we use natural units @xmath11 ) .
the shape of the action to model this physical structure is given by @xmath12 being @xmath13 the exterior or interior region of the brane respectively , @xmath14 is the determinant of the five - dimensional ( 5d ) metric and @xmath15 the determinant of the four - dimensional ( 4d ) one , @xmath16 is the 5d planck mass , @xmath17 is the 5d scalar curvature , @xmath18 is the extrinsic curvature and @xmath2 is the 5d cosmological constant , @xmath19 is the lagrangian of the fields content in the exterior brane ( spin zero ) and @xmath20 is the lagrangian of the fields content in the interior brane ( spin one ) . the bulk can be described with the natural coordinates in the form @xcite
@xmath21 . \label{2}\end{aligned}\ ] ] now we solve the 5d einstein equations @xmath22 with @xmath23 for the most general @xmath24 function , we obtain @xmath25 where @xmath26 is the mass parameter of the ads - s@xmath27 black hole @xcite and @xmath28 is the 5d cosmological constant in the bulk . on the other hand , if we consider the location of the two 3-brane by @xmath29 , @xmath30 parametrized by the time @xmath31 in the brane , we have the induced four dimensional metric given by @xmath32 , \label{4}\ ] ] where the subindex @xmath33 denote the two branes , having in mind that @xmath34 . the functions @xmath35 and @xmath36 are then constrained by @xmath37 being @xmath38 the brane velocity .
the dots represent here differentiation with respect to @xmath31 .
we define the unit normal vector on the two branes @xmath39 such that @xmath40 and @xmath41 , their components are given by @xmath42 following @xcite , @xcite , @xcite we can construct the extrinsic curvature using the equation @xmath43 the no null components of @xmath44 can be written as @xmath45 @xmath46 with the equations of motion of the brane , it follows @xcite , @xcite , @xcite @xmath47 { g}_{\mu \nu } { -[k}_{\mu \nu } { ] = k}_{(5)}^{2 } { t}_{\mu \nu } { , } \label{10}\ ] ] @xmath48 { , } \label{11}\ ] ] where @xmath49 , @xmath50 = k_{\mu \nu } ^{+ } { -k}_{\mu \nu } ^{-}$ ] and @xmath51 is the energy - momentum tensor . using equation
we find @xmath52 where we are supposing that the matter content of the branes can be represented as perfect fluids with density @xmath53 and presion @xmath54 @xcite , @xcite , @xcite .
the equation represents the energy - momentum conservation on the brane , given by @xmath55 we will use the equation to give a physical interpretation of the model .
in this section we write the field equations of the model . first observe that we have to distinguish between three vacuum regions , the first one is located inside the interior brane , we call it region i
, the second one is located between the branes , region ii , and the third vacuum region surrounds the exterior brane , region iii , ( see fig .
[ fig:1 ] ) . now consider the following analogy with an electromagnetic system .
imagine two charged conducting parallel plates with opposite charges , we know that in the region in between the plates , there exist an electric field induced by the charges on the plates , while in the surrounded region , the electric field vanishes . if we now suppose that the branes are the plates and the electric charge is the gravitational field , the analogy tell us that it should exist an extreme difference between the expectation value for the vacuum in the region in between the branes and zero in the exterior region , having in mind that in the linear regime , the electromagnetic and the gravitational fields behave in the same way .
thus , for the region inside of the interior branes ( region i ) , equation becomes @xmath56 since there does not exist a gravitational potential inside of the interior brane .
but in the region in between the branes ( region ii ) , the equation can be written as @xmath57 where @xmath58 is the mass of the interior brane .
if we substitute the equations and into equation , we have @xmath59 on the other hand , inside of the exterior brane ( region ii again ) we have @xmath60 finally , outside of the exterior brane ( region iii ) , the total mass is @xmath61 .
therefore we have @xmath62 again , we substitute the equations and into the equation , to obtain @xmath63 expressions and are the friedman equations on the interior and exterior branes respectively . in both equations , ) and , the third , fourth , fifth and sixth terms are present only if the assumption of @xmath64-symmetry is dropped out . both equations , and , are in agreement with ida et . al @xcite , except for the third , fourth , fifth and sixth terms which are characteristic of the proposed model .
according to the brane world scenario , it is natural to assume that the matter component consists of the brane tension @xmath65 and the ordinary fields @xmath66 and @xmath67 , such that the brane densities are given by @xmath68 @xmath69 where @xmath70 and @xmath71 denote the energy density and the pressure of the matter with spin one interactions before the collision .
after the collision we can interpret @xmath70 and @xmath71 as radiation in the early universe ( high energy regime ) and as baryons plus radiation plus neutrinos , etc .
, in the late universe ( low energy regime ) .
@xmath72 and @xmath73 denote the energy density and the pressure of the self interacting scalar field respectively and @xmath10 is the constant brane tension @xcite .
now we assume two scenarios on the branes , the first one corresponds to very big densities and the second one when they are very small in comparison with the tension @xmath10 , that is * @xmath74 * and @xmath75 , @xmath76 .
first we analyse the branes when @xmath74 , @xmath76 .
this scenario corresponds to the universe before the collision .
we assume the next fine tuning conditions on the interior brane @xmath77 , @xmath78 , with @xmath79 where @xmath80 is of the order of magnitude of the quantum fluctuations of the vacuum predicted by the standard model @xcite .
we replace the previous conditions and equation in equation to obtain @xmath81 \notag \\ & & + \frac { { 3m}^ { { 2 } } } { { m } _ { { ( 5)}}^ { { 6 } } { a } _ { { 1}}^ { { 8 } } { \kappa } _ { { ( 4)}}^{2}}\frac { { \sigma } } { { 4\sigma \rho } _ { s { f}}\left ( { 1 + } \frac { { \rho } _ { s { f } } } { { 2\sigma } } \right ) { + 2\sigma } ^ { { 2 } } } { , } \label{22}\end{aligned}\ ] ] with @xmath82 this is because in the high energy limit it follows that @xmath83 for simplicity we assume the ansatz @xmath84 with non physical interpretation .
then , imposing @xmath85 and proposing @xmath86 we obtain @xmath87 with @xmath88 in the same way , for the exterior brane we can set the next fine tuning condition @xmath89 . again , if one replace the previous conditions and equation in equation , we obtain @xmath90 \nonumber \\&&+\frac{3m^{2}}{m_{(5)}^{6}a_{2}^{8}\kappa _ { ( 4)}^{2}}\frac{\sigma } { 4\sigma\rho _ { m}\left ( 1+\frac{\rho _ { m}}{2\sigma } \right ) + 2\sigma ^{2 } } , \label{26}\end{aligned}\ ] ] with @xmath91 again , we have set the limit condition @xmath92 thus , in the limit where @xmath93 and @xmath86 , we obtain @xmath94 with @xmath95 the relations and are the friedmann equations for the early universe of the model .
observe that the density @xmath96 appears quadratic in these equations , thus , the brane containing the scalar field inflates , even with a very small scalar field mass and collides with the exterior one , where the matter content is only of spin one interactions and the brane expands without inflation . in the same way we analyse the regime @xmath97 , @xmath76 for the late universe using the same conditions as in the previous regime .
we obtain the next two relations for the branes 1 and 2 .
@xmath98 \notag \\ & + & \frac { { 3m}^ { { 2 } } } { { m } _ { { ( 5)}}^ { { 6 } } { a } _ { { 1}}^ { { 8 } } { \kappa } _ { { ( 4)}}^{2}}\frac { { \sigma } } { { 4\sigma \rho } _ { { sf}}\left ( { 1+}\frac { { \rho } _ { { sf } } } { { 2\sigma } } \right ) { + 2\sigma } ^ { { 2 } } } { , } \label{30}\end{aligned}\ ] ] @xmath99 \nonumber \\&&+\frac{3m^{2}}{m_{(5)}^{6}a_{2}^{8}\kappa _ { ( 4)}^{2}}\frac{\sigma } { 4\sigma\rho _ { m}\left ( 1+\frac{\rho _ { m}}{2\sigma } \right ) + 2\sigma ^{2 } } , \nonumber \label{31}\end{aligned}\ ] ] with @xmath100 in the same way as before , this is because @xmath101 .
then , imposing the condition @xmath75 , @xmath76 and again @xmath86 , we obtain @xmath102 where now we find the remarkable result that @xmath103 thus , we can fix the cosmological constant @xmath104 such that @xmath105 have the value of the classical observed 4-dimensional cosmological constant @xcite .
the relations and are the friedmann equations for the late time universe and give the dynamics of the branes .
from the similitude of equations and we observe that both branes could have a very similar dynamics in the late time universe .
we conclude that we can fix the free parameter of the model in order that both branes expand together .
now we can follow the dynamics of the branes .
the main idea of the model we are dealing with here is not to fix the initial conditions @xmath106 , @xmath107 else , etc . instead of that we set the initial topology of the model as follows .
suppose there exist two concentric @xmath0 branes , in the interior one with a very small radius lives a scalar field ; in the exterior brane live the particles of spin one .
the matter contained in the exterior brane has a state equation @xmath108 , which density evolve as @xmath109 with the equation ( [ 13 ] ) . during the early universe the expansion of the brane follows equation ,
thus the exterior brane evolves as @xmath110 or , if we derive this expression with respect to @xmath111 , we arrive at @xmath112 if during this period the exterior brane is dominated by matter , the first term goes like @xmath113 , if this brane is dominated by radiation , this term goes like @xmath114 . in any case , during this period the first term dominates over the other two , indicating that this brane starts with a very big negative acceleration but decreasing its acceleration very fast . on the other brane
the situation is very different due to the fact that the interior brane is permeated with a scalar field , its evolution is governed by equation . during the epoch
when the density is enough big , such that the two last terms on the right hand side of equations ( [ 24 ] ) can be neglected , in the slow - roll approximation this equation reduces to @xmath115 , \label{38}\ ] ] @xmath116 where @xmath117 is the scalar field potential .
then , using the two slow - roll parameters @xmath118 , \label{41}\end{aligned}\ ] ] if , for example , the scalar field potential is an exponential one @xmath119 , @xmath120 is given by @xcite @xmath121 during this period @xmath122 @xcite , thus @xmath123 and the brane inflates till @xmath124 , this is , till the scalar field potential reaches the value @xmath125 .
if the potential is @xmath126 , the slow - roll parameter reads @xmath127 that means that if @xmath128 , inflation ends when the scalar field reaches the value @xmath129 . for
both potentials , the interior brane inflates and collide with the exterior one , heating the branes very fast .
the heating of the branes essentially depends on the interaction constant between the scalar field and the matter and , of course , on the speed of the collision .
inflation in the interior brane also grows the quantum fluctuations of the scalar field , in such a way that after inflation ends , the interior brane contains a spectrum of semiclassical potential wells which become the seeds for the structure formation in the exterior brane . between these two regimes , during the branes collision , the physics of the system is just like in the ekpyrotic model @xcite . during the collision , quantum gravitational effects due to the interactions of both space - times
take place . from this moment ,
both branes follow the dynamics of equations and , indicating that both branes expand with the same dynamics .
the scalar field riches the minimum of its potential implying that , from this period on , the scalar field potential has a behaviour like @xmath130 .
the scalar field in the interior brane induces potential wells into the exterior one via gravitational interaction , which evolve as cold dark matter ( see @xcite ) .
as we have explained before , the scalar field confined on the interior brane produces potential wells along the whole brane , caused by the extreme growing of the scalar field fluctuations during inflation . on the exterior brane , matter feels this potential wells as an extra gravitational force , but it can not detect the scalar field in any other way . in order to fit cosmological and galactic observations , the scalar field mass measured in the exterior brane must be ultra - light , such that @xmath131ev .
this ultra - light mass causes , of course , a problem of hierarchy , however this problem can be solved using the prescription of randall - sundrum @xcite as follows . the scalar field part of the action on the interior brane reads @xmath132 \label{43}.\ ] ] using normal gaussian coordinates for the ads metric @xmath133 , action ( [ 43 ] ) translated into the exterior brane reads @xmath134 , \label{44}\ ] ] with the identification @xmath135 and @xmath136 , where @xmath137 .
as long as the two branes expand together , @xmath138 remains constant after the collision .
thus , with this prescription it is enough that @xmath139 , for instance , with the scalar field mass @xmath140 @xcite .
that means , we obtain the required ultra - light mass of the scalar field on the exterior brane through a scalar field on the interior one .
furthermore , this ultra - light mass sets a minimum distance between the branes . with this mass , that means , with only one free parameter , the scalar field on the exterior brane behaves
exactly in the same way as scalar field dark matter ( sfdm ) @xcite with the following important features : the ultra - light scalar field mass ( @xmath141ev ) fits : 1 . the evolution of the cosmological densities @xcite .
2 . the rotation curves of galaxies @xcite and the central density profile of lsb galaxies @xcite , 3 . with this mass ,
the critical mass of collapse for a real scalar field is just @xmath142 , i.e. , the one obeserved in galaxies haloes @xcite .
4 . the central density profile of the dark matter is flat @xcite .
5 . the scalar field has a natural cut off , thus the substructure in clusters of galaxies is avoided naturally . with a scalar field mass of @xmath143ev the amount of substructure is compatible with the observed one @xcite .
sfdm forms galaxies earlier than the cold dark matter model , because they form bose - einstein condensates at a critical temperature @xmath144tev .
so , if sfdm is right , we have to see big galaxies at big redshifts .
the question we are facing on here is the following ; is it possible that the universe has the dynamics we observe because of its global topology ?
this question makes sense if we start from the 4d general einstein equations in vacuum , @xmath145 .
this equation is completely geometrical , there is no matter content in it .
the friedman equation for this case is @xmath146 where @xmath147 for a flat , @xmath0 or @xmath5 topology respectively . or , in a more convenient form for our goal here , we can write it as @xmath148 which can be seen as a @xmath149 dynamical equation .
we observe that even when there is no matter , the space - time posses a dynamics .
even if there is no cosmological constant @xmath150 , it is enough that @xmath151 to have a dynamical universe . in other words , it is sufficient that at the beginning , the global topology is fixed to have a specific dynamics . in this work
we consider a special global topology of the universe , namely , two concentric spherical branes and some content of matter with different interactions in each region of the universe : fundamental interactions of spin zero confined in the interior brane , fundamental interactions of spin one confined in the exterior brane ( the brane that will lead to the standard model ) and gravitation in the bulk .
we find that the spin zero interactions inflate the interior brane , making this brane to collide with the exterior one due to the concentric topology , we assume the branes heat because of a small interactions constant between the spin zero and the spin one particles .
after the collision , both branes expand together as shown by equations and .
the scalar field fluctuations build potential wells which are seen as sfdm in the exterior brane via the rs prescription , with the idea showed in the equations ( [ 43 ] ) , ( [ 44 ] ) .
however , in the exterior brane we expect that the collision provokes the existence of particles with fractional spin and spin one , giving the zoo of particles predicted by the standard model like in the ekpyrotic or cyclic models @xcite .
this idea is very interesting , but so far there does not exist mathematical evidence to back it up .
it is important to remark that during the collision it is impossible to analyse all the interactions with the classical theory of general relativity due to the quantum - gravitational effects and the interactions between the fields that permeates both branes .
thus , in this model the universe expands as we see it , but the particles of the standard model can not interact with the scalar field particles because they are confined in the interior brane and the standard model ones are confined in the exterior one .
the five - dimensional cosmological constant @xmath2 is pure geometrical and can be seen as a second free constant of this toy model .
the extreme difference between the expectation values in the interior and exterior branes can be explained using an analogy with electromagnetism .
if the fields in the 5d bulk produced by the branes , behaves like a electromagnetic field , remember that for a weak field , linearised einstein equations are analogous to the electromagnetic one , then there exist a huge expectation value inside of the region between the two branes which is associated with @xmath152 , while in the exterior region the expectation value is zero .
the benefit of working with spherical branes is the natural acceleration predicted by this kind of models where the acceleration starts at redshift @xmath153 in agreement with the observations @xcite .
it is important to see that the acceleration does not depend much on the particular matter content , only on the geometrical features @xcite . in future
works it is necessary to analyze the perturbations on the exterior and interior branes @xcite , we expect that the scalar field fluctuations reproduce the large scale structure , cmb @xcite , sachs - wolfe effect @xcite etc . in the same way it is necessary to generalize the rs prescription in a global point of view where there exist a natural expansion and acceleration of the branes .
we want to acknowledge the enlightening conversation with roy maartens during his visit in leon guanajuato , mxico and many helpful discussions with ruben cordero , abdel prez lorenzana and luis urea .
this work was partially supported by conacyt mxico , under grants 49865-f , 216536/219673 and i0101/131/07 c-234/07 , instituto avanzado de cosmologia ( iac ) collaboration .
cordero ruben and vilenkin alexander . , _ et al_. , phys .
d * 65 * , 083519 .
arxiv.hep-th/0107175 randall lisa and sundrum raman . , _ et al_. , phys .
lett . * 83 * 3370 - 3373 ( 1999 ) , arxiv : hep - ph/9905221v1 rsnen s. , ph.d .
thesis , ( 2002 ) .
arxiv : astro - ph/0208282v2 alcubierre miguel , siddhartha guzmn f. , matos tonatiuh , nez daro , urea luis a. and wiederhold petra . class .
* 19 * 5017 - 5024 ( 2002 ) , gr - qc /0110102 .
bernal argelia , matos tonatiuh and nez daro . ,
* 44 * , 149 - 160 ( 2008 ) , arxiv : astro - ph/0303455 carroll sean m. living reviews in relativity , ( 1999 ) , arxiv : astro - ph/0004075v2 coley alan a. proceedings of ere-99 , arxiv : gr - qc/9910074v1 copeland edmund j , liddle andrew r. and lidsey james e _ et al_. , phys .
d * 64 * , 023539 ( 2002 ) .
dodelson scott . academic press , copyright 2003 , elsevier dvali g. , gabadadze g. , porrati m. _ et al_. , phys .
b * 485 * 208 - 214 , ( 2000 ) , arxiv : hep - th/0005016v2 gogberashvili merab . , _
et al _ phys .
b * 636 * 147 - 149 ( 2006 ) arxiv : gr - qc/0511039v2 ida d. extra large dimensions , classical theories of gravity jhep09 ( 2000 ) 014 .
khoury , ovrut , steinhardt , turok . , _ et al_. , phys .
d * 64*:123522 ( 2001 ) matos tonatiuh and siddhartha guzmn f. _ et al_. , class .
17 * , l9-l16 ( 2000 ) , arxiv : gr - qc /9810028 matos tonatiuh , vazquez alberto and magaa j.a .
et al _ mnras . , * 389 * , 13957 ( 2009 ) arxiv:0806.0683 matos tonatiuh and urea luis a. , _ et al_. , phys rev .
d * 63 * , 063506 ( 2001 ) , arxiv : astro - ph/ 0006024 langlois d. , maartens r. , sasaki m. , wands d. _ et al_. , phys .
d * 63 * , 084009 ( 2001 ) , arxiv : hep - th/0012044 liddle a. and ure - lpez l. , _ et al _ . ,
lett . * 97 * , 161301 ( 2006 ) .
lidsey james , matos tonatiuh and urea luis a. _ et al_. , phys rev .
d * 66 * , 023514 ( 2002 ) , arxiv : astro - ph/ 0111292 maartens roy . , _
et al _ living rev .
rel.7:7 , ( 2004 ) , arxiv : gr - qc/0312059v2 maartens r. , wands d. , bassett b.a . and heard i.p.c .
et al _ phys .
d * 62 * , 041302 ( 2000 ) .
mac fadden p. , phd thesis .
arxiv : hep - th/0612008v2 niz quevedo gustavo . ,
phd thesis , ( 2006 ) lorenzana abdel . , _ et al_. , j.phys.conf.ser .
* 18 * 224 - 269 ( 2005 ) , arxiv : hep - ph/0503177v2 ] rodriguez montoya ivan and matos chassin tonatiuh . ,
( 2009 ) arxiv:0908.0054v1 [ astro-ph.co ] shiromizu , maeda , sasaki . , _
et al _ phys .
d * 62 * 024012 ( 2000 )
|
we explore the possibility that the dynamics of the universe can be reproduced choosing appropriately the initial global topology of the universe . in this work
we start with two concentric spherical three - dimensional branes @xmath0 , with radius @xmath1 immersed in a five - dimensional space - time .
the novel feature of this model is that in the interior brane there exist only spin - zero fundamental fields ( scalar fields ) , while in the exterior one there exist only spin - one fundamental interactions . as usual ,
the bulk of the universe is dominated by gravitational interactions . in this model , like in the ekpyrotic one , the big bang is consequence of the collision of the branes and causes the existence of the particles predicted by the standard model in the exterior brane ( our universe ) .
the scalar fields on the interior brane interact with the spin - one fields on the exterior one only through gravitation , they induce the effect of scalar field dark matter with an ultra - light mass on the exterior one .
we discuss two different regimes where the energy density and the brane tension are compared , with the aim to obtain the observed dynamics of the universe after the collision of the branes .
|
the nflation model of dimopoulos et al .
@xcite corresponds to a collection of uncoupled massive fields which drive inflation via the assisted inflation mechanism @xcite .
the existence of multiple fields is motivated by the axions of string theory , and their presence enables sufficient inflation to be obtained without using super - planckian field values .
models of this type had first been considered by kanti and olive @xcite and then kaloper and liddle @xcite in the context of kaluza klein models .
they showed the massive fields evolve faster to the minima of their own potential and light ones later .
easther and mcallister @xcite introduced random matrix theory as a way of computing the possible distribution for masses in the nflation model .
for some related constructions see ref .
@xcite .
it is obviously important to develop observational predictions from such models .
this has been thought difficult , because in multi - field models the predictions depend in general upon the field initial conditions as well as the model parameters .
easther and mcallister @xcite only studied two types of initial conditions , where either the field values or the field energy densities were equal .
neither is well motivated physically .
however , in the case of an exponential mass spectrum , kim and liddle @xcite showed that provided there were enough fields , with randomly - chosen initial conditions , the observational predictions become essentially _ independent _ of initial conditions again .
the reason is that with enough fields , the space of possible initial conditions is well sampled by a single realization , and they named this the ` thermodynamic regime ' . the purpose of this article is to apply the random initial conditions approach of ref .
@xcite to the random matrix mass spectrum of ref .
@xcite , in order to explore its observational predictions and test its viability .
the main observables are the density perturbation spectral index @xmath0 , the tensor - to - scalar ratio @xmath1 , and the non - gaussianity parameter @xmath2 .
some quite general results are already known , applying to arbitrary mass spectra and initial conditions provided sufficient inflation is obtained .
alabidi and lyth @xcite showed that @xmath1 always has the same value as in the single - field case , and kim and liddle @xcite showed that the same was true of @xmath3 ( this having previously been shown for two fields in ref .
@xcite ) . only @xmath0 has model and initial conditions dependence , and it has been shown that its predicted value can not be larger than the single - field value @xcite .
there are some further generalisations of these results @xcite .
once one considers the scalar fields in inflation as axions in string theory , their masses can be written in matrix form , which depends on specific details on compactification ( following ref .
@xcite , we also assume that higher - order terms can be neglected , so that what are really cosine functions can be approximated as massive uncoupled fields ) .
the shape of the mass distribution depends only on the basic structure of the mass matrix , which is specified by the supergravity potential . in the simplest assumption ,
the entries in the mass spectrum are independent and identically distributed , i.e. a random matrix .
the fields can be uncoupled by diagonalization of this matrix , with the mass spectrum given by the distribution of eigenvalues .
the distribution of the eigenvalues for random matrices of this kind is characterized by the marcenko
pastur law @xcite when the matrices are large .
the distribution function depends on a parameter @xmath4 , the ratio of the number of axions to the dimension of the moduli space .
easther and mcallister @xcite devised this formalism and computed the observational predictions in terms of @xmath4 for specific choices of initial conditions for the fields where the field values or the energy densities are identical ( see also ref .
@xcite ) .
we follow their notation for the mass spectrum , and label the average value of the mass - squared @xmath5 .
we will throughout put @xmath6 where @xmath7 is the reduced planck mass ; a simple rescaling of @xmath8 would be sufficient to match the observed normalization of perturbations .
the mass spectrum is determined by two quantities , the number of fields @xmath9 and a parameter @xmath4 which governs its shape .
the shape parameter @xmath4 lies in the range zero to one , with values around @xmath10 perhaps the most plausible @xcite .
figure [ f : mbeta ] shows the shape of the spectrum for @xmath11 and some values of @xmath4 . the mass - squared distribution for 1000 fields ; the dashed - dotted ( black ) flat line is for @xmath12 , the dotted ( blue ) for @xmath13 , the solid ( red ) for @xmath14 and the dashed ( green ) for @xmath15.,width=283 ]
for a set of uncoupled fields with quadratic potentials , the number of @xmath16-foldings @xmath17 , in the slow - roll approximation , is @xcite @xmath18 where @xmath19 is the @xmath20-th field . for our initial results ,
we choose the field initial values randomly from a uniform distribution in the range @xmath21 to @xmath22 .
the total number of @xmath16-foldings is then accurately given by the linear relation @xmath23 @xcite , this result holding for an arbitrary mass spectrum .
we will throughout assume the observable scales crossed outside the horizon 50 @xmath16-foldings before the end of inflation .
accordingly , sufficient inflation requires a minimum of around 700 fields .
in passing we note that increasing the number of fields increases the energy scale at the end of inflation , and hence might increase the number of @xmath16-foldings relevant to observable perturbations .
however this effect is well within current uncertainties from the unknown behaviour of the universe between inflation and nucleosynthesis .
we follow the usual formulae for the observational predictions of @xmath24 , @xmath25 , @xmath1 , and @xmath2 .
these are @xcite @xmath26\ , , \label{e : ns}\\ & \simeq & 1 - \frac{1}{n}-4 m_{\rm pl}^2 \left [ \frac{\sum_i m_i^4 \phi_i^2}{(\sum_k m_k^2 \phi_k^2)^2}\right]\ , , \label{e : ns1}\\ r & \simeq & \frac{32 m_{\rm pl}^2}{\sum_i \phi_i^2 } \simeq \frac{8}{n}\,,\label{e : r}\\ -\frac{5}{6}f_{\rm nl}^{(4 ) } & \simeq & \frac{2 m_{\rm pl}^2}{\sum_i \phi_i^2 } \simeq \frac{1}{2n } \simeq \frac{r}{16}\ , , \label{e : fnl}\end{aligned}\ ] ] where @xmath27 and @xmath28 are the @xmath20-th mass and the second term of the nonlinearity parameter @xmath2 respectively ( the first contribution to @xmath2 is model - independent and small @xcite ) . more detailed calculations of the non - gaussianity have confirmed that this is the leading contribution to the non - gaussianity @xcite .
we have simulated these observables and will show in particular @xmath29 to study the effect from this mass distribution .
the most important observable in nflation models is the spectral index , which we calculated as described above and show in fig .
[ f : n ] .
we see that it depends on both model parameters @xmath9 and @xmath4 . the spectral index @xmath25 ( from top to bottom , @xmath4=0 , 0.1 , 0.3 , 0.5 , 0.7 , 0.8 , 0.9 and 0.95 ) .
the dotted line shows the observational lower limit on @xmath25 from wmap3.,width=283 ] the curves have a generic shape where they first dip , and then , after a minimum typically around @xmath30 , increase to join the single - field value .
the single - field value is always obtained in the case @xmath12 , in which all masses become identical and this result is well known ( e.g. ref .
these curves look a little different from fig . 4 in ref .
@xcite , studying an exponential mass spectrum where the values asymptoted to constants for large @xmath31 .
this is just because @xmath9 is defined here in a rather different way ; in ref .
@xcite increasing it added new fields at the top of the mass range , and if they were heavy enough they fell to their minima before observable scales left the horizon .
the distinction here is that when we increase @xmath9 we are packing more fields into the same mass interval , forcing us to the equal - mass case . that the curves begin to rise after @xmath32 indicates that fields heavier than the 1500th have typically reached their vacuum state before the 50 @xmath16-foldings point . to compare with observations , the analysis in ref .
@xcite shows that for @xmath33 , the 95% confidence lower limit for @xmath25 lies at about 0.93 , shown as the dotted line in the figure .
provided @xmath34 , the spectral index is always large enough regardless of the number of fields . for large @xmath4
, the model is excluded only for a range of @xmath9 running from about one thousand to several thousand .
accordingly , most of the model parameter space is currently viable .
the results shown in fig . [ f : n ] are the mean values over realizations of the initial conditions . additionally we find that the spread in @xmath25 values is quite small ; the standard deviation of @xmath25 is never more than one percent of its displacement from unity .
this confirms the existence of the thermodynamic regime for this mass spectrum , just as in ref .
@xcite . as fig .
[ f : n ] , but now showing sets of curves for two choices of initial condition distributions .
the dashed lines reproduce the curves from fig .
[ f : n ] , and the solid ones show initial conditions with @xmath35 drawn from a uniform distribution.,width=283 ] although the above analysis shows that for our chosen initial distribution , the observational predictions are independent of realization , one might further ask whether there is dependence on the choice of that distribution . in order to test that
, we carried out two further series of simulations . in the first series
, we took @xmath35 to be chosen uniformly between zero and one .
the results are shown in fig .
[ f : comp ] .
the curves are shifted to a smaller number of fields , retaining both their shape and minima .
we found that when the number of fields is multiplied by 3/2 , making the total number of @xmath16-foldings the same , these solid curves become matched to the dotted ones .
this can be related to the different initial distributions by an approximate analytic argument given in the appendix . as fig .
[ f : n ] , but showing results for uniformly - chosen field energy densities .
only @xmath36 , @xmath37 , @xmath38 , @xmath39 and @xmath40 ( from top to bottom ) are shown .
the left - hand end of the curves corresponds to models only just achieving 50 @xmath16-foldings in total.,width=283 ] in the second series , we chose the field energy densities to be uniformly distributed , with the lightest field ranging from @xmath21 to @xmath22 and the other fields given smaller ranges to ensure the same expected energy density .
this is distinct from ref .
@xcite who gave each field the _ same _ energy density , rather than the same only on average .
in this case the more massive fields start closer to their minima , and a greater total number of fields is needed to get sufficient inflation . the results are shown in fig .
[ f : ne ] .
the curves have a rather different shape , and there is some difficulty in achieving a minimum of 50 @xmath16-foldings unless @xmath41 is large .
the analytic analysis given in the appendix does not apply in this case , as the field initial values are correlated to the masses .
nevertheless the overall conclusion is the same : that only for large @xmath4 is there a danger that the models are ruled out through predicting too small a value for @xmath0 , and even then only in a narrow range of @xmath41 .
the expected value @xmath42 is therefore viable for all initial condition distributions we tested . concerning the tensor - to - scalar ratio @xmath1 and the non - gaussianity parameter @xmath2
, unsurprisingly the results are same as in our previous work @xcite . comparing to the analytical results of eqs .
( [ e : r ] ) , ( [ e : fnl ] ) , we confirm numerically that @xmath1 is indeed independent of the model parameters , @xmath43 , and that @xmath44 which will never be sufficiently large to be detected .
all simulations of @xmath1 and @xmath2 individually give those results , without needing to average over realizations of the initial conditions .
we have carried out a detailed numerical investigation of the inflationary perturbations in nflation models with the mass spectrum of random matrix theory .
we produced the predictions for the spectral index as a function of both the number of fields @xmath9 and the distribution parameter @xmath4 in the mass spectrum .
we have found that the model remains viable in the majority of its parameter space , and that the thermodynamic regime , where the predictions become independent of the initial condition realization , holds for this spectrum as for the exponential case @xcite .
a.r.l . was supported in part by pparc ( uk ) .
s.a.k . acknowledges the hospitality of filippo vernizzi and the abdus salam ictp in italy , and a.r.l . of the institute for the astronomy , university of hawaii , while part of this work was being carried out .
we thank richard easther for providing his code generating the mp mass spectrum , and for many discussions relating to this work including the analytic evolution formulation described in the appendix .
this appendix is based on unpublished work by richard easther , whom we thank for providing details . in the main text we considered different initial distributions for the scalar fields . when the initial values of @xmath45 and @xmath35 are randomly chosen uniformly between zero and one , the expectation value @xmath46 is 1/3 and 1/2 respectively ( other choices will give a different @xmath47 ) .
( [ e : efold ] ) becomes @xmath48 .
this difference by a factor 2/3 is closely related to the shift in the curves shown in fig . [
f : comp ] . this can be shown via an analytic approximation to the evolution , in order to extract estimates of the summations @xmath49 and @xmath50 which appear in the expression for the spectral index . using the slow - roll approximation , the field value is @xmath51 where @xmath52 and @xmath53 have the same definition as in ref .
@xcite : they are the ratio of the value of the heaviest field at the time @xmath54 to its initial value , and the upper limit of the probability distribution of the mass - squared spectrum , respectively . also @xmath55 is the initial time . then using the exponential function with @xmath56/b$ ] , we can rewrite the summation terms such as @xmath57\,.\ ] ] if one ignores correlations between the mass distribution and the initial field distribution , so that @xmath58 \phi_i^2(t_0 ) \rangle = \langle m_i^2 \exp[m_i^2 c ] \rangle \ , \langle \phi_i^2(t_0 ) \rangle$ ] , then with the marcenko pastur distribution , the expectation value is given by @xmath59 where @xmath60 is the hypergeometric function .
this follows from eq.(6.14 ) of ref .
@xcite , where the summation can be rewritten as a hypergeometric function . therefore the summation terms with the expectation values of field initial conditions and of distribution of mass spectrum are @xmath61 note that @xmath54 is related to @xmath62 ; that is , @xmath54 counts forward from when the fields start to evolve from their own initial conditions , while @xmath17 is the number of @xmath16-foldings before the end of inflation . the @xmath25 data in fig .
[ f : comp ] are taken at different times , but at the same @xmath17 .
using eqs .
( [ e : m2phi2 ] ) and ( [ e : m4phi2 ] ) , the @xmath25 formula becomes @xmath63 where @xmath64 ^ 2 } \,.\ ] ] here the function @xmath65 represents how @xmath25 evolves with respect to time ( or @xmath17 ) .
even though it is hard to evaluate it in general , it will have the same value whenever @xmath54 and @xmath4 are same .
hence in cases with same @xmath66 , but not necessarily the same @xmath9 or @xmath47 , then @xmath25 must be same .
this explains why the solid curves in fig . [ f : comp ] are matched with dotted ones after multiplying 3/2 to @xmath9 , because this makes @xmath66 the same .
different @xmath47 shifts the @xmath25 curves along the @xmath9 axis , with bigger @xmath47 shifting the curves in the smaller @xmath9 direction .
we can conclude then that , insofar as the approximations hold , @xmath67 should satisfy the current observations regardless of the field initial condition distribution , while for larger @xmath4 some choices of @xmath9 will be ruled out and that those @xmath31 values depend on the @xmath47 value of the initial condition distribution .
however the above analysis relies on using a slow - roll approximation for all fields , which is likely to become increasingly inaccurate as increasing numbers of fields evolve towards their minima , and assumes the mass and initial field value distributions to be uncorrelated .
the latter approximation fails badly for the uniform field energy density initial conditions .
50 s. dimopoulos , s. kachru , j. mcgreevy , and j. wacker , hep - th/0507205 .
a. r. liddle , a. mazumdar , and f. e. schunck , phys .
rev . d*58 * , 061301(r ) ( 1998 ) , astro - ph/9804177 .
p. kanti and k. a. olive , phys .
rev . d*60 * , 043502 ( 1999 ) , hep - ph/9903524 ; p. kanti and k. a. olive , phys . lett .
b*464 * , 192 ( 1999 ) , hep - ph/9906331 .
n. kaloper and a. r. liddle , phys .
rev . d*61 * , 123513 , ( 2000 ) , hep - ph/9910499 .
r. easther and l. mcallister , jcap * 0605 * , 018 ( 2006 ) , hep - th/0512102 .
a. jokinen and a. mazumdar , phys .
b*597 * , 222 ( 2004 ) , hep - th/0406074 ; k. becker , m. becker , and a. krause , nucl .
b*715 * , 349 ( 2005 ) , hep - th/0501130 .
s. a. kim and a. r. liddle , phys .
rev d*74 * , 023513 ( 2006 ) , astro - ph/0605604 .
l. alabidi and d. h. lyth , jcap * 0605 * , 016 ( 2006 ) , astro - ph/0510441 .
s. a. kim and a. r. liddle , phys .
rev d*74 * , 063522 ( 2006 ) , astro - ph/0608186 .
f. vernizzi and d. wands , jcap * 0605 * , 019 ( 2006 ) , astro - ph/0603799 .
d. h. lyth and a. riotto , phys
. rep . * 314 * , 1 ( 1999 ) , hep - ph/9807278 .
piao , phys .
d*74 * , 047302 ( 2006 ) , gr - qc/0606034 .
v. a. marcenko and l. a. pastur , math .
sbornik * 1 * , 457 ( 1967 ) .
gong , phys .
d*75 * , 043502 ( 2007 ) .
j. m. maldacena , jhep * 0305 * , 013 ( 2003 ) , astro - ph/0210603 .
d. n. spergel et al .
( wmap collaboration ) , astrophys .
* 170 * , 377 ( 2007 ) , astro - ph/0603449 .
t. battefeld and r. easther , jcap * 0703 * , 020 ( 2007 ) , astro - ph/0610296 ; d. battefeld and t. battefeld , jcap * 0705 * , 012 ( 2007 ) , hep - th/0703012 .
|
we carry out numerical investigations of the perturbations in nflation models where the mass spectrum is generated by random matrix theory .
the tensor - to - scalar ratio and non - gaussianity are already known to take the single - field values , and so the density perturbation spectral index is the main parameter of interest .
we study several types of random field initial conditions , and compute the spectral index as a function of mass spectrum parameters .
comparison with microwave anisotropy data from the wilkinson microwave anisotropy probe shows that the model is currently viable in the majority of its parameter space .
|
wolf - rayet ( wr ) stars are characterized by broad emission lines that are indicative of fast stellar winds and high mass loss rates .
the fast stellar wind sweeps up the ambient medium into a dense expanding shell , called a wind - blown bubble .
wind - blown bubbles in a homogeneous medium have been modeled by , for example , castor , mccray , & weaver ( 1975 ) , steigman , strittmatter , & williams ( 1975 ) , and weaver et al .
however , these models can not be readily applied to bubbles blown by wr stars because wr stars are evolved massive stars and their progenitors mass loss has drastically modified their gaseous surroundings .
the physical conditions of a wr star s ambient medium are highly dependent on the star s evolutionary history and mass loss history .
wr stars are divided into wn ( nitrogen ) and wc ( carbon - oxygen ) sequences ; the wn sequence is further divided into excitation classes wn29 and the wc sequence into wc49 ( van der hucht et al . 1981 ) .
these diverse types of wr stars have different initial masses and evolutionary paths ( langer et al .
it is conceivable that ring nebulae around wr stars must have a variety of origins and physical conditions .
the formation of wr bubbles has been calculated both analytically ( garca - segura & mac low 1995a ) and numerically ( garca - segura & mac low 1995b ) .
hydrodynamic models of bubbles have been produced for wr stars descendent from luminous blue variables ( lbvs ; garca - segura , mac low , & langer 1996 , hereafter gml96 ) and from red supergiants ( rsgs ; garca - segura , langer , & mac low 1996 , hereafter glm96 ) , respectively .
these models conclude that the abundances , morphology , and kinematics of a wr bubble can be used to diagnose the evolutionary status of the bubble and to determine whether the central wr star is a descendant of a lbv ( e.g. , rcw 58 ) or a rsg ( e.g. , ngc6888 ) .
it is of interest to compare these models to a large number of wr bubbles to determine the nature of the bubbles and the applicability of the models .
recent ccd surveys have found many new wr ring nebulae in our galaxy ( miller & chu 1993 ; marston et al . 1994a , 1994b ) and in the large magellanic cloud ( lmc ; dopita et al .
these ring nebulae provide excellent samples of wr bubbles for comparisons with models .
we have begun to study the lmc sample because the extinction is small , the distance is well - known , and the lmc wr bubbles can be easily studied with long - slit spectrographs .
table 1 lists the most well - defined small wr ring nebulae in the lmc .
the wr stars br 2 , br 10 , br 13 , br 48 , br 52 , and br 100 were from breysacher ( 1981 ) , and the star br 40a was from conti & garmany ( 1983 ) .
we have obtained high - dispersion echelle / ccd observations for five of these small wr rings to determine their dynamic structures and to diagnose their nature . in this paper
we report the observations ( 2 ) , interpret the data ( 3 ) , and compare the lmc wr ring nebulae to the hydrodynamic models of gml96 and glm96 , and compare them to the ring nebulae around galactic wr stars , lbvs , and blue supergiants ( 4 ) .
a summary is given at the end ( 5 ) .
we have obtained emission - line images of five small wr ring nebulae in the lmc using ccd cameras on the 0.9 m telescope at cerro tololo inter - american observatory ( ctio ) . in december 1991
, the nebulae around br 2 and br 13 ( breysacher 1981 ) were imaged in the h@xmath0 and [ oiii]@xmath15007 lines with a tektronix 1024@xmath21024 ccd ( tek1024#1 ) .
the pixel size was 04 pixel@xmath3 and the field of view was 68@xmath268 . in january 1996 , the nebulae around br 40a , br 48 , and br 52 were imaged in the [ oiii]@xmath15007 line with a tektronix 2048@xmath22048 ccd ( tek2k3 ) .
the pixel size was 04 pixel@xmath3 and the field of view was 135@xmath2135 .
the exposure time for each image was typically 900 s. the ring nebula around br 13 is so faint that two 900 s exposures were averaged together to produce the final image in each filter .
images of the five lmc wr ring nebulae are presented in figure 1 . note that the h@xmath0 images include contributions of the [ nii ] @xmath46548 , 6583 lines , as the h@xmath0 filter was centered at 6575 with a fwhm of 14 . using the line intensities measured by garnett & chu ( 1994 )
, we find that the [ nii ] contribution to the h@xmath0 images is @xmath51.5% for the br 2 nebula and @xmath57% for the br 13 nebula .
high - dispersion spectroscopic observations of five lmc wr rings were obtained with the echelle spectrograph on the 4 m telescope at ctio in january 1996 .
the spectrograph was used in a long - slit mode by inserting a post - slit h@xmath0 filter ( 6563/75 ) and replacing the cross - disperser with a flat mirror . a 79 lines mm@xmath3 echelle grating and the long focus red camera were used .
the detector was a tektronix 2048@xmath22048 ccd ( tek2k4 ) .
the pixel size was 0.08 pixel@xmath3 along the dispersion and 026 pixel@xmath3 in the spatial axis .
the slit length was effectively limited by vignetting to @xmath6 . both h@xmath0
@xmath16563 and [ nii ] @xmath46548 , 6583 lines were covered in this setup .
the slit - width was 250 @xmath7 m , or 164 , leading to an instrumental fwhm of @xmath814 km s@xmath3 at the h@xmath0 line .
thorium - argon lamp exposures were obtained for wavelength calibration and geometric distortion correction .
the journal of echelle observations is given in table 2 , and the echellograms of the h@xmath0 and the [ nii ] @xmath16583 lines are shown in figure 2 .
each panel covers 38.48 along the dispersion ( the horizontal axis ) , and 2@xmath9 for br 2 and br 13 and 4@xmath9 for br 40a , br 48 , and br 52 along the slit ( the vertical axis ) .
the narrow , unresolved lines are telluric h@xmath0 and oh lines ( osterbrock et al .
1996 ) , which provide convenient references for fine - tuning the wavelength calibration .
the oh 6 - 1 p2(3.5)6568.779 line is unfortunately blended with the lmc h@xmath0 line .
this oh line , corresponding to the h@xmath0 line at a heliocentric velocity of v@xmath10 km s@xmath3 , can be seen on the left side of the h@xmath0 line of the br 52 nebula ( figure 2 ) .
the ring nebula around br 2 was at first identified to be the 3@xmath9 shell nebula to the south of br 2 ( rosado 1986 ) .
it was later discovered in higher resolution images that a 28@xmath11 arc surrounds br 2 ( dopita et al .
this small arc has a morphology similar to that of ngc6888 , but has a much higher excitation and emits heii @xmath14686 line ( pakull 1991 ) .
the contrast between this arc and the background hii region is higher in h@xmath0 images than in [ oiii ] images ( see figure 1 ) .
we assume that the small arc is the bona fide ring nebula of br 2 .
the echelle observations along position angles pa = 120@xmath12 and 135@xmath12 ( near the minor axis ) show a broader h@xmath0 line within the arc ; the fwhm of the h@xmath0 line is 42@xmath13 km s@xmath3 near
br 2 , and 32@xmath13 km s@xmath3 outside the arc .
the centroid of the h@xmath0 line near br 2 is blue - shifted by 3 km s@xmath3 with respect to the background hii region velocity at v@xmath14 = 253@xmath152 km s@xmath3 .
this small velocity shift prohibits an unambiguous decomposition of the broadened line into multiple components .
fortunately , the heii @xmath16560.18 and heii lines are narrow .
we find the heii line to be at [email protected] shorter wavelength than the h@xmath0 line .
for an h@xmath0 wavelength of 6562.78 , the heii line will be at 6560.18 .
] line is detected within and beyond the arc .
the heii line is less confused by the background hii region emission , hence provides better kinematic diagnostics for br 2 s ring .
the heii line is curved . in the background
hii region exterior to the arc , the centroid velocity of the heii line , 250@xmath153 km s@xmath3 , is similar to that of the h@xmath0 line .
near br 2 the heii line is blue - shifted by 16@xmath152 km s@xmath3 with respect to the heii velocity exterior to the arc .
a similar heii velocity structure is seen in the echellogram taken along pa = 45@xmath12 near the major axis the arc .
this velocity structure indicates that the stellar wind of br 2 has accelerated the ambient medium by 16 km s@xmath3 .
the accelerated ambient medium is probably interstellar matter , as it has normal lmc interstellar abundances ( garnett & chu 1994 ) .
it is worth noting that a supernova remnant ( snr ) candidate is detected to the northwest of br 2 s ring along pa = 120@xmath12 and 135@xmath12 .
this snr candidate appears as a high - velocity feature , with velocity offsets up to 150 km s@xmath3 , extending from the br 2 s ring to the northwest for at least 80@xmath16 .
our long - slit low - dispersion spectra of this region also show enhanced [ sii ] and [ oi ] lines characteristic of snrs ( garnett & chu 1994 ) .
the ring nebula of br 13 is elongated along the position angle 33@xmath12 .
its semi - major axis is 20@xmath16 long and its semi - minor axis 16@xmath16 , corresponding to 5 pc and 4 pc , respectively .
only the northwestern half of the ring is detected in our h@xmath0 and [ oiii ] images .
our echelle observation along the e w direction shows a stationary component from the background hii region and a velocity - position ellipse from the ring nebula .
the velocity - position ellipse , indicating an expanding shell , is detected in both the h@xmath0 and [ nii]@xmath16583 lines , although the latter has a much lower s / n ratio . as in the direct images , the surface brightness of the expanding shell is not uniform .
the receding side of the shell is detected for 19@xmath16 on the west side of br 13 and @xmath820@xmath16 on the east , but the approaching side is detected only on the west side .
the largest velocity split , 160 km s@xmath3 , is detected near the shell center . to determine the expansion velocity
, we need to know the systemic velocity of the nebula .
br 13 and its ring nebula are projected against the outskirts of the hii region deml56 ( davies , elliott , & meaburn 1976 ) which has a heliocentric velocity of 304 km s@xmath3 . however , this velocity does not necessarily represent the velocity of br 13 s ring nebula because the ring nebula consists of ejected stellar material ( garnett & chu 1994 ) and br 13 may not have the same origin and systemic velocity as deml56 .
the stellar heii @xmath16560 line profile of br 13 , having a fwhm of @xmath81080 km s@xmath3 and containing an unknown amount of stellar h@xmath0 emission , can not be used to determine the radial velocity of br 13 .
we have to resort to the nebular velocities to assess the systemic velocity of br 13 and its ring nebula .
the echelle data show that the western tip of the ring nebula s velocity - position ellipse converges to the hii region velocity , but the eastern tip is not detected .
the faint velocity - position ellipse in the [ nii ] line appears to be slightly tilted , with the western end toward lower velocity and the eastern end higher velocity .
since the [ nii ] line is detected at a very low s / n , it is difficult to determine the exact amount of line tilt .
if the expansion is symmetric with respect to the central star , the radial velocities of the approaching and receding sides of the shell at the center , 237 and 397 km s@xmath3 , would imply a systemic velocity of 317 km s@xmath3 and an expansion velocity of 80 km s@xmath3 along the line of sight toward br 13 .
br 13 s ring nebula would be moving at + 13 km s@xmath3 relative to the background hii region .
this relative motion may be responsible for the higher surface brightness of the leading side of the shell .
nevertheless , we can not confidently rule out the possibility that br 13 and its ring nebula do have the same systemic velocity as the hii region deml56 , and the expansion of br 13 s circumstellar bubble is asymmetric , with the expansion velocity being 67 km s@xmath3 on the approaching side and 93 km s@xmath3 on the receding side .
br 40a is located in the northwestern quadrant of the hii complex n206 ( henize 1956 ) , or deml221 .
images show a parabolic - shaped arc around br 40a , which has been suggested to be a ring nebula ( dopita et al .
however , our [ oiii ] image shows that the h@xmath0 arc is composed of two filaments with different excitations .
the eastern part of the h@xmath0 arc has such a low excitation that it is not detected in our [ oiii ] image .
the shape of the h@xmath0 arc is probably fortuitous .
our echelle observation has a n s oriented slit passing through br 40a . within the nebula around br 40a ,
no line split is detected , but the centroid of the h@xmath0 line is red - shifted by several km s@xmath3 with respect to that of the background hii region .
the h@xmath0 velocity varies from 232 km s@xmath3 southward of br 40a s ring to 242245 km s@xmath3within the boundary of br 40a s ring , and to 237 km s@xmath3at large distances to the north .
a heii - emission region around br 40a has been reported by niemela ( 1998 ) ; however , our echelle observation did not detect the heii@xmath16560 line .
the velocity pattern of br 40a s ring nebula does not suggest an expanding shell , but does suggest interactions between br 40a and its surrounding medium . the anomalous velocity in the small nebula round br 40a is probably caused by the fast stellar wind accelerating the dense interstellar medium on the far side of the star .
the ring nebula around br 48 is inside the hii region deml231 .
the ring has a
double rim " morphology that resembles a tilted short cylinder .
this distinct morphology is remarkably reminiscent of the ring nebula , although their nature and dynamic structures are completely different .
br 48 s ring has a dimension of 95@xmath17 , or 24@xmath217 pc , and the surrounding hii region is about 50 pc in diameter .
the velocity field of deml231 has been previously studied with the same echelle spectrograph on the same telescope but using photographic plates ( chu 1983 ) . based on those photographic data
, it was deduced that br 48 interacts with a tilted slab of interstellar gas and that no three - dimension expanding shell is present .
this basic picture is still supported by our new observations using a more sensitive detector .
our new echelle observation samples a e w cut through the central star .
an apparent velocity gradient is detected in the hii region , with the heliocentric velocity varying from 311 km s@xmath3 on the east to 297 km s@xmath3 on the west .
within the central cavity , the h@xmath0 line is centered at 295300 km s@xmath3 .
relative to this central velocity , the inner rim of br 48 s ring is blue - shifted at v@xmath14 @xmath8 294 km s@xmath3 on the east side , and red - shifted at v@xmath14 @xmath8 306 km s@xmath3 on the west side .
this indicates that br 48 indeed interacts with a slab of gas and that this slab is tilted with the east side toward us .
no high - velocity components are detected near the central cavity .
the apparent velocity fwhm at the central cavity reaches 45 km s@xmath3 , as opposed to 30@xmath151 km s@xmath3 in the brighter and more quiescent parts of the ring or the hii region .
quadratically subtracting a thermal fwhm of 21 km s@xmath3 ( for 10@xmath18 k ) and an instrumental fwhm of 14 km s@xmath3 , we derive a turbulent fwhm of 37 km s@xmath3 for the central cavity and 16 km s@xmath3 in the bright ( quiescent ) region . if the broadening of velocity profiles at the central cavity is caused by an expansion , the expansion velocity can not be larger than 18@xmath151 km s@xmath3 .
the ring nebula of br 52 was once thought to be the 40@xmath16 triangular - shaped nebula to the east of the star ( chu & lasker 1980 ) .
recent ccd images have detected the fainter part of the ring nebula to the southwest of br 52 ( dopita et al .
the [ oiii ] image in figure 1 shows a complete shell structure , although the southwestern half is fainter .
our new echelle observation has a e w oriented slit centered on br 52 .
the h@xmath0 line is dominated by one bright component .
the heliocentric velocity of this main component is 320 km s@xmath3 outside the ring nebula . within the ring nebula ,
the velocity of the main h@xmath0 component is marginally blue - shifted , by 2 km s@xmath3 , on the west side of br 52 and red - shifted by up to 5 km s@xmath3 on the east side of br 52 .
an additional faint , red - shifted h@xmath0 component is present within the ring nebula on the west side of br 52 .
the heliocentric velocity of this component reaches 370 km s@xmath3 , which is offset from the main component by + 50 km s@xmath3 .
the velocity structure of br 52 s ring nebula indicates a blister " structure .
this ring nebula is most likely an interstellar bubble blown by br 52 in a medium with a steep density gradient .
the bubble / blister expands into the lower - density medium with an v@xmath19 of @xmath850 km s@xmath3 .
the formation of a wr ring nebula can be qualitatively described by the following scenario ( gml96 ; glm96 ) .
a massive star evolves off the main sequence , passes through a rsg phase or a lbv phase , then lands on a wr phase . during the main sequence stage ,
the fast stellar wind sweeps up the ambient interstellar medium to form an interstellar bubble .
the copious mass loss during the rsg or lbv phase would form a circumstellar envelope within the central cavity of the main sequence bubble . as the central star evolves into a wr star
, the fast wr wind compresses the circumstellar envelope of previous rsg or lbv wind to a dense shell , forming a circumstellar bubble . as the circumstellar bubble expands past the outer edge of the circumstellar envelope , instabilities set in and the dense shell fragments .
the circumstellar bubble may collide and merge with the main sequence interstellar bubble or evaporate in the hot shocked wr wind in the interior of the main sequence bubble .
an observed wr ring nebula could be in any of these aforementioned stages .
it is conceivable that interstellar bubbles have large dynamic ages and normal ( interstellar ) abundances , while the circumstellar bubbles have small dynamic ages and anomalous abundances . as
a circumstellar bubble merges with the surrounding interstellar bubble , the observed abundances will asymptotically approach the normal ( interstellar ) abundances . the most telltale physical properties that may distinguish these stages are thus nebular dynamics and abundances .
however , it must be born in mind that large variations exist in the history of stellar mass loss and the distribution of ambient interstellar medium ; therefore , even detailed information on dynamics and abundances may not lead to a unique interpretation .
we are extending the study of wr ring nebulae to the lmc sample and have selected the seven most well - defined wr rings in the lmc from the survey by dopita et al .
these seven ring nebulae , listed in table 1 , not only have distinct ring or arc morphology but also have sizes in the range of 5 to 40 pc , so that these ring nebulae may be the true counterparts of the archetypical wr rings , such as ngc2359 , ngc6888 , and s308 in the galaxy ( johnson & hogg 1965 ) .
we will first discuss the nature of these seven lmc wr ring nebulae and compare them to the galactic wr ring nebulae .
we will further compare the wr ring nebulae to ring nebulae around lbvs and blue supergiants ( bsgs ) .
as these different spectral types represent different evolutionary stages of massive stars , the comparison of physical properties among their ring nebulae may help us understand the evolutionary aspects of these stars .
the size , expansion velocity , dynamic timescale " , and n / o abundance ratios of these nebulae are tabulated in table 3 .
the dynamic timescale " is defined as the expansion velocity divided by the radius .
this dynamic timescale scales with the dynamic age of the nebula , but is not equal to the dynamic age which depends on whether the expansion has been accelerated or decelerated since the initial formation . the kinematic structure of each of the seven selected lmc wr ring nebulae ( chu 1983 ; this paper ) suggests interactions between the fast stellar wind and the ambient medium . in some nebulae , especially those around br 40a and br 48 , the kinematic structure is so irregular that it can not be approximated as an expanding shell .
these irregular motions are clearly caused by the large density variations in the ambient interstellar medium ; consequently , the kinematic structure of these nebulae can not be used to derive unambiguous information on the stellar mass loss history . for the rest of this discussion
we will concentrate on only the other five nebulae that have well - behaved expansion properties . for comparison ,
we have selected six galactic wr ring nebulae , listed in table 3 , based on their well - observed dynamics and abundances .
relative to a galactic interstellar n / o ratio of @[email protected] ( shaver et al . 1983 ) ,
s308 , rcw58 , m1 - 67 , and ngc6888 are obviously enriched and must contain stellar ejecta . ngc2359 and ngc3199 , on the other hand , show little abundance anomaly , indicating that these nebulae are dominated by interstellar material .
these two nebulae also have the most irregular expansion patterns and the smallest expansion velocities .
these dynamic properties and n / o abundance ratios suggest that the circumstellar bubbles of ngc2359 and ngc3199 have merged with the fossil main sequence interstellar bubbles . using the detailed fragmentation morphology , glm96 and gml96
conclude that ngc6888 s wr star has evolved through a rsg phase , and rcw58 s wr star has evolved through a lbv phase .
the lmc wr ring nebulae do not have as many observations of abundances available as the galactic nebulae .
only two lmc nebulae , around br 2 and br 13 , have been observed ( garnett & chu 1994 ) .
compared to the lmc interstellar value of n / o @xmath80.04 ( garnett 1998 ) , the br 13 nebula has clearly anomalous abundances while the br 2 nebula is marginally anomalous .
the br 13 nebula must be a circumstellar bubble ; its regular expansion pattern and small dynamic timescale both support this explanation .
the spectral type of br 13 , wn8 , is the same as that of wr40 , the central star of rcw58 ; the n / o ratio of the br 13 nebula is similar to that of rcw58 .
if the progenitor of wr40 was a lbv , it is then possible that the progenitor of br 13 was also a lbv .
a high - resolution image of the br 13 nebula would be useful in determining whether the fragmentation of the nebula is consistent with those expected for a lbv progenitor ( gml96 ) .
the n / o ratio of the br 2 nebula indicates that it might be a circumstellar bubble merging with a main sequence interstellar bubble .
this explanation is supported by the large dynamic timescale and small expansion velocity of the br 2 nebula .
the other three lmc wr ring nebulae , around br 10 , br 52 , and br 100 , do not have abundance observations .
their large sizes and surface brightness variations indicate that they must be interacting with the ambient interstellar medium
. however , these three nebulae have relatively large expansion velocities and quite regular expansion pattern , especially the br 10 nebula .
if the lmc wr nebulae behave similarly to the galactic wr nebulae , we may expect that these three nebulae are not yet dominated by interstellar material .
the spectral types of these three central stars are wc5 and wn3 - 4 .
no abundances have been derived for circumstellar nebulae of wc5 or wn3 - 4 stars in either the galaxy or the magellanic clouds .
new abundance observations of these three wr nebulae would be most interesting , as they could place constraints on the evolution of progenitors for these spectral types .
we have included in table 3 a number of galactic and lmc lbv nebulae whose nebular dynamics have been well observed .
the galactic lbv nebulae are all smaller and younger than the galactic wr ring nebulae . among the galactic lbv nebulae ,
the smallest one has the smallest dynamic timescale and the lowest expansion velocity .
if these four galactic lbvs evolve similarly , their nebulae must have gone through a rapid acceleration during the lbv phase .
the lmc lbv nebulae appear to be generally larger than the galactic lbv nebulae .
this could be completely caused by a combination of small - number statistics and difficulty in resolving and detecting small lbv nebulae in the lmc where 1 pc subtends only 4@xmath16 .
it is noteworthy , however , that the lmc lbv nebulae have smaller expansion velocities than the galactic counterparts .
this relation , if it holds for a larger number of lmc lbv nebulae , may indicate a difference in the mass loss properties between the lmc lbvs and the galactic lbvs .
it is interesting to compare the br 13 nebula to the lmc lbv nebulae .
the dynamic timescale of the br 13 nebula is comparable to those of the lmc lbv nebulae of r127 and s119 , although the br 13 nebula is much larger and expands much faster .
this comparison suggests that the progenitor of br 13 could not have been a lbv similar to r127 or s119 . finally , we examine two ring nebulae around lmc blue supergiants ( bsgs ) .
the ring nebula around the o9f star sk@xmath2069279 is recently discovered by weis et al .
( 1997b ) .
its high [ nii]@xmath16583/h@xmath0 ratio , @xmath80.7 , is consistent with that expected in a n - enriched ejecta nebula .
its expansion velocity , only 14 km s@xmath3 , is lower than those of all wr and lbv nebulae listed in table 3 .
its size is larger than that of every known lbv nebula , but comparable to those of small wr ring nebulae . the exact abundances of sk@xmath2069279 s nebula are unknown , hence it is uncertain whether the nebula has swept up a significant amount of interstellar material and its expansion has subsequently been slowed down .
future abundance observations are needed to determine the evolutionary status of sk@xmath2069279 with respect to rsg , lbv , and wr phases .
the ring nebula around the b3i star sk@xmath2069202 , better known as the progenitor of sn 1987a , is small compared to ring nebulae around other massive stars .
its expansion velocity is the smallest . clearly , the size and dynamics of sk@xmath2069202 s ring nebula suggest that sk@xmath2069202 could not have gone through a wr phase . this is consistent with the relatively low mass ( 20 m@xmath21 ) inferred for the supernova sn 1987a s progenitor .
the echelle observations of br 2 reveal a high - velocity feature to the northwest of br 2 .
the echellogram of the slit position centered on br 2 along the position angle 120@xmath12 shows high - velocity material projected from the vicinity of br 2 to almost 90@xmath16 northwest of br 2 . the slit position centered at 10@xmath16n , 35@xmath16w of br 2 along the n - s direction shows high - velocity material over 110@xmath16 , or 28 pc , along the slit ( figure 2 ) .
the velocity structure and size are very similar to those of known snrs in the magellanic clouds , particularly the snr n19 ( [email protected] ) in the small magellanic cloud ( chu & kennicutt 1988 ) .
the radio continuum emission of this region is brighter than those of hii regions of comparable or even higher h@xmath0 surface brightnesses ( haynes et al .
1991 ) , indicating a nonthermal radio emission .
thus , this high - velocity feature most likely originates from a snr at an age of a few @xmath22 yr .
br 2 s ring nebula is unusual in two respects .
first , heii emission is detected in the ring nebula .
second , br 2 s ring overlaps the projected position of a snr candidate .
however , the heii emission can not be caused by a dynamic interaction between the ring nebula and the snr , as the heii@xmath16560 line does not show violent velocities and the h@xmath0 line of the ring nebula does not show continuous high - velocity wings .
the snr is probably just projected by chance to the vicinity of br 2 and its ring nebula .
we have selected the seven most well - defined wr ring nebulae in the lmc to study their physical nature and evolutionary stages .
new images and echelle observations have been obtained for five of these nebulae ; previous observations ( chu 1983 ) are available for the two remaining nebulae .
only five of these nebulae ( br 2 , br 10 , br 13 , br 52 , and br 100 ) have well - behaved expansion pattern to warrant further discussion .
of these five nebulae , br 2 s and br 13 s ring nebulae have abundance information available ( garnett & chu 1994 ) .
based on nebular dynamics and abundances , we suggest that the br 13 nebula is a circumstellar bubble , and that the br 2 nebula may represent a circumstellar bubble merging with a fossil main - sequence interstellar bubble .
the nebulae around br 10 , br 52 , and br 100 all show influence of the ambient interstellar medium .
their regular expansion patterns suggest that they still contain significant amounts of circumstellar material .
the abundances of these nebulae would be extremely interesting , as their central stars are wc5 and wn3 - 4 stars whose nebular abundances have never been derived before .
the lmc wr ring nebulae do not differ significantly from their galactic counterparts .
comparisons between wr ring nebulae and ring nebulae around other massive stars , such as lbvs and bsgs , yield intriguing and tantalizing implications on stellar evolution and mass loss history . however , the credibility of these implications is limited by a small number statistics .
future observations of a larger number of nebulae are needed to confirm these results
. references .
( 1 ) this paper ; ( 2 ) garnett & chu 1994 ; ( 3 ) chu 1983 ; ( 4 ) chu et al .
1982 ; ( 5 ) esteban et al .
1992 ; ( 6 ) chu 1988 ; ( 7 ) chu 1982 ; ( 8) solf & carsenty 1982 ; ( 9 ) smith 1991 ; ( 10 ) smith et al .
1997 ; ( 11 ) smith et al . 1994 ; ( 12 ) hutsemekers et al .
1994 ; ( 13 ) weis et al .
1997a ; ( 14 ) nota et al .
1997 ; ( 15 ) appenzeller et al . 1987 ; ( 16 ) smith et al . 1998 ; ( 17 ) nota et al . 1994 ; ( 18 ) weis et al .
1997b ; ( 19 ) crotts & heathcote 1991 ; ( 20 ) panagia et al .
appenzeller , i. , wolf , b. , & stahl , o. 1987 , in instabilities in luminous early - type stars , ed .
h. j. g. l. m. lamers & c. w. h. de loore ( dordrecht : reidel ) , 241 breysacher , j. 1981 , a&as , 43 , 203 castor , j. , mccray , r. , & weaver , r. 1975 , apj , 200 , l107 chu , y .- h .
1982 , apj , 254 , 578 chu , y .- h .
1983 , apj , 269 , 202 chu , y .- h .
1988 , pasp , 100 , 986 chu , y .- h . , gull , t. r. , treffers , r. r. , kwitter , k. b. , & troland , t. h. 1982 , apj , 254 , 562 chu , y .- h .
, & kennicutt , r. c. jr . 1988 , aj , 95 , 1111 chu , y .- h . , & lasker , b. m. 1980 , pasp , 92 , 730 conti , p. s. , & garmany , c. d. 1983 , pasp , 95 , 411 crotts , a. p. , & heathcote , s. r. 1991 , nature , 350 , 683 davies , r. d. , elliott , k. h. , & meaburn , j. 1976 mmras , 81 , 89 dopita , m. a. , bell , j. f. , chu , y .- h . , & lozinskaya , t. a. 1994 , apjs , 93 , 455 esteban , c. , vlchez , j. m. , smith , l. j. , & clegg , r. e. s. 1992 , a&a , 259 , 629 garca - segura , g. , & mac low , m .-
1995a , apj , 455 , 145 garca - segura , g. , & mac low , m .-
1995b , apj , 455 , 160 garca - segura , g. , langer , n. , & mac low , m .-
1996 , a&a , 316 , 133 ( glm96 ) garca - segura , g. , mac low , m .-
, & langer , n. 1996 , a&a , 305 , 229 ( gml96 ) garnett , d. r. 1998 , in iau symp .
190 , new views of the magellanic cloud , ed . y .-
chu et al . , in press garnett , d. r. , & chu , y .- h . 1994 , pasp , 106 , 626 haynes , r. f. , et al .
1991 , a & , 252 , 475 hutsemekers , d. , van drom , e. , gosset , e. , & melnick , j. 1994 , a&a , 290 , 906 johnson , h. m. , & hogg , d. e. 1965 , apj , 142 , 1033 langer , n. , hamann , w .- r . , lennon , m. , najarro , f. , pauldrach , a. w. a. , & puls , j. 1994 , a&a , 290 , 819 marston , a. p. , chu , y .- h .
, & garca - segura , g. 1994a , apjs , 93 , 229 marston , a. p. , yocum , d. r. , garca - segura , g. , & chu , y .- h .
1994b , apjs , 95 , 151 miller , g. , & chu , y .- h . 1993 , apjs , 85 , 137 nota , a. , drissen , l. , clampin , m. , leitherer , c. , pasquali , a. , robert , c. , paresce , f. , & robberto , m. 1994 , in circumstellar media in the late stages of stellar evolution , ed .
r. clegg et al .
( cambridge : cambridge univ . press ) , 89 nota , a. , smith .
, l. j. , pasquali , a. , clampin , m. , & stroud , m. 1997 , apj , 486 , 338 osterbrock , d. e. , fulbright , j. p. , martel , a. r. , keane , m. j. , trager , s. c. , & basri , g. 1996 , pasp , 108 , 277 pakull , m. w. 1991 , in iau symp .
143 , wolf - rayet stars and interrelations with other massive stars in galaxies , ed .
k. a. van der hucht & b. hidayat ( dordrecht : kluwer ) , 391 panagia , n. , scuderi , s. , gilmozzi , r. , challis , p. m. , garnavich , p. m. , & kirshner , r. p. 1996 , apj , 459 , l17 rosado , m. 1986 , a&a , 160 , 211 shaver , p. a. , mcgee , r. x. , newton , l. m. , danks , a. c. , pottasch , s. r. 1983 , mnras , 204 , 53 smith , l. j. 1991 , in iau symp .
143 , wolf - rayet stars and interrelations with other massive stars in galaxies , ed
. k. a. van der hucht & b. hidayat ( dordrecht : kluwer ) , 385 smith , l. j. , crowther , p. a. , & prinja , r. k. 1994 , a&a , 281 , 833 smith , l. j. , nota , a. , pasquali , a. , leitherer , c. , clampin , m. , & crowther , p. a. 1998 , apj , 503 , 278 smith , l. j. , stroud , m. p. , esteban , c. , vlchez , j. m. 1997 , mnras , 290 , 265 smith , l. j. , & willis , a. j. 1983 , a&as , 54 , 229 solf , j. , & carsenty , u. 1982 , a&a , 116 , 54 steigman , g. , strittmatter , p. a. , & williams , r. e. 1975 , apj , 198 , 575 van der hucht , k. a. , conti , p. s. , lundstrom , i. , & stenholm , b. 1981 , space sci .
, 28 , 227 weaver , r. , mccray , r. , castor , j. , shapiro , p. , & moore , r. 1977 , apj , 218 , 377 weis , k. , duschl , w. j. , bomans , d. j. , chu , y .- h . , & joner , m. d. 1997a , a&a , 320 , 568 weis , k. , chu , y .- h . , duschl , w. j. , bomans , d. j. 1997b , a&a , 325 , 1157
|
we have selected the seven most well - defined wr ring nebulae in the lmc ( br 2 , br 10 , br 13 , br 40a , br 48 , br 52 , and br 100 ) to study their physical nature and evolutionary stages .
new ccd imaging and echelle observations have been obtained for five of these nebulae ; previous photographic imaging and echelle observations are available for the remaining two nebulae . using the nebular dynamics and abundances
, we find that the br 13 nebula is a circumstellar bubble , and that the br 2 nebula may represent a circumstellar bubble merging with a fossil main - sequence interstellar bubble .
the nebulae around br 10 , br 52 , and br 100 all show influence of the ambient interstellar medium .
their regular expansion patterns suggest that they still contain significant amounts of circumstellar material .
their nebular abundances would be extremely interesting , as their central stars are wc5 and wn3 - 4 stars whose nebular abundances have not been derived previously .
intriguing and tantalizing implications are obtained from comparisons of the lmc wr ring nebulae with ring nebulae around galactic wr stars , galactic lbvs , lmc lbvs , and lmc bsgs ; however , these implications may be limited by small - number statistics . a snr candidate close to br 2
is diagnosed by its large expansion velocity and nonthermal radio emission .
there is no indication that br 2 s ring nebula interacts dynamically with this snr candidate .
-0.5 in 6.5 in
|
for a finite set @xmath3 let @xmath4 be the set of probability distributions on @xmath3 .
a _ specification g _
( also commonly known as a @xmath1-function ) is a measurable function from @xmath5 to @xmath4 .
a specification @xmath1 is _ regular _ if there exists @xmath6 such that for every sequence @xmath7 and every @xmath8 we have that @xmath9 .
we focus on specifications that are regular and continuous with respect to the product topology .
a _ gibbs measure _ for a specification @xmath1 is a shift invariant probability measure @xmath10 on @xmath11 such that for every @xmath12 , @xmath13 it is easy to show that every continuous specification has a gibbs measure .
given a past , @xmath14 , the specification @xmath1 tells us the probability distribution for @xmath15 , the next state of the process . thus the specification and the past
determine the stochastic evolution of the process .
one common example of a specification is a finite state space markov chain .
the specification for a @xmath16-step markov chain is determined by @xmath17 .
for this reason dblin and fortet referred to specifications as `` chains with infinite connections '' @xcite . the question of whether a specification @xmath1 uniquely determines the stationary process under a natural `` mixing '' assumption has been prominent since the pioneering work of dblin and fortet @xcite .
however , in the last three decades reasonable progress was achieved only in providing sufficient conditions for the uniqueness of gibbs measure ( see @xcite for detailed discussions and references ) .
harris @xcite studied the behavior of lumped markov chains and introduced important coupling ideas that were used by several later authors .
m. keane @xcite introduced the notion of a continuous @xmath1-function and gave conditions under which a @xmath1-function has a unique measure .
one natural way to express uniqueness conditions is in terms of the modulus of continuity of @xmath1 . to quantify the modulus of continuity of a specification @xmath1
, we define the _ variation of @xmath1 at distance @xmath16 _ to be @xmath18 the continuity of @xmath1 is equivalent to the sequence @xmath19 . in @xcite , walters showed that if @xmath20 , then the gibbs measure is unique .
walters and ledrappier @xcite established a strong connection between specifications and the thermodynamic formalism of statistical mechanics .
in particular they relate specifications with symbolic dynamics as developed in works of sinai @xcite , ruelle @xcite , bowen @xcite , and others .
walters s work was sharpened by lalley @xcite and berbee @xcite , who showed that the gibbs measure is unique if @xmath21 on the other hand , by the early 1980 s it was known that the equilibrium measures of appropriately chosen one dimensional long - range ising models are not unique ( see @xcite for the hierarchical type models and @xcite for the @xmath22 decaying cases ) .
existence of phase transition was later established for one - dimensional long - range percolation model ( see @xcite and @xcite ) and finally for one - dimensional fk - random cluster model @xcite .
nevertheless a little more than a decade ago it was widely believed that continuous and regular specifications admit a unique gibbs measure .
however , in 1993 , bramson and kalikow @xcite provided a remarkable ( and until now unique ) example of a continuous and regular specification that admits multiple gibbs measures .
the variation of the function @xmath1 in bramson and kalikow s construction is not in @xmath2 for any @xmath23 .
in fact , in their example @xmath24 for some constant @xmath25 .
this gave rise to the following question : for which values of @xmath26 does @xmath27 imply uniqueness .
a few years ago stenflo further sharpened berbee s work @xcite .
however , the results of berbee and stenflo , while improving over walters result , are still in the realm of @xmath28 . recently
johansson and berg @xcite showed that if @xmath1 is regular and @xmath29 is in @xmath30 then @xmath1 admits a unique gibbs measure .
our main result is the following : [ thm : main ] for every @xmath0 , there exists a regular specification @xmath1 such that @xmath31 and @xmath1 admits multiple gibbs measures .
this shows that the result of berg and johansson is tight .
* remark . * in @xcite
@xmath32and maillard proved a dobrushin type uniqueness condition .
this condition is not comparable with the variation conditions .
we will use the alphabet of size four @xmath33 .
we fix a parameter @xmath34 . given this we pick a positive integer @xmath35 such that the inequalities in lines ( [ star1 ] ) through ( [ star3 ] ) below are true for all @xmath36 .
we now begin to define a regular continuous specification @xmath37 the choice of @xmath38 given @xmath39 consists of four steps : 1 .
[ item : gety ] choose @xmath40 independently of @xmath39 , so that @xmath41 with probability @xmath42 and @xmath43 with probability @xmath42 .
[ item : gets ] using the values of @xmath44 choose a ( deterministic ) set of odd size @xmath45 .
[ item : getupsilon ] using the values of @xmath44 choose a ( deterministic ) value @xmath46 .
[ item : getx ] let @xmath47 be the majority value of @xmath48 . choose @xmath49 with probability @xmath50 and @xmath51 with probability @xmath52 . in order to complete the first step the second coordinate ( @xmath53 ) must be i.i.d . with distribution ( 1/2,1/2 ) .
we ensure this if for all @xmath54 and @xmath53 @xmath55 first we pick @xmath40 .
we write @xmath56 to represent @xmath53 and @xmath40 .
the most intricate part of the construction is the choice of the set @xmath57 . before we choose @xmath57 we need some notation . for every positive integer @xmath16 , let @xmath58 be the sequence of length @xmath59 such that the last element is @xmath60 and all other elements are @xmath61 .
we also define @xmath62 and @xmath63 now we define a block structure which will allow us to choose the set @xmath64 and a parameter @xmath65 .
[ compblock ] a * complete @xmath16 block in @xmath56 * is a subsequence @xmath66 such that 1 .
@xmath67 2 . for no @xmath68
does @xmath69 we also define a partial @xmath16 block as follows .
a * partial @xmath16 block in @xmath56 * is a subsequence @xmath70 such that 1 .
@xmath71 2 . for no @xmath72
does @xmath69 a * @xmath16 block in @xmath56 * is an interval that is either a complete @xmath16 block or a partial @xmath16 block .
if @xmath56 is well understood then we write @xmath73 to denote a @xmath16 block @xmath66 .
note that this definition is invariant under a shift of @xmath56 in the following sense : given @xmath53 and any integer @xmath74 define @xmath75 by @xmath76 for all @xmath77 .
then for @xmath78 , if @xmath79 is a @xmath16 block for @xmath80 then @xmath81 is a @xmath16 block for @xmath56 .
also note that the length of a @xmath16 block is likely to be close to @xmath82 , and that the number of @xmath83 blocks inside a @xmath16 block is likely to be close to @xmath84 .
a precise statement of these claims will be used extensively in the next section .
next we label all of the @xmath16 blocks . given a sequence @xmath56 and a positive integer @xmath16 we will define @xmath85 to be the @xmath86-th @xmath16 block in @xmath56 . more precisely we define @xmath87 and @xmath88 such that 1 .
@xmath89 is a @xmath16 block in @xmath56 .
@xmath90 for all @xmath86 3 .
. these sequences can be either finite or infinite .
then we define @xmath92 given @xmath53 and @xmath93 define @xmath94 to be the number of @xmath83 blocks in the @xmath16 block containing @xmath95 .
more precisely , we take @xmath94 to be so that @xmath96 or @xmath97 if ( [ defnk ] ) has no solution
. we also set @xmath98 [ def : begin ] we say that the * beginning * of a @xmath99 block @xmath73 is the interval @xmath100 . for @xmath93
we say that the * beginning * of a @xmath16 block @xmath73 is the @xmath101 first @xmath83 blocks in @xmath102 , i.e. @xmath103 the * opening * @xmath104 of a @xmath16 block @xmath102 is the set of points @xmath105 such that 1 .
@xmath106 is in the beginning of its @xmath107-block for every @xmath108 .
2 . if @xmath109 is the smallest element in the @xmath107-block containing @xmath106 , then @xmath110 for all @xmath111 .
we also define @xmath112 to be the union of @xmath104 over all @xmath16 blocks @xmath102 .
note that if @xmath113 then @xmath114 for all @xmath115 .
the event that @xmath116 is determined by @xmath117 , @xmath118 $ ] .
this fact will be used to show that our specification is continuous and to show how quickly @xmath119 approaches 0 . define @xmath120 to be the highest value such that @xmath121 .
if there is no such value then we take @xmath122 .
if @xmath123 for every @xmath16 then we take @xmath124 .
we now define @xmath57 to be the following set : if @xmath125 then @xmath57 is the empty set .
if @xmath126 is odd then we take @xmath127 . if @xmath126 is even , @xmath128 .
next we choose the value of @xmath65 .
if @xmath129 then we take @xmath130 . otherwise , we take @xmath131 finally we set @xmath132 where sign(@xmath54 ) is the function that takes on values + 1 or -1 depending on whether @xmath54 is positive or negative .
thus we have defined our specification @xmath1 by the coordinate functions @xmath133 we leave it to the reader to check the following lemma which says that @xmath1 is symmetric in the @xmath54 coordinate .
[ lem : symmetry ] for all @xmath134 and @xmath135 @xmath136 we now show that @xmath1 is continuous .
the two most important elements of the construction are that for all @xmath53 1 .
there exists @xmath137 such that the set @xmath57 is determined by @xmath138 .
the larger @xmath137 is the closer @xmath65 is to zero . in the next lemma
we quantify these two statements and show that they imply that @xmath1 is continuous .
[ lem : cont ] the specification @xmath1 is regular and continuous .
moreover the sequence @xmath139 for all @xmath140 note that as @xmath141 approaches @xmath95 , the bound in ( [ eq : noeps ] ) goes to @xmath142 .
let @xmath93 .
we want to estimate @xmath143 for @xmath144 .
let @xmath145 and @xmath146 be such that @xmath147 and @xmath148 for every @xmath149 .
it is enough to estimate @xmath150 for @xmath151 .
fix @xmath152 and for @xmath153 let @xmath154 if @xmath155 then @xmath156 and @xmath157 depends only on @xmath158 .
therefore , on this case @xmath159 .
if , on the other hand , @xmath160 , then @xmath161 as well and therefore @xmath162 .
the goal of this section is to prove the following lemma : [ prop : main ] for every @xmath34 the specification @xmath164 admits multiple measures .
theorem [ thm : main ] is a direct consequence of lemmas [ lem : cont ] and [ prop : main ] . to see that the function @xmath1 admits multiple gibbs measures , we introduce the following notation .
choose an arbitrary @xmath106 .
we let @xmath165 denote the @xmath16 block containing @xmath106 , and we let @xmath166 denote the opening of @xmath165
. we will prove that @xmath167 with extremely high probability .
next we note that for any @xmath106 and @xmath168 that @xmath169 ( and thus @xmath170 ) for all @xmath16 large enough .
( see figure [ fig : arrows ] below . )
thus by the symmetry of lemma [ lem : symmetry ] there exist at least two invariant measures : one where @xmath171 a.s .
and one where @xmath172 a.s .
assume that @xmath10 is a stationary and ergodic measure on @xmath173 which is a gibbs measure for @xmath1 .
note that in the definitions of a complete @xmath16 block ( definition [ compblock ] of page ) there was no requirement that @xmath174 and @xmath175 be negative .
thus we can use the same definition of a complete block for a sequence in @xmath176 * remark : * from now on we will abbreviate the term * complete @xmath16 block * to * @xmath16 block*. this is a natural abuse of notation that results from the fact that we now speak about objects taking places in @xmath173 instead of @xmath177 . in the same spirit , @xmath54 and @xmath53 will now denote two - sided sequences . in subsections [ subsec : gb ] and [
subseq : beautipoints ] we prove results about the block structure .
the block structure depends only on the @xmath53 sequence , which is an i.i.d .
@xmath178 sequence . in subsection
[ subsec : last ] we use the results obtained in subsections [ subsec : gb ] and [ subseq : beautipoints ] to understand the ( more complicated ) structure of the @xmath54 sequence and show the existence of two different gibbs measures . in this subsection
we define the notion of a good block .
a good block is a block whose length is close to its expected length and the size of its opening is also close to its expected value .
our aim in this subsection is to show that a @xmath16 block is good with high probability . in the next subsections
we show that good blocks exhibit behavior that yields multiple gibbs measures .
[ defn : goodblock ] a @xmath16 block @xmath102 is * good * if 1 .
[ item : size ] @xmath179 and 2 .
[ item : coresize ] @xmath180 in order to show that most blocks are good we put a measure on @xmath16 blocks by conditioning on @xmath53 such that @xmath181 . for a @xmath16 block @xmath182 such that 1 .
@xmath183 , 2 .
@xmath184 and 3 .
there is no occurrence of @xmath58 in @xmath102 we define the measure @xmath185 for @xmath186 , we let @xmath165 denote the @xmath16 block containing @xmath106 . with these definitions
we can check that for any @xmath10 which is a gibbs measure for @xmath1 and any @xmath16 and @xmath102 @xmath187 [ lem : pvg ] for all @xmath188 @xmath189 we prove the lemma by induction on @xmath16
. for @xmath190 we need only to check the first condition .
the probability that a @xmath16 block is longer than @xmath191 is less than @xmath192 .
thus the probability that the first condition in the definition is not satisfied is at most @xmath193 this is true by our choice of @xmath194 . if @xmath195 and 1 .
@xmath179 2 .
there are at least @xmath101 k-1 blocks in @xmath102 and [ ii ] 3 .
at least half of the @xmath101 k-1 blocks in the beginning of @xmath102 are good [ iii ] then @xmath196 and @xmath102 is good . as the bound in line [ star1 ] holds for any @xmath195 we have that the probability that the first condition is not satisfied is at most @xmath197 .
the probability that conditions [ ii ] and [ iii ] are satisfied is the probability that if we select @xmath101 @xmath83 blocks independently according to @xmath198 that none of them are a @xmath16 block and that at least half of them are good .
the probability that a @xmath83 block is also a @xmath16 block is @xmath199 .
thus the probability of having a @xmath16 block among @xmath101 k-1 blocks is less than @xmath200 by the induction hypothesis a @xmath83 block is not good with probability less than @xmath201 .
thus the probability that half of a sequence of @xmath101 @xmath83 blocks chosen independently are not good is at most @xmath202 [ lem : goodhighp ] for every @xmath36 @xmath203 first , we want to estimate @xmath204 @xmath205 is the probability that @xmath58 appears somewhere in @xmath206 . for @xmath207 ,
let @xmath208 be the event that there exists @xmath209 with @xmath210 such that @xmath211 .
@xmath212 are negatively associated , and for every @xmath86 , @xmath213 therefore @xmath214 on the other hand , for @xmath215 let @xmath216 to be the event that @xmath58 appears in @xmath217 then the @xmath216-s are independent and using a first - moment argument we see that for every @xmath86 , @xmath218 . the event measured by @xmath205 is the union of the @xmath216-s plus the event that @xmath58 appears on the seams between the blocks .
therefore , @xmath219 for any @xmath107 , let @xmath220 be the probability that the first appearance of @xmath58 is in the interval @xmath221
. then @xmath222 where the last term comes from the event that @xmath58 appears on the seam between two consecutive intervals of length @xmath82 .
from ( [ eq : geqyt ] ) , ( [ eq : leqyt ] ) and ( [ eq : conyz ] ) , we get that for any @xmath107 we have that @xmath223 from this we get @xmath224 and @xmath225 thus for any set @xmath57 of @xmath16 blocks such that @xmath226 we have that @xmath227 combining this last statement with lemma [ lem : pvg ] @xmath228 in this subsection we define the notion of a beautiful point
. our goal in this subsection will be to show that most points are beautiful .
[ def : beau ] we say that @xmath186 is * @xmath16 beautiful * if for every @xmath229 , 1 .
[ item : ingood ] @xmath230 is good and 2 .
[ item : inend ] @xmath106 is not in the beginning of @xmath230 .
( the _ beginning _ of a block was defined in definition [ def : begin ] of page ) in this subsection we state two easy lemmas : [ lem : triv ] if @xmath106 and @xmath231 belong to the same @xmath16 block , and @xmath106 is @xmath232 beautiful , then @xmath231 is also @xmath232 beautiful .
lemma [ lem : triv ] follows immediately from definition [ def : beau ] .
[ lem : singlepoint ] almost surely , for every @xmath106 there exists @xmath233 such that @xmath106 is @xmath233 beautiful .
first we show that @xmath234 lemma [ lem : goodhighp ] tells us that the probability that the @xmath107 block containing @xmath95 is good is greater than @xmath235 .
as there are at most @xmath236 @xmath83 blocks in the beginning of @xmath237 and the expected number of @xmath83 blocks in @xmath237 is @xmath238 , the probability that @xmath95 is in the beginning of @xmath239 is less than @xmath240 therefore @xmath95 is @xmath16 beautiful with probability at least @xmath241 thus by borel - cantelli we get that @xmath242 exists a.s .
the lemma is true because @xmath10 is shift invariant . in the previous subsections we only discussed the structure induced by the @xmath53 values . in this subsection
we will shed some light on the structure of the @xmath54 values . for a @xmath16 block @xmath102
, we define the * signature * of @xmath102 to be @xmath243 where @xmath244 note that since @xmath245 is always odd , @xmath246 can only be @xmath61 or @xmath60 .
we assume that @xmath10 is an ergodic gibbs measure for @xmath1 .
[ lem : majinblock ] for all @xmath106 @xmath247 exists a.s . and is equal to 1 or @xmath60 . by lemma [ lem : singlepoint ]
we get @xmath248 such that @xmath106 is @xmath248 beautiful . by lemma [ lem : symmetry ]
it causes no loss of generality to assume that @xmath249 . using that @xmath250 is good for every @xmath251 , that @xmath10 is a gibbs measure with respect to @xmath1 , and line ( [ eq : defups ] ) in the definition of @xmath1 , we get that given @xmath53 and @xmath252 the values of @xmath253 are independent and identically distributed . for every @xmath231 in @xmath254
, @xmath255 also , the fact that @xmath165 is good implies @xmath256 thus we get @xmath257 we have that @xmath258 the standard deviation of the sum is less than @xmath259 .
thus by markov s inequality we have that @xmath260 by borel - cantelli there are only finitely many values of @xmath16 such that @xmath261 and @xmath262 exists . as @xmath263 is odd for all @xmath16 and @xmath106 the limit must be either 1 or @xmath60 .
we are now ready to prove lemma [ prop : main ] . for every @xmath106 and @xmath231 @xmath264 for all @xmath16 sufficiently large .
therefore @xmath265 for all @xmath231 and @xmath106 .
( see figure [ fig : arrows ] at page . ) since @xmath10 is ergodic and @xmath266 is shift invariant , @xmath266 is a @xmath10 almost sure constant , which we denote by @xmath267 .
let @xmath268 then , by lemma [ lem : symmetry ] , @xmath269 is also a gibbs measure for @xmath1 . on the other hand , @xmath270 , and therefore @xmath271 .
our construction is not monotone - by changing a @xmath53 value from a @xmath60 to a @xmath61 we may change the set @xmath57 at which we look , and then reduce the probability that the function outputs @xmath61 at the @xmath54 coordinate .
the construction of bramson and kalikow , on the other hand , is monotone .
therefore we ask the following question : * acknowledgment : * our research on this problem started while n.b . and c.h .
visited v.s . in impa , rio de janeiro .
we thank yuval peres and maria eulalia vares for useful discussions . n.b . and
are grateful for the hospitality and financial support during their stay at impa .
the work was partially supported by grants of faperj and cnpq .
. stenflo .
uniqueness of invariant measures for place - dependent random iterations of functions .
fractals in multimedia ( minneapolis , mn , 2001 ) , 1332 , i m a vol .
appl . , 132 , springer , new york , 2002 .
|
for every @xmath0 , we construct a regular and continuous specification ( @xmath1-function ) , which has a variation sequence that is in @xmath2 and which admits multiple gibbs measures . combined with a result of johansson and berg , @xcite
, this determines the optimal modulus of continuity for a specification which admits multiple gibbs measures .
|
let @xmath5 be a smooth bounded strictly convex region in the plane , and let @xmath6 be the eigenvalues , with multiplicity , of the dirichlet problem for the laplacian on @xmath5 : @xmath7 in the spirit of @xcite ) we will define the _ wave trace _ of the dirichlet laplacian @xmath0 to be the sum @xmath8 by a theorem of andersson and melrose @xcite , @xmath9 is a tempered distribution whose singular support in @xmath10 is the closure of the set @xmath11 where @xmath12 is the set of critical values of the length function given by @xmath13
@xmath14 are distinct boundary points .
each critical point of the length function defines a polygonal trajectory consisting of the line segments joining @xmath15 to @xmath16 and @xmath14 to @xmath17 .
these are , by definition , the _ closed geodesics _ on @xmath5 , and @xmath18 is the set of lengths of these geodesics , i.e. the _ length spectrum _ of @xmath5 .
each set @xmath12 is closed , but there are limiting geodesics as @xmath19 .
these limits are closed curves , called _ gliding rays _ , that lie entirely in @xmath20 .
near @xmath21 in @xmath18 , the nature of the singularity of @xmath22 is fairly well understood in generic cases ( see @xcite ) ; but , at least at present , this is not so for @xmath21 in the closure of the length spectrum .
for example , if @xmath21 is the perimeter of @xmath5 , then there are trajectories forming a convex polygon with @xmath23 sides whose perimeters tend to @xmath21 as @xmath19 . in general , the limiting lengths @xmath21 are integer multiples of the perimeter of @xmath24 , and the set @xmath25 is infinite for every @xmath26 .
the cumulative effect of infinitely many singularities could be quite complicated and has not been analyzed completely even in simple model examples .
however to the right of @xmath21 one does nt encounter this problem .
@xmath27 for sufficiently small @xmath26 , so that @xmath9 is @xmath4 for @xmath28 .
therefore , several years ago the second author posed the question , what is the asymptotic behavior of @xmath22 as @xmath29 from the right ?
in one simple case , the case of the unit disk , we showed in @xcite that @xmath22 is @xmath4 to the right , i.e. , for some @xmath30 , is @xmath4 on @xmath31 .
we will publish details of that result elsewhere ; however in this article we will discuss another example for which we have been able to show that this assertion is true , and , in fact , for which the verification is simpler and more transparent than in the case of the disk .
this is the operator @xmath32 introduced by f. g. friedlander in @xcite , defined by @xmath33 on the manifold @xmath34 with dirichlet boundary conditions at @xmath35 .
it represents the simplest approximation to the laplace operator near the boundary of a convex domain .
the operator @xmath32 has continuous spectrum on the space of functions that depend on @xmath36 alone . on the orthogonal complement of this space , the spectrum is discrete , and the eigenfunctions of @xmath32 are described in terms of the airy function @xmath37 as follows ( see appendix a ) .
[ friedl spectrum ] the dirichlet eigenfunctions of @xmath32 on @xmath38 are the functions @xmath39 and their eigenvalues are @xmath40 in which @xmath41 are the roots of @xmath42 .
we will define the trace of the wave operator associated to @xmath32 by @xmath43 our main result is as follows .
[ friedl theorem ] let @xmath44 be the friedlander length spectrum , as computed in section [ sec : geodesics ] .
\a ) the limit points of @xmath44 are @xmath45 .
\b ) @xmath46 is @xmath4 in the complement of @xmath47
\c ) for each @xmath48 , there is @xmath49 such that @xmath46 extends to a @xmath4 function on @xmath50 . the main issue in the proof is to identify appropriate non - classical symbol properties for the function @xmath51 .
the friedlander model is significantly easier to treat than other integrable models such as the disk because the asymptotics of the spectral function @xmath51 follow easily from well known asymptotics of the airy function .
the paper is organized as follows . in section [ sec : geodesics ] , we find the length spectrum of @xmath32 . in section [ sec : smoothsymbols ]
, we show how the poisson summation formula is used to deduce smoothness of the trace when the spectral function is a classical symbol .
next , we show , in section [ sec : symbols ] , that the friedlander spectral function @xmath51 belongs to an appropriate non - classical symbol class of product type . in section [ sec :
proof ] , we prove the smoothness properties of the friedlander wave trace @xmath46 ( theorem [ friedl theorem ] ) by dividing the @xmath52 lattice into sectors and using poisson summation methods adapted in each sector to the symbol properties of @xmath53 .
we conclude , in section [ sec : actions ] , by computing the action coordinates of the completely integrable system corresponding to the operator @xmath32 and show that they capture the first order properties of eigenvalues and eigenfunctions of @xmath32 in sectors where @xmath51 behaves like a classical symbol .
this final computation reveals more details about which parts of the spectrum are linked to which aspects of the geodesic flow . in appendix a , we recall the properties of the airy function and prove proposition [ friedl spectrum ] characterizing the spectrum of @xmath32 . in appendix
b we review the semiclassical analysis of joint eigenfunctions of commuting operators in a general integrable system , showing how to find the parametrization of the eigenfunctions by a discrete lattice and how to carry out the computations of section [ sec : actions ] linking the classical and quantum systems in a more general context .
our first task is to calculate the length spectrum @xmath44 for the friedlander model .
define the hamiltonian function @xmath54 as the square root of the symbol of @xmath32 , @xmath55 the integral curves of the hamiltonian vector field of @xmath54 restricted to the hypersurface @xmath56 are defined as solutions to @xmath57 with the constraint @xmath58 we begin by analyzing a single trajectory between boundary points . [
prop : reflection ] consider an integral curve of the hamiltonian system - on @xmath59 , with @xmath60 if the initial velocity is @xmath61 , @xmath62 , then final velocity is given by @xmath63 and @xmath64 solving the equations , we get @xmath65 by translation invariance in @xmath66 , we may as well assume @xmath67
. then @xmath68 and @xmath69 substituting for @xmath21 we obtain @xmath70 and @xmath71 finally , @xmath72 and @xmath73 imply @xmath74 which concludes the proof .
we call a continuous curve @xmath75\to m$ ] a _ geodesic _ if there are finitely many points @xmath76 for which the curve meets the boundary ( i. e. @xmath77 ) and for @xmath78 , the curve is the projection onto @xmath79 of a solution to and , and , finally , at each @xmath80 the curve undergoes a reflection in the boundary : @xmath81 the hamiltonian equations are translation invariant in @xmath82 and @xmath66 . proposition [ prop : reflection ] shows that a geodesic with starting velocity @xmath83 has final velocity @xmath84 . by , the reflected velocity @xmath85 , the same as the starting velocity .
consequently , each arc of a geodesic between encounters with the boundary is a translate of the others .
this is analogous to equilateral polygonal trajectories in the disk .
the figure below depicts a trajectory reflected at the boundary @xmath35 .
note also that it is tangent to the caustic , @xmath86 , depicted by a dotted line .
[ prop : length spectrum ] a closed geodesic with @xmath87 reflections and winding number @xmath88 in the @xmath66 variable , has length @xmath89 determined by the equations @xmath90 proposition [ prop : reflection ] shows that the length of a closed geodesic with @xmath23 reflections is @xmath91 , with @xmath21 given as in the proposition .
hence , using @xmath92 , @xmath93 moreover , if the winding number is @xmath94 , then if @xmath95 is as in proposition [ prop : reflection ] with @xmath67 , @xmath96 substitituting @xmath97 into the last equation , it can be written @xmath98 the friedlander length spectrum is defined by @xmath99 where @xmath100 solve the equations of proposition [ prop : length spectrum ] for some @xmath101 , @xmath102 .
note that if @xmath103 is replaced by @xmath104 , then @xmath94 is replaced by @xmath105 and @xmath106 .
thus we may confine our attention to @xmath107 .
the implicit definition of @xmath100 can be restated as follows .
let @xmath108 be the homogeneous function of degree @xmath109 that is equal to @xmath109 on the curve @xmath110 then @xmath111 .
[ length structure ] the friedlander length spectrum @xmath44 can be written @xmath112 furthermore , \a ) @xmath113 .
\b ) for each @xmath114 , there is @xmath49 such that @xmath115 .
\c ) @xmath116 because the curve is the graph of a decreasing function , @xmath117 .
furthermore , @xmath118 as @xmath119
. therefore , every sequence of distinct values of @xmath100 tends to infinity unless @xmath94 remains bounded .
it follows from this and the fact that the curve is asymptotic to the horizontal line of height @xmath109 that all the limit points are given by @xmath120 the sequence @xmath100 is increasing in @xmath23 for fixed @xmath94 , so ( a ) and ( b ) are established .
finally , to confirm ( c ) , note that if @xmath121 for some @xmath122 , then the equations imply that @xmath123 is algebraic , a contradiction .
proposition [ length structure ] ( a ) and ( c ) imply theorem [ friedl theorem ] ( a ) .
the symbol class @xmath124 is the set of functions @xmath125 satisfying @xmath126 and some constants @xmath127 .
@xmath128 consists of the symbols in @xmath124 with an asymptotic expansion , that is , there are homogeneous functions @xmath129 of degree @xmath130 , smooth except at the origin , and a cut - off function @xmath131 such that @xmath132 on @xmath133 , and @xmath134 the function @xmath135 is the principal symbol of @xmath136 .
functions defined on a cone @xmath137 belong to the symbol class @xmath138 ( @xmath139 ) if they are restrictions of symbols in the class @xmath124 ( @xmath128 ) .
let @xmath140 be a real - valued symbol of first order , @xmath141 with positive principal symbol .
let @xmath142 be the 2-torus , @xmath143 , and let @xmath144 be the constant coefficient self - adjoint first order elliptic pseudodifferential operator given by @xmath145 the eigenvalues of @xmath146 are @xmath147 , @xmath148 , and hence its wave trace is @xmath149 since the summands are smooth functions of @xmath82 , the singularities of @xmath22 depend only on the asymptotic behavior of @xmath147 as @xmath150 .
for simplicity we will assume that @xmath151 on @xmath152 , where @xmath153 is homogeneous of degree @xmath109 and positive .
( with small modifications the proof we will give applies in general . ) to identify the length spectrum , consider the hamiltonian flow on the cotangent bundle of @xmath142 associated with the symbol , @xmath153 , defined by the ordinary differential equations @xmath154 its periodic trajectories are curves of the form , @xmath155 , @xmath59 , satisfying @xmath156 for some @xmath157 and some @xmath158 .
let @xmath159 be the set of @xmath21 for which has a solution for some @xmath160 .
( note that @xmath161 is homogeneous of degree @xmath162 , so that we may as well assume @xmath163 . )
the length spectrum is given by @xmath164 note that because @xmath161 is nonzero , bounded , and continuous on @xmath163 , the length spectrum is closed and bounded away from @xmath162 .
our goal is to prove that the function @xmath22 in is @xmath4 in @xmath10 on the complement of the length spectrum .
to do so we will apply the poisson summation formula to the function , @xmath165 .
the fourier transform of this function is @xmath166 so by the poisson summation formula , @xmath167 where @xmath168 we will first show that on the complement of the length spectrum these individual summands are @xmath4 .
if @xmath169 and @xmath170 then there exists an interval , @xmath171 , such that for all @xmath172 , either @xmath173 or @xmath174 is non - zero .
hence for @xmath82 on this interval @xmath175 can be written as a sum @xmath176 where @xmath177 and @xmath178 are smooth homogeneous functions of degree zero on @xmath172 and the first of the functions above is non - zero on @xmath179 and the second non - zero on @xmath180 .
finally @xmath181 is supported in a compact neighborhood of the origin .
the third integrand is compactly supported , and the corresponding integral converges along with all derivatives with respect to @xmath82 , regardless of the range of @xmath82 .
let @xmath182 then @xmath183 hence by integration by parts the first summand of ( [ smoothsplitsum ] ) can be written as @xmath184 or inductively as @xmath185 where @xmath186 since @xmath135 is a homogeneous function of degree zero in @xmath187 outside a small neighborhood of the origin , it follows by induction from ( [ andef ] ) that @xmath188 is a homogeneous function of degree @xmath189 , and hence @xmath190 uniformly on the interval @xmath191 .
differentiation @xmath23 times with respect to @xmath82 introduces the factor @xmath192 in the integrand , which is dominated by @xmath193 ; hence the integral converges . in all , the first of the summand in ( [ smoothsplitsum ] ) is a @xmath4 function of @xmath82 in the range specified .
the proof for the term @xmath178 is similar , replacing @xmath135 by @xmath194 , @xmath195 by @xmath196 and @xmath188 by @xmath197 one concludes by the same argument that the second of these summands is @xmath4 using integration by parts in @xmath198 . for @xmath199 large ,
@xmath200 is non - vanishing for all @xmath187 and all @xmath82 in @xmath201 , and integration by parts with respect to @xmath202 yields @xmath203 with @xmath204 and @xmath205 the estimate is replaced by @xmath206 similarly , for @xmath207 large , integration by parts with respect to @xmath198 yields an analogous formula with better decay as @xmath208 .
hence if we break the sum @xmath209 into subsums , @xmath210 , @xmath211 , and @xmath209 , @xmath212 , and make use of these estimates we conclude that not only are the individual summands of @xmath4 but that the sum itself is @xmath4 on the complement of the length spectrum @xmath18 .
let @xmath213 be a cone ( sector with vertex at the origin ) in @xmath214 .
a smooth function @xmath215 belongs to @xmath216 if there exists constants @xmath217 so that for all @xmath218 , the following upper bounds hold @xmath219 a typical example is @xmath220 if @xmath221 and @xmath222 .
[ symbol properties ] the following properties hold .
\a ) @xmath223 .
\b ) if @xmath224 and @xmath225 , the product @xmath226 belongs to @xmath227 .
\c ) if @xmath228 and @xmath229 with @xmath230 in @xmath213 ( @xmath136 is said to be elliptic positive ) , then @xmath231 .
moreover , @xmath232 uniformly for all constants @xmath233 . to establish the symbol properties of the phase @xmath234 in the formula for the friedlander wave trace @xmath46
, we will need a well known asymptotic formula for the roots of the airy function .
( a proof is given in appendix a ; see also @xcite page 450 . )
[ airy roots ] there is a symbol @xmath235 such that @xmath236 for all @xmath237 , @xmath238 are the roots of @xmath42 , and the principal symbol of @xmath239 is @xmath240 ( @xmath241 ) . denote @xmath242 then @xmath243 and the sum can be written @xmath244 let @xmath245 .
the first quadrant is covered by the union of three overlapping sectors @xmath246 one can deduce easily from proposition [ airy roots ] the following .
[ friedl symbol ] let @xmath247 .
then \a ) @xmath248 , and @xmath249 is positive , elliptic in @xmath250 .
\b ) @xmath251 with principal symbol @xmath252 given by @xmath253 \c ) @xmath254 , and @xmath255 is positive elliptic in @xmath256 .
we are now ready to finish the proof of theorem [ friedl theorem ] .
first , we divide the sum representing @xmath46 into three regions corresponding to different behaviors of the symbol @xmath257 .
let @xmath258 be supported in @xmath259 and such that @xmath260 let @xmath261 satisfy @xmath262 for all @xmath263 and @xmath264 for all @xmath265 .
then @xmath266 the method of section [ sec : smoothsymbols ] applies in the sector @xmath267 .
the times @xmath82 at which singularities can occur are the stationary points @xmath159 of the phases of the representation using the poisson summation formula , namely , solutions to @xmath268 furthermore , on @xmath269 , @xmath270 thus if @xmath271 , then @xmath272 , exactly the points of the length spectrum @xmath44 .
the set of @xmath159 is @xmath273 ( evidently , @xmath274 is a stationary point ; there are no roots @xmath275 , @xmath276 and no roots @xmath277 , @xmath278 of .
the rest of the range @xmath279 is covered by the symmetries @xmath280 and @xmath281 . )
the restriction to @xmath267 gives a further limitation on the set of stationary points .
define a smaller collection of stationary points by @xmath282 @xmath283 is a closed , discrete set in @xmath284 . for @xmath285
sufficiently large depending on @xmath286 , all the stationary values that occur in the the range @xmath287 are in @xmath283 . because @xmath283 has no limit points , and the symbol of the phase has classical behavior in @xmath267 , the method of section [ sec : smoothsymbols ] applies and yields the following . [ i2smooth ] for @xmath285 sufficiently large depending on @xmath286 , @xmath288 is smooth in the complement of @xmath289 in particular , for any @xmath290 and any @xmath291 there is @xmath26 such that @xmath292 all the remaining singularities in @xmath293 are accounted for in @xmath294 , associated with @xmath295 .
the main issue is the limit points at @xmath296 .
our theorem in @xmath294 is stated in sufficient generality to apply as well to the case of the disk .
[ mainbyparts ] let @xmath297 be a classical symbol of degree @xmath162 in one variable supported in @xmath298 and let @xmath299 be a classical symbol of degree @xmath162 supported in a cone @xmath294 given by @xmath300 .
suppose that the distribution @xmath301 is defined by @xmath302 where @xmath303 belongs to @xmath304 , with @xmath305 is positive elliptic in @xmath306 on the support of @xmath307 . for any @xmath157 there exist @xmath26 and @xmath291 sufficiently small depending on @xmath21 and the constant in the symbol upper bound on @xmath249 such that \a ) if @xmath308 , @xmath290 , then @xmath301 is smooth on @xmath309 ; \b ) if @xmath310 , then @xmath301 is smooth on @xmath311 . in the proof of part ( a )
we will assume for simplicity that @xmath312 .
the other cases are similar .
let @xmath313 .
the poisson summation formula says @xmath314 with @xmath315 denote @xmath316 and define two differential operators in order to integrate by parts as follows .
@xmath317 @xmath318 then @xmath319 .
let us start with the most singular case @xmath320 . because @xmath321 , we have @xmath322
the main point is that by proposition [ symbol properties ] , @xmath323 uniformly in @xmath324 .
let @xmath325 and @xmath326 denote the adjoint operators to @xmath327 and @xmath328 .
then proposition [ symbol properties ] also implies that @xmath329 with constants in the estimates independent of @xmath330 . @xmath331 since @xmath332 , the symbol @xmath333 belongs to @xmath334 with constants @xmath335 .
note that symbols in @xmath336 supported in the cone @xmath337 are integrable if @xmath338 .
therefore it suffices to take @xmath339 to get a convergent integral . in case @xmath340 , take @xmath341 and @xmath342 large .
otherwise , take @xmath343 so that the sum over @xmath330 converges .
so far we have shown that that sum over @xmath344 is bounded . to prove smoothness in @xmath82 , differentiate @xmath23 times .
this yields an extra factor @xmath345 in the integrand .
therefore , @xmath346 is represented by sum of integrals of symbols @xmath347 with constants @xmath335 .
the integrals converge for @xmath342 large , for instance @xmath348 .
again , take @xmath341 when @xmath340 and @xmath343 when @xmath278 . next ,
consider @xmath349 .
the same estimates as before hold for @xmath325 . recall that we may assume that @xmath350 is zero on a fixed compact subset .
provided @xmath82 is close enough to @xmath351 , and @xmath286 sufficiently small , and @xmath352 sufficiently large on the support of @xmath350 , we have @xmath353 and hence @xmath354 and @xmath355 hence , @xmath356 with constants @xmath357
. so the symbol @xmath358 belongs to @xmath359 with constants @xmath360 .
the integral converges provided @xmath361 .
when @xmath340 , take @xmath341 ; when @xmath278 take @xmath362 . then the sum over @xmath363 , with @xmath364 also converges .
lastly , if we apply @xmath365 , we obtain integrands that belong to @xmath366 with constants @xmath360 , so the symbols are integrable if @xmath367 . to sum the series , for @xmath278 , @xmath364 , let @xmath343 . for @xmath340 , @xmath364 ,
use @xmath341 .
this concludes the proof of part ( a ) of theorem [ mainbyparts ] . for part ( b ) , once again for any @xmath368 we can choose @xmath291 sufficiently small that on the support of @xmath350 ( where @xmath369 is sufficiently large ) @xmath370 let @xmath371 be the integer such that @xmath372 is smallest , and choose both @xmath368 and @xmath26 less than @xmath373 .
then for @xmath374 , @xmath375 on the support of @xmath350 . hence , @xmath376 and the rest of the proof proceeds as in the case @xmath364 above .
this concludes the proof of theorem [ mainbyparts ] . the next proposition will take care the integral @xmath377 .
[ i3smooth ] let @xmath378 satisfy @xmath379 for @xmath198 sufficiently large and @xmath380 for @xmath381 .
suppose that where @xmath140 belongs to @xmath382 and @xmath383 is positive elliptic in @xmath384 on the cone @xmath385 .
for each @xmath386 there exists @xmath387 such that if @xmath299 is a classical symbol of degree @xmath162 supported in a cone @xmath213 given by @xmath388 , then the distribution @xmath389 satisfies @xmath390 .
as usual , the poisson summation formula implies @xmath314 with @xmath391 denote @xmath316 .
then @xmath392 and @xmath393 with @xmath394 , @xmath395 .
since @xmath255 is positive elliptic in @xmath396 , @xmath397 belongs to @xmath398 with support on @xmath213 and bounds depending on @xmath399 . by induction ,
@xmath400 with support in @xmath401 .
thus @xmath402 is represented by a convergent integral . moreover ,
if @xmath403 , we may differentiate @xmath23 times with respect to @xmath82 to obtain an integrand dominated by @xmath404 which is integrable if @xmath405 .
for @xmath276 , use @xmath406 to integrate by parts @xmath342 times to obtain @xmath407 \ , d\xi d\eta\ ] ] the integrand is majorized by @xmath408 which is convergent in @xmath213 for sufficiently large @xmath285 depending on @xmath409 and @xmath342 .
( note that because @xmath350 is supported in @xmath410 , @xmath411 has compact support , and every term in which a derivative falls on @xmath412 is convergent . ) the factor @xmath413 makes the sum over @xmath414 convergent as well .
we use the same integration by parts as in case 1 .
note that the set of @xmath415 where @xmath416 and @xmath417 is finite .
any finite sum of exponentials is smooth in @xmath82 , so we may assume without loss of generality that @xmath418 is supported on @xmath419 for some large @xmath420 .
it follows that @xmath421 on the support of @xmath422 .
furthermore , if @xmath387 is sufficiently large , depending on @xmath21 , @xmath423 thus @xmath424 with bounds @xmath425 . by induction ,
@xmath426 with bounds @xmath427 .
the rest of the proof proceeds as in case 1 . if @xmath403 , then no more integrations by parts are needed .
if @xmath276 , then one uses integration by parts in @xmath198 as in case 1 to introduce a factor @xmath428 so as to be able to sum in both @xmath330 and @xmath414 .
finally , we assemble these ingredients to prove theorem [ friedl theorem ] .
part ( a ) was already proved at the end of section [ sec : geodesics ] . for part(b ) , choose @xmath429 , and choose @xmath430 so that @xmath431 \cap ( \bar\l_f \cup ( -\bar\l_f ) \cup\{0\ } ) = \emptyset\ ] ] choose @xmath286 sufficiently small depending on the symbol bounds and the distance from @xmath432 $ ] to @xmath433 and @xmath387 sufficiently large depending on the size of @xmath21 , we may apply theorem [ mainbyparts ] ( b ) , proposition [ i2smooth ] and proposition [ i3smooth ] , respectively , to @xmath295 , @xmath288 and @xmath377 to obtain smoothness in @xmath311 . note that proposition [ friedl symbol ] implies that the correponding phases satisfy the appropriate symbol estimates . for part ( c ) ,
apply theorem [ mainbyparts ] ( a ) to @xmath295 , to obtain smoothness in @xmath434 . by propositions [ i2smooth ] and [ i3smooth ]
@xmath288 and @xmath377 are smooth near @xmath435 on both sides .
this concludes the proof of theorem [ friedl theorem ] .
the classical mechanical system associated to the friedlander model has two commuting conserved quantities , @xmath436 in this section we will show how the action variables associated to this system determine the first order asymptotics of the eigenvalues and eigenfunctions of the friedlander operator @xmath32 .
( see appendix b for the general framework of this calculation . )
let @xmath437 and @xmath438 be real numbers satisfying @xmath439 .
the lagrangian submanifold @xmath440 is a torus foliated by trajectories with initial velocity @xmath441 on @xmath35 , with @xmath442 , @xmath443 and @xmath444 .
these trajectories are tangent to the caustic @xmath445 depicted as a dotted line in the figure below and in section [ sec : geodesics ] .
fix @xmath56 , so that @xmath92 .
denote by @xmath446 the trajectory given by proposition [ prop : reflection ] on the interval @xmath447 .
define by @xmath448 the loop ( whose projection onto @xmath449 is pictured in bold in the figure below ) that follows @xmath450 on its two curved portions and then the segment of the caustic @xmath451 , with @xmath452 , @xmath453 ( oriented by @xmath454 ) going from @xmath455 to @xmath456 .
define by @xmath457 the loop that follows the caustic @xmath451 , @xmath458 , @xmath452 , and @xmath459 , with orientation @xmath460 .
the loops @xmath448 and @xmath457 are a basis for the homology of the torus @xmath56 , @xmath461 , and the action coordinates are defined by @xmath462 since @xmath463 is constant , @xmath464 , and @xmath465 , we have @xmath466 moreover , @xmath467 more generally , by homogeneity , @xmath468 thus @xmath54 and @xmath469 can be written as functions of the action variables , @xmath470 define @xmath471 as the values of @xmath472 when the action variables belong to the integer lattice @xmath473 , @xmath415 .
( since @xmath474 , we consider only @xmath475 . ) then @xmath476 the bohr - sommerfeld energy levels of @xmath54 are defined as the values of @xmath54 on the lattice , namely @xmath477 .
[ prop : asympt1 ] in any sector of the form @xmath478 , the eigenvalues @xmath53 of the friedlander operator satisfy @xmath479 the proposition is an immediate consequence of the formula for @xmath480 and the fact that @xmath481 with @xmath482 , @xmath483 ( see proposition [ airy roots ] ) . next , we show that the phases of the eigenfunctions are equal , asymptotically , to the generating function of the canonical relation of the lagrangian submanifold . indeed ,
if @xmath484 is the generating function , then by definition , the lagrangian is the graph of @xmath485 : @xmath486 thus , @xmath487 and @xmath488 notice that the upper limit of the range of @xmath36 is exactly the caustic @xmath489 so that the graph over this interval in @xmath36 ( and all @xmath66 ) covers the whole lagrangian submanifold .
now let the submanifold be given by @xmath473 .
then @xmath490 and we have two generating functions @xmath491 to compare this with the phase of @xmath492 , recall that for @xmath493 , @xmath494 ( see proposition [ friedl spectrum ] and appendix a. ) thus @xmath492 is a sum of two terms with phases @xmath495 -\pi/2 + ny = \rho_\pm(x , y ) + o(1),\ ] ] as @xmath150 in sectors of the form @xmath496 ( since @xmath497 ) .
an airy function is a solution @xmath95 of the differential equation @xmath498 .
the classical properties of airy functions are treated in @xcite , section 10.4 and @xcite , section 7.6 , and reviewed briefly here .
the _ airy function _ ( or _ airy integral _ ) , denoted @xmath37 , was introduced in 1838 by g.b .
airy in the remarkable paper @xcite in order to describe the intensity of the light near a caustic .
this allowed him to propose a good theory for the rainbow .
he defined the function @xmath37 by the oscillatory integral @xmath499 in other words , @xmath37 is the inverse fourier transform of @xmath500 .
the airy function @xmath501 is positive for @xmath502 and decays exponentially as @xmath503 ; it is oscillating for @xmath504 with an expansion , given by the method of stationary phase , of the form @xmath505 , \quad t < 0,\ ] ] with @xmath506 and @xmath507 a classical symbol of order @xmath508 .
differentiating , we find @xmath509 , \quad t < 0,\ ] ] with @xmath510 a classical symbol of order @xmath508 .
the function @xmath37 has infinitely many real negative zeros @xmath511 to derive the asymptotic formula for @xmath512 , denote @xmath513 the foregoing asymptotic expansions can be written @xmath514 where @xmath515 is a complex valued classical symbol of order @xmath508 .
define @xmath516 @xmath517 for all @xmath518 , and specify the branch of @xmath519 by @xmath520 in that range .
thus @xmath519 is a classical symbol of order @xmath109 with principal part @xmath521 as @xmath522 , and @xmath523 [ prop : airy appendix ] let @xmath519 be as above .
then @xmath524 for all @xmath525 , and there is an absolute constant @xmath526 such that @xmath527^{2/3 } > 0 , \
\mbox{for } \ \xi\ge c_1,\ ] ] is well defined ; @xmath235 with principal term @xmath528 . finally , @xmath529 where @xmath530 are the zeros of @xmath37 .
@xmath531 , \quad a= \ai(-s^{2/3 } ) , \quad b = \ai'(-s^{2/3}).\ ] ] therefore , @xmath524 for @xmath532 .
because @xmath519 is increasing , the zeros of @xmath37 are given by @xmath533 let @xmath534 .
@xmath535 because @xmath536 .
the inverse function @xmath537 of @xmath519 is a classical symbol of order @xmath109 on @xmath538 .
then @xmath539 is defined for all @xmath540 and satisfies all the properties of the proposition .
now we turn to proof of proposition [ friedl spectrum ] describing the spectrum of the friedlander operator @xmath32 .
begin by noting that if @xmath32 is restricted to functions @xmath541 of @xmath36 alone , it becomes @xmath542 which has completely continuous spectrum on @xmath543 .
next , consider @xmath544 in the orthogonal complement , @xmath38 by separation of variables , the eigenfunctions are of the form @xmath545 , where @xmath546 is a non - zero integer , and @xmath547 satisfies @xmath548 for @xmath549 , the operator @xmath550 acting on functions satisfying @xmath551 is self - adjoint and has compact resolvent , so a complete system of eigenfunctions is obtained by solving for @xmath552 and @xmath547 such that @xmath551 and @xmath553 . making the change of variables @xmath554
equation becomes @xmath555 choose @xmath82 so that @xmath556 , then the equation is the airy equation @xmath557 since @xmath547 is in @xmath543 , the same is true of @xmath327 .
up to a constant multiple , there is a unique such function @xmath327 , the airy function @xmath37 .
the dirichlet condition @xmath551 implies @xmath42 , and the roots of this equation are @xmath558 .
hence the formulas in the proposition for the eigenfunctions and eigenvalues are confirmed .
let @xmath559 be an n - dimensional manifold and @xmath560 , , @xmath561 be commuting self - adjoint zero order semiclassical pseudodifferential operators with leading symbols @xmath562 , , @xmath563 . in section
[ sec : actions ] we considered the example @xmath564 , @xmath565 ( these operators are of order zero with respect to @xmath566 , @xmath567 and multiplication by @xmath568 . ) let @xmath575 be the graph of @xmath576 in @xmath577 .
@xmath575 is a lagrangian submanifold and determines @xmath578 up to an additive constant ; @xmath578 is known as the _ generating function _ of @xmath575 .
setting @xmath579 , one finds the _ eikonal equation _
, the first of the heirarchy of equations for @xmath136 , @xmath578 , and @xmath580 : @xmath581 let us confine ourselves to the sequence @xmath582 for integers @xmath583
. then defines @xmath544 provided @xmath578 is well - defined modulo @xmath433 . to see what this entails ,
denote by @xmath584 the projection , @xmath585 , and by @xmath586 the inclusion , @xmath587 , and denote @xmath588 the canonical 1-form on @xmath577 . the fact that @xmath575 is the graph of @xmath589 implies @xmath590
if @xmath591 is a regular value of @xmath592 and @xmath575 is compact , then @xmath575 is a torus of dimension @xmath546 .
thus if @xmath593 , , @xmath594 is a basis of the first homology group of @xmath575 , then implies the functions @xmath596 are known as action coordinates on @xmath577 .
let @xmath597 . in the generic case
in which @xmath598 defines an @xmath546-dimensional lagrangian torus in @xmath577 , the torus is known as a bohr - sommerfeld surface , and we will denote it by @xmath599 . proposition [ prop : globalcanonical ] says that the generating function @xmath600 of @xmath599 is well - defined modulo @xmath601 .
thus one can follow the ansatz to seek a joint eigenfunction @xmath544 whose phase function @xmath578 equals @xmath600 to first order as @xmath573 and whose eigenvalue @xmath602 with respect @xmath571 is given to first order by @xmath603 evaluated on @xmath599 as in . under suitable further hypotheses , all but finitely
many joint eigenfunctions are obtained in this way .
typically , @xmath544 and @xmath599 exist for a cone of values of @xmath597 ( see @xcite ) .
|
in a recent preprint , @xcite , we showed that for the dirichlet laplacian @xmath0 on the unit disk , the wave trace @xmath1 , which has complicated singularities on @xmath2 , is , on the interval @xmath3 , the restriction to this interval of a @xmath4 function on its closure . in this paper
we prove the analogue of this somewhat counter - intuitive result for the friedlander model .
the proof for the friedlander model is simpler and more transparent than in the case of the unit disk .
|
given a potential @xmath1 , with @xmath2 , the surface defined by @xmath1 is called the potential energy landscape ( pel ) of the given system @xcite .
the special points of a pel , defined by the solutions of the equations @xmath3 for @xmath4 , provide important information about the pel .
these special points , called critical points or stationary points ( sps ) of the pel , can be further classified according to the number of negative eigenvalues of the hessian matrix , @xmath5 .
the sps at which @xmath6 is positive ( negative ) definite are called minima ( maxima ) of the pel and the sps at which @xmath6 has exactly @xmath7 negative eigenvalue are called saddles of index @xmath7 .
sps at which @xmath6 has at least one zero eigenvalue , after removing the global symmetries from the system corresponding to overall translation and rotation , are called singular sps or non - morse points .
the sps of the pel can be employed to calculate or estimate certain physical quantities of interest .
a variety of techniques have been deveoped within the framework of potential energy landscape theory @xcite , with applications to many - body systems as diverse as metallic clusters , biomolecules and their folding transitions , and glass formers . except for rare examples , like the one - dimensional xy model @xcite
, it is not usually possible to obtain the sps analytically because solving the nonlinear stationary equations can be an extremely difficult task .
hence , one has to rely upon numerical methods .
when a numerical method finds a solution of a given system , it essentially means that it has found a numerical approximation of an exact solution . after achieving a numerical approximate
, one can heuristically validate it by either monitoring iterations of newton s method or by substituting the approximations into the equations to see if they are satisfied up to a chosen tolerance .
usually , such a validation works well in practice .
however , as will be clear from the examples provided below , such heuristic approaches do not guarantee that the numerical approximation will indeed converge quadratically to the associated solutions using arbitrary precision . more specifically ,
even if a numerical approximation is heuristically validated , it could turn out to be a nonsolution at higher precision , or newton iterations may have unpredictable behavior , such as attracting cycles and chaos , when applied to points that are not in a basin of attraction @xcite of some solution .
we note that if the given system is a set of polynomial equations , then one can use numerical polynomial homotopy continuation @xcite to compute all the isolated solutions ( see e.g. @xcite for some related approaches ) .
briefly , the method works as follows : first , one determines an upper bound on the number of isolated complex solutions of the given system .
the highest upper bound on the number of solutions is the so - called classical bezout bound , which is the product of the degree of each polynomial equation of the system , but there are several other tighter upper bounds available for structured systems .
then , one constructs another system that has exactly the same number of solutions as the upper bound , such that the new system is easy to solve .
finally , one uses continuation to track each solution of the new system to the original one . some of the paths may diverge to infinity , which will happen whenever the upper bound is larger than the true number of solutions .
the paths that converge will tend to solutions of the original system in appropriate limits .
this method is quite different from conventional numerical approaches , in that it is guaranteed to find all solutions , in principle .
however , due to the numerical computations used with this method in path tracking , in the end , only numerical approximations are obtained , and hence the above mentioned difficulties may also arise .
the goal of this paper is to develop a rigorous way to validate the numerical approximates of the sps , independent of the numerical method used to obtain them . in the mathematics literature , proving that a given numerical approximation will converge quadratically to the nearby associated solution using arbitrary precision is called _
certification_. it is well known that quadratic convergence doubles the number of correct digits after each iteration .
hence the associated solution can be approximated to a given accuracy quite efficiently after a certain number of newton iterations .
smale and others , in the 1980 s , developed a method that certifies a numerical approximation as an actual solution of the system @xcite .
the method is now known as smale s @xmath0-theory .
interestingly , it turned out that the certification could be done via computing three numbers from a given numerical approximate : for a given system of equations @xmath8 and a given point @xmath9 , one computes two numbers @xmath10 and @xmath11 ) , which guarantee that newton s method starting from @xmath9 will quadratically converge to a solution of @xmath8 if the number @xmath12 is less than @xmath13 . applying this certification scheme ensures that our numerical solutions are good enough so that more accurate approximations can be obtained easily and efficiently @xcite . in this paper ,
we first give details of smale s @xmath0-theory in the context of the pels in section [ sec_smales_theory ] .
then , by providing examples in section [ sec_examples ] , we show how numerical approximates may turn out to be non - solutions even in seemingly simple situations . in passing ,
we certify the solutions of the well - known examples of the wilkinson polynomial under a small perturbation and the roots of the chebyshyv polynomials of the first kind for the first time . in section [ sec : xy_model ] , we consider a more physically relevant potential , i.e. , the two - dimensional xy model without disorder .
we certify all the known sps of this model and provide a guide for conventional numerical methods .
section [ sec_conclusions ] provides an outlook and conclusions .
in this section , we describe smale s @xmath0-theory following ref . @xcite . we restrict ourselves to square systems , i.e. , systems that have the same number of equations as variables , since sps of a pel satisfy a square system of equations .
smale s @xmath0-theory is usually used to certify complex solutions for systems of analytic functions , so we start by describing this approach @xcite .
we then discuss the certification of real solutions separately .
finally , we will discuss @xmath0-theory applied to polynomial systems and systems involving exponentials and trigonometric functions .
we start by considering a system @xmath14 of @xmath15 multivariate analytic equations in @xmath15 variables .
we denote the set of solutions of @xmath8 as @xmath16 and the jacobian of @xmath14 at @xmath17 as @xmath18 .
consider the newton iteration of @xmath14 starting at @xmath17 defined by @xmath19 for @xmath20 , the @xmath21-th newton iteration is simply @xmath22 now , a point @xmath23 is called an _ approximate solution _ of @xmath14 with _ associated solution _ @xmath24 if , for each @xmath20 , @xmath25 where @xmath26 is the standard euclidean norm on @xmath27 , i.e. , @xmath17 is an approximate solution to @xmath14 if it is in the quadratic convergence basin defined by newton s method of some solution @xmath28 .
the following theorem provides a sufficient condition for proving that a given point is an approximate solution without knowledge about @xmath28 .
* theorem : * if @xmath29 for a square analytic system @xmath14 and point @xmath17 such that @xmath30 exists , then @xmath17 is an approximate solution to @xmath14 , where @xmath31 if @xmath17 is an approximate solution of @xmath14 , then @xmath32 , where @xmath24 is the associated solution to @xmath17 .
moreover , in @xmath33 , the term @xmath34 is the symmetric tensor whose components are the partial derivatives of @xmath14 of order @xmath21 . finally , for convenience
, one can remove the condition on @xmath18 by defining @xmath0 , @xmath35 , and @xmath36 appropriately .
if @xmath37 such that @xmath18 is not invertible , define @xmath38 , @xmath39 and @xmath40 . if @xmath41 such that @xmath18 is not invertible , then @xmath42 , @xmath43 and @xmath33 are taken as @xmath44 . since this theorem provides a sufficient condition for a point to be an approximate solution ,
the set of _ certifiable _ approximate solutions is generally much smaller than the set of approximate solutions , as demonstrated in figures [ fig : z4 - 1 ] and [ fig : cheby ] .
however , if it is a true approximate solution , then a few newton iterations usually generate a point that can be certified . *
distinct complex solutions * given two approximate solutions @xmath45 and @xmath46 , one often needs to verify that the corresponding associated solutions @xmath47 and @xmath48 are distinct . one way to check this condition uses the triangle inequality together with @xmath49 .
for arbitrary analytic systems , @xmath33 as described in the above theorem may be difficult to compute or bound above .
when @xmath14 is polynomial , @xmath36 is defined as a maximum over finitely many terms and thus can be computed in theory .
often , however , such as in the software package alphacertified @xcite , @xmath33 is bounded above via proposition 8 of @xcite , which depends on the degrees and coefficients of the polynomials in @xmath14 , @xmath50 , and @xmath51 , which we now briefly summarize .
let @xmath52 and @xmath53 be a polynomial in @xmath15 variables of degree @xmath54 .
define @xmath55 and , by writing @xmath56 , define @xmath57 for a system @xmath14 of @xmath15 polynomials in @xmath15 variables , let @xmath58 if @xmath51 is invertible , define @xmath59 where @xmath60 with @xmath61 and @xmath62.\ ] ] * proposition : * if @xmath14 is a polynomial system with @xmath61 , @xmath63 , and @xmath52 such that @xmath51 is invertible , then @xmath64 an upper bound on @xmath33 also exists for systems of polynomial - exponential equations @xcite .
a system is polynomial - exponential if it is polynomial in both the variables @xmath65 and finitely many exponentials of the form @xmath66 where @xmath67 .
many standard functions such as @xmath68 , @xmath69 , @xmath70 , and @xmath71 can be formulated as systems of polynomial - exponential functions since they are indeed polynomial functions of @xmath72 for suitable @xmath67 .
the above theorem provides a bound on the distance between an approximate solution @xmath17 and its associated solution @xmath28 , namely @xmath73 .
apart from certifying that @xmath17 is indeed an approximate solution , one often wants to prove additional information about @xmath28 .
the theory of newton - invariant sets @xcite provides one approach to this problem .
one particular newton - invariant set of particular interest is @xmath74 when @xmath75 defines a real map , that is , @xmath76 for @xmath77 , which was first observed in @xcite . in this case , one is able to certifiably determine if @xmath78 or @xmath79 given any approximate solution @xmath17 associated with @xmath28 .
the newton iteration corresponding to a potential energy function @xmath80 is a real map if @xmath1 is real for all real @xmath17 .
therefore , one is able to certifiably determine if an associated solution is real or nonreal .
this reality test and other @xmath0-theoretic computations are implemented in the software alphacertified @xcite , which we describe next .
the basin of attraction of each minimum is an important quantity in potential energy landscape studies since the sum of the volumes of all the basins of attraction is related to the entropy of the system .
there is a crucial difference between the basins of attraction and the certifiable region of the same minimum .
the basin of attraction of a minimum can not overlap with that of another minimum , and aside from boundaries the union of all the basins of attraction covers the whole configuration space .
however , the certified region of a minimum is contained within the basin of attraction of the minimum .
hence , the sum of the certifiable regions is generally a subset of this space .
the software alphacertified @xcite performs computations related to @xmath0-theory .
the input system must be presented exactly with rational coefficients . when the internal computations are performed using exact rational arithmetic via gmp @xcite , all results are rigorous and
thus provide a _ mathematical proof of the computed results_.
this framework provides an alternative to other analytic or symbolic computations for yielding computational proofs . since each solution can be independently certified , the procedures are parallelizable . from the exact input system , which is either
a system of polynomial or polynomial - exponential functions , the jacobian is constructed exactly . for these systems ,
the value of @xmath36 is bounded above , as discussed in
[ sec : specialnonlinear ] , eliminating the need to compute the higher - order derivatives .
hence , @xmath0 is also bounded above . since the magnitude of the rational numbers can increase during a sequence of exact computations , alphacertified also permits the use of arbitrary precision floating - point arithmetic via mpfr @xcite . when using floating - point arithmetic , the round - off errors are not explicitly controlled but can be reduced by increasing the precision .
in this section , we provide several other example systems illustrating many possible scenarios in which a conventional numerical method may face difficulty in obtaining numerical approximates .
we start with a simple example of one equation in one variable followed by a few other examples exhibiting different numerical issues . for the system @xmath81 , we demonstrate the explicit computation of @xmath82 and @xmath36 . since @xmath83 is zero if and only if @xmath84 and @xmath85 , we can safely assume that @xmath86 .
thus , @xmath87 .
now , the term @xmath88 in @xmath89 is simply the @xmath21-th derivative of @xmath14 at @xmath90 , namely @xmath91 . since @xmath14 has degree @xmath92
, we only need to take the maximum over @xmath93 to compute @xmath89 .
it is easy to show that the maximum is attained at @xmath94 with @xmath95 .
thus , @xmath96 for @xmath86 . for @xmath97
, we have @xmath98 and thus @xmath97 can not be certified as an approximate solution .
it is indeed outside of all of the quadratic convergence basins .
however , at the point @xmath99 , @xmath100 .
thus , @xmath99 is certifiably an approximate solution of @xmath8 . in this case , we know that the associated solution is @xmath101 , and table [ tab : i ] confirms the quadratic convergence for a few iterations . figure [ fig : z4 - 1 ] plots the basins of convergence starting at points @xmath102 for @xmath103 of newton s method applied to @xmath14 . in this plot , the white areas are the certifiable quadratic convergence basins which lie inside of the respective quadratic convergence basins .
these quadratic convergence basins lie inside of the respective linear convergence basins with chaotic behavior separating the linear convergence basins .
the structure is similar to that observed for convergence of alternative optimisation algorithms for atomic clusters in previous work @xcite .
.convergence to @xmath101 for @xmath104 starting at @xmath105 .
[ cols="^,^,^,^,^,^",options="header " , ] using @xmath0-theory , one can also certifiably determine the index .
this certificate would protect against a true solution having a small eigenvalue that changes sign when using an approximation of the solution .
due to complications of expanding determinants for large matrices , we use a eigenvalue vector / value formulation at the expense of increasing the size of the system .
the resulting system is of the form @xmath106 = { \bf 0}\ ] ] where @xmath107 is the hessian of @xmath80 .
since @xmath108 is defined up to scaling , we dehomogenize using a sufficiently random patch and then convert to a polynomial system with the same subsitution as above .
solving nonlinear equations is one of the most frequently arising mathematical problems in theoretical chemistry , e.g. , in finding minima and transition states of a potential energy function or finding steady states of chemical rate equations , etc .
it is customary to resort to a numerical method to solve such systems , aside from the rare instances when the equations can be solved exactly . for a numerical method , a solution of a given system means a numerical approximation of an actual solution .
it is quite likely that such a numerical approximation is a nonsolution of the system , i.e. , it may lie in the linear convergence basin or in a chaotic region , instead of the quadratic region of convergence . in many cases , such false numerical approximations
may leave us with fundamentally different , and incorrect conclusions . in our previous paper @xcite , we employed smale s @xmath0-theory which _ certifies _ if a given numerical approximation is in the quadratic convergence region of an actual solution of the system . in the present work ,
we have elaborated the mathematical concepts and a related software called alphacertified .
we then used the certification procedure to explore the solution space of a simple problem , @xmath109 , where @xmath110 , which already shows several regions where a numerical method may fall into chaotic regions .
we also showed the certified regions of the known four solutions of this system .
then , we picked two celebrated examples , the wilkinson polynomial and a chebyshev polynomial of the first kind . for both systems , we explored the solution spaces and identified the certified regions as well as regions of linear convergence and chaotic convergences .
finally we considered the xy model in two dimensions without disorder .
there , we took the already known numerical approximations of the saddles of all the possible indices , for the lattice sizes up to @xmath111 and tried to certify the solutions .
we chose this system to demonstrate the applicability of the @xmath0-theory to systems that are not in the polynomial form . in these systems , the higher the lattice dimension is the higher the required accuracy from a numerical method , otherwise the numerical approximations do not fall into the quadratic convergence region , as for the lennard - jones ( lj ) potential @xcite .
we used alphacertified to refine the numerical approximations whenever they were not in the quadratic convergence regions of the corresponding solutions .
we also observe that , in contrast to the lj case @xcite , for the xy model , the size of the quadratic convergence region ( or the certified region ) remains fairly constant as @xmath112 increases .
one needs to be careful here because this difference may also arise if the technique used to find the sps did not find the more ill - conditioned solutions ( the ones with smaller quadratic convergence basins ) .
we anticipate that the certification method presented in this paper will turn out to be a standard tool to verify the numerical sps obtained from various numerical methods .
another possible application is to make rigorous statements about sps random potential surfaces , such as the ones studied in statistical physics and cosmology , as well as in pure mathematics @xcite .
dm and jdh were supported by darpa young faculty award .
jdh was also supported by the national science foundation through dms-1262428 .
djw and dm gratefully acknowledge support from the epsrc and the erc . d. martinez - pedrera , d. mehta , m. rummel and a. westphal , jhep * 1306 * , 110 ( 2013 ) y .- h .
he , d. mehta , m. niemerg , m. rummel , and a. valeanu , jhep * 1307 * , 050 ( 2013 ) .
d. mehta , d. a. stariolo and m. kastner , phys .
e 5 * 87 * 052143 ( 2013 ) .
b. greene , d. kagan , a. masoumi , d. mehta , e. j. weinberg and x. xiao , phys .
d * 88 * , 026005 ( 2013 ) .
a. j. sommese and c. w. wampler , , world scientific publishing company , 2005 .
|
typically , there is no guarantee that a numerical approximation obtained using standard nonlinear equation solvers is indeed an actual solution , meaning that it lies in the quadratic convergence basin .
instead , it may lie only in the linear convergence basin , or even in a chaotic region , and hence not converge to the corresponding stationary point when further optimization is attempted . in some cases
, these non - solutions could be misleading .
proving that a numerical approximation will quadratically converge to a stationary point is termed _
certification_. in this report , we provide details of how smale s @xmath0-theory can be used to certify numerically obtained stationary points of a potential energy landscape , providing a _ mathematical proof _ that the numerical approximation does indeed correspond to an actual stationary point , independent of the precision employed .
|
10 url # 1#1urlprefix[2][]#2 aad g _
et al . _ ( atlas collaboration ) 2012 _ phys.lett . _ * b716 * 129 ( _ preprint _ )
|
in this talk the methods and computer tools which were used in our recent calculation of the three - loop standard model renormalization group coefficients are discussed .
a brief review of the techniques based on special features of dimensional regularization and minimal subtraction schemes is given .
our treatment of @xmath0 is presented in some details .
in addition , for a reasonable set of initial parameters the numerical estimates of the obtained three - loop contributions are presented .
the renormalization group ( rg ) proves to be a useful and powerful tool in studying high - energy behavior of the standard model . before the discovery of the higgs boson rg equations ( rge ) were used , among other things , to bound the value of the higgs self - coupling .
however , the bounds significantly depends on the scale at which one expects the appearance of new physics .
the observation of the higgs boson in 2012 @xcite in some sense finalizes the sm since the information about the values of all the sm couplings become available from experiments . due to this fact
, the interest to rg studies of the standard model arises again , but at a new level of precision .
one- and two - loop results for sm beta - functions have been known for quite a long time @xcite and are summarized in @xcite . the first paper with full three - loop calculation of gauge coupling beta - functions within the sm was published in ref .
@xcite .
the next step was carried out by another group from karlsruhe @xcite , which considered , in the specific limit , the three - loop beta - functions for the top quark yukawa coupling , the higgs self - coupling and mass parameter . at this state of things
our group entered the game .
we not only confirmed the results of refs .
@xcite but also provided the three - loop expressions for beta - functions of all the yukawa couplings @xcite corresponding to the fermions of third generation .
contrary to chetyrkin and zoller , we include the dependence on electroweak couplings . in the beginning of 2013
our group started the calculation of missing three - loop terms in beta - functions of the higgs potential parameters .
it turns out that the same problem was considered by the authors from karlsruhe .
unfortunately to us , they managed to obtain and make their results @xcite public a week earlier than our group @xcite .
nevertheless , in such a complicated calculation it is important to have a confirmation from an independent source . in what follows , we are going to discuss the peculiarities of the procedure used to obtain the results published by our group in a series of papers @xcite .
first of all , let us mention that the calculation of the three - loop sm beta - functions requires evaluation of millions of feynman diagrams .
this task definitely requires automatization by means of a computer .
fortunately , all the necessary tools were available on the market , so we only needed to combine them in a proper way .
it is due to nice features of dimensional regularization and minimal subtraction scheme so one can significantly simplify the calculation . since in @xmath1-scheme all the renormalization constants can be extracted from the ultraviolet ( uv ) divergent parts of corresponding green functions , one can modify infra - red ( ir ) structure of the model to simplify the calculation of counter - terms .
this is the essence of the so - called infrared - rearrangement ( irr ) trick , which was originally proposed by vladimirov a.a @xcite .
this kind of modifications can lead to a spurious ir divergences which should be removed consistently by the so - called @xmath2 @xcite .
however , in many practical cases one can avoid this kind of complications .
in our series of paper we made use of two variants of ( naive ) irr procedure .
for the calculation of gauge and yukawa coupling beta - functions it is possible to convert all the required two- and three- point green functions to the massless propagator - type feynman integrals .
it is done via neglecting all internal masses and setting the higgs boson external momenta entering yukawa vertex to zero .
the evaluation of massless three - loop propagators is performed via a form @xcite package ` mincer ` @xcite .
this kind of manipulations does not introduce spurious ir divergences .
moreover , since we are only interesting in uv counter - terms it is possible to work from the very beginning within the unbroken phase of the sm , in which all the fields are massless and the higgs doublet @xmath3 does not have a vacuum expectation value .
the second approach to irr , which was used in calculation of higgs potential parameter beta - functions , is the introduction of an auxiliary mass parameter @xmath4 in every propagator via iterative application of the following formula @xcite @xmath5 where @xmath6 and @xmath7 are linear combinations of internal and external momenta correspondingly .
it is clear that if one applies this kind of decomposition a sufficient number of times the last term can be neglected in the calculation of uv divergences ( after subtraction of subdivergences ) .
it turns out that for the scalar four - point green functions considered only the first term in eq .
is necessary .
consequently , we are left with massive vacuum integrals , which can be calculated by either public ` matad ` package @xcite or private ` babma ` code written by velizhanin . in such an approach
no spurious ir divergences appear so it can be used in the situations when a naive application of the first variant of irr fails . however , the price to pay for this advantage is the necessity of explicit calculation of diagrams with counter - term insertions .
this is due to the fact that one needs to introduce counter - terms for divergences contributing to the auxiliary masses for vector and scalar bosons . for further details
see
@xcite .
it is worth mentioning that we can still exploit the symmetries of the unbroken sm .
for example , all the components of the higgs boson doublet should have the same auxiliary mass counter - term .
the same is true for the su(2 ) gauge bosons .
moreover , as it is stated in ref .
@xcite , the auxiliary mass appearing in the numerator in rhs of eq . can be safely neglected since it can only contribute to the unphysical mass counter - terms .
in addition , one can also skip feynman diagrams with vacuum subdiagrams . in spite of the fact that these subdiagrams are non - zero when the auxiliary mass is introduced , they still can be neglected due to the same reasons . as it was noticed above
there are a lot of diagrams which should be generated and evaluated in order to find three - loop contributions to the considered quantities . in our calculation
we made use of two popular codes , ` feynarts`@xcite and ` diana ` @xcite/`qgraf ` @xcite , which generate necessary diagrams and produce corresponding analytic expressions .
both packages require a model file prepared in a special format to do their job . since we were wanting to simplify the calculation as much as possible we prepared a model file for the unbroken sm in the background field gauge ( bfg ) @xcite .
this kind of gauge allows one to find gauge coupling beta - functions solely from uv - divergences of corresponding gauge field propagators .
we used a very fast ` lanhep ` code @xcite by a. semenov to derive all the sm vertices from the considered sm lagrangian ( see .
@xcite ) in ` feynarts ` notation .
it is worth mentioning that the karlsruhe group made use of alternative package ` feynrules ` @xcite to solve similar problem .
latter on a simple script was written to convert the ` feynarts ` model file to that of ` diana ` .
a typical problem which arises in this kind of calculation is internal momenta identification . in order to evaluate a feynman diagram one needs to use the momenta notation of the chosen code ( ` mincer`/`matad`/`bamba ` ) . in the case of gauge and
yukawa couplings the problem was solved with the help of routine , ` mapmincer ` , which associate with every ` feynarts ` topology the corresponding ` mincer ` topology and distribute ` mincer ` momenta accordingly .
it is worth mentioning that ` mapmincer ` can deal with three - point vertices with one external leg carrying zero momentum ( i.e. , when internal lines have dots ) .
the corresponding routine for ` diana`/`qgraf ` , ` mapdiana ` , performs similar task , but maps every generated topology to fully massive vacuum integrals , which appear after the mentioned``exact '' decomposition of internal propagators . given a model file for ` feynarts`/`diana ` together with the correct mapping of internal momenta to the notation of the utilized three - loop codes
it is tenuous but straightforward to generate and calculate one- , two- , and three - loop contributions to the 1pi green - functions presented in fig .
[ fig:1 ] .
for the su(3 ) color algebra the ` color ` @xcite package was used .
the green functions , considered in the beta - function calculations , together with the corresponding renormalization constants .
left- and right - handed fermions , denoted by @xmath8 , renormalise differently in the sm .
the same is true for background @xmath9 and quantum @xmath10 gauge fields in bfg employed .
the su(2 ) invariance implies the presented equalities , which serve as an additional cross - check . ] from two - point green functions we extract the corresponding wave function renormalization constants in the @xmath1-scheme .
it is worth pointing that the self - energies of both background and quantum gauge fields are considered .
the renormalization constants of the former , @xmath11 , are directly connected to that of gauge couplings and the renormalization of the latter corresponds to the z - factors of three gauge - fixing parameters , @xmath12 , @xmath13 , and @xmath14 , which we keep non - zero during the whole stage of calculation .
the independence of the final results on these parameters serves as an important cross - check of the obtained expressions .
the three - point 1pi functions for the yukawa vertices with neutral higgs bosons @xmath15 and @xmath16 are also presented in fig .
[ fig:1 ] .
it is interesting to note that our calculation explicitly demonstrated that the semi - naive treatment of @xmath0 discussed below is only applicable if one takes into account the gauge anomaly cancellation condition @xmath17 . here
@xmath18 denotes the number of colors .
the beta - functions of the sm parameters are extracted from the corresponding renormalization constants . for the gauge and yukawa couplings we used the following relations @xmath19 where @xmath20 , @xmath21 are su(2 ) and u(1 ) gauge couplings correspondingly
, @xmath22 denotes the strong coupling , and @xmath23 is the yukawa coupling associated with the ( right - handed ) fermion @xmath24 .
both neutral components , @xmath25 , gave the same result , and , thus , provided us with a confirmation of the validity of the obtained expressions .
for the higgs self - coupling it is impossible to use ` mincer ` naively , so feynman diagrams for the four - point functions ( see fig .
[ fig:1 ] ) converted to the fully massive vacuum integrals were calculated with the help of private code ` bamba ` ( by velizhanin , who considered the @xmath26 vertex ) and public package ` matad ` ( by bednyakov and pikelner who considered the fully symmetric @xmath27 vertex ) .
these two independent evaluations lead to the same final expression for the vertex renormalization constants , i.e. , confirming the su(2 ) relation @xmath28 .
a comment is in order on the higgs mass parameter @xmath29 .
it is possible to obtain the corresponding anomalous dimension by considering the renormalization of the @xmath30 composite operator within the unbroken(=massless ) sm ( see , e.g. , @xcite ) .
this kind of result can be found at almost no cost from the calculation of @xmath26 vertex .
it is sufficient to select the diagrams , which have @xmath31 and @xmath32 external fields connected to the same four - point vertex ( see fig .
[ fig:2 ] ) , and weight different contributions with a correct combinatorial factor .
this restricted set of diagrams give rise to the @xmath33}$ ] renormalization constant .
the restricted set of feynman diagrams which is used to obtain the renormalization constant @xmath33}$ ] .
the diagrams of the first type have to be multiplied by @xmath34 . ] at the end of the day , the renormalization constants for @xmath35 and @xmath29 are obtained with the help of the following relations @xmath36}}{z_{h } } \label{eq : lambda_mass_rc}\ ] ] from renormalization constants @xmath37 for the dimensionless sm parameters , @xmath38 it is straightforward to obtain the corresponding beta - functions @xmath39 here @xmath40 is the @xmath1 renormalization scale , @xmath41 is the parameter of dimensional regularization , and @xmath42 denotes the coefficient for the single pole in @xmath43 in the expression for @xmath37 , which enters the relation between the bare parameters @xmath44 and the renormalized ones .
for the anomalous dimension of the higgs mass parameter @xmath29 one can use similar formulae @xmath45 the full analytical results for the considered quantities can be found in ancillary files of the arxiv versions of our papers .
the intermediate expressions , e.g. , all the renormalization constants , can also be obtained , if needed , from the authors .
it is worth mentioning that the beta - functions of all the fundamental sm parameters are free from gauge parameter dependence , which is a crucial test for our calculation .
in addition , the anomalous dimension of the higgs doublet can be of some interest , so we also include the corresponding expression in the ancillary files of ref . @xcite . before going to the results we would like to stop on the problem related to the definition of the @xmath0 matrix withing the dimensional regularization .
it is known from the literature ( see , e.g. , ref .
@xcite and recent explicit calculation @xcite ) that the traces with an odd number of @xmath0 appearing for the first time in the three - loop diagrams require special treatment .
we closely follow the semi - naive approach presented in refs .
@xcite .
first of all , we anticommute @xmath0 to the rightmost position in a fermion chain and use @xmath46 . in the `` even '' traces all @xmath0 are contracted with each other , so the corresponding expressions can be treated naively in dimensional regularization . in `` odd '' traces
we are left with only one @xmath0 .
these traces are evaluated as in four dimensions and produce totally antisymmetric tensors via the relation @xmath47 with @xmath48 . due to the fact that we use both the @xmath0 anticommutativity and the four - dimensional relation , the cyclicity of the trace
should be relinquished @xcite .
one has to choose a certain `` reading prescription '' for an `` odd '' dirac trace , i.e. , start reading a closed fermion chain from a proper place , in order to achieve the correct final result .
however , in our calculations the problem of @xmath0 positioning within the `` odd '' traces is solved implicitly since the diagram generation routines split the traces for us at certain points .
it should be stressed that a non - trivial contribution to the considered quantities can only appear when there are two `` odd '' traces in a diagram , since two @xmath43-tensors should be `` contracted '' with each outher to produce a non - zero effect .
we can distinguish two situations .
two `` odd '' traces in our three - loop calculations can appear either as two internal fermion loops in a diagram with external bosons or , if one considers green functions with two external fermions , one dirac trace from internal fermion loop can be combined with the trace appearing after contraction with an appropriate projector .
it is easy to convince oneself that in the case of two internal traces only triangle subloops with three external vector particles can potentially produce `` eps''-tensors .
however , it is known that in the sm these kind of traces cancel upon summation over all the fermion species due to the absence of gauge anomalies @xcite .
the same is true if one considers fermion self - energies up to three - loops .
it turns our that the non - trivial contribution due to the contraction of two @xmath49-tensor appears from the yukawa vertex ( see fig .
[ fig:3 ] ) .
this kind of diagrams was also considered in @xcite . the diagrams for the fermion - fermion - higgs vertex ( @xmath50 ) , which produce a non - zero contribution due to contraction of two @xmath49-tensors appearing from two dirac traces with @xmath0 .
the ambiguity in positioning of @xmath0 within the traces does not affect the uv - divergent part . ] a final remark about @xmath0 is again related to the fact that the ambiguity in the choice of `` reading '' point in the `` odd '' traces can only spoil our result for the three - loop uv - divergence in the case of the mentioned triangle subgraphs , for which one can set all the `` odd '' traces to zero from the very beginning . for the only non - trivial case all the `` reading '' points are equivalent since the difference gives the contribution @xmath51 which we neglect here . the scale dependence of the relative contributions of different terms in the three - loop corrections to the beta - functions of the ( squared ) gauge and yukawa couplings
. only the most sizable corrections are shown .
the boundary conditions at scale @xmath52 gev are given in . ]
it is obvious , that the resulting expression for the three - loop contributions to the considered renormalization group quantities are too lengthy to be presented here . for the demonstration purposes only
, we would like to show how different relative contributions to the three - loop corrections to the sm coupling beta - functions evolve with renormalization scale @xmath40 ( see fig .
[ fig:4 ] ) . for a reasonable choice of the sm initial running parameters at the scale @xmath52 gev @xmath53 we solve the corresponding rge numerically up @xmath54 gev . with the help of these solutions one can evaluate the three - loop beta - functions at any scale and
find how different terms contribute to to the total value of @xmath55 . from fig .
[ fig:4 ] one can see that the dominant contributions is due to the strong and top yukawa couplings .
however , with the increase of the renormalization scale the su(2 ) coupling can also play a role .
it is fair to say that the most interesting sm beta - function is that of the higgs self - coupling , since from the evolution of the latter one can deduce the so - called `` vacuum stability bound '' ( see recent papers @xcite and references therein ) . in fig .
[ fig:5 ] one can find the same evolution of the relative contributions to @xmath56 .
in addition , the slice @xmath57 of the phase space @xmath58 is presented together with trajectories , obtained with the help of one- , two- , and three - loop evolution from 100 gev to @xmath54 gev .
one can see that for the given set of initial conditions the running @xmath59 is driven to zero faster when one - loop rges are employed instead of two- or three - loop ones .
the difference between two- and three - loop evolution is not sizable , but the fact that with three - loop corrections the scale at which @xmath35 hits zero slightly higher , favours the latter . [
cols="^,^ " , ] to conclude , we obtained the three - loop beta - functions for the sm parameters .
the results for gauge and higgs - potential couplings coincide with that obtained by two karlsruhe groups .
the beta - functions for yukawa couplings were obtained for the first time .
moreover , we established a framework that allow us to carry out a similar calculation within an `` arbitrary '' qft model .
however , it should be stressed that in a self - consistent rge analysis of the chosen model the obtained rges should be accompanied by the so - called threshold ( matching ) corrections ( see , e.g.,refs .
@xcite for the recent sm results ) .
avb is grateful to the organizers and conveniers of the acat2013 workshop for the invitation and for arranging such a nice event .
in addition , we would like to thank m. kalmykov for drawing our attention to the problem and stimulating discussions . the work is supported in part by rfbr grants 11 - 02 - 01177-a , 12 - 02 - 00412-a , and by jinr grant no .
13 - 302 - 03 .
|
in order to understand the fundamental mechanism for high-@xmath0 superconductors ( htsc ) , many experiments have attempted to find out the properties of the superconductors pair wave function . notwithstanding this effort
the nature of the orbital order parameter symmetry ( or energy gap ) has not yet been settled , despite increasing evidences toward a major @xmath1 symmetry as it explains a number of different experimental results@xcite .
furthermore the d - wave state has also theoretical support from weak and strong - coupling approaches@xcite and from monte carlo and numerical studies on a two dimensional hubbard model@xcite . on the other hand several experiments
have also suggested the presence of real or complex mixture of order parameters .
for instance , c - axis josephson tunneling data between twinned ybco and a s - wave superconductor were interpreted by a condensate containing a mixture of @xmath1 and @xmath7 order parameter@xcite .
this type of order parameter symmetry was argued to be a consequence of the orthorhombicity of the ybco crystal structure@xcite .
recently , klemm at al@xcite have studied the symmetry operation of the crystal groups relevant to hg1001 , ybco and bscco to analyze the possibility of different order parameter symmetry combinations for these htsc and their results agree with the c - axis josephson tunneling experiments .
furthermore the possibility of complex mixtures of different symmetries has also been suggested@xcite as it would induce time reversal symmetry breaking in connection with magnetic defects or small fractions of a flux quantum @xmath8 observed in ybco powders .
similarly a complex admixture of @xmath1 and @xmath7 symmetry was also proposed@xcite to explain the ab - oriented ybco / i / cu tunnel junctions data@xcite .
they proposed an interesting and very simple physical picture : andreev scattering near the ybco surface causes strong pair dissociation .
the resulting quasiparticles may then be paired again also by a subdominant @xmath7-channel that is less sensitive to surface pair breaking then the dominant @xmath1-channel .
calculations minimizing the free energy show@xcite that such complex mixture can coexist at low temperatures . using a tight binding model in a mean field treatment honerkamp et al@xcite have arrived at a similar conclusion that a surface phase transition towards a time - reversal symmetry breaking surface state carrying an @xmath9i@xmath1 state may appear ( below @xmath0 ) .
recently@xcite a complex mixture of @xmath10 was also proposed in order to explain a phase transition characterized by magnetic field plateaus or kinks in the thermal conductivity@xcite of @xmath11 compounds . based on the above discussion
, we study in this paper the competition of the @xmath1 and others subdominants complex symmetries of the gap function .
we develop in these calculations a novel approach based on a change in the position of the attractive potential @xmath12 from its usual nearest neighbor position in the extended hubbard hamiltonian .
as it is well known , the on - site coulomb correlations may explain the antiferromagnetism of the low doping regime , the large magnetic fluctuations and the semiconductor like properties of the metallic phase@xcite .
furthermore the presence of a small attractive interaction leads to phase separation and , on a square lattice , yields a superconducting phase with a @xmath1 symmetry@xcite .
thus , the hubbard hamiltonian with such phenomenological attraction is a natural candidate to deal with the htsc .
the idea to change the position of the attractive potential in the extended hubbard hamiltonian was originally motivated by the experimental fact that different compounds have different @xmath13 phase diagram and the optimal doping values may vary according the compound .
this fact can be physically interpreted due to a possible change in the range of the attractive interaction that leads to pair formation @xcite ; for compounds with short range correlations the carrier density must be larger than those compounds with a longer ranged interaction in order to undergo a superconducting phase transition .
the same type of physical argument can be used to interpret the measured different values of the coherence length @xmath14 for different family of compounds since a short range interaction requires larger densities than a long range one in order to produce the coherence motion that leads to superconductivity .
furthermore , the typical parameter which characterizes the strength of the attractive interaction@xcite , @xmath15 , when estimated for most of htsc , varies between 10 - 3000 and such high values indicate that the size of the bound states ( or cooper pairs ) are indeed related to the size of the minimum of the attractive potential .
another interesting aspect from the theoretical point of view is that this parametric change of the attractive potential can be related with the order parameter expansion introduced by the spin - fluctuation theory@xcite . in this approach
it has been proposed a @xmath1-wave gap of the form @xmath16 where @xmath17 is expanded in powers of @xmath18 . taking a close look at this expansion
, one can easily verify that it contains terms like @xmath19 which are proportional to @xmath20 and what can be seen as a type of d - wave gap symmetry that arises from a potential like @xmath21 . by the same token
, we can find terms proportional to @xmath22 which can be originated by a potential @xmath23 and so on .
there will also be crossed terms proportional to @xmath24 that may be associated to a next - nearest potential .
thus , we see that the gap expression proposed by the spin - fluctuation theory contains several terms and some may be associated to a potential of the form @xmath25 , where @xmath26 is a parameter related to the position of the attractive potential ( @xmath27 leads to the usual nearest - neighbor expression ) .
consequently we have studied in this paper the phase diagram of complex admixtures of a d - wave like order parameter with some minor complex component with this parameterized change of the potential position , that is , with different values of @xmath5 .
these calculations are made with the extended hubbard model on a square lattice of side @xmath28 , which describes a tight binding model with the strong correlations taken into account and is defined by @xmath29 where @xmath30 is the hopping integral between sites @xmath31 and @xmath32 and are estimated by comparison with either band structure calculations or fermi surface measurements , u is the on - site correlated repulsion and v is an attractive phenomenological interaction which in principle can be due to spin fluctuations , as mentioned above or any other type of mechanism . in order to obtain quantitative results ,
we have used the tight - binding dispersion relation derived from fermi surface measurements by norman et al@xcite for @xmath11 but this choice is not a fundamental requirement , another studies have used different dispersion@xcite with small quantitative changes .
we used a bsc mean - field approximation@xcite to study the superconducting phase associated with the hamiltonian of eq .
( [ hamiltonian ] ) described above . in this model the superconducting state is characterized by a gap order parameter , which at a finite temperature t satisfies the bcs self - consistent equation @xmath33 where @xmath34 is the fourier transform of the potential of eq .
( [ hamiltonian ] ) and @xmath35 where @xmath36 is the dispersion relation taken from ref.@xcite and @xmath37 is the chemical potential .
following along the lines of ref.@xcite , one can easily show that @xmath34 may be written in a `` separable '' form , @xmath38 which leads to the usual d and s - wave .
this procedure can easily be generalized for the case of @xmath39 discussed above and will be developed below in order to study the superconducting transition with respect to the position of the attractive potential .
for a two - component order parameter with a separable form , the corresponding gap equations may be written as@xcite @xmath40 here , since we want to study the competition between the @xmath1 and others symmetries , we have always taken @xmath41 , @xmath42 and @xmath43 and the complex component can either be @xmath44 or @xmath45 with @xmath46 or @xmath47 and the gap functions are @xmath48 or @xmath49 respectively . now , separating the real and imaginary parts and combining eq.[gapf ] and eq.[gapv ] we obtain the following two gap equations @xmath50 these two equations has to be solved with the density of carriers @xmath51 equation which is given by@xcite @xmath52 thus we solve self - consistently the above 3 equations ( eq.[gapff ] and eq.[dens ] ) in order to study the phase diagram of a complex admixture of the order parameter .
these equations are general in the sense that they can be applied to any type of two component symmetry mixtures .
the results will be discussed below .
= 3.6truein we start with the analysis for the complex mixing with the two most relevant@xcite symmetries , the @xmath1 and the @xmath3 which we write as @xmath53 .
we have solved the above self consistent equations with fixed @xmath54 and @xmath55 , which these values were chosen in order to obtain reasonable values for the critical temperatures @xmath0 in the range of parameters studied . in this way
we obtain a maximum @xmath56k for b=1 and by the dominant @xmath1 symmetry near the optimum doping . in fig.1
we show the results for the values of @xmath57 in order to compare with the experimental results of allgeier et al@xcite on @xmath58 ( bi2212 ) .
notice that their maximum @xmath0 occur at @xmath59 while our calculations yield a maximum @xmath0 at @xmath6 and this discrepancy is probably due the dispersion relation that we used since a similar bcs mean field calculation using the hubbard hamiltonian ( for b=1 ) gives a maximum which agrees with the experimental result@xcite .
we should point out that the comparison with a htsc experimental phase diagram , as in fig.1 , has only meaning if the material has a bcs type behavior , that is , when the appearance of the superconducting gap and phase coherence occur simultaneously at the onset of superconductivity . on the other hand , angle - resolved photoemission spectroscopy ( arpes ) on underdoped bi2212
has revealed the presence of a `` pseudogap '' above @xmath0@xcite which has also been confirmed by recently fermi surface arpes measurements@xcite .
this unusual behavior may be an indication of non - bcs behavior which @xmath0 is controlled by the doping level , at underdoped regime , rather then by phase coherence with a single particle gap @xmath60@xcite .
nevertheless this is not a completely settled scenario which only more experiments will give the correct interpretation specially because the pseudogap has not yet been seen in several cuprate superconductors@xcite .
= 3.6truein in the fig.2 we show the results for the zero temperature gap amplitudes @xmath61 ( continuous line ) and @xmath62 ( dashed line ) as function of the density @xmath63 for three values of the parameter @xmath5 , namely @xmath64 and @xmath65 . where there is superconductivity region , the values of @xmath66 have their maxima at zero temperature and vanishes at @xmath0 in agreement with the bcs method .
in fact , the self - consistent calculations yields the temperature , the density @xmath63 and the value of the gap as one solves eqs . 5 and 6 simultaneously . in fig.3
we show how the gap changes with the temperatue .
the zero temperature gap value is proportional to @xmath0 and the constant of proportionality differs slightly from one channel to the other .
we get @xmath67 in the optimally region .
= 3.6truein there are very interesting differences from the @xmath7 and @xmath68 channels which arises as we vary the parameter @xmath5 . for @xmath69 , the values of @xmath61
are basically suppressed while for @xmath70 its its strength is decreased by about half of its value for @xmath27 ( which agrees with the calculations of ref.@xcite ) and does not move appreciably from the optimum doping value of hole content @xmath71 ( @xmath63 is the fraction of holes per cu atom in the @xmath72 sheet ) .
this result is a little higher then the experimental result@xcite which yields @xmath73 for the bi optimum doping value .
the maximum @xmath0 is related with the envelope of the curve shown in fig.2 . on the other hand , for the extended @xmath7 , @xmath62 ,
the change in @xmath5 causes the value of the corresponding optimum doping density to change continuously ; it is at @xmath74 for @xmath69 and and about @xmath73 for @xmath70 as it is seen in figs.2a , 2b and 2c . at @xmath27 ( fig.2a ) ,
the s - wave has the optimum doping at @xmath75 , for @xmath69 the maximum @xmath0 occurs at @xmath76 and it dominates over @xmath77 which is almost entirely suppressed . for @xmath70
the competition between the two channels becomes very interesting ; the maximum of @xmath78 has moved to @xmath79 while the maximum of @xmath77 remains fixed at @xmath80 which gives rise to a completely novel _ doubled crossover phenomena _ ; up to @xmath81 the self - consistent equations yields pure extended s - wave solutions . between @xmath82
the mixing of order parameters starts at low temperatures but with the s - channel dominating over the d - channel as its critical temperature @xmath0 is larger as shown in fig.2c and at fig.3a for @xmath83 . at @xmath84
a crossover arises and the d - channel solutions dominates over the s - channel which becomes the minor component as shown in fig.2c and fig.3b
. there is always mixing at low temperatures ( the continuous curve of the minor s - wave never vanishes completely ) but the d - channel is the only one at @xmath0 . at @xmath85
there is a new crossover and the dominant solution changes again as the s - wave becomes the dominant solution and the only one at @xmath0 and the d - wave appears again at low temperatures as the minor components as displayed in fig.2c and fig.3c .
notice that the appearance of this subdominant component is exactly as the mechanism proposed to explain the ab - oriented ybco / i / cu tunnel junction data@xcite discussed in the introduction .
= 3.6truein next we have studied the complex gap function of symmetry @xmath86 .
the results are quite similar to those above ; for @xmath27 the d - state dominates and has its optimum doping at @xmath87 as shown in fig.4a . for @xmath69 the d - channel is again suppressed and the @xmath4-state is the only one and its maximum doping is also at @xmath88 as displayed in fig.3b . for @xmath70 there is again the very interesting _ doubled crossover _ between these two competing symmetries . for @xmath89
there is only s - wave instability but at @xmath90 a small d - wave channel develops at low temperatures . between @xmath84 and @xmath91
we find only pure @xmath1 solutions as opposed to the case shown above in fig.2c where a low temperature mixing was always present .
for @xmath92 the @xmath4-channel starts to develop , initially as a minor component ( see fig.5b ) and above @xmath93 a new crossover appears as it becomes the dominant channel and the only channel at @xmath0 .
thus we verified that the states @xmath94 has also the _ doubled crossover _ and the dominant regions are similar to those for @xmath53 however , there is a new feature with this type of admixture : we can see from fig.4c that , for @xmath70 , one kind of gap symmetry influences the other .
one can easily see the abrupt change in @xmath49 near @xmath79 and the change in slope for both channels near @xmath85 .
the new phenomenon , that is , the slightly suppression on @xmath95 of the dominant channel can also be seen on figs.5a , 5b and 5c .
this is a new feature not seen in figs.3c , and this suppression is the cause of the pure d - wave solutions at the optimal region which did not occur for previous studied extended s - wave .
this interesting feature may be similar to what happens in andreev scattering near surfaces as discussed in the introduction .
= 3.6truein
lastly we have also studied the complex order parameter mixing @xmath10 for the three values of b , however we have not found even a single crossover . for @xmath27 ,
the @xmath1 channel dominates and has also its optimum value at @xmath87 and the @xmath2 appears only at low temperatures . for @xmath69 and @xmath65
the @xmath2 dominates completely around the same optimal region and the other channel gives only a very small component at low temperatures .
thus we do not observe any crossover for @xmath10 complex mixture , simply the interchange of dominant symmetry as we change @xmath5 .
the complex mixtures in superconducting order parameter were studied here as function of hole density in the context of the bcs mean - field calculations for the extended hubbard model .
we have revealed very interesting properties that may be relevant for htsc ; if the interaction is sufficiently short ranged that can be mapped by a nearest neighbor potential , at low doping , the system is described by either _ pure _ @xmath1 or @xmath3 order parameter . if the range of the potential is larger the system changes its order parameter symmetry ( for @xmath69 ) or may admit a mixed state with both @xmath7 and @xmath68 components present ( for @xmath70 ) . in this last case
the dominant component depends strongly on the density or doping level @xmath63 and may change abruptly the dominant symmetry at @xmath0 . since in most
htsc it is difficult to determine @xmath63 accurately@xcite and as our calculations demonstrate , a small change in @xmath63 may be accompanied by a crossover of the dominant component .
if we assume an expansion with @xmath96 like that of the spin fluctuation theory of ref.@xcite , then one must look at their weight to see which one is more important and concomitantly , which symmetry of the order parameter is the dominant one .
this present work provides a hint to understand why different order parameter symmetries appear in different experiments performed on the same type of compounds .
we acknowledge partial financial support from the brazilian agencies capes and cnpq and the hospitality of the cnrs / crtbt at grenoble , france .
[email protected] d. l. cox and m. b. maple , phys . today , 32 , fev .
d. j. van harlingen , rev .
phys , * 67 * , 515 ( 1995 ) .
d. j. scalapino , phys . rept . * 250 * , 329 ( 1995 ) .
e. dagotto , j. riera , y. c. chen , a. moreo , a. nazarenko , f. alcaraz and f. ortoloni , phys . rev . *
b49 * , 3548 ( 1994 ) .
sun , d. a. gajewski , m. b. maple and r. c. dynes , phys .
lett . * 72 * , 2267 , ( 1994 ) .
k. a. kouznestsov , g.g .
sun , b. chen , a. s. katz s. r. bahcall , john clarke , r. c. dynes , d. a. gajewski , s. h. han , m. b. maple , h. giapintzakis , j. t. kim and d. m. ginsberg , phys .
79 * , 3050 , ( 1997 ) .
m. b. walker , phys .
rev . * b53 * , 5835 , ( 1996 ) . c. odonovan , d. branch , j. p. carbotte and j. s. preston , phys .
* b51 * , 6588 , ( 1995 ) .
r. a. klem , c. t. rieck and k. scharnberg , cond - mat 9811303 , submitted to phys .
b. m. sigrist , d. b. bailey , and r. b. laughlin , phys .
lett . * 74 * , 3249 , ( 1995 ) .
m. fogelstrm , d. rainier , and j. a. sauls , phys .
rev . lett.*79 * , 281 ( 1997 ) .
m. covington , m. aprili , e. paraonu , l.h .
greene , f.yu , j. zhu and c. a. mirkin , phys .
rev . lett.*79 * , 277 ( 1997 ) . c. honerkamp , k. wakabyashi and m. sigrist , cond - mat 9902026 .
r. b. laughlin , phys .
* 80 * , 5188 , ( 1998 ) .
k. krishana , n. p. ong , q. li , g. d. gu , and n. koshizuka , science , * 277 * , 83 , ( 1997 ) . p. w. anderson science * 235 * 1196 ( 1987 ) .
r. micnas , j. ranninger , s. robaszkiewicz and s. tabor , phys . rev . *
b37 * 9410 ( 1988 ) .
r. micnas , j. ranninger and s. robaszkiewicz , rev .
* 62 * , 113 ( 1990 ) .
e.v.l.de mello , physica * c259 * , 109(1996 ) and chec .
j. phys.*46 * , 945 ( 1996 ) .
e.v.l . de mello and c.acha , phys .
rev * b56 * , 466 ( 1997 ) and physica * c 282 - 287 * , 1819 , ( 1997 ) .
`` mcanique quantique '' , l. landau and i. lifshitz , ( ed .
mir , moscow 1966 ) , page 190 .
p. monthoux , a. v. balatsky and d. pines , phys .
* 67 * , 3448 ( 1991 ) .
m. r. norman , m. randeria , h. ding and j. c. campuzano , phys .
* b52 * , 615 , ( 1995 ) .
angilella , r. pucci and f. siringo , phys . rev . *
b54 * , 15471 ( 1996 ) .
gennes , `` superconductivity of metals and alloys '' , w.a.benjamin , new york,1966 .
a. j. legget , rev .
phys . * 47 * , 331 , ( 1975 ) . h. gosh , europhys . lett . *
43 * , 707 ( 1998 ) . c. allgeier and j. s. schilling , physica * c168 * , 499 ( 1990 ) .
d. s. marshall , d. s. dessau , a. g. loeser , c - h park , a. y. matsura , j. n. eckstein , i bozovic , p. fournier , a. kapitulnik , w.e spicer and z. x. shen , phys .
lett . * 76 * , 4841 , ( 1996 ) .
a. g. loeser , z.x .
shen , d. s. dessau , d. s. marshall , c. h. park , p. fournier and a. kapitulnik , science * 273 * , 325 ( 1996 ) .
h.ding t. yokoya , j. c. campuzano , t. takahashi , m. randeria , m. r. norman , t. mochiko , k. kadowaki and j. giapintizakis , nature * 382 * , 51 , ( 1996 ) .
m. r. norman , h. ding , m. randeria , j. c. campuzano , t. yokoya , t. takeuchi , t. takahashi , t. mochiko , k. kadowaki , p. guptasarma and d. g. hinks , nature * 392 * , 157 , ( 1998 ) . v. j. emery and s. a. kivelson , nature * 374 * , 434 , ( 1995 ) .
p. coleman nature * 392 * , 134 , ( 1998 ) .
j. l. tallon , c. bernhard , h. shaked , r.l hitterman and j. jorgensen , phys .
b51 * , 12911 , ( 1995 ) .
|
a number of recent experiments have suggested the presence of either real or complex components in the gap symmetry of high-@xmath0 superconductors ( htsc ) . in this paper
we introduce a novel approach to study the competition of such complex order parameter mixtures by varying the position of the two - body attractive potential in a two dimensional extended hubbard hamiltonian .
we show that this procedure explain a number of experimental results and on the theoretical side , it may be related with certain htsc microscopic models like the spin fluctuation theory . following current trends we concentrate on the study of @xmath1 order parameter with a component of the type @xmath2 or a s - wave like @xmath3 and @xmath4 symmetry .
we show that the position of the optimal s - component peak changes with the position parameter @xmath5 while the d - component occurs always in the optimally region around hole content @xmath6 . these studies may be useful to interpret some experimental data and to explain why similar experiments yield different gap symmetries . 2
|
bipolar plasma outflows , loosely termed jets , are a feature common to a variety of different astrophysical objects .
jets are observed in accreting compact objects of all scales , as well as at stellar birth and death .
they seem equally able to form out of massive accretion disks as well as from quasi - spherically collapsing plasma , which suggests that their creation is a basic byproduct of some routinely occurring ingredients .
these necessary inputs seem to be rotation , collapsing or infalling / accreting plasma , and magnetic fields .
jets likely are formed at least in part from a necessity to shed the system of excess angular momentum ( e.g. , * ? ? ?
the exact details of their formation , as well as the nature of the coupling between the infalling / collapsing plasma and the outflow is still an active area of controversy and research on many fronts .
however , the fact that so many types of systems share this mechanism extends hope that clues gained from one class of objects can be applied to the others in the search for a unified picture .
the advantage of looking for answers in x - ray binary ( xrb ) jets is twofold .
first , their formation seems to be a recurrent phenomenon on time scales we can observe repeatedly over the course of our own human lifetimes .
for instance in the galactic xrb gx 339@xmath04 and cyg x-1 , accretion without observable jets proceeds into accretion with discernible jets and back again every few years @xcite .
thus the same object can provide tests of theories developed based on its earlier activity .
secondly , the existence of quality simultaneous broadband data for many sources linking the `` traditional '' jet outflow bands ( radio and likely also infrared , or ir ) and inflow bands ( x - ray and optical ) gives us a way of studying the relationship and interplay between the two sides of accretion .
we use the word `` traditional '' to refer to the picture which existed up until recently that the jets and accretion flow radiate in different energy bands and can generally be studied as distinct phenomena .
this picture has rapidly evolved in the last few years , beginning with the discovery of an unexpectedly intimate connection between the low and high frequency wavebands in the hard state ( hs ) of xrbs ( see , e.g. , @xcite , for a detailed definition ) . using radio observations in conjunction with 1.3 - 12.2kev data from the all sky monitor ( asm ) on - board the rossi x - ray timing explorer ( rxte ; * ? ? ?
* ) , @xcite first discovered a near - linear correlation between the radio and x - ray emission of gx 339@xmath04 .
these studies were followed by joint radio observations and pointed x - ray observations with rxte @xcite .
this intensive simultaneous monitoring showed that , in fact , the correlation is non - linear , with @xmath1 @xcite .
additionally , this non - linear correlation holds over orders of magnitude changes in the source luminosity with time .
in fact , even after fading into quiescence , gx 339@xmath04 seemingly returns to the same correlation in subsequent outbursts @xcite , although it is not clear whether the correlation always maintains the same normalization in each instance @xcite .
despite some possible variation in its normalization , the correlation between radio and x - ray fluxes appears to be universal to all hs xrbs with comparable broadband data @xcite .
the @xmath2 slope of the @xmath3-@xmath4 correlation follows directly from analytic predictions of scaling synchrotron jet models @xcite , and can be generalized for other x - ray emission processes in terms of their dependence on the accretion rate @xcite .
the correlation s normalization in the @xmath5 plane also depends on the central engine mass .
when this scaling is accounted for , unbeamed , extra - galactic , supermassive black holes sources agree remarkably well with the same radio / x - ray correlation found in galactic , stellar - mass sources @xcite . the underlying physics governing the connection between the radio / ir and x - ray bands
must therefore be fundamental to accreting sources regardless of mass .
in @xcite we showed that jet synchrotron emission could account for the broad continuum features of the simultaneous radio through x - ray hard state observation of the galactic xrb , xte j1118 + 480 . in a later work
, we showed that this same model could also explain the broad spectral features of thirteen simultaneous or quasi - simultaneous radio / x - ray observations of gx 339@xmath04 , and that the radio / x - ray correlation was naturally produced by only changing the power input into the jet @xcite .
these models , however , did not attempt to address the fine features present in the x - ray spectrum which are the hallmarks of an optically thick accretion disk : in particular fluorescent iron lines and a characteristic flattening above 10kev attributed to compton reflection @xcite .
the reflection is generally assumed to result from a hard x - ray continuum originating above the cooler accretion disk . we began to explore the interaction of the jet emission with the putative accretion disk in @xcite . in the hard state ( hs ) , the spectrum is hard enough and the observed reflection signature weak enough to be problematic for coronal models where there is significant coverage of the disk by hard x - ray - emitting material ( e.g. , * ? ? ? * and references therein ) .
various mechanisms have been proposed to decrease both the cooling of the hard x - ray - emitting material by soft disk photons and the fraction of reflected x - rays .
these mechanisms include patchy coronae @xcite , high disk ionization @xcite and beaming of the coronae away from the disk with mildly relativistic velocities @xcite .
this latter approach seems extremely close in principle to the characteristics of the base of a jet .
compact accretion disk coronae are theoretical concepts based on observations that suggest the presence of hot electrons radiating near the inner parts of a thin accretion disk .
the existence of hot electrons in the same region is also empirically required by the jets we image in the radio wavebands .
the high brightness temperatures and occasionally measured linear polarization ( e.g. , * ? ? ?
* ) argue strongly that nonthermal synchrotron is the dominant radio - emitting process . because the base of the jet is significantly more compact than the outer region which dominates the cm - bands , conservation arguments imply an even hotter , denser medium near the accretion disk .
we therefore wish to determine whether the base of the jets can `` subsume '' the role of the corona .
this idea was proposed already back in @xcite , and by subsume we mean provide the spectral characteristics ( both direct and reflected ) which are traditionally attributed to a compact comptonizing corona .
thus we will compare coronal fits using a standard comptonizing model ( see below ) to fits made by our jet model .
regardless of the ultimate relationship determined to exist between the base of the jets and the corona , this is a critical step in the road to understanding .
the results of this study will provide valuable clues as to the nature of the corona , both in terms of geometrical as well as internal characteristics which differ from the base of the jets .
as an additional note , the concept of cyclo - synchrotron photons feeding inverse compton processes in a corona has existed for over 20 years @xcite , inspired by multiwavelength observations showing similar optical / x - ray behavior @xcite .
it has recently been reconsidered for several sources @xcite in the context of magnetic flares in the corona . while a valuable step towards understanding the role of magnetic fields , these models do not address the radio emission from the jets with the same model .
the remainder of this paper is structured as follows . in section [ sec : model ]
we very briefly summarize the model , with a full description included in appendix [ app ] . in section [ sec : data ]
we present our data analysis and fitting techniques , the results of which will be presented in section [ sec : results ] .
we discuss our conclusions in section [ sec : discuss ] .
our model for the jets in hs xrbs is based upon five main assumptions : 1 ) the total power in the jet scales with the total accretion power at the inner edge of the disk , @xmath6 , 2 ) the jet is expanding freely and is only weakly accelerated via the resulting pressure gradient , 3 ) the jet contains cold protons which carry most of the kinetic energy while leptons do most of the radiating , 4 ) particles are eventually accelerated into a power - law distribution , and 5 ) this power - law is maintained along the length of the jet thereafter .
these assumptions are motivated by observations which are discussed further in appendix [ app ] , where we have also included a detailed description of the model and its development . in this section
we only summarize the key parameters and assumptions of the jet model .
beyond a small nozzle region of constant radius , which establishes the base of the jet , the jet expands sideways with the initial proper sound speed for a relativistic electron / proton plasma , @xmath7 .
the plasma is weakly accelerated along the resulting longitudinal pressure gradient , which allows an exact solution for the velocity profile via the euler equation ( see , e.g. , * ? ? ?
this results in a roughly logarithmic dependence of velocity upon distance @xmath8 . after a period of more rapid acceleration immediately beyond the nozzle ,
the velocity gradient lessens , and the velocity saturates at lorentz factors of @xmath92 - 3 .
the size of the base of the jet , @xmath10 , is a free parameter and once fixed determines the radius as a function of distance along the jet , @xmath11 .
the the model is most sensitive to the fitted parameter @xmath12 , which acts as a normalization .
it dictates the power initially given to the particles and magnetic field at the base of the jet , and is expressed in terms of a fraction of the eddington luminosity @xmath13 erg s@xmath14 .
the total power input at the base of the jets is in fact approximately an order of magnitude larger than @xmath15 , due to the requirement of distributed acceleration along the jets ( see appendix ) .
once @xmath12 is specified and conservation is assumed , the macroscopic physical parameters along the jet are determined .
we assume that the jet power is evenly shared between the internal and external pressures , and that the radiating particles are close to equipartition with the magnetic field .
the radiating particles enter the base of the jet where the bulk velocities are lowest , with a quasi - thermal distribution .
around @xmath16 @xmath17 , a significant fraction of the particles are accelerated into a power - law tail whose maximum energy is determined self - consistently with the local cooling rates .
the particles in the jet radiatively cool via adiabatic expansion , the synchrotron process , and inverse compton upscattering ; however , adiabatic expansion is assumed to dominate the observed effects of cooling because of distributed acceleration .
while thermal photons from the accretion disk are included as seed photons in our compton calculations , the beaming reduces their energy density compared to the rest frame synchrotron photons ( synchrotron self - compton ; ssc ) , except at the very base of the jet where they can be of the same order .
reprocessed disk radiation will contribute even less and thus its feedback on the x - ray spectrum will be negligible .
we therefore do not include this latter component in our calculations . in the case of near - equipartition of energy between the magnetic field and particles ,
inverse compton processes only will dominate the direct synchrotron emission close to the compact object where relativistic beaming is at its minimum .
as the jet plasma is only weakly beamed , some fraction of the radiation will impact upon the cooler material in the accretion disk and lead to fluorescent line emission and a reflection hump ( e.g. , * ? ? ?
* ) . in @xcite
we calculated the resulting reflection spectrum for typical jet models , and found that when synchrotron emission from the acceleration region dominates the hard x - ray spectrum , the fraction of reflected emission is very small ( @xmath18few percent ) .
in contrast , when compton processes ( predominantly ssc ) originating near the base dominate the hard x - rays , ` reflection fractions ' of @xmath19 are possible .
as discussed by @xcite , calculating the reflection spectrum for our jet models is complex , as one needs to calculate a reflected spectrum associated with each emission region along the axis of the jet .
furthermore , for our detailed jet models the disk sees a qualitatively different jet spectrum than the direct spectrum viewed by a distant observer .
this is in contrast to the simple ` moving corona ' models of , e.g. , @xcite , where both the direct spectrum and the spectrum impinging upon the disk have the same simple power law shape , and the total reflected spectrum can be calculated analytically . for the initial spectral fit studies presented here ,
a detailed reflection calculation for our jet model was deemed too computationally prohibitive . for the purposes of the study presented here
, we adopt a simplified approach . in our fits , we add a single gaussian line to our continuum model , and we allow the line energy to vary between 67 kev and the line width , @xmath20 , to vary between 01.5kev .
we then convolve the entire _ directly viewed _ spectrum with a non - relativistic reflection model derived from the green s functions of @xcite .
the amplitude of the reflected spectrum is left as a fit parameter , expressed in the usual manner as a fractional solid angle , @xmath21 , subtended by the reflector .
( @xmath22 indicates that the radiation impinging upon the reflector is equal to that directly viewed by the distant observer . )
this general approach of phenomenologically fitting a reflection spectrum is similar to that employed by many pure comptonization models ( e.g. * ? ? ?
* ; * ? ? ?
* ) , albeit here it is less likely that the directly viewed spectrum is a completely adequate proxy for the spectrum viewed by the cold disk .
we will consider more sophisticated approaches to calculating the reflection features in future works . in our previous papers ( e.g. , * ? ? ? * ; * ? ? ? * ) , we only compared the jet model to spectra that had been ` unfolded ' with ` xspec ` using simple exponentially cutoff , broken power - law models .
this is in fact the usual practice among many researchers when comparing complicated theoretical models to multiwavelength data ( e.g. , * ? ? ?
* ; * ? ? ?
unfortunately , such a procedure does not allow the determination of a statistical goodness - of - fit , and more importantly it does not allow one to compare the model to the fine features of the x - ray spectral density distribution ( sed ) . here
we address this issue by forward - folding the jet model through the detector response matrices of the x - ray instrument , and then comparing to the data in ` detector space ' .
specifically , we have imported our jet + multi - color blackbody thermal disk ( similar to diskbb ; * ? ? ?
* ; * ? ? ?
* ) model as a subroutine for use in standard x - ray data analysis packages . the current routine works in both xspec @xcite and isis @xcite . in this work ,
we perform our fits using isis v.1.2.6 as this latter package can read in and fit lower frequency simultaneous data sets ( e.g. , radio through optical data ) as ascii files , without the need for creating dummy response matrices .
furthermore , the isis ` unfolded spectra ' ( shown in the figures throughout this work ) are independent of the assumed spectral model ( i.e. , the unfolding is done solely with the response matrix and effective area files ; see @xcite for details ) .
all shown residuals , however , are for comparisons between the properly forward - folded model and the data in detector space .
see @xcite for a discussion of further differences between ` isis ` and xspec .
we are making detailed comparisons between our jet model and detector space data for the first time .
indeed , to our knowledge , this is the first time that such a complex , multi - wavelength model has been compared properly to the x - ray data in detector space . for this paper ,
we therefore have chosen to initially leave most jet model parameters free so that we can fully explore their effect on the fits .
we can then determine which parameters are most fundamental for describing the data and which can effectively be frozen in future applications .
this is essentially in the same spirit as prior studies with comptonization models , where models such as ` eqpair ` @xcite have significantly more parameters than are used in a typical fit . for such comptonization models , even fairly basic parameters , such as thermal vs. nonthermal compactness
( i.e. , the ratio of coronal energy to coronal radius ) , can be degenerate with one another in real data fits , and one often chooses to freeze the nonthermal parameters to negligible values . for a detailed discussion of these points ,
see the review article by @xcite .
currently , our multi - component model runs significantly slower than single - component models which calculate comptonization in the corona . determining meaningful error bars for all parameters was not always possible because for some fit parameters the resulting spectrum can vary noncontinuously .
given that this is the first time that broadband data are being properly compared to our forward - folded model , such initial difficulties are not surprising .
nevertheless , the model fits presented here represent an exploration of a significant amount of the parameter space , especially given the fitting time scales involved .
we model the data from rossi x - ray timing explorer ( rxte ) observations for the hard state galactic black hole candidates ( bhcs ) gx 339@xmath04 and cygnus x-1 .
we use data from both the rxte proportional counter array ( pca ; @xcite ) and from the high energy x - ray timing experiment ( hexte ; @xcite ) .
the data have been extracted with the recent ` heasoft 5.3.1 ` software release .
compared to earlier releases of this software , the relative calibration of the pca and the hexte is now in excellent agreement ( see analysis in , e.g. , * ? ? ? * ) .
power - law fits to spectra of the crab pulsar and nebula show the pca data to have a systematic uncertainty of 0.5% , which we added in quadrature to the data .
we used the energy bands 322kev for the pca ( and grouped the data to a minimum of 30 counts / bin ) and 18200kev for the hexte ( where we added data from the two clusters and grouped the counts to a minimum signal - to - noise , after background subtraction , of 10 in each bin ) .
we use the top layer of the pca only . during part of the observations ,
some proportional counter units of the pca were switched off . in these cases
we added the spectra extracted for the different pcu combinations , appropriately weighting the response matrices .
the cygnus x-1 data consist of three of the many observations taken during our several years long rxte monitoring campaign ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) and were chosen to be representative spectra bracketing the typical hard state spectral variations of cyg x-1 .
the data from gx 339@xmath04 also have been discussed by us elsewhere @xcite , and bracket the typical variations in the overall hardness and luminosity of the gx 339@xmath04 hard state .
( they do not , however , encompass the most extreme bright and relatively soft hs data discussed by @xcite,@xcite , and @xcite . )
table [ tab : obslog ] contains a log of the observations .
cclcc cyg x-1 & 40099 - 01 - 19 & 1999 september 25 & 8080 & 13 + cyg x-1 & 60090 - 01 - 26 & 2003 february 23 & 9696 & 1/4 + cyg x-1 & 60090 - 01 - 41 & 2003
september 22 & 11600 & 13/14 + gx 339@xmath04 & 20181 - 01 - 02 & 1997 february 10 & 10528 & + gx 339@xmath04 & 40108 - 02 - 01 & 1999 april 2 & 9152 & + gx 339@xmath04 & 40108 - 02 - 03 & 1999 may 14 & 9776 & + in order to compare our jet models to the more standard comptonization models , we also fit our observations with the hybrid thermal / non - thermal comptonization model ` eqpair ` @xcite .
we slightly changed the publicly available version of this code to make the model run under xspec v.11.3.1 , and hence also under isis v.1.2.6 . to allow for a partial covering of the disk by the compton corona , we add emission from a disk using the diskpn model @xcite .
the peak temperature of this disk model is also set equal to the peak temperature of the seed photons input to the compton corona .
these seed photons are similarly assumed to have a spectral energy distribution that follows that of the diskpn model .
the corona is presumed to be comprised solely of thermal electrons , and its properties are described by two parameters : a seed electron scattering optical depth , @xmath23 ( pair production can increase the total optical depth from this value , although pair production is negligible for the fits described in this paper ) , and the coronal compactness ( i.e. , coronal energy content divided by coronal radius ) , expressed as a ratio relative to the soft , seed photon compactness , @xmath24 .
we also allow for an additive gaussian fe k@xmath25 emission line whose centroid energy is constrained to be between 6 and 7kev and whose width is constrained to be @xmath26kev . in our use of the eqpair model , the reflection component is relativistically smeared using the velocity profile of the ` diskpn ` disk model , which in turn follows the rotational velocity of a standard thin disk circularly rotating under the influence of a paczyski - wita pseudo - newtonian potential @xcite .
the reflecting medium is assumed to have solar abundances ; however , the reflecting medium can be ionized , as parameterized by the ionization parameter , @xmath27 @xcite .
table [ tab : eqpair ] gives the best fit parameters of these ` eqpair ` fits , and the fits are shown in the third column of figs .
[ gx339_all ] & [ cyg_all ] .
the reduced @xmath28 values are close to unity , indicating very good agreement between the model and the data .
the parameters found for the comptonization model are in general agreement with those reported for earlier observations .
the fits yield moderate optical electron depths of @xmath29 with coronal compactnesses of @xmath30@xmath31 , which correspond to coronal temperatures of @xmath32 - 100kev .
( note that the @xmath33 for obs i d 40108 - 02 - 03 has rather large error bars , and is partly being driven by the poor statistics in the hexte band . )
all of the coronal model fits yield reflection covering fractions @xmath19 .
none of the fits strongly require ionized reflection ; however , the reflection model used by the eqpair code is very simplified ( being based upon the pexriv code ; @xcite ) .
the fitted equivalent widths of the fe k@xmath25 lines are on the order of 150ev , and consistent with the highly smeared reflection models employed , the lines are generally found to be rather broad . as will be discussed elsewhere in more detail @xcite , we believe that part ( although not all ) of the fitted line broadness is caused by the comptonization model employed here having too simplified a description of the transition between the accretion disk emission and the comptonization continuum . finally , the changes in accretion disk temperature and flux , coupled with the changes in coronal compactness and optical depth , are in response to the fact that we have deliberately chosen observations that span a wide range of flux and spectral hardnesses .
detailed discussions of coronal model fits to cyg x-1 data ( e.g. , comparison to phenomenological broken power - law fits , comparison to timing data ) are discussed in @xcite .
their are three main points that we wish to emphasize with the fits presented here .
first , the eqpair fits describe the data well as well as any coronal models that we have explored ( although not better than simple exponentially cutoff , broken power - law models ; @xcite ) .
second , the fits do not require comparatively as large reflection fractions .
third , and related to the previous point , the coronal models themselves describe a continuum spectrum more complicated than a simple , exponentially cutoff power - law .
some fraction of the spectral hardening above 10kev , normally attributed to the ` reflection hump ' , is in fact described by curvature of the comptonization continuum . in compton
coronal models , this curvature is in part due to transiting from a spectrum strongly affected by the soft seed photons to one strongly dominated by photons that have been multiply compton scattered . as we discuss below , this continuum curvature an additional interpretation in jet models .
as mentioned above , our strategy in testing our jet models against the data was to allow the maximum number of model parameters to vary freely during the fit .
this allowed us to explore which parameters had the greatest influence on the fits .
several parameters settled on fairly similar values for both sources and all observations , which suggests that those parameters could be fixed in future applications of this model .
we will discuss this possibility explicitly below
. fits with a complicated model can easily lead to false , local minima ; therefore , we began the fitting procedure outside of isis , using unfolded data sets in order to find a set of starting parameters that would yield reduced @xmath34 .
for this paper we have chosen to explore only models with rough equipartition between the magnetic and radiating particle pressures ( @xmath35 ) .
this assumption is obviously not applicable for poynting flux dominated jets ; however , we consider the very weak acceleration required by observations to be an indication that magnetic domination ( which would generally imply stronger acceleration mechanisms ) is not likely to be very extreme .
we use the results of @xcite as a rough guide for consistency , and only consider models with reflection fractions @xmath19 for jet nozzle radii @xmath36 to be ` successful ' fits . for smaller values of the radii , slightly lower values for the reflection fraction would be expected . in all cases the jet models seemed to naturally prefer reflection fractions @xmath37 , so in practice this restriction on reflection fraction was not an issue .
[ gx339_comps ] shows a representative model for gx 339@xmath04 in order to illustrate the contribution from the various components which go into the later figures showing actual fits .
the radio through ir originates exclusively from self - absorbed synchrotron radiation beginning at @xmath38 and continuing outwards along the jet .
optically thin synchrotron emission from the accelerated power - law tail of leptons also contributes to the soft x - ray band .
the base of the jet radiates direct synchrotron photons , giving a slight hump in the optical / uv , which are then included as seed photons ( along with disk photons ) for upscattering by the emitting electrons into an ssc / ec `` hump '' in the hard x - rays .
the shape of this hump is rounded , due to the quasi - thermal particle distribution assumed in the base , and thus reduces the need for a large fraction of disk - reflected photons to contribute to the spectral break / hardening at @xmath1810 kev .
the possibility that there are two correlated continuum components in the x - ray band is supported by our ensemble of simultaneous radio / x - ray observations of cyg x-1 , whose spectra are very well - described by exponentially cut - off , broken power - laws ( @xcite ) . in these models ,
the soft x - ray spectral slope is very well - correlated with the break of the hard x - ray spectral slope , and the hard x - rays are very well - correlated with the radio @xcite .
the multicolor blackbody included in the jet model is also shown in the soft x - ray band .
the photons from this component are included in the inverse compton upscattering within the jet .
the statistical fits are shown in figs .
[ gx339_all ] & [ cyg_all ] , for gx 339@xmath04 and cyg x-1 , respectively .
each figure encompasses nine panels , with each row representing a single simultaneous radio and x - ray observation of the source .
the first column shows the entire radio through x - ray jet model + soft disk + reflection fit , and the second column focuses just on the x - ray band .
the last column shows the thermal compton corona model from eqpair for comparison .
values for all fitted parameters are given in tables 24 .
the results presented in figs .
[ gx339_all ] & [ cyg_all ] and tables [ tab : mffjeti ] , [ tab : mffjetii ] , & [ tab : eqpair ] immediately demonstrate three important results of this work . *
* jet models describe the data equally well as pure compton corona models , even when employing the broad - band and high statistics of the rxte data .
* there has been question in the literature ( e.g. , @xcite ) as to whether or not jet models can adequately describe the spectral cutoff of the hard tail .
while pure synchrotron jet models may not be able to describe the steep cutoff present in some ( but not all ) observations , jets can also successfully account for this hard tail cutoff via comptonization .
the primary differences between ` traditional ' coronal models and jet coronal models are that the coronal temperatures are higher in the latter ( @xmath39k vs. @xmath40k ) , and the disk seed photons in jet models are significantly augmented if not dominated by synchrotron seed photons . * * the jet models , similar to the compton corona models , describe a complex continuum with curvature specifically , a hardening above 10kev . * again , the hardening often predominantly ascribed to a ` reflection hump ' is being somewhat subsumed by continuum emission . in the jet model ,
the steeper slope of the soft x - rays is primarily due to synchrotron emission from the jet , rather than being influenced by the disk photons , while the hardening is due to the ssc / ec component .
reflection is present , but represents a smaller fraction of the total hard flux compared to the corona models .
note that the fits discussed here are in contrast to earlier incarnations of the jet model which attempted to describe observations solely via the synchrotron component ( e.g. , the studies of xte j1118 + 480 ; @xcite ) .
for xte j1118 + 480 , the lack of any discernible break or hardening near 10kev in the continuum spectrum was used to argue for the lack of a reflection component @xcite , which would be consistent with a jet spectrum dominated by the synchrotron component @xcite .
in contrast , the fact that all of the xrb spectra shown here harden above 10kev argue for the importance of both the synchrotron and ssc / ec components of the jet spectrum for these fits , as well as the presence of reflection . *
* although the jet and compton coronal models describe the x - ray data equally well , only the jet model naturally describes both the amplitude and slope of the radio data without the need for additional free parameters . *
it is important to note that the radio spectrum , as well as the bulk of the 3 - 200kev x - ray continuum ( i.e. , the overall amplitude and continuum curvature ) , are almost solely driven by the parameters of the jet : energy input to the jet ( @xmath41 ) , radius of the jet base ( @xmath10 ) , electron temperature at the jet base ( @xmath42 ) , slope of the power - law component of the electron energy distribution ( @xmath43 ) , the equipartition parameter ( @xmath44 ) , and the location of the particle acceleration zone in the jet ( @xmath45 ) . if one begins instead with a corona model and attempts to fit the amplitude and slope of the radio data by assuming a relationship to an extant jet model , this will in essence require the addition of many more parameters to describe just two physical quantities .
we argue that nothing can be gained by this approach , which is why we have chosen to study the test case of full coupling . from our set of jet model
fits that successfully describe the gx 339@xmath04 and cyg x-1 data , we can now explore the implications of the fitted parameters .
cyg x-1 and gx 339@xmath04 are somewhat different sources while in their hard states .
the luminosity of gx 339@xmath04 varies much more than that of cyg x-1 .
this is apparent in the fitted values for the power normalization @xmath12 and output power @xmath46 shown in the first two columns of table [ tab : mffjeti ] .
both sources show best fit @xmath12 in the range of @xmath47 @xmath48 , but gx 339@xmath04 shows significantly more variation in this parameter .
differences can also be seen in the other main parameters , for instance on average gx 339@xmath04 favors slightly larger values of the nozzle radius , @xmath10 .
this reflects the slightly higher ratio of x - ray to radio flux observed in cyg x-1 .
a smaller scale jet base , for a given jet power , increases the jet compactness which in turn gives a slightly higher flux ( with the x - ray flux being more sensitive to this effect ) and pushes all jet emission to slightly higher frequencies
. other fit parameters also indicate differences between these two sources ; however , the parameters surprisingly fall roughly in the same range given the _ a priori _ possibility for much greater parameter variations .
note that the lowest luminosity gx 339@xmath04 observation , obs i d 40108 - 02 - 03 , has poor enough statistics that the following general statements about the gx 339@xmath04 fits will be based mostly on the brighter two observations .
for example , both the electron temperature , @xmath42 , and the location of the start of the acceleration zone , @xmath38 , tend to be larger in gx 339@xmath04 than in cyg x-1 .
( although we were unable to find adequate error bars for the latter parameter , the trend of finding larger values in gx 339@xmath04 was persistent . ) additionally , we note that the electron power law index , @xmath43 , shows more variation in gx 339@xmath04 than in cyg x-1 .
although the range in electron temperatures at the base of the jets falls within a range of a factor of @xmath49 for both sources , one of the criticisms of this class of models has been that @xmath50-@xmath51 k is not as `` natural '' a value as the @xmath52kev typically used in thermal comptonization models .
the jet model electron temperature , however , is comparable to or slightly greater than the value typically derived for radiatively inefficient accretion flows ( e.g. , * ? ? ?
if some fraction of the accreting plasma is fed directly into the jets , and also perhaps heated slightly in the process , we would expect such a temperature .
on the other hand , several jet model free parameters seemed to settle quite quickly into similar values for both cyg x-1 and gx 339@xmath04 : the equipartition factor @xmath44 , the fraction of accelerated particles @xmath53 , and the ratio of nozzle length to radius @xmath54 .
physically , this suggests that these jets either are close to equipartition , or we have found a local minima of solutions for @xmath55 , and we may find other reasonable minima for @xmath56 , e.g. , magnetically - dominated jets . we plan to explore this particular question in a future work .
it seems reasonable , however , to freeze the fraction of accelerated particles at @xmath57 and make the statement that reasonably efficient acceleration is expected to occur in hard state bhc jets .
similarly , a compact base / corona region with scale height similar to the radius seems a reasonable assumption .
it is likely that this parameter can also be frozen in future applications of the jet model .
we listed the two `` acceleration parameters '' , the shock speed relative to the bulk plasma flow , @xmath58 , and the ratio of the scattering mean free path to the gyroradius , @xmath59 , as separate fit parameters .
however , as discussed in the appendix , these parameters are perhaps not physically meaningful as currently defined , since we are no longer as convinced that the acceleration process is diffusive fermi acceleration .
these parameters enter into the acceleration rate as the factor @xmath60 , which is compared to the sum of the cooling rates from synchrotron and inverse compton radiation , and adiabatic expansion in order to calculate the local maximum accelerated lepton energy .
therefore , this parameter can be loosely interpreted as a factor related to the efficiency of the acceleration process and merged for future fits into a single parameter . for the current fits ,
the individual components were left free to vary , with mixed results . for cyg x-1 , @xmath61 , suggesting a meaningful range of fits . however , for gx 339@xmath04 , @xmath62 ranges over two orders of magnitude , and we suspect that we did not fully explore parameter space meaningfully for this source . in future applications of the model ,
it is likely that we will combine these two parameters into a single parameter that will absorb our uncertainty about the acceleration process .
while we list best fit values for the multicolor blackbody disk model parameters @xmath63 and @xmath64 , as well as the derived inner radius @xmath65 , it is important to keep in mind that because the data only extend down to a few kev , we can not meaningfully constrain this component from spectral fitting . as described in the appendix ,
the disk photons are not as important to the overall photon field as the locally produced synchrotron photons and thus can mainly be constrained by their direct spectral contribution . this contribution , although weak , is in fact required for a good fit , but it is not unique .
@xcite have also fit some of the x - ray data presented here , and for similar reasons fixed the thermal disk parameters . on the other hand , for the disk emission plus comptonization models presented here ( as well as for those presented in @xcite ) , the temperature of the disk was tied to the temperature of the seed photons i.e .
, the thermal photons were the _ only _ source of seed photons for comptonization .
thus by virtue of these imposed restrictions , much more stringent formal limits for the disk components were attainable .
the main result to take away about the accretion flow modeling is that our fits are generally consistent with a sub - eddington accretion disk with temperatures somewhat less than 1 kev .
similarly , the total power entering into the jets is roughly consistent with being of the same order as the observed luminosity required in the disk to be consistent with the data .
the presence of a weak disk component is necessary for a good fit , and thus disk photons will contribute to the inverse compton component from the jet .
we do not , however , feel we can confidently make any statements about the disk geometry and thus the accretion rate at the inner radius assumedly feeding the jets . for both gx 339@xmath04 and cyg x-1 ,
the amount of reflection required by the jet models is roughly comparable to , although in general slightly lower than , that required by comptonization models .
this trend can be understood by the fact that the jet base ssc / ec component dominates above 10kev , and to some extent subsumes the role played by the `` compton hump '' normally attributed to disk reflection .
this effect of course brings up an interesting point : there is a clear degeneracy in how the spectral hardening above 10kev can be understood in terms of continuum models .
the fact that the jet ssc / ec component has a similar appearance to the compton reflection hump does not preclude the presence of both .
clearly , the fluorescent fe line implies that there must be a degree of reflection .
what these results ( as well as the comptonization fits ; @xcite ) do suggest is that one can not uniquely determine a reflection fraction independently of the presumed continuum model .
the key result of this work is that we have clearly demonstrated that even the high statistics of rxte x - ray data can not distinguish between the jet model and thermal comptonization .
( indeed , the rxte data also can not distinguish between thermal and nonthermal comptonization models , as has been previously discussed by , e.g. , @xcite ) . in this sense
, the base of the jets can be said to effectively `` subsume '' , at least spectrally , the role created for the corona .
the main difference between these two pictures then comes down to geometry , inflow versus outflow , and the relationship of the corona to the lower - frequency - radiating regions of the outer jets .
the jet model presented here is consistent with the physical picture seen in relativistic mhd simulations @xcite , where the corona is not static but instead is a windy hot material blowing away from the inner regions of the accretion flow .
this wind / corona is likely what enters into , or is collimated and becomes , the jets and therefore it is intimately related to the larger scale outflows , similar to what we propose here . in addition , the jet base provides a natural illuminator for the disk . our ability to constrain the role of the accretion disk
is significantly limited by the lack of low - energy data .
ideally we would like to be able to determine the inner temperature , and perhaps also assess the need for a recessed inner edge , explicitly from the observations .
having this information would allow us to better constrain the ratio of external to synchrotron - produced photons in the jet base .
additionally , we expect differences between jet models and compton corona models to become more pronounced as one considers data above the nominal comptonization cutoff at 100200kev . from both of these perspectives ,
the observational situation will likely soon improve with the launch of astro - e2 , which will have the ability to observe from 0.5 - 600 kev and thus more stringently constrain both jet and compton corona models .
additionally , although the broad - band effects of reflection , e.g. , the reflection hump , can not be fit independently from the assumed continuum , there is some hope that finer spectral resolution ( perhaps coupled to correlated studies of the continuum variability ) will be able to more uniquely to determine the reflection fraction based upon detailed modeling of the fe line @xcite . in
the scenario considered here , the nozzle / base of the jet is the dominant region for the creation of the hard x - ray emission via comptonization processes . constructing a completely realistic description of this region , however , is difficult given the that the physics of jet formation , acceleration , and collimation are not yet understood .
the nozzle essentially represents the initial conditions which fix parameters along the rest of the jet .
we have shown that the form we have developed for this region over the last several years can successfully mimic the corona in purpose , but it is not as well - determined feature as we would like .
an alternative , but also currently rather ad - hoc , possibility would be to invoke special conditions linking an outflowing corona with more typical thermal plasma conditions to an outer jet . this would require some kind of acceleration and/or heating of the particles at the interface ( and thus more free parameters ) , but may satisfy many of the spectral requirements . one possible way to discern between these two situations , or at the very least limit the possible flux from a nozzle , could be to calculate the amount of local ionization expected from nozzle uv photons and compare that to what is observed ( j. krolik , priv .
if it can be shown for a given source that all observed line emission can be accounted for by the known companion , this limit may be rather stringent .
we note , however , that in high mass x - ray binaries such as cyg x-1 , the companion luminosity , especially in the uv , is extremely high .
it is therefore more likely that meaningful limits can be obtained for low mass x - ray binaries , such as gx 339@xmath04 .
while we have focused on modeling simultaneous radio / x - ray data sets in this paper , for gx 339@xmath04 and several other sources observations do exist which in addition have simultaneous ir / optical data .
near - ir data in particular can provide a valuable constraint on the jet synchrotron component , especially when the transition turnover from optically thick to thin is resolved ( e.g. * ? ? ? * ) . using the work presented here as groundwork
, we will explore elsewhere whether more parameters can be constrained by such data sets .
similarly , with _ hess _ already online and _ glast _ imminent , predictions of @xmath66-ray fluxes will be another important way of testing not only the current model , but models which incorporate hadronic processes .
however , the next major frontier for application of our jet model clearly is to consider questions of time - dependence . by studying the lag between the radio and x - ray correlated variability
, we can constrain the plasma speed and/or distance between the two regions .
the existing radio / x - ray flux correlations constrain the long time scale , while simultaneous radio / rxte pointed observations constrain thousand of second time scales @xcite .
we are currently using long duration ( several hundred ksec ) , quasi - continuous radio / x - ray observations of cyg x-1 to explore intermediate time scales ( markoff et al .
, in prep . ) .
the models considered here , however , are steady - state and focus on the hard accretion state only .
some of the most interesting and revealing behavior is seen during transitions between these states .
for instance , during transition into the hard state we observe a hard x - ray tail before the jet forms as a detectable radio structure @xcite .
the presence of the tail suggests that the jet base / corona may form as a viable region before the outer stable radio - emitting structures are built up .
there are also significant clues coming from noise and faster scaled time variability such as qpos @xcite which need to be integrated into the picture . in general the state transitions reveal many interesting properties which are clearly not due to a steady state model , but which do share several similar properties to the more persistent hard states . studying the time - dependent behavior of these systems
will likely be an effective way to further constrain the conditions at the jet base / corona .
we acknowledge partial funding and travel grants from the deutscher akademischer austauschdienst ( daad ) and the national science foundation , under nsf grant int-0233441 .
m.a.n . is also supported by nasa grant sv3 - 73016 .
is supported by an nsf astronomy & astrophysics postdoctoral fellowship , under nsf award ast-0201597 .
we thank the aspen center for physics for its hospitality during the final stages of the preparation of this paper , as well as the anonymous referee for helpful comments .
like all models for optically thick outflows , our jet model builds on the initial work of @xcite .
their paper demonstrated how a superposition of idealized , self - absorbed , conical jet components results in a flat synchrotron spectrum in the radio bands . in reality ,
compact jets usually deviate from this idealized case , and show a slight spectral inversion in the radio wavebands , with spectral index @xmath67 ( @xmath68 ) .
a more realistic treatment of the internal physics can account for this inversion . using the framework for hydrodynamical jets developed in @xcite
, @xcite developed a `` free jet '' model which is more self - consistent .
for instance , the assumption that the jet is roughly conical via lateral expansion with a constant sound speed requires weak longitudinal acceleration .
the euler equation can then be solved for the longitudinal velocity gradient , assuming only adiabatic losses , and is analogous to a pressure driven wind solution with the proper jet speed replacing the wind speed .
this weak acceleration combined with relativistic beaming effects will result in the spectral inversion typical of compact jet sources .
these earlier models were very useful for studying the general physical scalings predicted by compact jet models .
however , without full particle distributions and radiative transfer , they can not be tested against observed spectral data .
our models have been developed on the basis of these earlier papers , but specifically focus on the problem of spectral predictions .
these models are only appropriate for systems where the jets are not expected to be highly collimated nor highly accelerated , e.g. , the jets in hard state xrbs and llagn .
we start by assuming that the base of the jets are at the speed of sound for a relativistic gas with adiabatic index 4/3 , giving a proper sound speed @xmath69 .
furthermore , we assume the jets expands laterally with the constant sound speed , which is appropriate under the assumption of maximally efficient jets ( see * ? ? ?
the resulting longitudinal acceleration along the axes is largest near the bases , and asymptotically approaches a maximum of lorenz factor @xmath70 in the outer radio emitting regions , and @xmath71 for the synchrotron emission .
such weak beaming factors are suggested by observations of hard state jets @xcite , in contrast to the near light - speed velocities implied by superluminal motion in higher luminosity radio - emitting states .
the main free parameters of this model are the power normalization @xmath12 , the radius of the jet nozzle @xmath10 , the equipartition of energy between the radiating leptons and the magnetic field @xmath72 , and the parameters determining the initial lepton distribution as described below .
we also left the length of the nozzle region ( with fixed radius ) , @xmath54 free , but it turns out the fits are not very sensitive to its value .
other parameters which come in , such as the central mass @xmath73 , the inclination of the jets to the line of sight @xmath74 and the distance to the source @xmath75 , are determined from observations and remain fixed for any given source .
finally there are two parameters related to the presence of weak disk emission , the total luminosity radiated by the disk and its inner radius temperature , @xmath76 and @xmath77 respectively .
these can not be well constrained by our data because although they are spectrally fit assuming a multi - color disk blackbody @xcite , the low - energy x - ray spectrum is not well covered in our observations .
similarly , although they do contribute to the photon field upscattered by energetic jet leptons , they are not the dominate component .
the most important free parameter is @xmath12 , parameterized in terms of the eddington luminosity , which determines the power initially input into the particles and magnetic field at the base of the jets .
the total power input at the base of the jets is in fact approximately an order of magnitude larger than @xmath78 , and for our model can be estimated only _ a posteriori _ from the resulting emission in the emitting rest frame .
the difference between @xmath41 and the total output radiation @xmath46 expresses the lack of knowledge of what initially collimates the nozzle , as well as what causes the distributed reenergization and consequent energy requirement to maintain the observed ( e.g. * ? ? ?
* ) particle distributions along the jet ( see below ) .
@xmath12 plays a similar role to the compactness parameter in thermal comptonization models , which parameterizes the total energetics in the absence of an understanding of the mechanism for energizing the corona @xcite , as a consequence of assuming maximal jets , there should be approximate equipartition between the kinetic and internal energy .
once @xmath79 is fixed , therefore , the energetics along the jets are fixed as well . for a given @xmath10 and @xmath44 at the base , conservation of energy and particle flux
together with the proper velocity profile along the jet , @xmath80 , determine the radius , density , and magnetic field profiles along the entire jets .
what is left to be determined is how the energy in the radiating particles is distributed . as mentioned earlier , we assume that the jets contain both leptons and hadrons , but that the hadrons remain non - relativistic and serve only to carry the bulk kinetic energy
. the leptons , which are assumed to be mildly relativistic and quasi - thermally distributed ( see , e.g. , * ? ? ?
* ) are thus more likely to be accelerated . in any case , we assume that leptons are predominantly responsible for the observed radiation .
this is begging the question of hadronic acceleration , because the end effect of significant hadronic acceleration and subsequent inelastic collisions will invariably be relativistic leptons .
the expected energy distributions , whether from direct acceleration or creation via hadronic collisions , are not trivial to discern from x - ray emission alone .
therefore we do not specify whether the radiating particles are electrons or positrons , since this will not effect the observable outcome .
eventually the consistency with these results can be checked against hadronic models , but this is a very difficult problem ( see , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and is not considered here .
the one hint for what will be required beyond neutral matter is that we do not require @xmath81 , and once a fit is made , we can see if additional pair creation is necessary .
the current model was developed in simplified form by @xcite , where we first incorporated simple particle distributions for the radiating leptons and calculated the radiative transfer along the jets .
our goal was to determine if the same kind of model which could explain the inverted radio core of the galactic center supermassive black hole , sgr a * , also predicted significant x - ray emission .
this idea was motivated by the first identification of sgr a * in the x - ray band by the _ chandra _ observatory @xcite .
we explored two `` canonical '' types of particle distributions , power - laws and maxwellians , for sgr a * in its quiescent state ( see , e.g. , * ? ? ? * ) .
these two ostensibly different distributions result in similar fits as long as the characteristic particle energy , ( @xmath82 for the power - law or @xmath83 for the maxwellian ) is similar . unlike `` typical '' low - luminosity agn ( llagn ; * ? ? ?
* ) , the quiescent sgr a * does not show any indication of a power - law of optically thin synchrotron emission after the break from its flat / inverted radio spectrum .
therefore if the radiating particles have a power - law distribution , it must be so steep as to be indistinguishable from a maxwellian in the optically thin regime .
either scenario strongly suggests , for example , that acceleration in the jets of sgr a * is absent or very inefficient at @xmath84 .
based on sgr a * , we therefore assume that the baseline particle distribution , below a certain accretion rate which must be very low , is quasi - thermal .
this scenario is consistent with particles being advected into the jets directly from the accretion flow , or created and roughly thermalized near the base .
sources with jets at higher luminosity , such as agn and xrbs , do , however , show the standard indicator of particle acceleration : optically thin synchrotron emission from a power - law distribution of particles ( e.g. , * ? ? ?
* ; * ? ? ?
* and refs .
therein ) .
interestingly , sgr a * shows daily flaring in which the x - ray spectrum rapidly increases in flux and hardens dramatically @xcite .
the flare emission stems from energization of the particles via acceleration , heating or both at once @xcite , resulting in enhanced synchrotron and/or ssc emission in the x - rays .
because the increased x - ray flux during the flares brings sgr a * more in line with typical levels of llagn activity @xcite , one interpretation is that acceleration of particles is the main factor contributing to the comparative weakness of sgr a*. we also therefore assume that above some critical accretion rate above that of sgr a * , particle acceleration will be commonplace in the jets .
this is also necessary to explain the high synchrotron efficiency in xrbs ( see , e.g. , * ? ? ?
* ) . in the only xrb where
the optically thick - to - thin turnover from compact jet synchrotron has been directly observed ( gx 339@xmath04 ; * ? ? ?
* ; * ? ? ?
* ) , it occurs at a significantly lower frequency than would be expected if the acceleration began in the jet base .
we therefore assume that when particle acceleration occurs , it begins at some location @xmath38 along the jet , which is a free parameter but which appears to fall in the range of a few @xmath85 @xmath17 for most sources we have so far considered .
the fraction of particles accelerated into the tail is a free parameter in our model , but is generally driven to fairly high values so we will likely fix it at some value above 80% in future applications .
the power - law index of the tail @xmath43 ( @xmath86 ) is also a free parameter , and can be constrained by the data .
particle acceleration must compete with radiative and adiabatic cooling to energize the quasi - thermal particles into a power - law tail . where the sum of the cooling rates equals the acceleration rate defines the maximum achievable particle energy , @xmath87 .
unfortunately since the acceleration process is still an open question , the acceleration rate is open to significant interpretation . in @xcite and @xcite
, we used acceleration rates appropriate for diffusive shock acceleration in the most conservative case of the magnetic field direction parallel to the shock normal ( see * ? ? ?
the rate of diffusive shock acceleration is always proportional to the magnetic field .
if synchrotron cooling dominates the cooling term , @xmath87 will thus be independent of the magnetic field and can give information about the plasma parameters if the location of the cutoff can be determined .
in @xcite we showed that this results in a synchrotron emission cutoff around 100 kev for @xmath88 and if the ratio between the particle s mean free path for diffusive scattering to the gyroradius , @xmath89 . in general @xmath27 is thought to be limited by the ratio of the particle to shock velocities , and so likely no higher than @xmath90 , but this value is certainly not fixed . at the time of writing , other processes such as stochastic resonant acceleration ( e.g. , * ? ? ?
* ) are coming into favor . given this uncertainty in the mechanism
, we are loosening the constraints on the acceleration rate we used previously and are letting @xmath27 and @xmath91 vary .
if the acceleration rate we are using is not really physical , then these parameters have lost their meaning and become essentially fudge factors absorbing the free parameters controlling the acceleration rate for other mechanisms , to be determined later . if shock acceleration does still hold , then these tell us about the plasma conditions in the jets .
regardless of the process , the inferred cooling times for the accelerated leptons will be too high to maintain the power - law along the jets unless there is continuous distributed acceleration .
distributed , continuous _ in situ _ injection is a standard requirement for explaining the persistence of particle distributions along agn jets @xcite and seems to be a common feature in outflows .
so to summarize the particle distribution discussion , we assume that the particles enter the jet with a relativistic quasi - maxwellian distribution , the temperature of which , @xmath92 , is a fitted parameter . some fraction of the particles are then assumed to be accelerated continuously beyond a location in the jets , @xmath38 . where exactly can be constrained by the frequency of the optically thick - to - thin break in the synchrotron spectrum , if detected , as well as from the synchrotron component contributing to the soft x - rays . as the leptons travel outwards along the jets , they interact with the local magnetic field and photon fields and cool via synchrotron and inverse compton radiation .
inverse compton processes are strongest near the base of the jets , where the density is highest , and thus one x - ray component is due to the upscattering of both jet synchrotron photons as well as thermal disk photons by the quasi - thermal leptons in the base . for the assumptions in this paper (
the thin disk is recessed with inner radius @xmath65 calculated from fitted parameters @xmath76 and @xmath77 ) , the disk thermal photon field is only an important factor at the very base of the jets .
this is because even the mild beaming in the jets quickly serves to weaken the disk photon field s energy density compared to the locally produced synchrotron photons .
once the acceleration zone is reached , it is difficult to suppress x - ray synchrotron radiation from the tail of accelerated leptons in the diffusive acceleration case ( see * ? ? ?
as described above , we are relaxing the constraint on the synchrotron cutoff by allowing for an alternate acceleration process . in this scenario ,
a power - law of synchrotron radiation will contribute significantly at least to the soft x - ray band , but not necessarily to the hard x - rays which are in this model dominated by inverse comptonized disk and synchrotron photons .
this distinction is important because , as shown by @xcite , a very typical hard state model where the x - rays are exclusively due to synchrotron radiation can not easily reproduce disk reflection fractions greater than a few percent ( assuming perpendicular geometry and no disk flaring ) .
the larger distance of the first synchrotron emission region @xmath38 from the disk , combined with moderate beaming , reduces its importance for disk feedback . on the other hand ,
inverse compton processes near the base of the jets can produce reflection fractions of up to @xmath9320% .
this value is actually a lower limit because it does not include the effects of light bending , which will serve to increase the reflection fraction possibly quite significantly @xcite beyond relaxing the assumption about the acceleration rate , there are a few other differences between this model and the synchrotron - dominated models considered in @xcite and @xcite .
the comparatively weak inverse compton component in prior models followed mainly from our choice of a small jet nozzle radius , @xmath94 , which was based on the assumption that the jet radius was on the order of the event horizon .
if , on the other hand , the jet base is contiguous with , or generated in , an extended corona , a larger scale seems more sensible . with this in mind
, we here consider models with with typical values of @xmath95 .
the dependence of the calculated spectrum upon the model parameters , and their interdependence , are complex . increasing the scale of the jet base decreases the electron density as well as the magnetic field , for a fixed equipartition relationship .
this allows one to consider electron temperatures several times higher than those used in our previous models ( to make up for lost synchrotron flux ) .
the higher electron temperatures lead to greater inverse compton emission relative to synchrotron processes in the x - ray band .
alternately , one can compensate for the weakened radiating power of the decreased particle density by increasing the total power input into the jet , @xmath96 .
again , for a fixed equipartition relationship , the particle density is more sensitive to this change than the magnetic field ( because the pressure of the field is @xmath97 ) and thus the inverse compton component will experience a greater boost relatively . ,
k. , swank , j. h. , giles , a. b. , stark , m. j. , strohmayer , t. , zhang , w. , & morgan , e. h. 1996 , in proc .
2808 , p. 59
- 70 , euv , x - ray , and gamma - ray instrumentation for astronomy vii , oswald h. siegmund ; mark a. gummin ; eds .
, 5970 rccccccccc 20181 - 01 - 02 & & _ 7.9 _ & & & & & & _ 25 _ & + 40108 - 02 - 01 & & _ 4.7 _ & & & & & & _ 302 _ & + 40108 - 02 - 03 & & _ 2.1 _ & & _ 4.0 _ & & _ 2.2 _ & _ 0.7 _ & _ 87 _ & + 40099 - 01 - 19 & & _ 7.4 _ & & & & & & _ 16 _ & + 60090 - 01 - 26 & & _ 7.1 _ & & & & & & _ 9 _ & + 60090 - 01 - 41 & & _ 6.4 _ & & & & & & _ 14 _ & + rcccccccccl 20181 - 01 - 02 & & & & & _ 486 _ & & & & & 179/109 + 40108 - 02 - 01 & & & & & _ 0.04 _ & & & & & 118/87 + 40108 - 02 - 03 & _ 0.57 _ & _ 230 _ & _ 0.1 _ & _ 0.36 _ & _ 0.4 _ & & & & & 22/36 + 40099 - 01 - 19 & & & & & _ 0.5 _ & & & & & 225/154 + 60090 - 01 - 26 & & & & & _ 0.1 _ & & & & & 207/177 + 60090 - 01 - 41 & & & & & _ 0.2 _ & & & & & 186/179 + rccccccccccl 20181 - 01 - 02 & & & & & & & & & & & 170/111 + 40108 - 02 - 01 & & & & & & & & & & & 117/90 + 40108 - 02 - 03 & & & & & & & & & & & 17/37 + 40099 - 01 - 19 & & & & & & & & & & & 226/158 + 60090 - 01 - 26 & & & & & & & & & & & 191/181 + 60090 - 01 - 41 & & & & & & & & & & & 186/183 +
|
the hard state of x - ray binaries ( xrbs ) is characterized by a power law spectrum in the x - ray band , and a flat / inverted radio / ir spectrum associated with occasionally imaged compact jets .
it has generally been thought that the hard x - rays result from compton upscattering of thermal accretion disk photons by a hot , coronal plasma whose properties are inferred via spectral fitting .
interestingly , these properties especially those from certain magnetized corona models
are very similar to the derived plasma conditions at the jet footpoints . here we explore the question of whether the ` corona ' and ` jet base ' are in fact related , starting by testing the strongest premise that they are synonymous .
in such models , the radio through the soft x - rays are dominated by synchrotron emission , while the hard x - rays are dominated by inverse compton at the jet base with both disk and synchrotron photons acting as seed photons .
the conditions at the jet base fix the conditions along the rest of the jet , thus creating a direct link between the x - ray and radio emission .
we also add to this model a simple iron line and convolve the spectrum with neutral reflection .
after forward - folding the predicted spectra through the detector response functions , we compare the results to simultaneous radio / x - ray data obtained from the hard states of the galactic xrbs gx 339@xmath04 and cygnus x-1 .
results from simple compton corona model fits are also presented for comparison .
we demonstrate that the jet model fits are statistically as good as the single - component corona model x - ray fits , yet are also able to address the simultaneous radio data .
|
near infrared ( near ir ) imaging surveys have resulted in the discovery of a population of infrared bright , extremely red objects ( ` eros ' ) , which may be of significance to studies of galaxy evolution .
eros , which we define in this paper as having observed optical near - ir colors @xmath10 , have been identified both in the field , and around high - redshift radio galaxies and quasars ( elston , rieke , & rieke 1988 , 1989 , 1991 ; mccarthy , persson , & west 1992 ; eisenhardt & dickinson 1992 ; graham et al .
1994 ; hu & ridgway 1994 ; dey , spinrad , & dickinson 1995 ) .
images of eros from ground based telescopes show that they are spatially extended on scales of @xmath11 .
this suggests that eros are probably galaxies rather than stellar objects ( e.g. , brown dwarfs ) .
most eros were not _ a priori _ selected as radio sources and are generally radio quiet .
extremely faint optically ( @xmath12 ) , their distances and spectral properties ( i.e. , whether they are normal or active galaxies ) remain unknown . the extreme colors of these objects might be attributed either to an old stellar population , or to a younger stellar population reddened by dust ( graham & dey 1996 , hereafter gd96 ; yamada et al .
1997 ) . the spectral energy distributions ( hereafter seds ) of the reddest eros
known are inconsistent with unreddened old populations at any redshift , and suggest that they must be highly reddened starburst galaxies , perhaps the distant counterparts of the local ultraluminous infrared galaxies ( ulirgs ; e.g. , soifer et al .
1984 ) discovered by the _ infrared astronomical satellite _ ( ) .
most eros have been discovered serendipitously , and their space densities are therefore uncertain .
some measurements suggest that the most extreme eros ( , those as red as 10 ) are as abundant as quasars , with a surface density in blank fields @xmath13arcmin@xmath14 ( hu & ridgway 1994 ; cowie et al . 1994 ) .
more recent measurements suggest that the surface density of objects with @xmath15 and @xmath16 is @xmath17arcmin@xmath14 ( beckwith 1998 ) with fainter eros possibly having even higher surface densities ( @xmath18arcmin@xmath14 for eros with @xmath19 ; eisenhardt 1998 ) .
there is also some evidence that their surface density is higher in regions around high - redshift radio galaxies and quasars compared with the general field ( e.g. , aragon - salamanca et al .
1994 ; dey , spinrad , & dickinson 1995 ; yamada et al .
an obvious interpretation of this result is that the eros are clustered around and hence physically associated with the distant , luminous agn , which would imply that eros are at @xmath20 and their rest - frame optical luminosities are 520 times more luminous than unevolved @xmath21 ellipticals at the same redshifts .
alternatively , this excess number density may suggest that our samples of distant luminous agn are biased due to gravitational lensing by foreground mass concentrations with which the eros are associated .
given that the eros may constitute a significant population in our universe , it is important to understand their nature . in this paper
we present new optical , near - ir and sub - mm observations ( 2 ) of ero j164502 + 4626.4 , an ero discovered in the field of the @xmath22 quasar pc1643 + 4631 ( object # 10 of hu & ridgway 1994 ) ' 10 ( e.g. , graham & dey 1996 ; cimatti 1998 ) .
the nomenclature used in this paper follows the iau recommendations for source names . ] .
our new observations convincingly demonstrate that 10 is a @xmath0 luminous , dusty starburst galaxy , a distant counterpart of local ulirgs like arp 220 and mrk 231 , as originally suggested by gd96 ( 3 ) .
the detection of 10 at sub - mm wavelengths , and the relatively large surface density of the ero population suggests that these objects may be candidates for the sub - mm emitters recently discovered in deep scuba surveys ( smail 1997 , blain 1999a , hughes 1998 , barger 1998 ) , and we briefly discuss this possibility and its consequences in 3.5 .
we assume @xmath23 , @xmath24 , and @xmath25 throughout .
for this cosmology , the luminosity distance of 10 is 10.53@xmath26 gpc . for = 0.1 ,
the luminosity distance is larger by a factor of 1.286 .
the field of pc 1643 + 4631 was observed using the wide field planetary camera 2 ( wfpc2 ; trauger 1994 , holtzman 1995 ) on the refurbished _ hubble space telescope ( hst ) _ through the f814w filter on ut 1997 may 7 . a position angle ( p.a . ) of 68.2 was used in order to simultaneously observe both ero j164502 + 4626.4 and ero j164457 + 4626.0 ( objects # 10 and 14 of hu & ridgway 1994 ) .
both objects were targetted on the wf ccds in order to optimize the detection of faint diffuse emission .
we obtained 4 exposures over 2 orbits with the telescope dithered by @xmath27 between orbits .
the total integration time was 5300 sec .
the images resulting from the calibration pipeline were corrected for cosmic rays , registered and coadded .
flux calibration was performed using the photometric zero points of holtzman ( 1995 ) .
the final image reaches a 3@xmath28 limiting surface brightness of 26.77 ab @xmath29 in an 1 aperture .
the wfpc2 f814w image of 10 ( rotated to a normal orientation where north is up and east is to the left ) is shown in figure [ hr10f814 ] .
the galaxy is extended by roughly 09 in @xmath30 in a reflected ` s shaped ' morphology that is suggestive of an interacting system or spiral / tidal arms .
we spatially coregistered the _
image with the near ir
@xmath31band image of gd96 using 16 common objects ; the relative positional accuracy is better than 002 .
a comparison of the f814w and @xmath31band images reveals that the bulk of the near ir emission arises from a region of low optical flux , in between the two bright lobes of optical emission ( the ends of the reflected ` s ' ; figure [ hr10cont ] ) .
the brightest region of optical emission is _ not _ cospatial with the peak of the near ir emission , but instead lies @xmath32 south of it ( figure [ hr10cont ] ) . in a 3 diameter aperture , 10
is found to have a magnitude of @xmath33 ab mag ( at @xmath34 ) or , equivalently , a cousins @xmath35-band magnitude of @xmath36 .
the color of 10 in this aperture is then @xmath37 , consistent with the values quoted by gd96 and hr94 .
the magnitudes of the northern and southern optical lobes , measured in apertures of diameter 05 , are @xmath38 ab mag and @xmath39 ab mag respectively .
we obtained optical spectra of 10 on u.t .
1998 may 4 using the low resolution imaging spectrometer ( lris ; oke 1995 ) at the cassegrain focus of the keck ii telescope .
the observations were made in good seeing ( fwhm @xmath40 ) through a 10 wide slit oriented in p.a .
lris was configured with the 400 line @xmath41 grating ( @xmath42 ; @xmath43 ) , and the spectra cover the wavelength range @xmath44 .
the total exposure time was 3 hours , and the target was dithered along the slit after every 30 min sub - exposure in order to allow for proper fringe removal and sky subtraction .
wavelength calibration was performed using near lamps obtained immediately following the 10 observations , and relative flux calibration was performed using observations of the standard stars wolf 1346 and feige 34 ( massey 1988 , massey & gronwall 1990 ) .
although the observations were made under non - photometric conditions , the spectrum was scaled to an absolute scale using published photometry ( gd96 , hu & ridgway 1996 ) .
the sky - subtracted two - dimensional spectrum is shown in figure [ 2dspectrum ] , and the extracted one - dimensional spectrum ( extracted in a 17 aperture ) and the corresponding 1@xmath28 errors are presented in figure [ optspectrum ] .
the notable features in the optical spectrum are a strong emission line at @xmath459090 ( which corresponds to the [ oii]@xmath463726,3729 doublet at @xmath47 ) and a very red continuum emission ( @xmath48 in the wavelength range @xmath49 ) .
the signal - to - noise ratio in the continuum is too poor to convincingly detect any absorption lines or spectral breaks . the measured flux , fwhm and rest frame equivalent width of [ oii]@xmath463726,3729 are presented in table [ emlines ] .
near - ir spectroscopy of 10 was obtained using the cryogenic spectrograph ( crsp ; joyce 1995 ) at the cassegrain focus of the kpno 4 m mayall telescope on the night of u.t .
1997 june 21 .
crsp was configured with grating # 4 ( 200 @xmath50 , @xmath51 , dispersion @xmath52 ) used in 2nd order and a 10 slit to provide a resolution fwhm of 0.0042 over the wavelength range 1.444 to 1.750 . observations of 10 were obtained by dithering the object along the slit ; each dither sequence consisted of five equally spaced target positions along the slit separated by 8 .
the slit was oriented at p.a.=59.3 in order to include a nearby galaxy ( object 6 in figure [ hr10f814 ] ; our notation follows that of hr94 ) to aid in accurately coadding the data .
each individual observation was 3 min long , and the total exposure time was 195 min .
the seeing during the observations was typically 08 in the @xmath31band .
the atmospheric absorption was corrected using observations of the telluric standard hr6064 ( g1v ) obtained at various airmasses both preceding and following the 10 observations .
the spectral frames were dark - subtracted , flat - fielded , rectified , sky - subtracted , and coadded by using object 6 ( see figure [ hr10f814 ] ) to determine the accurate relative offsets .
the sky subtraction of each spectral frame was performed using a ` local ' sky constructed from the median of the four preceding and four following frames .
the summed two - dimensional spectrum is shown in the upper panel of figure [ 2dspectrum ] .
the extracted spectrum was divided by the mean spectrum of the telluric standards to correct for the telluric absorption , and corrected for the stellar features by multiplying by the mean spectrum of a g0v star constructed by averaging the spectra of hd109358 and hr5868 from the near - ir spectral atlas of lancon and rocca - volmerange ( 1992 ) .
since the continuum emission from 10 is only marginally detected in these observations , the telluric corrections are largely unnecessary , but also provide us with a rough relative flux calibration .
the resulting spectrum was then scaled by a constant to be consistent with the @xmath53-band flux of 10 .
the calibrated near
ir spectrum of 10 is shown in figure [ irspectrum ] , and clearly shows the emission line detected by gd96 .
the line is narrow , and appears to be resolved into at least two components .
the ratio of the wavelengths of the two brighter components roughly matches the ratio of [ ] @xmath456584/ , and is consistent with the original identification ( by gd96 ) of the line as + [ ] .
the extracted spectrum also shows a weak feature at the expected location of the [ sii]@xmath466717,6731 doublet . however , given the signal - to - noise ratio of the present data , we caution that the [ nii]@xmath456584 and the [ sii]@xmath466717,6731 detections are marginal , and should be treated accordingly ( , figure [ irspectrum ] ) .
the emission line measurements are presented in table [ emlines ] .
the numbers in this table were derived by using the specfit software ( kriss 1994 ) as implemented in _
iraf _ to fit gaussians to the emission lines in the calibrated unsmoothed spectrum .
the h@xmath54 and [ nii]@xmath456584 emission lines were fit jointly by requiring the two features to have equal width and holding the ratio of their central wavelengths constant .
the new high - resolution near - ir observation presented here provides a more accurate measurement of the equivalent widths and fwhms of the h@xmath54 and [ nii ] emission lines . in these new higher spectral resolution data ,
the emission lines are found to be narrower and have smaller equivalent widths than those derived by gd96 .
data were obtained during 1997 june , july and december , and 1998 january and february using the sub - mm common - user bolometer array ( scuba ; holland et al .
scuba has two arrays of bolometric detectors which are operated at 0.1 k to achieve sky background - limited performance on the telescope at 350450 and 750850 .
three extra bolometers ` photometric pixels ' optimized for use at 1100 , 1350 and 2000 are positioned around the long - wave array .
photometry of sources significantly smaller than the beam are generally performed using the central pixels of each array , which are aligned to within an arcsecond of each other , or the photometric pixels .
the best photometric accuracy is achieved by averaging the source signal over a slightly larger area than the beam , so the secondary mirror was ` jiggled ' in a filled - square , 9-point pattern , covering @xmath55 . during the ` jiggle '
the secondary mirror was chopped azimuthally by 60 at 6.944hz .
after the first 9 sec ` jiggle ' , the telescope was nodded azimuthally to the reference position ( subsequently every 18 sec ) . at 450 and 850
, we spent 260min on source , with a further 165min at 1350 .
skydips were performed before , during and after the target measurements to determine the zenith opacities , and the telescope pointing accuracy was checked regularly using 1633 + 382 and 3c345 .
all data were calibrated against observations of mars and uranus .
the observing conditions during the scuba runs were excellent , especially during the early 1998 observing periods .
reduction of the 450 and 850 data consisted of taking the measurements from the central bolometer , rejecting spikes , and averaging over 18 sec time intervals .
the signal detected by all the bolometers is dominated by spatially correlated sky emission ( ivison et al .
1998a ) , so data from the adjacent rings of bolometers were treated in a similar manner .
the residual sky background was removed using the median of the inner two rings of pixels , after rejecting those with excessive noise .
this reduced the noise - equivalent flux density at 850to around 90mjyhz@xmath56 , suggesting that the effects of rapid sky variability have been removed entirely .
reduction of the 1350data followed a similar process , though without the removal of sky noise . at 850
, we find a clear detection of 10 at a level well above the expected extrapolation of the weak radio emission ; at 450 and 1350 , the detection significance is lower , but the measured flux densities are @xmath57 detections , and are entirely consistent with thermal emission from optically thin dust .
the flux density measurements at 450 , 850 and 1350 are reported in table [ submmdata ] .
the 850 and 1350 flux densities reported in this paper are roughly half the values reported by cimatti ( 1998 ) .
our more extensive dataset was obtained during exceptional observing conditions , as evident from the 450 detection , and we can exclude a substantially higher flux with some confidence . furthermore , our data were obtained using a single telescope with internally consistent flux calibration , and the 450 and 850 photometric data were obtained simultaneously .
hence , the relative photometry ( and the derived spectral index ) are self consistent and more accurate than other sub - mm measurements of 10 presented to date . there also exist 95 and 175 observations of 10 obtained with the _ infrared space observatory _ ( _ iso _ ) , which can provide additional consistency checks on the sub - mm observations .
we will report on these _ iso _ observations in a forthcoming work by ivison ( 1999 ) .
the spectroscopic observations presented in this paper confirm the @xmath58 redshift of the galaxy by resolving the emission line reported by gd96 into two components which match the h - alpha + [ nii ] identification , the detection of the [ oii]@xmath463726,3729 doublet , and the marginal detection of the [ sii]@xmath466717,6731 emission lines at the same redshift . the h@xmath54/[nii]@xmath456584 ratio ( @xmath59 ) and the relatively narrow linewidths of the [ oii ] and h@xmath54 emission lines suggest that the ionization is more likely to be due to young , hot stars than an active galactic nucleus ( agn ) .
the spectrum of 10 is similar to that observed in local star forming galaxies ( , kennicutt 1992 ) .
the fwhm ( deconvolved fwhm @xmath60 ) of the [ oii ] doublet and equivalent width of the h@xmath54 emission line from 10 are more typical of star forming galaxies than agn ( , liu & kennicutt 1995 ) . however , since an agn could be hidden behind dust and rendered invisible at rest frame optical and ultraviolet wavelengths , a firm statement on the nature of the energy source must await future mid- and far - infrared spectroscopy . in the following discussion
, we assume that young stars power the entire far - infrared emission from 10 .
the _ hst _ wfpc2 image ( figures [ hr10f814 ] and [ hr10cont ] ) clearly shows that 10 has an elongated and distorted morphology . at a redshift of 1.44 , the wfpc2 f814w filter samples the rest - frame wavelength range @xmath462900@xmath613900 ( i.e. , roughly rest - frame @xmath62-band light ) , which includes the [ ] @xmath463726,3729 emission doublet . the equivalent width of this emission feature in 10 implies that the emission line contamination in the f814w filter is @xmath638% ; the observed morphology is therefore dominated by the continuum emission from the galaxy . since the filter samples the continuum at near - uv wavelengths just shortward of the 4000 break ,
the morphology observed in the _ hst _ image only reflects the distribution of uv bright populations ( , hot , young stars ) , and in addition may be heavily modified by the distribution of dust in the galaxy .
the peculiar uv morphology may be due to spiral arms , tidal features or something more complex , and suggests that 10 may be an interacting or distorted system . the most striking aspect of the rest frame near
uv morphology is that it is very different from the rest frame far red morphology observed in the @xmath31band : the red emission is more symmetric ( at least at the 06 resolution of the gd96 ground - based image ) and peaks in a region where the near uv emission falls to a minimum , approximately at the center of the reflected s shaped structure seen in the wfpc2 image ( figure [ hr10cont ] ) . since the far red emission in galaxies is generally dominated by late type giant stars and an older main sequence population and the near uv emission is dominated by younger , hotter stars , the observed morphological difference could be partly due to spatially distinct populations , with the older stars being more centrally concentrated ( perhaps in a bulge ? ) compared to the younger stars ( perhaps in a disk or tidal arms ? ) .
is it possible that the uv morphology is dominated by young super star clusters ?
the specific luminosities of the northern and southern lobes observed in the near uv ( @xmath64 ) image are @xmath65 and @xmath66 respectively .
therefore , these regions have luminosities that are at least 3 times larger than the brightest of the super star clusters observed in local starburst galaxies ( , meurer 1995 ) , and it is possible that these regions are composed of groups of star clusters .
we have not corrected the observed flux densities for the effects of extinction due to dust , and the luminosities quoted should be regarded as lower limits . however , since the extreme red color of 10 requires a significant amount of dust , the observed morphological difference could also be produced by a spatially non - uniform dust distribution which reddens the central regions more than the outer structure .
indeed , the colors in any resolution element are redder than any reasonable stellar population , implying that the entire visible spatial extent of the galaxy is strongly reddened .
it is intriguing that similar uv / optical morphological differences are observed in the inner regions of some nearby ulirgs where they are commonly interpreted as resulting from spatial differences in the dust extinction .
the sed of 10 is remarkably similar to those of low redshift ultraluminous galaxies like arp 220 and mrk 231 ( figure [ hr10sed ] ) , and this strongly suggests that 10 is a dust enshrouded system .
the measured sub - millimeter flux densities at 450 , 850 and 1350 are greatly in excess of the power - law extrapolation of the non - thermal radio emission , and it is therefore safe to assume that the sub - millimeter flux results from thermal emission from dust heated by an agn or in star forming regions . the mm / sub - mm spectral index is also consistent with this hypothesis .
if we assume that the sed of 10 is indeed identical to that of the local ulirgs , we can estimate the luminosity and mass of the warm dust responsible for the thermal emission , and attempt to constrain the star
formation rates necessary to power this galaxy . using the template
sed of an luminous _ iras _ galaxy compiled by guiderdoni ( 1998 ) , we find that a least squares fit of the guiderdoni sed to the three sub - mm points give a bolometric luminosity of @xmath67 with reduced @xmath68 ( figure [ hr10sedmod ] ) .
almost all of this emission emerges in the far infrared : @xmath69 , where we have defined @xmath70 as the luminosity emerging between rest wavelengths of 10 and 2 cm .
this ranks 10 among the most luminous ( unlensed ) ir galaxies known , and classifies it as an ulirg .
note that the above estimate of the bolometric luminosity requires a large extrapolation .
the sub - mm luminosity ( estimated from @xmath71 $ ] , , the best - determined sub - mm flux ) is @xmath72 , and so this is an extrapolation of two orders of magnitude .
however , the sub - mm flux densities are well fit by optically thin thermal dust emission with an emissivity law with index 1.5 ( , @xmath73 ) and temperature of @xmath74 k. this dust temperature is typical of ulirgs ( klaas & elsasser 1993 ) .
these modified black body model fits to the sub - mm data have steeper spectra than the guiderdoni sed at @xmath75 , and result in slightly lower values of @xmath70 .
however , although the sub - mm data are all on the rayleigh - jeans tail of the dust emisssion and therefore do not strongly constrain the dust temperature , the small @xmath76 of the fit and the reasonable value of the derived dust temperature suggest that our estimate of the far - infrared luminosity is uncertain by at most a factor of two . the corresponding mass of dust
is @xmath77 for optically thin emission from grains with normal interstellar parameters ( draine & lee 1984 , hildebrand 1983 ) .
the star formation rates can be estimated from either the h@xmath54 luminosity or the sub - mm continuum luminosity only if we assume that young , hot stars provide the energy source for the line and thermal continuum emission . for our adopted cosmology ( = 50@xmath78 and = 0.5 ) ,
the luminosity in the h@xmath54 emission line is @xmath79 .
if this emission line is powered entirely by young , hot stars , then its luminosity implies a total star formation rate of @xmath80 ( kennicutt 1983 ) .
this is almost certainly a lower limit to the true star formation rate in the galaxy , since we have not made any corrections for the dust extinction .
a more reliable estimate for the star formation rate in 10 can be derived under the assumption that the sub - mm continuum excess is due to thermal emission from dust grains heated by young , hot stars .
if stars are formed at a constant rate over @xmath81 yr with a salpeter initial mass function ( @xmath82 ; @xmath83 ) , a luminosity of @xmath84 corresponds to a star formation rate between @xmath85 ( leitherer and heckman 1995 ) .
therefore the observed bolometric luminosity corresponds to a total star formation rate of @xmath86 ( or a factor of three smaller for the formation rate of only massive stars ) .
a different estimate for the formation rate of massive stars can be derived by relating the total luminosity to the mass consumption rate in early type stars .
scoville & young ( 1983 ) derive @xmath87 , which implies a formation rate of @xmath88 for the massive stars .
note that this estimate does not account for the formation of lower mass stars , the mass cycled through or locked up in lower mass stars and their remnants .
this star formation rate is therefore a lower limit , and as such is consistent with the estimates derived above using the leitherer & heckman ( 1983 ) starburst models .
if the sub - mm luminosity is indeed powered by young stars , the derived star formation rate is @xmath89 times greater than that inferred from the h@xmath54 luminosity ; if this difference is due to dust extinction by a foreground screen , it implies an extinction of @xmath90 mag . for a galactic extinction curve , this would imply that the intrinsic ratio of [ ] to h@xmath54 of @xmath91}/f_{\rm h\alpha})_0 \approx 1.3 $ ] , which is at the extreme limit of observed values for local star forming galaxies ( , kennicutt 1992 ) .
since the dust is more likely to be intermixed with the line - emitting gas than external to it , the extinction derived by comparing the sub - mm and h@xmath54 luminosities may not be directly applicable to the reddening correction of the emission line ratios .
the @xmath92 inferred here is about 2.5 mag larger than that estimated by fitting the 10 optical and near - ir flux densities with a reddened sb galaxy model sed ( gd96 ) .
the star formation rate derived from the sub - mm continuum emission is extremely large and may lead us to the conclusion that the dust emission is powered by an agn rather than by star formation .
indeed , the source of the large luminosities in local ulirgs is still a matter of debate ( , sanders & mirabel 1996 ) .
however , there is no evidence from the existing spectroscopy that this is the case : as noted in 3.1 , the emission line ratios and widths are more typical of starburst galaxies than agn .
the dust mass estimate from the previous section and the star formation rate derived here can be combined to yield a crude order - of - magnitude estimate for the lifetime of the starburst . assuming that 10 has a ` normal ' gas - to - dust ratio of 0.01 , the total mass of gas is @xmath93 , and the lifetime of the burst is @xmath94 yr . the total mass of stars produced is @xmath95 , comparable to that of a present - day massive galaxy . in this section we have assumed that the sub - mm continuum emission is due to thermal emission from dust heated by young , hot stars .
one curious property of 10 which may be relevant to this hypothesis is that the rest frame 60 emission predicted by the dust emission and the nonthermal radio emission at a rest wavelength of 6 cm ( extrapolated from the flux density observed at 1.4 ghz assuming a @xmath96 synchrotron spectrum ) does not follow the well - known 60
6 cm correlation obeyed by local star forming galaxies ( , de jong 1985 ) .
the rest frame 6 cm emission predicted by this correlation for 10 is @xmath97jy , more than an order of magnitude above the detected radio emission ( table [ submmdata ] ) .
the only ways of decreasing the predicted 60 emission would be to decrease the dust temperature and increase the dust optical depth , both of which result in worse fits to the sub - mm data .
the existing _ iso _ data may be able to help constrain the dust temperature and optical depth estimates ( ivison 1999 ) .
this issue can also be resolved with better far - infrared observations of 10 , which will soon be possible with the _ space infrared facility _ ( _ sirtf _ ) .
if the large deviation from the 60 6 cm correlation is real , its interpretation is unclear .
it is unlikely that this deviation is due to the sub - mm flux being heated by an agn instead of a starburst , since in the local universe agn tend to depart from the correlation by having excess nonthermal radio emission , not excess far - infrared emission ( , dey & van breugel 1994 ) .
some local ulirgs ( , arp 220 ) do show low frequency turnovers in their radio spectra which are generally attributed to thermal absorption of the radio emission ( sopp & alexander 1991 ) . on the other hand ,
the first reliably identified sub - mm selected high redshift galaxy , smm j02399@xmath610136 ( @xmath98 ; ivison 1998b ) , has a ` normal ' 60 6 cm ratio , yet shows some evidence for an agn contribution to the total emission . since the far - infrared flux densities are poorly constrained at present , and the origin of the correlation in local galaxies is also not completely understood , we hesitate to draw any firm conclusions from this observation . measurements of the global star - formation history of the universe , using deep redshift surveys , e.g. , the canada france redshift survey ( lilly et al . 1996 ) reaching @xmath99 , the statistics of lyman - limit galaxies ( steidel & hamilton 1992 ) at @xmath100 , the hubble deep field ( hdf ) @xmath101 ( madau et al .
1996 , connolly et al .
1997 ) , imply that the star - formation and metal - production rates were about 10 times greater at @xmath99 than in the local universe , that they reach a maximum somewhere in the redshift range @xmath102 and remain roughly constant ( or perhaps slowly decline ) at higher redshifts ( steidel 1999 ) . these conclusions , which are based almost entirely on samples selected at optical and near - uv wavelengths ,
may be misleading ( smail , ivison & blain 1997 ; blain 1999a ) .
absorption by dust in regions of star - formation may have distorted our picture of galaxy evolution in the high- and low - redshift universe in two ways .
first , neglect of dust leads to an underestimate of the the star - formation rate in known high- and low - redshift objects .
second , it is possible that an entire population of heavily dust - enshrouded high - redshift objects , such as eros , has escaped undetected in the optical / uv surveys .
a useful diagnostic of the redness of the sed is the rest frame infrared to blue luminosity ratio ( @xmath103 ) , where @xmath104 at @xmath105 m and @xmath106 at @xmath107 . by this measure , the reddest galaxy in the ugc is arp 220 ( @xmath108 ; soifer 1984 ) .
the mean @xmath103 value for the ultralumimous sample is 25 ( sanders 1987 ) , and the reddest ultraluminous galaxy in the _ iras bright galaxy catalog _ is iras 12112 + 0305 , which has @xmath109 .
if we estimate @xmath110 for 10 by interpolating the observed @xmath35 and @xmath111 fluxes , we find that 10 has @xmath112jy .
hence , @xmath113 , and @xmath114 ! the name ero is therefore justified for this remarkable object . the relevance of 10 to cosmology and galaxy formation depends critically on the space density of the ero population , which is highly uncertain at present . under the assumptions that the eros form a homogeneous population and the redshift and sub - mm luminosity of 10is typical of this class
, gd96 estimated that the space density of these objects is @xmath115 .
hence , eros may be more abundant than the _ iras _ ultraluminous galaxies and quasars by two orders of magnitude . if the star formation rate for these galaxies is similar to that observed in 10 ( , @xmath116 ) , the star formation rate at @xmath117 associated with the ero population is @xmath118 .
this sub - mm estimate of the star formation density exceeds the rest - frame uv estimates of the high redshift star formation rate ( from the cfrs and hdf surveys ) by more than a factor of six .
it is noteworthy that this crude estimate is roughly consistent with the predictions of pei & fall ( 1995 ) ( see also hughes 1998 ) .
deep surveys at 850 of the sub - mm sky have resulted in the detection of several faint sources ( @xmath119 mjy ) , about 20% of which appear to have no obvious optical counterpart brighter than @xmath120 ( smail 1998 ) .
it is possible that these ` missing ' optical identifications are galaxies similar to 10 , perhaps even fainter at optical wavelengths , but with similarly strong thermal dust emission at sub - mm wavelengths .
since the cumulative surface density of the sub - mm continuum emitters with 850 flux densities @xmath121 mjy is @xmath122 , the surface density of the optically faint fraction is @xmath123 ( estimates here are based on the surface density determined by blain 1999b ; however , see barger 1998 for a lower estimate ) .
the surface density of eros is highly uncertain : hu & ridgway ( 1994 ) estimated that the surface density of objects with @xmath124 and @xmath125 is @xmath126 , whereas estimates of the surface density of less extreme eros ( with @xmath127 ) discovered serendipitously tend to be larger by at least an order of magnitude ( , dey , spinrad & dickinson 1995 ; knopp & chambers 1997 ; beckwith 1998 ) .
the roughly comparable space densities of these two populations suggests that it is possible that eros comprise a significant fraction , perhaps all , of the optically faint sub - mm emitters discovered in recent surveys .
a large fraction of both local ulirgs and distant scuba sources ( smail 1998 ) appear to be interacting or merging systems , perhaps similar to 10 .
since the sub - mm sources thus far detected at flux densities @xmath128 mjy arguably account for upto 100% of the 850 cosmic sub - mm background ( smail 1997 , blain 1999 ) , the ero population may comprise a substantial component of the resolved emission .
the relevance of the properties of 10 for our general understanding of both the ero and sub - mm populations will hinge on whether it represents a particularly high luminosity or low redshift member of this group . the very red sed of 10 also implies that it will be extremely difficult to detect the most actively star - forming galaxies at higher redshifts , since they may be completely enshrouded by dust and undetectable at optical and near - ir wavelengths .
for instance , if 10 were at @xmath129 , it would have a @xmath130 and @xmath131 ( = 50 , = 0.5 ) , and therefore undetectable in even the deepest existing ground- and space - based surveys .
in contrast , the large negative sub - mm k - correction would imply that the 850 flux density of such an object would be @xmath132 mjy , comparable to that at @xmath0 .
the existence of dusty star - forming systems at @xmath133 is not an unrealistic expectation : the @xmath134 galaxy 0140 + 326rd1 ( dey 1998 ) has been recently argued to be a dusty star forming system based on its red @xmath135 color ( armus 1998 ) .
it is important to note , however , that the inferred dust content of 0140 + 326rd1 is much lower than that of 10 : the implied extinction is @xmath136 mag for 0140 + 326rd1 , compared with @xmath137 for 10 .
investigating the properties of the higher redshift counterparts of systems like 10 will require sensitive ground - based sub - mm facilities with high - angular resolution and space - based telescopes operating at near - ir and mid - ir wavelengths . in the near - future , the _ space infrared telescope facility _
( _ sirtf _ ) will provide flux density observations of eros at mid- and far - infrared wavelengths ( @xmath138 ) allowing us to place more critical constraints on the dust masses , temperatures and source luminosities , and perhaps determine the relationship of these objects to present - day galaxies .
we have presented new optical , near ir and sub - mm observations of ero j164502 + 4626.4 ( object # 10 in hu & ridgway 1994 ) , an extremely red object in the field of the qso pc 1643 + 4631a . the new optical and near ir spectroscopy show that 10 is a distant galaxy lying at a redshift of @xmath0 .
the peculiar rest frame near uv and far red morphologies suggest that 10 is a disturbed or interacting system .
the far ir and sub - mm evidence are consistent with the hypothesis that this ero is a dust - enshrouded object , with its luminosity ( @xmath139 ) powered by either a starburst or an agn .
the existing spectral data for 10 show no strong evidence for agn emission at rest - frame near uv or optical wavelengths .
if the observed sub - mm continuum flux is due to optically thin thermal emission from dust heated by a young , star forming population , the inferred star formation rate is extremely large ( @xmath140 ) . in all its known properties , 10appears to be a high - redshift luminous counterpart of the dusty , ultraluminous galaxies discovered in the local universe by _
iras_. although it is difficult to draw general conclusions from a single object , less luminous or higher redshift galaxies similar to 10 may be missed from even the deepest existing optical and surveys .
it is important to determine the space densities of these eros and their relevance to our understanding of galaxy formation and evolution in the distant universe .
we thank brett huggard , gillian rosenthal & dick joyce for their expert assistance during our kpno observing runs , richard elston for advice on using crsp , the numerous people who have made possible the process of obtaining and calibrating _ hst _
wfpc2 data , and the jcmt staff for assistance during the scuba observations .
we are grateful to wayne wack , david sprayberry and bob goodrich for their help with the keck lris observations .
we thank bill reach for help with determining the galactic far - infrared foreground towards 10 , and the referee , peter eisenhardt , for constructive comments on our manuscript .
ad acknowledges the support of nasa hf-01089.01 - 97a and partial support from a postdoctoral research fellowship at noao , operated by aura , inc . under cooperative agreement with the nsf .
jrg acknowledges support from nasa go-06598.02 - 95a .
the w. m. keck observatory is a scientific partnership among the university of california , the california institute of technology , and the national aeronautics & space administration , and was made possible by the generous financial support of the w. m. keck foundation .
the james clerk maxwell telescope is operated by the joint astronomy centre on behalf of the particle physics and astronomy research council of the united kingdom , the netherlands organisation for scientific research and the national research council of canada .
aragon - salamanca , a. , ellis , r. s. , schwartzenberg , j .-
m . , & bergeron , j. a. 1994 , apj , 421 , 27 armus , l. , matthews , k. , neugebauer , g. & soifer , b. t. 1998 , , 506 , l89 barger , a. j. 1998 , nature , 394 , 248 barvainis , r. , antonucci , r. , hurt , t. , coleman , p. , & reuter , h .-
1995 , apj , 451 , l9 beckwith , s. v. w. , thompson , d. , manucci , f. & djorgovski , s. g. 1998 , , 504 , 107 blain , a. w. , kneib , j .-
ivison , r. j. , & smail , i. 1999a , , in press ( astro - ph/9812412 ) blain , a. w. , smail , i. , ivison , r. j. & kneib , j .- p . 1999b , , 302 , 632 broadhurst , t. , & lehr , j. 1996 , apj , 450 , l41 cimatti , a. , andreani , p. , rttgering , h. & tilanus , r. 1998 , nature , 392 , 895 connolly , a.j . , szalay , a.s . , dickinson , m. subbarao , m.v , brunner , r.j .
1997 , apj , 486 , l11 cowie , l. l. , gardner , j. p. , hu , e. m. , songalia , a. , hodapp , k .- w . , & wainscoat , r. j. 1994 , apj , 434 , 114 dey , a. , spinrad , h. , & dickinson , m. 1995 , apj , 440 , 515 dey , a. , spinrad , h. , graham , j. r. & chaffee , f. h. 1998 , , 498 , l93 dey , a. & van breugel , w. 1994 , in `` mass transfer induced activity in galaxies '' , i. shlosman , ed .
( cambridge univ press ) p.263 draine , b. t. & lee , h. m. 1984 , , 285 , 89 eisenhardt , p. r. m. , & dickinson , m. 1992 , apj , 399 , l47 eisenhardt , p. , elston , r. , stanford , s. a. , dickinson , m. , spinrad , h. , stern , d. , & dey , a. 1998 , in the birth of galaxies , " proceedings of the xth recontres de blois , eds .
b. guiderdoni , f. r. bouchet & t. x. thuan , paris : editions frontieres , in press .
elston , r. , rieke , g. h. , & rieke , m. j. 1988 , apj , 331 , l77 elston , r. , rieke , m. j. , & rieke , g. h. 1989 , apj , 341 , 80 elston , r. , rieke , g. h. , & rieke , m. j. 1991 , in astrophysics with infrared arrays , ed .
r. elston ( asp conf .
ser . , 14 ) , 3 fixsen , d. j. , dwek , e. , mather , j. c. , bennett , c. l. & shafer , r. a. 1998 , , in press ( astro - ph/9803021 ) frayer , d. t. 1996 , phd thesis , university of virginia graham , j. r. , et al . 1994 , apj , 420 , l5 graham , j. r. , & dey , a. 1996 , apj , 471 , 720 graham , j. r. & liu , m. c. 1995 , , 449 , l29 guiderdoni , b. , hivon , e. , bouchet , f. r. , maffei , b. 1998 , , 295 , 877 hauser , m. g. 1998 , , in press ( astro - ph/9806167 ) hildebrand , r. h. 1983 , qjras , 24 , 267 holland , w. s. et al .
1998 , mnras , submitted holtzman , j. , 1995 , , 107 , 156 hu , e. m. , & ridgway , s. e. 1994 , aj , 107 , 1303 hughes , d. 1998 , nature , 394 , 241 ivison , r. j. 1995 , mnras , 275 , l33 ivison , r. j. et al . 1998a , apj , 494 , 211 ivison , r. j. et al .
1998b , , 298 , 583 ivison , r. j. et al .
1999 , in preparation joyce , r. 1995 , `` cryogenic spectrometer user manual '' , noao kennicutt , r. c. 1983 , , 272 , 54 kennicutt , r. c. 1992 , , 79 , 255 kessler , m. f. , et al .
1996 , a&a , 315 , l27 klaas , u. & elsasser , h. 1993 , aa , 280 , 76 knopp , g. p. & chambers , k. c. 1997 , , 109 , 367 lancon , a. & rocca volmerange , b. 1992 , , 96 , 593 lemke , d. , et al .
1996 , a&a , 315 , l64 leitherer , c. & heckman , t. m. 1996 , apjs , 96 , 9 lilly , s.j . , le fvre , o. , hammer , f. , crampton , d. 1996 , , 460 , l1 liu , c. t. & kennicutt , r. c. 1995 , , 450 , 547 madau , p. et al .
1996 , mnras , 283 , 1388 massey , p. , strobel , k. , barnes , j. v. & anderson , e. 1988 , , 328 , 315 massey , p. & gronwall , c. 1990 , , 358 , 344 mccarthy , p. j. , persson , s. e. , & west , s. c. 1992 , apj , 386 , 52 meurer , g. r. , heckman , t. m. , leitherer , c. , kinney , a. , robert , c. & garnett , d. r. 1995 , , 110 , 2665 oke , j. b. , et al . 1995 , pasp , 107 , 375 pei , y. c. & fall , s. m. 1995 , , 454 , 69 puget , j. l. 1996 , , 308 , l5 rowan - robinson m. , et al . 1993 ,
mnras , 261 , 513 sanders & mirabel 1996 , araa , 34 , 749 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 smail , i. , ivison , r. j. & blain , a. w. 1997 , apj , 490 , l5 smail , i. , ivison , r. j. , blain , a. w. & kneib , j .- p . 1998 , apj , 507 , l21 sopp , h. m. & alexander , p. 1991
, , 251 , 112 soifer , b. t. , et al .
1984 , apj , 283 , l1 soifer , b. t. , et al .
1992 , apj , 399 , l55 steidel , c.c . ,
hamilton , d. 1992 , , 104 , 941 steidel , c. c. , adelberger , k. l. , giavalisco , m. , dickinson , m. & pettini , m. 1999 , apj in press ( astro - ph/9811399 ) trauger , j. t. , 1994 , , 435 , l3 yamada , t. , tanaka , i. , aragon - salamanca , a. , kodama , t. , ohta , k. , & arimoto , n. 1997 , apj , 487 , l125 lccccc [ oii]@xmath463726,3729 & [email protected] & 1.439 & [email protected] & 420@xmath141100 & 47@xmath1415 h@xmath54 & 16030@xmath1417 & 1.443 & 33@xmath1418 & 597@xmath141140 & 89@xmath14120 [ nii]@xmath456584 & 16081.1 & & 14@xmath1415 & 597 & 37@xmath14113 [ sii]@xmath466717,6731 & 16437@xmath1417 & 1.444 & 14@xmath1416 & 532@xmath141250 & 38@xmath14117 [ emlines ] ccccl 4400 & 1800 & @xmath142jy & & gd96 , hu & ridgway ( 1992 ) 7930 & 3250 & @xmath143jy & wfpc2/_hst _ & this paper 1.2 & 4920 & @xmath144jy & & gd96 , hu & ridgway ( 1992 ) 1.6 & 6560 & @xmath145jy & & gd96 , hu & ridgway ( 1992 ) 2.2 & 9010 & @xmath146jy & & gd96 12 & 4.9 & @xmath147 mjy & _ iras _ & 25 & 10.2 & @xmath148 mjy & _ iras _ & 60 & 24.6 & @xmath149 mjy & _ iras _ & 100 & 41.0 & @xmath150 mjy & _ iras _ & 450&184 & @xmath151 mjy & scuba / jcmt & this paper 850&348 & @xmath152 mjy & scuba / jcmt & this paper 1350&553 & @xmath153 mjy & scuba / jcmt & this paper 3.6 cm & 1.5 cm & @xmath154jy & & frayer ( 1996 ) 20 cm & 8.6 cm & @xmath155jy & & frayer ( 1996 ) [ submmdata ]
|
we present new near - infrared and optical spectroscopic observations which confirm the redshift of the @xmath0 extremely red object ero j164502 + 4626.4 ( object # 10 of hu & ridgway 1994 ; formerly known as ` hr 10 ' or ` [ hr94 ] 10 ' ) and a _ hubble space telescope _ image which reveals a reflected - s shaped morphology at ( rest frame ) near - ultraviolet wavelengths .
the contrast between the rest frame far - red ( @xmath1 ) and near - uv ( @xmath2 ) morphologies suggests that the central regions of the galaxy are heavily obscured by dust and that the galaxy is most likely an interacting or disturbed system .
we also present new photometry of this object at 450 , 850 and 1350 obtained using the scuba submillimeter camera on the james clerk maxwell telescope .
our sub - mm data are extremely sensitive to emission from cold dust at high redshift . the rest
frame spectral energy distribution of ero j164502 + 4626.4 is best understood in terms of a highly reddened stellar population with ongoing star formation , as originally suggested by graham & dey ( 1996 ) .
the new submillimeter data presented here indicate that the remarkable similarity to ultraluminous infrared galaxies ( ulirgs ) such as arp 220 and mrk 231 extends into the rest - frame far - infrared which bears the signature of thermal emission from dust , presumably heated by young stars .
ero j164502 + 4626.4 is extremely luminous ( @xmath3 ) and dusty ( @xmath4 ) .
if its luminosity is powered by young hot stars , then ero j164502 + 4626.4 is forming stars at the prodigious rate of @xmath5 . we conclude that ero j164502 + 4626.4 is a distant analogue of the nearby ulirg population , the more distant or less luminous counterparts of which may be missed by even the deepest existing optical surveys .
the sub - mm emitters recently discovered by deep scuba surveys may be galaxies similar to ero j164502 + 4626.4 ( but perhaps more distant ) .
this population of extremely dusty galaxies may also contribute significantly to the cosmic sub - mm background emission .
10ero j164502 + 4626.4 # 1 # 1 # 1to 0pt#1 3@xmath6 2 @xmath7 3mpch^-3mpc^3 0@xmath8 0@xmath9 # 1 # 110^#1 # 1to 0pt#1 a&a # 1;#2;#3;#4 # 1 , # 2 , # 3 , # 4 # 1#2#3#4 , # 3 , # 4 # 1#2#3 # 1 _ accepted for publication in the astrophysical journal _
|
supersymmetry ( susy ) has been intensively studied as a fundamental law of nature for following reasons : susy reduces degrees of divergences in quantum field theories @xcite , and hence stabilizes the vast separation between the electroweak scale and the planck scale or the grand unification scale @xcite .
if susy is broken by some strong gauge dynamics , the smallness of the electroweak scale is explained by dimensional transmutation . in the minimal susy extension of the standard model ( mssm ) ,
three gauge coupling constants unify , which is consistent with the prediction of the simplest grand unified theory ( gut ) , namely the @xmath2 gut @xcite . with an @xmath3 parity preserved ,
the lightest susy particle ( lsp ) is a good candidate of dark matter @xcite . the discovery of the standard - model - like higgs as heavy as @xmath4 gev @xcite strongly constrains the mass spectrum of mssm particles . in the mssm ,
quartic couplings of higgs fields are given by @xmath5 term potentials of standard model gauge interactions in the susy limit .
thus the mass of the standard - model - like higgs is lighter than the mass of the @xmath6 boson at the tree level . the observed large higgs mass indicates that quantum corrections from susy breaking effects to the higgs mass are significant @xcite . to explain the observed higgs mass , soft scalar masses of scalar top quarks
are required to be larger than about @xmath7 tev @xcite .
we note that gravity mediated soft scalar masses larger than @xmath8 tev are compatible with physics in the early universe .
planck scale - suppressed mediation , where the gravitino mass is as large as soft scalar masses , with the gravitino mass larger than @xmath8 tev , is free of the constraint from the big - bang - nucleosynthesis even for a high reheating temperature @xcite . with scalar masses and the gravitino mass larger than @xmath8 tev
, one may naively expect that the lsp mass is also of the same order , which leads to the overproduction of the lsp from thermal bath in the early universe .
tev @xcite .
] however , if susy breaking fields are charged or composite , gaugino masses are far smaller than soft scalar masses due to an approximate classical super - weyl symmetry . actually , this is the case with many dynamical susy breaking models ( see e.g. refs .
it should be noted that susy breaking models with charged or composite susy breaking fields are free from the polonyi problem @xcite . in this thesis
, we consider the susy breaking scenario where the gravitino mass is larger than @xmath8 tev and susy breaking fields are charged or composite , which we refer to as the heavy sfermion scenario " .
the pev susy @xcite , the pure gravity mediation model @xcite , the minimal split susy model @xcite and the spread susy model @xcite belong to this scenario . in sec .
[ sec : pgm ] , we review the theoretical framework of the heavy sfermion scenario , including its cosmology . further , we assume that the dirac mass term of the higgsino , so - called the @xmath9 term , is also as large as the gravitino mass .
indeed , the @xmath9 term as large as the gravitino mass is naturally explained if any charges of the up and the down type higgs chiral multiplets add up to zero @xcite ( see also ref .
@xcite ) . the heavy sfermion scenario with this origin of the @xmath9 term
is called the pure gravity mediation model @xcite . assuming that the higgsino is heavy , we pay attention to gaugino masses in the heavy sfermion scenario . in the heavy sfermion scenario ,
gaugino masses are at least given by the anomaly mediation @xcite .
the anomaly mediation yields gaugino masses proportional to the gravitino mass and the beta functions of the corresponding gauge coupling constants .
the lsp is the neutral wino with a mass of @xmath10 tev for the gravitino mass of @xmath11 tev .
due to its large ( co)annihilation cross - section , the thermal abundance of the neutral wino is smaller than the observed dark matter abundance as long as the wino mass is smaller than @xmath12 tev @xcite .
[ sec : amsb ] is devoted to the derivation of the anomaly mediated gaugino mass in the superspace formulation of supergravity with a wilsonian effective action @xcite . in the superspace formulation of supergravity
@xcite , the anomaly mediated gaugino mass in the wilsonian effective action invariant under the super - diffeomorphism was not known @xcite .
this is because the graivitino mass , which is the origin of the anomaly mediated gaugino mass , is the vacuum expectation value ( vev ) of the scalar component of the supergravity multiplet .
couplings of the supergraivity multiplet to the gauge multiplet are strongly constrained by the super - diffeomorphism invariance .
thus , it is difficult ( actually impossible ) to write down the anomaly mediated gaugino mass in the wilsonian effective action in a manifestly super - diffeomorphism invariant way .
we give a super - diffeomorphism invariant expression of the anomaly mediated gaugino mass by taking the path - integral measure into account . in sec .
[ sec : correction ] , we discuss the deviation of gaugino masses from the prediction of the anomaly mediation in the mssm . in the presence of a flat direction coupling to gauge charged matters , gaugino masses receive corrections as large as the anomaly mediation @xcite . actually , this is the case with a ksvz - type qcd axion @xcite .
the ksvz - type qcd axion couples to vector - like matter fields charged under the standard model gauge interaction .
thus , gaugino masses in general deviate from the prediction of the anomaly mediation in the mssm if the ksvz - type qcd axion exists @xcite .
gaugino masses also receive corrections as large as the anomaly mediation if there are vector - like matter fields whose masses are smaller than the gravitino mass @xcite . actually , vector - like matter fields as heavy as the gravitino are predicted in models with an anomaly free discrete @xmath3 symmetry @xcite . with the above two corrections , the gaugino mass spectrum drastically changes .
for example , the gluino mass can be lighter than the prediction of the anomaly mediation in the mssm , which enhances the detectability of the gluino at the large hadron collider ( lhc ) .
we pay attention to the case where some gauginos are degenerated in their masses with each other , because the thermal relic abundance of the lsp as well as the the strategy to search gauginos drastically change in this case .
we refer to this region of gaugino masses as the gaugino coannihilation region " . in sec .
[ sec : pheno ] , we calculate the thermal relic abundance of the lsp in the gaugino coannihilation region , and discuss the phenomenology of gaugino searches at the large hadron collider ( lhc ) and cosmic ray experiments .
in this section , we review the heavy sfermion scenario .
we first review the relation between sfermion masses and the observed higgs mass , and show that heavy sfermion masses are suggested .
then we discuss the mass spectrum of susy particles in the heavy sfermion scenario , where susy breaking fields are charged or composite fields .
as we will see , gaugino masses are one - loop suppressed in comparison with soft scalar masses .
finally , we investigate the compatibility of the heavy sfermion scenario with cosmology .
we show the upper bound on the wino mass from the thermal relic abundance of the wino .
we also show that the gravitino problem and the saxion / axino problem are is considerably relaxed in the heavy sfermion scenario .
we first calculate the tree level higgs mass . at the tree level in the mssm ,
the potential of the up - type higgs @xmath13 and the down - type higgs @xmath14 is given by @xmath15 \nonumber \\ & & + \frac{1}{8 } ( g^2 + g'^{2})\left(|h_u^0|^2 + |h_u^+|^2 - |h_d^0|^2 -|h_d^-|^2 \right)^2 + \frac{1}{2}g^2 | h_u^+ h_d^{0 * } + h_u^0 h_d^{- * } |^2,\end{aligned}\ ] ] where @xmath9 , @xmath16 , @xmath17 , @xmath18 , @xmath19 and @xmath20 are the supersymmetric higgsino mass , the soft scalar squared masses of the up - type and the down - type higgs , the holomorphic quadratic soft mass , the @xmath21 and @xmath22 gauge coupling constants , respectively .
the minimum of the potential in eq .
( [ eq : higgs potential ] ) is calculated in appendix [ sec : higgs ] . in this section
we assume the decoupling limit , where higgs bosons except for the standard - model like one are far heavier that the @xmath6 boson .
then the standard - model like higgs @xmath23 is given by @xmath24 where @xmath25 .
the potential of @xmath23 is given by @xmath26 where @xmath27 gev is the vev of the higgs @xmath23 determined by quadratic mass terms .
remembering that the mass of the @xmath6 boson is given by @xmath28 , the mass of @xmath23 , @xmath29 , is given by @xmath30 at the tree level , the mass of the standard - model like higgs is lighter than the mass of the @xmath6 boson , @xmath31 gev .
the tree level mass of the higgs is bounded from above because the quartic coupling of higgs @xmath32 is given by the @xmath5 term potential , whose value is restricted by susy .
susy breaking effects can raise the quartic coupling through quantum corrections . at the one - loop level ,
corrections are dominated by the following two effects .
one is the finite threshold correction involving stops , whose one - loop diagrams are shown in fig .
[ fig : threshold ] .
they correct the quartic coupling as @xcite @xmath33 where @xmath34 , @xmath35 , @xmath36 and @xmath37 are the top yukawa , the soft trilinear coupling of higgses and stops , the soft squared masses of the left - hand and right - hand stops , respectively .
another is the running of the quartic coupling . for simplicity
, we assume that susy particles have the same mass and hence the running is determined by the standard model interaction from the soft mass scale down to the electroweak scale , @xmath38,\end{aligned}\ ] ] where @xmath9 is the renormalization scale . the correction in eq .
( [ eq : lambda threshold ] ) should be added at the soft mass scale and the renormalization equation ( [ eq : lambda running ] ) should be solved with the corrected boundary condition at the soft mass scale . in fig .
[ fig : higgs mass ] , we show the higgs mass obtained from eqs .
( [ eq : lambda threshold ] ) and ( [ eq : lambda running ] ) with @xmath39 and @xmath40 . here
, we have also included the threshold correction at the electroweak scale summarized in ref .
the blue band shows the observed higgs mass , @xmath41 gev @xcite .
it can be seen that @xmath42 gev is obtained for sufficiently large @xmath43 .
for example , for @xmath44 , the observed higgs mass requires @xmath45 tev .
for the recent accurate calculation , see ref .
@xcite . .
the the blue band shows the observed higgs mass , @xmath41 gev @xcite . _ ] in most of dynamical susy breaking models ( see e.g. refs .
@xcite ) , susy breaking fields are charged under some symmetry or are composite fields .
we refer to this type of susy breaking as susy breaking without singlet " . in this subsection , we discuss the mass spectrum of susy particles expected in susy breaking without singlet .
we denote the susy breaking field as @xmath6 and assume the following simplest effective superpotential , @xmath46 where @xmath47 is the susy breaking scale and @xmath48 is a constant term .
have a charge under some symmetry , @xmath47 has a charge so that the superpotential is invariant under the symmetry .
that is , the susy breaking is associated with the breaking of the symmetry . ]
the gravitino mass is given by @xmath49 , where @xmath50 is the reduced planck mass .
vanishing of the cosmological constant requires that @xmath51 , where we have normalized @xmath6 so that it is canonical .
let us first discuss soft squared scalar mass terms of chiral multiplets @xmath52 in the mssm .
the susy breaking field @xmath6 in general couples to @xmath52 through the kahler potential , @xmath53 we assume that the susy breaking sector couples to the standard model sector only through planck scale suppressed interactions and hence @xmath54 is at largest @xmath1 .
the soft squared mass term of @xmath52 is given by @xmath55 thus , in general , the soft squared scalar mass term is as large as the gravitino mass . among mssm higgses ,
only the standard model - like higgs is light while other higgses have the same mass as large as the gravitino mass .
we note that generic @xmath54 without any structures induce flavor changing neutral currents , cp violations and lepton flavor violations . for discussions on these issues in the heavy sfermion scenario
, we refer to refs .
@xcite .
in order for the electroweak symmetry to be broken by vevs of mssm higgses , the @xmath9 term must be as large as or smaller than soft scalar mass terms of higgses ( see appendix [ sec : higgs ] ) .
the origin of the correct magnitude of the @xmath9 term is one of the key issues in susy model - building , which is dubbed as the @xmath9 problem " .
a trivial way to obtain the @xmath9 term is to assume that the combination @xmath56 has an @xmath3 charge of 2 while is neutral under other symmetries .
charge where the superpotential has an @xmath3 charge of 2 . ] then the @xmath9 term of any magnitude is allowed .
the natural value would be as large as the fundamental scale such as the plack scale and the gut scale .
indeed , this is the case with the minimal @xmath2 gut model @xcite . in the landscape point of view @xcite , the small @xmath9 term may be selected from the landscape by the anthropic principle @xcite .
@xmath57 is obtained by the supergravity effect . here , instead of readily adopting the anthropic principle
, we show two ways to naturally obtain a small @xmath9 term . in both cases ,
the @xmath9 term is forbidden by a symmetry and is given by the breaking of the symmetry .
problem in this way requires the gut group to be a product one @xcite .
] we do not discuss the origin of the smallness of the breaking scale in detail here .
the smallness is naturally explained , for example , if the breaking scale is generated by dimensional transmutation .
the first model assume the peccei - quinn ( pq ) symmetry @xcite .
the pq charge of mssm fields is given in tab .
[ tab : pq ] . with this charge assignment ,
yukawa couplings are allowed while the @xmath9 term is forbidden .
the @xmath9 term is provided by the following coupling to a pq breaking field @xmath58 with a pq charge @xmath59
@xcite , @xmath60 for @xmath61 , the @xmath9 term is of @xmath62 tev for the pq breaking scale of @xmath63 gev .
this range is consistent with the lower bound from the burst duration of sn1987a ( ref .
@xcite and references therein ) and the upper bound " from the cosmic abundance of the axion by an initial misalignment angle @xcite .
the @xmath18 term is given by the vev of the scalar component of the supergravity multiplet and the @xmath64 term of @xmath58 as large as @xmath65 ( see sec .
[ sec : axion ] ) .
both contributions result in the @xmath18 term of @xmath66 . if @xmath67 , @xmath18 is smaller than the soft squared masses of sfermions , and hence @xmath68 is large unless cancellation occurs ( see eq .
( [ eq : tanb ] ) )
. then the gravitino mass must be less than @xmath8 tev to explain the observed higgs mass . in this case , however , gaugino masses discussed later are too small that it is inconsistent with constraints from the lhc .
thus , the @xmath9 term must be as large as the gravitino mass .
.pq charge assignment of mssm fields [ cols="^,^,^,^,^",options="header " , ] chiral superfields are defined by the constraint , @xmath69 their components are defined by @xmath70 for a given superfield @xmath71 without local lorentz indices , the following superfield is a chiral superfield , @xmath72 real superfields are defined by the constraint , @xmath73 let us construct a super - diffeomorphism and super local lorentz invariant action in the super coordinate @xmath74 . under the super - diffeomorphism and the super local lorentz transformation , an einstein and lorentz scalar @xmath75 transforms as @xmath76 where @xmath77 parameterizes the super - diffeomorphism .
it can be seen that a mere integration of @xmath75 by @xmath78 does not yield an invariant action . in order to obtain the supergravity action , we consider the super determinant of the super vielbein , with bosonic and fermionic components , @xmath79 where latin indices and greek indices denote bosonic and fermionic indices .
the super determinant is defined by @xmath80 ] @xmath81 the transformation law of the super vielbein is given by @xmath82 then the transformation of @xmath83 is given by @xmath84 the product of an einstein and lorentz scalar @xmath75 and @xmath83 transforms as @xmath85 which is a total derivative .
hence , the super - diffeomorphism and the super local lorentz invariant action is given by @xmath86 where @xmath87 is an einstein and lorentz scalar .
the expression of the action in eq .
( [ eq : action real ] ) is not convenient for practical calculations .
one needs tedious calculations to expand given superfields by @xmath88 and @xmath89 .
calculations are simplified by constructing an action in a chiral representation .
a draw back is that a part of a manifest super - diffeomorphism and super local lorentz invariance are lost because we construct an action in a gauge - fixed form .
a manifest diffeomorphism , supergauge transformation , and local lorentz transformation invariance are maintained .
the chiral coordinate @xmath90 is a coordinate system such that the expansion of a chiral scalar field @xmath91 is given by @xmath92 then the supergravity transformation of chiral scalar fields is given by @xcite @xmath93 where the ellipses denote terms given by @xmath94 and the supergravity multiplet . here and hereafter , indices @xmath95 denote the chiral coordinate @xmath90 . in the flat limit ,
a global supergravity transformation ( @xmath96 ) in the chiral representation reduces to a global susy transformation in the chiral representation ( see eq .
( [ eq : susy trans chiral ] ) ) . in order to construct an action invariant under the supergravity transformation
, we define chiral densities .
chiral densities are functions of @xmath97 whose supergravity transformation is given by @xmath98.\end{aligned}\ ] ] a product of a chiral density @xmath99 and a chiral scalar @xmath91 is also a chiral density ; @xmath100.\end{aligned}\ ] ] then , a supergravity invariant action is given by @xmath101 in order to achieve the diffeomorphism invariance , we consider a chiral density whose lowest component is a determinant of the vielbein , @xmath102 .
such chiral density @xmath103 is constructed by comparing the supergravity transformation law of the supergravity multiplet with eq .
( [ eq : density sugra ] ) , @xmath104.\end{aligned}\ ] ] note that @xmath103 is scalar under the local lorentz transformation .
apparently , the diffeomorphism of @xmath103 and @xmath91 is given by @xmath105,~~ \delta \phi = - \eta^m\partial_m \phi,\nonumber \\ \eta^m ( z ) & = & \eta^m ( x),~~ \eta^\alpha(z)=0.\end{aligned}\ ] ] then , an action invariant under the diffeomorphism , the local lorentz transformation , and the supergravity transformation is given by @xmath106 for example , the einstein gravity and the kinetic term of the gravitino are given by @xmath107 the kinetic term of a chiral scalar multiplet @xmath91 is given by @xmath108
in this section , we summarize the super - weyl transformation law of various superfields and their components .
the super - weyl transformation of the vielbein and the connection is the transformation of them such that @xmath109 and the torsion constraint in eq .
( [ eq : torsion constraint ] ) is preserved .
the super - weyl transformation of the vielbein is given by @xmath110 where @xmath111 is an arbitrary chiral multiplet .
the super - weyl transformation of the connection is given by @xmath112 the super - weyl transformation of the gravity multiplet is obtained by calculating the transformation of torsions from eqs .
( [ eq : sw vielbein ] ) and ( [ eq : sw connection ] ) .
the transformations of the superspace curvature @xmath3 and the superspace vector @xmath113 are given by @xmath114 by taking the lowest component of eqs .
( [ eq : sw vielbein ] ) and ( [ eq : sw rg ] ) , we obtain the transformation law of component fields , @xmath115 where @xmath116 , @xmath117 , @xmath118 and @xmath119 are the vielbein , the gravitino , the auxiliary scalar , and the auxiliary vector , respectively . @xmath120 denotes the lowest component of a superfield @xmath71 .
the super - weyl transformation of a chiral multiplet @xmath121 is defined by @xmath122 where @xmath123 is the weyl weight of @xmath121 . by remembering the definition of component fields , @xmath124 and the transformation law of the connection in eq .
( [ eq : sw connection ] ) , we obtain @xmath125 the super - weyl transformation of a vector multiplet @xmath126 is defined by @xmath127 where @xmath128 is the weyl weight of @xmath126 .
then the transformation of the chiral projection of @xmath126 is given by @xmath129 = \left ( { \cal d}^{\dag 2 } - 8r \right ) \left [ ( w'-4 ) + ( w'+2 ) \sigma^\dag \right ] u\end{aligned}\ ] ] the super - weyl transformation of a gauge multiplet @xmath87 is defined by @xmath130 then the superspace fields strength defined by @xmath131 transforms as @xmath132 from the definition of component fields , @xmath133,\end{aligned}\ ] ] and eq .
( [ eq : sw connection ] ) , we obtain @xmath134 we may also consider the super - weyl transformation of chiral superfields with fixing the definition of the chiral coordinate @xmath135 while transforming components fields according to eqs .
( [ eq : sw sugra ] ) , ( [ eq : sw chiral ] ) and ( [ eq : sw gauge ] ) .
this transformation is convenient when the supergravity action is given in a chiral coordinate .
the expansion of the chiral density @xmath103 is given by @xmath136 we define the transformed chiral density @xmath137 by @xmath138 where @xmath139 for a component field @xmath71 . then the transformation of @xmath103 is written as @xmath140 in the similar way , we obtain transformation laws of @xmath3 , @xmath121 and the chiral projection of @xmath126 as @xmath141 & = & \left ( { \cal d}^{\dag 2 } - 8r \right ) \left [ ( w'-4 ) + ( w'+2 ) \sigma^\dag \right ] u - s^\alpha \frac{\partial}{\partial \theta^\alpha } \left [ \left({\cal d}^{\dag 2 } - 8r \right ) u \right ] . \nonumber\end{aligned}\ ] ] the transformation given in eqs .
( [ eq : chiral sw1 ] ) and ( [ eq : chiral sw2 ] ) is the one we use in sec .
[ sec : amsb ] .
in this appendix , we show that the measure given in eq .
( [ eq : sd measure ] ) is invariant under the super - diffeomorphism .
, the super - diffeomorphism " given in eqs .
( [ eq : sdiff ] ) and ( [ eq : sdiff2 ] ) is not the full super - diffeomorphism , but a part of it . ] before discussing supergravity , we construct the diffeomorphism invariant measure of a real scalar field @xmath142 and a weyl fermion @xmath143 @xcite . under an infinitesimal diffeomorphism , @xmath144 , @xmath145 and a vielbein @xmath116 transforms as @xmath146 where @xmath147 parameterize the diffeomorphism .
then the determinant of the vielbein @xmath148 transforms as @xmath149 let us first discuss the measure of the scalar @xmath142 . in order to define the path - integral measure
, we consider a complete set @xmath150 normalized as @xmath151 the following field , @xmath152 transforms under the diffeomorphism as @xmath153 we expand @xmath154 by @xmath150 , @xmath155 and define the path - integral measure by @xmath156 \equiv \prod_n { \rm d}a_n.\end{aligned}\ ] ] let us calculate the jacobian in the transformation of the measure . @xmath157 and @xmath158 are related by @xmath159\nonumber\\ & = & \left [ \delta_{nk } - \int { \rm d}^4 x \phi_n(x ) \left\ { \xi^m(x ) \partial_m + \frac{1}{2 } \left ( \partial_m \xi^m \left(x\right ) \right ) \right\}\phi_k(x ) \right]a_k \nonumber \\ & \equiv & ( \delta_{nk } + m_{nk } ) a_k.\end{aligned}\ ] ] then the jacobian is given by @xmath160 in the last line , we have assumed that @xmath161 vanishes sufficiently fast at infinity .
we have shown that the measure given in eq .
( [ eq : measure scalar ] ) is invariant under the diffeomorphism . for later convenience
, we show the invariance of the measure in a slightly differrent way .
the transformation of the measure in eq .
( [ eq : measure scalar ] ) is given by @xmath162 & = & \prod_x [ d \tilde{\phi}'(x ) ] = \prod_{x } [ d \tilde{\phi}(x ) ] { \rm det}_{x ' x '' } \frac{\delta \tilde{\phi}'(x '' ) } { \delta \tilde{\phi } ( x ' ) } \nonumber \\ & = & [ d \tilde{\phi } ] { \rm det}_{x ' x } \left [ 1 - \xi^m(x ) \partial_m - \frac{1}{2 } \partial_m \xi^m(x ) \right ] \delta^4 \left(x'-x\right)\nonumber \\ & = & [ d \tilde{\phi } ] j , \nonumber \\ { \rm ln}j & = & -{\rm tr}_{x'x } \left [ \xi^m(x )
\partial_m + \frac{1}{2 } ( \partial_m \xi^m(x ) ) \right]\delta^4 \left(x'-x\right)\nonumber \\ & = & - \int { \rm d}^4 x { \rm d}^4 x ' \delta^4 \left(x'-x\right ) \left [ \xi^m(x ) \partial_m + \frac{1}{2 } ( \partial_m \xi^m(x ) ) \right]\delta^4 \left(x'-x\right).\end{aligned}\ ] ] the product of delta functions is handled by the completeness relation , @xmath163 by replacing one of the delta function in eq .
( [ eq : functional j ] ) , we obtain @xmath164 .
similarly , the diffeomorphism invariant measure of the weyl fermion @xmath145 is given by @xmath165[d\tilde{\psi}^\dag],\\ \tilde\psi ( x ) \equiv e^{1/2}(x ) \psi(x).\end{aligned}\ ] ] the diffeomorphism invariance of the measure can be proven in the same way as the case of the scalar .
a crucial difference is that the cancellation like that in eq .
( [ eq : expansion j ] ) occurs between the contribution from @xmath166 $ ] and @xmath167 $ ] .
finally , we show the diffeomorphism invariance of the measure @xmath168 $ ] from a very definition of the path - integral measure .
we define a path - integral measure of scalar fields by @xmath169 { \rm exp}\left [ \frac{1}{2 } i a \int { \rm d}^4x c(x)^2 \right ] = n,\end{aligned}\ ] ] where @xmath170 and @xmath171 are constants
. then the measure @xmath172 $ ] satisfies @xmath173 { \rm exp}\left [ \frac{1}{2 } i a \int { \rm d}^4x \tilde{\phi}(x)^2 \right ] = \int [ d \tilde{\phi } ] { \rm exp}\left [ \frac{1}{2 } i a \int { \rm d}^4x e \phi(x)^2 \right ] = n.\end{aligned}\ ] ] since the exponent as well as @xmath171 are diffeomorphism invariant , the measure @xmath172 $ ] is also diffeomorphism invariant .
next , let us discuss the super - diffeomorphism invariance of the measure in eq .
( [ eq : sd measure ] ) , @xmath174 = [ d\left(2{\cal e}\right)^{1/2}q].\end{aligned}\ ] ] under the transformation given in eq . ( [ eq : sdiff ] ) , @xmath175 the variable @xmath176 transforms as @xmath177 then , the path - integral measure @xmath178 $ ] transforms as ( see eq .
( [ eq : functional j ] ) ) , @xmath179 & = & [ dq_{\rm diff}]\times { \rm exp}\left[\,{\rm str}_{z',z}\,{\cal o}(z',z)\,\right]\ , \\
\label{eq : ozz } { \cal o}(z',z ) & \equiv & - \left [ \eta^m\partial_m + \frac{1}{2 } ( -1)^m \left(\partial_m\eta^m\right ) \right]\delta^6(z'-z)\ , \\ \label{eq : str } { \rm str}\,{\cal o}(z',z ) & = & \int { \rm d}^6z { \rm d}^6z ' \delta^6 ( z'-z ) { \cal o}(z',z)\ .\end{aligned}\ ] ] where we have collectively represented @xmath180 and @xmath181 by @xmath182 .
a naive conclusion is that the super - trace vanishes due to the saturation of grassmann variables @xmath181 and @xmath183 from the delta functions in eqs.([eq : ozz ] ) and ( [ eq : str ] ) .
however , since there is also a factor of @xmath184 , which is well - defined only after integrating over @xmath180 or @xmath185 , one should carefully investigate the integration . to examine the the integration ,
let us expand the delta function by plane waves , @xmath186 by substituting this expression into eq.([eq : str ] ) , the above super - trace is expressed by , @xmath187 \psi_{k,\tau}(z)\ .\end{aligned}\ ] ] now , let us notice an identity , @xmath188 \psi_{k,\eta}(z)&= & \frac{1}{2}(-)^m \int { \rm d}^6 z \partial_m \left[\psi_{k,\eta}\left(z\right)\eta^m\psi_{k,\eta}\left(z\right)\right]\nonumber\\ & = & 0\ , \end{aligned}\ ] ] where we have used the property that an integration of a total derivative vanishes . by using this identity several times ,
the super - trace can be rearranged as @xmath189 \left(\psi_{k,\tau}\left(z\right)+\psi_{-k,-\tau}\left(z\right)\right)\nonumber\\ & = & -\frac{1}{4}(-)^m\int { \rm d}^6z \int \frac{{\rm d}^4 k}{(2\pi)^4 } d^2 \tau \partial_m \left [ \left(\psi_{k,\tau}\left(z\right)+\psi_{-k,-\tau}\left(z\right)\right ) \eta^m \left(\psi_{k,\tau}\left(z\right)+\psi_{-k,-\tau}\left(z\right)\right ) \right]\nonumber\\ & = & 0\ .\end{aligned}\ ] ] this shows that the measure given in eq .
( [ eq : sd measure ] ) is actually invariant under the super - diffeomorphism .
it should be noted that the transformation law in eq.([eq : qdiffsd ] ) is crucial to use eq.([eq : identity ] ) , and hence , the super - diffeomorphism invariance does not hold for measures with different weights @xmath190 $ ] @xmath191 .
in fact , the super - diffeomorphism transformation of @xmath190 $ ] @xmath191 is accompanied by konishi - sizuya anomaly @xcite .
this argument provides an superfield expression of the arguments in ref .
@xcite reviewed in the provious subsection .
as is the case with the diffeomorphism invariance , there is a quicker route to show the super - diffeomorphisim invariance of @xmath178 $ ] .
the path - integral measure of superfields in the chiral superspace is defined by @xmath192 { \rm exp } \left [ \frac{i}{2 } a\int { \rm d}^6 z c^2 \right ] = n\ , \end{aligned}\ ] ] where @xmath170 and @xmath171 are normalization constants .
then the measure @xmath193 $ ] satisfies @xmath194 { \rm exp } \left [ \frac{i}{2 } \int { \rm d}^6 z q_{\rm diff}^2 \right ] = \int [ d q_{\rm diff } ] { \rm exp } \left [ \frac{i}{2 } \int { \rm d}^6 z\ , 2{\cal e}\ , q q \right ] = n\ .\end{aligned}\ ] ] now , since @xmath195 is invariant under the super - diffeomorphism as @xmath121 is a chiral scalar multiplet , so is the path - integral measure @xmath193=[d(2{\cal e})^{1/2}q]$ ] .
in fact , under the super - diffeomorphism , @xmath196 we have the following identities @xmath197 { \rm exp } \left [ \frac{i}{2 } \int { \rm d}^6 z\ , 2{\cal e } \,q q \right]\ , \nonumber\\ & = & \int
[ d \left(2 { \cal e'}\right)^{1/2 } q ' ] { \rm exp } \left [ \frac{i}{2 } \int { \rm d}^6 z \,2{\cal e'}\ , q ' q ' \right ] \ , \nonumber\\ & = & \int [ d \left(2 { \cal e'}\right)^{1/2 } q ' ] { \rm exp } \left [ \frac{i}{2 } \int { \rm d}^6z \,2{\cal e}\ ,
q q \right]\ .\end{aligned}\ ] ] here , the second equality is just a change of variable .
we have used a super - diffeomorphism invariance of the exponent in the third equality .
thus , from these identities , we find that @xmath198 = d [ \left(2 { \cal e}\right)^{1/2 } q ] \ , \end{aligned}\ ] ] which again shows the super - diffeomorphism invariance of the measure @xmath178 $ ] . in the same token
, we can derive the super - diffeomorphism invariance of the measure of a scalar multiplet @xmath87 in a real superspace , @xmath199 = [ de^{1/2 } v]\ .\end{aligned}\ ] ]
in this section , we discuss the anomaly mediated gaugino mass when susy is broken by @xmath64 terms of singlet chiral fields .
we start from the supergravity action with the super - weyl compensator @xmath200 , @xmath201 + \frac{1}{16g^2}\int { \rm d}^2 \theta\ , 2{\cal e}\ , w^\alpha w_\alpha + { \rm h.c.}.\end{aligned}\ ] ] here , we take the unit with @xmath202 .
we parametrize the components of @xmath200 as @xmath203 in the chiral coordinate .
note that these components are gauge degree of freedoms , and we can freely choose them .
@xmath144 can be used to choose the frame of gravity theory .
the most convenient choice is , @xmath204,\end{aligned}\ ] ] which yields the einstein frame gravity .
@xmath205 can be chosen in order to remove mixings between the gravitino and chiral matter fields . in order to find a convenient choice for @xmath64 ,
let us solve the equation of motion of the auxiliary component of the gravity multiplet and the chiral matter multiplets , @xmath118 and @xmath206 . from eqs .
( [ eq : expansion ] ) , ( [ eq : chiral projection ] ) and ( [ eq : sw classical action2 ] ) , lagrangian terms which depend on @xmath118 , @xmath206 and @xmath64 are given by @xmath207\nonumber\\ & & + \left[3\phi^2 f w + \phi^3 w_i f^i -m^\dag \phi^3 w + { \rm h.c.}\right]\end{aligned}\ ] ] by solving the equation of motion of @xmath118 and @xmath206 , we obtain @xmath208\end{aligned}\ ] ] in this formulation , it is not necessary to solve the equation of motion of @xmath64 , because it is a gauge degree of freedom .
we take a gauge in which @xmath209 , that is , @xmath210 this choice simplifies calculation and hence is frequently adopted in the literature . in the gauge where @xmath211 and @xmath209 , @xmath206 and @xmath64 are given by @xmath212,\\ \label{eq : compensator f } f & = & e^{k/6 } \left [ e^{k/2 } w^\dag + \frac{1}{3}k_if^i \right ] = e^{k/6 } \left [ m_{3/2 } + \frac{1}{3}k_if^i \right ] .\end{aligned}\ ] ] we adopt this gauge in the following .
note that the lagrangian in eq .
( [ eq : sw classical action2 ] ) possesses the following formal super - weyl symmetry , @xmath213 this symmetry must be anomaly free .
otherwise , one can not gauge away the super - weyl compensator @xmath200 to come back to the original theory without @xmath200 . to see the anomaly of the formal super - weyl symmetry ,
let us consider the @xmath214 part , @xmath215 where @xmath216 is a real constant .
fermion parts of the chiral multiplet , @xmath217 , and the gaugino @xmath32 transforms as @xmath218 this transformation is anomalous ; @xmath219 [ d\chi^{\bar{i}\dag } ] [ d \lambda ] [ d \lambda^\dag ] = [ d\chi^{'i } ] [ d\chi^{'\bar{i}\dag } ] [ d \lambda ' ] [ d \lambda^{'\dag } ] e^{i\delta s_{\rm ano } } , \nonumber \\ \delta s_{\rm ano } = \alpha\int { \rm d}^4 x\frac{1}{32\pi^2 } \epsilon^{abcd}f_{ab}^af_{cd}^a \times3\left(t_g -\sum_i t_i\right),\end{aligned}\ ] ] where @xmath220 and @xmath221 are the dynkin index of the adjoint representation and the chiral multiplet @xmath52 , respectively . in the superspace ,
( [ eq : ranomaly ] ) is expressed as @xmath222 this anomaly is cancelled by adding a counter term @xmath223 whose variation under the @xmath214 symmetry cancels with eq .
( [ eq : superanomaly ] ) .
finally , let us separate the chiral multiplets from the super - weyl compensator by a transformation @xmath224 this transformation is anomalous , @xmath225 [ d q^{i \dag } ] = [ d q_s^i ] [ d q_s^{i \dag } ] e^{i \delta s_c } , \nonumber \\ \delta s_s = \frac{1}{128\pi^2}\int { \rm d}^4 x{\rm d}^2 \theta 2 { \cal e } { \rm ln}c w^{a\alpha } w^a_\alpha \sum_i t_i + { \rm h.c.}.\end{aligned}\ ] ] adding eqs .
( [ eq : local counter term ] ) and ( [ eq : cano anomaly ] ) , we obtain @xmath226 from eq .
( [ eq : compensator f ] ) and ( [ eq : compensator ] ) , the gaugino mass from the @xmath64 term of the super - weyl compensator is given by @xmath227 .\end{aligned}\ ] ] when susy is broken by @xmath64 terms of singlet chiral fields , @xmath228 in general .
let us consider the coupling of @xmath64 terms of susy breaking fields with @xmath229 in the kahler potential .
we denote the susy breaking fields and their @xmath64 terms as @xmath230 and @xmath231 . from eq .
( [ eq : sw classical action2 ] ) , couplings between the @xmath64 terms and @xmath229 are given by @xmath232.\end{aligned}\ ] ] here , @xmath52 is a chiral multiplet while @xmath233 , @xmath234 and @xmath235 are lowest components of the corresponding superfields .
then canonically normalized fields are given by @xmath236,\nonumber\\ k_{i\bar{j } } & \equiv & { u_i}^k c_k \delta_{k\bar{l}}{u^{\dag \bar{l}}}_{\bar{j}},\end{aligned}\ ] ] where @xmath126 is the unitary matrix and @xmath237 are positive real constant .
canonicallization induces the @xmath64 term of the gauge kinetic function , @xmath238 & = & \prod_i [ d q_c^i ] \times e^{i \delta s_c } , \\
\delta s_c & = & \frac{1}{128\pi^2}\int { \rm d}^4x{\rm d}^2 \theta 2 { \cal e } \left [ - \frac{1}{3 } k_i f^i \sum_i t_i \theta^2 + \sum_it_i ( { \rm ln~det } k_{i\bar{j}})_i f^i \theta^2 \right ] w^{a\alpha } w^a_\alpha+ { \rm h.c.}.\nonumber\end{aligned}\ ] ] here , we have used the identity @xmath239 = { \rm tr}{\rm ln}_{kl } \left [ { \delta_k}^l + c_k^{-1/2 } { u_k^{\dag}}^i k_{i\bar{j}i } f^i u^{\bar{j}\bar{l } } c_l^{-1/2 } \theta^2 \right ] \nonumber \\
= { \rm tr}_{kl } \left [ c_k^{-1/2 } { u_k^{\dag}}^i k_{i\bar{j}i } f^i u^{\bar{j}\bar{l } } c_l^{-1/2 } \theta^2 \right]= ( { \rm tr}~{\rm ln } k_{i\bar{j}})_i f^i \theta^2 = ( { \rm ln } ~{\rm det } k_{i\bar{j}})_i f^i \theta^2.\end{aligned}\ ] ] then the gaugino mass is given by @xmath240 + we note that it is crucial to start from the path - integral measure @xmath241 $ ] .
if one starts from measures @xmath242 $ ] with a generic function @xmath243 , one obtains different results from eq .
( [ eq : gaugino2 ] ) . assuming the measure @xmath241 $ ] , adding eqs .
( [ eq : gaugino1 ] ) and ( [ eq : gaugino2 ] ) , we obtain @xmath244 which is consistent with ref . @xcite .
in this section , we derive the leading order annihilation cross section of the gluino used in sec .
[ subsec : bino - gluino ] .
we formulate the sommerfeld effect with the non - relativistic effective theory of the gluino .
let us first derive a non - relativistic effective actin of the gluino .
we assume that squarks are heavy enough that they do not affect the interaction of the gluino except for its decay .
the interaction of the gluino @xmath245 with the gluon @xmath246 is described by the action , @xmath247,\ ] ] where @xmath248 and @xmath118 and @xmath249 are the mass of the gluino and the @xmath250 gauge coupling constant , respectively . here
, we have taken the feynman - t hooft gauge and the gluino is expressed as a majorana field .
ghost fields are irrelevant for the leading order calculation . in the calculation of the annihilation of the gluino ,
the time - like degree of freedom of the gluino field and the soft degree of freedom of the gluon field do not appear in external lines . by integrating them out ,
we obtain @xmath251 \nonumber \\ & + & \int { \rm d}^4x{\rm d}^4y\big [ -\frac{i}{2}g_s^2f^{abc}f^{ade}\overline{\widetilde{g}}^d(y)\slashed{g}^e(y)s_{f}(y - x)\slashed{a}^b(x)\widetilde{g}^c(x ) \nonumber \\ & & -\frac{i}{8}g_s^2f^{abc}f^{ade}\overline{\widetilde{g}}^d(y)\gamma_{\mu}\widetilde{g}^e(y)d_{f}^{\mu\nu}(y - x)\overline{\widetilde{g}}^b(x)\gamma_{\nu}\widetilde{g}^c(x ) \big],\end{aligned}\ ] ] where @xmath252 in the dirac representation , the non - relativistic gluino field @xmath205 is defined by @xmath253 where the charge conjugation is defined by @xmath254 by substituting eq .
( [ eq : nrgluino ] ) into eq .
( [ eq : raction ] ) , we obtain @xmath255 \nonumber \\ &
+ & \int { \rm d}t{\rm d}^3{\bf x}{\rm d}^3{\bf y}\frac{-\alpha_3({\bf x}-{\bf y})}{\| { \bf x}-{\bf y}\|^3}e^{-m{\| { \bf x}-{\bf y } \|}}\left(1+m{\| { \bf x}-{\bf y } \|}\right)\frac{i}{2}f^{abc}f^{ade}g^{ei}({\bf x},t)g^{bj}({\bf y},t)\nonumber \\ & & \times\big[i\epsilon^{ikj}e^{2imt}\chi^{d\dag}({\bf x},t)(\chi^c)^c({\bf y},t)+i\epsilon^{ikj}e^{-2imt}(\chi^d)^{c\dag}({\bf
x},t)\chi^c({\bf y},t ) \nonumber \\ & & + e^{2imt}\chi^{d\dag}({\bf x},t)(\delta^{kj}\sigma^i+\delta^{ki}\sigma^j-\delta^{ij}\sigma^k)(\chi^c)^c({\bf y},t ) \nonumber \\ & & + e^{-2imt}(\chi^d)^{c\dag}({\bf x},t)(\delta^{kj}\sigma^i+\delta^{ki}\sigma^j-\delta^{ij}\sigma^k)\chi^c({\bf y},t)\big ] \nonumber \\ & + & \int { \rm d}t{\rm
d}^3{\bf x}{\rm d}^3{\bf y}\frac{1}{4}f^{abc}f^{ade}\frac{\alpha_3}{\| { \bf x}-{\bf y } \|}\\ & & \big[\chi^{b\dag}({\bf x},t)(\chi^d)^c({\bf y},t)(\chi^c)^{c\dag}({\bf x},t)\chi^e({\bf y},t ) -\chi^{b\dag}({\bf x},t)\sigma^i(\chi^d)^c({\bf y},t)(\chi^c)^{c\dag}({\bf x},t)\sigma^i\chi^e({\bf y},t)\big].\nonumber\end{aligned}\ ] ] let us replace a pair of gluinos with a two body state field .
we introduce auxiliary fields by inserting the identity , @xmath256[ds_r^{\mu\alpha_r\dag}]{\rm exp}\big[\frac{i}{2}\int { \rm d}t{\rm d}^3{\bf x}{\rm d}^3{\bf
y}\phi_r^{\mu\alpha_r}({\bf x},{\bf y},t ) \nonumber \\ & \times\left(s_r^{\mu\alpha_r\dag}({\bf y},{\bf x},t)-c^{\alpha_r}_{ab}\chi^{a\dag}({\bf x},t)\sigma^{\mu}(\chi^b)^c({\bf y},t)\right)\big],\end{aligned}\ ] ] and its conjugate .
here , @xmath3 indicates representations of @xmath250 and @xmath257 denotes indices of the representation @xmath3 .
@xmath258 are the clebsch - gordan coefficient for a decomposition @xmath259 . by integrating out @xmath260 , and their conjugate
, we obtain the effective action , @xmath261 \nonumber \\ & + \sum_{r,\alpha_r,\mu}\int { \rm d}^3{\bf r}{\rm d}^4x\big [ \phi_r^{\mu\alpha_r\dag}({\bf r},x)(i\partial_{t}+\frac{{\bf \nabla}^2_x}{4 m } + \frac{{\bf \nabla}^2_r}{m } -v_r({\bf r } ) ) \phi_r^{\mu\alpha_r}({\bf r},x ) \nonumber \\ & + \phi_r^{\mu\alpha_r\dag}({\bf r},x)d_r^{\mu\alpha}({\bf r},x ) + \phi_r^{\mu\alpha_r}({\bf r},x)d_r^{\mu\alpha\dag}({\bf r},x)\big],\end{aligned}\ ] ] where @xmath262 .\nonumber \end{aligned}\ ] ] @xmath263 and @xmath264 correspond to a pair of gluinos with a spin @xmath265 and @xmath266 , respectively .
accodring to landau - yang s theorem @xcite , a spin - one state can not decay into two gluons .
therefore , the matrix element calculated from the second term in eq .
( [ eq : spinone ] ) should vanish if the two gluons are on - shell .
let us take the annihilation of the gluino into account .
since we consider non - relativistic processes , the annihilation is dominated by s - waves and hence the annihilation is expressed by a local four fermi term , @xmath267 where @xmath268 . and the bilinear should be replaced as @xmath269 , where @xmath270 is the field which creates an anti - particle . ] with two body state fields , the fermi term is expressed as @xmath271 let us calculate @xmath272 by matching tree level calculations of the annihilation cross section of non - relativistic gluinos with forward scattering amplitudes calculated from eq .
( [ eq : fermi ] ) , with an aid of the optical theorem . for spin-0 states ,
annihilation cross sections are given by @xmath273 while those of spin-1 states are given by @xmath274 note that spin-1 states do not annihilate into two gluinos , which is consistent with the landau - yang s theorem .
annihilation cross sections are related with forward scattering amplitudes by the optical theorem for two body scatterings : @xmath275 where @xmath276 is the center of mass energy . here , @xmath277 is the invariant matrix element normalized by @xmath278 with the lorentz invariant normalization of one - particle states , @xmath279 with this normalization , the invariant amplitude calculated by the action in eq .
( [ eq : fermi ] ) is given by @xmath280 hence , @xmath272 is given by @xmath281 specifically , @xmath282 non - relativistic annihilation processes are dominated by s - waves .
thus , it is useful to extract an s - wave component of a given initial state . for that purpose ,
we insert a complete set of asymptotic two body s - wave states , @xmath283 where @xmath284 and @xmath285 denote the total momentum and the relative momentum of the two body state .
@xmath32 labels the discrete quantum number of the two body state , such as the color and the spin .
@xmath285 is related with the relative velocity @xmath286 by @xmath287 . in the following ,
we omit the index @xmath32 to simplify notation . by inserting the complete set , the forward scattering amplitude of a state @xmath288 by its s - wave component
is given by @xmath289 the state @xmath290 is constructed by @xmath291 where @xmath292 is the creation operator of a particle with a momentum @xmath293 and a quantum number @xmath216 .
@xmath294 is a coefficient to construct the quantum number @xmath32 , which is normalized by @xmath295 . with a normalization @xmath296
it can be shown that the normalization in eq .
( [ eq : normalization ] ) holds .
let us consider the two - body state in the center - of - momentum system ; @xmath297 .
the state is normalized in the lorentz invariant way and has a overlap with the state @xmath298 , @xmath299 by remembering that the total energy of the initial state is given by @xmath300 , we obtain @xmath301 hence , the forward scattering amplitude of the state @xmath288 is given by @xmath302 where @xmath303 .
next , let us relate @xmath277 with the green s function of the two body state fields , @xmath304 the relation is given by the lehmann - symanzik - zimmermann ( lsz ) reduction formula @xcite we derive in this subsection .
we start from the effective lagrangian , @xmath305 as an asymptotic initial state , we consider the state with two free gluinos . therefore , to derive the lsz reduction formula ,
we first expand @xmath144 by ignoring the potential and the annihilation terms .
further , since we consider only the s - wave state , we reduce @xmath144 as @xmath306 with this reduction , the effective lagrangian is given by @xmath307 we define the annihilation operator by @xmath308 the conjugate momentum and the canonical quantization condition are given by @xmath309 | _ { x^0=y^0 } = i 2\pi \delta ( k - k ' ) \delta^{(3)}({\bf x}-{\bf y}),\end{aligned}\ ] ] which requires @xmath310 = ( 2\pi)^4 \delta^{(3)}({\bf p}-{\bf p}')\delta ( k - k').\end{aligned}\ ] ] it can be easily shown that the state , @xmath311 ) .
then the matrix element of the field @xmath312 between the vacuum and the asymptotic two gluon free state is given by @xmath313 with the exactly same procedure as the derivation of the lsz formula for relativistic field theories , one can derive a formula , @xmath314 which yields @xmath315 by remembering the inverse of the reduction in eq .
( [ eq : reduction ] ) , @xmath316 we obtain , @xmath317 from eq .
( [ eq : def amp ] ) and ( [ eq : lsz formula ] ) , @xmath318 is given by @xmath319 with the contribution from the free propagation subtracted . finally , with the optical theorem
, we obtain @xmath320,\nonumber\\ g^{\mu
r}(e;{\bf r},{\bf r}')&= & i\int { \rm d}^4 x e^{iex^0}<t \phi_r^{\mu \alpha_r}({\bf r},x)\phi_r^{\mu \alpha_r}({\bf r}',0)^\dag>,\end{aligned}\ ] ] where we have denoted the @xmath321 dependence explicitly .
let us calculate the green s function of the gluino pairs : @xmath322 where we have again omitted indices @xmath323 and @xmath257 . from the action of @xmath144 given in eq .
( [ eq : action phi ] ) , the green s function satisfies @xmath324 as we have mentioned , we consider only the s - wave state since it dominantly contributes to the annihilation of the gluino .
for the s - wave , eq .
( [ eq : sch ] ) becomes @xmath325 with @xmath326 , eq .
( [ eq : ssch ] ) becomes @xmath327 let us first neglect @xmath328 and solve eq .
( [ eq : ssch simp ] ) analytically .
we denote the solution as @xmath329 .
@xmath328 is taken into account as a perturbation later . for a dimensionless variable @xmath330 , eq .
( [ eq : ssch simp ] ) with @xmath331 is @xmath332 where @xmath333 and @xmath334 .
the solution for eq .
( [ eq : ssch dimless ] ) is expressed as @xcite @xmath335 note that the function @xmath336 is not yet fixed .
since @xmath337 corresponds to an out going wave sourced by the delta function , we put a condition that @xmath336 contains only an out going modes @xmath338 at @xmath339 . prescription .
if an incoming mode exists at @xmath339 , the green s function diverges at infinity due to the small imaginary part in @xmath340 , @xmath341 .
] then the solution for @xmath337 is given by @xmath342 where w is the whittaker function .
for @xmath339 , @xmath336 behaves as @xmath343 to the first order in @xmath328 , @xmath19 is given by @xmath344 let us note subtlety read off from the asymptotic form of the green s function .
as can be seen from eq .
( [ eq : asymptotic ] ) , for @xmath339 , the green s function have a non - trivial phase @xmath345 $ ] , in addition to a trivial phase @xmath346 which corresponds to free particles .
this means that our assumption of free incident particles is not good .
the non - trivial phase appears due to a long - range force mediated by gluons . in thermal bath in the early universe ,
however , gluons obtain their thermal masses and hence the long - range force is screened .
thus , the non - trivial phase should vanish once thermal effects are taken into account . in the following
, we simply drop the non - trivial phase in eq .
( [ eq : asymptotic ] ) . then , the wave function of gluinos at infinity is simply enhanced ( or declined ) by a constant @xmath170 , in comparison with the trivial one , @xmath347 .
we are now at the point to calculate the annihilation cross section of the gluino . from eqs .
( [ eq : xsection ] ) and ( [ eq : green perturbation ] ) , we obtain @xmath348 ^ 2,\end{aligned}\ ] ] where we have omitted indices @xmath349 . here , we have subtracted a term proportional to @xmath329 which corresponds a scattering between gluinos .
let us evaluate the factor in the parenthesis .
note that the factor vanishes unless the integration diverges .
thus , we may replace the integrand with its asymptotic form for @xmath350 given in eq .
( [ eq : asymptotic ] ) . observing that @xmath351 we find @xmath352.\end{aligned}\ ] ]
the cross section given in eq .
( [ eq : xsection all ] ) includes not only the annihilation into gluons and quarks , but also includes the scattering into a pair of gluinos .
let us extract the former contribution in a intuitive way .
as we have mentioned , the factor @xmath170 expresses the enhancement ( or decline ) of the wave function of gluinos by a constant , @xmath353 .
the factor @xmath170 would enter the annihilation amplitude simply as a multiplication factor .
then , the annihilation cross section into gluinos and quarks is given by @xmath354 in the end , we obtain @xmath355 here , @xmath356 is the fine structure constant of the qcd evaluated at the scale @xmath357 which is determined by @xcite @xmath358 a spin and color averaged annihilation total annihilation cross section is given by @xmath359\nonumber\\ & \simeq & \frac{81\pi}{16 } \frac{\alpha_{s,8}}{v } \times \frac{\pi \alpha_{s,{\rm uv}}^2}{m^2 } = \frac{18\pi}{7 } \frac{\alpha_{s,8}}{v } \sigma v _ { \rm tot , tree},\end{aligned}\ ] ] where @xmath360 . here , we have used the approximation that @xmath361 and neglected the contribution from @xmath362 representation .
this approximation induces only an error of one percent . in fig .
[ fig : enhancement ] , we show the enhancement factor @xmath363 . .
here , we summarize annihilation cross sections of the wino . as already mentioned in main text , there are six annihilation modes : @xmath364 , @xmath365 , @xmath366 , and @xmath367 . initial states of @xmath364 and @xmath367 form only spin-0 states , while those of other modes form both spin-0 and spin-1 states .
below , we carefully present the cross sections in each mode . since this is the annihilation between identical particles , its initial sate forms only a spin-0 state , and it annihilates into @xmath369 pair .
the cross section shown here can also be applied to its conjugate case , namely @xmath370 .
@xmath371^{3/2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } \right]^{-2}.\end{aligned}\ ] ] cross sections presented here can also be applied their conjugate cases .
when the initial state forms a spin-0 state , it annihilates into @xmath373 and @xmath374 .
@xmath375^{3/2 } \left [ 1 - \frac{m_w^2+m_z^2}{4m_2 ^ 2 } \right]^{-2 } , \\
\sigma_0 v|_{w\gamma } & = & \frac{2\pi \alpha_2 ^ 2 s_w^2}{m_2 ^ 2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } + \frac{m_w^4}{16m_2 ^ 4 } \right]^{3/2 } \left [ 1 - \frac{m_w^2}{4m_2 ^ 2 } \right]^{-2}.\end{aligned}\ ] ] on the other hand , when the initial state forms a spin-1 state , it annihilates into @xmath376 , @xmath377 , and @xmath373 . those cross sections are given as follows .
below , we only show the cross section of @xmath378 as a representative of @xmath379 .
@xmath380^{-2 } \left [ 1 - \frac{m_e^2}{4m_2 ^ 2 } \right]^2 \left [ 1 + \frac{m_e^2}{8m_2 ^ 2 } \right ] , \\ \sigma_0 v|_{wh } & = & \frac{\pi \alpha_2 ^ 2}{12m_2 ^ 2 } \left [ 1 - \frac{m_w^2}{4m_2 ^ 2 } \right]^{-2 } \left [ \left ( 1 + \frac{m_w^2-m_h^2}{4m_2 ^ 2 } \right)^2 + 2 \frac{m_w^2}{m_2
^ 2 } \right ] \nonumber \\ & & \times \left [ 1 - \frac{m_w^2 + m_h^2}{2m_2 ^ 2 } + \frac { \left ( m_w^2 - m_h^2 \right)^2}{16m_2 ^ 4 } \right]^{1/2 } , \\ \sigma_0 v|_{wz } & = & \frac{\pi \alpha_2 ^ 2}{12m_2 ^ 2 } \left [ 1 - \frac{m_w^2+m_z^2}{2m_2 ^ 2 } + \frac { \left ( m_w^2 - m_z^2 \right)^2}{16m_2 ^ 4 } \right]^{3/2 } \left [ 1 - \frac{m_w^2+m_z^2}{4m_2 ^ 2 } \right]^{-2 } \nonumber \\ & & \times \left [ 1 - \frac{m_w^2}{4m_2 ^ 2 } \right]^{-2 } \left [ 1 + \frac{5}{2}\frac{m_w^2+m_z^2}{m_2 ^ 2 } + \frac{m_w^4 + m_z^4 + 10 m_w^2 m_z^2}{16m_2 ^ 4 } \right].\end{aligned}\ ] ] since the neutral wino is a majorana particle , its initial state forms only a spin-0 state .
a pair of the neutral wino annihilates only into @xmath381 .
@xmath382^{3/2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } \right]^{-2}.\end{aligned}\ ] ] when the initial state forms a spin-0 state , it annihilates into @xmath383 , @xmath384 , @xmath385 , and @xmath386 .
corresponding cross sections of these annihilation channels are as follows : @xmath387^{3/2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } \right]^{-2 } , \\
\sigma_0 v|_{zz } & = & \frac{4\pi \alpha_2 ^ 2 c_w^4}{m^2 } \left [ 1 - \frac{m_z^2}{m_2 ^ 2 } \right]^{3/2 } \left [ 1 - \frac{m_z^2}{2m_2 ^ 2 } \right]^{-2 } , \\ \sigma_0 v|_{z \gamma } & = & \frac{8\pi \alpha_2 ^ 2 c_w^2 s_w^2}{m^2 } \left [ 1 - \frac{m_z^2}{2m_2 ^ 2 } + \frac{m_z^4}{16m_2 ^ 4 } \right]^{3/2 } \left [ 1 - \frac{m_z^2}{4m_2 ^ 2 } \right]^{-2},\end{aligned}\ ] ] on the other hand , when the initial state forms a spin-1 state , it annihilates into @xmath384 , @xmath388 , and @xmath389 .
below , @xmath390 and @xmath391 denote electric charge and @xmath21 charge respectively , while @xmath392 is defined as @xmath393^{1/2}$ ] .
@xmath394^{3/2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } \right]^{-2 } \left [ 1 - \frac{m_z^2}{4m_2 ^ 2 } \right]^{-2 } \nonumber \\ & & \times \left [ 1 - \frac{m_z^2 - m_w^2}{2m_2 ^ 2 } \right]^{2 } \left [ 1 + \frac{5m_w^2}{m_2 ^ 2 } + \frac{3m_w^4}{4m_2 ^ 4 } \right ] , \\
\sigma_0 v|_{zh } & = & \frac{\pi \alpha_2 ^ 2}{12m_2 ^ 2 } \left [ 1 - \frac{m_z^2}{4m_2 ^ 2 } \right]^{-2 } \left[\left ( 1 + \frac{m_z^2 - m_h^2}{4m_2 ^ 2 } \right)^2 + 2 \frac{m_z^2}{m_2 ^ 2 } \right ] \nonumber \\ & & \times \left [ 1 - \frac{m_z^2+m_h^2}{2m_2 ^ 2 } + \frac{\left ( m_z^2 - m_h^2 \right)^2}{16m_2 ^ 4 } \right]^{1/2 } , \\
\sigma_0 v|_{f \bar{f } } & = & \frac{\pi \alpha_2 ^ 2}{6m_2 ^ 2}\frac{k_f}{m_2 } \left[4 s_w^4 q_f^2 \left ( 3 - \frac{k_f^2}{m_2 ^ 2 } \right ) - 3 \frac{m_f^2}{m_2 ^ 2 } i_{3f}^2 \left ( 1 - \frac{m_z^2}{4m_2 ^ 2 } \right)^{-2 } \right . \nonumber \\ & & + 4 s_w^2 q_f \left ( i_{3f}- 2 s_w^2 q_f \right ) \left ( 3 - \frac{k_f^2}{m_2 ^ 2 } \right ) \left ( 1 - \frac{m_z^2}{4m_2 ^ 2 } \right)^{-1 } \nonumber \\ & & \left .
+ 4 \left ( s_w^4 q_f^2 - i_{3f } s_w^2 q + i_{3f}^2/2 \right ) \left ( 3 - \frac{k_f^2}{m_2 ^ 2 } \right ) \left ( 1 - \frac{m_z^2}{4m_2 ^ 2 } \right)^{-2 } \right].\end{aligned}\ ] ] [ [ mixing - between - widetildew0-widetildew0-and - widetildew - widetildew- ] ] mixing between @xmath364 and @xmath365 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ since @xmath364 and the spin-0 state of @xmath365 have the same quantum number , they are mixed each other . in order to evaluate the sommerfeld factor for the states
, we have to calculate the imaginary part of the transition amplitude between @xmath364 and @xmath365 . as can be easily understood from the interaction of the neutral wino ,
the intermediate state of the amplitude is the @xmath395 boson pair , which is evaluated as @xmath396^{3/2 } \left [ 1 - \frac{m_w^2}{2m_2 ^ 2 } \right]^{-2}.\end{aligned}\ ] ]
|
the heavy sfermion scenario is naturally realized when supersymmetry breaking fields are charged under some symmetry or are composite fields . there , scalar partners of standard model fermions and the gravitino are as heavy as @xmath0 tev while gauginos are as heavy as @xmath1 tev .
the scenario is not only consistent with the observed higgs mass , but also is free from cosmological problems such as the polonyi problem and the gravitino problem . in the scenario ,
gauginos are primary targets of experimental searches . in this thesis
, we discuss gaugino masses in the heavy sfermion scenario .
first , we derive the so - called anomaly mediated gaugino mass in the superspace formalism of supergravity with a wilsonian effective action
. then we calculate gaugino masses generated through other possible one - loop corrections by extra light matter fields and the qcd axion .
finally , we consider the case where some gauginos are degenerated in their masses with each other , because the thermal relic abundance of the lightest supersymmetric particle as well as the the strategy to search gauginos drastically change in this case . after calculating the thermal relic abundance of the lightest supersymmetric particle for the degenerated case
, we discuss the phenomenology of gauginos at the large hadron collider and cosmic ray experiments .
|
the salience of satellite dwarf galaxies for understanding galaxy formation in a cosmological context has been made all the more evident in the past decade with the discovery of numerous faint local group galaxies .
these faint systems are not only important to understand the faint end of galaxy formation ( e.g. @xcite ) but also their distribution around their host can test the hierarchical formation induced by the favored cosmological paradigm ( e.g. @xcite ) .
the andromeda satellite system is arguably the ideal system to study in order to tackle these questions as its proximity renders it easily observable , while our outsider s perspective facilitates its complete mapping , provided one has access to high - quality , multi - color , and wide - field photometry .
the number of known andromeda satellite dwarf galaxies has soared over the last decade , thanks mainly to two large photometric surveys of this region of the sky .
the sloan digital sky survey ( sdss ) enabled the discovery of a handful of relatively bright ( @xmath16 ) systems , initially from a dedicated stripe along the major axis of the andromeda galaxy ( and ix , and x ; @xcite ) , which was later complemented by a more systematic coverage of parts of the region around m31 in data release 8 ( and xxviii , and xxix ; @xcite ) . the pan - andromeda archaeological survey ( pandas )
is , on the other hand , a survey dedicated to the study of the stellar populations within the halo of the andromeda galaxy , out to distances of @xmath17 .
alone , it has enabled the discovery of 16 unambiguous new andromeda satellite dwarf galaxies @xcite with total magnitudes ranging from @xmath18 to @xmath19 .
the great success of these surveys in unveiling the m31 satellite system also highlights their coverage limitations . and vi , vii & xxviii are all examples of dwarf galaxies at projected distances of at least about @xmath20 ( corresponding to @xmath21 m31-centric distances ) , yet sdss has only surveyed regions that stay well away from the milky way disk , mostly south of m31 , and pandas is limited to a cone of @xmath22 from m31 . the panoramic survey telescope and rapid response system 1 ( pan - starrs1 ; @xcite ) is a photometry survey that does not have these limitations and , with final stacked data reaching slightly deeper than the sdss @xcite
, it has the potential to provide a much more complete understanding of the m31 dwarf galaxy satellite system out to and beyond the virial radius of andromeda ( @xmath23 ) . in this paper , we present the discovery of the two dwarf galaxies found in stacked pan - starrs1 imaging data . as they are located in the lacerta and cassiopeia constellations , but are also likely andromeda satellites , we follow the dual - naming convention emphasized in @xcite to avoid ambiguity .
we therefore dub the first dwarf galaxy both lacerta i and andromeda xxxi ( lac i / and xxxi ) and the second one both cassiopeia iii and andromeda xxxii ( cas iii / and xxxii ) .
the data set used for the discovery is summarized in 2 , while the derivation of the dwarf galaxies properties is described in 3 , before we conclude in 4 .
the ps1 @xmath15 survey ( chambers et al . , in preparation ) is a systematic imaging survey of the sky north of @xmath24 in five optical and near - infrared photometric bands ( @xmath25 ; @xcite ) .
conducted on a dedicated 1.8 m telescope located on haleakala , in hawaii , the system relies on a 1.4-gpixel , @xmath26-field - of - view camera for rapid mapping of the sky .
any location in the survey is observed repeatedly for a planned four times per year per filter , conditions permitting , with exposure times of @xmath27 seconds in the @xmath28-bands , respectively @xcite .
the median seeing values in these bands are @xmath29 arcseconds .
images are automatically processed through the image processing pipeline @xcite to produce the stacked data catalogue . at the time of this paper ,
the region around andromeda had been observed for three consecutive seasons from may 2010 , with up to twelve exposures per filter .
chip gaps , poor observing conditions , and technical problems , however , mean that a varying number of exposures are available for stacking at any given location . for point - sources ,
this leads to @xmath30-completeness limits of 22.1 in the @xmath0 band , and 22.2 in the @xmath1 band for the surroundings of lac i. around cas iii , these limits are 22.0 and 21.9 , respectively . checking the spatial completeness from the presence of 2mass stars , bright enough to be observed with ps1 , yields an area coverage factor @xmath31 percent in either band .
flags were used to remove spurious detections that are above the saturation limit , blended with other sources , thought to be defect , saturated pixels , or are considered significant cosmic rays or extended sources .
sources for which either the point - spread function fit is bad , the moments , the local sky , or the local sky variance could not be measured were also removed from the data set .
finally , detections for which the peak lands on a diffraction spike , a ghost , a glint , or near a chip edge were discarded as well , in addition to sources with signal - to - noise below 5 .
the photometric calibration of the stacked ps1 data is still under development , with spurious offsets in the photometry for subsets of detections . in the region of lac
i , a comparison of the stacked photometry with that of the single exposure ps1 observations , calibrated at the 0.01-magnitude level by @xcite , yields no significant offset in the @xmath1 band , whereas the @xmath0 band has @xmath32 percent of sources offset by @xmath33magnitudes . in the cas
iii surroundings , a small fraction of measurements are affected at the 0.05 level in the @xmath0 band , and at the 0.1 level in the @xmath1 band . however , since the majority of point sources have a reliable photometry , we have decided to nevertheless keep theses discrepant sources in the catalogue . throughout , the distance to m31
is assumed to be @xmath34 and the uncertainties on this values are modeled by drawing directly from the posterior probability distribution function provided by @xcite .
all magnitudes are dereddened using the @xcite maps , with the extinction coefficients for the ps1 filters provided by @xcite : @xmath35 , @xmath36 , and @xmath37 .
map of dwarf galaxies located in the surroundings of m31 and m33 , projected on the tangential plane centered on m31 .
known satellites are shown as hollow circles while the two new discoveries , lac i and cas iii , are represented by filled circles .
the limits of the sdss and pandas are represented by the thin - line and the double - line polygons , respectively .
the dotted lines highlight lines of constant galactic latitude at @xmath38 , @xmath39 , and @xmath40 from top to bottom . ] in a first pass at searching for unknown m31 satellites within @xmath20 from their host , a map was produced from the catalogue sources selected in the color - magnitude diagram ( cmd ) to loosely correspond to potential red giant branch ( rgb ) stars at the distance of andromeda ( @xmath41 , @xmath42 ) .
this map showed a handful of high - significance peaks , including the well - known m31 dwarf galaxies and i , ii , iii , v , vi , and lgs3 .
the cmd of significant peaks which did not overlap with known satellites were inspected by eye , leading to the clear discovery of two compact stellar system located at icrs @xmath43 and @xmath44 that have no counterparts in ned or simbad .
we designate them as lac i / and xxxi and cas iii / and xxxii and show the location of the two dwarf galaxies comparatively to the m31 satellite dwarf galaxy system in figure [ map ] .
the top , left - hand panels of figure [ cmd_map ] present the cmd of stars within two half - light radii of lac i ( see below for the estimation of its structural parameters ) ; for comparison , the panels also include the cmd of a field region with the same coverage , but offset by a degree towards the south - east .
lac i produces a clear overdensity of stars within the rgb color - selection , typical of stars close to the tip of the red giant branch ( trgb ) at the distance of m31 .
isolating these stars yields the map shown to the right of the cmds and reveals that they are also clumped in a clear spatial overdensity , with a size of a few arcminutes , typical of m31 satellite dwarf galaxies .
the same is true for cas iii ( top , right - hand panels ) , despite a more severe contamination from the prominent milky way disk . for comparison ,
the bottom panels of figure [ cmd_map ] show the same set of plots for the well - known andromeda i dwarf galaxy , also present in the stacked ps1 data , albeit in a region of deeper data .
the cmds of both and i and its background region are less populated owing to the lower galactic foreground contamination ( @xmath45 for lac i and @xmath46 for cas iii vs. @xmath47 for and i ) , but otherwise displays a similar overdensity of stars when compared to the two discoveries .
.[properties]derived properties of lac i / and xxxi and cas iii / and xxxii [ cols="<,^,^",options="header " , ] the structural parameters of lac i and cas iii are determined following the maximum likelihood technique presented by @xcite , and updated in martin et al .
( 2013 , in prep ) to allow for a full markov chain monte carlo estimate of the parameters , allowing for missing data . for systems that contain relatively few stars ,
a likelihood - based estimate of the structural parameters is far less biased and much more robust than estimates that rely on binning the data @xcite . in this instance
, the algorithm explores the likelihood of a family of exponential models with a constant background level to reproduce the distribution of ps1 stars around lac i or cas iii .
the parameters of the dwarf galaxy models are its centroid , its exponential half - light radius ( @xmath48 ) , the position angle of its major axis ( @xmath49 ) , its ellipticity ( @xmath50 , with @xmath51 and @xmath52 being , respectively , the major and minor axis scale lengths of the system ) , and the number of member stars within the chosen cmd selection box ( @xmath53 ) , which is tied to the background level through the number of stars in the data set .
uncertainties on these parameters are determined from the one dimensional , marginalized likelihood distribution functions and the location at which the likelihood reaches @xmath54 of the peak value .
for a gaussian distribution , the corresponding parameter values bound the @xmath55 confidence interval .
the likelihood distribution functions for these parameters , constrained on the ps1 data of the region within @xmath56 of the centroid of lac i or cas iii , are presented in figure [ andxxxi_struct ] for the most relevant parameters . these are well constrained and ,
in particular , the high significance of @xmath53 in both cases leaves no ambiguity to the existence of the new dwarf galaxies .
their radial density profiles are also displayed in the same figure and , for both dwarf galaxy , compares the data , binned following the most likely centroid , ellipticity , and position angle , with the exponential model of most likely half - light radius and number of stars . in both cases , the very good agreement between the two is testimony to the quality of the fit .
the measured structure of lac i and cas iii are rather typical of bright andromeda satellite galaxies with moderate ellipticities ( @xmath57 and @xmath58 , repectively ) .
the size of the two systems is on the high side of the size distribution of m31 satellite dwarf galaxies : @xmath9arcmin or @xmath10 for lac i and @xmath11arcmin or @xmath12 for cas iii ( see below for their heliocentric distance estimates ) . the properties of lac i and cas iii are summarized in table [ properties ] and the mcmc chains are available on demand to the corresponding author .
the magnitude of the trgb is a very good distance indicator in the case of andromeda dwarf galaxies ( e.g. , @xcite ) , especially in the absence of deep photometric data .
the tip s location also has the benefit of being relatively insensitive to the metallicity and age of stellar populations , as long as these are reasonably old and metal - poor , a condition that is fulfilled for all known m31 dwarf spheroidal galaxies .
we apply the likelihood technique implemented by @xcite to determine the distance to and xxviii , later confirmed to be accurate from deeper observations of this system ( slater et al .
, in preparation ) .
the trgb of lac i is estimated at @xmath59 , and that of cas iii at @xmath60 . in both cases , the measurement is in good agreement with the dwarf galaxy s luminosity function , as shown in figure [ andxxxi_lf ] .
the trgb magnitude has been estimated at @xmath61 by @xcite .
its conversion onto the ps1 photometric system using the color equations of @xcite yields @xmath62 and allows us to calculate the distance modulus of lac i : @xmath63 .
this distance modulus corresponds to a heliocentric distance of @xmath5 and , when also accounting for the m31 distance uncertainty , an m31-centric distance of @xmath7 .
given the dearth of isolated local group dwarf galaxies , it still appears that likely lac i is an andromeda satellite despite its large m31-centric distance ; its properties are similar to other remote satellites like and vi , vii , xxviii , or xxix , which have now all been confirmed as andromeda satellites from radial velocity measurements @xcite . in the case of cas
iii , the distance modulus is @xmath64 , leading to a heliocentric distance of @xmath6 , and an m31-centric distance of @xmath8 . at this distance ,
cas iii is even more likely to be a member of the andromeda satellite system . of course , radial velocities are needed to add evidence that both lac i and cas iii are truly bound to m31 . to estimate the total magnitude of the two dwarf galaxies
, we first sum up the flux contribution of all stars within their half - light radius that are in the interval @xmath65 and also broadly fall in the rgb s color range .
the contaminating flux from foreground sources is calculated in the same way from a large local field region , later scaled to the area within the half - light radius of lac i and cas iii , and subtracted from the flux of the dwarf galaxies .
we correct for unobserved stars by integrating a metal - poor ( @xmath66 ) and old ( 12.7gyr ) parsec luminosity function in the ps1 photometric system @xcite .
this yields the corrected apparent magnitude : @xmath67 and @xmath68 for lac i and @xmath69 and @xmath70 for cas iii .
data are not deep enough in @xmath71 to yield a reliable estimate of the apparent magnitude in that band , complicating the transformation to the widely used @xmath72-band magnitude . in order to circumvent this difficulty
, we rely on a differential measurement with the @xmath73 magnitudes of and i , ii , iii and v as measured from ps1 data , in the ps1 bandpasses .
we repeat the above exercise for these four dwarf galaxies , account for their distance modulus difference with those of lac i and cas iii , and tie the total magnitude of the two discoveries to that of these well - known dwarf galaxies , yielding @xmath74 for lac i and @xmath75 for cas iii . the crudeness of this method and the multitude of assumptions it relies on
are made evident by the large uncertainties , measured as the dispersion of the magnitude derivations from the four reference dwarf galaxies .
nevertheless , the cmds of the two new dwarf galaxies apparently containing a similar number of stars to that of and i ( figure [ cmd_map ] ) acknowledges the fact that they have a total magnitude similar to that of and i ( @xmath76 ) .
we report the discovery of two new dwarf galaxies , lac i / and xxxi and cas iii / and xxxii , from stacked ps1 imaging data . at a heliocentric distance of @xmath5 , lac
i is located in the environs of the andromeda galaxy , at an m31-centric distance of @xmath7 , making it a likely satellite of andromeda .
cas iii is located closer to m31 in projection and , combined with our measurement of its heliocentric distance ( @xmath6 ) , is well within the m31 virial radius with an m31-centric distance of only @xmath8 .
the membership of both dwarf galaxies to the m31 satellite system shall be confirmed kinematically through our ongoing spectroscopic program of both lac i and cas iii .
the discovery of cas iii is of particular interest as it is located on the rotating disk of dwarf galaxies that includes about half of andromeda s dwarf galaxies @xcite .
both dwarf galaxies are quite large ( @xmath77 and @xmath78 ) , yet this is not unexpected for dwarf galaxies or their brightness .
@xcite estimated that , for a total magnitude of @xmath79 , andromeda dwarf galaxies have an average size of @xmath80 , with a lognormal dispersion of 0.23 , implying that lac i is barely above the average size and cas iii just slightly more than one dispersion away .
one also has to keep in mind that , of course , it is the largest systems that are the most likely to have so far avoided detection , biasing new discoveries towards larger systems than were discovered from photographic surveys .
poss ii f ( red ) image of and vi ( left ) and lac i ( right ) . the images are 30 arcmin on the side , centered on the two dwarf galaxies , with north to the top and east to the left .
and vi is readily visible while lac i appears as a faint amount of unresolved light .
more significant wispy features due to the foreground milky way dust appear at the edge of the lac i image . ] to a certain degree , the only unexpected property of lac i and cas iii are their total magnitude , @xmath81 and @xmath82 , respectively , which are an order of magnitude more luminous than any member of the cohort of local group dwarf galaxies discovered in the last decade .
although surprising at first , one must remember that , prior to ps1 , the region of the two discoveries had only been systematically observed in the optical by photographic surveys .
regardless , one might wonder why such bright dwarf galaxies were not discovered earlier and , in particular , why the searches based on photographic plates that led to the discovery of , e.g. , the fainter and vi @xcite , did not also pick up lac i. the explanation lies in the large half - light radii of the new systems , which are 23 times larger than that of and vi , leading to their much fainter surface brightness ( @xmath83 mag / arcsec@xmath14 ) .
in fact , if one is to look meticulously at the location of lac i in the poss plates , one can see a very faint fuzz of unresolved light that betrays the presence of the dwarf galaxy ( figure [ poss_images ] ) .
the signal of this detection was likely too easily confused with the wispy milky way dust filaments , evermore present at this latitude , and disregarded at the time .
cas iii , with its even larger size , does not appear as a significant detection in the poss plates . providing a detailed assessment of the stellar populations present in lac i and
cas iii is beyond the ps1 data and the scope of this discovery paper , even though we can nevertheless notice that the dwarf galaxies do not appear to harbor a population of very young and blue stars ( @xmath84myr ) . the strong signal of lac i and cas iii in the ps1 data , along with the detectability of fainter systems such as and xxix , highlights the ability of the ps1 @xmath15 stacked survey to expand our knowledge of the andromedan satellite system and , beyond , that of the 1-mpc heliocentric sphere .
the new observations should prove particularly interesting in the region around m31 not observed by the deeper pandas , nor by the sdss that shied away from the close - by milky way plane .
nfm gratefully acknowledges the cnrs for support through pics project pics06183 .
nfm & hwr acknowledge support by the dfg through the sfb 881 .
cts and efb acknowledge support from nsf grant ast 1008342 .
the pan - starrs1 surveys ( ps1 ) have been made possible through contributions of the institute for astronomy , the university of hawaii , the pan - starrs project office , the max - planck society and its participating institutes , the max planck institute for astronomy , heidelberg and the max planck institute for extraterrestrial physics , garching , the johns hopkins university , durham university , the university of edinburgh , queen s university belfast , the harvard - smithsonian center for astrophysics , the las cumbres observatory global telescope network incorporated , the national central university of taiwan , the space telescope science institute , the national aeronautics and space administration under grant no .
nnx08ar22 g issued through the planetary science division of the nasa science mission directorate , the national science foundation under grant no .
ast-1238877 , and the university of maryland . , a. r. , ibata , r. a. , lewis , g. f. , parker , q. a. , zucker , d. b. , martin , n. f. , mcconnachie , a. w. , irwin , m. j. , tanvir , n. , fardal , m. a. , ferguson , a. m. n. , chapman , s. c. , & valls - gabaud , d. 2012 , , 758 , 11 , r. a. , lewis , g. f. , conn , a. r. , irwin , m. j. , mcconnachie , a. w. , chapman , s. c. , collins , m. l. , fardal , m. , ferguson , a. m. n. , ibata , n. g. , mackey , a. d. , martin , n. f. , navarro , j. , rich , r. m. , valls - gabaud , d. , & widrow , l. m. 2013 , , 493 , 62 , n. , burgett , w. , chambers , k. , denneau , l. , heasley , j. , jedicke , r. , magnier , e. , morgan , j. , onaka , p. , & tonry , j. 2010 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
7733 , society of photo - optical instrumentation engineers ( spie ) conference series , e. f. , finkbeiner , d. p. , juri , m. , magnier , e. a. , burgett , w. s. , chambers , k. c. , grav , t. , hodapp , k. w. , kaiser , n. , kudritzki , r .- p . ,
martin , n. f. , morgan , j. s. , price , p. a. , rix , h .- w . , stubbs , c. w. , tonry , j. l. , & wainscoat , r. j. 2012 , , 756 , 158 , j. l. , stubbs , c. w. , lykke , k. r. , doherty , p. , shivvers , i. s. , burgett , w. s. , chambers , k. c. , hodapp , k. w. , kaiser , n. , kudritzki , r .- p . ,
magnier , e. a. , morgan , j. s. , price , p. a. , & wainscoat , r. j. 2012 , , 750 , 99 , d. b. , kniazev , a. y. , bell , e. f. , martnez - delgado , d. , grebel , e. k. , rix , h .- w . , rockosi , c. m. , holtzman , j. a. , walterbos , r. a. m. , annis , j. , york , d. g. , ivezi , . ,
brinkmann , j. , brewington , h. , harvanek , m. , hennessy , g. , kleinman , s. j. , krzesinski , j. , long , d. , newman , p. r. , nitta , a. , & snedden , s. a. 2004 , , 612 , l121 , d. b. , kniazev , a. y. , martnez - delgado , d. , bell , e. f. , rix , h .- w . ,
grebel , e. k. , holtzman , j. a. , walterbos , r. a. m. , rockosi , c. m. , york , d. g. , barentine , j. c. , brewington , h. , brinkmann , j. , harvanek , m. , kleinman , s. j. , krzesinski , j. , long , d. , neilsen , jr .
, e. h. , nitta , a. , & snedden , s. a. 2007 , , 659 , l21
|
we report the discovery of two new dwarf galaxies , lacerta i / andromeda xxxi ( lac i / and xxxi ) and cassiopeia iii / andromeda xxxii ( cas iii / and xxxii ) , in stacked pan - starrs1 @xmath0- and @xmath1-band imaging data . both are luminous systems ( @xmath2 ) located at projected distances of @xmath3 and @xmath4 from m31 .
lac i and cas iii are likely satellites of the andromeda galaxy with heliocentric distances of @xmath5 and @xmath6 , respectively , and corresponding m31-centric distances of @xmath7 and @xmath8 .
the brightest of recent local group member discoveries , these two new dwarf galaxies owe their late discovery to their large sizes ( @xmath9arcmin or @xmath10 for lac i ; @xmath11arcmin or @xmath12 for cas iii ) , and consequently low surface brightness ( @xmath13 mag / arcsec@xmath14 ) , as well as to the lack of a systematic survey of regions at large radii from m31 , close to the galactic plane .
this latter limitation is now alleviated by the @xmath15 pan - starrs1 survey , which could lead to the discovery of other distant andromeda satellite dwarf galaxies .
|
we argue that the results for the vacuum forces on a slab and on an atom embedded in a magnetodielectric medium near a mirror , obtained using a recently suggested lorentz - force approach to the casimir effect , are equivalent to the corresponding results obtained in a traditional way .
we also derive a general expression for the atom - atom force in a medium and extend a few classical results concerning this force in vacuum and dielectrics to magnetodielectric systems .
this , for example , reveals that the ( repulsive ) interaction between atoms of different polarizability type is at small distances unaffected by a ( weakly polarizable ) medium .
although modifications of the casimir and van der waals forces due to the presence of a medium between the interacting objects were the subject of interest for a long time @xcite , this issue is still of great importance owing to the dominant role of these forces at small distances and to the rapid progress in micro and nanotechnologies .
a common way of extending the lifshitz theory @xcite of the casimir effect @xcite to material cavities is to ( eventually ) employ the minkovski stress tensor @xmath0 when calculating the force @xcite .
recently , however , a lorentz - force approach to the casimir effect was suggested @xcite ( see also @xcite ) in which the relevant stress tensor is of the form ( brackets denote the average with respect to fluctuations and we use the standard notation for the macroscopic field operators ) @xmath1 in this work , we reinterpret the results for the vacuum forces on a slab and on an atom embedded in a semi - infinite magnetodielectric cavity ( see figure 1 ) as recently obtained using @xmath2 @xcite and argue their equivalence to the corresponding results obtained using @xmath3 . also , we derive a general expression for the atom - atom force in a medium and extend a few well - known results for this force to magnetodielectric systems . when calculating @xmath2 for planar geometry @xcite , the zero - temperature force on the slab per unit area in the configuration of figure [ sys ] can be written as @xcite @xmath4 @xmath5(i\xi , k),\ ] ] @xmath6(i\xi , k)\nonumber\\ \fl & + \frac{\hbar}{8\pi^2c^2}\int_0^\infty d\xi\xi^2\mu(n^2 - 1)\int^\infty_0\frac{dkk}{\kappa } \sum_{q = p , s}\delta_q \left[\frac{(1+r^q)^2-{t^q}^2}{(r^q\rme^{-2\kappa d})^{-1 } -r^q}\right](i\xi , k ) , \end{aligned}\ ] ] corresponding to the decomposition of the stress tensor in ( [ t ] ) . here
@xmath7 is the perpendicular wave vector in the cavity at the imaginary frequency , @xmath8 , @xmath9 and @xmath10 are fresnel coefficients for the ( symmetrically bounded ) slab and @xmath11 are those for the mirror .
as seen , the first term in @xmath12 may be combined with the traditional ( minkowski ) force @xmath13 to form a medium - screened casimir force , with the contributions of tm and te polarized waves scaled , respectively , by @xmath14 and @xmath15 , whereas the remaining term in @xmath12 may be regarded as a medium - assisted force @xcite . however , owing to this ( additional ) screening of the force , it turns out that the medium - screened van der waals and casimir forces have a rather unusual dependence on the material parameters of the system @xcite and , therefore , this formal combination demands reconsideration . here ,
we interpret ( 2 ) by recalling that , according to the @xmath16 quantum - field - theoretical approach to the casimir force , the minkowski stress tensor @xmath3 corresponds to the effective stress tensor in a medium which is in mechanical equilibrium @xcite .
if so , the ( unbalanced ) second term in ( [ t ] ) , which leads to @xmath12 , gives a force on the medium .
therefore , the force on the slab is given solely by the traditional force @xmath13 in ( 2 ) , whereas @xmath12 describes a force on the medium .
indeed , note that @xmath12 vanishes when there is no medium , @xmath17 , and is nonzero when there is no slab @xmath18 [ @xmath19 and @xmath20 in ( [ tf ] ) ] , in this case clearly representing the force on a layer of the medium .
the force on an atom near a mirror can be obtained from ( 2 ) by assuming that the slab consists of a thin ( @xmath21 ) layer of the cavity medium with a small number of atoms @xmath22 embedded in it @xcite .
then , from the lorentz - lorenz ( clausius - mossotti ) relation @xcite , it follows that @xmath23,\ ] ] where @xmath24 is the number density and @xmath25 the electric vacuum polarizability of embedded atoms .
assuming a similar relation between @xmath26 , @xmath15 and the effective atomic magnetic polarizability @xmath27 , for the force on an _ embedded _ atom at distance @xmath28 from the mirror we find ( for details , see the derivation of equation ( 14 ) in @xcite and @xmath29 must be made in the last term of this equation . ] ) [ fal ] @xmath30 @xmath31r^p\right .
\nonumber\\ \fl & + \left.\left[\alpha^a_m \left(2\frac{\kappa^2c^2}{\mu\xi^2}-\varepsilon\right ) -\alpha^a_e\mu\right]r^s\right\}(i\xi , k),\end{aligned}\ ] ] @xmath32r^p\right.\\ \fl & \left.+(\mu-1)\left[\alpha_m \left(2\frac{\kappa^2c^2}{\mu\xi^2}-\varepsilon\right ) -\alpha_e\mu\right]r^s+\left(\mu-\frac{1}{\varepsilon}\right ) \left[\alpha_e\mu r^p-\alpha_m\varepsilon r^s\right]\right\}(i\xi , k).\nonumber\end{aligned}\ ] ] as before , this can be combined to give a medium - screened and a medium - assisted force on the atom @xcite .
however , according to the interpretation adopted here , the true force on the atom is @xmath33 , whereas @xmath34 may be regarded as an atom - induced force on the medium ( per atom ) .
we note that ( [ fa ] ) extends ( in different directions ) previous results for the atom - mirror force in various circumstances @xcite by fully accounting for the magnetic properties of the system .
enables one to calculate the force between two atoms in a magnetodielectric medium
. this force can be found by assuming a single - medium mirror consisting of the cavity medium with a small number of , say , type @xmath35 atoms embedded in it , so that @xmath36 and @xmath37 . in this case , the reflection coefficients of the mirror can be approximated by @xcite @xmath38\ ] ] and @xmath39 .
also , the potential energy of the atom @xmath40 is given by @xmath41 , where @xmath42 and @xmath43 is the interaction energy between the atoms @xmath22 and @xmath35 .
using the identity @xmath44 and combining these relations , from ( [ fa ] ) and ( [ rp ] ) we obtain the extension of the feinberg and sucher formula @xcite to material systems @xmath45 where @xmath46 and @xmath47 . at small distances between the atoms ,
@xmath48 ( @xmath49 is an effective upper limit in the integration over @xmath50 @xcite ) , we may let @xmath51 , @xmath52 and @xmath53 in ( [ uab ] ) .
this gives for the atom - atom force @xmath54 at small distances @xmath55 which generalizes the well - known results for the van der waals - london force @xcite .
we observe that , whereas the force between the atoms of the same polarizability type is strongly affected by the surrounding medium , the ( repulsive ) force between the atoms of different polarizability type ( in weakly - polarizable media ) remains the same as in vacuum . at large distances between the atoms ,
the main contribution to the integral in ( [ uab ] ) comes from the @xmath56 region .
approximating the @xmath50-dependent quantities by their static values ( denoted by the subscript @xmath57 ) and then performing the integration , we arrive at @xmath58_0 , \label{faal}\ ] ] which gives the medium corrections to the retarded atom - atom force @xcite and extends the previous considerations of these corrections @xcite to magnetodielectric systems .
as seen , the repulsive component of the force is in a ( weakly - polarizable ) medium simply scaled by @xmath59 , whereas the modification of its attractive part is more complex .
we end this short discussion by noting that the symmetry of ( [ fa ] ) under the transformation : @xmath60 , @xmath61 and @xmath62 [ and consequently that of ( [ uab])-([faal ] ) with respect to the replacements @xmath60 and @xmath61 ] is a consequence of the invariance of the minkowski stress tensor @xmath3 with respect to a duality transformation , e.g. @xmath64 , @xmath65 , @xmath66 and @xmath67 @xcite .
this symmetry is lost in the second term of @xmath2 , so that the force @xmath34 is not invariant under the replacement of the electric and magnetic quantities .
since the medium - screened atom - mirror and atom - atom forces are combinations of the corresponding @xmath13 s and @xmath12 s , this explains the unusual ( asymmetric ) medium effects on these forces found in @xcite . in conclusion ,
when properly interpreted , the result for the vacuum force on an object ( a slab or an atom ) embedded in a medium near a mirror obtained using the lorentz - force approach to the casimir effect agrees with that obtained in a traditional way .
extensions of the well - known results for the atom - atom force to magnetodielectric systems reveals that the ( repulsive ) interaction between atoms of different polarizability type remains ( in weakly - polarizable media ) the same as in vacuum at small and is scaled by @xmath59 at large distances .
medium corrections to the force between atoms of the same polarizability type are more complex and depend on the polarizability type of atoms .
+ the author thanks to c. raabe and anonymous referees for constructive criticism and suggestions .
this work was supported in part by the ministry of science and technology of the republic of croatia under contract no . 0098001 .
|
|
diffusive systems can not support resonances in the usual sense , a periodic force can not build oscillations that grow with each cycle .
the frequency response of systems vary monotonously .
we present an over - damped system which has an oscillating frequency response due to the presence of a position dependent effective force .
such conditions arise for objects under the influence of retarded effects from boundaries an example of which we also provide .
the fokker - planck ( fp ) equation @xcite describes the time dependence of the probability distribution @xmath0 of the position of a particle in a diffusive environment : @xmath1 \label{fp}\ ] ] where @xmath2 is the drift constant relating velocity to force and @xmath3 is the coefficient of the diffusion term .
the potential energy @xmath4 is related to the time - independent force field the particle is in , while @xmath5 represents a time and position dependent driving force which will be relevant to our problem .
specifically , we will consider a force which emanates from sinusoidally varying boundary effects , propagating with a constant velocity into the interval of interest , so that the space dependence of the force is also sinusoidal . considerable amount of work has been carried out on the behaviour of @xmath6 , under the effect of various types of potentials .
analytical solutions are usually restricted to simpler forms of potentials .
the equation may be treated as a schrdinger equation with imaginary time , and the time dependent solutions have a relaxational type of behaviour .
this is the consequence of the assumption of absence of any inertia ( or `` memory '' ) in the system : the time rate of change of @xmath6 is described in terms of the value of @xmath6 at time @xmath7 .
however , we find resonance " behavior , in a looser context , as a consequence of eqn .
[ fp ] , which corresponds to amplified response at a sequence of frequencies , based on matching of the waveform of the driving force to the interval between the boundaries .
a number of other stochastic systems with damped dynamics do display non - monotonous frequency response to sinusoidal drives . in brownian motors ,
the particle moves in a ( usually ratchet - type ) static potential and the sinusoidal force is used to drive the particle over the easier " barrier @xcite .
( the asymmetry in the direction may be achieved either by the form of the static potential or the form of the time - dependent forces . ) alternatively , a high - frequency drive may effectively act to average over portions the static potential , on a length scale which depends on the properties of the drive .
this results in a
vibrational resonance " @xcite with transport properties strongly dependent on parameters associated with the drive .
the force may take the form of a superposition of a number of sinusoidal ( or rectangular ) functions , in which case the non - linear effects may lead to interesting effects at certain combination of frequencies @xcite . besides the models we mention here , the reader is referred to a review by hnggi and marchesoni @xcite for a detailed discussion of these and other similar systems .
sinusoidal drive on a damped system has also been studied in relation to stochastic resonance " @xcite .
this relates basically to the motion of a damped particle in a double well potential on which a random diffusive force as well as the sinusoidal force is acting .
it had been demonstrated that there would be an optimal magnitude for the random force which amplifies transitions between the potential minima in synchronism with the periodic drive .
jung and hnggi @xcite solved the corresponding fp equation , demonstrating that the time autocorrelation function of a periodically driven bistable over - damped system sustains undamped oscillations .
the term `` resonance '' here is associated with amplified response at a certain value of the magnitude of the random force .
the process has been used to model a very wide spectrum of phenomena , ranging for example from the effect of seasonal changes on population systems @xcite to the dependence of stock prices on periodic information flow @xcite , as well as a multitude of biological phenomena@xcite .
all of the above processes incorporate time - independent multiple potential wells and position independent periodic forces .
in contrast to , the phenomena we report is associated with a static potential confined to the boundaries and a periodic retarded force also emanates from the boundaries .
the effects of a force field propagating with a finite velocity ( which in our case is equivalent to a position dependent time - periodic force ) has not received much attention .
relevant work include the analysis of a diffusive system which injects particles ( which are annihilated upon contact with one another ) at a boundary @xcite , and the derivation of an effective force in a hamiltonian system with a position - dependent sinusoidal force coupled to a bath of oscillators ( resulting in the random force ) @xcite .
it may seem contradictory to discuss the presence of propagating effects within the context of a memoryless , over - damped system .
however , the underlying mechanism which drives the diffusion dynamics can be distinct from the process which generates the effective force .
this is the case , for example , in totally asymmetric simple exclusion process ( tasep ) systems where the diffusion of a persistent shock front in the system @xcite is influenced by the boundary conditions after a delay .
thus , tasep forms a discrete realization of the aforementioned fp system .
we first analyse the fp system ( sec .
ii / a ) , we then report the results of a monte carlo study carried out on the tasep system as well ( sec .
we will first describe our analysis of the fp equation .
in particular , we take the time - independent potential @xmath8 to be given by @xmath9 the quadratic structure softens the boundaries at @xmath10 with a range related to @xmath11 .
we also assume that a time dependent force acts on the particle , synchronously from the two boundaries , but retarded in time ( with a propagation speed @xmath12 ) in proportion to the distance to the boundaries : @xmath13 \nonumber \\ & & + f_0 \sin \left [ \omega\left ( t-\frac{l/2-x}{v}\right ) \right ] \nonumber \\ & = & 2f_0 \cos \frac{\omega x}{v } \sin \left [ \omega\left ( t-\frac{l}{2v}\right ) \right ] . \label{force}\end{aligned}\ ] ] it can be observed that the retarded force results in a position dependent amplitude in the oscillation .
this amplitude is not frequency dependent , however we will show that the magnitude of the response to this force does depend on how the wavelength compares to the size of the system .
we write eqn .
[ fp ] in scaled form @xmath14\nonumber \\ & + & \overline d \frac{\partial^2 p}{\partial z^2 } \label{scaled_fpe}\end{aligned}\ ] ] where we have used the dimensionless quantities in table [ table : tablo ] with unitless potential : @xmath15
l l + + @xmath16 & @xmath17 + @xmath18 & @xmath19 + @xmath20 & @xmath21 + [ 0.25 cm ] the parameter @xmath22 , besides representing the wavelength of the time dependent force relative to the distance variable @xmath23 , is also proportional to the period of oscillation : @xmath24 .
we have solved equation [ scaled_fpe ] numerically for various values of the parameters .
the probability density @xmath25 was solved on a mesh of 256 points in the @xmath23 direction , corresponding to the @xmath26-coordinate values for @xmath27 plus the two boundary regions which were taken to be @xmath28 wide each .
the equation was integrated numerically in the time variable @xmath29 with step sizes such that @xmath30 .
the integration for the period of @xmath31 was repeated 10 times , which was sufficient for convergence , _
i.e. _ for obtaining no appreciable change in @xmath25 .
we then calculate the expectation value of position as a function of @xmath29 : @xmath32 the size of this oscillation is parametrized though the fundamental fourier coefficients : @xmath33 the magnitude of the oscillation of the position expectation value as a function of wavelength @xmath22 for various values of @xmath34 , the boundary smoothness .
( other parameters which enter eqn .
[ scaled_fpe ] are @xmath35 and @xmath36 . )
the continuous ( dotted ) line represents the in phase ( out of phase ) fundamental component @xmath37 ( @xmath38 ) in eqn .
[ sincos ] of the oscillation . for large values @xmath22 , the value of @xmath38 saturates due to the system width .
note the different scales for the response in the figures.,width=325 ] fig .
[ fig : coklu_plot ] displays that the response of the system has strong frequency dependence .
the amplitude of the oscillations grow as @xmath34 decreases , when the boundaries become sharper .
a wider boundary allows more oscillations ( as a function of position ) in the probability density , diminishing the variations in @xmath39 . as the boundary region expands , so do the features on the plot , implying longer wavelengths . for values of @xmath22
much larger than one , the response monotonically increases to its asymptotic value .
note also that the `` in phase '' component @xmath37 dominates the response for smaller @xmath34 .
the size of @xmath40 was chosen to obtain a response magnitude comparable to that obtained from the monte carlo analysis .
the extrema of the response correspond to matching of the wavelength of the driving force to the effective length of the diffusion region @xmath41 , plus the boundary region .
[ fig : ikili_fp ] displays the probability densities for points * a * and * b * in fig . [ fig : coklu_plot ] . note that for @xmath42 and @xmath43 , the probability densities increase with @xmath23 in both plots @xmath44 and @xmath45 , consistent with a sinusoidal drive .
however , when an even ( odd ) number of maxima are present , the position expectation value for these cases become negative ( positive ) , producing oscillations as a function of wavelength .
as more and more wavelengths `` fit '' into the system ( smaller @xmath22 ) , the change in the expectation value of the particle position becomes less and less discernible .
the probability densities of marked points in eqn .
[ scaled_fpe ] .
( a ) corresponds to point * a * ( @xmath46 ) , consecutive dark and light patterns along the x - axis , indicate standing waves of two wavelengths that fits to the lattice size .
( b ) corresponds to point * b * ( @xmath47 ) , it can be seen that only one wavelength is supported by the system .
, width=325 ] we now turn to the tasep system which provides a discrete realization of this phenomenon on a one dimensional lattice .
its dynamics is described by the following transition probabilities in time @xmath48 : on the leftmost site @xmath49 with probability @xmath50 , inside the bulk @xmath51 with probability @xmath48 , and on the rightmost site @xmath52 with probability @xmath53 .
these systems have been studied extensively for time - independent boundary rates .
recently , the effects of time dependence has also received interest : popkov _ et al .
_ studied a vehicular traffic on highways under periodically changing green and red - lights @xcite and basu _ et al . _ also showed a frequency dependent modality on similar transport systems @xcite .
shock profile distributions for various numbers of particle number .
, width=325 ] the time - independent system can be solved exactly @xcite .
a first order phase transition line separates the high and low density phases for @xmath54 . along this line
the density may be shown to correspond to a superposition of shock profiles which move from one end of the system to the other .
these structures are not quantities that are measurable at any instant of time : occupation statistics of states with a specific number of @xmath55 particles lead to a shock profile associated with that @xmath55 . fig .
[ fig : shockdist ] shows these profiles for a system with 50 sites , for various values of @xmath55 .
the linear dependence of the position of the profile on particle number is apparent .
[ fig : lineer ] displays this dependence .
a perturbation to the boundary conditions leads to a change in the number of particles in the system .
this then acts as an effective force which results in the change of the shock position after a delay .
moreover for small values of @xmath56 these shock profiles tend to be evenly distributed within the lattice .
in other words , the shock front carries out a random walk in the lattice @xcite .
[ fig : p_n_dist ] displays the probability distribution @xmath57 for the existence of @xmath55 particles ( equivalently contribution from a shock profile at a corresponding position ) in the time independent system .
this distribution indeed may be interpreted as the result of a random walk of the profile in a potential of the form @xmath8 .
this random walk is ordinarily constrained by the time - independent boundaries : if the shock wave moves too close to the left , increased density at that boundary reduces the particle entry rate , effectively `` pushing '' the profile to the right .
a similar mechanism on the exit boundary constrains the profile to the central region of the lattice .
this constraint has been represented by a free - energy functional " @xcite which is quite similar in form to @xmath8 .
the linear relationships between particle number @xmath55 and shock position @xmath58 for the profiles in fig .
[ fig : shockdist ] .
the shock position @xmath58 is defined as the lattice position at which the density @xmath59 has the value @xmath60 where @xmath61 and @xmath62 are respectively the maximum and and minimum values of @xmath59 .
the inset displays the relationship when boundary conditions change sinusoidally with period @xmath63 as discussed in the text .
the plots are for four different time values as @xmath64 , @xmath65 , @xmath66 and @xmath67 . here
@xmath68 and @xmath69 .
it can be seen that the linearity of the relationship does not change appreciably with time , resulting in an approximate relation of the form @xmath70 . since @xmath58 is determined by interpolation , it in general is not an integer .
, width=325 ] probability distribution @xmath57 of particle number @xmath55 . , for a lattice size of @xmath68 and @xmath71.,width=325 ] response of the probability distribution function @xmath57 to a pulsed @xmath72 rate .
time averaged values have been subtracted to better display the response .
pulse s magnitude is @xmath73 , other parameters are the same as those in fig .
[ fig : p_n_dist ] .
the arrows indicate the maxima of the curves to better display the propogation.,width=325 ] the change in the probability distribution in fig .
[ fig : p_n_dist ] in response to a pulsed change in the particle entry rate @xmath72 is shown in fig .
[ fig : bump_response ] .
the disturbance caused by an influx of extra particles for a short period of time travels along the lattice , with a damped response .
( note the oscillatory nature of the response indicating a travelling wave mechanism as well as damping . ) while this motion along the lattice justifies the association of this system with the `` retarded force '' mechanism of eqn .
[ fp ] , the relatively strong damping leads to results not as `` clean '' as the ones given in fig.[fig : ikili_fp ] for the fp equation .
we now proceed with our discussion of the details of our computations of the sinusoidally varying drive .
the mc procedure we used is known as the
kinetic monte carlo " method @xcite , in which events occur at time intervals @xmath74 where @xmath75 is a random number uniformly distributed between @xmath76 and @xmath77 , and @xmath78 is the sum of rates of all possible transitions in the system .
the event which does take place is also selected randomly with a probability proportional to its rate .
transition rates at the boundaries are assumed to be constant within the duration @xmath79 , since @xmath80 .
the mersenne twister algorithm was used for generating @xmath75 @xcite .
time dependent probability density calculations are carried out over @xmath81 monte carlo steps ( mcs ) . for a lattice of size @xmath82
, we define a mcs as @xmath83 changes in the system .
we have observed that the time - dependent boundary effects decay appreciably inside the lattice .
a small lattice is therefore necessary for boundary effects to be seen inside the bulk .
moreover , broadest range of a random walk is known to take place for smaller values of @xmath72 and @xmath84 . in order to extend the range of the random walk to the full range between the boundaries ,
@xmath85 is taken for an anchor point on a lattice of size @xmath68 .
for the effect to be prominent , boundary rates were varied with a significant amplitude such as @xmath86 and @xmath87 , comparable to the average values of @xmath72 and @xmath84 .
note that our time - dependent perturbation drives the system between two values of the boundary parameters corresponding to low and high density phases .
this drive should then result in an oscillation of the number of particles within the system , hence of the probability density of the shock position . in fig .
[ ikili_mc ] , the probability distribution for finding @xmath55 particles in the system ( which is linearly related to shock position as can be seen in fig .
[ fig : lineer ] ) as a function of time , is plotted .
the results agree qualitatively with the fp results @xcite .
we have used compatible values for corresponding parameters in the fp analysis : the exact diffusion constant @xmath88 of the system @xcite yields the value @xmath89 for @xmath71 . we have also observed atypical behaviour within two lattice sites of the boundaries in the mc data , which were excluded from plots in fig .
[ ikili_mc ] .
this corresponds to a boundary smoothness of @xmath90 , which too was used in the fp analysis .
fundamental components of the system s response were calculated using the time - dependent probability function @xmath91 for the occupation of @xmath55 sites at time @xmath7 : @xmath92 where @xmath93 change in probability density @xmath94 from its time average for @xmath68 and different periods calculated over @xmath81 mcs .
( a ) when @xmath95 , similar to fig .
[ fig : ikili_fp ] a , two wavelengths are supported by the system .
( b ) when @xmath96 , similar to the results of fig .
[ fig : ikili_fp ] b , only one wavelength is supported by the system . in both plots , two extremal coordinates in both sides were excluded to avoid artifacts due to boundary effects .
color scales are arbitrary.,width=325 ] fundamental components ( @xmath38 and @xmath37 in eqn . [ disc ] ) with respect to different period values .
ten maximum and minimum values of particle number are excluded from the calculation .
extrema in the plot , indicate resonance like response of the system .
points @xmath97 and @xmath98 correspond to the density distributions displayed in fig .
[ ikili_mc ] .
inset shows that for smaller values of period , the interesting sinusoidal behavior is present.,width=325 ] oscillations in these components are shown in fig .
[ fig : mc_response ] .
these oscillations are not as prominent as those obtained in the fp analysis , and displayed in fig .
[ fig : ikili_fp ] .
we attribute this to the decay of the magnitude of the effective force away from the boundaries , damping the effect at higher frequencies .
time dependence of the boundary rates cause sufficient fluctuations in the tasep system to cause the appearance of the density distributions that are quite different from those which result from constant boundary conditions .
the details of these effects in the tasep model will be reported separately .
here we have only discussed results relevant to the diffusive motion of the shock profile in this system .
we have demonstrated that the response of an over - damped system to a retarded oscillatory force from the boundaries leads to resonant effects in the oscillation amplitudes of statistical quantities such as the average position .
these are apparent in numerical solutions of the associated fp equation as well as the monte carlo analysis of the motion of shock fronts in a tasep system .
the basis of this phenomenon is the matching of a number of density oscillations in space within the system size .
the density oscillations are driven by a force field whose amplitude is space dependent , but does not have a frequency dependence . the frequency dependence in the response is the result of the matching stated above .
the authors acknowledge support from turkish academy of sciences .
the authors would also like to thank roland netz for useful discussions and thank to m. ozgur oktel for a critical reading of the manuscript .
99 a. d. fokker , ann .
phys . * 43 * , 810 ( 1914 ) .
m. planck , sitzber .
akad . wissens . 324 ( 1917 ) .
andrei kolmogorov , on analytical methods in the theory of probability " , 448 - 451 , ( 1931 ) .
h. risken , _ the fokker - planck equation _ , springer , berlin ( 1989 ) .
p. hnggi , h. thomas , phys .
rep . * 88 * 207 ( 1982 ) r. bartussek and p. h'anggi , phys .
* 51 * ( 6 ) , 506 ( 1995 ) .
p. landa and p. v. e. mcclintock , j. phys .
a * 33 * , l433 ( 2000 ) .
cubero , j. casado - pascual , and f. renzoni , phys .
* 112 * , 174102 ( 2014 ) .
p. hnggi and f. marchesoni , rev .
phys . * 81 * , 387 ( 2009 ) .
r. benzi , a. sutera and a. vulpiani , j. phys .
a : math . gen .
* 14 * , l453 ( 1981 ) .
p. jung and p. hnggi , europhys .
* 8 * , 505 ( 1989 ) .
l. gammaitoni , f. marchesoni , e. menichella - saetta , and s. santucci phys . rev
. lett .
* 62 * , 349 ( 1989 ) .
jung , phys . rep . * 234 * , 175 ( 1993 ) .
frank moss , contemporary problems in statistical physics .
, 205 - 253 , ( 1994 ) l. gammaitoni , p. hnggi , p. jung , and f. marchesoni , rev . mod . phys . * 70 * , 223 ( 1998 ) .
b. spagnolo , d. valenti , and a. fiasconaro , math .
eng . 1 , 185 ( 2004 ) . j - c .
li , c. li , and d - c .
mei , phys .
a * 378 * , 1997 ( 2014 ) .
v. ushakov , a. a. dubkov , and b. spagnolo , phys .
lett . , * 107 * , 108103 ( 2011 ) .
s. martignoli , f. gomez , and r. stoop , sci . rep . * 3 * , 2676 ( 2013 ) .
n. zygouras , ann .
henri poincare c26 , 271 ( 2009 ) .
a. shit , s chattopadhyay , and j. r. chaudhuri , eur .
j. b * 86 * , 23 ( 2013 ) .
r. a. blythe and m. r. evans , j. phys .
a : math . theor . * 40 * r333 ( 2007 ) .
v. popkov , m. salerno , and g. m. schuetz , phys .
e. ( 2008 ) .
u. basu , d. chaudhuri , and p. k. mohanty , phys .
e. ( 2011 ) .
b. derrida , e. domany , and d. mukamel , j. stat .
* 69 * , 667 ( 1992 ) .
b. derrida , m. r. evans , and d. mukamel , j. phys . a * 26 * , 4911 ( 1993 ) .
b. derrida , m. r. evans , v. hakim , and v. pasquier , j. phys . a * 26 * , 1493 ( 1993 ) .
p. f. arndt , t. heinzel , and v. rittenberg , j. stat .
a. b. bortz , m. h. kalos and j. l. lebowitz , j. comp . phys .
* 17 * , 10 ( 1975 ) .
t. nishimura , acm trans .
mod . comp .
10 ( 2000 ) .
m. matsumoto and t. nishimura , acm trans .
. 8 ( 1998 ) . in fig .
[ ikili_mc ] , the time average of the density was subtracted from @xmath94 to magnify its time - dependece .
note that the scales in fig .
[ ikili_mc ] are with respect to the variable @xmath7 , while those in fig .
[ fig : ikili_fp ] are with respect to @xmath29 , which contains a phase shift .
b. derrida , m. r. evans , and k. mallick , j. stat .
phys.*79 * , 833 ( 1995 ) .
|
strong frequency dependence is unlikely in diffusive or over - damped systems .
when exceptions do occur , such as in the case of stochastic resonance , it signals an interesting underlying phenomenon .
we find that such a case appears in the motion of a particle in a diffusive environment under the effect of periodically oscillating retarded force emanating from the boundaries .
the amplitude for the expectation value of position has an oscillating frequency dependence , quite unlike a typical resonance .
we first present an analysis of the associated fokker - planck equation , then report the results of a monte carlo simulation of the effect of a periodic perturbation on a totally asymmetric simple exclusion process ( tasep ) model with single species .
this model is known to exhibit a randomly moving shock profile , dynamics of which is a discrete realization of the fokker - planck equation .
comparison of relevant quantities from the two analyses indicate that the same phenomenon is apparent in both systems .
|
we congratulate the organizers for arranging and conducting this conference , celebrating the 150 years of the discovery by ( gustav robert ) kirchoff and ( robert wilhelm eberhard ) bunsen that various gases can be easily and positively identified by a detailed study of the light they emit and absorb . as paul merrill ( @xcite ) noted , a new era in astronomy began when bunsen saw ` in yellow flame of an ordinary alcohol lamp whose wick was sprinkled with salt , and possibilities of the chemical analysis of the most distant stars ' .
thus began the era of astronomical spectroscopy .
merrill s own contributions to our discipline are legendary with his discovery of tci lines in the spectra of s stars signaling that element synthesis in stars was a continuing phenomena ( @xcite )
. a few historical remarks might be appropriate .
astronomical spectroscopy in india started during the famous total solar eclipse that occurred on 1868 august 18 in which a spectroscope was first used to study the nature of solar prominences leading to the discovery of helium .
madras observatory , from which the present host institution evolved , also played a role .
norman robert pogson , then director of the observatory , used a visual spectroscope to look at the prominences and noted a bright line in the yellow close to but not coincident with sodium d lines .
this was the d3 line - the same line as found by janssen during the 1868 eclipse but by lockyer without the aid of eclipse .
the first stellar spectrum recorded in india was obtained at madras ( now chennai ) ; this was a spectrum of the wolf - rayet star @xmath0 velorum observed in 1871 .
spectroscopic analysis of some of the most enigmatic stars yet discovered is the theme of our contribution .
the chosen stars are the r coronae borealis stars and their putative cooler relatives , the hydrogen - deficient cool carbon stars ( hdcs ) .
the enigma presented by these stars has two principal parts : ( i ) how did these stars become so h - deficient ? and
( ii ) how do the rcbs but not the hdcs develop thick clouds of soot that obscure the stellar photosphere from view ?
answers to these questions are emerging . here
, we highlight particular contributions to their solution from analysis of high - resolution optical and infrared spectra .
@xcite has lamented ( tongue in cheek , we suppose ) that ` the tragedy of astronomy is that most information lies in spectra ' .
we would counter by substituting ` allure ' for ` tragedy ' and add that one should expect to apply all available observational tools in seeking answers to our questions .
r coronae borealis stars are a rare class of peculiar variable stars .
the two defining characteristics of rcbs are ( i ) a propensity to fade at unpredictable times by up to about 8 magnitudes as a result of obscuration by clouds of soot , and ( ii ) a supergiant - like atmosphere that is very h - deficient and he - rich .
the @xmath1 of the class ranges from 8000 to 4000 k and @xmath2 from 0.5 to 1.5 ( cgs units ) except for the two hotter stars dy cen ( 19000 k ) and mv sgr ( 14000 k ) .
the m@xmath3 ranges from -5 to -2.5 and @xmath4 from 4.0 to 3.2 l@xmath5 . presently , there are 52 known rcbs in the galaxy ( @xcite ) , 21 known members in the large magellanic cloud and 6 in the small magellanic cloud
. the total number of rcbs in the galaxy may be about 3200 ( @xcite ) there are only 5 known hdc stars in the galaxy . their detection may be hampered because they do not show rcb - like light variations , an infrared excess or a stellar wind ( @xcite ) .
they overlap the rcbs in t@xmath6 with a spread from about 6500 to 5000 k. at the high @xmath1 end of the rcbs sit the extreme helium stars ( ehes ) of which 21 are known in the galaxy with @xmath1 from 32000 to 9000 k and @xmath2 from 0.75 to 4.0 .
their @xmath4 ranges from 4.4 to 3.0 l@xmath5 . here
, we give no more than passing attention to the ehes and other , even hotter h - deficient stars that may form an evolutionary sequence with the hdcs and rcbs .
a fascinating aspect of rcbs is that they present several faces to observers , like the hindu mythological god karthikeya ( son of kritthikas -pleiades ) with his six faces .
the faces presented to our view are ( i ) a chemically peculiar supergiant , ( ii ) a variable star obscured without warning from our view by clouds of soot , ( iii ) a small amplitude pulsation both in light and radial velocity with a period around 40 days , ( iv ) an infrared source producing dust and blowing it away after several days , weeks , months or even years as clouds , ( v ) a hot stellar wind , and ( vi ) and may even be a central star of a low density nebula as evidenced by the presence of nebular lines of [ o ii ] , [ n ii ] ( fig . 1 ) and [ s ii ] .
they also have putative relatives of higher and lower temperatures , as intriguing as karthikeya s relatives .
our emphasis is on insights obtainable from high - resolution spectra with the assumption that relevant atomic and molecular data of adequate precision and completeness are to be provided from laboratory and theoretical spectroscopic studies .
this assumption is patently false , as is no surprise to practicing astronomical spectroscopists . on the chance that a reader of this paper may wish to take up a challenge , we mention two aspects of the incompleteness of knowledge of the spectrum of neutral carbon that compromise some analyses of the spectra of rcbs .
spectra of rcbs are crossed by many lines of ci , as noted by @xcite .
one might refer to the spectrum of a rcb as arising from a column of somewhat contaminated high - temperature carbon vapour .
a great step forward in identifying the ci lines was made when @xcite extended the laboratory spectrum and knowledge of the ci term system .
with modern spectra of rcbs , many suspected ci lines remain unidentified and cry out for a new laboratory investigation of the ci spectrum . today , astronomical spectroscopists often require more than a set of wavelengths and a table of energy levels .
this is certainly true for the pursuit of an abundance analysis leading to the chemical composition of a rcb .
abundance analysis at a minimum requires knowledge of the @xmath7-values for identified lines including the ci lines .
@xcite report on an extensive abundance analysis of rcbs and their uncovering of a ` carbon problem ' which may arise because the adopted theoretical @xmath7-values or the photoionization cross - sections for neutral carbon were too large by a factor of about four .
( asplund et al . also suggested possible ways to reduce the carbon problem without appealing to deficiencies in atomic physics . ) where are the physicists who will attack the spectrum of the carbon atom with the precision and thoroughness that we seek ? ,
[ n ii ] representing low density nebular gas are also present ( see text).,width=340,height=264 ]
as stated in the introduction , the two principal puzzles presented by the rcbs concern their origins and their ability to produce soot clouds . the hdcs share the puzzle about their origins but not that of soot production
of course , the ` dust ' puzzle might incorporate the hdcs with the rcbs by asking ` why is it that rcbs but not the hdcs , which are on average cooler than the rcbs , are prone to soot production ? '
the basic problem is to understand how a h - deficient star evolves from a h - rich single or binary star ; h - poor he - rich gas clouds capable of sustaining stellar nurseries have not been found .
complete resolution of the problem would show how the classes of h - deficient stars
hdcs - rcb - ehe - beyond are or are not related . presently , there are two principal hypotheses for the origins:(i ) the double - degenerate ( dd ) scenario and ( ii ) a late thermal pulse or final flash ( the ff scenario ) in a luminous star on the white - dwarf cooling track . in the dd scenario ,
a he white dwarf is accreted by a c - o white dwarf as loss of orbital energy drives the white dwarfs to merge .
there are rather specific ranges of initial masses and separations for normal ( h - rich ) stars in a binary to merge and produce a h - deficient he - rich star .
( other combinations produce other odd stars such as a single white dwarf with a mass exceeding chandrasekhar s limit that must go bang in the night or day . ) the post - merger product has the greatly extended envelope required for a supergiant rcb and hdc star .
subsequent evolution follows the canonical post - agb contraction to the white dwarf track ( @xcite ) . in the ff scenario ,
h - rich post - agb single stars that reach the white dwarf cooling track with the potential to experience a final thermal pulse in their he - shell .
under particular conditions , the thin surviving h - envelope may experience h - burning .
onset of the pulse leads to a return of the star to a supergiant - like form but , if conditions are ripe , this supergiant will be h - poor and he - rich .
subsequent evolution takes the supergiant across the hr - diagram to the white dwarf cooling track .
@xcite , @xcite , @xcite , and @xcite among others discuss aspects of the ff scenario .
enigmatic stars such as fg sge and sakurai s object ( v4334 sgr ) are commonly identified as ff scenario products .
quantitative comparison of predictions of the ff and dd scenario and the observations of the rcb and hdc stars involves several dimensions .
our focus here is on the chemical composition obtainable only from examination of the spectrum of rcb and hdc stars .
the ( preferably ) high - resolution spectra from the ultraviolet , optical and near - infrared are analyzed with model atmospheres computed for the appropriate composition ( @xcite ) .
recent elemental abundance studies of c , n , o in particular seem to favour the dd scenario for the rcbs and hdcs ( asplund et al .
2000 ; saio & jeffery 2002 ; pandey et al .
2006 ) .
two remarkable pieces of observational evidence seems to tilt the balance in favour of the dd scenario .
first , @xcite made the dramatic discovery that @xmath8o is more abundant than @xmath9o from moderate resolution spectra of hdc and rcb stars showing the first - overtone co vibration - rotation bands at about 2.3 microns .
second , @xcite for ehes and @xcite for rcbs showed that @xmath10f was drastically overabundant .
the @xmath8o and f results show that the dd scenario was not a quiescent mixing process but either involved nucleosynthesis during the merger and/or subsequently during the evolution of the merged product to and from its supergiant phase .
it is amusing to see how this discovery of @xmath8o was made in hd137613 , as told us by tom geballe .
he got an email from geoffrey clayton at 5:30 p.m on 2004
november 10 : ` hi tom .
i have this spectrum of the first overtone co bands in the hdc star , hd 137613 .
i took a look at it and the bands were split , so i said to myself , oh well there s @xmath11co , but the extra components are nt at the right wavelengths .
so i m a bit mystified .
take a look at the attached plot and see if you have an idea of what this is ' .
to which tom replied : that is amazing ! !
i happen to have a book of co wavelengths that i calculated when i was a grad student in the 70s .
all dv=1 and dv=2 lines of all isotopic species .
these bandheads match 12c18o ! !
the first three bandheads ( 2 - 0 , 3 - 1 , 4 - 2 ) of @xmath12c@xmath8o are at 2.349 , 2.378 , and 2.408 .
i also looked 12c17o and it does nt match .
how could a star have as much @xmath8o as @xmath9o ?
what kind of star is this ?
cheers and congrats , tom ( geballe 2008 ) .
high resolution spectroscopy is essential for properly estimating the isotope ratios @xmath9o to @xmath8o , @xmath9o to @xmath13o ( as well as detection of the weak lines of @xmath10f ) .
garca - hernndez et al .
( 2009 ) used a spectral resolution of 50000 in their analysis of the 2.3 micron co bands .
they confirmed clayton et al.s result of a high @xmath8o overabundance for hdcs and the rcb s aps obtaining a ratio of @xmath9o to @xmath8o of 0.5 in hdcs and the ratio 16 for s aps ( fig.2 ) .
the high resolution optical spectra obtained with the harlan j. smith telescope at the mcdonald observatory and with the vainu bappu telescope at kavalur led to the discovery that @xmath10f is enhanced by a factor of several hundreds to a 1000 times solar in these stars . in passing we note that harlan j. smith and m.k .
vainu bappu were great friends .
c@xmath8o at 2.3485 microns .
the top plots show synthetic spectra computed for the hdc star hd137613 for two co isotopologues and cn .
this figure is from garca - hernndez et al .
, width=340,height=264 ] there can be no doubt that @xmath8o and @xmath10f were synthesized in or subsequent to the creation of the rcbs and hdcs . before full acceptance of the dd scenario is made , the scenario has to account for the nucleosynthesis of these and other observable elemental and isotopic abundances . in principle , as first noted by @xcite , @xmath8o may be synthesized by @xmath14-capture from abundant @xmath15n .
this may occur during the merger in the dd scenario but not with ease in the ff scenario .
the reservoir of @xmath15n is finite and probably not easily restored , thus , one wonders what cuts off the high temperatures at the point that there remains an observable amount of nitrogen .
other complexities remain .
in particular , some rcbs and one hdc show an appreciable amount of lithium , presumably @xmath16li .
discussions of the nucleosynthesis in and following the merger are provided by @xcite , @xcite , and garca - hernndez et al .
additional high - resolution spectroscopy , especially in the infrared , is desirable to determine the o isotopic abundances across the rcb sample .
the ff scenario does plausibly account for other stars .
most notable among the ff candidates are fg sge , v4334 sgr and v605 aql ( clayton et al .
it is not impossible that rcbs from the ff scenario lurk among the known galactic , lmc and smc examples .
historically , rcb stars were the first stars in which circumstellar dust production was invoked to explain brightness variations , i.e. , the dramatic declines of several magnitudes ( @xcite ) .
it is now well established that soot ( carbon ) clouds form close to the star and obstruct the star light reaching us .
such clouds have now been imaged in the infrared around ry sgr at distances of about 1000 stellar radii ( @xcite ) .
these clouds were formed presumably closer to the star ; observational constraints prohibited their detection close to the star .
questions abound concerning the dust . what provides the trigger for dust production in or around the star
how may dust condense in or around such hot stars as rcbs ?
the trigger appears to be related to the atmospheric pulsations which are revealed by light and radial velocity variations .
@xcite noted a correlation between the beginning of a light decline and pulsation phase in ry sgr .
it has now been shown by @xcite for four stars , by longterm monitoring of the brightness , that when a decline happens it does so at a particular phase although not every pulsation results in a decline .
@xcite developed a model for dust formation following a pulsation - induced shock in the upper atmosphere .
when the pulsation amplitudes are large , the temperatures and densities in the post - shock gas are predicted to be conducive for nucleation to occur .
the question intriguing spectroscopists is how may we observe a strong shock develop from a pulsation ? what evidence may we offer that dust condensation actually takes place in the atmosphere of the star ?
can we detect the presence of cool gas from which dust could condense ?
high - resolution spectroscopic observations prior to the onset of the minimum are crucial to studying such questions .
recently , r crb has entered a prolonged light decline that commenced on july 2007 and has not yet ended ( july 2009 ) .
this is one of the deepest minima the star has had in several decades , reaching more than 9 magnitudes . at minimum
the star shows a variety of emission lines including broad ( fwhm of 250 km s@xmath17 ) emission lines of hei , h & k lines of ca ii , nai d lines ( fig.1 ) and sharp ( fwhm @xmath18 15 km s@xmath17 ) emission lines due to feii , tiii etc .
( @xcite ) .
we postpone the discussion of this intriguing emission spectrum for a later occasion .
we were fortunate to catch the star just on the verge of the decline .
our optical spectra obtained at the mcdonald observatory and at vainu bappu observatory have a spectral resolution over 60000 .
3 shows the light curve with the dates of our spectroscopic observations marked .
the spectral changes that r crb underwent even prior to light minimum may provide important clues to the phenomenon of dust production .
the present series of spectra that precede the light minimum starts on 2007 march 4 .
spectra at maximum light are dominated by lines of ci , oi , ni , sii and si in addition to the many lines of neutral and ionized metals seen in spectra of normal f - g supergiants .
the star remained at maximum until 2007 july 10 when the decline in light started . in the following account , we discuss briefly the spectral changes that took place during maximum light .
even four months prior to the onset of the minimum , the star appears to be disturbed with emission in some line cores suggesting a component with an inverse p - cygni profile .
line doubling of the absorption lines is seen two months prior to the onset .
the blue and red components of the doubled lines showed a different level of excitation for the rising gas ( @xmath19 of 4300 k ) from the infalling gas ( @xmath19 of 5600 k ) .
some lines , like tiii 5154 @xmath20 , are not doubled but show emission in the line core .
even the cores of the na i d lines show emission compared to the reference spectrum at ` normal ' maximum light .
moreover , the radial velocity of the emissions is same as that of the star . the absorption lines also broadened without altering the equivalent widths , a month and half before the onset of minimum as though the starlight was passing through a scattering medium ( see fig .
4 ) . study of such line broadening requires high spectral resolution .
some aspects of the spectroscopic changes months ahead of the decline beginning in 2007 july are not unusual .
for example , line doubling appears to be a regular feature of stars such as r crb and ry sgr where the atmospheric pulsation is seen clearly .
it will be of interest to see if these changes occur at the same or similar phases in the pulsation .
perhaps , there is a quasi - steady amplification of the changes at the particular phase lasting a sequence of several pulsations and culminating in the physical conditions amounting to the trigger for a decline .
examination of this idea will call for routine observations of a rcb such as r crb , ry sgr or v854 cen , a star prone to frequent declines . considerable observing time may be wasted in that the declines are unpredictable and irritation of telescope allocation committees seems assured . but
these datasets are most likely to reveal new phenomena that may tighten the link between pulsation and the trigger for dust production .
for example , the spectroscopic variations line asymmetry , doubling , radial velocity and strength may be more extreme at particular phases . more significantly , major disturbances were present in the spectrum of r crb three days prior to the descent .
comparison of pre - maximum and 2007 july 7 spectra shows that most lines in 2007 july 7 spectrum have shifted - absorption components ( fig .
5 ) in addition to the usual photospheric lines . fig .
5 shows a comparison of the spectrum on 2007 july 7 with respect to an undisturbed spectrum obtained on 1995 august 17 illustrating the additional absorption components to several lines that appear in the 2007 july 7 spectrum .
however , the ci lines do not show such additional absorption components .
note that , while the neutral metal lines ( e.g. , fei , and cri ) show strong blue - shifted components , the ionized metal lines ( e.g. , feii , and crii ) show red - shifted components on 2007 july 7 . such differences suggests that rising and falling gases have different levels of ionization . as further shown in fig .
6 during the descent to minimum the blue ( ionized ) and redshifted ( neutral ) components are separated by 53 km s@xmath17 probably as a result of passage of a strong shock . of seemingly particular relevance to dust production , the 2007 july 7 spectrum also displays lines of the c@xmath21 phillips system showing the presence of gas at temperatures ( 1000 to 800 k , figs . 7 and 8) suggestive of a site for dust nucleation .
one supposes these unusual line profiles and the appearance of cool gas are manifestations of the shock induced by the pulsation that triggered the decline . that represents the atmospheric shock velocity .
, width=340,height=264 ] the @xmath12c@xmath21 swan bands are fairly strong in the spectrum of r crb .
bandheads of the 0 - 0 band at 5165 , the 1 - 0 band at 4737 , and the 0 - 1 band at 6174 are readily identified .
more interestingly , the spectrum of 2007 july 7 shows weak bands of the phillips system of c@xmath21 in absorption
see fig . 7 for examples of lines from the 2 - 0 band .
these c@xmath21 lines not present in the regular stellar spectrum are shifted in velocity from the stellar lines and present with a red and a blue component . our spectra show
that the phillips lines were not present on 2007 june 6 spectrum , a month before the onset of the decline .
superposed on the reference spectrum obtained on 1995 august 17 ( blue ) .
absorption lines of the @xmath12c@xmath21 phillips 2 - 0 band are seen as extra absorptions in the 2007 july 7 spectrum but are absent from the 1995 august 17 spectrum .
line doubling in the c@xmath21 lines can also be seen .
, width=340,height=264 ] 2 - 0 lines of the phillips system from the spectrum of r crb obtained on 2007 july 7 .
blue filled circle , open circle and red filled circle symbols refer to q , r and p branch lines , respectively . , width=340,height=264 ] rotational temperatures of 1048@xmath2230 k and 784@xmath2230 k were obtained ( fig . 14 ) for the blue component at @xmath2314.5 km s@xmath17 and the red component at @xmath245 km s@xmath17 , respectively , using the molecular data given by @xcite .
the corresponding molecular column densities are 7.7@xmath2510@xmath15 @xmath26 and 4.5@xmath2510@xmath15 @xmath26 .
the radial velocities of c@xmath27 molecules of @xmath2314.5 km s@xmath17 and @xmath245
km s@xmath17 suggest cool gas rising relative to the star with stellar lines giving the radial velocity of 21 km s@xmath17 .
the cooler gas is less rapidly rising than the warmer molecular gas .
these rotational temperatures show that there was gas at the condensation temperatures conducive to formation of carbon soot .
similar detections of cool c@xmath27 molecules have been reported by us for v854 cen ( @xcite ) , r crb ( @xcite ) , and v cra ( @xcite ) . but these detections were during minimum light .
in contrast , the present detection of cool c@xmath27 molecules at maximum light but on the verge of a minimum is significant and leads to the conclusion that the light minima are indeed caused by formation of dust grains .
woitke et al s ( 1996 ) model seems to be consistent with observations !
intense spectroscopic coverage of the days leading up to a descent to a minimum is an important unfulfilled challenge made more by difficult by the unpredictable onset of the descent .
a mark that distinguishes rcbs from hdcs is the absence from the latter s defining characteristics , the declines deep or even shallow that are such a strong feature of the rcbs .
absence of declines accounts for the lack of an infrared excess from the hdcs .
another distinguishing mark between rcbs and hdcs is the presence of a stellar wind from rcbs but not from hdcs .
are these two marks different sides of one difference between the two groups of h - deficient stars ?
@xcite reported the presence of a p - cygni profile for the hei 10830 line at maximum light for several rcbs .
these observations show that the rcbs possess a stellar wind with an outflow velocity of 200 to 300 km s@xmath17 .
then , rao , lambert & shetrone ( 2006 ) showed that in r crb the strong photospheric lines , particularly the oi 7771 line , had a pronounced blue wing suggesting a component with an expansion velocity of 120 km s@xmath17 .
examination of profiles of other lines showed a component associated with the stellar wind with the velocity proportional to the excitation energy of the lines , i.e. , the range of lines sample the region in the upper atmosphere where the wind is accelerating to the ( possibly terminal ) velocity measured from the hei 10830 line .
our high - resolution spectra now show that a stellar wind may be common , perhaps ubiquitous , among rcbs .
the temporal dependence of the rcb stellar wind is not yet fully known .
there is no indication that it depends greatly or at all on the phase of the pulsation cycle .
long - term behaviour is also not known .
observations at the 19951996 minimum of r crb suggest that the wind - affected wings of the o i lines are undisturbed even though the core is affected by transient emission in the early decline ( rao et al .
what drives this wind ?
is there any connection with magnetic fields ? in the r crb wind , low excitation lines ( e.g. , the ali resonance lines ) show variability on a short time scale but there is no change in the high excitation lines .
examination of magnetically - sensitive and magnetically - insensitive fei lines of comparable strength and excitation potential from a series of high - resolution spectra of r crb taken at maximum light suggests the variability in the former , but not in the latter ( rao 2008 ) .
this result suggests a connection between surface magnetic field and the wind .
the relation between the ( apparently ) permanent and ( seemingly ) pulsation phase - independent wind and trigger for a decline is as yet unknown .
in contrast , @xcite examined the five known hdcs at the hei 10830 line on high - resolution spectra and found no evidence of a stellar wind . the absence of a wind and the lack of deep declines among the hdcs and the presence of a wind among the rcbs with their deep declines is suggestive that the wind or the presence of an extended atmosphere as a result of the wind is a necessary condition for occurrence of deep declines . for al and 120 km s@xmath17 for o. oi line profile of a normal supergiant @xmath28 cyg ( black )
is shown for comaprison .
, width=340,height=264 ]
answers are slowly emerging for the important questions posed by rcb and hdc stars concerning the origin of these h - deficient stars , and the mechanisms by which a cloud of soot forms and obscures the star .
evidence suggests that the dd rather than the ff scenario provides a superior accounting for the elemental abundances of c , n and o for rcb stars and their likely relatives , the extreme helium ( ehe ) stars and cool hydrogen deficient ( hdc ) stars .
the ff scenario does plausibly account for other stars .
most notable among the ff candidates are fg sge and v4334 sgr , also known as sakurai s object and v605 aql ( @xcite ) .
it is not impossible that rcbs from the ff scenario lurk among the analysed sample attributed in the main to the dd scenario .
the process of soot formation is most likely associated with the atmospheric pulsation which , on occasions at a certain phase of the pulsation , leads to a stronger than usual shock in the atmosphere such that the physical conditions are conducive to molecule formation and dust nucleation , as envisaged by woitke et al.(1996 ) .
how the dust grains grow and soot clouds form represent questions for future study . as our essay has hopefully shown , high - resolution optical and infrared spectra have and will prove crucial to addressing the leading questions about the h - deficient rcb and hdc stars . in preparing the paper
, we came across a collection of articles on modern high - resolution spectroscopic techniques .
there , the leading chapter by sir harry kroto carried the title ` old spectroscopists forget a lot but they do remember their lines ' ( @xcite ) .
this pair of old spectroscopists remember at least the important lines .
we acknowledge with thanks the variable star observations from the aavso database .
this research has made use of the simbad database , operated at cds , strasbourg , france .
our sincere thanks to the editors of this volume for accepting with admirable patience our delay in submitting this article .
this research has been supported in part by a grant ( f-634 ) to dll from the robert a. welch foundation of houston , texas .
|
high - resolution spectroscopy is a very important tool for studying stellar physics , perhaps , particularly so for such enigmatic objects like the r coronae borealis and related hydrogen deficient stars that produce carbon dust in addition to their peculiar abundances .
examples of how high - resolution spectroscopy is used in the study of these stars to address the two major puzzles are presented : ( i ) how are such rare h - deficient stars created ? and ( ii ) how and where are the obscuring soot clouds produced around the r coronae borealis stars ? = 1
|
in the current research of nuclear physics and astrophysics , there is of great interest to study the density dependence of the nuclear symmetry energy @xmath0 that essentially characterizes the isospin dependent part of the equation of state ( eos ) of asymmetric nuclear matter .
the exact knowledge on the symmetry energy is important for understanding not only many problems in nuclear physics , but also many critical topics in astrophysics @xcite as well as some interesting issues regarding possible new physics beyond the standard model @xcite . during the last decade , although significant progress has been made both experimentally and theoretically on constraining the density dependence of the symmetry energy @xcite ( see , e.g. , refs .
@xcite for review of recent progress ) , large uncertainties on @xmath0 still exist , especially its super - normal density behavior remains elusive and largely controversial @xcite .
to reduce the uncertainties of the constraints on @xmath0 is thus of critical importance and remains a big challenge in the community , and this provides a strong motivation for studying isospin nuclear physics in radioactive nuclei at the new / planning rare isotope beam facilities around the world , such as csr / lanzhou and brif - ii / beijing in china , ribf / riken in japan , spiral2/ganil in france , fair / gsi in germany , frib / nscl in usa , spes / lnl in italy , and koria in korea .
theoretically , the uncertainties of the constraints on @xmath6 are mainly due to the lack of knowledge about the isospin dependence of in - medium nuclear effective interactions and the limitations in the techniques for solving the nuclear many body problem .
very recently , it has been proposed that it is very useful to directly decompose @xmath0 in terms of some relevant parts of the commonly used underlying nuclear effective interaction in nuclear medium @xcite .
based on the hugenholtz - van hove ( hvh ) theorem , indeed , the @xmath0 can be decomposed analytically in terms of the single - nucleon potential in asymmetric nuclear matter and the resulting expressions are quite general and independent of the detailed nature of the nucleon interactions , providing an important and physically more transparent approach to extract information on the symmetry energy from the isospin dependence of strong interaction in nuclear medium and understand why the symmetry energy predicted from various models is so uncertain @xcite . in these works , the decomposition of @xmath0
is based on non - relativistic framework .
it is thus of great interest to explore more general decomposition within relativistic covariant framework , which is the main motivation of the present work .
the relativistic covariant formulation has made great success during the last decades in understanding many nuclear phenomena @xcite . in particular ,
the microscopic relativistic covariant dirac - brueckner - hartree - fock ( dbhf ) approach @xcite has achieved impressive success in describing the saturation properties of nuclear matter without any need to introduce a three - nucleon force required in the microscopic non - relativistic bhf calculations ( see , e.g. , refs .
it has been argued that in non - relativistic calculations the three - nucleon forces must be introduced to mimic the variation of the dirac spinors in the nuclear medium contained in relativistic covariant approach @xcite .
in addition , the lorentz covariant decomposition of the nuclear mean field potential has been shown to be very important for understanding the dynamics in heavy ion collisions at relativistic energies @xcite .
these features imply that the lorentz covariance could be important for understanding the higher energy / density nuclear phenomena , e.g. , the high density behaviors of the symmetry energy . owing to the translational , rotational and time - reversal invariance , parity conservation , and hermiticity , the lorentz covariant nucleon self - energy in the rest frame of asymmetric nuclear matter with baryon density @xmath9 and isospin asymmetry @xmath10 can be written generally as @xcite , @xmath11where @xmath12 is isospin index ; @xmath13 is the scalar self - energy , @xmath14 and @xmath15 are , respectively , the zeroth component and the space component of the vector self - energy @xmath16 , and they all generally depend on @xmath17 , @xmath18 and the magnitude of the nucleon momentum @xmath19 ( minkowski metric is @xmath20 ) ; @xmath21 is the unit vector of momentum @xmath22
. a proof of eq . ( [ defse2 ] ) can be found in ref .
accordingly , the single - nucleon energy can be expressed as @xcite @xmath23with @xmath24 ^{2}+\left [ m_{j}^{\ast } ( \rho , \alpha , |\mathbf{k}|)\right ] ^{2}},\]]where the nucleon effective ( dirac ) mass @xmath25 and effective momentum @xmath26 are defined , respectively , as@xmath27with @xmath28 being the nucleon mass . due to the exchange symmetry between protons and neutrons in nuclear matter , the eos of asymmetric nuclear matter , defined by its binding energy per nucleon , can be expanded as a power series of even - order terms in @xmath18 as @xmath29where @xmath30 is the eos of symmetric nuclear matter , and the symmetry energy is expressed as @xmath31around the nuclear matter saturation density @xmath32 , the symmetry energy can be expanded as @xmath33where @xmath34 is a dimensionless variable , and @xmath35 is the density slope parameter of the symmetry energy at @xmath32 and thus carries important information on the symmetry energy at both high and low densities . generally , the slope parameter of the symmetry energy at arbitrary density @xmath17 is defined as @xmath36 according to the hvh theorem @xcite , the nucleon chemical potential in asymmetric nuclear matter should be equal to its fermi energy ( the single - nucleon energy at fermi surface ) , i.e. , @xmath37}{\partial \rho _ { j}}+m,\ ] ] where @xmath38 is the nucleon fermi energy , and
@xmath39 ( we assume @xmath40 and @xmath41 in this work ) is the nucleon fermi momentum with @xmath42 being the fermi momentum in symmetric nuclear matter at density @xmath17 .
it should be noted that the hvh theorem is independent of the detailed nature of the interactions used and is valid for any interacting self - bound infinite fermi system , such as infinite nuclear matter @xcite . expanding @xmath43 as a power series of even - order terms in @xmath18 on the right - hand side of eq .
( [ hvhanm ] ) , we can obtain @xmath44}{\partial\rho}+2 m \notag \\ + & 2\frac{\partial}{\partial\rho}\sum_{i=1}\rho e_{\text{sym},2i}(\rho)\alpha^{2i } , \label{detl}\end{aligned}\ ] ] where @xmath45[{\partial ^{2i}e(\rho , \alpha ) } /{\partial \alpha ^{2i}}]_{\alpha = 0}$ ] are the symmetry energies of different orders and particularly we have @xmath46 .
furthermore , expanding @xmath47 as a power series of @xmath48 on the left - hand side of eq .
( [ detesym ] ) and eq .
( [ detl ] ) , and comparing the coefficients of the first - order @xmath48 terms on both left- and right - hand sides of eq .
( [ detesym ] ) , we then obtain @xmath49\right|_{\alpha=0 } , \label{foresym}\end{aligned}\ ] ] while comparing the coefficients of second - order @xmath48 terms on both sides of eq .
( [ detl ] ) leads to the following expression : @xmath50\right|_{\alpha=0 } + 3e_{\text{sym}}(\rho ) .
\label{forl}\end{aligned}\ ] ] substituting eq .
( [ disrelaanm ] ) into eq .
( [ foresym ] ) , we can obtain @xmath51 where @xmath52 , @xmath53 and @xmath54 ( here @xmath55 denotes k , s or v ) represent , respectively , the contributions from the kinetic part , the momentum dependence of the nucleon self - energies in symmetric nuclear matter and the first - order symmetry self - energies , and they can be expressed analytically as @xmath56where @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , @xmath62 , and the @xmath63-th order symmetry self - energy is defined as ( here @xmath64 ) @xmath65 \right\vert _ { \alpha = 0}. \label{symmse}\ ] ] furthermore , eq . ( [ expresym ] ) can be rewritten as @xmath66 where @xmath67 is the nucleon landau mass in symmetric nuclear matter , i.e. , @xmath68^{-1}$ ] ( see , e.g. , ref .
@xcite ) with @xmath69 , and one can easily verify the relation @xmath70 .
in this way , we have decomposed analytically the symmetry energy @xmath71 in terms of the lorentz covariant nucleon self - energies in asymmetric nuclear matter . similarly , by substituting eq .
( [ disrelaanm ] ) into eq .
( [ forl ] ) , the slope parameter @xmath72 can be decomposed as @xmath73 with @xmath74 @xmath75_{|\mathbf{k}|=k_{\text{f } } } \notag \\ & + \frac{k_{\text{f}}}{6}\left[\frac{k_{\text{f}}^{\ast}}{\mathcal{e}_{\text{f}}^{\ast } } \frac{\partial\sigma_{\text{k}}^0 } { \partial|\mathbf{k}| } + \frac{m_0^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\frac{\partial\sigma_{\text{s}}^0}{\partial|\mathbf{k}| } + \frac{\partial\sigma_{\text{v}}^0 } { \partial|\mathbf{k}|}\right]_{|\mathbf{k}|=k_{\text{f } } } \notag \\ & + \frac{k_{\text{f}}^2}{6\mathcal{e}_{\text{f}}^{\ast3}}\left[m_0^{\ast2}\left ( \frac{\partial\sigma_{\text{k}}^0}{\partial |\mathbf{k}|}\right)^2+k_{\text{f}}^{\ast2}\left(\frac{\partial\sigma_{\text{s}}^0 } { \partial |\mathbf{k}|}\right)^2\right]_{|\mathbf{k}|=k_{\text{f } } } \notag \\ & -\frac{k_{\text{f}}^2k_{\text{f}}^{\ast}m_0^{\ast } } { 3\mathcal{e}_{\text{f}}^{\ast3}}\left [ \frac{\partial\sigma_{\text{s}}^0}{\partial \label{deflmden } \\ &
l^{\mathrm{1st}}(\rho)=\frac{3}{2\mathcal{e}_{\text{f}}^{\ast3}}\left[m_0^{\ast}\sigma_{\text{k}}^{\mathrm{{sym},1}}- k_{\text{f}}^{\ast}\sigma_{\text{s}}^{\mathrm{{sym},1}}\right]^2 \notag \\ & + \frac{3}{2}\left[\frac{k_{\text{f}}^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\sigma_{\text{k}}^{\mathrm{{sym},1}}+ \frac{m_0^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\sigma_{\text{s}}^{\mathrm{{sym},1 } } + \sigma_{\text{v}}^{\mathrm{{sym},1}}\right ] \notag \\ & + \frac{k_{\text{f}}m_0^{\ast2}\sigma_{\text{k}}^{\text{sym,1 } } } { \mathcal{e}_{\text{f}}^{\ast3 } } -\frac{k_{\text{f}}k_{\text{f}}^{\ast}m_0^{\ast}\sigma_{\text{s}}^{\text{sym,1 } } } { \mathcal{e}_{\text{f}}^{\ast3 } } , \label{defls1 } \\ & l^{\text{cross}}(\rho)= \notag \\ & k_{\text{f}}\left[\frac{k_{\text{f}}^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\frac{\partial \sigma_{\text{k}}^{\mathrm{{sym},1}}}{\partial |\mathbf{k}|}+\frac{m_0^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\frac{\partial \sigma_{\text{s}}^{\mathrm{{sym},1}}}{\partial|\mathbf{k}|}+\frac{\partial \sigma_{\text{v}}^{\mathrm{{sym},1}}}{\partial |\mathbf{k}|}\right]_{|\mathbf{k}|=k_{\text{f } } } \notag \\ &
-\frac{k_{\text{f}}\sigma_{\text{k}}^{\mathrm{{sym},1}}}{\mathcal{e}_{\text{f}}^{\ast } } \left[\frac{k_{\text{f}}^{\ast2}}{\mathcal{e}_{\text{f}}^{\ast2}}\left(\frac{\partial\sigma_{\text{k}}^0 } { \partial|\mathbf{k}|}+\frac{m_0^{\ast}}{k_{\text{f}}^{\ast}}\frac{\partial\sigma_{\text{s}}^0 } { \partial|\mathbf{k}|}\right ) -\frac{\partial\sigma_{\text{k}}^0 } { \partial|\mathbf{k}|}\right]_{|\mathbf{k}|=k_{\text{f } } } \notag \\ & -\frac{k_{\text{f}}\sigma_{\text{s}}^{\mathrm{{sym},1}}}{\mathcal{e}_{\text{f}}^{\ast } } \left [ \frac{m_0^{\ast2}}{\mathcal{e}_{\text{f}}^{\ast2}}\left(\frac{k_{\text{f}}^{\ast}}{m_0^{\ast } } \frac{\partial\sigma_{\text{k}}^0 } { \partial|\mathbf{k}|}+\frac{\partial\sigma_{\text{s}}^0 } { \partial|\mathbf{k}|}\right)-\frac { \partial\sigma_{\text{s}}^0 } { \partial|\mathbf{k}|}\right]_{|\mathbf{k}|=k_{\text{f } } } , \label{deflcross } \\ & l^{\mathrm{2nd}}(\rho)=3\left[\frac{k_{\text{f}}^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\sigma_{\text{k}}^{\mathrm{{sym},2 } } + \frac{m_0^{\ast}}{\mathcal{e}_{\text{f}}^{\ast}}\sigma_{\text{s}}^{\mathrm{{sym},2}}+\sigma_{\text{v}}^{\mathrm{{sym},2}}\right ] .
\label{defls2}\end{aligned}\ ] ] on the right - hand side of eqs .
( [ deflkin])-([defls2 ] ) , the density and momentum dependence have been suppressed with @xmath76 ( @xmath64 ) .
( [ expresym ] ) ( or ( [ esymdecomp ] ) ) and eq .
( [ exprl ] ) are two main results of this work .
the nucleon self - energies can be calculated theoretically from a certain relativistic covariant approach or extracted experimentally ( around @xmath77 ) from the dirac phenomenology of nucleon - nucleus scattering .
the lorentz covariant nucleon self - energy decompositions of @xmath78 in eq .
( [ expresym ] ) ( or ( [ esymdecomp ] ) ) and @xmath72 in eq .
( [ exprl ] ) are general and they are useful for determining the density dependence of the symmetry energy and understanding its lorentz structure and the microscopic origin . as an example , we consider here the nonlinear @xmath2-@xmath3-@xmath4-@xmath5 relativistic mean field ( rmf ) model which is based on effective interaction lagrangians involving nucleon and meson fields , and has been widely discussed in the literature ( see , e.g. , ref .
a very useful feature of this model is that the nucleon self - energies in asymmetric nuclear matter can be obtained analytically and this makes our analysis physically transparent .
the lagrangian density of the nonlinear @xmath2-@xmath3-@xmath4-@xmath5 rmf model can be expressed as ( see , e.g. , ref .
@xcite ) : @xmath79\psi \notag \\ + & \frac{1}{2}\left(\partial_{\mu}\sigma\partial^{\mu}\sigma - m_{\sigma}^2\sigma^2\right ) -\frac{1}{4}f_{\mu\nu}f^{\mu\nu}+\frac{1}{2}m_{\omega}^2\omega_{\mu}\omega^{\mu } \notag \\
-&\frac{1}{3}b_{\sigma}m(g_{\sigma}\sigma)^3 -\frac{1}{4}c_{\sigma}(g_{\sigma}\sigma)^4+\frac{1}{4}c_{\omega}(g_{\omega}^2\omega_{\mu}\omega^{\mu})^2 \notag \\ + & \frac{1}{2}\left(\partial_{\mu}\vec{\mkern1mu\delta}\cdot\partial^{\mu}\vec{\mkern1mu\delta } -m_{\delta}^2\vec{\mkern1mu\delta}^2\right ) + \frac{1}{2}m_{\rho}^2\vec{\mkern1mu\rho}_{\mu}\cdot\vec{\mkern1mu\rho}^{\mu } -\frac{1}{4}\vec{\mkern1mu g}_{\mu\nu}\cdot\vec{\mkern1mu g}^{\mu\nu } \notag \\ + & \frac{1}{2}\left(g_{\rho}^2\vec{\mkern1mu\rho}_{\mu}\cdot\vec{\mkern1mu\rho}^{\mu}\right)\left(\lambda_{\text{s}}g_{\sigma}^2\sigma^2 + \lambda_{\text{v}}g_{\omega}^2\omega_{\mu}\omega^{\mu}\right ) \notag
\\ -&g_{\rho}\vec{\mkern1mu\rho}_{\mu}\cdot\bar{\psi}\gamma^{\mu}\vec{\tau}^j\psi + g_{\delta}\vec{\mkern1mu\delta}\cdot\bar{\psi}\vec{\tau}^j\psi,\end{aligned}\ ] ] where @xmath80 and @xmath81 are strength tensors of @xmath3 field and @xmath4 field , respectively while @xmath82 , @xmath2 , @xmath83 , @xmath84 and @xmath85 are nucleon field , isoscalar - scalar field , isoscalar - vector field , isovector - vector and isovector - scalar field , respectively , and the arrows denote the isospin vector . @xmath86 and @xmath87 are two cross - coupling constants for varying the density dependence of @xmath88 , and @xmath89 , @xmath90 , @xmath91 , and @xmath92 are masses of mesons
. in the rmf model , meson fields are replaced by their expectation values , i.e. , @xmath93 , @xmath94 , @xmath95 , where the subscript @xmath96 `` denotes the zeroth component of the four - vector while the superscript ( @xmath97 ) '' denotes the third component of isospin , furthermore , the space - like self - energy @xmath98 vanishes ( due to the hartree approximation in the rmf model ) while the scalar and time - like self - energies are momentum independent , i.e. , @xmath99 the symmetry energy then can be decomposed as @xmath100where the ( @xmath101st - order ) symmetry self - energies are @xmath102 with @xmath103 , @xmath104 being the scalar density , and @xmath105 .
we note that the above analytical expression for @xmath88 is exactly the same as the one obtained from the normal approach ( see , e.g. , ref .
similarly , the slope parameter @xmath72 can be decomposed as @xmath106 with @xmath107 \notag \\ + & \frac{3k_{\text{f}}^2}{2\mathcal{e}_{\text{f}}^{\ast3}}\left[\sigma_{\text{s}}^{\mathrm{{sym},1}}(\rho ) \right]^2-\frac{m_0^{\ast}k_{\text{f}}^2\sigma_{\text{s}}^{\mathrm{{sym},1}}(\rho)}{\mathcal{e}_{\text{f}}^{\ast3 } } , \label{rmfldecom1stsymm1 } \\
l^{\mathrm{2nd}}(\rho)=&3\left[\frac{m_0^{\ast}\sigma_{\text{s}}^{\mathrm{{sym},2}}(\rho)}{\mathcal{e}_{\text{f}}^{\ast}}+\sigma_{\text{v}}^{\mathrm{{sym},2}}(\rho)\right ] , \label{rmfldecom2ndsymm1}\end{aligned}\ ] ] where the @xmath108nd - order symmetry self - energies are @xmath109 with @xmath110 and @xmath111 . to the best of our knowledge ,
the above formulas give , for the first time , the analytical expression of the slope parameter @xmath72 in the nonlinear rmf model . the above analytical expressions of @xmath88 and @xmath72 can be easily generalized to the case of the density dependent rmf model that has similar isospin structure as the nonlinear rmf model @xcite .
it should be mentioned that these analytical expressions for @xmath88 and @xmath72 are very useful for determining the isovector parameters in the rmf model by fitting the empirical properties of asymmetric nuclear matter ( see , e.g. , ref .
@xcite for such a procedure in the case of the isoscalar sector ) . and its self - energy decomposition according to eq .
( [ rmfesymdecom1 ] ) in the nonlinear rmf model with different interactions.,width=302 ] shown in fig .
[ figesymdecom ] is the density dependence of @xmath88 and its self - energy decomposition according to eq .
( [ rmfesymdecom1 ] ) for four interactions , i.e. , fsugold @xcite , iu - fsu @xcite , nl@xmath112 @xcite and ha @xcite .
fsugold and iu - fsu are two accurately calibrated interactions based on the ground state properties of closed - shell nuclei , their linear response , and the structure of neutron stars .
since fsugold and iu - fsu do not consider the isovector - scalar @xmath5 meson field , one thus has @xmath113 . on the other hand , both nl@xmath112 and
ha include the @xmath5 meson field with the latter fitting successfully some results obtained from the microscopic dbhf approach while the former fitting the empirical properties of asymmetric nuclear matter and describing reasonably well the binding energies and charge radii of a large number of nuclei @xcite . from fig .
[ figesymdecom ] , one can see that while the kinetic contribution @xmath52 is roughly the same for different interactions , the different interactions predict significantly different values for @xmath114 and @xmath115 .
in particular , compared with fsugold and iu - fsu , ha predicts very similar total @xmath88 but significantly different @xmath114 and @xmath115 .
similarly , we show in fig .
[ figldecom ] the density dependence of the slope parameter @xmath72 and its self - energy decomposition according to eq .
( [ rmfldecom ] ) for fsugold , iu - fsu , nl@xmath112 and ha .
again , it is seen that the different interactions predict roughly same kinetic contribution @xmath116 but significantly different values for @xmath117 and @xmath118 . in particular
, one can see that the higher - order contribution @xmath118 from the second - order symmetry self - energies generally can not be neglected , agreeing well with the recent non - relativistic calculations @xcite . and its self - energy decomposition according to eq .
( [ rmfldecom ] ) in the nonlinear rmf model with different interactions.,width=302 ]
using the hugenholtz - van hove theorem , we have shown that the symmetry energy @xmath0 and its density slope @xmath1 can be decomposed analytically in terms of the lorentz covariant nucleon self - energies in asymmetric nuclear matter , and the corresponding expressions have been derived for the first time .
these general expressions for the covariant self - energy decomposition of @xmath0 and @xmath1 are useful for determining the density dependence of the symmetry energy , deciphering the lorentz structure of the symmetry energy , and understanding the microscopic origins of the symmetry energy . as an example
, we have analyzed the lorentz covariant nucleon self - energy decomposition of @xmath0 and @xmath1 within the nonlinear @xmath2-@xmath3-@xmath4-@xmath5 relativistic mean field model and derived the corresponding analytical expressions for @xmath0 and @xmath1 , which are potentially useful for fixing the isovector parameters in the rmf model from fitting the empirical properties of asymmetric nuclear matter . from analyzing the self - energy decomposition of @xmath119 and @xmath1 within the nonlinear @xmath2-@xmath3-@xmath4-@xmath5 relativistic mean field model
, we have found that the results strongly depend on the interactions used and also whether the isovector - scalar @xmath120 meson is included or not .
these results imply that it is of great importance to determine individually each part of the lorentz covariant nucleon self - energy decomposition of @xmath0 and @xmath1 from experiments ( e.g. , dirac phenomenology ) or microscopic calculations based on nucleon - nucleon interactions derived from scattering phase shifts ( e.g. , dbhf ) . on the other hand ,
the lorentz covariant nucleon self - energies in asymmetric nuclear matter can also be determined from quantum chromodynamics ( qcd ) by means of qcd sum - rule techniques @xcite .
the general expressions of the lorentz covariant nucleon self - energy decomposition of @xmath0 and @xmath1 presented in this work are thus very useful for determining the symmetry energy from qcd .
these studies are in progress .
the authors would like to thank r. chen , b.a .
li and c. xu for useful discussions .
this work was supported in part by the nnsf of china under grant nos .
10975097 and 11135011 , the shanghai rising - star program under grant no .
11qh1401100 , the shu guang " project supported by shanghai municipal education commission and shanghai education development foundation , the program for professor of special appointment ( eastern scholar ) at shanghai institutions of higher learning , the science and technology commission of shanghai municipality ( 11dz2260700 ) , and the national basic research program of china ( 973 program ) under contract no .
|
using the hugenholtz - van hove theorem , we derive analytical expressions for the nuclear symmetry energy @xmath0 and its density slope @xmath1 in terms of the lorentz covariant nucleon self - energies in isospin asymmetric nuclear matter .
these general expressions are useful for determining the density dependence of the symmetry energy and understanding the lorentz structure and the microscopic origin of the symmetry energy in relativistic covariant formulism . as an example
, we analyze the lorentz covariant nucleon self - energy decomposition of @xmath0 and @xmath1 and derive the corresponding analytical expressions within the nonlinear @xmath2-@xmath3-@xmath4-@xmath5 relativistic mean field model .
|
integrated optical waveguide technologies offer improvements in size , scalability , integrability and nonlinear interaction strength over their bulk counterparts .
these properties have been used to create a myriad of impressive devices , such as high - speed switches @xcite , optical parametric oscillators @xcite and frequency conversion devices @xcite .
frequency conversion is often used to produce laser fields at wavelengths inaccessible to current materials @xcite , but can also be used to interface individual parts of a network @xcite or to transfer information from one field to another at a different wavelength @xcite . traditionally , the main advantage of waveguide devices has been very high nonlinear interaction strengths , allowing for relatively high conversion efficiencies at low input and low output powers .
however , for many applications both high conversion efficiency and high output powers are desired .
this is where the performance of waveguides has traditionally been limited , due to a number of factors such as large waveguide losses , poor mode shape profile and high power effects such as photorefraction .
only recently have large improvements in stable , high conversion efficiency and high output power waveguides been realised . the most common architecture used to this aim
is mgo : linbo@xmath0 ridge waveguides @xcite .
in contrast to other architectures , the ridge waveguide design allows for magnesium oxide doping , which significantly reduces the photorefractive effect @xcite .
for example sun _
et al _ have recently produced single - pass waveguides with internal second harmonic conversion efficiencies of around 70% and with output powers exceeding 400mw at 466 nm @xcite .
one can further enhance the strength of the nonlinear interaction by resonating the optical field within the waveguide .
this method of enhancement has been demonstrated by various groups in a number of different architectures @xcite .
resonators also allow one to shape the spectrum of the output light and to tailor the system for optimum conversion for a given pump power and waveguide loss @xcite . despite these benefits investigation of waveguide resonators for _ high power _ and _ high conversion efficiency _
applications has not been experimentally realised .
this is likely due to increased sensitivity to temperature fluctuations and waveguide losses .
however , waveguide losses less than 0.1db / cm in both titanium indiffused and ridge waveguides have been demonstrated in lithium niobate @xcite .
this opens up the possibility of producing high performance waveguide resonators . in this paper
we present the realisation of a high power waveguide resonator , whose external second harmonic conversion efficiency ( efined here as the amount of second harmonic power exiting the device divided by the amount of fundamental power incident on the device ) of 75% is , to the best of our knowledge , the highest achieved from an integrated device .
the device stably produces up to 110mw of sh power , after which the output power begins to fluctuate in time , possibly due to photorefraction .
the cavity length has also been locked using a pound - drever - hall ( pdh ) type locking scheme @xcite , stabilising the output power over many hours . whilst the wavelength of the converted light is easily accessible by various laser technologies , this device provides access to a phase congruent fundamental ( which lies in the communications band ) and second harmonic field .
this is useful , for example , in nonlinear optics experiments that pump any process with the second harmonic of the field such as parametric amplification and oscillation @xcite , as well as detection of these states using , for example , balanced homodyne detection @xcite .
the performance of the waveguide shg resonator will depend on many factors such as loss , nonlinear strength , photothermal properties , photorefraction and waveguide mode shape .
in particular , the reduction of losses becomes a critical factor in resonator devices .
titanium indiffused waveguides have the lowest losses seen from any nonlinear waveguide technology , with reliably measured ( using cavity resonances @xcite ) losses as low as 0.02db / cm @xcite .
ridge waveguides are not far behind , with measured losses as low as 0.1db / cm , @xcite . the increased losses of ridge waveguides is , however , compensated in part by a larger nonlinearity .
ridge waveguides have demonstrated second harmonic conversion efficiencies of around 240%w@xmath1@xmath2 ( for sh at 340 nm ) @xcite whilst the titanium indiffused waveguides produced in our group have shown second harmonic conversion efficiencies of up to 14%w@xmath1@xmath2 ( sh at 775 nm ) .
the maximum output power of a device will be decided by thermal and photorefraction effects within the waveguide . for indiffused waveguides
it is possible to reduce photorefraction via sample heating , whereas doping techniques can be utilised in ridge waveguides @xcite . due to the lower losses expected from titanium indiffused waveguides ,
the resonator was produced using this technology .
indiffused waveguides also have a very high spatial overlap with single mode fibres ( of up to around 94% ) .
the disadvantage of titanium indiffused waveguides is the presence of photorefraction .
photorefraction is the migration of free carriers due to impurities in the wafer , typically fe , that act to vary the electric field across the sample @xcite . due to this fact , it is virtually impossible to estimate the effect of photorefraction on device stability .
even with relatively small input powers this effect can cause a change in the refractive index of the material over time scales of tens of minutes as the dynamics stabilise @xcite . whilst this effect can be compensated for , at higher powers
it is expected that catastrophic photorefraction will occur and will produce large amounts of noise on the produced shg @xcite .
waveguides are fabricated in z - cut linbo@xmath0 by an indiffusion of lithographically patterned 7 @xmath3 m wide , 80 nm thick titanium strips .
the diffusion is performed at 1060 @xmath4c for 9 hours in oxygen atmosphere . in a subsequent second lithography step ,
an insulating photoresist pattern is defined which is used for field assisted periodic domain inversion .
we have chosen a poling period of approximately 16.9 @xmath3 m to achieve the desired type 0 quasi - phase matching , where all interacting fields are in the tm polarization .
a short sample length of 8 mm long is chosen as a smaller sample is easier to stabilise in temperature and ensures that phase matching is uniform across the entire sample . in order to optimise the conversion efficiency , one needs to correctly choose the mirror coating reflectivities .
we choose to have the second harmonic exit the rear of the sample whilst the pump enters through the front .
the rear surface therefore has an anti - reflection ( ar ) coating at 775 nm and a high - reflectivity ( hr ) coating for 1550 nm .
the front surface is coated with a 77% mirror at 1550 nm and a hr coating at 775 nm .
the 77% coating is chosen to achieve critical coupling of the pump assuming intra - cavity losses of 0.07db / cm @xcite .
the coatings for the device are produced in - house using ion - beam assisted vapour deposition .
glass `` blanks '' are coated in the same coating run and from these one can accurately determine the properties of the coatings .
the hr coatings are expected to have a reflectivity greater than 99@xmath5 , and ar coatings a reflectivity of around 0.5@xmath5 .
we begin by measuring the linear properties of the device at the fundamental wavelength of 1550 nm .
the experimental layout is shown in figure [ experiment ] .
the waveguide is pumped by a tunics tunable c - band laser that is amplified using an edfa .
the power is fine - tuned using a half - wave plate followed by a polarising beam splitter . a dichroic mirror before the input to the waveguide can be used to measure the amount of second harmonic light leaking from the coating on the front surface .
a second dichroic mirror at the exit of the waveguide separates the transmitted pump light from the generated sh light .
the amount of power in this field is then attenuated and measured using a photodetector . for accurate conversion efficiency measurements the coupling between the (
fibre - coupled ) laser mode and the waveguide mode is first precisely determined .
the coupling of the laser beam into the mode of the waveguide is optimised and the light transmitted through the cavity is then coupled into a fibre as shown in figure [ experiment ] .
the laser light is then disconnected and reconnected to this rear fibre - coupling and the light is shone through the waveguide in the reverse direction . by removing the faraday isolator one can measure the amount of power before and after the first fibre - coupling to achieve an accurate measure for the overlap . from this measurement , the overlap between the fibre mode and the waveguide mode
is determined to be 0.93@xmath6 .
the fundamental wavelength resonator properties are determined at a wavelength 10 nm away from the phasematching temperature , well outside the phase matching bandwidth but near enough to the operational wavelength such that the properties of the coatings do not significantly vary . the wavelength is then scanned over a cavity resonance , allowing one to measure the transmitted power on - resonance and the amount of power on and off - resonance in reflection . from these parameters
the loss of the waveguide and the reflectivity of the cavity hr mirror can be precisely determined if the reflectivity of the front mirror and the incident power is precisely known .
approximately 300@xmath7w of power is coupled into the waveguide cavity . on reflection
we find that the power on resonance is 36@xmath8 of the power off resonance and the transmitted power on resonance is found to be 6.3@xmath9 of the power entering the cavity . calculating the best fit using the standard treatment ( see for example @xcite ) one finds that the intra - cavity loss at the fundamental wavelength is 0.16@xmath10db / cm and the reflectivity of the rear mirror at the fundamental wavelength is 99.4@xmath11 .
these values result in an expected ( off phase matching ) cavity finesse of 20 , a free spectral range of 8.7 ghz and a cavity linewidth of 440 mhz .
the loss is unfortunately higher than desired , but it is necessary to find a waveguide with the right phase matching conditions , high nonlinear efficiency and low losses .
this waveguide was the best compromise of the three parameters on this particular device .
with the fundamental cavity well characterised , we now turn our attention to the nonlinear properties of the device .
the fundamental power entering the waveguide is varied through rotation of a half - waveplate in the beam path and the fibre amplifier gain .
the wavelength of the laser is then scanned such that the optimal cavity resonance , the one that is closest to the centre of the phase - matching curve , is found .
the wavelength of the laser is then adjusted at each power setting to ensure that thermal effects do not shift the cavity from the phase - matching condition .
the generated second harmonic power on resonance is detected on a photodiode after attenuation via a calibrated neutral density filter .
s@xmath12 ) measured at the input powers shown in the inset .
waveguide loss at the sh is assumed to be equal to that found for the fundamental ( of 0.16db / cm ) . ]
figure [ conveff ] shows the results of the second harmonic power run .
the dotted line shows the expected performance of the device , where the only free parameter is the nonlinear coupling strength . in order to uncouple possible high power effects , such as photorefraction
, the nonlinear coupling strength was found from a fit with low input power , as shown in the inset . at these low powers
the nonlinear coupling strength is found to be @xmath13 s@xmath12 , matching very well with the data .
this value for the nonlinear efficiency was then used , along with the previously determined losses of the system to create the model indicated by the dotted line . of note
is that the output of the waveguide resonator is in strong agreeance with the predicted theory .
this indicates that power related effects that may reduce the conversion efficiency , such as photorefraction , are not significant at these power levels and at this operating temperature . whilst photorefraction does not seem to reduce the produced sh power in the power range shown in figure [ conveff ] ,
the output power was seen to fluctuate at fundamental input powers of 140mw and above .
this is perhaps an indicator of the onset of catastrophic photorefraction for this particular waveguide .
however , due to the nature of the impurities , this limit will change from waveguide to waveguide and sample to sample .
nevertheless , we have shown that it is possible to produce large levels of second harmonic power from titanium indiffused waveguides .
the high power performance of the device is limited by photothermal and photorefractive effects .
these effects are difficult to distinguish between because they are both expected to shift the cavity resonance in a similar fashion .
it is expected that as the device produces more power the magnitude of these effects will increase and the task of keeping the resonator on resonance will become more difficult . whilst it is difficult to get a quantitative analysis of these effects , it is possible to look at them qualitatively . by scanning the frequency of
the laser one can scan over the cavity resonance approaching from both high and low frequency sides whilst detecting the transmitted power .
the presence of photothermal heating and/or photorefraction will produce a cavity scan that is asymmetric .
such a cavity scan is shown in figure [ cavscan ] , where approximately 110mw of fundamental light is incident on the cavity .
the direction of the laser frequency scan ( scanned using a sawtooth function ) is reversed at around zero seconds .
the black trace shows the transmitted fundamental power when the cavity temperature is held at phase matching , resulting in a second harmonic power of around 70mw .
we note that the airy function at this power level is not symmetric , indicating some minor amount of photothermal and/or photorefractive effects . to investigate the wavelength dependence of the asymmetry the frequency of the laser was varied , such that the device was far from phase matching and no second harmonic power was produced .
the transmitted power was again measured as the cavity was scanned and is shown by the blue trace .
we immediately see that the cavity bandwidth has reduced due to the absence of nonlinear interaction , as expected .
perhaps surprisingly , however , a large asymmetry is still seen .
photorefraction is typically expected to have a very strong dependence on wavelength , whilst one might expect absorption to have a slight dependence on wavelength over this range .
therefore the observed results are consistent with a similar absorption at both wavelengths and only a minor contribution from photorefraction effects on the timescales of the scan .
this is perhaps a surprising result and will be investigated further .
the stability of frequency conversion devices is a critical factor in many applications .
one requires device stability that is at least equal to the required measurement time .
therefore , a preliminary investigation into device stability over longer time periods is undertaken .
due to the narrow linewidth of the cavity resonance this requires very stable , fine temperature tuning and for greater stability , some form of feedback that locks to the cavity resonance .
the issue of stability is addressed in two steps .
the first step is to optimise the standard temperature loop .
the 2-stage oven is enclosed in a teflon housing and is then encased in a larger housing with openings only for the input and output fields .
with this setup the temperature of the cavity is stable to a couple of millikelvin or better and this is enough to ensure that the cavity resonance is held , at low input powers and without the pdh loop , on the minute time scale .
this scheme is limited because it does not sense the resonance condition . as such , system drifts may force the cavity off resonance in a way that is not sensed by the temperature sensor in this loop and are therefore not compensated .
a second step is required to provide stability over longer time scales and at higher powers .
the second step is the pdh locking loop which directly senses the resonance condition , thereby providing a means of compensating system drift over longer times .
the field entering the waveguide resonator is phase modulated via an electro - optic modulator at 25 mhz , as shown in figure [ experiment ] .
these sidebands are well within the cavity linewidth and as such are transmitted through the cavity with the carrier and are detected at a photodiode .
this signal is then fed into a toptica digilock system , therein undergoing mixing and filtering to produce the error signal .
this error signal is then used to lock the system to resonance via feedback .
it is possible to achieve this feedback via two methods ; the first is by feeding back to the temperature controller to provide an offset , and the second by feeding back to the laser frequency . feeding back to the laser provides a much faster lock but results in the centre frequency of the laser shifting , which may be an unwanted effect for some applications .
the performance of both locking schemes is illustrated in figure [ lock ] .
the temperature feedback lock is performed with approximately 30@xmath14mw of fundamental field entering the cavity , resulting in approximately 10@xmath14mw of second harmonic power whilst the laser feedback scheme has approximately @xmath15mw of fundamental field entering the cavity and 9@xmath14mw second harmonic power exiting the system . as expected
, the temperature feedback scheme does not perform as well as the laser frequency feedback , an accumulated offset over time is observed , likely due to the cavity resonance condition moving away from the peak of the phase matching .
the laser feedback scheme has been locked stably for more than 6 hours , the drift of which is not greater than shown in figure [ lock ] , and can be locked stably with output powers up to and including 50mw . beyond this output power
the device becomes unstable , the reason for which is currently under investigation .
the performance of both schemes far exceeds the requirements for many experiments .
heating of the sample at higher input powers makes locking more difficult , but is currently under investigation .
in conclusion , we demonstrated a very high external conversion efficiency , high power second harmonic waveguide resonator device .
this was achieved in an 8 mm long titanium indiffused lithium niobate waveguide device . with a 1550 nm pump field ,
an external conversion efficiency of 75% was measured output powers up to 110mw were directly measured .
photorefraction was likely seen at output powers beyond this .
the device was locked stably over an hour using a pdh scheme producing 10mw of second harmonic via both a laser frequency and a temperature offset feedback scheme .
the authors acknowledge funding from the gottfried wilhelm leibniz - preis . ed .
l. wooten , k. m. kissa nad a. yi - yan , e. j. murphy , d. a. lafaw , p. f. hallemeier , d. maack , d. v. attanasio , d. j. fritz , g. j. mcbrien , and d. e. bossi . a review of lithium niobate modulators for fiber - optic communications systems .
, 6:69 , 2000 .
m. iwai , t. yoshino , s. yamaguchi , m. imaeda , n. pavel , i. shoji , and t. taira .
high - power blue generation from a periodically poled mgo : linbo@xmath0 ridge - type waveguide by freqeuncy doubling of a diode end - pumped nd : y@xmath0al@xmath16o@xmath17 laser .
, 83:3659 , 2003 .
k. mizuuchi , t. sugita , k. yamamoto , t. kawaguchi , t. yoshino , and m. imaeda .
efficient 340-nm light generation by a ridge - type waveguide in a first - order periodically poled mgo : linbo@xmath0 .
, 28:1344 , 2003 .
nava , p. minzioni , i. cristiani , n. argiolas , m. bazzan , m. v. ciampolillo , g. pozza , c. sada , and v. degiorgio .
photorefractive effect at 775 nm in doped lithium niobate crystals .
, 103:031904 , 2013 .
m. s. stefszky , c. m. mow - lowry , s. s. y. chua , d. a. shaddock , b. c. buchler , h. vahlbruch , a. khalaidovski , r. schnabel , p. k. lam , and d. e. mcclelland .
balanced homodyne detection of optical quantum states at audio - band frequenices and below . ,
2012 .
y. kondo , s. miyaguchi , a. onoe , and y. fujii .
quantitatively measured photorefractive sensitivity of proton - exchanged lithium niobate , proton - exchanged magnesium oxide - doped lithium niobate , and ion - exchanged potassium titanyl phosphate waveguides .
, 33:3348 , 1994 .
|
we report on a highly efficient waveguide resonator device for the production of 775 nm light using a titanium indiffused linbo@xmath0 waveguide resonator .
the device stably produces up to 110mw of second harmonic power with 140mw _ incident _ on the device - an external conversion efficiency of 75% .
the cavity length is also locked , using a pound - drever - hall type locking scheme , via one of two methods involving feedback to either the cavity temperature or the laser frequency .
both locking schemes result in a stable output power of at least 10mw over one hour .
|
as soon as kuhn understood the flexibility of polymer chains , and the origin of rubber elasticity , his student a. katchalsky thought about the possibility of transforming chemical energy into mechanical energy , using gels swollen by water .
his first idea is explained on fig.[fig1 ] .
starting from chains which carry acid groups @xmath0 and adding @xmath1 ions , one obtains a charged network @xmath2 where the chains stretch by electrostatic repulsions .
if one then adds @xmath3 , the system returns to neutral , and the gel contracts .
[ h ] pgg1nov.eps this system , however , does not allow for many cycles .
adding @xmath1 really means adding soda @xmath4 and adding @xmath3 means adding hydrochloric acid @xmath5 . at each cycle ,
one thus adds one mole of @xmath6 , and this ionic solute screens out the electrostatic interactions : the system dies out fast .
katchalsky solved this problem by an intelligent trick : he used ion exchange @xmath7 against @xmath8 where @xmath9 binds two @xmath2 groups and contracts the gel .
his group produced active fibers of this type@xcite @xmath10 however , the process did not gave rise to useful applications , for a number of reasons : \1 ) time constants : what is implied here is diffusion of ions from a bath to a fiber , and diffusion is always very slow . and
even if the ions were injected locally ( by highly divided electrodes , or , by conducting polymers ) the diffusion of water remains necessary to swell or deswell .
\2 ) fatigue : if we swell a gel by water , the swelling process starts at the outer surface , and creates huge mechanical tensions in a thin region : buckling instabilities occur@xmath11 and fractures show up . there is little hope for an artificial muscle which breaks locally at each cycle . in the following section
, we present two attempts where these difficulties are taken into account
. it may well be that none of them gives a durable answer , but the trends are interesting .
the starting point here is a nematic network schematized on fig.[fig2 ] . at low temperatures ,
the system is elongated . at higher temperatures , above the nematic isotropic transition point @xmath12 , the network contracts .
networks of this type have been synthetised by a very intelligent technique @xcite .
[ h ] pgg2nov.eps of course , thermal effects have their difficulties : the diffusion of heat is faster than the diffusion of solvents , but is still slow .
this led us to think about a _ semi - fast _ system@xcite@xmath13 here we heat up rapidly the system by a light pulse ( having some adsorbing dyes inside ) and induce the nematic isotropic transition .
the contraction time of rubber in a sling -is related to the velocity of shear waves in the rubber , and is typically of order 1 millisecond .
when we want to close the cycle , we have to cool down the sample ( by a few degrees ) and this takes a long time ( seconds ) . but this semi fast actuator might be of some use .
to avoid fatigue , we conceived a system based on block copolymers ( fig.[fig3 ] ) which is hopefully well protected .
the fabrication of these sophisticated copolymers requires artistic chemistry , and is under way@xcite@xmath10 [ h ] pgg3nov.eps
a completely different approach has been used by m. shahinpoor and coworkers @xcite@xmath10 here , the basic material is commercially available , cheap , and robust .
it is a nafion : a fluoropolymer containing some fixed @xmath14 groups plus @xmath15 counterions and small water pockets which are inter connected ( fig.[fig4 ] ) .
shahinpoor was able to set up electrodes , with large contact areas , on both sides of a thin nafion sheet , by formation of platinum nanoparticles .
when such a sheet is put under a mild voltage ( @xmath161 volt ) it deforms as shown on fig[fig5 ] .
the basic process appears to be simple . when a @xmath15 ion moves , it drags some water with it towards the cathode : thus the cathode swells , while the anode contracts .
the features can be described rather simply (
like all electro osmosis effects ) in terms of two coupled fluxes ( normal to the plate ) : the electric current @xmath17 and the water flux @xmath18 the corresponding forces are the electric field @xmath19 and the pressure gradient @xmath20 @xmath21 @xmath22 one can construct order of magnitude estimates for the various coefficients . for the very small pores of interest here ,
the standard ( smoluchow ski ) description of electro osmosis is not very adequate .
the following is an alternate view point :
|
solid state actuators ( piezoelectric , ferroelectric , ... ) deform under an external field , and have found many applications .
they respond fast , but their mechanical deformations are very small .
there is a need for soft actuators , giving larger responses , but necessarily less fast .
this paper describes the early attempts by katchalsky and others , and the problems which showed up -related to time constants , and , most importantly , to fatigue .
two current attempts are reviewed .
_ key words _ : actuators , artificial muscles , nematic polymers , electro osmosis .
_ pacs numbers _ : 64.70 md , 65.70 + y , 66.30 qa .
|
passive scalars are ` tracer particles ' which are advected by the flow of another fluid .
a typical example might be ink mixed by flowing water .
the mixing of a scalar field @xmath0 in a velocity field @xmath1 is governed by the advection - diffusion equation , @xmath2 in which @xmath3 is a diffusion constant and @xmath4 is an external source of scalar .
the velocity field is determined independently of the scalar field ( hence the adjective _ passive _ ) and ( in this discussion ) is incompressible ( @xmath5 ) .
the incompressible flow simply transports the scalar , without changing any of the scalar s values .
the diffusion term smooths the resulting scalar differences , while the source term generates new scalar to compensate for this smoothing .
the mixing of a scalar field @xmath0 is characterized experimentally by its structure functions , defined as @xmath6 for positive integers @xmath7 .
it has been established experimentally that passive scalar structure functions are scaling functions of @xmath8 over a wide range of length scales when the advecting flow is turbulent .
this range of length scales , known as the inertial range , extends from a macroscale ( usually determined by the flow geometry ) down to a dissipative cut - off scale .
the lower end of the inertial scaling range is set either by the viscous scale of the velocity field ( the scale where viscosity damps out the fluid motion ) or by the scalar s dissipation scale ( the scale where molecular diffusivity damps out scalar fluctuations ) .
the ratio of these length scales , known as the prandtl number , is a material property of the system .
this work will focus exclusively on systems in which the molecular diffusivity imposes the lower bound on the inertial range ( prandtl number less than one ) .
in such systems the size of the inertial range is measured by the peclet number , defined as the ratio of the macroscale to the dissipation scale .
the inertial range scaling exponents @xmath9 , defined by @xmath10 , are measured experimentally .
the simplest possible prediction is based on the belief that the scalar difference between two points separated by a distance @xmath8 will typically have a magnitude @xmath11 which scales with @xmath8 , @xmath12 ( for some @xmath13 ) .
this implies @xmath14 .
consequently , the scaling exponents satisfy @xmath15 , which is known as ` regular scaling . '
experimentally , the exponents are seen to exhibit ` anomalous scaling , ' @xmath16 .
such scaling indicates that there is no typical magnitude for @xmath11 , and this wide variation in magnitudes is known as _
multiscaling_. experimentally , measured scaling indices for passive scalars in turbulent flows are quite anomalous . @xcite
until recently , relatively little attention was paid to this observed anomalous scaling in the passive scalar field .
this is perhaps because the phenomena had often been attributed to anomalous scaling of the underlying velocity field .
however , a model proposed by robert kraichnan @xcite suggests that a hypothetical regular - scaling velocity field would still generate anomalous scaling in the scalar structure functions .
in addition , the scaling exponents would be universal , independent of details such as the geometry of the flow . in this model ,
the passive scalar is advected by a stochastic velocity field with regular scaling in space but with an extremely short correlation time .
it is possible to form equations for the structure functions in this model , provided a closure assumption is made .
the proposed closure @xcite @xcite leads to structure functions which scale with @xmath8 , but with scaling exponents @xmath9 which grow like @xmath17 rather than @xmath7 .
others , however , have applied different techniques to the same model ( avoiding a closure ansatz ) and claim to have found different results .
@xcite @xcite @xcite @xcite a review of the disagreement is given by shraiman and siggia .
@xcite to study this closure ansatz and the passive scalar structure functions numerically , i use a one - dimensional model @xcite in which the structure functions @xmath18 obey statistical equations with the same closure problem as in kraichnan s model .
the model is simple enough to allow direct numerical study of the structure function scaling ( over 2 orders of magnitude in the separation @xmath8 ) and the proposed closure ansatz . mixing occurs through a random mapping function chosen to mimic an incompressible flow and induce scaling in the structure functions . because it is restricted to one dimension , the model is best viewed as representing scalar mixing in a turbulent pipe flow . in this model
the second order structure function can be determined analytically . for large peclet number it scales with @xmath8 with a scaling exponent @xmath19 determined by the advective mapping function .
the higher order structure functions @xmath20 also scale ( numerically ) , with exponents @xmath9 which appear to approach a constant value at large @xmath7 .
the closure ansatz proposed by kraichnan is seen to fail in numerical simulations of this model .
the source of this failure can be traced to the finite size of the largest mixing events .
this upper length scale is analogous to the largest eddy size in a fluid flow ( in a pipe this size is determined by the pipe diameter ) .
very large scalar differences can not be produced by a single mixing event , but instead arise from the combined action of several events .
for this reason large scalar differences are exponentially improbable .
deviations from the closure ansatz are observed only for these large scalar differences .
however , the behavior of the large @xmath7 structure functions is determined by these rare events . in this model
the scaling exponents approach a constant at large @xmath7 , in contrast to the @xmath21 behavior predicted for kraichnan s model .
in addition , the value of the constant might depend on the largest eddy size , so that the large @xmath7 behavior of the exponents would be geometry dependent .
because these conclusions rest primarily on the fact that the largest mixing events are much smaller than the system size , one might expect them to be independent of the details of the small - scale mixing process .
mixing in a turbulent pipe flow should exhibit this separation of length scales , and there is some experimental evidence to support the idea that large scalar differences are exponentially unlikely in such a flow .
@xcite robert kraichnan first introduced the ` white - advected ' passive scalar model in 1968 @xcite , and introduced the anomalous - scaling solution in a pair of recent papers .
@xcite @xcite all of the principal results of this section can be found in the recent literature , although some of the notation has been altered here . in this model
the passive scalar @xmath22 obeys the usual advection - diffusion equation , @xmath23 in which @xmath3 is a diffusion constant .
the external source @xmath4 was not explicitly considered by kraichnan , but is necessary to maintain a state of statistical equilibrium .
the incompressible velocity field ( @xmath5 ) is random with zero mean and is white in time .
the second order velocity structure function is specified as @xmath24 \cdot { { \bf r } \over r } [ { \bf u}({\bf x+r},t')-{\bf u}({\bf x},t ' ) ] \cdot { { \bf r } \over r } \rangle \equiv c \delta(t - t ' ) \left ( { r \over l_v } \right ) ^\eta\ ] ] where @xmath25 is a constant and @xmath26 is a parameter of the model .
this scaling in space holds until @xmath8 approaches some large correlation length @xmath27 , representing the largest motions in the system .
the scaling correlation in space is similar to real turbulent flows , while the very short range of temporal correlations is unphysical but desirable for technical reasons .
the velocity statistics are chosen to be gaussian , so higher order velocity structure functions can be expressed in terms of this second order function .
consequently , the velocity structure functions exhibit regular scaling , and so any anomalous scaling found in the passive scalar structure functions must arise from the structure of the scalar advection - diffusion equation and not from the underlying velocity scaling .
the source term is also a gaussian random variable with zero mean .
its correlations are specified by @xmath28 the spatial function @xmath29 is assumed to be smooth , so that @xmath30 for @xmath31 , where @xmath32 represents the correlation length of the source field .
calculation of the structure functions proceeds by forming an equation for scalar differences at equal times , defined as @xmath33 .
one can take the ensemble average of this equation by integrating over the recent history of the scalar and using the defined correlation functions for @xmath34 and @xmath4 .
the @xmath35-correlation in time makes it possible to do the resulting integrals . using the definition @xmath36 ,
the structure functions are found to obey @xmath37 in this equation , @xmath38 is the richardson eddy - diffusivity operator , defined as @xmath39 where @xmath40 and @xmath41 is the dimension of space .
the source term @xmath42 was makes it possible to establish a state of statistical equilibrium ( @xmath43 ) .
the dissipation function @xmath44 must be expressed in terms of @xmath18 before the system of equations for @xmath18 will be closed .
the lowest non - trivial structure function , @xmath45 , can be determined exactly .
the dissipation function @xmath46 can be evaluated by commuting derivatives with @xmath47 and with the averaging process @xmath48 , so that @xmath49 the mean - squared scalar dissipation @xmath50 is constant , and can be determined by balancing against the scalar source ( since the flux of scalar is conserved ) . at very large separations
@xmath51 , the source correlation vanishes ( @xmath52 ) and @xmath53 must approach a constant .
consequently , the equation for @xmath53 implies @xmath54 . the complete equation for @xmath53
can then be re - cast as @xmath55 assumming @xmath30 .
the two differential operators which act on @xmath53 possess different scalings .
the length scale at which the operators are equal is defined as the dissipation scale , @xmath56 : @xmath57 the upper limit of the inertial ( scaling ) range is set by the lesser of the two macroscales , @xmath32 and @xmath27 . in the inertial range ( @xmath58 ) , the second operator can be neglected , and the solution is approximately @xmath59 where @xmath60 is the second - order scaling exponent . alternately , in the dissipation range ( @xmath61 )
, the first operator can be neglected , and the solution is approximately @xmath62 there are two possible approaches for determining the higher order scaling exponents .
the first is to balance the eddy - diffusivity operator against the forcing term .
if the @xmath63-th structure function has scaling exponent @xmath9 , then the corresponding eddy - diffusivity term scales with @xmath8 with the scaling exponent @xmath64 .
the source term has scaling @xmath65 . if these terms are equal ( _ i.e. _ if @xmath66 is negligible ) then the scaling exponents are given by @xmath67 , which is regular scaling .
the other possibility , first proposed by kraichnan @xcite , is to balance the eddy - diffusivity operator against the dissipation term @xmath66 .
this basically assumes that the dominant scaling is set not by the source term , but by a homogeneous solution ( or _ zero mode _ ) of the structure function equations .
consequently , the scaling could be _ universal _ , independent of the details of the source term .
unfortunately , evaluation of the @xmath66 requires a closure ansatz stating how the laplacian of the scalar field at two distinct points , @xmath68 depends on the scalar difference @xmath69 between those points .
the proposed closure ansatz is best understood in terms of the probability distribution function ( pdf ) for scalar differences @xmath70 .
this function gives the probability for seeing a particular value @xmath47 for the scalar difference between two points separated by a distance @xmath8 .
the structure functions @xmath20 are the moments of this pdf : @xmath71 the unknown closure information is expressed in the conditional probability @xmath72 which gives the ensemble - averaged value of @xmath68 subject to the constraint that the scalar difference between the two points take on the particular value @xmath47 .
the dissipation functions can be expressed using @xmath73 and the pdf for scalar differences @xmath70 : @xmath74 the closure ansatz proposed in @xcite is that @xmath73 is approximately a linear function of @xmath47 : @xmath75 with some unknown slope @xmath76 .
this assumption can be viewed as a truncation of the taylor series expansion of @xmath77 ( by symmetry , the series has only odd terms ) . by substitution , the dissipation functions are @xmath78 .
the function @xmath76 can be determined by considering the @xmath45 case .
the result is that @xmath79 ( recall @xmath46 is approximately constant in the inertial range ) .
hence both @xmath66 and @xmath80 have scaling exponent @xmath64 .
consequently , the scaling exponents are set by the _
coefficients _ of the terms .
the result is that @xmath81 the asymptotic behavior is that @xmath82 , which depends only on the spatial dimension @xmath41 and the velocity scaling @xmath26 ( through @xmath83 ) .
the anomalous scaling for @xmath9 in the kraichnan model depends on the closure ansatz , @xmath84 .
unfortunately , it is not possible to consider higher order corrections to the taylor series for @xmath77 because the resulting equations for @xmath18 can not be solved .
however , others have approached the model using techniques which avoid the need for a closure ansatz and , in certain limits , have found values of @xmath9 which disagree with kraichnan s result .
the competing approach to determining scaling exponents in this model focuses on the @xmath7-point correlation functions of the scalar ( known as @xmath7th order moments ) , defined as @xmath85 the @xmath7th order structure function can be expressed in terms of the corresponding moment by allowing points to fuse together .
the moments convey more information than the structure functions and can be viewed as the more fundamental objects , but moments are generally not measured experimentally .
the @xmath35-correlation in time makes it possible to construct closed equations for the moments ( known as _
hopf _ equations ) .
there is no closure problem associated with the dissipative terms .
unfortunately , the equation of the @xmath7th order moment is a partial differential equation in @xmath86 vector separations , and it is beyond the reach of known mathematics to determine the solutions of such complicated equations . to determine scaling exponents ,
several groups have assumed an overall scaling for the @xmath7th order moment when all separations lie within the inertial range , and then attempted to calculate the scaling exponent in certain limits using perturbation theory . in each case
it is assumed that the dominant contribution comes from a zero mode of the hopf equation .
these exponents differ from the structure function scaling exponents proposed by kraichnan .
the most readable description of this approach can be found in the paper by chertkov , _ et .
al._. @xcite one of the first of these calculations was done in the limit @xmath87 .
@xcite the result was that the 4th order scaling exponent was given by @xmath88 ( @xmath89 ) .
this indicates that the scaling is nearly regular in this limit , with a small correction for @xmath90 .
the solution proposed by kraichnan ( equation [ kraexp ] ) is not regular in this limit ; instead it yields @xmath91 which is quite anomalous as @xmath92 ( in three dimensions @xmath93 , not 4 ) .
this disagreement is very substantial .
unfortunately , the huge difference in approach makes it difficult to ascertain the source of the disagreement .
another limit in which a descrepancy was found is the limit of large space dimension ( @xmath94 ) .
@xcite @xcite in this case the scaling is again nearly regular , with a small correction in @xmath95 .
the corresponding limit of equation [ kraexp ] is also regular , but the correction term differs from that obtained from the hopf equation calculation . one proposed reconcilliation between these calculations is a generalization of kraichnan s closure based on the ` fusion rules . '
@xcite the dissipation functions are given by a generalized form of equation [ j2n ] @xmath96 with an unknown constant of proportionality @xmath97 .
values @xmath98 would change the result for the scaling exponents to @xmath99 no precise form for @xmath97 has been proposed , but ching @xcite argues that @xmath100 for large @xmath7 , indicating @xmath101 for large @xmath7 .
numerical simulations of the kraichnan model in two dimensions have been attempted by two groups .
@xcite @xcite although the size of the inertial range is limited , both groups claim to support kraichnan s solution for intermediate values of @xmath19 . the limit @xmath87 has not been approached in either case .
to address the debate , i propose a simple passive scalar model which reproduces the closure problem described above .
the closure ansatz can be studied numerically on two levels .
first , one can calculate the dissipation functions @xmath66 and measure the constants of proportionality to determine if @xmath98 .
second , one can study the conditional probability @xmath102 directly to determine if it is a linear function of @xmath47 . for numerical simulations ,
it is necessary to develop a model which is simple enough computationally to permit a large inertial range .
consequently , a one - dimensional scalar field is preferable
. however , a one - dimensional incompressible velocity field would be quite dull , so we are forced to choose some other form of mixing that retains certain traits of an incompressible flow .
the attribute preserved in this model is that the mixing simply re - arranges the scalar field , without altering any scalar values .
any advection of this type will necessarily be non - local .
the particular method of re - arrangement is based on a picture of a turbulent flow consisting of many swirling eddies , and is constructed to induce scaling in the passive scalar structure functions . because the model is one - dimensional , it is best thought of as being analogous physically to a turbulent pipe flow .
this model is loosely based on the linear eddy model of kerstein .
@xcite the model is designed to produce structure functions which obey statistical equations with the same form as equation [ stateq ] .
the dissipation functions @xmath66 are identical , so that the closure problem is reproduced , but the eddy - diffusity operator @xmath38 is altered ( although it remains a scaling operator ) .
the motivation for this particular model is developed as follows : imagine that the one - dimensional scalar field is embedded in a plane .
the advection consists of an eddy in that plane , centered on the scalar field , which rotates one - half turn .
this maps the scalar field @xmath103 onto itself , according to the rule @xmath104 where @xmath105 is the eddy size ( centered on @xmath106 ) .
the eddy acts only on the region @xmath107 ; elsewhere nothing happens .
applying one of these eddies in each time step @xmath108 , with randomly chosen size and position , along with diffusion , gives a rule for advancing the state of the passive scalar by one time step , @xmath109+d\tau \partial_x^2 t(x,0)\ ] ] in which the advection operator @xmath110 $ ] is @xmath111 \equiv \left\ { \begin{array}{ll } t(2x_o + l - x)-t(x ) & \mbox{if $ x_o \le x \le x_o + l$ , } \\ 0 & \mbox{otherwise .
} \end{array } \right.\ ] ] the model consists of applying this rule many times , with the size l and position @xmath112 chosen randomly at each step from appropriate probability distribution functions . all possible @xmath112 in the system have equal probability , but the eddy sizes @xmath105 are chosen according to a scaling probability with scaling index @xmath113 : @xmath114 this generates scaling behavior within an inertial range determined by the smallest and largest possible values of @xmath105 : @xmath115 . in the analogy to turbulent pipe flow
, @xmath27 would play the role of the pipe diameter and @xmath116 the role of a viscous cut - off scale ( the size of the smallest motions ) .
the passive scalar structure functions would be expected to exhibit inertial range scaling between @xmath116 and @xmath27 , so long as the dissipation scale is not larger the @xmath116 . in practice ,
the constant @xmath117 is chosen so that the dissipation scale is roughly equal to @xmath116 ( corresponding to a prandtl number near unity ) , so that the ratio of length scales @xmath118 plays the role of the peclet number in the model . to maintain a state of statistical equilibrium , some forcing is required . in this model ,
the forcing is done by imposing an overall gradient @xmath119 on the scalar , as in @xcite .
a new variable @xmath120 is then defined by the deviation from the gradient , @xmath121 and structure functions are defined in terms of @xmath120 : @xmath122 equations for the structure functions @xmath123 can be constructed in the same way as in section 1.2 .
the scalar difference @xmath124 obeys the equation @xmath125\ ] ] when time is advanced by one unit @xmath108 .
the action of the convective term @xmath126 on @xmath47 depends on whether x or y ( or both ) lie within the eddy : @xmath127 \equiv \left\ { \begin{array}{ll } \delta ( x , y ) & x , y \notin [ x_o , x_o +
l ] \\
\delta ( 2x_o + l - x , y ) + g(l+2x_o -2x ) & x \in [ x_o , x_o + l ] , \\ & y \notin [ x_o , x_o + l ] \\ \delta ( x,2x_o + l - y ) - g(l+2x_o -2y ) & x \notin [ x_o , x_o + l ] , \\ & y \in [ x_o , x_o + l ] \\ \delta ( 2x_o + l - x,2x_o + l - y ) - 2g(x - y ) & x , y \in [ x_o , x_o + l ] .
\end{array } \right.\ ] ] raising this equation to the @xmath7th power and taking the ensemble average gives @xmath128 \rangle + nd\tau \langle \psi ^{n-1 } [ \delta(x , y;0 ) ] ( \partial_x^2 + \partial_y^2 ) \delta ( x , y;0 ) \rangle + ... \ ] ] keeping only the lowest order term in @xmath117 . in the diffusive term ,
@xmath129 $ ] can be replaced with @xmath47 , because only a small portion of the scalar field lies within the eddy at that time step ( the eddy is another higher order correction ) .
then the equation becomes @xmath130 \rangle + j_n(x , y;0)\ ] ] where @xmath131 as in equation [ jdef ] and @xcite .
the quantity @xmath132 \rangle$ ] can be computed from the definition of @xmath126 by taking the ensemble average over the eddy probabilities for @xmath112 and @xmath105 . defining the difference variable @xmath133
, the result divides into three regions : a diffusive interval @xmath134 , the inertial range @xmath135 , and a large scale region @xmath136 .
the structure functions obey the equation @xmath137 = \lambda j_n(r,0 ) + f_n(r,0)\ ] ] where @xmath138 is the system size . in statistical equilibrium @xmath139 , and this equation is identical in form to equation [ stateq ] .
the eddy - diffusivity operator @xmath38 and the source @xmath140 differ according to the value of @xmath8 . for @xmath141
, they are @xmath142 \equiv -{\int_{l_o}^{l_v } p(l ) dl } \ { { \int_{l - r}^{l+r } dz s(z ) } -(r+l)s(r)-(r - l)s(-r ) \}\ ] ] @xmath143 in the inertial range , @xmath144 , they are @xmath145 \equiv -{\int_{l_o}^{r } p(l ) dl \ { { \int_{r - l}^{r+l } dz s(z ) } -2ls(r ) \ } } \nonumber \\ - { \int_{r}^{l_v } p(l ) dl \ { { \int_{l - r}^{l+r } dz s(z ) } -(r+l)s(r ) -(r - l)s(-r ) \ } } \end{aligned}\ ] ] @xmath146 in the large scale region , @xmath147 , they are @xmath142 \equiv -{\int_{l_o}^{l_v } p(l ) dl \ { { \int_{r - l}^{r+l } dz s(z ) } -2ls(r ) \ } } \ ] ] @xmath148 the first order structure function @xmath150 vanishes , so the lowest non - trivial structure function is @xmath53 .
it can be determined approximately in both the inertial and dissipative ranges . as a first step ,
it is necessary to re - write @xmath46 by commuting derivatives and using spatial homogeneity , as in the kraichnan model : @xmath151 where @xmath152 is the mean - square dissipation of the scalar ( a constant )
. this constant can be evaluated by looking at the large scale region ( where @xmath53 must approach a constant ) and balancing it with the forcing term @xmath153 , so that @xmath154 this result serves as a check on the accuracy of numerical simulations . the solution far into the dissipative region @xmath155 can then be evaluated by neglecting both @xmath38 and @xmath156 ( setting @xmath157 ) : @xmath158 in the inertial range , the approximate solution is found by balancing the convective term @xmath38 against the dissipation term @xmath159 : @xmath160 \cong -{2\over3 } \langle l^3 \rangle g^2.\ ] ] by assuming a scaling solution , @xmath161 , the convective term becomes @xmath162 \cong -{a_2 l_o^y \over { y-1 } } i(\rho_2,y)r^{2+\rho_2-y}\ ] ] in the limits @xmath163 and @xmath164 .
the definite integral @xmath165 is defined as @xmath166 and can be evaluated numerically .
the first integral in @xmath167 diverges at the lower limit for @xmath168 , setting an upper limit on the scaling range .
( for @xmath168 the lower limit must be replaced by @xmath169 , and the ultraviolet region determines the solution . ) since @xmath170 $ ] balances a constant , the scaling is fixed at @xmath171 ( which only makes sense for @xmath172 ) and the solution for @xmath173 is @xmath174 where @xmath175 has been evaluated explicitly .
these solutions for @xmath53 compare very well with the numerical results in both the inertial and dissipative ranges ( see figure [ fig5 ] ) . in the inertial range ,
the scaling of higher even - order structure functions is determined by balancing @xmath176 $ ] against @xmath66 , as in @xcite : @xmath177 \cong j_{2n}(r).\ ] ] the closure assumption for @xmath66 is that @xmath178 this generalized form with @xmath98 was suggested in @xcite , while @xmath179 gives the original kraichnan ansatz of @xcite . assuming scaling functions for @xmath18 , the scaling exponents @xmath9 are determined by the coefficients in equation [ leqj ] : @xmath180 this equation for the exponents is analogous to equation [ modexp ] in the kraichnan model .
this result can be used ( in principle ) to evaluate the coefficients of proportionality @xmath97 given numerically measured @xmath9 .
unfortunately , the sensitivity of @xmath167 to @xmath181 limits the accuracy of this method .
however , the numerical results for the exponents and @xmath97 ( computed from equation [ jn ] ) are consistent with this result , indicating that the scaling is indeed set by balancing the convective term @xmath38 against the dissipation term @xmath66 .
odd order structure functions exhibit scaling only in the dissipative range . because the eddy - diffusivity operator @xmath38 differs depending on whether it operates on an odd or even function of @xmath8 , the inertial range scaling solution exists only for even order structure functions .
odd order functions are positive scaling functions in the dissipation range ( small @xmath8 ) , but pass through zero in the inertial range and then approach zero from below at large @xmath8 .
this behavior of the odd structure functions is peculiar to this particular model and differs from physical passive scalars . at small enough
length scales the dissipative terms determine the scaling . in this ( dissipative )
range the @xmath182 term dominates the solution .
it is convenient to re - write it as : @xmath183 \rangle .\ ] ] balancing the two parts of @xmath182 against each other will lead to a solution if the conditional probability @xmath184|\delta \rangle\ ] ] can be calculated . in a purely dissipative system , @xmath185 can be determined from a taylor expansion of @xmath47 at small separations : @xmath186 , which implies @xmath187 .
however , the presence of convection competes with the smoothing effects of diffusion and generates higher terms in the taylor expansion of @xmath47 .
given this fact , a reasonable closure approximation ( which is supported numerically ) is @xmath188 where @xmath189 , @xmath190 , and @xmath191 are constants .
the assumed @xmath8 dependence is necessary to generate a solution with regular scaling . for a regular scaling solution , @xmath192
, the unknown constants can be expressed in terms of @xmath193 ( calculated above ) and @xmath194 ( unknown ) : @xmath195 hence all higher order structure functions are expressible in terms of @xmath53 and @xmath196 . unfortunately , no analytic solution for @xmath196 has been found .
however , in the absence of any spatial assymmetry ( which is generated by the gradient forcing used in this model ) the odd order structure functions would vanish and the solution would be @xmath197 which has regular scaling but non - gaussian statistics .
a computational model was created by discretizing the dynamical equation for @xmath103 .
the advective term is handled straightforwardly , since it is just a re - arrangement of the values of @xmath103 .
diffusion was done subsequent to the advection process in each time step , using a second - order implicit finite differences scheme .
an overall gradient was applied to generate forcing , and periodic boundary conditions for the fluctuating field @xmath198 were used .
the system was evolved to a state of statistical equilibrium before any averaging computations were done .
equilibrium was indicated by the establishment of stable ( analytically known ) values of @xmath199 and @xmath149 .
structure functions were computed by taking space - time averages over the entire simulation , implicitly assuming an ergodic system .
[ fig2 ] simulations were conducted in a system with 50000 grid points and a grid spacing of @xmath200 .
the smallest eddy size was @xmath201 , permitting a significant dissipative interval .
the largest eddy size was either @xmath202 or @xmath203 , allowing for inertial range scaling over two orders of magnitude .
the imposed gradient was @xmath204 , and the diffusion constant @xmath117 was of order @xmath205 ( varying slightly with @xmath113 to set the dissipation scale approximately equal to @xmath116 ) .
the eddy scaling exponent was varied between @xmath206 and @xmath207 for @xmath202 , and between @xmath206 and @xmath208 for @xmath203 .
the number of time steps varied between @xmath209 and @xmath210 . a large number of time steps is needed to suitably average over the eddy size probability distribution @xmath211 , especially for larger @xmath27 and @xmath113 .
in addition , several simulations were conducted with large diffusion constants ( @xmath212 ) to study the dissipative range solution .
a typical ` snapshot ' of the scalar profile @xmath103 is shown if figure [ fig2 ] .
the profile reveals structure on a wide range of length scales , including significant ` flat ' regions , and appears similar to experimentally observed profiles .
second order structure functions @xmath53 have been computed analytically and can be compared directly with simulation results .
figure [ fig5 ] shows @xmath53 as a function of @xmath8 on a log - log plot ( base @xmath213 ) for @xmath202 and four values of @xmath113 .
the smooth curves show the simulation data ( 50 data points each ) .
the analytic results are shown as two straight lines ( equation [ s2d ] for the dissipative region , and equation [ s2r ] for the inertial range ) .
there are no free parameters .
the agreement is good ( within 5% in the heart of the inertial range ) except at the inertial range boundaries , where the analytic calculation has no validity .
interestingly , the boundary terms apparently grow in importance as @xmath113 increases , and this encroachment reduces the effective size of the inertial range .
higher even - order structure function also exhibit inertial range scaling .
figure [ fig7 ] shows the even order scaling indices @xmath9 as a function of @xmath7 for five values of @xmath113 .
the error bars represent the range of observed values over several simulations with different initial conditions .
the scaling indices are independent of the upper length scale @xmath27 ( within the error ) .
the second order indices @xmath19 lie within 3% of the theoretical values @xmath214 .
the deviation from regular scaling @xmath215 is quite pronounced . for larger
@xmath113 , the scaling exponents appear to approach a constant value ( dependent on @xmath113 ) as @xmath7 increases .
[ fig7 ] figure [ fig8 ] shows the probability distribution function ( pdf ) for @xmath47 on a log - linear scale for several values of @xmath8 from the @xmath206 , @xmath202 simulation .
the core of the pdf , defined roughly by @xmath216 ( @xmath217 ) , is rounded .
the slight asymmetry between positive and negative @xmath47 is due to the imposed gradient .
figure [ fig9 ] shows the pdfs for the same values of @xmath8 from the @xmath208 , @xmath202 simulation .
the core of this pdf is much more sharply peaked .
[ fig8 ] [ fig9 ] all the numerically generated pdfs exhibit exponential tails for @xmath218 : @xmath219 where @xmath191 is independent of the separation @xmath8 .
exponential tails have also been derived in another model in a particular limit .
@xcite this form of the pdf suggests that , for large @xmath7 , the structure function scaling exponents @xmath9 approach a constant independent of @xmath7 . in this case
the structure functions obey @xmath220 in the limit of large @xmath7 . the pdf s and the structure functions
can both be used to independently estimate @xmath191 , and the results are shown in figure 3.9 ( as a function of @xmath19 ) .
the two sets of data represent the two different values of @xmath27 , @xmath202 and @xmath203 .
the exponential tails can be understood in terms of a random walk of fluid elements . to generate a particular temperature difference
@xmath47 , fluid elements initially separated by a distance of order @xmath221 must be brought close together .
since the largest correlated motion in the system is of size @xmath27 , scalar differences larger in magnitude than @xmath222 can only be generated by the uncorrelated action of several eddies .
this behavior is essentially a random walk , and the multiplication of probabilities leads to the exponential pdf . to be more quantitative , consider the probability for the motion of a fluid element along a lagrangian trajectory from position zero to position @xmath223
. for simplicity assume that all eddies are the same size @xmath105 . to move a distance @xmath223 with @xmath224 ( for integer @xmath225 )
will typically require the point to be moved by @xmath225 eddies , each carrying it a distance of order @xmath105 . however , the probability that the point will lie within a particular eddy is @xmath226 , and these probabilities multiply : @xmath227 the motion must occur quickly , in ( order of magnitude ) @xmath225 steps , or else diffusion will cause the fluid element to equilibrate with its new environment . by assuming the two points involved in constructing @xmath47 move independently , this probability can be converted to a pdf for @xmath47 by using @xmath228 regardless of the separation @xmath8 .
the result is @xmath229 @xmath230 assuming that each eddy is of size @xmath27 suggests estimates of @xmath231 for @xmath202 and @xmath232 for @xmath203 .
these values are shown as straight lines in figure [ fig10 ] , and are in qualitative agreement with the data .
of course , not all eddies are of size @xmath27 ; in fact the number of large eddies decreases as @xmath19 increases .
this results in the trend of increasing @xmath191 shown in figure [ fig10 ] .
[ fig10 ] once two particles have come close together , they become subject to correlated motions and diffusion .
this is presumably the source of the @xmath8 dependence in the pre - factor @xmath233 of the exponential tail .
correlated motions have a weak influence on the pdf tail , and can not alter its exponential character .
also , there is a slight asymmetry between @xmath47 and @xmath234 in the prefactor ; this is attributed to the fact that producing a negative @xmath47 requires the two points to pass by each other ( in one dimension ) , and during this time their motion is correlated . to test the closure ansatz in @xcite and evaluate the constants of proportionality @xmath97 , the dissipation functions @xmath66 were computed directly from the simulation data .
rather than use the laplacian , @xmath66 was re - written by commuting derivatives as @xmath235 \rangle .\ ] ] in the limit @xmath236 , only the second term on the right side remains . in actual simulations ,
the first term makes a finite contribution which makes it more difficult to determine the inertial range scaling .
so in practice only the second term ( approximated using finite differences ) was used as a surrogate for @xmath66 in the inertial range .
[ fig12 ] in the inertial range , the @xmath66 are scaling functions of @xmath8 with scaling indices @xmath237 which , by the closure ansatz , ought to satisfy @xmath238 figure [ fig12 ] shows that this part of the closure ansatz holds , by plotting @xmath237 vs. @xmath239 for the @xmath203 simulations . the constants @xmath97 can be computed using @xmath240 and averaging over the inertial range ( @xmath97 is approximately constant over the region averaged ) .
the resulting values to not appear to depend on any parameters of the system other than @xmath113 ( a weak dependence on @xmath27 is suspected but not detectable ) .
figure [ fig13 ] shows computed values for @xmath97 averaged over several simulations .
the deviation from @xmath241 is small but significant , contradicting the ansatz of @xcite .
the growing values of @xmath97 at small @xmath19 suggests near - regular scaling in this limit , at least for moderate values of @xmath7 .
this is consistent with the scaling exponents in figure [ fig7 ] .
the intermediate case @xmath242 lies very close to the kraichnan prediction of @xmath241 . for larger @xmath19
, @xmath97 decreases with @xmath7 , consistent with the scaling exponents approaching a constant value ( as in figure [ fig7 ] ) .
[ fig14 ] in @xcite , the closure ansatz in @xcite is derived from an assumption about a conditional probability , namely that @xmath243 this conditional probability @xmath102 has be computed numerically in the simulations .
figure [ fig14 ] shows @xmath77 as a function of @xmath47 , normalized by @xmath244 , for @xmath245 and both @xmath202 and @xmath203 .
two values of @xmath8 are shown for each simulation .
the resulting averages lie very close to @xmath246 for small @xmath47 , as assumed in @xcite .
however , they deviate from the straight line at @xmath247 .
this is approximately the point at which correlated motion becomes unimportant and the dynamics are controlled by the ` random walk ' described on the previous section .
the failure of kraichnan s ansatz appears to be due to the existence of a finite upper size of the inertial range , @xmath27 , which is much smaller than the system size .
other values of @xmath19 exhibit a similar behavior . upon closer examination , the large @xmath7 behavior of the constants of proportionality @xmath97
can be related to the failure of the closure ansatz .
using the definition of the constants , @xmath248 the @xmath97 can be related to the closure ansatz through the definitions of @xmath18 and @xmath66 .
these definitions are @xmath249 and @xmath250 where @xmath251 is the probability distribution function for scalar differences .
if one defines the ` error ' in kraichnan s closure as @xmath252 @xmath253 then the closure constants can be written as @xmath254 where the functions @xmath255 are defined as @xmath256 representing the contribution of the ` error ' @xmath252 to @xmath97 .
numerically , @xmath252 is aproximately zero for small @xmath257 , but appears to grow linearly with @xmath47 for @xmath258 .
asymptotically , @xmath77 might grow as fast as linearly with @xmath47 ( but with a slope less than one ) .
an approximate large @xmath257 form for @xmath73 consistent with the simulation results is @xmath259 where @xmath260 .
then @xmath261 for @xmath258 . assuming this form for @xmath252 ,
the function @xmath262 should be less than one .
however , as @xmath263 , @xmath264 .
if the numerically observed trend for @xmath77 persists as @xmath265 , this implies @xmath266 which is a constant .
it would be consistent with the numerics to have @xmath267 and hence @xmath268 it is also possible that @xmath76 is non - zero and that @xmath269 depends on @xmath19 .
this might explain the possibly different asymptotic values for @xmath97 for different values of @xmath19 seen in figure [ fig13 ] .
unfortunately , in the kraichnan model the truly interesting quantity is @xmath270 . for @xmath271 , there are two possible assymptotic behaviors . if @xmath270 is bounded at very large @xmath7 , the scaling exponents @xmath9 would approach a constant as @xmath263 ( as suggested for the eddy model ) .
if instead @xmath270 is unbounded ( but growing no faster than @xmath7 ) , then @xmath9 would also be unbounded , but growing no faster than @xmath272 .
the possibility that @xmath273 is excluded by the holder inequalities , since it would imply @xmath9 decreasing with @xmath7 . if instead @xmath274 , @xmath270 must grow linearly with @xmath7 and the scaling exponents ultimately grow like @xmath17 , although the precise values may differ from the @xmath275 prediction .
the conditional probability @xmath276 needed for closure in the dissipative range is shown in figure [ fig15 ] for several values of @xmath8 ( from the simulation with @xmath277 ) .
the conditional probabilities are well approximated by the parabola @xmath188 with @xmath278 , @xmath279 , and @xmath280 .
these values are consistent with the independently known values of @xmath193 and @xmath194 .
[ fig15 ] the structure functions are given by @xmath192 , and the constants @xmath281 are shown in figure [ fig16 ] .
in addition , the calculated values @xmath282 from equation [ dis ] are also shown , using the analytically known value of @xmath193 and the value of @xmath194 determined from fitting @xmath283 .
the deviations from the analytic solution at large @xmath7 are possibly due to the neglected terms @xmath284 and @xmath38 in the equation for the structure functions .
the original motivation for studying this ` eddy model ' was to test the validity of the closure ansatz proposed in @xcite . in the model , the principal failure of the closure
could be traced to the finite size of the largest eddy , @xmath27 .
specifically , the conditional probability @xmath77 is not a linear function of @xmath47 , deviating at large values ( @xmath218 , where @xmath119 is the imposed gradient ) .
the deviation arises because the largest velocity eddy in the system ( @xmath27 ) is much smaller than the system size , so that large scalar differences only occur as a result of the uncorrelated action of several eddies .
this finite @xmath27 effect causes the constants of proportionality @xmath97 proposed in @xcite to deviate from @xmath285 , growing with @xmath7 for small @xmath19 but shrinking with @xmath7 at large @xmath19 ( in contradiction to the predictions of @xcite and also the simulations of @xcite ) .
other apparent consequences of this effect include exponential tails for scalar difference pdfs and scaling exponents @xmath9 which approach a constant value at large @xmath7 .
because this effect depends only on the existence of a finite eddy size @xmath27 and not on the details of the smaller scale mixing , it is reasonable to expect that it might apply generically to other models of scalar mixing , including the one proposed by kraichnan . for small values of the scalar difference
@xmath47 ( @xmath287 ) the numerical simulations of the eddy model appear to support the conditional probability ansatz @xmath84 .
low order structure functions should therefore have exponents @xmath9 which lie very close to the values predicted by kraichnan .
the fact that the errors become significant at large @xmath257 means that the scaling exponents @xmath9 will exhibit deviations from the values predicted by this ansatz when one looks at sufficiently high order structure functions ( large @xmath7 ) .
mathematically , structure function scaling is defined for an infinite inertial range , or @xmath286 .
when considering the large @xmath7 behavior of the scaling indices , one must take the limits @xmath286 and @xmath263
. the order matters .
if one considers small enough values of @xmath7 ( for any particular @xmath27 ) it will appear that kraichnan is correct .
hence if one lets @xmath286 first , all values of @xmath7 will satisfy his predictions . for moderate values of @xmath19 the numerical simulations of the ` eddy model ' support krachnan s ansatz if the limits are taken in this order .
however , for any finite @xmath27 one can find sufficiently large @xmath7 for which the scaling exponents deviate from the kraichnan values due to the finite @xmath27 effect .
hence if one lets @xmath263 first , the kraichnan solution will fail .
it is my personal opinion that keeping @xmath27 finite while letting @xmath263 is the more physical limit , since any real system will have a finite largest eddy size .
there is some numerical evidence that regular scaling is approached in the limit @xmath288 in the eddy model .
the constants of proportionality deviate significantly from @xmath275 even for the lowest values of @xmath7 , and the lowest scaling exponents fall just below the regular scaling line @xmath289 . in addition , the core of the scalar difference pdf becomes very rounded , perhaps gaussian , in this limit .
the source of these effects is not understood , but it is very possible that they have nothing to do with the finite @xmath27 effect .
if so , they may indicate a failure of the kraichnan ansatz which would persist in the limit @xmath286 .
however , it must be remembered that direct calculation of the conditional probability @xmath77 does not indicate any measurable discrepancy with the closure ansatz for small values of @xmath47 in this limit . in most of the competing calculations for the exponents of the kraichnan model @xcite @xcite @xcite ,
the gradient forcing is replaced by a random source term for the passive scalar . in this formulation
there are two upper length scales in the problem .
they are the largest correlation length of the velocity field , @xmath27 , and the largest correlation length of the source field , @xmath32 . whether or not one would expect to see deviations from kraichnan s ansatz of the type observed in the eddy model depends on which of these two length scales is larger . first , consider @xmath290 .
the source of the scalar is smooth over length scales much larger than the largest motions of the velocity field . in this case
very large scalar differences can only be generated by uncorrelated transport over scales much larger than @xmath27 .
this is analagous to the transport against the gradient by random events in the eddy model .
hence in this case one might expect to see scalar difference pdfs with exponential tails , and deviations from kraichnan s ansatz like those seen in the eddy model .
the opposite case is @xmath291 .
then the largest ` globs ' of scalar ( of size @xmath32 ) can be mixed by coherent motions of the velocity field .
the sort of transport described above would not happen , and the kraichnan ansatz could be correct in this case .
however , this is the order of limits taken in @xcite @xcite @xcite , and these authors claim to find deviations from the kraichnan result .
if they are correct , the deviations come from a different source than the finite @xmath27 effect , so the eddy model simulations do nt necessarily support their claims . a recent paper by chertkov , _ et .
_ , @xcite , has more in common with the eddy model .
it considers a scalar advected by a one - dimensional compressible velocity field with correlations analogous to the kraichnan model . by assuming that the tails of the scalar difference pdf are dominated by the most rapidly stretched lagrangian trajectories ,
they conclude that the tails must be exponential and that the scaling exponents @xmath9 must be independent of @xmath7 .
the calculations were done in the limit @xmath288 .
although the level of mathematical sophistication is quite different , the basic assumption is identical to the one used to understand the pdf tails in the eddy model , and the conclusions are also similar .
as suggested earlier , one can draw physical analogies between the eddy model and scalar mixing in turbulent pipe flow .
in particular , the separation of length scales ( between the largest eddy size and the system size ) required for exponential tails in the scalar difference pdf can be achieved in pipe flow .
the finite diameter of the pipe limits velocity eddies to this size , while the pipe itself can be much longer ( analogous to this one - dimensional model ) .
the mixing in the eddy model at small scales is quite different from real flows , but the likelihood of transport across distances larger than @xmath27 ( analogous to the pipe diameter ) could be similar ( statistically ) to real pipes .
this suggests that , for large @xmath7 , the scaling exponents in pipe flows would approach a constant value .
further , since the slope of the pdf tails depends on the pipe diameter ( @xmath222 ) , this value might be geometry - dependent .
recent experiments on pipe flows @xcite have revealed pdf s of scalar values with exponential tails .
no attempt has been made to study scalar differences , but the arguments of the previous chapter suggest that exponential tails in the scalar value pdf would imply exponential tails in the scalar difference pdf . if true , this would force structure function scaling exponents in turbulent pipe flow to approach a constant ( non - universal ) value at large @xmath7
. flows in other geometries ( such as a box , for example ) , would not be expected to display this effect .
if the largest velocity eddy is comparable to the system size , large - scale mixing can occur under the correlated motion of a single large eddy , and the mechanism for generating exponential tails in the pdf would not exist .
hence there is no reason to expect that scaling exponents @xmath9 would approach a constant at large @xmath7 in such geometries . the structure function scaling exponents of the passive scalar might therefore be geometry - dependent , with pipes exhibiting a different large @xmath7 asymptotic behavior than more open flow geometries .
it is a pleasure to acknowledge usefull discussions with l.p .
kadanoff , g. falkovich , d. lohse , p. constantin , m. chertkov , t. zhou , and n. schoegerhofer .
this work was supported in part by a fannie and john hertz foundation fellowship .
it also utilized mrsec shared facilities under nsf - dmr grant number 9400379 .
|
this paper presents a simple , one - dimensional model of a randomly advected passive scalar .
the model exhibits anomalous inertial range scaling for the structure functions constructed from scalar differences .
the model provides a simple computational test for recent ideas regarding closure and scaling for randomly advected passive scalars .
results suggest that high order structure function scaling depends on the largest velocity eddy size , and hence scaling exponents may be geometry - dependent and non - universal .
_ pacss : _ 47.27.gs
; 05.40.+j _ keywords : _ isotropic turbulence ; random processes ; passive scalars
|
radial velocity ( rv ) measurements of sun - like stars are now reaching the 1 m s@xmath0 accuracy milestone and are expected to become more accurate in the future ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
naturally , the increased sensitivity allows to identify more and more low amplitude effects that need to be accounted for when modeling the measured rv curves .
we present here such an effect , induced by what is known as stellar convective blueshift ( hereafter cb ; explained in more detail below ) , which manifests itself as a small anomalous rv signal during a planetary transit .
this small rv signal is superposed on , and independent of , the already well known rossiter - mclaughlin ( rm ) effect ( @xcite ; @xcite ; see also @xcite ) , seen during a planetary transit ( e.g. , * ? ?
* ; * ? ? ?
the rm effect originates from the host star rotation , and allows the measurement of @xmath1 , the sky - projected angle between the host star s rotation axis and the planet s orbital angular momentum to mark this angle .
] , which is an important piece of information in the study of planetary formation and orbital evolution ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . during transit the planet blocks the light coming from different regions of the host star surface , with a different rv component due to the star s rotation .
the observed line profile is distorted , it becomes asymmetric , and its center is shifted , resulting in the rm rv signal .
the rm rv curve shape depends primarily on the spin - orbit angle @xmath1 , the sky - projected rotation velocity @xmath2 , and the planet to star radii ratio @xmath3 @xcite . the new effect we present here is expected to be at the 1 m s@xmath0 level and to occur during transit , thus affecting the interpretation of the rv curve .
we describe the cb effect in , and in we use a simple numerical model to calculate the impact of the cb effect on rvs measured during transit .
we give a discussion in and a summary in .
the history of the convective blueshift effect dates back to when astronomers started investigating in detail the spectrum of the sun , and identified disagreements between the measured line positions and their expected positions based on lab experiments ( @xcite ; for a review see @xcite ) . we know today that several physical processes contribute to these disagreements , one of them being the relativistic gravitational redshift ( e.g. , * ? ? ?
still , even after the gravitational redshift and the doppler shift induced by the rotation of the sun and the earth were accounted for the above discrepancy was not completely resolved .
the corrected spectra showed that spectra of the sun observed close to the sun s limb are redshifted relative to spectra from the center of the sun s disk , and this redshift scales with @xmath4 , where @xmath5 is the angle between the normal to the sun s surface and the line of sight .
this unexpected trend was studied by many ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and its origin was a puzzle for the better part of the 20th century ( for an early historic review see * ? ? ?
before it was understood it was named the solar redshift problem , or the solar limb effect .
the solution to the problem came from the realization that the surface of a sun - like star is composed of convective cells , or granules , where hot gas is rising upward , and cooler gas is sinking downward in the intragranular lanes .
therefore , the stellar surface is composed of local blueshifts and redshifts .
since the rising stream of hot gas is brighter than the cooler sinking gas , a net blueshift is observed ( e.g. , * ? ? ?
* ; * ? ? ?
following the understanding of the physical process responsible for the effect it was referred to as convective blueshift .
we denote the local " blueshift velocity as @xmath6 , which is the blueshift observed at the center of the star s disk , when averaging across a surface area larger than the size of a granulation cell .
we note that the @xmath6 velocity is negative . depending on the value of this velocity ,
the spectral lines are broadened in an asymmetric way and their center is shifted .
for the sun @xmath7 @xmath8 m s@xmath0 ( e.g. , * ? ? ?
* ) , and it ranges from about @xmath9 m s@xmath0 for k - type stars to @xmath10 m s@xmath0 for f - type stars ( @xcite ; @xcite ; @xcite ) . for comparison
the solar gravitational redshift velocity at infinity is @xmath11 m s@xmath0 @xcite .
therefore , the cb effect should be accounted for when determining a star s absolute rv , relative to its center of mass velocity @xcite , and it should be taken into account when measuring the relative rv between the two components of a binary star , when the required accuracy is better than 1000 m s@xmath0 @xcite . our goal here is to quantify the impact of the cb effect on doppler shift measurements during a planetary transit .
this effect was already briefly mentioned by @xcite , but its influence on the measured rvs was not studied .
we adopt a model where the planet is a spherical dark body , and the host star is a uniformly rotating spherical body with a limb darkened surface .
we further assume that the host star has a uniform cb effect , where the radiation emanating from each position on its surface is blueshifted by @xmath6 directed perpendicular to the surface .
a hidden assumption here is that the transiting planet size ( or the stellar disk area the planet transverses during a single exposure ) is much larger than the typical size of a granular cell , where for the sun the latter has a diameter of typically up to about 1,000 km ( e.g. , * ? ? ?
we note that the adopted model is somewhat simplified as it is meant to present a first step at understanding the impact of the cb effect on spectroscopic transit observations .
the model ignores possible more subtle effects ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , such as meridional flows on the stellar surface ( e.g. , * ? ? ?
* ) , a possible different @xmath6 for different spectral lines ( e.g. , * ? ? ?
* ) , or at different stellar latitudes @xcite , or a dependence of the local observed blueshift on @xmath12 ( e.g. , * ? ? ?
* see also ) . in our model
the blueshift radial component of each surface element is @xmath12@xmath6 , so the absolute rv _ decrease _ induced by the cb effect is : @xmath13 where the integration is across the stellar hemisphere seen by the observer , @xmath14 is the limb darkening law , and @xmath15 is an order of unity coefficient .
since in our model @xmath6 is a constant we get : @xmath16 which equals 2/3 when there is no limb darkening . as the granulation pattern constantly changes , the value of @xmath15 will vary with a time scale of a granulation cell lifetime , which for the sun is about 10 min @xcite .
however , even for short exposure times this variation is small if not negligible since it is averaged out by the integration across the stellar surface , containing many cells .
although granulation cell characteristics may change significantly for giant stars those are not targeted by high precision rv measurements .
we note that the corresponding integral for the rm effect is zero when the planet is out - of - transit , meaning the origin of the rm effect the host star rotation does not affect the absolute rv , as expected .
as an example , we have adopted an imaginary star - planet system whose parameters are typical for a transiting system with a close - in giant planet , and are listed in .
we used a planetary radius and mass of @xmath17 and @xmath18 , and assumed a 3.5 day orbit which is completely edge - on ( @xmath19=90 deg ) , circular ( @xmath20=0 ) and aligned ( @xmath1=0 deg ) . for the host star we assumed it is a g0 type star with radius and mass of @xmath21 and @xmath22 @xcite , rotational velocity of @xmath2 = 2,000 m s@xmath0 and cb velocity of @xmath6 = @xmath23 m s@xmath0 .
we calculated the keplerian orbit of such a system , and at each phase during the transit we numerically integrated over the visible and eclipsed stellar surfaces in order to calculate both the rm and cb rv shifts .
we assumed a quadratic limb darkening law and determined the coefficients using the grids of @xcite for the sdss - g band , where the host star emits most of its radiation . in this model the coefficient @xmath15 in eq .
[ eq : rvzp ] is 0.72 and @xmath24 = @xmath25 m s@xmath0 . the resulting rv curves are plotted in fig .
[ fig : cb ] .
the top panel shows the rv curve during transit induced by the cb effect ( blue ) , while the rm and keplerian signals are ignored .
the middle panel shows the cb rv curve ( blue ) and the rm rv curve ( green ) and their sum ( black ) , while ignoring the orbital rv signal . for completeness
we show in the bottom panel the cb and rm rv curves including the keplerian orbit , with the same color code as the middle panel .
during transit the planet blocks light coming from different regions of the host star s surface .
this changes the integrated surface area in eq .
[ eq : rvzp ] , and since surface elements with different sky - projected distance from the star center have a different rv component ( following a different value for @xmath12 ) , the result of the integration varies , giving an anomalous rv signal during transit .
this signal is superposed on the keplerian and rm signals .
we note that similarly to the rm effect , the rv variation originates from a variation in the spectral line centers following a variation in their shape , and not a shift of the lines as a whole , which is the case in orbital motion .
since the cb rv shift depends only on @xmath12 ( and @xmath6 , which is constant ) the cb rv curve is a _
function about the mid transit time ( for circular orbits ) . as shown by fig .
[ fig : cb ] top panel , the cb effect during transit induces a small anomalous rv signal , at the 1 m s@xmath0 level , although it could be larger for other stars with larger @xmath6 .
the signal reaches maximum at mid transit time since in our model the observed blueshift velocity is largest close to the center of the stellar disk , which is also where most of the light coming from the star is emitted due to limb darkening .
the reason for the sign change in the cb rv curve can be understood as follows . during the transit
start and end the planet blocks light coming from the stellar limb , where the radial blueshift velocity component is smaller than the average across the entire stellar disk .
this in turn _ increases _ the overall blueshift , causing a _ decrease _ in the measured velocity .
when the planet approaches the stellar disk center , the blueshift rv component of the eclipsed stellar surface _ increases _ , thereby _ decreasing _ the average blueshift across the rest of the disk , inducing an increase in the average velocity in eq .
[ eq : rvzp ] . as seen in fig .
[ fig : cb ] top panel , the rv increase at mid transit is larger than the rv decrease close to transit start and end .
this reflects the geometry of the model and the limb darkened stellar surface .
the maximum increase in rv can be approximated as : @xmath26 where @xmath15 is determined by the integrals in eq .
[ eq : f ] and @xmath27 is the planet to star radii ratio . for the example presented in fig .
[ fig : cb ] , @xmath28 m s@xmath0 .
we caution that eq .
[ eq : acb ] is a rough estimate , and @xmath29 decreases for inclined orbits with @xmath30 deg .
an order of magnitude comparison to the rm rv amplitude ( * ? ? ? * eq . 6 ) , gives : @xmath31 this ratio is 25 % for the example presented in fig .
[ fig : cb ] , which uses @xmath2 = 2,000 m s@xmath0 , and it is larger for stars rotating more slowly . a more accurate estimate of the above ratio can be achieved by multiplying @xmath6 by @xmath15 in the right hand side of eq .
[ eq : comp ] , resulting in a ratio of 18 % for our adopted system , while the numerical integration gives a ratio of 12 % .
the cb - induced rv curve does not depend on @xmath1 or @xmath2 , meaning that on its own it does not hold information about the system s spin - orbit alignment .
in fact , the cb rv signal should be present , in principle , even if the host star does not rotate at all .
since the measured rv curve includes both effects , the cb and rm , not accounting for the cb effect in the fitted model may result in a small bias of the spin - orbit angle .
this is especially true for bright host stars that are also quiet , meaning show a low stellar activity level , as those stars allow a measured rv precision comparable to and even below the amplitude of the cb - induced rv signal . the symmetric shape of the rv curve induced by the cb effect distorts the shape of the rm signal . specifically , for completely edge - on orbits ( not necessarily aligned ) or aligned orbits ( not necessarily edge - on ) the rm curve is an _
function about the mid transit time , so the combined rm + cb curve will have no definite symmetry , and the cb rv curve can not be completely absorbed into an rm model simply by changing @xmath2 or @xmath27 .
therefore , the cb effect can bias the interpretation of the rv curve measured during transit and the estimate of @xmath1 . for orbits close to edge - on
the distortion is strongest during the central part of the transit where it increases the measured rv .
[ fig : cbzoom ] presents a zoomed - in view of the middle panel of fig .
[ fig : cb ] around the mid transit time .
it shows that the cb + rm rv curve ( black solid line ) crosses the rv zero point ( dashed black line ) , meaning it equals the keplerian velocity , later than when the rm curve does , which is at the mid transit time ( dotted vertical black line ) , when the rm effect briefly cancels out .
this time difference is approximated by dividing the cb signal at mid transit time ( @xmath29 ) by the slope of the rm rv curve : @xmath32 where @xmath33 is the duration of the full transit , @xmath34 is the orbital period , and @xmath35 is the host star mass .
[ eq : timediff ] shows that the above time difference can easily be a few minutes or more .
it is 6 min for the system presented in fig .
[ fig : cb ] . of course
this time difference is dependent on the inclination angle , and for systems far from edge - on it will depend on @xmath1 as well .
however , for systems close to edge - on it will depend only weakly on @xmath1 , unless @xmath1 is close to 90 or 270 deg .
for example , for the system simulated here the above time difference is mimicked by the rm effect alone by increasing the impact parameter to about 0.25 and taking @xmath36 deg . therefore , an accurate estimate of the mid transit time _ and _ the expected keplerian rv curve during transit are important for the correct interpretation of the spectroscopic transit rv curve and the identification of the cb effect .
an accurate mid transit time can be obtained through simultaneous photometry the measurement of the photometric light curve of the same transit event observed spectroscopically @xcite since transit light curves are known to provide mid transit time estimates with an accuracy of 1 min and better .
the required uncertainty on the keplerian rv curve during transit is smaller than the cb effect amplitude .
such a low uncertainty on this rv curve can be achieved by obtaining rvs out - of - transit , both right before and right after the transit , and also on adjacent nights in orbital phases further away from the transit , to allow for an accurate orbital characterization , including specifically the rv slope during transit and the system s center of mass velocity . in addition
, we note that for edge - on orbits with @xmath1 of 90 or 270 deg the rm signal vanishes and the in - transit rv curve will be dominated by the cb effect .
as we already mentioned the model we used here is probably a simplified version of the true behavior of the cb effect .
specifically , it was noted by several authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ) that the dependency of the observed local blueshift velocity scales as @xmath37 , where @xmath38 .
this leads to a faster decrease in the observed doppler shift towards the stellar limb , faster than in the model we adopted here where @xmath39 .
the reason for this is the longer optical path in the stellar atmosphere closer to the limb . using models with @xmath38 in our calculations results in increasing @xmath29 .
although beyond the scope of this paper , it is worth noting that the cb effect can influence the analysis of the rm effect in stellar binaries , where the latter is measured through a careful analysis of the variation in the spectral lines shape ( * ? ? ?
* and references therein ) .
we have presented here the cb effect and its impact on the rv curve during a planetary transit .
we adopted a simplified model where every surface element is radiating with the same blueshift velocity , directed along the normal to the surface . using numerical integration across the observed stellar surface we calculated the rv curves for an imaginary transiting star - planet system , similar to some of the well known transiting systems with a close - in giant planet .
the amplitude of the anomalous signal induced by the cb effect during transit is at the 1 m s@xmath0 level for giant transiting planets .
not accounting for the cb effect can lead to inaccurate interpretation of the rv curve , possibly impacting the estimate of the spin - orbit angle , which will be especially visible in the case of bright and quiet host stars with accurate rvs . for close to
edge - on orbits the cb rv signal during transit is strongest at mid transit time , where it causes a time shift between the mid transit time and the time when the rv curve crosses the keplerian curve .
therefore , an accurate estimate of the mid transit time and a well determined orbit are important for an accurate analysis of the spectroscopic transit .
so far the cb effect has been studied in detail only for the sun .
future high - precision rvs will allow us to study it also for bright stars orbited by transiting planets , showing yet another way through which transiting planets can be used to probe their host stars .
a measurement of the cb velocity for a large sample of stars will help to better understand processes such as convection and granulation , which in turn will help to better determine absolute rvs .
lcc + orbital period , @xmath34 & 3.5 & day + inclination , @xmath19 & 90 & deg + eccentricity , @xmath20 & 0 & + spin - orbit angle , @xmath1 & 0 & deg + + rotational velocity , @xmath2 & 2,000 & m s@xmath0 + convective blueshift , @xmath6 & @xmath40 & m s@xmath0 + star radius , @xmath41 & 1.10 & @xmath42 + star mass , @xmath35 & 1.05 & @xmath43 + + planet radius , @xmath44 & 1.2 & @xmath45 + planet mass , @xmath46 & 0.8 & @xmath47 + _ top : _ the cb - induced rv curve during transit ( blue ) , while the rm and keplerian signals are ignored . _
middle : _ the cb rv curve ( blue ) and rm rv curve ( green ) and their sum ( black ) , while ignoring the keplerian signal .
_ bottom : _ similar to the middle panel , with the keplerian orbit rv added to all three curves .
all panels show rv in m s@xmath0 versus orbital phase from mid transit time . ]
zoomed - in view of fig . 1 middle panel around mid transit time .
the different rv curves are represented using the same color code as in fig . 1 .
the figure shows that for orbits close to edge - on or spin - orbit aligned , while the rm rv effect ( solid green line ) briefly cancels out and crosses the zero point ( dashed black line , which is equivalent to the keplerian signal ) at mid transit time ( vertical dotted black line ) , the rv curve including the rm and cb effects ( solid black line ) crosses the keplerian signal at a later time .
this time difference is quantified in eq .
[ eq : timediff ] and equals 6 min in this example .
the blue solid line is the rv curve induced by the cb effect . ]
|
we present here a small anomalous radial velocity ( rv ) signal expected to be present in rv curves measured during planetary transits .
this signal is induced by the convective blueshift ( cb ) effect a net blueshift emanating from the stellar surface , resulting from a larger contribution of rising hot and bright gas relative to the colder and darker sinking gas .
since the cb radial component varies across the stellar surface , the light blocked by the planet during a transit will have a varying rv component , resulting in a small shift of the measured rvs .
the cb - induced anomalous rv curve is different than , and independent of , the well known rossiter - mclaughlin ( rm ) effect , where the latter is used for determining the sky - projected angle between the host star rotation axis and the planet s orbital angular momentum axis .
the observed rv curve is the sum of the cb and rm signals , and they are both superposed on the orbital keplerian curve . if not accounted for , the presence of the cb rv signal in the spectroscopic transit rv curve may bias the estimate of the spin - orbit angle .
in addition , future very high precision rvs will allow the use of transiting planets to study the cb of their host stars .
|
homoepitaxial films of inp are commonly grown as buffer and cladding / waveguiding layers , setting the base for numerous types of semiconductor devices.@xcite the morphology of these films can have a crucial impact on the quality of the overgrown material .
for example , poor quality buffers could result in structural defects propagating to the device layers , reducing the carrier mobility or affecting optical properties , and thereby degrading the performance of the device.@xcite in general , an epitaxial process is required to fabricate materials with reproducible properties of the highest quality in terms of crystal structure , purity , alloy composition , surface morphology , etc . in most cases , material quality is controlled by the adjustment of the growth conditions , such as substrate temperature , gas / molecular flow and their ratios which , in metalorganic vapor - phase epitaxy ( movpe ) , will affect precursors diffusion , surface decomposition processes , nucleation , diffusion of adatoms and their insertion at specific crystallographic sites and steps.@xcite for molecular - beam epitaxy , a well - founded conceptual and computational framework has emerged @xcite since the pioneering work of burton , cabrera , and frank ( bcf).@xcite systematic studies based on surface diffraction measurements and scanning tunneling microscopy , in conjunction with extensive computational modeling have identified and characterized many of the atomistic processes , even in the complex setting of iii - v systems.@xcite the corresponding development for movpe has been comparatively slow , due in part to the limited availability of _ in situ _
measurements.@xcite nevertheless , systematic _ ex situ _ measurements of growth morphologies , again in conjunction with theoretical modeling , have revealed the importance of precursor diffusion and decomposition at step edges in determining the nature and scale of surface morphologies.@xcite epitaxy - ready substrates are prepared by cutting individual slices out of the bulk crystal at the desired angle with respect to the main crystallographic plane.@xcite in the case of inp , the crystallographic steps exposed after this cutting may be ideally terminated either with indium ( so called a - steps , with the substrate misoriented toward the [ 111]a planes ) , or phosphorus atoms ( b - steps , with the substrate misoriented toward the [ 111]b planes ) , or any combination of these two.@xcite depending on the miscut , the exposed surface can be close to singular or contain a dense periodic array of crystallographic steps .
epitaxial overgrowth on such substrates may then proceed ( when truly `` epitaxial '' ) in one of three basic growth modes : step flow , where the overgrowing layer is advancing each exposed step at the same rate , creating an exact copy of the underlying surface ; step bunching , where terrace formation is observed due to clustering of individual steps ; or island formation , when the growth of new layer is initiated only at the step edges , but also between steps , in which case no long range order is observed .
if the growth conditions are not optimal , due to defect formation , the surface morphology can be corrupted and will not resemble any of the three main modes . for most applications ,
step flow is the desired mode , as it leads to creation of atomically flat layers , without irregularities ( such as , in an extreme example , antiphase boundaries).@xcite on the other hand , the role of small substrate misorientations for the iii - v system have been recently revealed for arsenide alloys ( e.g. gaas , alinas or inas ) grown by movpe,@xcite where complex surface organization , striking effects and state - of - the - art results on the material optical / transport properties have been reported . in those reports ,
small changes of growth conditions significantly affected the epitaxial morphology . as usual , the growth conditions revolved around a few parameters , where the common variables were substrate miscut , growth temperature , growth rate and molar ratio between precursors injected into the reactor chamber ( reactor pressure , carrier gas and precursor choice were fixed in the given setup ) . crucially
, the authors found a strong correlation between surface organization and material properties .
what is even more relevant is that the authors linked the unusual variety of surface organization ( e.g. from islanding , through step flow to periodic step bunches just by varying substrate miscut ) , to the intrinsic two - step movpe growth process : first adsorption , diffusion and decomposition of molecular precursors at step edges and , subsequently , adatom diffusion and incorporation .
their model reproduced the condition for island nucleation as well as for the step bunching instabilities , linking this last process to the difference in molecular dissociation from above and below step edges.@xcite despite the broad and primary commercial interest and exploitation of inp,@xcite only limited data are available in the literature on inp grown by movpe .
only a small range of parameters / surface organization appear to have been correlated .
for instance , the influence of small differences of substrate misorientation was partly discussed in ref .
[ ] ( for substrates with offcuts of [email protected]@xmath1 toward ( 01@xmath2 ) ) and ref .
[ ] ( for singular wafers and with misorientations of 0.2@xmath1 toward [ 110 ] , [ 111]a and [ 111]b ) .
morphologies created during homoepitaxy on singular wafers ( with misorientation below 0.1@xmath1 ) were also investigated in ref . with respect to the direction of the miscut . in ref .
[ ] , the authors analyzed the effect of temperature on step bunching on vicinal surfaces in a larger miscut range ( from 0.2@xmath1 to 2@xmath1 toward [ 111]a ) .
nevertheless , there is no comprehensive study of the organization of the movpe - induced crystallographic steps on vicinal surfaces , so a fundamental understanding of the fundamental kinetic processes is still lacking . in this paper , we report a systematic study of the impact of growth conditions on the surface morphology of inp surfaces with small misorientations during movpe .
we show several surprising effects and a remarkably broad variety of surface organizational modes , well beyond what has been observed in the arsenide ( gaas ) counterpart .
the organization of our paper is as follows .
the procedure used for our experiments is described in sec .
the surface morphologies observed on vicinal inp(001 ) are described in sec .
[ sec3 ] , with the presentation of the results grouped according to growth rate and v / iii ratio ( secs .
[ sec3a ] , [ sec3b ] , [ sec3c ] ) and the effect of doping ( sec .
[ sec3d ] ) .
these results are analyzed in sec .
[ sec4 ] , which includes a theoretical discussion based on an extension of the model in ref .
one of the interesting aspects of our experiments is that we can examine the effect of the group - v species on the various steps of the growth kinetics .
we summarize our results in sec .
[ sec5 ] , where we also discuss future modelling strategies .
all samples in this study were grown by movpe at low pressure ( 80 mbar ) in a commercial horizontal reactor with purified n@xmath3 as the carrier gas and trimethylindium ( tmin ) and phosphine ( ph@xmath4 ) as precursors.@xcite 300-nm - thick single layers of inp were grown on perfectly oriented or slightly misoriented inp(001 ) substrates , which in most cases were semi - insulating fe - doped , with reference samples grown on zn- ( @xmath5-type ) and si - doped ( @xmath6-type ) wafers ( we anticipate no observed differences to result from the substrate doping , as no diffusion to the epitaxial layer is expected ) .
the majority of the inp layers were grown as nominally undoped .
the unintentional doping level is estimated to be well below @xmath7 , which is our detection threshold .
we have also grown several samples using diethylzinc ( dezn ) and disilane ( si@xmath3h@xmath8 ) as precursors to investigate the impact they have on the surface morphology when incorporated into the inp matrix .
doping levels were @xmath9 for zn and @xmath10 for si and constant across the layer , as confirmed by electrochemical capacitance voltage measurements .
such carrier concentrations are considered moderate for inp - based semiconductor devices.@xcite growth conditions were varied systematically across samples : the estimated actual growth temperature @xmath11 varied in the range 520@xmath1c to 720@xmath1c , the v / iii ratio @xmath12 from 150 to 450 , and sample misorientation from nominally `` perfectly oriented '' to 0.6@xmath1 ( all with a tolerance of @xmath13 ) toward [ 111 ] a or b. the growth rate @xmath14 was kept constant at 0.7 @xmath15m / hr for most of the samples , with reference / comparison growth carried out at 0.35 @xmath15m / hr and 1.4 @xmath15m / hr . all together , more than 150 samples were grown in various conditions for this study , with several control growth runs performed to ensure the full reproducibility of the reported results . in all cases ,
particular attention was paid to reactor environment quality and to temperature control by growing the samples only with reactor walls already baked and covered by previous growth runs .
the growth temperature was estimated by emissivity corrected pyrometry .
the relevant details are referenced below when a given example is discussed . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] our model will be used to analyze the morphologies in table [ table1 ] , as this provides the most comprehensive temperature - misorientation ( toward [ 111]a ) matrix of data .
these data and our fits are shown in fig .
[ fig15 ] , with the arrhenius parameters for the kinetic coefficients appearing in ( [ eq1])([eq6 ] ) compiled in table [ table5 ] .
the focus of most of our discussion will be on processes occurring at step edges , but the prefactor for the surface diffusion of tmi merits some comment .
the value @xmath16 , which may seem quite high in comparison with the range @xmath17@xmath18 typically used in such theories and simulations , actually conforms with estimates@xcite for large molecules .
such large prefactors are due to jumps over several lattice cites , which is in contrast to the usual assumption of jumps of a single lattice unit .
otherwise , the precise values in table [ table5 ] are not as important as the trends they reveal as the growth conditions are varied .
we first consider the morphologies at high phosphine flow rate @xmath19 .
the trends in the left panel of fig .
[ fig15 ] are similar to those seen on gaas(001 ) ( cf .
fig . 3 of ref .
[ ] ) , which was also grown under group - v - rich conditions .
the direct incorporation of the adatom from the precursor occurs at a greater rate from above the step than below for both systems , but there is no asymmetry in the adatom step incorporation kinetics . indeed ,
large ( i.e. several tenths of electron volts ) variations of the energy barriers for the adatom diffusion and incorporation only weakly affect on the stability calculation , provided that the incorporation kinetics at the step edge are symmetric .
we have enclosed the barriers for these processes in parentheses in table [ table5 ] to indicate this fact . for these growth conditions , the instability of step flow to step bunching in inp(001 ) and gaas(001 ) is driven by an effective negative step - edge barrier induced by the asymmetric precursor incorporation kinetics.@xcite an altogether different scenario emerges from the stability analysis at low phosphine flow rates , as can be seen immediately in the right panel of fig .
[ fig15 ] . at lower temperatures ( @xmath20c and 560@xmath1c )
the morphology of all substrates shows either discrete or diffuse islands ( table [ table1 ] ) .
discrete islands indicate a growth regime far removed from step flow .
the adatom density on the terrace is large , leading to a high probability of island nucleation and growth on the terraces . as the temperature
is increased , the adatom density on the terraces decreases , so the growth nucleation and growth of islands decreases accordingly , with the islands showing some coalescence with steps , thereby indicating a transition regime between discrete islanding and step flow .
the separation of stable and unstable regions of step flow is expected to be qualitatively different from that obtained with higher phosphine flow rates .
table [ fig5 ] shows that , while the asymmetry in the molecular step - edge incorporation rates has been maintained , an asymmetry in the _ atomic _ step - edge incorporation rates has been introduced , but in _ opposition _ to the molecular incorporation rates .
this is the reason for the qualitative difference in shape of the curve that separates the stable and unstable regions @xmath21 .
our calculations show that the profile of the curve @xmath22 separating the stable and unstable regions is very sensitive to the atomic parameters . with decreasing temperature ,
the net incorporation current at steps becomes more negative , while that due to precursor incorporation becomes more positive .
for certain growth conditions , the atomic current dominates , which suggests the onset of another instability through the appearance of mounds , a fact that has also been noted by vladimirova _
et al._@xcite in fact , mounds have been observed on inp(001 ) during metalorganic molecular - beam@xcite and chemical - beam@xcite epitaxy .
the interesting point from our perspective is that this morphology is triggered by the low phosphine flow rate and can therefore be avoided .
homoepitaxy on perfectly oriented substrates resulted in the most homogenous surfaces
stable step flow was observed in the majority of cases over some temperature range , with step - bunching composed of a maximum of three steps observed near 650@xmath1c .
the modification of the v / iii ratio had the most striking effect on the surface morphology .
for @xmath23 , a growth temperature below 600@xmath1c was found to be insufficient to create useful surface organization for any misorientation of the substrate ( the exception being the substrate with a - steps with a misorientation of 0.6@xmath1 , where regular step organization was observed at 585@xmath1c ) .
temperatures above 685@xmath1c led to defects related to three - dimensional growth .
the mid - range growth temperatures provided more of a distinction between various substrates .
step - bunching was observed in most of the cases but , depending on the surface misorientation and exact growth temperature , different features were formed on the surface .
when examined in detail , the overlap of crystallographic steps could take diverse forms , like clustering of just two monolayers ( on perfectly oriented substrates ) , well organized , periodic micron - scale bunching ( cliffs observed on a - type substrates at 655@xmath1c ) , or a braid - like bunching of up to 40 monolayers ( on all vicinal surfaces grown at 630@xmath1c ) .
this limited the optimal growth temperature to a quite narrow window of about 100@xmath1c . by changing the phosphine flow to @xmath24 ,
most of the samples were eliminated by that restriction
stable , organized step - flow was observed even at 520@xmath1c and good surface morphology was obtained for growths conducted at temperatures of 720@xmath1c .
there are several aspects of the surface morphology that can be addressed with modelling .
the basis of our theoretical work@xcite has been a continuum model of the burton , cabrera , and frank@xcite type .
this allows an analytic solution to be obtained and a linear stability analysis to be carried out , but can not address the variety of morphologies seen for fully developed step bunching , for which the solution of the two - dimensional problem is required .
there are two approaches .
one is to use a continuum formulation , for example , one based on the phase - field method.@xcite this has the advantage of building the extended terrace length scales on the substrate from the outset , but suffers from the difficulty of including the details of step kinetics while retaining a direct connection to the atomistic parameters .
kinetic monte carlo ( kmc ) simulations provide an obvious alternative in that the atomistic processes can be included to whatever level of detail required , but the large terrace lengths present a computational challenge for traditional formulations .
one alternative is to use a hybrid scheme that combines the flexibility of the kmc approach and the extended length and time scales of continuum methods.@xcite these methods are currently under investigation .
this research was supported by the irish higher education authority program for research in third level institutions ( 2007 - 2011 ) via the inspire programme , by the science foundation ireland under grants 05/in.1/i25 , 10/in.1/i3000 and 07/src / i117 and by the european 7th framework programme under grant 228033 ( mode - gap ) .
the authors are grateful to dr .
k. thomas for support of the movpe system .
thanks b. a. joyce for many helpful discussions about precursor chemistry .
it should be noted that the detailed evolution at high @xmath11 , high v / iii and small misorientation ( fig .
[ fig15 ] left ) is actually not fully taken into consideration in our model , and will probably need a more comprehensive 2d model of surface dynamics , including e.g. phosphine desorption .
|
we present a systematic study of the morphology of homoepitaxial inp films grown by metalorganic vapor - phase epitaxy which are imaged with _
ex situ _ atomic force microscopy .
these films show a dramatic range of different surface morphologies as a function of the growth conditions and substrate ( growth temperature , v / iii ratio , and miscut angle @xmath0 and orientation toward a or b sites ) , ranging from stable step flow to previously unreported strong step bunching , over 10 nm in height .
these observations suggest a window of growth parameters for optimal quality epitaxial layers .
we also present a theoretical model for these growth modes that takes account of deposition , diffusion , and dissociation of molecular precursors , and the diffusion and step incorporation of atoms released by the precursors .
the experimental conditions for step flow and step bunching are reproduced by this model , with the step bunching instability caused by the difference in molecular dissociation from above and below step edges , as was discussed previously for gaas ( 001 ) .
|
the mechanical properties of amorphous solids have been harnessed extensively in designing materials which are ubiquitous in our everyday life . however , a complete microscopic understanding of the mechanisms leading to the macroscopic response of these materials is still missing . in order to develop materials with specific functions ,
it is necessary to have an improved knowledge of these underlying processes .
this remains a challenging task .
it is known that the material properties of amorphous solids , such as colloidal or metallic glasses , depend on their history of production , e.g. the cooling rate by which they were quenched from a fluid phase @xcite .
this dependence on the history , i.e. the age of the amorphous solid , is an important issue in computer simulations of glasses , especially because the accessible cooling rates in simulations are many orders of magnitudes larger than those accessible in experiments of real systems . with respect to the comparison between simulation and experiment
, it is therefore crucial to systematically understand the dependence of structural and dynamic properties on the age of the glassy solid .
it is thus expected that the response of a glass to an external mechanical loading is affected by the age of the glass .
if one shears an amorphous solid under a constant strain rate , it is in general transformed into a flowing fluid @xcite . while at sufficiently high strains , the flowing fluid reaches a steady state without any memory of the initial unsheared state , the transient response to the shear is affected by the history of the initial glass state .
the characteristic stress - strain relation of a glassy system , in response to an externally applied shear rate , exhibits typically a maximum at a strain of the order of 0.1 @xcite . in numerical simulations of sheared thermal glasses , the amplitude of this maximum
is observed to depend on the age of the material , typically growing logarithmically with increasing age @xcite .
moreover , the transient response of glasses to an external shear field is often associated with the occurrence of shear bands , i.e. band - like structures with strain or mobility higher than other regions , observed both in experiments @xcite and numerical simulations @xcite .
such spatially localised structures are seen to emerge after the occurrence of the stress overshoot , as the stress relaxes to the steady state value .
also , the formation of these transient shearbands have been observed to be influenced by the thermal history of the glassy state , with states which are obtained by faster cooling being less susceptible to shearband formation .
further , it has been noted that such a spatially heterogeneous response is more likely to occur in the transient regime beyond the stress overshoot , at any given temperature .
the focus of our study are thermal glasses which are often characterized as simple yield stress fluids , e.g. colloids , emulsions . for such materials , the steady state flow curve (
i.e. stress vs. imposed shear rate ) is a monotonic function @xcite .. thus there are no persistent shear - bands , which would be the case for non - monotonic flow curves @xcite . in the case of simple yield stress fluids , the transient shearband that emerges are seen to broaden with time and eventually the entire material is fluidized , with the timescale of fluidization depending on the imposed shear - rate or stress @xcite .
such spatio - temporal fluctuations are also visible during steady flow , both in experiments and simulations @xcite . however , in this work , our objective is to characterize the transient spatial heterogeneities , prior to onset of steady flow .
the formation of transient shearbands has been addressed within the scope of various theoretical models . within the framework of spatially - resolved fluidity and soft - glass - rheology ( sgr ) models @xcite , the age - dependent spatially heterogeneous response has been obtained , with the occurrence of the stress - overshoot under an applied shear - rate being associated with an instability leading to the formation of these transient heterogeneities .
the model recovers the observation that more pronounced and long - lived shearbanding occurs for more aged glassy samples .
similarly , manning et al .
@xcite also observe various transient heterogeneous states , by analysing a shear - transformation - zone ( stz ) model of glassy materials , which depend upon the initial state of the system ( characterised by an initial effective temperature ) and the imposed shear - rate .
further , they were able to map their results to those obtained from numerical simulations @xcite .
the same phenemenology has also been reproduced by other mesoscopic models @xcite . in this work ,
we address the question how the combination of ambient temperature , applied shear - rate and age of the glass affects the transient response , specifically the observation of spatio - temporal heterogeneities .
this has not been systematically studied in earlier numerical simulations .
consequently , we also compare our observations with those from the theoretical models . to this end , we perform molecular dynamics computer simulations of the kob - andersen binary lennard - jones ( kablj ) model @xcite , a well - studied glass former .
amorphous states are prepared by quenching a supercooled liquid to different temperatures below the mode coupling temperature , followed by a relaxation of the sample over a waiting time @xmath0 .
then , the resulting glass samples are sheared with different shear rates @xmath1 .
the onset of plastic flow occurs around the location of @xmath2 , i.e. at a strain @xmath3 ( with @xmath4 the time at which the maximum is obtained ) , with the appearance of a peak in the stress - strain response .
we demonstrate that at the different temperatures @xmath5 , the peak height , @xmath6 , for all ages and shear - rates , obeys the functional behavior @xmath7 ( with @xmath8 a function depending on @xmath1 and @xmath5 and @xmath9 a temperature - dependent amplitude ) .
note that this finding is consistent with earlier studies @xcite .
further , as we have shown recently @xcite , transient ( but long - lived ) shear bands are formed for @xmath10 , provided that shear rate is sufficiently low .
we quantify the contrast in spatial mobilities and demonstrate that the extent of spatial heterogeneities is not only dependent on the age of the glass , but also on the ambient temperature .
the rest of the paper is organized as follows . in sec .
[ sec2 ] we describe the kablj model and the details of the simulation .
then , we present the results for the stress - strain relations and the analysis in terms of mobility maps in sec .
[ sec3 ] . finally , in sec .
[ sec4 ] , we summarize the results and draw some conclusions .
we consider a binary mixture of lennard - jones ( lj ) particles ( say a and b ) with 80:20 ratio .
this is a well - studied glass former .
particles interact via lj potential which is defined as : @xmath11\ : r < r_{c},\end{aligned}\ ] ] where @xmath12 .
the interaction parameters are given by @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 , and @xmath19 .
masses of both type of particles are equal , i.e. , @xmath20 .
all quantities are expressed in lj units in which the unit of length is @xmath21 , energy is expressed in the units of @xmath22 and the unit of time is @xmath23 .
more details about the model and parameters can be found in ref .
@xcite .
we perform molecular dynamics ( md ) simulation in the @xmath24 ensemble using the package lammps ( `` large - scale atomic / molecular massively parallel simulator '' ) @xcite .
the simulations are done for the box geometry having the dimension @xmath25 .
temperature is kept constant via a dissipative particle dynamics ( dpd ) thermostat @xcite .
our method for the preparation of the glass samples is as follows : at a density @xmath26 , we first equilibrate the system at the temperature @xmath27 , which is in the super - cooled regime . then , we quench it to the target temperatures @xmath28 , below the mode coupling transition temperature @xcite . for exploring the effect of aging on the mechanical response , we sample glassy states having different ages , @xmath29 and @xmath30 , as the system evolves after the quench to each target temperature . using these initial states sampled at different @xmath0
, we apply shear on @xmath31-@xmath32 plane in the direction of @xmath31 with different constant strain rates @xmath33 and @xmath34 . to simulate a bulk glass under shear ,
we use lees - edwards periodic boundary conditions @xcite .
in order to characterize the mechanical response of the aged amorphous solids , we measured different structural and dynamical properties , both at the macroscopic and local scales .
we now discuss these observation in detail . when the externally applied shear is imposed on the aging quiescent glass , the deformation response of the material is characterized by measuring the stress , @xmath35 , generated in the system with increasing strain ( @xmath36 ) . in fig .
[ fig1](a ) , we show how the stress evolves when we impose @xmath37 on initial states sampled from different ages , at @xmath38 .
we observe the characteristic stress overshoot @xcite , with the peak height ( @xmath2 ) increasing for larger @xmath0 . across temperatures , within the glassy regime , this characteristic response does not change .
however , as expected the material becomes less rigid with increasing temperature .
for example , for a fixed age , with increasing temperature , the steady state value of @xmath35 decreases , but still one observes the stress - overshoot , albeit with a lesser value of @xmath2 ; see fig .
[ fig1](b ) for the corresponding data at @xmath39 for an imposed @xmath37 . in fig .
[ fig1](c ) , we illustrate the variation of @xmath2 with age , for a wide range of @xmath1 at @xmath38 .
thereafter , we demonstrate that at each temperature , the data for @xmath2 can be collapsed onto a master curve using the relation @xmath40 with @xmath41 a temperature - dependent amplitude that is independent of @xmath1 and @xmath0 and @xmath42 a function that solely depends on @xmath1 and @xmath5 .
such a logarithmic relationship is motivated by ree - eyring s viscosity theory , modified appropriately to take into account the role of the sample s age , @xmath0 , as noted by varnik et al .
a similar dependence was also proposed by rottler et al .
@xcite for the same binary lj mixture subjected to uniaxial strain at a different state - point .
thus our scaling results are consistent with earlier observations .
we note that such a logarithmic dependence is not observed in experiments involving other yield stress fluids like colloidal pastes @xcite or carbopol @xcite where a power - law increase is observed , which is also captured in the numerical simulations of model gels @xcite .
thus , it seems that the response of dense thermal glasses are different from the low density materials with more complex structures .
having observed the scale of the stress overshoot for various ages , imposed shear - rates and ambient temperatures , we will later explore whether such observations necessarily lead to the occurrence of transient shearbands . to have a measure of the structural changes in the system , both during the aging process and the onset of flow
, we monitor the potential energy ( @xmath43 ) of the system . in fig .
[ fig2](a ) , we show how the potential energy decreases when the system is quenched from the supercooled state at @xmath27 to the lower temperatures @xmath44 . after the fast decrease during the initial aging regime , the potential energy decreases slowly , logarithmically depending on the age @xmath0 of the system , as is typically observed during aging . when the external shear is imposed on the system , the potential energy increases , with the steady state value depending upon the magnitude of
the imposed shear - rate ; see fig .
[ fig2](b),(c ) for @xmath45 , respectively .
further , we check whether the potential energy of the system measured during the transient regime , prior to the onset of steady flow , is dependent on the initial age of the quiescent glass . as can be seen in fig .
[ fig2](b ) , ( c ) , for both the temperatures , the dependence on @xmath0 becomes more visible with decreasing shear - rate .
this thus indicates that the system evolves through different intermediate structures , dependent on the initial state having different @xmath0 , before steady states structures start getting explored .
we now explore how these age - dependent transient structures are dynamically different .
we also note that for the applied shear - rates , the approach to steady state seems faster for @xmath46 than compared to @xmath38 . while in the former case , at long strain , the steady state potential energy , under applied shear , matches at large strains , this is not the case for @xmath38 .
this combination of temperature , shear - rate and age of the sample will be further explored in later sections . in order to characterise the microscopic dynamical response of the glass ,
when yielding under shear , we monitor the single - particle dynamics .
this is quantified by measuring the non - affine mean - squared displacement ( msd ) , @xmath47 , in the direction transverse to the applied shear ; the data is shown for @xmath37 , at @xmath38 ( fig .
[ fig3](a ) ) and @xmath46 ( in fig .
[ fig3](c ) ) . at both temperatures , during early times , particles exhibit ballistic motion , which is then followed by a caging regime .
subsequently , the onset of flow is marked by a super - linear regime , and eventually diffusive dynamics is observed .
the onset of the super - linear or `` super - diffusive '' regime occurs around the stress overshoot in the stress - strain curve , when the stress , building up in the system , is released via the breaking of local cages leading to the subsequent diffusive motion of the particles @xcite .
we check how the initial age of the glassy state influences the measured @xmath47 ; the corresponding data is plotted in fig .
[ fig3](a ) , ( c ) for the two temperatures . at both temperatures , both in the initial response and in the steady state
, there is no visible difference in the measured msd , for the samples having different ages . only during the transient regime , where the onset of super - diffusive dynamics occurs ,
we observe variation with age of the quenched glass , with the effect more visible for @xmath46 ; see fig .
[ fig3](c ) .
we also note that the onset timescale depends upon age , shifting to longer timescales with increasing @xmath0 , as is clearly visible . to further probe the spatial features of the local dynamics during yielding , we divide the simulation box into eight layers of thickness @xmath48 along the @xmath32-direction .
for each of these layers , we compute the averaged msd for the particles populating it at @xmath49 , i.e. before shear is imposed . in fig .
[ fig3](b ) , ( d ) , we show the corresponding data , at @xmath45 , respectively , for an imposed @xmath37 and age of @xmath50 .
we see that there is variation in the dynamics across the different layers , the variation being greater at @xmath38 than at @xmath46 .
while for @xmath46 , eventually , the long time dynamics seems to converge to the diffusive limit , within the timescale of observation , that is not the case for @xmath38 . in the latter case , while in one case , we see the onset of diffusive dynamics at long times , in some of the other slices , the dynamics continues to be sub - diffusive .
to summarise , this demonstrates that , during yielding of the glass , at any temperature , dynamics is heterogeneous .
however , the extent of heterogeneity and persistence increases with decreasing temperature .
we will quantify that in more detail in the subsequent sections .
to visualise the dynamical heterogeneities , we construct three - dimensional maps of local msd @xcite .
these maps are constructed in the following manner .
we divide the simulation box into small cubic sub - boxes having linear size of @xmath21 . at any time @xmath51
, we calculate the average msd of the particles located in each sub - box at @xmath49 ( unsheared glassy state ) , which provides a cumulative picture of how yielding proceeds locally starting from the quiescent glass . as discussed earlier , we measure the @xmath32-component of msd of each particle .
the evolution of the three - dimensional maps with increasing strain , is shown in fig .
[ fig4 ] and fig .
[ fig4b ] , for different ages of the glassy state at @xmath38 and @xmath52 , respectively , corresponding to an imposed shear rate of @xmath37 .
first , we focus on the situation at the lower temperature ( @xmath38 ) ; see fig .
[ fig4 ] . at early strain ( @xmath53 ) ,
we observe localised `` hot spots '' of large mobilities @xcite . at a strain of @xmath54 ,
more such local regions of faster dynamics have emerged . in the case of well aged initial state
( e.g. @xmath50 ) , we see that the hot spots are organised in the form of a shear band . on the other hand , for a younger initial state ,
i.e. @xmath55 , these hot spots are much more dispersed . at a later strain ( @xmath56 ) , this difference is further amplified . for the younger sample ,
large regions of the system have been fluidized , whereas for the older system , enhanced mobility is still localised in the form of a shear band .
thus , the age of the initial glassy state , does influence the spatio - temporal organisation of regions of increased mobility .
this is consistent with earlier findings in numerical simulations using different model systems @xcite or predictions from theoretical models @xcite .
we now contrast this with the situation at the higher temperature ( @xmath46 ) ; see fig .
[ fig4b ] . at early strain ( @xmath53 ) , we observe the hot spots , for all @xmath0 ; however , there are more in number as compared to the case of @xmath38 , for any @xmath0 . with increased strain @xmath57 , we see that the hot spots have proliferated and cover almost the entire domain , including for the most aged sample .
thus , access to increased thermal fluctuations at higher temperatures lead to the tendency for a more homogenised and faster fluidisation of the glassy state under applied shear .
we will now carry out a more quantitative characterisation of this spatio - temporal response of the glassy states , sampled from different temperatures , and aged over different timescales .
so far , by spatially resolving the single particle msds , we demonstrated the existence of heterogeneous dynamics during yielding of the glass and how that depends on the temperature of the glassy state . using single particle msds
, we construct the instantaneous spatial profile @xmath58 , which gives us a measure of fluctuations in local dynamics . the evolution of these profiles , with increasing strain , is shown in fig .
[ fig6 ] , for two different ages of the glass , viz .
@xmath59 , under an imposed shear - rate of @xmath37 , at @xmath38 .
we observe that the spatial fluctuations are much larger and more persistent in the older sample , consistent with the qualitative discussion , above , involving msd maps . using these spatial profiles
, we can now quantify the contrast in local mobilities in the deforming glass , under applied shear , and compare the response across temperatures . in order to do that , we compute the fluctuations in local msd , @xmath60 with increasing strain @xmath36 .
such a quantity measures the extent of fluctuations in average mobilities , as characterised by the local msd , across the regions parallel to the direction of flow .
defined in this manner , @xmath61 would therefore easily capture the contrast in the dynamics during the formation of shear - bands . to note , such a quantity is different from the dynamical susceptibility , @xmath62 , which is a measure of fluctuations in single particle dynamics , measured within the steady - state ensemble . here , we are concerned with the fluctuations across regions during the onset of flow from a quiescent state .
the corresponding data , for temperatures @xmath44 , across ages @xmath63 , for a range of imposed shear - rates , viz .
@xmath64 , are shown in fig .
[ fig7 ] . typically , @xmath61 behaves non - monotonically with increasing strain , with the maximal spatial fluctuations ( @xmath65 ) occuring at a strain which corresponds to the super - linear regime in the msd ( see fig .
[ fig3 ] ) ; i.e. , once yielding of the glass has occurred .
further , for a fixed @xmath1 , the magnitude of @xmath66 increases with age , i.e. the contrast in mobilities across a sample becomes largest for a more aged sample .
importantly , we also note that @xmath66 also starts increasing when we shear the glasses with smaller and smaller shear - rates ; e.g. see fig .
[ fig7](a)-(c ) , for the variation of @xmath66 with age and imposed shear - rate . with decreasing temperature , this spatial fluctuation increases even further . to get a more complete picture of the variation of these spatial fluctuations , across age and temperature ,
we plot the corresponding contour maps for @xmath67 ; these are shown in fig .
[ fig8 ] . from the contour maps ,
one can infer how strong and persistent the spatial fluctuations are with varying temperature , age and imposed shear - rate .
it is evident that for large shear - rates ( @xmath68 ) , these fluctuations are negligible at any age of the glassy state sampled at any of the temperatures ; see fig . [ fig8](a),(d),(h ) .
on the other hand , at smaller shear - rates ( @xmath37 ) , large fluctuations are visible at all temperatures .
however , the age of the sample then becomes a factor in determining the degree of fluctuations and persistence with increasing strain . at a lower temperature , @xmath38 ,
these are prominent over the range of ages we investigated and also persistent over large strains ; see fig .
[ fig8](i ) . on the other hand , at higher temperatures , @xmath46 ( fig .
[ fig8](c ) ) , that scale of fluctuations is only slightly visible at very large ages over a small strain window .
the range of ages and strain - interval over which similar fluctuations are seen broadens out at an intermediate temperature of @xmath69 ( fig .
[ fig8](f ) ) and , even more extensively , at @xmath38 .
this trend is also evident at an intermediate shear - rate of @xmath70 , with the scale of fluctuations being less but visible ( fig .
[ fig8](h ) ) ; howevever , it is important to note that the fluctuations are nearly negligible at the higher temperature @xmath46 at this shear - rate ( fig .
[ fig8](b ) ) .
thus , we demonstrate , that the scale of transient dynamical heterogeneities depend on a combination of temperature of the glass , its age and the imposed shear - rate . since the quantity we compute , @xmath67 , measures the contrast in mobilities in layers parallel to the flow direction , the conclusions drawn from the contour maps allow us to infer that the propensity to shear - band during yielding is determined on how rigid the glass is and how small / large the applied shear - rate is .
finally , we study how a transient shear - band , once formed , spatially evolves with time .
such analysis can only be done for the case where a well - defined shear - band can be identified .
thus , we focus at the case of @xmath38 and small shear - rate ( @xmath37 ) , where we observe transient shear - bands for a range of @xmath0 . in order to further quantify and characterise the spatio - temporal evolution of the shear - band , we define a region to be mobile or not , by setting a threshold @xmath71 on the local @xmath72 . as can be seen in fig .
[ fig3](a ) , such a choice of @xmath73 is larger than the plateau value in the msd and thus corresponds to motions beyond cage - breaking . we then define the local mobility @xmath74 as @xmath75 where @xmath76 is the average msd of particles in a sub - box . following this convention , we digitize the whole system into mobile and immobile regions .
an example of the mobility map , thus constructed , is shown in fig .
[ fig9](a ) .
we quantify the extent of fluidisation in the system by measuring the fraction of mobile cells , @xmath77 , at any given instant .
the evolution of @xmath77 as a function of strain , for an imposed shear of @xmath37 , is shown in fig .
[ fig9](b ) .
we observe that independent of age ( @xmath0 ) , the sudden initial increase of @xmath77 occurs nearly at the same strain .
however , the subsequent increase in @xmath77 does depend upon the age and thus , the fraction of system which is mobilised at @xmath78 is substantially larger for @xmath55 as compared to @xmath50 .
further , we try to identify the spatial organisation of these mobile cells . by locating contiguous layers of mobile cells
, we identify the formation of the shear band and , thereafter , by marking the interfaces of this band , we measure how the band - width , @xmath79 , evolves with time . for different ages of the initial glassy state ,
this time evolution is shown in fig .
[ fig9](c ) , for a fixed @xmath37 .
we see that @xmath79 initially grows quickly and then eventually it reaches a regime where the data can be fitted with @xmath80 , implying that the propagating interface of the shear band has a diffusive motion .
this diffusive regime is not extensively dependent on @xmath0 , although samples having largest @xmath0 do seem to display a slower motion of the spreading interface .
to summarize , we have used numerical simulations of a model glass former , to probe the mechanical response of amorphous solids , having different aging histories , by imposing a wide range of external shear rates .
the response is studied at different temperatures below the mode coupling temperature , in order to ascertain how thermal fluctuations affect the spatio - temporal response .
we underline , here , that the numerical simulations are done by integrating the newton s equations of motions using lees - edwards periodic boundary conditions and the local dpd thermostat .
this , thus , allows for unhindered spatio - temporal fluctuations as the glassy state responds to the applied shear , without introducing any biases or suppressing any fluctuations , which would be the case if one were using walls @xcite or integrating the sllod equations along with some global thermostat ( like nose - hoover ) @xcite .
therefore , qualitatively , the nature of dynamical heterogeneities observed could be different from the earlier works .
the macroscopic response of the glass is characterized by the occurrence of a overshoot in the stress - strain curves , observed for the range of shear - rates and ages that we have studied , at the different temperatures .
consistent with earlier works , we find that , at each temperature , the peak height can be scaled on to a master - curve as a function of @xmath81 .
the corresponding potential energies of the system , measured during the deformation process , show a dependence on age with decreasing shear rate .
this implies that the structures visited during the transient regime , prior to the onset of steady flow , start becoming different , depending upon the initial history .
we also observe that for lower temperatures , it takes longer for potential energies to reach the steady state value at smaller shear - rates , reflecting the role of thermal fluctuations in affecting the duration of the transient regime for these parameters .
this is further revealed , by measuring dynamical quantities at the microscopic scale .
our probe of choice is the time evolution of non - affine mean squared displacement ( msd ) of the particles relative to the initial quiescent state .
spatial profiles of the msd show that for a fixed imposed shear and age , the transient dynamics is more heterogeneous and more long - lived at lower temperatures .
this is illustrated by constucting sptatially resolved three - dimensional maps which provide a handle to visualize the extent of dynamical heterogeneity as the glassy states respond to the imposed shear .
such maps demonstrate that the heterogeneities are spatially localised in the form of shearbands , for old enough systems at low temperatures .
we quantify the extent of dynamical heterogeneity organised in the form of shear - bands , by measuring the degree of fluctuations , @xmath61 , in local mobilities across regions oriented parallel to the direction of imposed flow .
thus , large values in @xmath61 reflect more propensity of the system to exhibit shear - banding .
we do such measurements by scanning across entire range of shear - rates and thermal histories explored by us .
we observe that the fluctuations are maximum around the yielding strain and the peak in fluctuations increase with decreasing shear - rate as well as larger ages .
further by constructing contour maps of @xmath82 for varying shear - rates at different temperatures , we demonstrate that the emergence of significant shear - banding depends upon the combination of temperature , age and imposed shear - rate .
the contrast in local mobilites is most prominent at low shear - rates and low temperatures , where the age of the system does not seem to matter , over the range explored by us .
with increasing temperature , the age of the sample starts to determine whether transient shear - banding will be visible or not . for even higher temperatures
, one has to go to lower shear - rates and even older systems in order for such largely localized heteroegeneities to be visible . also , in the cases where shear - bands can be identified , we observe that the spreading of the shear - band across the system seems to depend on age , albeit weakly . thus , our study shows that the transient shearbanding , with sharply contrasting mobilities across regions , only occur for specific combinations of temperature , shear - rate and age , as discussed above . note that , for most of the glassy samples that we study , the macroscopic response under an imposed strain - rate is characterized by the stress overshoot .
yet , the transient shear - bands are only visible in only a subset of the cases we study .
thus , the occurrence of a stress overshoot does not necessarily lead to the transient shearbands , quantified via the mobilities relative to the initial quiescent state , as predicted by the spatially resolved fluidity and sgr models @xcite . on the other hand , as predicted by these and other models @xcite , we do observe that the propensity to shearband is more in the case of increasingly aged samples .
however , the choice of disorder distributions in these models ( e.g. distribution of yielding thresholds ) , with changing age , remains arbitrary , and it would be useful to obtain these as inputs from atomistic simulations
. the transient shear - banding observed in our numerical simulations as well as other , bears resemblance to dynamical heterogeneities observed in thermal glasses @xcite , except that in the case of a sheared glass , strong anisotropies come into play leading to the spatial localisation . in the case of supercooled liquids ,
the scale of dynamical heterogeneities increases as the temperature is decreased towards the putative glass transition temperature .
similarly , in the case of applied shear , the heterogeneous dynamics is enhanced as one approaches vanishingly small shear - rates , with the yield stress being identified as a critical threshold @xcite .
also , it is possible that the regime of imposed shear - rates over which the critical threshold influences flow @xcite could change with temperature , with decreasing thermal fluctuations extending this regime over longer scales .
this would rationalise why with increasing temperatures , the contrast in heterogeneities progressively decreases , for a fixed shear - rate . over and above
, the aging effects are more prominent at higher temperatures , with the glass remembering the history from where it was quenched over longer aging timescales @xcite . at low temperature , thus , structural arrest and therefore rigidity emerges at shorter timescale .
hence , the mechanical response of the glass , having the same age and same imposed shear - rate , would be different across temperatures .
this interplay of different timescales therefore brings about the complex response as a function of all these control parameters .
eventually , it would be useful to quantify the exact ranges in these control parameters , where we expect transient but prominent shearbanding to be visible .
the transient dynamical heterogeneities that we track and analyse are similar to measurements done via confocal microscopy in colloidal glasses @xcite or scattering measurements in granular systems @xcite .
this is different from using velocity profiles as a diagnostic tool , which capture plasticity over short timescales .
further , in our case , because of periodic boundary conditions , the shear bands can emerge anywhere in the system , unlike many cases where initial mobile regions happen to occur near confining surfaces @xcite .
however , similar to experiments , we do observe the propagation of the mobile front to eventually fluidize the system .
* acknowledgement : * + we achnowledge financial support by the deutsche forschungsgemeinschaft ( dfg ) in the framework of the priority programme spp 1594 ( grant no .
ho 2231/8 - 2 ) .
pc thanks for 1394 for financial support during visit to university of dsseldorf
. 99 binder , k. and kob , w. , _ glassy materials and disordered solids : an introduction to their statistical mechanics _ , ( world scientific , singapore , 2011 ) .
rodney , d. , tanguy , a. , and vandembroucq , d. , modeling the mechanics of amorphous solids at different length scale and time scale , _ modelling simul .
_ * 19 * , 083001 ( 2011 ) .
barrat , j .- l . and lemaitre , a. dynamical heterogeneities in glasses , colloids , and granular materials , chap . 8 , edited by l. berthier , g. biroli , j .-
bouchaud , l. cipelletti , and w. van saarloos ( oxford university press , oxford , 2011 ) .
zausch , j. , horbach , j. , laurati , m. , egelhaaf , s. u. , brader , j. m. , voigtmann , t. , and fuchs , m. , from equilibrium to steady state : the transient dynamics of colloidal liquids under shear , _ j. phys . : condens
. matter _ * 20 * , 404210 ( 2008 ) .
varnik , f. , bocquet , l. , and barrat , j .- l .
, a study of the static yield stress in a binary lennard - jones glass , _ j. chem .
phys . _ * 120 * , 2788 ( 2004 ) .
rottler , j. and robbins , m. o. , unified description of aging rate effects in yield of glassy solids , _ phys .
lett . _ * 95 * , 225504 ( 2005 ) .
schuh , c. a. , hufnagel , t. c. , and ramamurty , u. , mechanical behavior of amorphous alloys , _ acta mater .
_ * 55 * , 4067 ( 2007 ) .
mller , p. c. f. , rodts , s. , michels , m. a. j. , and bonn , d. , shear banding and yield stress in soft glassy materials , _ phys .
e _ * 77 * , 041507 ( 2008 ) .
divoux , t. , fardin , m. a. , manneville , s. , and lerouge , s. , shear banding of complex fluids , _ ann .
rev . of fluid mech .
_ , * 48 * , 81 ( 2016 ) . fielding , s. m. , shear banding in soft glassy materials , _ rep .
phys . _ * 77 * , 102601 ( 2014 ) .
chikkadi , v. , wegdam , g. , bonn , d. , nienhuis , b. , and schall , p. , long - range strain correlations in sheared colloidal glasses .
lett . _ * 107 * , 198303 ( 2011 ) .
besseling , r. , isa , l. , ballesta , p. , petekidis , g. , cates , m. e. , and poon , w. c. k. , shear banding and flow - concentration coupling in colloidal glasses , _ phys .
* 105 * , 268301 ( 2010 ) .
bokeloh , j. , divinski , s. v. , reglitz , g. , and wilde , g. , tracer measurements of atomic diffusion inside shear bands of a bulk metallic glass , _ phys .
lett . _ * 107 * , 235503 ( 2011 ) .
divoux , t. , tamarii , d. , barentin , c. , and manneville , s. , transient shear banding in a simple yield stress fluid .
* 104 * , 208301 ( 2010 ) .
varnik , f. , bocquet , l. , barrat , j .- l . , and berthier , l. , shear localization in a model glass .
lett . _ * 90 * , 095702 ( 2003 ) .
chaudhuri , p. and horbach , j. , onset of flow in a confined colloidal glass under an imposed shear stress , _ phys .
e _ * 88 * , 040301(r ) ( 2013 ) .
irani , e. , chaudhuri , p. , and heussinger , c. , impact of attractive interactions on the rheology of dense athermal particles , _ phys .
* 112 * , 188303 ( 2014 ) .
shrivastav , g. p. , chaudhuri , p. , and horbach , j. , formation and growth of shear bands in glasses : existence of an underlying , directed percolation transition , preprint arxiv:1506.03049 .
shi , y. and falk , m. l. , atomic - scale simulations of strain localization in three - dimensional model amorphous solids , _ phys .
b _ * 73 * , 214201 ( 2006 ) . bailey , n. p. , schiotz , j. , and jacobsen , k. w. , atomistic simulation study of the shear - band deformation mechanism in mg - cu metallic glasses , _ phys .
b _ * 73 * , 064108 ( 2006 ) .
chaudhuri , p. , berthier , l. , and bocquet , l. , inhomogeneous shear flows in soft jammed materials with tunable attractive forces , _ phys .
e _ * 85 * 021503 ( 2012 ) .
dasgupta , r. , hentschel , h. g. e. , and procaccia , i. , microscopic mechanism of shear bands in amorphous solids , _ phys .
lett . _ * 109 * 255502 ( 2012 ) .
nordstrom , k. n. , verneuil , e. , arratia , p. e. , basu , a. , zhang , z. , yodh , a. g. , gollub , j. p. , and durian , d. j. , microfluidic rheology of soft colloids above and below jamming , _ phys .
lett . _ * 105 * , 175701 ( 2010 ) .
bonn , d. , paredes , j. , denn , m. m. , berthier , l. , divoux , t. and manneville , s. , yield stress materials in soft condensed matter , preprint arxiv:1502.0528 .
martens , k. , bocquet , l. , and barrat , j .- l . , spontaneous formation of permanent shear bands in a mesoscopic model of flowing disordered matter , _ soft matter _ * 8 * , 4197 ( 2012 ) .
coussot , p. and ovarlez , g. , physical origin of shear - banding in jammed systems , _ eur .
j. e _ * 33 * , 183 ( 2010 ) .
divoux , t. , tamarii , d. , barentin , c. , teitel , s. and manneville , s. , yielding dynamics of a herschel - bulkley fluid : a critical - like fluidization behaviour , _ soft matter _ * 8 * , 4151 ( 2012 ) .
tsamados , m. , plasticity and dynamical heterogeneity in driven glassy materials , _ euro .
j. e _ * 32 * , 165 ( 2010 ) .
moorcroft , r. l. , cates , m. e. , and fielding , s. m. , age - dependent transient shear banding in soft glasses , _ phys .
lett . _ * 106 * , 055502 ( 2011 ) .
moorcroft , r. l. and fielding , s. m. , criteria for shear banding in time - dependent flows of complex fluids .
lett . _ * 110 * , 086001 ( 2013 ) . manning , m. l. , langer , j. s. and carlson , j. m. , strain localization in a shear transformation zone model for amorphous solids , _ phys .
e _ * 76 * , 056106 ( 2007 ) .
manning , m. l. , daub , e. g. , langer , j. s. and carlson , j. m. , rate - dependent shear bands in a shear - transformation - zone model of amorphous solids , _ phys .
e _ * 79 * , 016110 ( 2009 ) .
shi , y. , katz , m. b. , li , h. and falk , m. l. , evaluation of the disorder temperature and free - volume formalisms via simulations of shear banding in amorphous solids , _ phys .
* 98 * , 185505 ( 2007 ) .
jagla e. a. , shear band dynamics from a mesoscopic modeling of plasticity , _ j. stat .
_ p12025 ( 2010 ) .
vandembroucq , d. and roux , s. , mechanical noise dependent aging and shear banding behavior of a mesoscopic model of amorphous plasticity , _ phys .
b _ * 84 * , 134210 ( 2011 ) .
kob , w. and andersen , h. c. , scaling behavior in the @xmath83-relaxation regime of a supercooled lennard - jones mixture , _ phys .
lett . _ * 73 * , 1376 ( 1994 ) .
plimpton , s. , fast parallel algorithms for short - range molecular dynamics , _ j. comp .
* 117 * , 1 ( 1995 ) .
soddemann , t. , dnweg , b. , and kremer , k. , dissipative particle dynamics : a useful thermostat for equilibrium and nonequilibrium molecular dynamics simulations , _ phys .
e _ * 68 * , 046702 ( 2003 ) .
lees , a. w. and edwards , s. f. , the computer study of transport processes under extreme conditions , _
c : solid state phys . _ * 5 * , 1921 ( 1972 ) .
derec , c. , ducouret , g. , ajdari , a. and lequeux , f. , aging and nonlinear rheology in suspensions of polyethylene oxide protected silica particles , _ phys .
e _ * 67 * , 061403 ( 2003 ) .
divoux , t. , barentin , c. and manneville , s. , stress overshoot in a simple yield stress fluid : an extensive study combining rheology and velocimetry , _ soft matter _ , * 7 * , 9335 ( 2011 ) .
park , j. d. and ahn , k. h. , structural evolution of colloidal gels at intermediate volume fraction under start - up of shear flow , _ soft matter _ * 9 * , 11650 ( 2013 ) .
whittle , m. and dickinson , e. , stress overshoot in a model particle gel , _ j. chem
. phys . _ * 107 * , 10191 ( 1997 ) amon , a. , bruand , a. , crassous , j. , and clment , j. , hot spots in an athermal system , _ phys .
lett . _ * 108 * , 135502 ( 2012 ) .
sentjabrskaja , t. , chaudhuri , p. , hermes , m. , poon , w. c. k. , horbach , j. , egelhaaf , s. u. , and laurati , m. , microscopic dynamics during creep in colloidal glasses , _ sci .
rep . _ * 5 * , 11884 ( 2015 ) .
bocquet , l. , colin , a. , ajdari , a. , kinetic theory of plastic flow in soft glassy materials _ phys .
lett . _ * 103 * , 036001 ( 2009 ) .
liu , c. , ferrero , e. e. , puosi , f. , barrat , j .- l . ,
martens , k. , driving rate dependence of avalanche statistics and shapes at the yielding transition , _ phys .
lett . _ * 116 * , 065501 ( 2016 ) .
|
molecular dynamics computer simulations of a binary lennard - jones glass under shear are presented .
the mechanical response of glassy states having different thermal histories is investigated by imposing a wide range of external shear rates , at different temperatures .
the stress - strain relations exhibit an overshoot at a strain of around 0.1 , marking the yielding of the glass sample and the onset of plastic flow .
the amplitude of the overshoot shows a logarithmic behavior with respect to a dimensionless variable , given by the age of the sample times the shear rate .
dynamical heterogeneities having finite lifetimes , in the form of shear bands , are observed as the glass deforms under shear . by quantifying the spatial fluctuations of particle mobility ,
we demonstrate that such shearbanding occurs only under specific combinations of imposed shear - rate , age of glass and ambient temperature .
|
in this section we focus on the attractive case @xmath7 . the transition from the psf phase to the cdw phase , obtained for decreasing hopping ratio @xmath82 , is characterized by an inversion of the hierarchy between the pairing correlations , @xmath111 , and the density - density correlations @xmath112 . in the psf phase @xmath113 , while in the cdw phase @xmath114 .
the @xmath31 exponent , extracted from the low-@xmath74 behavior of the density structure factor as discussed in the text , is shown in fig . [ f.psf-cdw ] as a function of @xmath82 for various system sizes .
a finite - size extrapolation allows to identify the critical @xmath82 value at which @xmath31 traverses the critical value @xmath115 .
the composite @xmath2-@xmath3 pairs form therefore a tll with effective exponent @xmath116 . in the cdw phase
, we obtain a tll with @xmath117 , which is quite remarkable in a system with on - site interactions only .
indeed in one - component tlls such a situation is realized only in presence of non - local interactions , such as nearest - neighbor or dipolar ones ( r. citro _ et al . _ , new j. phys . 10 , 045011 ( 2008 ) ; t. roscilde and m. boninsegni , new j. phys .
* 12 * , 033032 ( 2010 ) ; m. dalmonte _
et al . _ , phys rev lett . * 105 * , 140401 ( 2010 ) ) or when considering highly - excited states such as a super - tonks - girardeau gas ( g. e. astrakharchik _ et al . _ ,
lett . * 95 * , 190407 ( 2005 ) ) . in our case , although the interactions between the @xmath2 and @xmath3 particles are on - site only , the bound @xmath2-@xmath3 pairs have effective longer - range interactions , decaying exponentially with the distance . these effective interactions
are responsible for the transition to crystalline order ( phase tc2 ) at extreme mass imbalance .
this transition occurs at the critical value @xmath118 , where @xmath119@xcite . since in our case @xmath120
, one obtains @xmath121 at the transition line , and @xmath122 in the crystalline phase , marking the onset of long - range density - density correlations .
this implies the appearance of a peak diverging linearly in @xmath123 at wavevector @xmath124 in the structure factor for the total density @xmath104 , or alternatively in the density structure factor of both species @xmath125~.\ ] ] the onset of a linear - in-@xmath123 divergence of the structure factor peak is shown in fig .
[ f.cdw-tc ] , where the peak value divided by @xmath123 is seen to converge to a finite value in the thermodynamic limit .
this criterion has been used to determine the phase boundary between tc2 and cdw in fig .
[ f.phd ] .
a further clear signature of the tc2 phase is the exponential suppression of the pairing correlator @xmath34 , in sharp contrast with the algebraic decay of both cdw and psf phases .
we have used dmrg calculations with up to @xmath20 sites for _ fermionic _ @xmath2 and @xmath3 species in order to determine the extent of the tc2 phase ( diamonds in fig .
[ f.phd ] ) . a typical set of data for @xmath126
is presented in fig .
[ f.gab_corr ] : there it is seen that for @xmath127 and @xmath128 the correlator @xmath34 decays exponentially ; interestingly , for _ stronger _ attraction , _
i.e. _ for more tightly bound pairs , the correlation function acquires again an algebraic decay , showing the re - entrant nature of the tc2 phase .
we observe that , close to the onset of the tc2 , the decay of the @xmath34 correlator is indeed consistent with an exponent @xmath121 . and @xmath3 particles at @xmath127 and various @xmath82 values .
solid lines are fits to the form @xmath129.,title="fig:",width=264 ] and @xmath3 particles at @xmath127 and various @xmath82 values .
solid lines are fits to the form @xmath129.,title="fig:",width=264 ] as mentioned in the main text , the entanglement entropy ( ee ) allows to extract the central charge of the system@xcite . in fig .
[ f.ee_sup ] , we consider two different approaches to extract @xmath66 . in the left panel ,
we plot @xmath131 as a function of the block length @xmath67 at a fixed hopping asymmetry @xmath46 and fixed length @xmath123 , for different values of @xmath40 ; in the repulsive case , the numerical results are compatible with a completely gapless system , whereas in the attractive case the emergence of the spin gap reduces the central charge to 1 . in the right panel , we plot the ee of the half chain @xmath132 which , for @xmath133 , does not exhibit any oscillatory correction . for @xmath6 , all datas fit very well with the cft prediction with @xmath134 , whereas a clear bending at large system sizes indicates that a gap opens in the attractive case .
. left panel : ee scaling for @xmath135 and different values of @xmath40 ; for @xmath136 and @xmath7 , results are compatible with @xmath48 ( black thick line ) and @xmath49 ( dashed line ) respectively .
right panel : scaling of ee for @xmath137 at different system sizes . in the repulsive case , datas are compatible with @xmath48 , whereas in the attractive one datas for larger system deviates toward @xmath49 at large @xmath123 ; dashed lines are linear fits with @xmath138 from top to bottom , whereas the black , thick line represents @xmath48 .
data points for @xmath139 and @xmath140 have been shifted by @xmath141 to improve readability .
, title="fig:",width=158 ] . left panel : ee scaling for @xmath135 and different values of @xmath40 ; for @xmath136 and @xmath7 , results are compatible with @xmath48 ( black thick line ) and @xmath49 ( dashed line ) respectively .
right panel : scaling of ee for @xmath137 at different system sizes . in the repulsive case , datas are compatible with @xmath48 , whereas in the attractive one datas for larger system deviates toward @xmath49 at large @xmath123 ; dashed lines are linear fits with @xmath138 from top to bottom , whereas the black , thick line represents @xmath48 . data points for @xmath139 and @xmath140 have been shifted by @xmath141 to improve readability . , title="fig:",width=158 ] fig .
[ f.cnnbar ] shows the nearest - neighbor correlation function for @xmath3 holes , @xmath142 in the repulsive case .
the calculation is performed via exact diagonalization on a chain with @xmath143 sites , which allows to access properly the true ground state in the limit @xmath108 , both in the case of crystalline order and of phase separation .
the latter limit is hard to reproduce appropriately with qmc or dmrg simulations , due to the abundance of metastable states close in energy to the ground state .
we see that @xmath106 is very sensitive to the succession of phases traversed by the system .
as explained in the text , for small @xmath40 values @xmath106 is suppressed as @xmath108 due to the formation of the crystal of @xmath144 trimers , implying that @xmath106 has to attain its minimum value @xmath145 .
as @xmath82 is reduced , two competing effects result in a non - trivial evolution of @xmath146 .
the formation of @xmath144 trimers leads to an enhancement of the @xmath106 correlator : the kinetic repulsion between the @xmath93 holes , which would be normally present in a 1d single - component gas with hardcore repulsion , is strongly suppressed due to the decrease of the hopping ratio @xmath82 , and therefore it is overcome by the binding effect of the light @xmath2 particles , which gain energy by delocalizing over two ( or more ) @xmath93 holes . yet the trimers that form via this mechanism experience an effective mutual repulsion at sufficiently small @xmath40 values , due to the kinetic repulsion between light @xmath2 particles .
this leads to the suppression of configurations with adjacent trimers ; a complete suppression of such configurations would lead the system to the trimer crystal state , which , as mentioned before and in the text , corresponds to an absolute minimum of @xmath146 . given that trimers are only loosely bound for small @xmath40 , the effect of the repulsion between trimers dominates over the effect of trimer formation , so that @xmath146 is seen to decrease monotonically for decreasing @xmath82 ( @xmath147 data in fig .
[ f.cnnbar ] ) . on the other hand , for larger @xmath40
, the enhancement of @xmath146 due to trimer formation dominates over the suppression due to trimer repulsion , so that @xmath146 is seen to increase for decreasing @xmath82 down to the transition point to the crystalline phase , at which it drops down to the @xmath148 limit ( @xmath86 data in fig .
[ f.cnnbar ] ) . on the other hand , for even larger
@xmath40 the effective attraction between @xmath93 holes mediated by the @xmath2 particles begins to involve more than two holes at once , as the @xmath2-@xmath3 interaction overcomes progressively the kinetic repulsion between @xmath2 particles and favors their aggregation over extended clusters of holes .
this phenomenon culminates into the phase separation between hole - rich and hole - free regions@xcite upon decreasing @xmath82 , namely upon suppression of the kinetic repulsion between holes .
this succession of phenomena leads to a much stronger enhancement of @xmath146 as @xmath82 decreases , with a steep increase when phase separation is achieved .
when full phase separation is realized in the limit @xmath108 , @xmath146 takes the value @xmath149/l$ ] ( namely @xmath150 for @xmath143 , as seen in fig .
[ f.cnnbar ] ) . in the following
, we will study the dependence on the tll parameters of the generalized string correlation function : @xmath151\tilde{s}^z_{i+r}\right\rangle\ ] ] where we have defined @xmath152 . after applying standard bosonization identities to @xmath153 operators by introducing the bosonic @xmath154 fields@xcite
, one can define _ effective _ spin and charge fields as @xmath155 such that : @xmath156 where @xmath157 are vertex operators related to the charge and spin fields and @xmath158 is a constant ; additional contributions with higher scaling dimension play no significant role in the remaining , and can thus be neglected . the non - local contribution in eq . [ corr_string ]
can be reduced to : @xmath159 & = & \exp[i\pi\beta \int^{i+r}_i dy \frac{\partial_y\phi_{\sigma}(y)}{\sqrt{\pi}}]=\nonumber\\ & = & \exp[i\sqrt{\pi}\beta [ \phi_{\sigma}(i+r)-\phi_{\sigma}(i)]]\nonumber.\end{aligned}\ ] ] in addition to a pure spin contribution which stems from the first term in eq . , and which decays faster than @xmath160 , the string correlator contains a more complicated , second contribution , stemming from vertex operators , and which we will indicate as @xmath161 . assuming spin - charge separation ,
@xmath161 can be written as : @xmath162 whose asymptotic decay is affected by both charge and spin contributions : the charge part has a fixed exponent , whereas the spin one @xmath163 has a @xmath164-dependent form ; in particular , for @xmath165 correlation considered in this work , the slowest decaying part of @xmath163 is a constant , so that the dominant term in the string correlation functions decays as : @xmath166 it is worth noticing that , from eq .
, one can easily recover standard spin correlations for @xmath167 decaying as @xmath168 .
the fact that the decay of string correlations is uniquely governed by the @xmath25 exponent is indeed remarkable ; as discussed in the main text , this leads to a string structure factor diverging with system size as @xmath102 ( and @xmath114 for all @xmath82 values ) .
this result has to be compared with the scaling of the peaks of the charge and spin structure factor , namely for @xmath33 and @xmath169 with @xmath124 .
these peaks come from the dominant oscillating term ( going like @xmath170 ) of the @xmath171 and @xmath172 correlation functions , whose amplitude decays like @xmath173 .
this means that in general the @xmath174-peak height scales with system size as @xmath175 . in the sg phase the @xmath91 exponent vanishes for @xmath176 , so that , strictly speaking , the true asymptotic scaling of @xmath33 and @xmath169 is indeed the same as that of @xmath177 . but _ in practice _
the scaling of @xmath178 to zero is extremely slow typically logarithmic , a fact which generates a huge finite - size correction to the effective scaling of @xmath33 and @xmath169 . in our simulations
we observe that these quantities grow more slowly than @xmath80 over the entire sg phase for all the system sizes we considered ( up to @xmath41 see fig .
[ f.sq ] ) . to see them
diverge at all with system size one might need exceedingly big system sizes , both for simulations and for experiments .
therefore string correlations stand as a unique tool to characterize the correlations in the sg phase .
in this section we provide further details of the estimate of the transition line between the sg and the tc1 phase in the repulsive case .
the transition is obtained by a joint finite - size scaling analysis .
we focus on the scaling of the peak in the structure factors @xmath180 appearing at @xmath181 , and diverging linearly with system size when entering the tc1 phase ; and we consider as well the scaling of the superfluid density for both species , @xmath182 , which scales to zero in the thermodynamic limit when in the tc1 phase . fig .
[ f.sg-tc1 ] shows that the scaling of both quantities give a consistent estimate of the critical @xmath183 value for the onset of crystalline order .
the transition to the tc1 phase was already estimated by one of the authors in ref .
@xcite for the same density .
the estimate of the transition line we present here is corrected with respect to that given in ref .
indeed in ref .
@xcite the estimate of @xmath183 was biased by the choice of a scaling ansatz on the superfluid stiffness which , albeit satisfactorily verified within the statistical uncertainties , turns out to not be fully justified .
the current estimate is not based on any scaling ansatz , and it delivers a transition line which is shifted to larger @xmath183 values ( typically by about 10 - 20 % ) with respect to the estimates in ref .
. moreover ref .
@xcite identified a narrow region of parameters close to the transition line as a _ super - tonks _
regime , characterized by a sub - linearly diverging peak in the density structure factor of both species .
the subtle effect of the opening of a spin gap before entering the crystalline phase was not observed in ref .
@xcite , and this led to the erroneous claim of algebraic off - diagonal correlations for both species in the super - tonks regime .
our current results point to the fact that the above regime is rather characterized by superfluidity of both atomic species ( which persists up to the transition , as shown in fig .
[ f.sg-tc1 ] ) but with exponentially decaying off - diagonal correlations , due to the presence of a spin gap .
|
we numerically determine the very rich phase diagram of mass - imbalanced binary mixtures of hardcore bosons ( or equivalently fermions , or hardcore - bose / fermi mixtures ) loaded in one - dimensional optical lattices .
focusing on commensurate fillings away from half filling , we find a strong asymmetry between attractive and repulsive interactions .
attraction is found to always lead to pairing , associated with a spin gap , and to pair crystallization for very strong mass imbalance . in the repulsive case
the two atomic components remain instead fully gapless over a large parameter range ; only a very strong mass imbalance leads to the opening of a spin gap .
the spin - gap phase is the precursor of a crystalline phase occurring for an even stronger mass imbalance .
the fundamental asymmetry of the phase diagram is at odds with recent theoretical predictions , and can be tested directly via time - of - flight experiments on trapped cold atoms .
one - dimensional quantum liquids occupy a special place in the context of quantum many - body systems : indeed interactions of any strength lead to quantum fluctuations as strong as to discard bose condensation for bosons and the fermi liquid picture for fermions down to zero temperature . for sufficiently weak interactions
a new unifying paradigm of the so - called tomonaga - luttinger liquids ( tll ) emerges @xcite , characterized by the fact that all elementary excitations are gapless , and both diagonal and off - diagonal correlations decay algebraically with the distance .
recent advances in the trapping of ultracold atoms in optical lattices allow to realize one - dimensional quantum liquids in a highly flexible way , with the possibility of fully controlling the statistics and the interaction strength @xcite .
a series of recent experiments has demonstrated the physics of one - dimensional bose gases with strong interactions up to the hardcore ( or tonks - girardeau ) limit @xcite . a special role in the context of one - dimensional systems
is played by binary mixtures , either bosonic , fermionic , or bose - fermi ones , for which tll theory predicts the separation of spin and charge modes @xcite . in the case of particles with equal masses and repulsive short - range interactions , both charge and spin sectors
can be gapless , and one recovers an effective picture of two decoupled tlls .
such a picture can be made unstable via several mechanisms : via mott localization in presence of an underlying lattice and for integer total filling ; via localization into a true ( long - range - ordered ) crystal ( tc ) in presence of a strong off - site repulsion ; via phase separation ; or via the formation of bound states ( _ e.g. _ cooper pairs for attractive interactions ) leading to the appearance of a spin gap . .
the phase boundaries are determined via qmc ; the open diamonds indicate points in parameter space for which dmrg finds a fully gapped crystal phase tc2 .
the dashed line indicate the fk gap @xmath0 for kink - antikink pairs .
the left panel presents a sketch of the phases : 2 tll = 2 tomonaga - luttinger liquids ; ps = phase separated ; sg = spin gap ; psf = pair superfluid ; cdw = algebraic charge density wave ; tc1 and
tc2 : true crystals.,width=321 ] here we show that the tll picture undergoes a complex series of instabilities in binary mixtures with _ mass imbalance _ between the two species .
we focus here on quantum particles on a lattice with intraspecies hardcore repulsion and on - site interspecies interactions , describing at the same time spin-1/2 fermions , spin-1/2 hardcore bosons , and mixtures of hardcore bosons and spinless fermions . the system hamiltonian reads @xmath1\ ] ]
in which @xmath2 and @xmath3 correspond to the two atomic species .
mass imbalance is controlled by the ratio @xmath4 and interaction by the ratio @xmath5 .
as far as the spectrum and the diagonal observables are concerned , we do not need to specify the statistics of the @xmath2 and @xmath3 operators ; for what concerns off - diagonal observables , unless otherwise specified we will refer explicitly to hardcore bosons , satisfying bosonic commutation relations off - site and fermionic anticommutation relations on - site .
the denomination of the many - body phases will also be mostly inspired by the case of hardcore - boson mixtures .
we consider both repulsive ( @xmath6 ) and attractive ( @xmath7 ) interactions , and we focus on the case of equal densities away from half filling , @xmath8 , @xmath9 .
the system with equal masses , @xmath10 , is integrable @xcite , and features spin - charge separation into two tll in the repulsive case ; in the attractive one , an @xmath2-@xmath3 bound state appears , associated with the opening of a spin gap , and only the charge sector remains gapless , giving rise to a paired superfluid ( psf ) phase .
the case of mass imbalance has been studied recently by bosonization @xcite , and numerically via the density - matrix renormalization group ( dmrg ) and related approaches @xcite . making use of numerically exact methods , here we determine comprehensively the rich phase diagram of the system with mass imbalance , @xmath11 , as shown in fig .
[ f.phd ] .
the latter figure refers to the case @xmath12 , but the qualitative features are generic for commensurate fillings .
our main findings are the following : 1 ) in the attractive case , the psf phase is found to persist up to strong imbalance , at which the system becomes unstable to the formation of a tc of pairs .
2 ) in the repulsive case , on the other hand , the double tll of the mass - balanced case survives up to a large imbalance , at which two different instabilities appear . for sufficiently large repulsion , the two species phase - separate . for weaker repulsion ,
first a spin - gap opens , associated with fluctuating magnetic order captured by string correlations ; this phase is the precursor of a tc phase with pinned @xmath3 particles , and @xmath2 particles localized in the interstitial regions . before discussing the derivation of the phase diagram ,
we point out that the physics described in this paper is largely accessible to current experimental setups on ultracold mixtures in one - dimensional optical lattices .
the mass imbalance can be realized with heteronuclear mixtures ( _ e.g. _ @xmath13k-@xmath14li , for the fermionic case , @xmath15k-@xmath16rb for the bosonic case , and @xmath13k-@xmath16rb for the bose - fermi case @xcite ) , and with homonuclear mixtures in different hyperfine states , and it can be continuously tuned by using lasers with a wavelength close to the magic value or to an atomic resonance for one of the two species . the interspecies interaction can be tuned by feshbach resonances , as widely demonstrated in the recent literature @xcite .
the phase diagram in fig .
[ f.phd ] is the result of a joint numerical study based on quantum monte carlo ( qmc ) and dmrg@xcite .
our qmc calculations are based on a canonical formulation @xcite of the stochastic series expansion approach with directed loops @xcite , applied to chains with up to @xmath17 sites with periodic boundary conditions , and at temperatures @xmath18 capturing the @xmath19 physics for both species .
our dmrg calculations apply to chains with up to @xmath20 with open boundary conditions and retaining up to @xmath21 states .
we start our discussion by the attractive case . for a very broad range of mass imbalance ,
the system displays a psf phase , characterized by quasi - condensation of bound @xmath2-@xmath3 pairs , giving rise to an algebraic decay of the pairing correlation function @xcite , @xmath22 ; density - density correlations are also decaying algebraically as @xmath23 ; here , @xmath24 , and @xmath25 is the charge tll parameter .
for equal masses , @xmath26 @xcite for all @xmath27 , so that the dominant correlations are the pairing ones .
moreover @xmath28 due to the presence of the spin gap ( see below ) .
we extract the luttinger exponent from the slope of the density structure factor at @xmath29 , @xmath30 , and we find that mass imbalance leads to a reduction of @xmath31 , consistently with what observed for other fillings in ref . @xcite . for large mass imbalance , @xmath25 becomes smaller than one : this corresponds to the loss of quasi - condensation , in favor of a quasi - solid phase ( or charge density wave , cdw ) , with dominant density correlations .
this phase is the precursor of a quantum phase transition to a tc of pairs
phase tc2 of fig .
[ f.phd ] with the onset of long - range density order at wavevector @xmath32 .
we determine the extent of the tc phase via qmc by determining the mass imbalance at which @xmath33 starts diverging linearly with system size ; and by dmrg detecting the onset of the exponential decrease of @xmath34 and @xmath35 , marking the opening of a charge gap @xcite .
the tc instability is well understood coming from the falicov - kimball ( fk ) limit , @xmath36 . in this limit , which reduces to a 1d lattice gas in a static potential
, we find that the ground state corresponds to the tc of pairs for all values @xmath7 .
the gap @xmath37 to the formation of kink - antikink pairs in the tc is found to be a non - monotic function of @xmath38 , displaying an intermediate maximum .
we observe that , for small @xmath38 , the boundary of the crystalline region follows closely the locus at which the gap @xmath37 equals @xmath39 , and it has a re - entrant shape mimicking the non - monotonic behavior of the gap as a function of @xmath40 .
this suggests that the quantum melting transition corresponds to a condensation of kink - antikink pairs in the ground state .
, @xmath41 ) ; ( b ) @xmath42 coefficient ( dashed lines are linear fits ) ; all data are for @xmath43 ; ( c ) block entropy @xmath44 for chains with @xmath45 sites ( the curves for @xmath46 and @xmath47 are shifted to improve readability ) ; thick lines are a reference for @xmath48 ( black ) and @xmath49 conformal field theories ( red ) ; ( d ) fermionic spin - spin correlation functions , calculated via dmrg between the sites @xmath50 and @xmath51 of a chain with @xmath41 . the data for @xmath52 and @xmath53 have been rescaled to improve readability ( by a factor @xmath54 and @xmath55 , respectively).,width=287 ] [ f.2tll ] the repulsive side of the phase diagram is more complex . combining bosonization with renormalization group calculations up to two loops , refs .
conclude that a spin gap should open for any infinitesimal mass imbalance as @xmath56 for @xmath57 , where @xmath58 . on the contrary , ref .
, also based upon bosonization , concludes that the spin gap is absent in the repulsive case .
all our numerical findings point toward the persistence of a fully gapless tll behavior for both the charge _ and _ spin sector over a dominant portion of the repulsive phase diagram .
our conclusion is based on a number of crossed evidences .
first of all , we observe that the one body correlation functions @xmath59 and @xmath60 can be very well fitted with the simple power - law form @xmath61 , where @xmath62 is the conformal distance ( see fig . [ f.2tll](a ) ) .
in particular we find that , for weak and moderate repulsions , @xmath63 , which implies that the momentum distribution @xmath64 displays a quasi - condensation divergent peak at @xmath65 , to be detected in time - of - flight experiments ( see below ) .
moreover we exploit the fact that an explicit counting of the number of gapless degrees of freedom in the system comes from the central charge @xmath66 of the conformal field theory corresponding to our model of interest .
this quantity can be directly extracted via dmrg , using the fundamental result that the entanglement entropy ( ee ) of a boundary block of the system grows with the size @xmath67 of the block @xcite as @xmath68 + \mathcal{o}(1/l)+ { \rm const}$ ] , where @xmath69 is the reduced density matrix of the boundary block , and open boundary conditions are employed .
fig .
[ f.2tll](b ) ( see also ref .
@xcite ) shows that the scaling of the ee is fully consistent with @xmath48 in the repulsive regime , providing further evidence for the fact that the tll has two gapless components even for a significant mass imbalance .
finally , using qmc we gain further insight into the gapless phase by investigating the spin - spin correlation function @xmath70 ; where @xmath71 , and @xmath72 is the spin tll parameter ; in absence of a spin gap , @xmath73 .
we extract the luttinger exponent from the low-@xmath74 behavior of the spin structure factor , @xmath75 , giving the finite - size estimate @xmath76/2 $ ] . at the su(2 ) invariant point @xmath10 , @xmath77 is known to obey the kosterlitz - thouless ( kt ) critical scaling @xmath78^{-1}$ ] @xcite , which implies that the quantity @xmath79^{-1}$ ] scales linearly with @xmath80 .
quite remarkably , this scaling law remains valid over a broad range of @xmath11 values , and even for strong inter - species coupling , as shown in fig .
[ f.2tll](c ) .
this means that , even if the su(2 ) symmetry is broken in the hamiltonian , it appears to be restored in the ground - state of the system , namely that a moderate mass imbalance is an _ irrelevant _ perturbation and that , under a renormalization group treatment , the system flows to the su(2)-symmetric fixed point @xmath81 over a broad range of @xmath82 values .
this picture is further corroborated by comparing the longitudinal spin - spin correlations @xmath83 and the transverse ones in the _ fermionic case _
, @xmath84 , where @xmath85 ; the two correlation functions show the same algebraic decay over a broad range of @xmath82 values , and they even appear to coincide at short range , as shown in fig .
[ f.2tll](d ) . for @xmath86 and decreasing @xmath82 ; dashed lines are fits to @xmath87 ; ( b ) @xmath3-hole density correlator for nearest neighbors , @xmath88 ; the shaded areas mark the transition regions among the various phases ; ( c ) string structure factor as a function of system size ( same parameters as in ( a ) ) ; ( d ) decay exponent of the string correlation function as function of @xmath82.,width=287 ] [ f.sg ] on the other hand , a careful study of the behavior of the system shows that the two - component tll becomes indeed unstable to the formation of a spin gap ( sg ) for _ strong _ mass imbalance ( @xmath89 for @xmath90 ) .
the opening of a spin gap leaves a clear signature in the luttinger exponent @xmath91 , which stops flowing to the su(2 ) symmetric value : this is seen in the fact that @xmath42 begins to scale _ faster _ than linearly in @xmath80 , which is a manifestation of the fact that @xmath77 flows to values smaller than 1 , as shown in fig .
[ f.sg](a ) .
we use this criterion to locate the opening of the spin gap in fig .
[ f.phd ] .
moreover the breaking of su(2 ) symmetric scaling is also seen in the spin - spin correlation functions , which acquire a different behavior at short and long distances ( see fig . [ f.2tll](d ) ) .
the opening of a spin gap signals the formation of a bound state of some sort : but what could this bound state be in a purely repulsive system ?
we argue that a bound state can appear between the @xmath2 particles and the _ holes _ of the @xmath3 species , which are in a commensurate proportion @xmath92 to the @xmath2 particles . as shown in ref . , a sufficiently strong mass imbalance can indeed bind together particle composites in 1d quantum fluids - in the specific case of @xmath12 we have trimers formed by one @xmath2 particle and two @xmath3 holes ( @xmath93 ) . for strong enough repulsion ( @xmath94 ) the apparition of @xmath2-@xmath95 composites
is seen in the nearest - neighbor density correlations of the @xmath93 holes , @xmath96 , which is enhanced for increasing mass - imbalance , as seen in fig .
[ f.sg](b ) .
the @xmath2-@xmath95 composites have mutual repulsive interactions , as observed in ref .
, so that their binding favors configurations of the type @xmath97 in which the @xmath2 and @xmath3 particles alternate spatially .
such configurations , albeit lacking long - range density and magnetic order , contain strong non - local correlations that can be captured using a _ _ string correlation function__@xcite @xmath98 . as shown in fig .
[ f.sg](c-d ) , string correlations are significantly enhanced when entering the sg phase : indeed the string structure factor , @xmath99 , shows a more pronounced divergence with system size , which signals the slower decay of @xmath100 . as shown in ref .
, in presence of spin - charge separation the dominant decay of the string correlations is governed solely by the @xmath25 exponent , @xmath101 ; therefore @xmath31 controls the divergence of @xmath102 , and in fig .
[ f.sg](d ) we observe that the @xmath103 exponent ( extracted from the low-@xmath74 behavior of @xmath104 ) is significantly enhanced in the sg phase .
given that both the spin and charge structure factor exhibit a peak whose divergence with system size is hindered by the slow size - scaling of the divergence exponent @xmath105 @xcite , we find that the string correlations best characterize the fluid " magnetic order of the sg phase .
the bound trimers appearing in the sg phase undergo crystallization - phase tc1 in fig .
[ f.phd ] - for moderate repulsion and extreme mass imbalance .
the tc1 phase is analogous to the one discussed in ref .
( see also ref . ) .
this phase is marked by a suppression of @xmath106 which attains its absolute minimum @xmath107 for @xmath108 , due to the suppression of the weight of configurations with contiguous trimers .
the resulting non - monotonic behavior of @xmath106 as a function of @xmath82 , shown in fig .
[ f.sg](b ) ( see also ref .
@xcite ) , provides further evidence that the sg phase is a liquid of pre - formed trimers - the enhancement of @xmath106 in that phase is due to trimer binding , while the suppression in the tc1 phase is due to crystal ordering of the trimers .
our theoretical results have immediate consequences for current experiments on one - dimensional mixtures of mass - imbalanced cold atoms .
such experiments can probe both attractive and repulsive interactions , within the same experimental conditions , via the use of feshbach resonances @xcite .
our phase diagram reveals a fundamental asymmetry between the attractive and the repulsive case for weak and moderate mass imbalance , with the formation of bound pairs on the attractive side and the absence of spin gap on the repulsive one .
this asymmetry is very well seen in time of flight experiments probing the momentum distributions @xmath109 , @xmath110 , which are very broad on the attractive side , while they exhibit sharp quasi - condensation peaks on the repulsive side as shown in fig .
[ f.nk ] . this asymmetry can be used as strong evidence of pairing on the attractive side .
a partial symmetry is recovered only for strong mass imbalance , with the opening of a spin gap in the repulsive case , and the occurrence of a crystalline phase for both signs of the interaction .
this is also well captured by the momentum distributions , showing this time a suppression of the quasi - condensation peaks on the repulsive side due to the appearance of the sg phase and of the crystalline phase . finally the availability of high - resolution _ in - situ _ imaging @xcite sensitive to the spin @xcite allows to detect the formation of bound trimers and the enhancement of string correlations characterizing the sg phase .
we thank p. azaria , s. capponi , t. giamarchi , f. heidrich - meisner , and e. orignac for fruitful discussions , f. ortolani for help with the dmrg code , and the psmn ( ens - lyon ) for generous computer support .
cdeb is grateful to cnism unit of the physics department , university of bologna , where this work was started .
99 t. giamarchi , _ quantum physics in one dimension _ , clarendon press , oxford ( 2004 ) .
i. bloch _
et al .
_ , rev .
mod .
phys . * 80 * , 885 ( 2008 ) .
t. kinoshita _
et al . _ ,
science * 305 * , 1125 ( 2004 ) ; b. paredes _ et al .
_ , nature * 429 * , 277 ( 2004 ) ; e. haller _ et al .
_ , _ ibid .
_ * 466 * , 597 ( 2010 ) .
f. h. l. essler , h. frahm , f. ghmann , a. klmper , and v. e. korepin , _ the one - dimensional hubbard model _ , cambridge , 2005 .
m. a. cazalilla _
et al . _ ,
phys .
rev .
lett . * 95 * , 226402 ( 2005 ) .
l. mathey , phys .
rev .
b * 75 * , 144510 ( 2007 ) .
z. g. wang _
et al . _ ,
phys .
rev .
b * 75 * , 165111 ( 2007 ) . m. rizzi and a. imambekov , phys . rev .
a * 77 * , 023621 ( 2008 ) .
b. wang _
et al . _ ,
phys .
rev .
a * 79 * , 051604(r ) ( 2009 ) .
l. barbiero _
et al . _ ,
phys .
rev .
b * 81 * , 224512 ( 2010 ) . c. chin _ et al . _ ,
rev .
mod .
phys . * 82 * , 1225 ( 2010 ) , and references therein .
s.r.white , phys .
rev .
lett . * 69 * , 2863 ( 1992 ) .
t. roscilde , phys .
rev .
a * 77 * , 063605 ( 2008 ) .
o. f. syljusen , phys .
rev .
e * 67 * , 046701 ( 2003 ) .
see supplementary material .
p. calabrese and j. cardy , jstat p06002 ( 2004 ) .
e. burovski _
et al . _ , phys .
rev .
lett . * 103 * , 215301 ( 2009 ) .
t. keilmann _
et al . _ , phys .
rev .
lett . * 102 * , 255304 ( 2009 ) . m. den nijs and k. rommelse , phys .
rev .
b * 40 * 4709 ( 1989 ) .
w. s. bakr _
et al . _ ,
nature * 462 * , 74 ( 2009 ) . c. weitenberg _ et al .
_ , nature * 471 * , 319 ( 2011 ) .
|
the interplay of quantum phase coherence and repeated random scattering is at the origin of many effects in mesoscopic physics.@xcite these effects include weak localization and universal conductance fluctuations , both of which are small but fundamental corrections with respect to the conductance obtained from drude - boltzmann theory .
they culminate in anderson localization , the phenomenon that the resistance of a one or two - dimensional electron gas grows exponentially with system size if the system size is sufficiently large.@xcite originally , anderson localization and other mesoscopic effects were discovered in the context of disordered metals , in which electrons scatter off impurities with a size comparable to their wavelength .
theoretically , quantum effects in disordered metals are described using the ` disorder average ' which deals with an ensemble of macroscopically equivalent but microscopically different impurity configurations .
the presence of impurities is not essential for the existence of quantum effects , however .
the same effects , with the same statistical properties , have been found to appear if the electron motion is ballistic and chaotic , the only source of scattering of electrons being specular reflection off the sample boundaries.@xcite besides being of theoretical interest for understanding the quantum properties of systems with chaotic classical dynamics,@xcite the case of ballistic electron motion is relevant experimentally for very clean artificially structured two - dimensional electron gases in semiconductor heterostructures , such as quantum dots or antidot lattices.@xcite unlike disordered metals , in which impurities scatter diffractively , electrons in ballistic conductors have a well - defined classical dynamics .
many quantum properties of ballistic conductors with chaotic classical dynamics can be understood in terms of classical mechanics by making use of well - chosen semiclassical methods.@xcite a well - known example is the gutzwiller trace formula , which relates the density of states to properties of periodic orbits of the classical dynamics.@xcite for the conductance @xmath0 and its quantum corrections , there is a variant of gutzwiller s formula , which expresses @xmath0 as a double sum over classical trajectories @xmath1 and
@xmath2 connecting the source and drain contacts,@xcite @xmath3 upon entrance and exit , the trajectories @xmath1 and @xmath2 have the same transverse momenta @xmath4 and @xmath5 , respectively , but the position at which they enter or exit the sample may be different .
further , @xmath6 and @xmath7 are the so - called `` stability amplitude '' and the classical action of the trajectories . with eq .
( [ eq : gsemi ] ) as a starting point , the conductance , including its quantum corrections , can be calculated solely from knowledge of classical trajectories .
quantum effects have been linked to the existence of families of classical trajectories that only differ near ` small - angle encounters ' of the trajectory with itself.@xcite this semiclassical approach has been very successful explaining quantum effects in the ` perturbative regime ' in which quantum interference provides a small correction to the classical mechanics.@xcite weak localization and universal conductance fluctuations are examples of such perturbative quantum effects .
recently , heusler _ et al .
_ showed that it is also possible to understand quantum effects in terms of classical trajectories outside the perturbative regime .
they considered the spectral form factor @xmath8 of a ballistic cavity , the fourier transform of the two - point correlation function of the density of states . a direct semiclassical evaluation of @xmath8 using gutzwiller s trace formula was known to be possible in the perturbative regime @xmath9 only , where the heisenberg time @xmath10 , @xmath11 being the cavity s mean level spacing.@xcite motivated by the field - theoretical formulation of the problem , heusler _ et al .
_ used a different way to express @xmath8 in terms of periodic orbits that allowed them to calculate @xmath8 for all times.@xcite in this communication , we consider anderson localization for ballistic electron gases , and show that it , too , can be understood in terms of interference of classical trajectories , with eq .
( [ eq : gsemi ] ) as a starting point .
examples of systems that may exhibit such ` ballistic anderson localization ' are antidot lattices or arrays of chaotic cavities .
although it is generally accepted that anderson localization exists irrespective of the details of the microscopic electronic dynamics,@xcite the similarity of the phenomena for ballistic and disordered electron systems should not obscure the vastly different starting points of the theories for the two cases .
this difference not only pertains to the microscopic dynamics ( classical - deterministic vs. quantum - probabilistic ) , but also to the statistical assumptions of the theory . unlike theories of quantum transport in disordered metals , semiclassical theories of ballistic conductors
are intended to describe one specific system.@xcite fluctuations appear solely from variations of the fermi energy ; no changes in the classical dynamics are invoked . for disordered metals ,
anderson localization is most prominent for a quasi - one dimensional geometry , with sample length @xmath12 much larger than the sample width @xmath13 . for quasi - one dimensional samples , a full theory of transport in the localized regime
was developed by dorokhov@xcite and mello , pereyra , and kumar,@xcite using a stochastic approach , and by efetov and larkin,@xcite using a field - theoretic approach .
our theory of ballistic anderson localization closely follows these approaches .
the fact that the semiclassical theory for ballistic electrons follows the corresponding quantum mechanical theory for disordered metals is not special to the present case .
it is also typical of the trajectory - based semiclassical theories in the perturbative regime , which have a structure that resembles the diagrammatic perturbation theory of quantum corrections in disordered metals.@xcite in addition to the ballistic electron gases considered here , anderson localization also occurs in certain dynamic systems with a periodic time - dependent hamiltonian , such as the kicked rotor.@xcite classically , the dynamics of the kicked rotor is chaotic , with a momentum that changes diffusively under the influence of the periodic kicks .
quantum mechanically , it exhibits ` dynamic localization ' , anderson localization in momentum space rather than in real space.@xcite the phenomenology of dynamic localization is equal to that of anderson localization in disordered metals .
this is confirmed by extensive numerical simulations,@xcite as well as a field theoretical analysis of the problem.@xcite the trajectory - based theory of ballistic anderson localization that we report here is constructed for a specific system , an array of chaotic cavities .
a schematic drawing of such an array is shown in fig .
[ fig:1 ] .
we use this model system because the trajectory - based theory of transport through a single chaotic cavity is well established in the literature.@xcite arrays of cavities have been used as a starting point for a field - theoretic description of anderson localization using random matrix theory,@xcite but not with dorokhov s method . in sec.[sec : dis ] we show how dorokhov s theory can be adapted to this system if the cavities are disordered and random matrix theory can be used to describe transport through a single cavity . in sec .
[ sec:2 ] we summarize the basic elements of the trajectory - based semiclassical formalism .
this formalism is used to construct the trajectory - based semiclassical theory of ballistic localization in sec .
[ sec : semi ] . in section
[ sec : field_theory ] we approach the localization phenomenon from a different perspective .
we describe the array in terms of a nonlinear @xmath14-model whose perturbative ( ` diagrammatic ' ) evaluation generates structures of paired feynman amplitudes similar to those appearing in the native semiclassical approach .
alternatively , the dynamical structure of the theory can be analyzed to identify and successively eliminate hierarchies of different types of dynamical fluctuations in the system . in this way we obtain an effective low energy theory which turns out to be equivalent to the nonlinear @xmath14-model of diffusive quantum wires ;
the latter model is known to predict exponential localization at large length scales .
we conclude in sec .
[ sec : concl ] .
( color online ) schematic drawing of an array of @xmath15 chaotic cavities . the cavities are connected via ballistic contacts with dimensionless conductance @xmath16 . the semiclassical theory requires the limit @xmath17 at a fixed ratio @xmath18 . ]
we now describe how dorokhov s approach can be used to describe anderson localization in an array of disordered cavities or quantum dots .
we take the disorder in each cavity to be weak ( cavity size much smaller than the localization length ) , so that transport through a single cavity is described by random matrix theory.@xcite a schematic drawing of the array of cavities is shown in fig.[fig:1 ] .
the cavities are connected via ballistic contacts with dimensionless conductance @xmath16 . since we want to compare with a semiclassical theory for an array of ballistic cavities , we require @xmath19 .
localization takes place if the conductance of the array is of order unity .
this condition is met if the number @xmath15 of cavities in the array is comparable to @xmath16 .
hence , in the calculations below we take the limit @xmath20 while keeping the ratio @xmath18 fixed .
the same limit is taken in the field theoretical description of localization,@xcite where it is known as the `` thick wire limit '' .
the transport properties of the array of cavities are encoded in its scattering matrix @xmath21 .
( the superscript `` q '' is used to distinguish the quantum mechanical scattering matrix from its semiclassical counterpart to be introduced in sec .
[ sec:2 ] . )
the matrix indices of @xmath21 represent the two contacts @xmath22 at the far left and right of the array and the transverse modes in each contact , @xmath23 , @xmath24 , where @xmath25 is the number of channels in contact @xmath26 , @xmath22 .
the matrix @xmath21 is a random quantity because it depends on the fermi energy and on the precise disorder configuration in each cavity .
following the approach of dorokhov , mello , pereyra , and kumar,@xcite we consider the hermitian matrix @xmath27 and calculate its statistical distribution by expressing @xmath28 in terms of @xmath29 and proceeding recursively .
the matrix @xmath30 is related to the dimensionless conductance @xmath31 of the array through the landauer formula , @xmath32 taking the scattering matrix of each individual cavity from the circular ensemble of random matrix
theory,@xcite one then finds that the recursion relation for @xmath30 takes the form @xmath33 where the hermitian matrix @xmath34 is a random ( noise ) term with a gaussian distribution , @xmath35 where @xmath36 , @xmath37 or @xmath38 in the presence or absence of time - reversal symmetry , respectively , and @xmath39 the averaging brackets @xmath40 denote an average with respect to the disorder configuration in the @xmath15th cavity only . in the limit @xmath20 while keeping @xmath18 fixed , the stochastic recursion relation ( [ eq : ttdiff ] ) can be mapped to the dorokhov - mello - pereyra - kumar ( dmpk ) equation , which is a stochastic differential equation for the eigenvalues of @xmath30.@xcite the solution of the dmpk equation is known,@xcite which completes the theory of localization for an array of disordered cavities .
alternatively , the stochastic recursion relation ( [ eq : ttdiff ] ) can be used to generate a coupled set of recursion relations for the disorder averages of the moments of @xmath41 , @xmath42 with @xmath43 ( the argument @xmath44 is suppressed on the right hand side of the second equality . ) the average in eq .
( [ eq : tgeneral ] ) is the full disorder average , applied to all cavities in the array . taking the limit @xmath20 at fixed @xmath18 , eq .
( [ eq : tgeneral ] ) is mapped to a coupled set of differential equations for the moments of @xmath45 which is identical to the corresponding set of differential equations for a disordered wire.@xcite a subset of these equations can be resummed into a partial differential equation for the generating function @xmath46 , with@xcite @xmath47 as shown in refs .
, the resulting theory of localization in quasi - one dimension is formally equivalent to that obtained from the one - dimensional nonlinear sigma model.@xcite
the central object in the trajectory - based semiclassical theory of localization in an array of ballistic chaotic cavities is a semiclassical representation of the scattering matrix @xmath48 . in the semiclassical representation ,
the discrete transverse momenta become continuous variables , so that the scattering matrix @xmath48 becomes a ` scattering kernel ' @xmath49 . following standard semiclassical approximations , this scattering kernel
is then represented as a sum over classical trajectories @xmath1 connecting contact @xmath50 to contact @xmath26,@xcite @xmath51 such that the transverse momentum of @xmath1 upon entrance and exit equals @xmath52 and @xmath53 , respectively .
further , @xmath54 is the classical action of @xmath1 , and @xmath55 the stability amplitude , @xmath56 where @xmath5 is the transverse position upon entrance into the sample .
maslov indices and other additional phase shifts are included in @xmath57 .
because the transverse modes in the quantum mechanical formulation are linked to the absolute value of the transverse momentum , not to the transverse momentum itself , the semiclassical counterpart of the products @xmath58 and @xmath59 consist of two contributions : one in which transverse momenta in contact @xmath50 are equal and one in which the transverse momenta are opposite , @xmath60({p'},{p } ) & = & \sum_{\pm } \int_{-p_f}^{p_f } \frac{dp''}{2 \pi \hbar } s_{ij}^{\vphantom{m}}({p'},p '' ) s_{kj}^{\dagger}({p } , \pm p '' ) , \nonumber\end{aligned}\ ] ] @xmath61 ( { p}',{p } ) & = & \sum_{\pm } \int_{-p_f}^{p_f } \frac{dp''}{2 \pi \hbar } s_{jk}^{\dagger}(\pm p'',{p } ' ) s_{ji}^{\vphantom{m}}(p'',{p } ) , \nonumber \\ \label{eq : sproduct}\end{aligned}\ ] ] together , eqs .
( [ eq : ssemi ] ) and ( [ eq : sproduct ] ) specify how products of the quantum - mechanical scattering matrix and its hermitian conjugate are expressed in terms of classical trajectories .
[ the `` @xmath62 '' terms in the summations were omitted from the semiclassical expression for the conductance , eq .
( [ eq : gsemi ] ) above . ] for a theory of anderson localization , we are interested in the trace of a product of alternating factors @xmath63 and @xmath64 or in the product of such traces . using the semiclassical representation ( [ eq : ssemi ] ) , a polynomial function @xmath65 that involves the alternating product of @xmath15 factors @xmath63 and @xmath15 factors @xmath64 is written as a summation over @xmath66 classical trajectories @xmath67 and @xmath68 , one trajectory for each factor @xmath63 or @xmath64 , respectively . each configuration of classical trajectories is weighed by a phase factor @xmath69 with @xmath70 building on work by richter and sieber,@xcite haake and coworkers have identified a hierarchy of families of classical trajectories @xmath71, ,@xmath72 that contribute to the average @xmath73,@xcite where the average is taken with respect to variations of the fermi energy while keeping the classical dynamics ( i.e. , the shape of the cavities ) fixed .
their identification is based on the recognition that families of trajectories contribute to @xmath74 only if their total action difference @xmath75 is of order @xmath76 systematically , which happens only if the trajectories @xmath77 are piecewise and pairwise identical to the trajectories @xmath78 , @xmath79 , up to classical phase space distances of order @xmath80.@xcite trajectories that are separated by larger phase space distances have typical action differences @xmath75 that are parametrically larger than @xmath76 , so that their contribution vanishes upon taking the average .
the simplest choice for a family of trajectories for which @xmath81 is of order @xmath76 systematically is if each trajectory @xmath77 equals another trajectory @xmath78 for the full length of the trajectory . calculating @xmath74 from the contribution from such families of trajectories
only is known as the `` diagonal approximation''.@xcite nontrivial families of trajectories emerge from the possibility of small - angle encounters between trajectories , at which more than two trajectories are within a phase space distance @xmath82.@xcite at such small - angle encounters , the pairing between the @xmath77 and @xmath78 can be changed so that now trajectories need to be piecewise identical only .
the duration of a small - angle encounter is the `` ehrenfest time '' @xmath83 , where @xmath84 is the lyapunov exponent of the classical dynamics , @xmath85 the fermi momentum , and @xmath86 a characteristic length scale of the classical dynamics .
the fundamental action integrals corresponding to each small - angle encounter are known , @xcite and the resulting theory takes the form of simple combinatorial rules with which any product of traces of products the scattering matrix and its hermitian conjugate can be calculated to arbitrary order in @xmath76 from the semiclassical representation of @xmath63 , provided that the ehrenfest time @xmath87 be much smaller than the sample s mean dwell time @xmath88.@xcite ( the case of finite @xmath89 is considerably more complicated , see , _
e.g. _ , refs . , but not relevant for a semiclassical theory of anderson localization . ) in the remainder of this text , we refer to a calculation of the energy - average @xmath74 using contributions from families of trajectories thus constructed as the `` trajectory - based semiclassical formalism '' .
although there is no formal proof that this formalism is exact , _
i.e. _ , that there are no other contributions to @xmath74 than from families of piecewise paired classical trajectories , the formalism satisfies all known conservation rules and calculations based on trajectory - based semiclassics have been found to agree with fully quantum mechanical calculations whenever applicable.@xcite the present calculation can be viewed as another demonstration of the validity of trajectory - based semiclassics , by showing that the same formalism can serve as the starting point of a theory of localization .
while we do not need the detailed results of the trajectory - based semiclassical formalism , there are two properties of ensemble averages calculated using that formalism that are particularly relevant for our calculations below : + ( i ) all averages are compatible with the condition of unitarity,@xcite @xmath90(p , p ' ) & = & \sum_{j } [ s^{\dagger}_{jk } s_{ji}^{\vphantom{m}}](p , p ' ) \nonumber \\ & = & \delta_{ik } \delta(p - p').\end{aligned}\ ] ] ( ii ) for a product @xmath91 or @xmath92 , the trajectory @xmath1 of the semiclassical representation for @xmath63 and the trajectory @xmath2 of the semiclassical representation of @xmath64 at contact @xmath50 satisfy @xmath93 in particular , this implies that there is no contribution from the second term in eq .
( [ eq : sproduct ] ) for a product of two scattering kernels.@xcite +
in the perturbative regime , the semiclassical theories of quantum corrections to transport and to the density of states closely followed the corresponding theory for disordered metals . the semiclassical formalism described in the previous section was instrumental in formalizing the relation between the two types of theories . motivated by this correspondence , we now look for the possibility to adapt the theory of localization in an array of disordered cavities [ sec .
[ sec : dis ] ] to the case of an array of ballistic cavities .
thus , paraphrasing the arguments of sec . [
sec : dis ] , the goal of our calculation will be to find the full probability distribution of the function @xmath94({p'},{p})\ ] ] for an array of @xmath15 cavities . as shown in sec .
[ sec : dis ] , there are two ways in which this can be accomplished : 1 . using a stochastic approach , in which one considers the stochastic evolution of the function @xmath95 as a function of @xmath15 , or through 2 . the construction of a set of recursion relations for all moments of @xmath41 . in both cases ,
the resulting theory is formally identical to the known theories of localization in quasi - one dimension .
although technically simpler , the stochastic approach is at odds with the goals of a semiclassical theory for ballistic localization : the goal of a theory of localization in an array of ballistic cavities is to describe an array of cavities with a fixed shape , using variations of the fermi energy as the only source of statistical fluctuations .
since the fermi energy is set globally , for all cavities at the same time , quantum corrections for different cavities are not independent , and a stochastic approach is ruled out a priori .
a stochastic approach is possible , however , if one relaxes the goals of the theory , allowing for small variations of the shape of each cavity , or for variations of a `` gate voltage '' that sets the fermi energy of each cavity independently .
below we first describe the stochastic approach . in sec.[sec : b2 ] we consider the case of broken time - reversal symmetry , which is technically simpler .
the discussion of localization in the presence of time - reversal symmetry is given in sec .
[ sec : time ] . in sec .
[ sec : hierarchy ] we discuss how a hierarchy of recursion relations for the moments of @xmath96 can be constructed , where the average is taken with respect to variations of the fermi energy of the entire array only . the stochastic approach deals with ( statistical ) properties of the function @xmath41 before averaging .
although the properties ( i ) and ( ii ) of the trajectory - based semiclassical formalism [ sec . [ sec:2 ] ] are satisfied for the average of _ any _ product of traces of products alternating factors @xmath63 and @xmath64 to arbitrary order in @xmath76 , they have not been shown to follow from the semiclassical scattering matrix ( [ eq : ssemi ] ) _ before _ averaging . however , since our goal is a statistical theory of the transport the final statements of the theory will refer to averaged quantities only we will accept these two properties on the level of the sample - specific semiclassical scattering kernel @xmath97 in the arguments that follow below . starting from eq .
( [ eq : ssemi ] ) , the kernel @xmath96 is expressed as a double sum over classical trajectories @xmath1 and @xmath2 that connect the entrance and exit contacts [ fig .
[ fig:2 ] ] , @xmath98 here @xmath4 and @xmath5 are the transverse momenta of @xmath1 and @xmath2 upon entrance .
the two trajectories have equal transverse momenta upon exit , and exit at positions a quantum uncertainty @xmath99 apart see property ( ii ) above .
below , we express the difference @xmath100 in terms of classical trajectories . we first calculate the average @xmath101 , where the average is taken with respect to variations of the fermi energy of the @xmath15th cavity ( or of its shape ) , while keeping the fermi energy and shape of the other cavities fixed . after that , we calculate the variance of @xmath102 and the higher cumulants .
( color online ) example of a pair of trajectories @xmath1 and @xmath2 contributing to @xmath103({p'},{p})$ ] . upon entering the sample , the two trajectories have different transverse momenta @xmath4 and @xmath5 , and may enter at different transverse positions . upon exiting the sample , @xmath1 and @xmath2 have the same transverse momentum , and exit at transverse positions a distance @xmath104 apart . ] _ average of @xmath102 . _
when calculating the average @xmath105 to leading order in @xmath106 in the absence of time - reversal symmetry , it will be sufficient to calculate @xmath107 in the diagonal approximation by considering trajectories @xmath1 and @xmath2 that are `` paired '' in the @xmath15th cavity .
note , however , that trajectories need not be paired in the first @xmath44 cavities , because no average is taken there .
each trajectory is classified by the number of times @xmath108 , @xmath109 that it enters the @xmath15th cavity from the @xmath110th cavity .
hence , for each trajectory there are @xmath108 and @xmath109 segments in the @xmath15th cavity , which we label as @xmath71 , , @xmath111 and @xmath112 ,
, @xmath113 . since trajectories are paired in the @xmath15th cavity , @xmath114 .
examples of trajectory pairs @xmath1 and @xmath2 with @xmath115 , @xmath38 , and @xmath116 are shown in fig .
[ fig:3 ] .
since trajectories need to be paired upon exit , @xmath117 has to be paired with @xmath118 .
while there are @xmath119 ways in which the remaining segments can be paired , we now show that the `` diagonal pairing '' , @xmath120 paired with @xmath78 , @xmath121 , gives the main contribution to @xmath101 , whereas all other pairings give contributions a factor @xmath106 smaller .
cutting the trajectories @xmath1 and @xmath2 at the contact between the @xmath110st and @xmath15th cavity also separates the part of each trajectory that resides in the first @xmath44 cavities into @xmath122 segments .
the first segment of each trajectory connects the entrance contact to the exit of the @xmath110st cavity ; all other segments connect the exit contact of the @xmath110st cavity to itself .
since the electron dynamics in the @xmath15th cavity is fully ergodic , the positions and transverse momenta with which these segments cross the interface between the @xmath110st and @xmath15th cavity are fully random , without correlations between different segments .
[ correlations can be ruled out down to quantum phase space distances @xmath123 because the dwell time @xmath124 . ] hence , these @xmath122 segments can be interpreted as the semiclassical representation of a product of @xmath125 , @xmath126 , and @xmath127 factors @xmath128 and @xmath129 , _ before _ taking any average .
note that , although trajectories @xmath130 that are `` paired '' in the @xmath15th cavity have a phase space distance of order @xmath131 or smaller when they enter / exit the @xmath15th cavity from / to the @xmath110st cavity , such trajectory pairs are sufficient for the semiclassical calculation of the complete kernels @xmath96 because of property ( ii ) of sec .
[ sec:2 ] .
( color online ) examples of trajectories @xmath1 and @xmath2 contributing to @xmath132 , where an average is taken over the fermi energy of the @xmath15th cavity only .
the four panels show trajectories contributing for @xmath115 ( top left ) , @xmath133 ( top right ) , @xmath134 with diagonal pairing ( bottom left ) , and @xmath134 with non - diagonal pairing ( bottom right ) .
the number @xmath122 counts the number of times @xmath1 and @xmath2 enter the @xmath15th cavity from the @xmath110st cavity .
the three dashed lines in each panel represent the entrance contact , the contact between the @xmath110st cavity and the @xmath15th cavity , and the exit contact ( from left to right ) . in the @xmath15th cavity , the trajectories @xmath1 and @xmath2 are piecewise equal up to quantum uncertainties . ] for the diagonal pairing , the @xmath122 segments of @xmath1 and @xmath2 that reside in the first @xmath44 cavities generate a particularly simple product of factors @xmath63 and @xmath64 , @xmath135^{m-1}$ ] , where @xmath136(p , p')\ ] ] is the reflection coefficient for the first @xmath44 cavities , seen from the exit contact of the @xmath110st cavity , and @xmath137 using the trajectory - based semiclassical formalism to evaluate the diagonal trajectory sums in the @xmath15th cavity,@xcite one then easily finds that the diagonal pairing gives @xmath138^{m-1 } \nonumber \\ & = & \frac{{\cal t}(n-1 ) g_{\rm c}}{2 g_{\rm
c } - \mbox{tr}\ , { \cal r}(n-1)}. \label{eq : x1p}\end{aligned}\ ] ] using unitarity , one has @xmath139 so that eq .
( [ eq : x1p ] ) can be rewritten as @xmath140 up to corrections of order @xmath141 .
this is precisely the semiclassical equivalent of the recursion relation of eq .
( [ eq : ttdiff ] ) , averaged over the fermi energy or shape of the @xmath15th cavity .
it remains to show that non - diagonal pairings , in which a segment @xmath120 is not paired with @xmath78 give a contribution to @xmath101 that can be neglected in the limit of large @xmath16 .
hereto we first consider the case @xmath134 , for which the only possible non - diagonal pairing is @xmath142 , @xmath143 [ fig .
[ fig:3 ] , bottom right ] .
for this pairing , the three segments of the trajectories that reside in the first @xmath44 cavities generate the semiclassical representation of a product of six scattering matrices , @xmath144 . from unitarity
, one has @xmath145 for comparison , the diagonal pairing for @xmath134 generates @xmath146 ^ 2 $ ] , which is a factor @xmath147 larger because @xmath148 . using the ergodic dynamics in the @xmath15th cavity
, the non - diagonal pairing of segments in the @xmath15th cavity for @xmath134 gives a contribution to @xmath149 equal to @xmath150.\end{aligned}\ ] ] this is a factor @xmath151 smaller than the leading contribution ( [ eq : ttdiffavg ] ) to @xmath101 .
the same arguments can be used for @xmath152 : non - diagonal pairings come at the expense of at least two factors @xmath153 and , hence , lead to contributions to @xmath101 that are at least a factor @xmath151 smaller than the contribution from diagonal pairing .
these arguments can also be used to show that contributions to @xmath107 that involve small - angle intersections of the trajectories in the @xmath15th cavity are a factor @xmath106 smaller than the leading contribution considered above .
_ fluctuations of @xmath102 . _
the fluctuations of @xmath102 are described by the covariance @xmath154 .
we calculate @xmath155 from the identity @xmath156 where we used the shorthand notation @xmath157 , @xmath158 , and @xmath159 . the average @xmath101 is given by eq .
( [ eq : ttdiffavg ] ) above , so it remains to calculate @xmath160 .
the product @xmath161 is represented as a sum over four classical trajectories , @xmath1 , @xmath2 , @xmath162 , and @xmath163 . as
before , we introduce the numbers @xmath108 , @xmath109 , @xmath164 and @xmath165 that indicate how often each trajectory enters the @xmath15th cavity .
since trajectories always enter or exit the @xmath15th cavity in pairs , one has @xmath166 .
unlike the average @xmath101 , for which the only contribution came from the diagonal approximation in the @xmath15th cavity with diagonal pairing of the segments @xmath120 and @xmath78 , the average of the second moment @xmath167 has contributions from both non - diagonal pairings in the diagonal approximation and from trajectories beyond the diagonal approximation , which have small - angle encounters in the @xmath15th cavity .
we first consider the diagonal approximation , for which each segment @xmath77 or @xmath168 is paired with another segment @xmath78 or @xmath169 . for the diagonal pairing of segments , @xmath170 , @xmath171 and @xmath172 , @xmath173 , we find @xmath174 this contribution to @xmath175 precisely cancels the second and third terms in eq.([eq : dtt ] )
. hence @xmath176 must be from non - diagonal pairings of the trajectory segments within the diagonal approximation , or from trajectory configurations beyond the diagonal approximation in the @xmath15th cavity .
since the latter class of trajectories have small - angle encounters , we write @xmath177 ( color online ) examples of four trajectories @xmath1 , @xmath2 , @xmath162 , and @xmath163 contributing to @xmath178 for an array of chaotic cavities , where the average is taken with respect to the fermi energy of the @xmath15th cavity only .
the examples shown here have a non - diagonal pairing with @xmath179 , @xmath180 ( left panel ) and @xmath181 ( right panel ) . in the @xmath15th cavity , the trajectories @xmath1 and @xmath162 are piecewise equal to @xmath2 and @xmath163 , up to quantum uncertainties . ]
the leading non - diagonal pairing appears if one pairs the first @xmath182 segments of @xmath1 with the first @xmath182 segments of @xmath163 , and the last @xmath183 segments of @xmath1 with the last @xmath183 segments of @xmath2 , as well as the first @xmath184 segments of @xmath162 with the first @xmath184 segments of @xmath2 , and the last @xmath185 segments of @xmath162 with the last @xmath185 segments of @xmath163 , where @xmath186 and @xmath187 are integers .
two examples , with @xmath188 , @xmath189 , @xmath190 , and @xmath191 , and with @xmath192 and @xmath193 , are shown in fig .
[ fig:4 ] .
one then finds @xmath194 + { \cal o}(1/g_{\rm c}^2 ) , \nonumber\end{aligned}\ ] ] where @xmath195 and @xmath196 , with @xmath197 other pairings give contributions of order @xmath141 or smaller .
( color online ) two examples of a set of four trajectories contributing to @xmath198 that have a small - angle encounter in the @xmath15th cavity . for the left panel
the encounter is in the interior of the @xmath15th cavity ; for the right panel it touches the exit contact . ] the second contribution to the fluctuations of @xmath102 comes from trajectories with a small - angle encounter in the @xmath15th cavity . since we only need fluctuations of @xmath102 to leading order in @xmath106 , it is sufficient to consider trajectories with one small - angle encounter only .
taking the small - angle encounter to be between the segments @xmath199 , @xmath200 , @xmath201 , and @xmath202 , with @xmath203 , and pairing the first @xmath204 segments of @xmath1 with the first @xmath204 segments of @xmath163 , and the last @xmath205 segments of @xmath1 with the last @xmath205 segments of @xmath2 , as well as the first @xmath206 segments of @xmath162 with the first @xmath206 segments of @xmath2 , and the last @xmath207 segments of @xmath162 with the last @xmath207 segments of @xmath163 , we find @xmath208 where the first line comes from encounters in the interior of the @xmath15th cavity and the second line comes from encounters that touch the exit contact.@xcite examples of the two terms for @xmath209 are shown in fig.[fig:5 ] .
[ we do not need to consider encounters that touch the contact between the @xmath110st cavity and the @xmath15th cavity , because the contribution from such encounters is included in the ( products ) of the kernel @xmath96 of the first @xmath44 cavities . ] combining everything we have @xmath210 \nonumber \\ & & \mbox { } + { \cal o}(1/g_{\rm c}^2 ) .
\label{eq : ddtresult}\end{aligned}\ ] ] equation ( [ eq : ddtresult ] ) is the semiclassical equivalent of eq .
( [ eq : caltstat ] ) .
higher cumulants of @xmath102 can be calculated in the same way . for the @xmath211th cumulant
, one finds that only pairings that involve trajectories out of all @xmath211 factors @xmath96 contribute .
each additional factor @xmath96 involved in the pairing scheme contributes an additional factor @xmath106 , which is why all cumulants with @xmath212 are of sub - leading order in @xmath106 .
together , eqs .
( [ eq : ttdiffavg ] ) and ( [ eq : ddtresult ] ) form the semiclassical equivalent of the stochastic recursion relation ( [ eq : ttdiff ] ) used in the fully quantum mechanical theory of localization in disordered quasi - one dimensional wires . in the limit @xmath20 at fixed @xmath18 , eqs.([eq : ttdiffavg ] ) and ( [ eq : ddtresult ] ) can be mapped to stochastic differential equation for the eigenvalues of @xmath96 , which will be formally equivalent to the dmpk equation .
the dmpk equation , in turn , provides a full description of localization in quasi - one dimensional wires.@xcite in the presence of time - reversal symmetry both the average @xmath213 and the covariance @xmath214 are different . in both cases
the difference appears because one can pair time - reversed trajectories when taking the ensemble average in the @xmath15th cavity .
there are two additional contributions to the average @xmath101 .
the first of these arises from the diagonal approximation for the average in the @xmath15th cavity .
as before , we define the number @xmath215 as the number of times the trajectories @xmath1 and @xmath2 enter the @xmath15th cavity from the @xmath110st cavity .
the first additional contribution to @xmath101 then involves the pairing of segments @xmath216 with the time - reversed of @xmath217 , @xmath218 , and @xmath219 , @xmath220 examples for trajectory pairs contributing to the first and second line in eq .
( [ eq : tt1 ] ) are shown in fig .
[ fig:6 ] .
the second contribution comes from small - angle encounters inside @xmath15th cavity involving the segments @xmath221 , @xmath222 , @xmath223 , and @xmath224 , where @xmath225 . for this contribution
one finds @xmath226 examples of trajectory pairs contributing to @xmath227 are shown in fig .
[ fig:7 ] . using @xmath228 and adding these two contributions to eq .
( [ eq : ttdiffavg ] ) one finds @xmath229 up to corrections of order @xmath141 .
( color online ) two examples of trajectory pairs contributing to @xmath101 in the presence of time - reversal symmetry . ] for the fluctuations of @xmath102 in the presence of time - reversal symmetry one finds an extra contribution from pairing @xmath120 with the time - reversed of @xmath230 , @xmath231 , or @xmath232 with the time - reversed of @xmath233 , while following diagonal pairing rules for all other segments . proceeding as before , we find @xmath234 \nonumber \\ & & \mbox { } + { \cal o}(1/g_{\rm c}^2 ) .
\label{eq : ddtresultb1}\end{aligned}\ ] ] where @xmath235(p , p').\end{aligned}\ ] ] the stochastic process defined by eqs .
( [ eq : ttdiffavgb1 ] ) and ( [ eq : ddtresultb1 ] ) precisely mirrors the stochastic process ( [ eq : ttdiff ] ) for the quantum mechanical matrix @xmath236 .
( color online ) two examples of trajectory pairs contributing to @xmath237 in the presence of time - reversal symmetry .
the trajectory pairs have a small - angle encounter in the @xmath15th cavity . ] the direct construction of recursion relations for the moments of @xmath96 is an alternative to the stochastic approach that avoids extending the use of the properties ( i ) and ( ii ) of sec .
[ sec:2 ] to sample - specific quantities and the necessity to define a statistical ensemble by varying the fermi energy or shape of each cavity individually . the construction of recursion relations for moments of @xmath96 proceeds in the very same manner as the construction of the stochastic recursion relations for @xmath96 , with the additional requirement that trajectories are piecewise paired in all @xmath15 cavities , not only in the @xmath15th cavity .
since the arguments of the preceding section did not rely on the structure of the trajectories in the first @xmath44 cavities , one immediately concludes that the recursion relations for the moments of @xmath96 derived this way are identical to the recursion relations for the moments of @xmath96 one obtains from the stochastic approach .
starting from the stochastic recursion relations ( [ eq : ttdiffavg ] ) and ( [ eq : ddtresult ] ) or ( [ eq : ttdiffavgb1 ] ) and ( [ eq : ddtresultb1 ] ) one arrives at the same hierarchy of recursion relations ( [ eq : tgeneral ] ) derived for an array of disordered cavities .
( color online ) examples of trajectory pairs contributing to the full average @xmath238 in the absence of time - reversal symmetry .
the trajectory pairs have @xmath115 ( top ) , @xmath133 ( center ) , and @xmath134 ( bottom ) .
the left panel shows trajectory pairs without small - angle self intersections ; the right panel shows trajectory pairs with two small - angle self intersections in the first @xmath44 cavities . ]
we illustrate this procedure for the recursion relation for the first moment @xmath239 in the absence of time - reversal symmetry . following the rules of the semiclassical formalism , the average @xmath240
is determined by trajectory pairs @xmath1 , @xmath2 that are piecewise paired throughout the entire array of cavities .
the trajectories can have small - angle self encounters , at which the pairing between @xmath1 and @xmath2 can be changed .
each pair of trajectories is classified by the number @xmath122 of times the trajectories enter the @xmath15th cavity from the @xmath110st cavity .
the @xmath122 segments of each trajectory in the @xmath15th cavity are labeled @xmath241 and @xmath242 . as in sec .
[ sec : b2 ] , it will be sufficient to consider pairs of trajectories in which each segment @xmath120 is paired with the corresponding segment @xmath78 , @xmath121 .
trajectory pairs with self encounters in the @xmath15th cavity or with non - diagonal pairings in the @xmath15th cavity give contributions to @xmath243 that are a factor @xmath106 smaller than the leading contribution . for the diagonal pairing , the remaining @xmath122 segments of the trajectories @xmath1 and @xmath2 that reside in the first @xmath44 cavities precisely generate @xmath244^{m-1 } \rangle$ ] , where the averaging brackets refer to variations of the fermi energy for the entire array of cavities .
hence , we find @xmath245^{m-1 } \right\rangle}{2^m g_{\rm c}^{m-1 } } \nonumber \\ & = & \left\langle \frac{g_{\rm c } \mbox{tr}\ , { \cal t}(n-1 ) } { 2 g_{\rm c } - \mbox{tr}\ , { \cal r}(n-1 ) } \right\rangle .
\label{eq : tavgdiff}\end{aligned}\ ] ] a schematic representation of trajectory pairs contributing to eq.([eq : tavgdiff ] ) with up to two self - encounters in the first @xmath44 cavities are shown in fig .
[ fig : avg ] . using unitarity to express @xmath246 in terms of @xmath96 and subtracting @xmath247 ,
one then finds @xmath248 this is the same equation as one obtains from taking the average of the increment @xmath249 in the stochastic approach .
in this section , we will approach the localization phenomenon from a different perspective .
we will use that the quantum dot array depicted in fig .
[ fig:1 ] supports hierarchies of different types of field - fluctuations in a field - theoretic description .
these fluctuations reflect the fate of density distributions in classical phase space under the dynamical evolution of the system .
each of these fluctuations , thus , comes with a characteristic ` relaxation time ' , _ i.e. , _ a time scale on which the fluctuation decays .
( for example , fluctuations inhomogeneous in the sector of phase space describing an individual quantum dot will decay on a time scale comparable to the time of flight through the dot , etc . ) in the description of low energy phenomena such as the zero frequency ( dc ) conductance , modes operating at short time scales can be treated perturbatively .
their feedback into the sector of long time scales then stabilizes a ` low energy theory ' . in the following
, we will derive a theory that is minimal in that it contains information equivalent to that stored in the fokker - planck equation of localization .
the strategy pursued here parallels one applied previously @xcite to the problem of dynamical localization in the quantum kicked rotor ( also known as the `` standard map '' . )
one difference is that the spectrum of different modes encountered in the present problem happens to be more complex .
our logics also resembles that of ref . , where it had been shown that the relevant low energy theory of an _ ergodic _ quantum system contains the information otherwise stored in random matrix theory .
technically , our discussion will be based on a formulation of the array in terms of the ballistic nonlinear @xmath14-model.@xcite a quadratic approximation in energetically high lying modes generates an effective low energy theory wherein each quantum dot is treated as a structureless ( ` ergodic ' ) entity .
this theory will be equivalent to the celebrated nonlinear @xmath14-model of disordered quantum wires,@xcite a model that predicts anderson localization on large length scales .
we will see that the parameters stabilizing the hierarchical mode integration are the same as those utilized in previous sections of this paper .
conceptually , the hierarchical scheme is an alternative to an indiscriminate perturbative integration over all modes in one go . that latter scheme would be essentially equivalent ( see ref .
@xcite for a discussion in the context of spectral statistics . ) to a semiclassical expansion in terms of paired trajectories . in this sense
, the hierarchical mode integration processes the information stored in the statistics of trajectories by different means .
our starting point will be the description of the quantum dot array in terms of the supersymmetric ballistic nonlinear @xmath14-model .
this theory is obtained by averaging exact functional representations of green functions over an energy interval of width @xmath250 centered around the uniform fermi energy @xmath251 of the array .
a subsequent saddle point approximation ( stabilized in the parameter @xmath252 ) then obtains a field theory in classical phase defined by @xmath253),\end{aligned}\ ] ] with @xmath254 & = & \frac { \beta\pi \hbar \nu}{2 } \int_\gamma ( d{{\bf x } } ) \,{\rm tr}\big(t\ast \lambda \ { h , t^{-1}\}\big ) \nonumber \\ & & \mbox { } + s_{\rm reg}[t].\end{aligned}\ ] ] the integration variables in ( [ eq:5 ] ) , @xmath255 are ( super ) matrix valued fields defined on shells @xmath256 of constant energy in classical phase space .
further , @xmath257 where @xmath258 and @xmath259 are coordinates and momenta , respectively , @xmath260 is the hamiltonian function of the system , the integral over the energy - shell is normalized to the ( spatial ) volume of the system , @xmath261 , and @xmath262 is the single particle density of states per volume , @xmath263 . for time reversal and spin rotation invariant systems ( orthogonal symmetry class , @xmath37 ) , the ` internal ' structure of the matrices @xmath264
is described by a composite index @xmath265 , where @xmath266 discriminates between the advanced and retarded sector of the theory , @xmath267 discriminates between commuting and anticommuting sectors , and @xmath268 accounts for the operation of time reversal .
time reversal symmetry reflects in the relation @xmath269 , where @xmath270 is a fixed matrix whose detailed structure will not be of concern throughout . for time reversal non - invariant systems ( unitary symmetry , @xmath271 ) no time - reversal structure exists and @xmath272 . in either case , the matrices @xmath273 carry a coset space structure in the sense that configurations @xmath273 and @xmath274 are to be identified if @xmath275=0 $ ] , where @xmath276 and `` ar '' stands for action in advanced / retarded space . finally , the regulatory action @xmath277\equiv
\delta \int_\gamma ( d{{\bf x}})\ , { \rm str}\big ( \lambda t^{-1 } \ast \lambda t\big),\ \ \delta \downarrow 0\ ] ] determines the ` causality ' of field fluctuations , but will otherwise not be of much relevance throughout .
the fluctuation behavior of the fields @xmath273 in ( [ eq:5 ] ) is governed by the classical liouville operator @xmath278 ( where @xmath279 is the poisson bracket . )
quantum mechanics enters the problem through the presence of moyal products `` @xmath280 '' in ( [ eq:5 ] ) .
in essence , this product operation@xcite limits the maximum resolution of the theory in phase space to scales of the order of a planck cell . in the following sections we reduce the above ` bare ' theory to an effective low energy theory describing localization phenomena . before turning to the technicalities of the mode integration ,
let us describe the relevant hierarchies in qualitative terms : fluctuations inhomogeneous in the phase space sector representing individual dots are expected to relax on short time scales comparable to the time of flight across the dot , @xmath281.@xcite on the other hand , relative fluctuations in the configuration of different dots can survive up to time scales of the order of the dwell time @xmath282 . to describe this hierarchical decay profile within our field theory framework , we focus on a section of the array containing two neighboring quantum dots ( cf .
[ coupling_phase_space ] ) .
the fields @xmath283 representing phase space fluctuations in this subsystem may be decomposed as @xmath284 , where @xmath285 are ` slow ' and ` fast ' fluctuations , respectively .
the slow fluctuations are ( i ) homogeneous within each dot separately .
in particular , ( ii ) they do not vary in any spatial cross section transverse to the array , and ( iii ) do not depend on momentum .
however , ( iv ) the weakness of the inter - dot coupling ( @xmath286 ) leaves room for gradual fluctuations of the slow modes as we pass from one dot into the other .
this suggests a parameterization @xmath287 , where @xmath5 is the component of @xmath258 parallel to the longitudinal direction of the two - dot system . point ( i ) above means that @xmath288 and @xmath289 , where @xmath290 are the constant slow mode configurations of the left and the right dot , respectively .
as we are going to check in a self consistent manner , ( v ) relative fluctuations between left and right dot are suppressed ( in a parameter of the order of the number of transverse quantum channels supported by the connector region ) so that a leading order expansion in relative fluctuations @xmath291 is sufficient .
( conceptually , this expansion is equivalent to the kramers - moyal expansion employed in the derivation of the fokker - planck equation above . )
the field @xmath292 encapsulates all other fluctuations , _
i.e. _ , fluctuations that do not meet the criteria ( i - v ) .
generic fluctuations of this type
think of fast fluctuations deep inside the phase space of one of the dots are strongly gapped and do not couple to the slow fluctuations .
( formally , this decoupling manifests itself in an effective ` orthogonality ' between the modes representing these fluctuations . )
however , there is one sector of phase space , @xmath293 , in which fast and slow fluctuations talk to each other .
the domain @xmath293 includes all points in phase space which pass from one dot to the other in a time of order @xmath294 , much smaller than the typical dwell time ( cf . the shaded area in fig . [ coupling_phase_space ] ) .
this region is special in that it overlaps with the domain of gradual variation of the slow fields ( cf .
point ( iv ) above . ) as we will see , a perturbative integration over fast fields in @xmath293 effectively determines the slow field coupling between the two dots . to prepare the integration over the fast fields in @xmath293
, we need to bring the notation to a more explicit level : assuming that the bulk dynamics is ballistic , @xmath295 where @xmath296 is the unit vector in momentum space , and @xmath297 the fermi velocity .
current conservation in the specular reflection at the system boundaries translates to the effective boundary condition @xmath298 , where @xmath299 is a phase function subject to the action of @xmath278 , @xmath300 a boundary point and @xmath301 the direction vector with flipped normal component .
we next parameterize fluctuations as @xmath302 , where the field generators carry a block structure ( in advanced / retarded space ) , @xmath303 ( for further technical details on this representation we refer to ref . .
) we now substitute these generators into the action , expand to leading order in @xmath304 , and integrate .
this leads to the effective action @xmath305=
\frac{1}{2}\left \langle \left(s^{(1)}[t_s , w_f]\right)^2\right\rangle + s_{\rm reg}[t_s],\ ] ] where @xmath306)(\dots)$ ] , and @xmath307 $ ] is of @xmath15th order in @xmath304 . due to the isotropy of @xmath308 in momentum space and the linearity of the poisson bracket in @xmath309 ,
the action does not contain a contribution of zeroth oder in the fast fields . the dominant contribution to the coupling between
fast and slow fluctuations is given by the linear term ,
@xmath310&= \beta \pi \hbar \nu v_f \int_{\gamma_c } d{{\bf q}}d{{\bf n}}\,n_\parallel \,{\rm str}\left[w_s({{\bf q}},{{\bf n } } ) \phi(q)\right ] \label{tsw}\end{aligned}\ ] ] where we have introduced the abbreviation @xmath311 , the integration over the direction of momentum is normalized as @xmath312 , @xmath313 is the component of @xmath309 parallel to the longitudinal direction and the moyal product has been omitted ( which is permissible due to the general relation @xmath314 for the integrated product of _ two _ functions presently , matrix elements of @xmath315 and @xmath13 in phase space . ) neglecting contributions of @xmath316 ( as compared to the @xmath317-terms above ) the quadratic @xmath318-action is given by @xmath319 & \simeq & s^{(2)}[\openone , w_f ] \nonumber \\ & = & \beta \pi \hbar
\nu \int_{\gamma_c } d{{\bf q}}d{{\bf n}}\,{\rm str}(\bar b_f({{\bf q}},{{\bf n } } ) ( v_f { { \bf n}}\cdot \nabla + \delta ) b_f({{\bf q}},{{\bf n}})).\end{aligned}\ ] ] fast field fluctuations can now be integrated out according to the prescription @xmath320 , where @xmath321 specifically , the effective action is given by @xmath322 & = & ( \beta \pi \hbar \nu v_f)^2\int_{\gamma_c } d{{\bf q}}d{{\bf q}}'d{{\bf n}}\ , n_\parallel^2 \mbox{str}\ , \left(\phi_{21}(q ) \phi_{12}(q ) \right ) \pi({{\bf q}},{{\bf q}}';{{\bf n } } ) \end{aligned}\ ] ] so far , we have not made reference to the specific properties of the phase space region @xmath293 .
we now assume that the corridor connecting the dots has wave - guide properties , in that it ( a ) does not contain significant backscattering , and ( b ) the restriction of the liouville operator to @xmath293 has plane - wave like eigenfunctions characterized by a conserved longitudinal / transverse momentum @xmath323 .
we also assume that ( c ) the slow fields smoothly interpolate between @xmath324 and @xmath325 in a region centered around the longitudinal coordinate @xmath326 .
the presumed proximity @xmath327 implies that @xmath328 can be linearized , where we suppressed the slow field index ` @xmath329 ' for notational transparency . under these circumstances , and noting that the integration over @xmath330 implies a projection onto the zero - momentum sector @xmath331 , we obtain @xmath322 & = -\beta \pi \hbar \nu s_c v^2_f\int_{-c}^c dq dq ' { \rm \,str}\left(\phi_{21}(q)\phi_{12}(q ' ) \right ) \int d{{\bf n}}\ , n^2_\parallel \int \frac { dk } { 2\pi } \frac{e^{i k ( q - q')}}{i v_f n_\parallel k+ \delta } = \\ & = -\beta \pi \hbar \nu s_c v_f\int_{-c}^c dq dq ' { \rm \,str}\left(\phi_{21}(q ) \phi_{12}(q ' ) \right ) \underbrace{\int d{{\bf n}}\ , |n_\parallel|\theta\left ( \frac{q - q ' } { n_\parallel}\right)}_{={\rm const.}}=\\ & = { \rm const.}\times \hbar \nu s_c v_f\ , { \rm str}((\bar b_r-\bar b_l)(b_r - b_l)),\end{aligned}\ ] ] where the indices refer to `` ar''-space , @xmath332 is a constant , @xmath333 the transverse cross section of the contact , and @xmath334 are the generators of the slow fields in the left and the right dot , respectively . noting that the density of states per volume , @xmath335 ( where @xmath336 is the fermi momentum ) and @xmath337 is proportional to the number of transverse channels , @xmath338 , supported by the connector region , the prefactor can be written as @xmath339 , a number which we assume large lest a semiclassical description of the contact becomes meaningless .
we also note that the quadratic form may be replaced by its unique rotationally invariant generalization to the full field manifold , @xmath340 , where @xmath341 .
while a quadratic expansion of the latter expression reproduces the bilinear term , the largeness of @xmath338 implies that typical values contributing to the @xmath342-integration , @xmath343 , which means that non - linear contributions become inessential in the limit of large channel numbers .
we thus conclude that the coupling term can be rewritten as @xmath344= { \rm const . }
\times n { \,\rm str}(q_l q_r).\ ] ] finally , the obvious generalization of the above two - dot construction to an entire quantum dot array reads as @xmath345= { \rm const . }
\times \frac{1}{\tau_{\rm d } \delta } \sum_{m } { \rm str}(q_m q_{m+1}),\ ] ] where @xmath346 is the @xmath347-matrix representing the @xmath122-th dot and we used that the channel number @xmath348 . before turning to the discussion of localization properties , a few remarks on the construction above
are in order . *
conceptually , the above reduction programs involves three steps : 1 ) identification of ` low energy modes ' , _ i.e. _ , modes that decay on the largest time scales of the problem , 2 ) identification of ` high energy ' , or quickly decaying modes , and 3 ) perturbative integration over those fast modes that will conceivably couple to the slow modes . in principle , that integration can be explicated for any fluctuation in the problem . in practice , however , only few modes will effectively couple to the slow degrees of freedom , and these relevant fluctuations are best identified by semiclassical considerations : * semiclassically speaking , a ` mode ' represents the coherent propagation of a retarded and an advanced feynman amplitude along classical trajectories in phase space . locally ,
the semiclassical dynamics of such composites is described by the liouville operator , as is manifest in the action ( [ eq:5 ] ) .
this trajectory interpretation helps in identifying the relevant fast modes .
e.g. , in the system depicted in fig .
[ coupling_phase_space ] , the coupling between the ergodic slow modes of each dot , @xmath346 , is dominated by trajectories swiftly propagating from one dot to the other , _ i.e. _ , modes emanating at phase space points @xmath349 . *
although this identification of fast modes rests on specific model assumptions ( no backscattering in the contact region , etc . ) , the result ( [ qdotarray ] ) is reasonably universal .
for example , a somewhat more elaborate construction will show that the same action describes connector regions containing chaotic scattering and momentum relaxation by themselves .
( the latter complication would manifest itself in an altered value of @xmath88 , though . ) generally speaking , the low energy physics of the system will reduce to @xmath350 , as long as the dots are isolated from each other in the sense @xmath351 .
the effective action ( [ qdotarray ] ) is equivalent to a ` lattice version ' of the diffusive nonlinear @xmath14-model of quasi - one dimensional disordered wires .
indeed , we may pass to a continuum limit @xmath352 where @xmath353 is replaced by a smooth field , @xmath354 , and @xmath355 the spacing between the dots . comparing to the standard form of the diffusive model,@xcite where the action is @xmath356 with @xmath357 the localization length , we are led to the identification @xmath358 .
in the preceding sections we showed how a theory of anderson localization can be constructed for a sample in which the microscopic electron dynamics is ballistic , rather than disordered - diffractive .
our theory of `` ballistic anderson localization '' paraphrases the scaling approach to localization in disordered quantum wires of dorokhov , mello , pereyra , and kumar,@xcite using the language of the trajectory - based semiclassical formalism .
as noted in the introduction , the interest of constructing such a semiclassical theory is not the structure of the theory itself or the phenomena it explains . like most semiclassical theories of quantum corrections in the perturbative regime
, the structure of the theory closely resembles the structure of its fully quantum mechanical counterpart for disordered metals , whereas the observed phenomena are the same in the ballistic and disordered cases .
instead , the interest of the semiclassical theory is that it shows how quantum effects that were known from disordered metals arise if the electron dynamics is ballistic .
on hindsight it should not come as a surprise that a theory of ballistic anderson localization can be constructed by adapting the derivation of the dorokhov - mello - pereyra - kumar equation for a disordered quantum wire .
this point is best made by reconsidering dorokhov s original derivation.@xcite in this derivation , impurity scattering is treated in the born approximation .
all quantum mechanical amplitudes are squared into quantum probabilities .
hence , it is sufficient if one can replace the quantum mechanical probabilities by classical ones .
such a replacement is a standard procedure when connecting quantum mechanical and semiclassical theories .
its implementation for the array of chaotic cavities is what is done here .
it is interesting to observe that , while small - angle encounters form a crucial link in our understanding of quantum interference corrections in ballistic conductors , their role is very limited in our description of localization in quasi - one dimension : they serve to cancel spurious contributions from trajectories that enter the last cavity of the array more than twice .
( in dorokhov s original approach such processes are excluded automatically because of the condition that the length of the wire is increased by an amount @xmath359 much smaller than the mean free path @xmath86.@xcite ) implicitly , small - angle encounters do play a much more important role in our theory , however , because they help to preserve unitarity in the semiclassical theory .
unitarity is a key ingredient of both the quantum - mechanical derivation of the dmpk equation and the present semiclassical derivation .
we note that unitarity has played an important role in other extensions of the semiclassical framework beyond its previously assumed domain of validity : it is used to relate weak localization and enhanced backscattering , thus enabling a semiclassical description of weak localization without reference to small - angle encounters,@xcite and it is used to obtain an alternative expression for the spectral form factor , allowing its calculation in the non - perturbative regime by considering periodic orbits of duration below the heisenberg time only.@xcite we have also shown that the dynamical information entering the semiclassical approach can be processed by different means to derive an effective low energy field theory of the system .
this latter approach is based on the concept of ` modes ' , fluctuations in phase space decaying on parametrically different time scales . a successive integration over short lived modes stabilizes an effective action of the most persistent modes in the fluctuation spectrum . in the present context , that low energy limit turned out to be equivalent to the diffusive nonlinear @xmath14-model of disordered wires .
while the field theory approach is arguably less explicit than the direct classification of trajectories , it enjoys the advantage of high computational efficiency , a paradigm previously exemplified on the problem of universal spectral correlations.@xcite for example , the above mentioned condition of unitarity , as well as the symmetries relating trajectories to their time reversed are built into the field theory approach from the outset ; there is no need for explicit bookkeeping in terms of encounter processes .
the price to be payed for this compactness in the description is a higher level of abstraction , though .
we thank fritz haake for bringing this problem to our attention . this work was supported by the packard foundation , the humboldt foundation , the nsf under grant no .
dmr 0705476 , and by the sonderforschungsbereich sfb / tr 12 of the deutsche forschungsgemeinschaft .
|
we show that anderson localization in quasi - one dimensional conductors with ballistic electron dynamics , such as an array of ballistic chaotic cavities connected via ballistic contacts , can be understood in terms of classical electron trajectories only . at large length scales ,
an exponential proliferation of trajectories of nearly identical classical action generates an abundance of interference terms , which eventually leads to a suppression of transport coefficients .
we quantitatively describe this mechanism in two different ways : the explicit description of transition probabilities in terms of interfering trajectories , and an hierarchical integration over fluctuations in the classical phase space of the array cavities .
|
wide - angle tailed ( wat ) radio sources were first defined as a class by owen & rudnick ( 1976 ) who described them as a sub - class of head - tail radio sources .
head - tail sources include a radio core and tails of radio emission trailing from the core as the host galaxy moves through the dense gas of a cluster s intracluster medium ( icm ) .
head - tail galaxies also include narrow - angle tails ( nats ) , which are generally associated with galaxies with large peculiar velocities in the outer parts of clusters ; the ram pressure from the interaction with the icm bends the radio lobes to create the nat structure .
owen & rudnick found that wats , however , are most often associated with bright elliptical or cd galaxies in the cores of clusters , and surmised that their smaller peculiar velocities would therefore produce smaller bends in the radio lobes , resulting in a wide - angle , rather than a narrow - angle , tailed morphology .
later work , by e.g. eilek et al .
( 1984 ) and odonoghue et al .
( 1993 ) , found that the peculiar velocities of the host galaxies of wats were too small to produce the observed bending of the lobes by icm ram pressure .
in addition , odonoghue et al . ( 1993 ) further defined the class of wats as radio sources usually fulfilling five criteria : ( 1 ) association with a dominant galaxy in a cluster center , ( 2 ) radius from the core to the edge of each lobe @xmath5 kpc , ( 3 ) a dramatic transition from a collimated jet to a radio hot - spot ( 4 ) large - scale bends of both radio lobes in a c - shape , ( 5 ) radio power within a factor of three of the fr break .
point ( 5 ) refers to the break in radio power found by fanaroff & riley ( 1974 ) ; fr i sources typically have radio powers @xmath6 w / hz , lobes that darken at the edges and that are often distorted , while fr ii sources are the classic collinear doubles with @xmath7 w / hz and edge - brightened lobes . at low-@xmath1 , fr i sources are usually found in clusters or groups of galaxies ; fr ii sources usually are not ( hill & lilly 1991 ) .
since a wat s peculiar velocity relative to the cluster mean has been found to be insufficient to distort the radio lobes into the observed angles from icm ram pressure , alternate scenarios have been proposed for the bending of the lobes .
eilek et al .
( 1984 ) suggested that the icm may have large - scale fluid flows , left over from the cluster s formation .
burns et al .
( 1994 ) added that large flows in the icm could result from large scale cluster - cluster mergers .
the merging gas could have velocities high enough ( @xmath8 km / s ) to produce ram - pressure bending of the lobes , as shown in the simulations of roettiger et al .
( 1993 ) .
wat radio sources should trace out clusters of galaxies , and in particular , they may locate clusters that have recently merged . using a sample of 384 wat and nat radio sources selected from the first 3000 deg@xmath9 of the vla faint images of the radio sky at twenty - cm ( first ) survey ( becker , white , & helfand 1995 ) ,
we have confirmed the existence of dozens of clusters with redshifts up to @xmath2 ( blanton et al .
2000 , 2001 , 2003 ) .
imaging and multi - slit mask spectroscopic follow - up on a moderate - to - high redshift subset of ten of these revealed eight clusters with redshifts as high as @xmath10 ( blanton et al .
a complete ( area- and magnitude - limited ) low-@xmath1 subset including 40 of the 384 objects was observed optically , and @xmath11 of them were found to be in richness class abell 0 or greater clusters ( blanton et al .
the majority of the low - redshift sources that are found in clusters , based on our optical richness measurements , are detected in the rosat all sky survey ( five out of six detected for @xmath12 ) .
those that are not found in clusters , based on our richness measurements , are not detected in the rass ( only one out of nine detected for @xmath12 ) .
in other words , the combination of radio and optical data gives us a very clear indication for the presence , or lack of , a cluster , as confirmed with x - ray observations . in order to identify high-@xmath1 candidates
, we obtained @xmath13-band images of fields surrounding wats that were blank to the limit of the digitized sky survey ( dss ) .
some of these had no optical identifications even down to the limit of our images ( @xmath14 ) .
we then performed near - infrared ( nir ) observations at the positions of these sources in hopes of uncovering very red , high-@xmath1 cluster galaxies . here
we present optical and nir observations taken at the position of the wat radio source , 1137 + 3000 .
a combination of imaging and spectroscopy confirms that this source is located in a distant cluster of galaxies with a redshift @xmath2 .
we assume @xmath15 km s@xmath4 mpc@xmath4 , @xmath16 , and @xmath17 throughout .
the radio source is detected in the _ first _ survey as a three component object , including a core and two lobes .
the source is clearly bent with an opening angle between the two lobes of @xmath19 .
the size of the source ( lobe to core to lobe ) is 30@xmath20 ( 240 kpc ) . the total flux density , as measured by _ first _ is 9.1 mjy . the flux density measured from the low - resolution nrao vla sky survey ( _ nvss _ ; condon et al .
1997 ) is 9.8 mjy , slightly higher than the _ first _ value .
using the _ nvss _ flux density value , and assuming @xmath21 with @xmath22 , the radio power of the source at 1440 mhz is @xmath23 w hz@xmath4 .
this power is slightly below the fr i / ii break for radio galaxies ( fanaroff & riley 1974 ) , and is typical of a wat source ( odonoghue et al .
an overlay of the _ first _ radio contours onto a keck near - infrared camera ( nirc ) image in the k@xmath24 filter of the cluster center is shown in figure [ fig : ovly ] .
the figure is 280 kpc on a side for our assumed cosmology .
the radio core is coincident with an elliptical galaxy , as seen in the near - infrared image .
this nir image , which covers only the inner 35@xmath20 region of the cluster , is used for display purposes only . vs. @xmath25 cmd .
the top panel includes the objects within a @xmath26 radius from the radio source ; the bottom panel includes the objects in the entire 15.6 arcmin@xmath9 field .
there is a concentration of objects with @xmath27 that are likely cluster members .
spectroscopically confirmed cluster members are circled.[fig : krk],width=283 ] the 1137 + 30 field was observed in the nir with irim at the kpno 2.1 m on ut 25 and 26 may 1997 .
the total exposures were 64 and 32 min . in the @xmath28-band and @xmath29-band , respectively .
the optical observations were obtained at the mdm 1.3 m with a total exposure in the @xmath13-band of 7200 s on ut 21 may 1998 and were kindly provided by j. kemp .
focas was used to measure magnitudes for the standard stars and objects in the target fields for the nir and optical data .
aperture magnitudes were measured for objects in the target fields .
all images were scaled to have the same pixel scale ( @xmath30/pix ) .
the @xmath13 and @xmath29 images were smoothed with a gaussian to match the seeing in the @xmath28-band image ( the worst of the three , with 1.@xmath207 ) .
we then used focas to measure magnitudes within an aperture with a diameter equal to 2@xmath31fwhm ( @xmath32 = 6.8 pixels ) .
this is close to a total magnitude and will measure the same part of the galaxy in each of the bands . at @xmath2 ( see [ sec : spectra ] ) , this aperture corresponds to 27 kpc for our assumed cosmology .
the aperture magnitudes , measured as described above , for the radio source counterpart are 22.90@xmath33 , 19.46@xmath34 , and 17.78@xmath34 , in @xmath13 , @xmath29 , and @xmath28 , respectively .
this gives colors of ( @xmath35 ) = 5.12@xmath36 , ( @xmath37 ) = 1.68@xmath38 , and ( @xmath39 ) = 3.44@xmath36 .
the absolute magnitude of the host galaxy is @xmath40 , assuming no evolution , a redshift @xmath2 ( see [ sec : spectra ] ) , and a @xmath28-correction from coleman , wu , & weedman ( 1980 ) .
this is typical of a bright elliptical galaxy , but fainter than a cd .
color - magnitude diagrams ( cmds ) were constructed for objects in the entire 15.6 arcmin@xmath9 field as well as for only the objects found within a @xmath26 radius aperture ( 480 kpc at @xmath41 , see [ sec : spectra ] ) from the radio source .
the @xmath28 vs. @xmath25 cmd is presented in figure [ fig : krk ] .
the top panel includes the objects within @xmath26 from the radio source , while the bottom panel includes the objects from the entire field .
the objects that are confirmed to be at the cluster redshift ( see [ sec : spectra ] ) are circled .
there is a much larger range in the colors of objects when looking at the cmd for the whole field .
the objects in the @xmath26 radius aperture are more concentrated on the diagram than those in the whole field , and are redder , consistent with membership in a distant cluster .
most of the objects within @xmath26 from the radio source have colors in the range @xmath42 , consistent with early - type galaxies at @xmath0 . a long - slit observation ( 60 min . ) of the cluster that included the radio counterpart and two other galaxies was made at the keck ii 10 m telescope with the low resolution imaging spectrometer ( lris ; oke et al .
1995 ) on 13 april 1998 .
a multi - slit mask observation of the field surrounding the radio source was made at the keck ii 10 m with lris on 29 may 2001 with a total exposure of 60 min .
objects were chosen for the slit mask based on their @xmath25 colors .
objects that had @xmath42 were candidates for inclusion on the mask , since they provide us the best opportunity of confirming galaxies at redshifts @xmath43 .
the mask included a total of twelve program object slits .
spectra were successfully extracted for ten of the program objects .
redshifts were initially estimated by identifying a few obvious lines , including the ca ii h+k break , and the [ o ii ] @xmath44 and [ o iii ] @xmath45 emission lines .
the spectra were then fourier cross - correlated with an elliptical galaxy template using fxcor in iraf .
all ten program object spectra that were successfully extracted had redshifts in the range @xmath46 , giving us a @xmath47 success rate of identifying cluster galaxies based on their nir / optical colors .
the spectrum of the radio host galaxy is shown in figure [ fig : spect ] , and is typical of an elliptical galaxy .
we calculated the line - of - sight velocity dispersion using @xmath48 , where @xmath49 .
we find @xmath3 km s@xmath4 ( @xmath50 confidence ) .
this is a typical value for an abell richness 0 cluster ( lubin & bahcall 1993 ) .
we have provided photometric and spectroscopic evidence at optical and near - infrared wavelengths for a high - redshift cluster of galaxies associated with a wide - angle tailed radio galaxy at @xmath41 .
the majority of high - redshift clusters found in traditional x - ray and optical surveys select the most x - ray luminous and optically rich clusters .
the radio selection technique employed here is an important complement to those methods , in part because it can locate high - z clusters with a wide range of x - ray luminosities and optical richnesses , thus aiding in the study of galaxy evolution and cluster formation .
|
the morphologies of wide - angle tailed ( wat ) radio sources ( edge - darkened , c - shaped , fr i radio sources ) are the result of confinement and distortion of the radio lobes by the dense x - ray - emitting gas in clusters or groups of galaxies .
these radio sources are easily seen at high redshifts ( @xmath0 ) in short - exposure images from the faint images of the radio sky at twenty - cm ( first ) survey . using a sample of approximately 400 wat sources from the first survey
, we have discovered a number of high-@xmath1 clusters . here
, we present the highest-@xmath1 cluster found so far using this method : 1137 + 3000 at @xmath2 .
we include photometric and spectroscopic results .
ten galaxies are confirmed at the cluster redshift , with a line - of - sight velocity dispersion of @xmath3 km s@xmath4 , typical of an abell richness class 0 cluster .
[ 1996/06/01 ] a&a
|
when i was asked to give this plenary lecture in place of ashoke sen , i had some considerable misgivings that i would be able to do the subject the justice that he would undoubtedly have .
however i had no doubts about what i should speak .
the subject formerly known as string theory , and increasingly frequently being referred to as m - theory , has made some stunning advances since gr14 .
hence my title .
i should perhaps apologize , especially in india , for its wording .
this was dreamed up rather hurriedly to meet the printer s deadline . as far as physics
is concerned it would be more accurate to refer to the fourth , fifth or sixth millennium since the mathematical study of physics and astronomy must be at least as old as the the great river valley civilizations associated with the indus , tigris and euphrates .
however coming as i do from a a rather obscure corner of north - western europe i had , when i gave my title to the organizers , very much in mind the forthcoming christian millennium celebration and its immediate predecessor .
that earlier occasion provides an apt metaphor for the current activity in this subject : it resembles in many ways the construction of the great mediaeval cathedrals which followed the failure of the universe to live up to the most important cosmological prediction of those times , that it should come to an end in the year 1000 ad , thus appearing to demonstrate that physics is invariant under arbitrary shifts of the origin of the time coordinate . like string theory , the construction of the great cathedrals was a collective endeavour which took literally ages to complete , in many cases centuries .
although the beauty of the individual elements of the design would have been apparent fairly immediately , few if any of those working on the project at the beginning would have had much idea of the final shape of the structure that eventually emerged and which now combines those individual elements in such a harmonious whole .
sometimes , as in the case of beauvais in northern france , the whole building fell crashing down and had to be completely rebuilt .
in fact these cathedrals were frequently located on the sites of much earlier churches or indeed pre - christian temples often going back to the romans and before . walking around
one sometimes finds , embedded in the floor or the lower parts of the wall fragments of these older structures put to a new use .
@xmath0 much of this , it seems to me is true of the present state of string theory , only slowly are we beginning to get some idea of the final structure .
even now we ca nt be sure that it will hold up .
there is another , and more particular , way in which string theory resembles the mediaeval cathedrals .
one of the key features of many of them is is the existence of a spire placed on top of a high tower .
if one visits the cathedral one may climb to the top of the tower and then one may ascend the spire to gain a spectacular view of the surrounding countryside , the city and the cathedral precincts below . for us
the analogue of height above ground level is spacetime dimension and this brings me to what is perhaps the most important message of this talk : really !
the word really " is meant in the technical sense : the point is that we need to consider spinors in eleven dimensional spacetime and it is a useful and convenient fact that the real clifford algebra ( 10,1 ; r ) ( 32 ) . what is meant here that in eleven spacetime dimensions the algebra generated by the gamma matrices @xmath1 , @xmath2 _ _ + _ _ = 2 _ , with spacetime signature chosen so that _ 0 ^ 2 = -1 , is isomorphic to that of @xmath3 real matrices .
thus we may take the gamma matrices @xmath4 be @xmath5 and they act on the real @xmath6-dimensional space @xmath7 consisting of _ majorana spinors _
@xmath8 , @xmath9 .
the space of majorana spinors @xmath7 carries a @xmath10-invariant symplectic form @xmath11 , the _ charge conjugation matrix _ , which may be used to raise and lower spinor indices .
thus for example one finds that if one lowers an index on the gamma matrices they become _ symmetric _ : ( c ^ ) _ ab = ( c ^ ) _ ba . for later use
we remark that we must make an arbitrary choice when constructing the clifford algebra @xmath12 .
the image in the clifford algebra of the volume form commutes with all other elements of the algebra and may be taken to be a multiple of the identity .
we make the choice : _ 10= _ 0 _ 1 _ 9 ( we could have chosen the minus sign ) .
now let s look down to four dimensions .
using the major arithmetical theorem that 32 = 4 8 , we see that @xmath7 we will decompose into eight copies of the four real dimensional space of majorana spinors for @xmath13 , and in a like fashion the eleven - dimensional charge conjugation matrix decomposes into the sum of eight copies of the familiar charge conjugation matrix of ordinary physics .
we are in fact using the isomorphism ( 3,1 ; r ) ( 4 ) .
many participanrts at gr15 will be more familiar with weyl spinors and perhaps with the opposite signature convention . in that language
a majorana spinor is a pair of two - complex -component weyl spinors while c= with ^= .
let s return to eleven dimensions .
supersymmetry transformations depend on a constant spinor @xmath14 and act on _ superspace _ , which is defined to be ^10,1 s with coordinates @xmath15 as + x^x^+ 12 ^t c ^. one now takes a suitably graded semi - direct product with the eleven dimensional poincar group @xmath16 to get the super - poincar group .
note that the spinors @xmath17 and @xmath14 are anti - commuting variables .
the supertranslations have generators @xmath18 which transform as majorana spinors under the super - adjoint action of spin(10,1 ) and trivially under the adjoint action of the ordinary spacetime translations .
the non - trivial anti - commutation relation is , in index notation , q_a q_b + q_b q_a = p _ ( c^ ) _ ab .
descending to four dimensions we get eight four - component supercharges @xmath19 and the algebra of @xmath20 supersymmetry .
given the above information it is in principle possible to construct classical field theories in eleven dimensions .
however any such theory must contain particles of spin two and it is widely believed that _ there is only one possibility _ , the theory of cremmer and julia @xcite .
this contains g _
f_[]= 4 _ [ a _ ] and a ^a _ .
an elementary exercise in linear theory shows that there are 128 boson and 128 fermion degrees of freedom . in higher dimensional lorentzian spacetimes , or in eleven dimensions but with more supersymmetry , particles with spins greater than two seem to be inevitable .
that is why at present the pinnacle of the spire stops here .
however that has not dissuaded modern - day icari ( if indeed that is the plural of icarus ) from launching themselves into the blue yonder .
future architectural innovations may well include multi - temporal theories in twelve or thirteen dimensions .
those with little head for heights may descend the spire to the tower where life is more varied . we still have @xmath6 component since _ 10 ^ 2 = ( _ 0 _ 9 ) ^2 = 1 we may decompose the real @xmath6-dimensional space @xmath7 of majorana spinors into two real @xmath21 dimensional spaces @xmath22 of _ majorana - weyl _
spinors s= s^+ s^- with _ 10 s^= s^.
we now have _ three _ types of superymmetry and three types of superspace depending upon the chirality of the supercharges .
* type i theories have just one supercharge and is therefore necessarily chiral .
the superspace is ^9,1 s^+ .
* type iia theories have two supercharges , one of each chirality .
the superspace is
^9,1 s^+ s^- . *
type iib theories have two supercharges of the same chirality , and are thus also chiral .
the superspace is ^9,1
s^+ s^+ .
these include * super - yang - mills theory .
this has f _ and both in the adjoint representation of some compact gauge group @xmath23 . in many ways
this is the big - daddy of all gauge theories .
for example , this theory , reduced to four spacetime dimensions , gives the @xmath24 supersymmetric yang - mills theory which has so many deep and beautiful links with mathematics and geometry .
we shall see shortly that it has a central role to play in m - theory .
* type i supergravity .
this has as bosonic fields g _ and h_= 3 _ [ a _ ] . * the @xmath25 open superstring .
these include * type iia supergravity .
+ this has the bosonic fields of type i supergravity , referred to in this context as the _ neveu - schwarz@xmath26neveu - schwarz sector _ together with a closed two - form and a closed four - form field strength , referred to in this context as the _ ramond@xmath26ramond sector_. + and * the type iia closed superstring .
these include * type iib supergravity .
this has the bosonic fields of type i supergravity , often called for obvious reasons , the common sectortogether with a ramond@xmath26ramond sector consisting of closed one , three and five - forms .
the five - form @xmath27 is self - dual c_5 = c_5 , where @xmath28 denotes the hodge - dual which satisfies @xmath29 in @xmath30 . *
the type iib closed superstring . for closed strings ,
one has the additional possibility of the left and right moving modes on the string behaving differently and this gives rise to the two further possibilities with gauge group @xmath31 or @xmath32 .
it is of the course the latter which has received most attention for phenomonological purposes .
we get down to four dimensions with @xmath33 supersymmetry by taking the ten dimensional spacetime @xmath34 as a product ^9,1= e^3,1 cy where @xmath35 is a calabi - yau space , i.e. a closed riemannian six - manifold with holonomy @xmath36 .
thus , if one regards the field theories as limiting low - energy cases of super - string theories one has five possibilities .
the next most important point of this lecture is that .5 cm = 10000 .5 cm called .5 cm .5 cm @xmath37 the five string theories and eleven - dimensional supergravity are conjectured to be related by a web of _ dualities _ which interchange the perturbative _ elementary _ states which we encounter in linear theory with the non - perturbative _ bps soliton _ states which we only see in the fully non - linear theory .
i will have more to say about what precisely m - theory is expected to be later .
the basic dualities are of two types referred to as _ t - dualities _ and _ s - dualities_. they are believed to be combined in a more general symmetry called _
u - duality_. these are symmetries of perturbative string theory and may be shown to hold to all orders in perturbation theory @xcite .
they appear when one considers theories for which ^9,1= t^d ^9-d,1 .
the torus is equipped with a constant metric @xmath38 and a constant two - form @xmath39 and t - duality is a generalization of the idea of interchanging the torus , equipped and its metric @xmath40 with its dual or reciprocal torus and its dual or reciprocal metric @xmath41 .
a familar example of this in the everday physics of three dimensions is the duality between the face centred cubic lattice and the body centred cubic lattice .
the voronoi or wigner - seitz cell of the former is the rhombic dodecahedron whose dual is the cuboctahedron , which is the voronoi cell of the latter .
string theories related by duality are believed to be identical . in other words
it is one of the gauge symmetries of string theory , the simplest example arises when we have a a circle @xmath42 of radius @xmath43 and t - duality acts as r , where @xmath44 is the fundamental length that enters string theory .
more generally , the moduli spaceof torus theories is specified by giving a @xmath45 matrix @xmath46 with positive definite symmetric part is acted on by @xmath47 acting by fractional linear transformations e ( ae + b ) ( c e + d)^-1 where , if m= and j= , then m^t j m = j. the group in question is @xmath48 it is is a new and unexpected non - perturbative symmetry of string theory interchanging weak and strong coupling whose existence and importance was pointed out by ashoke sen @xcite and who has used it so effectively .
perhaps its most exciting feature is that it also interchanges classical with quantum properties .
s - duality is not all apparent in perturbative string theory , it only manifests itself indirectly .
thus in n=4 supersymmetric four - dimensional gauge theory , if = 2 + 4 i g^2 where @xmath49 is the usual gauge coupling constant and @xmath50 is the theta angles - duality acts by fractional linear transformations where sl(2 , z ) .
gauge theories with couplings so related are believed to be identical .
no proof is known but many , apparently rather delicate , tests have been made of this conjecture and no contradiction has been found .
in fact these tests involve extremely subtle and unexpected properties of the @xmath51 co - homology of the hyper - khler manifolds which are the moduli spaces of bps monopoles @xcite .
in string theory and supergravity theory we have g= e^ and the role of @xmath50 is played by a pseudoscalar field called an axion .
thus for example it is known that @xmath52 is a symmetry of type iib classical supergravity theory .
the exterior derivative of the axion is the closed ramond@xmath26ramond one - form mentioned earlier .
it is expected that quantum - mechanical effects associated with dirac quantization of electric and magnetic charges will break this continuous symmetry down to @xmath53 . by dimensional reduction on tori
the s - duality symmetry descends to lower - dimensional super - gravity theories .
of course they also acquire an explicit ( and continuous ) t - duality symmetry as well .
studies of super - gravity theories in the past revealed that one often gets more .
for example the @xmath20 supergravity theory in four spacetime dimensions is invariant under the action of @xmath54 .
hull and townsend @xcite have suggested that a discrete subgroup of @xmath54 should persist in the full quantum mechanical m - theory .
the non - perturbative soliton states , analogous to bps monopoles in yang - mills theory , which are acted on by non - perturbative dualities correspond to _ p - branes_. roughly speaking these are extended objects with p spatial dimensions which as they move through time draw out a p+1 dimensional world volume .
thus in @xmath55 spacetime dimensions p&=&0 + p&=&1 + p&=&2 + p&=&n-2 + p&=&n-3 + p&=&-1
+ one says that a p - brane state @xmath56 is bps if is invariant under one or more supersymmetry transformations q |p = 0 .
bps states typically carry central charges @xmath57 per unit p - volume and if if @xmath58 is the energy per unit p - volume of such a state it typically attains a _ bogomolnyi bound _ @xcite giving a lower bound for @xmath58 among all states with the same central charge m |z| .
in supergravity theories bps p - branes are very well known to the present audience .
they correspond to extreme black holes and higher dimensional analogues .
the spacetime of such a solution is invariant under the action of the poincar group @xmath59 times rotations of the @xmath60 dimensional transverse space @xmath61 .
i wo nt give a complete survey of all possibilities but remind you that a typical metric looks like : ds^2 = h^-2 p+1 ( -dt^2 + d*x*_p^2 ) + h^ 2 d_t-2 d*y*_t^2 , where @xmath62 is an arbitrary harmonic function on @xmath63 .
thus more than one p - brane may rest in equipoise .
the solutions admit killing spinors of the relevant supergravity theory . near the horizons ,
the symmetry and the superymmetry is frequently enhanced , the metric tending to the product @xmath64 .
typically one may think of the p - branes as spatially interpolating between different vacua or compactifications of the associated supergravity theories .
the basic and fundamental example is of course the majumdar - papapetrou solution of the einstein - maxwell equations in four spacetime dimensions .
it is important to note that supergravity p - branes may carry electric or magnetic charges .
associated to a @xmath65- form or @xmath66 form respectively .
the magnetic charges are necessarily solitonic , while the electric ones may sometimes be envisaged as arising from sources . in general charges
may be of neveu - schwarz@xmath26neveu - schwarz or ramond@xmath26 ramond origin .
in string theory electric neveu - schwarz@xmath26neveu - schwarz charge is carried by a _ fundamental string _ which corresponds to an elementary state of string theory .
however perturbative string states can not carry magnetic neveu - schwarz@xmath26 neveu - schwarz charge neither can they carry electric or magnetic ramond@xmath26 ramond charge .
there are no such local sources in perturbative string theory .
we have here a case of what misner and wheeler might have called note that in the quantum theory these charges should satisfy an analogue of dirac s quantization condition for electric andc magnetic charges . it is perhaps worth explaining here the origin of the rather quaint looking tensor product notation .
it derives form the fact that a the world sheet of a closed string is topologically @xmath67 which admits two spin structures , * neveu - schwarz , which corresponds to antiperiodic spinors + and * ramond , which corresponds to periodic spinors , and both must be included when considering the fermionic oscillations of the superstring .
the bosonic fields are built up as tensor products of left and right moving fermionic states , and must be periodic .
they thus fall into two sectors .
shifting now to ten - dimensional spacetime we see that the fermionic states will be associated with the tensor product of the space of majorana spinors with itself i.e. with @xmath68 .
but by virtue of the charge conjugation matrix @xmath69 we may identify @xmath70 with the clifford algebra @xmath71 which , as a vector space is the same as the grassmann algebra of forms @xmath72 .
type iia theories involve the even forms and type iib theories the odd forms .
a key observation of polchinski @xcite , which resolves the puzzle above , is that open string theory admits an entirely new type of state corresponding to p - branes .
they have come to be called _
dirichlet p - branes_. their importance is that their quantum mechanical properties can be discussed within the comparitively mathematically secure framework of conformal field theory .
the basic idea is consider decomposing ten - dimensional minkowski spacetime as ^9,1 = e^p,1 ^d_t with coordinates @xmath73 , @xmath74 and @xmath75 .
we now decree that end of the the string remains fixed on the hyperplane @xmath76 . in other words the coordinates of the string fields @xmath77 , where @xmath78 are space and time coordinates on the string world sheet are to be subjected to a mixture of the usual * neumann boundary conditions _ x^=0 and * dirichlet boundary conditions _
t y^a=0 .
note that world sheet hodge duality interchanges @xmath79 and @xmath80 and hence neuman and dirichlet boundary conditions .
polchinksi was able to show that in string theory such d - brane states are bps and like their supergravity cousins they carry the correct ramond@xmath26ramond charges .
moreover these charges satisfy the analogue of dirac quantization conditions , called in this context the _ nepomechie - teitelboim _ quantization conditions .
we may suppose that integrating over the string fluctuations will give an effective action for for the motion of light p - branes moving in a a fixed external field .
such so - called _ dirac - born - infeld _ actions have been obtained by a number of authors .
in addition to the world volume fields giving the embedding of the brane in @xmath30 , @xmath81 susy dictates that one must include a world volume closed two - form @xmath82 .
in flat space with trivial ramond@xmath26ramond and neveu - schwarz@xmath26neveu - schwarz fields dirac - born - infeld action reduces to - d^p+1x , where g _ = _ z^x
^ z^x ^ is the pull - back to the world volume of the flat ten - dimensional metric @xmath83 .
it is perhaps a striking fact that we encounter here embedded in the structure of m - theory a a fragment of the old , an unsuccessful , attempt of born and infeld to construct a finite classical theory of the electron .
now is not the place to dwell on this in detail but it is a striking consequence of the new developments that what was a major blemish of the that ancient religion : the singular source of the electron may now be understood as the end of a fundamental string ending on a d - brane @xcite .
i have no time to dwell at length on the details but it is perhaps worth pointing out here the simple underlying idea that permits the construction of superymmetric p - brane actions , including super - string actions , in a unified way . it is to to replace the bosonic idea of a map from a @xmath84 dimensional manifold of the form @xmath85 where @xmath86 is the p - brane s spatial manifold by into spacetime , @xmath87 for example , by a map into superspace @xmath88 .
perhaps the most persuasive evidence for the essential soundness of our current foundation is the remarkable calculations of the thermal properties of black holes initiated by strominger and vafa @xcite and developed further by callan and maldacena @xciteand then by many other people @xcite . both for reasons of time and
because there have been some been some fine expositions inthe parallel sessions , i shall not review them in detail but merely note some essential points . the first thing to remind ourselves
is why this idea is even feasible without a complete quantum theory of gravity . for a general black hole , the bekenstein - hawking entropy s_bh= 14 g a where @xmath89 is the horizon area .
this expression involves newton s constant @xmath23 . in string theory
we expect that g so that @xmath23 should depend on the expectation value of the dilaton @xmath90 . at present
this seems to be a completely arbitrary number depending on which vacuum we are in and to be quite beyond calculation .
however for an extreme black hole , for example an electric majumdar - papapetrou black hole , we have a = 4 ( gm)^2 and moreover g m^2 q^2 where @xmath91 is the electric charge , ( not the supercharge ! ) . thus s= q^2 which does depend upon @xmath23 at all , only on the quantized dimensionless and essentially topological quantity @xmath91 .
the same feature may be shown to be true for all extreme black holes with non - vanishing entropy , both in four and five spacetime dimensions @xcite .
the entropy is independent of any accidential moduli fields characterizing the vacuum we are in but depends only on dimensionless quantized charges .
in fact this is a rather general statement and does not depend upon string theory in an essential way , just the general structure of the low energy effective lagrangians .
the reason that one is so interested in bps states in any theory is that they are typically protected against quantum corrections .
in particular the number of states in a super - multiplet is fixed by the bogomolnyi condition @xmath92 .
thus properties of bps states at weak coupling should remain true at strong coupling . in other words quantum mechanical string or strictly speaking d - brane calculations
should and indeed do agree with semi - classical calculations in supergravity theories .
this is particularly to be expected for calculations of the number of bps states , i.e. of the entropy of extreme black holes .
thus the basic idea is that counts the number @xmath93 of microstates of a certian bps configuration of d - branes having certain charges @xmath94 using techniques from conformal field theory and hence the entropy s_d = n. in fact no great sophistication is needed for these calculations , they simply involve a one - dimensional gas .
one now constructs a fully non - linear bps supergravity solution with the same charges representing an extreme black hole .
one calculates the area of the horizon and hence its bekenstein - hawking entropy @xmath95 .
then one compares and @xmath96 . in the limit of large charges one gets exact agreement , in other words the factor of proportionality is correct .
this distinguishes this calculation from almost every other similar non - stringy calculation in four or five dimensions .
the simplest example involves five - dimensional black holes which may be thought of as 1-brane lying inside a five - brane and carrying kaluza - klein momentum .
there are three charges , which count the number of one branes . the number of five - branes and the kaluza - klein momentum of the string .
more testing calculations can be done giving complete agreement with the emission rates , including grey - body factors , in the limit of low frequency for slightly non - extreme holes .
in fact the absorption cross - sections of black holes at zero frequency are universal and always given in terms of the area @xcite but at non - zero frequency this is no longer so but nevertheless the emission rates continue to agree . in fact these calculations may be extended to strongly non - extreme holes but perhaps not surprisingly some small discrepancies have emerged but these may quite plausibly be ascribed to extending the approximation beyond its reasonably expected range of validity .
i personally have found these results to be tremendously impressive and at face value they constitute good evidence for the ultimate promise of the string / m - theory project .
what then is m - theory ?
the letter m has variously been claimed to stand for membrane , magic , mystery or mother ( as in mother of all theories ) but since no - one at present has a definitive theory the question of the name should presumably remain in abeyance .
perhaps the snappiest characterization is .
m - theory is the strong coupling limit of type iia string theory .
however this is rather non - constructive ( in the pure mathematical sense ) ans something more concrete is desirable . to see why it is reasonable however ,
consider passing between this is done by compactification on a circle of radius @xmath97 @xmath98 eleven dimensional supergravity then gives rise to type iia supergravity with its dilaton @xmath90 coming form the matric component @xmath99 .
standard kaluaz - klein calculations lead to the relation r_10= l_planck < e^ 23 > where @xmath100 is the eleven dimensional planch length . now in string terms the string coupling constant @xmath101 is given by g_string= < e ^ > and = g_string^ 13 .
it follows @xcite that weak string coupling corresponds to a small string circle and strong coupling to a large circle . thus if one starts from type iia string theory and increases the coupling constant one expects to arrive at an effectively eleven dimensional theory .
the situation in the definition is summarized in the following table : [ cols="^,^,^",options="header " , ] one of the conjectured dualities relates m - theory to the @xmath31 heterotic theory compactified on a calabi yau manifold @xmath35 @xcite .
the basic idea is to consider ^10,1 = i cy where @xmath102 is a closed interval which may be thought of as a circle @xmath103 identified under reflection about a diameter .
the reflection has two fixed points at the ends of the interval .
thus the universe looks like a sandwich consisting of two ten dimensional sheets bounding an eleven - dimensional bulk spacetime .
one @xmath104 gauge field lives on one sheet and the second factor on the other sheet .
the size of the interval is taken to be rather larger than the size of the calabi - yau manifold .
this gives improved phenomonological models .
ihe separations of scales also suggests that one one might be able to study the era of compactification , if indeed compactification ever took place , about which almost nothing is known or conjectured , as a two step process in which one ignores the formation of the calabi - yau and considers a compactification from five to four spacetime dimensions on an interval .
in asking what kind of theory m - theory can be it is natural to ask what sort of superymetric p - branes it admits . to answer this question recall @xcite that in any spacetime dimension @xmath55
one may consider a modification of the superysymmetry algebra by adding something to the right hand side of the basic anti - commutation relation q_aq_b + q_bq_a = p_(c^ ) _ ab + z_ab since @xmath105 is an element of the the clifford algebra @xmath106 it may also be thought of as an element of the grassman algebra of all p - forms @xmath107 .
thus z_ab = _ p1 1p ! z__1
_ p ( c ) ^_1 _ p ) _ ab .
the bosonic quantities @xmath108}$ ] may be thought of in general non - lorentz - invariant charges per unit p - volume carried by p - branes states .
thus a p - branes lying in the @xmath109 plane is associated with the clifford algebra element @xmath110 . of course
if @xmath111 or @xmath112 we recover the usual lorentz - invariant electric or magnetic central charges we are familiar in four - dimensional supergravity theories which are carried by @xmath113-branes , i.e. ordinary particles .
the allowed values of @xmath114 are constrained by the condition that @xmath115 } ) _ { ab}$ ] be symmetric in its spinor indices . in eleven dimensions
this leaves just two possibilities , the algebra is q_aq_b + q_bq_a = p_(c^ ) _ ab + 12 z _ ( c^ ) _ ab + 15 ! z _
( c ^ ) _ ab , where * @xmath116 corresponds to the charge caried by the _
m-2-brane _ + and * @xmath117 corresponds to the charge carried by the _ m-5-brane_. one may desrcibe these in two ways * \(i ) as elementary objects with lagrangians analogous to the nambu - goto action for strings + or * \(ii ) bps solutions of the low energy supergravity theory admitting one half the maximum number of killing spinors .
the m-2-brane , i.e. the supersymmetric theory of maps from a 3-manifold @xmath118 into @xmath119 , the superspace of eleven dimensions has of course been known for some time . attempts to quantize it in the analogue of lightcone gauge used in superstring theory revealed that the residual gauge symmetry is @xmath120 , the group of volume preserving diffeomorphisms of the 2-brane spatial manifold .
this is the analogue of the two copies of @xmath121 one obtains in string theory .
in the final part of the lecture i want to turn to the only concrete proposal for what m - theory actually may be .
it has the merit of being simple to state but , as its authors would be the first to admit , there is much to be done to establish that the proposed theory actually exits and that it possesses all of the symmetries expected of it . * \(i ) take the ( unique ) supersymmetric yang - mills theory in ten dimensions with gauge group @xmath122 metioned earlier .
* \(ii ) dimensionally reduce form ten to 1 spacetime dimensions when it becomes a supersymmetric quantum mechanical model based on @xmath123 hermitian matrices . *
\(iii ) take the @xmath124 limit .
.5 cm the yang - mills action is , in the usual physicists notation so the trace is positive and we have inserted the conventional @xmath125 in front of the dirac action .
- 14 tr f^2 + + i 2 ^t ^(_- ia _ ) .
if everything depends only upon time one pick the gauge @xmath126 and the bosonic variables reduce to the 9 @xmath127 hermitian matrices @xmath128 which are renamed @xmath129 .
the resulting quantum mechanical models has as bosonic lagrangian _
i 1 2 tr x^2_i - v ( x^i ) where the potential energy function @xmath130 is given by v= _ i
< j tr ^2 note that the classical ground state corresponds to commuting coordinates @xmath129 but the full quantum mechanics involves _ non - commuting coordinates _ and is closely related to some of connes ideas about non - commutative geometry . here is not the place to give a full motivation for the matrix ansatz .
the approach of the originators was is via d-0-branes .
another , historically earlier approach is to regard it as a regularization of the supermembrane action in light - cone .
this is reasonable because of the as yet incompletely relationship between two lie algebras : sdiff(m_2 ) = _
i hope it is clear form what i have said that no - one , including the the most passionate advocates of the matrix model , believes that this is the final story .
there will no doubt be many dramatic twists and turns before the final shape of m - theory emerges .
however what does now seem clear is that sufficiently many of the essential underlying ideas are in place to show that some definite structure will finally emerge and that it will be able to answer many of the questions that we would like to ask of a quantum theory of gravity .
whether it is the theory that nature has adopted is a different matter .
my own assessment is the situation is that it is rather analogous to that which prevailed after the discovery ( if you are , to use the language of scholastic philosophers , a realist ) or invention ( if your are a nominalist ) of riemannian geometry .
this allowed the unification of the three non - euclidean geometries into a unified generalization .
combined with the basic ideas of group theory this rapidly led to lie groups considered as riemannian manifolds and the classification of symmetric spaces .
later there came complex and khaler geometries .
much of twentieth century mathematics has been preoccupied with these topics . however , although some speculative applications of riemannian geometry to cosmography were rapidly forthcoming , it was not until einstein s work on special relativity that the essential _ phyiscal insight _ that _ time _ must also be included in the picture and his later idea of the _ equivalence principle _ that general relativity finally emerged and these earlier speculations became firmly emebeded in our cosmological world picture .
it took even later for the development of quantum mechanics and the realization of the central role the khler geometry of the space of states to emerge together with the importance of of lie groups and their representations , leading eventually to gauge theory and our present standard model of particle interactions . today
we see that we have in hand the beginnings of a vast mathematical structure including in it all the mathematical ideas that we have found useful in physics in the past .
already it is clear that the first three basic underlying physical principles are experimental verification of supersymmetry in accelerators would be convincing evidence that the present chain of ideas is essentially correct , however we desperately need some more physical ideas .
let s hope that later day patent - clerk comes along soon !
as i have emphasised above countless people have contributed to our present understanding and i have had the privilege of hearing about these ideas directly from many of them .
in particular my collaborators and my colleagues in damtp , paul townsend , michael green , malcolm perry and george papapdopoulos have strongly influenced this account .
the reference list below is in no way intended to be comprehensive or to provide an an accurate historical record of what happened when and who did what .
that task must await future historians .
it is simply meant as a jumping - off point for those who wish to explore the subject further .
|
i review some of the recent progress in string / m - theory epsf = 0 = 0 = 0 = 0 # 1 # 2 # 3=#1 = # 2 # 1 # 2 # 3=#1 = # 2 by - by - # 1 = cmbx12 = cmbx12 scaled 1 = cmbx12 scaled 2 amssym.def amssym.tex
|
chaotic dynamical systems are known to exhibit typical random processes behaviour due to their strong sensitivity to initial conditions .
deterministic diffusion arises from the chaotic motion of systems whose dynamics is specified , and it should be distinguished from noise induced diffusion where the evolution is governed by probabilistic laws .
diffusive ( standard and anomalous ) behaviours have been observed in periodic chaotic maps ( see , _
e.g. _ , refs .
@xcite and references therein ) and in continuous time dynamical systems @xcite .
the analysis of deterministic diffusion is relevant for the study of non - equilibrium processes in statistical physics .
the major aim is to understand the relationship between the deterministic microscopic dynamics of a system and its stochastic macroscopic description ( think , for example , at the connection between lyapunov exponents , kolmogorov - sinai entropy and macroscopic transport coefficients , firstly highlighted by gaspard and nicolis @xcite ) . in the present paper
we introduce a model of deterministic diffusion in a one - dimensional lattice , suggested by a classical mechanics interpretation of the celebrated lorenz 1963 system @xcite .
the steady state chaotic dynamics of the lorenz system can indeed be recasted as the one - dimensional motion of a classical particle subjected to viscous damping in a past history - dependent potential field ( see refs .
@xcite , and ref .
@xcite for an earlier preliminar analysis ) .
we shortly recall that the ( scaled ) lorenz dynamical system is given by @xmath0 with @xmath1 . for @xmath2 , where @xmath3 , three fixed points exist : @xmath4 ( unstable ) and @xmath5 ( stable ) .
for @xmath6 all fixed points are unstable , and the lorenz system can exhibit either periodic or chaotic behaviour on a strange attractor set ( see , _
e.g. _ , ref .
@xcite for a comprehensive exposition on the subject matter ) .
+ in the steady state , the system can be reduced to a one - dimensional integro - differential equation for the @xmath7 coordinate @xcite @xmath8 _
\beta x\ , , % \frac{\upd^{2}x}{\updt ^{2}}+ % \eta \frac{\upd x}{\updt } + ( x^{2}-1)x=-\alpha [ x^{2}-1 ] _
% \beta x \label{eq.lorenz}\ ] ] where @xmath9 , @xmath10^{1/2}$ ] , @xmath11^{1/2}$ ] , and the time is scaled by a factor @xmath12^{1/2}$ ] with respect to the time coordinate in eqs . .
the square brackets in eq .
indicate the _ exponentially vanishing memory _
, which is defined , for any suitable time function @xmath13 , by @xmath14_k(t)\equiv k\int_0^\infty \upd s \,e^{-ks}f(t - s)\ , .\ ] ] according to eq .
, the lorenz system chaotic dynamics corresponds to a one dimensional motion in a constant - in - time quartic potential @xmath15 . in the presence of friction ( @xmath16 ) the motion
is sustained by a time dependent memory term , which takes into account the system past evolution .
although eq . has been deduced from the lorenz system , it can be generalized to a wider class of equations showing similar dynamical properties @xcite .
indeed , it can be usefully recasted in the form @xmath17_\beta
\}\phi'(x)=0 \ddot{x}+ \eta \dot{x}+ \ { q(x)+\alpha \left[q(x)\right]_\beta \}\phi'(x)=0\ ] ] where the prime indicates the derivative with respect to @xmath7 .
equation is obtained for @xmath18 and @xmath19 .
the generalized equation can be regarded as the description of the motion of a unit mass particle subjected to a viscous force @xmath20 and interacting with a potential field @xmath21 through a dynamically varying `` charge '' @xmath22_\beta$ ] .
this charge depends both on the instantaneous particle position @xmath23 and on the past history @xmath24 .
it is just the coupling of @xmath25_\beta$ ] with the fixed potential field @xmath21 the origin of an endogenous forcing term which can sustain the motion even in the presence of friction : the chaotic behaviour can actually arise from the synergy between this term and the viscosity .
+ moreover , one can easily verify that eq
. corresponds to the _ generalized lorenz system _ @xmath26\ , .
\end{array } \right.\ ] ] the specific lorenz model can thus be viewed as singled out from a quite general class of dynamical systems which can exhibit chaotic behaviour , their common essential property being an exponentially vanishing memory effect together with a viscous damping . in our previous works @xcite the main chaotic dynamical features of the original lorenz system
have been investigated through the analysis of the piecewise linear system corresponding to the choice @xmath27 and @xmath28 .
we have thus obtained the piecewise linear lorenz - like equation @xmath29_{\beta}\ } \mbox{sgn}(x)=0 % \frac{d^2 x}{dt^2}+\eta \frac{dx}{dt}+ % \{|x|-1 % + \alpha \left[|x|-1\right]_{\beta}\ } % \mbox{sgn}(x)=0.\ ] ] which has the great advantage of being exactly solvable on each side of @xmath30 .
if in eq .
( [ eq . generalizzata ] ) one substitutes the quantities @xmath31 and @xmath32 with @xmath7-periodic functions , the chaotic jumps between the two infinite wells of the original quartic potential @xmath33 correspond to chaotic jumps among near cells . the result is a deterministic diffusion in an infinite lattice , induced by a lorenz - like chaotic dynamics .
equation ( [ eq . generalizzata ] ) will be thus called the _ lorenz diffusion equation_. in order to use as far as possible analytical tools , we shall consider the unit wavelength periodic potential @xmath34 ( corresponding to @xmath35 and @xmath36 ) , where @xmath37 indicates the fractionary part of @xmath7 .
this potential field obviously consists of a lattice of truncated parabolae ( see fig . [
fig : reticolo ] ) . by simple substitution one easily derives the equation @xmath38_{\beta } \,=0\ , .
\ddot{x}+\eta \dot{x}+ \{x\}-\textstyle\frac{1}{2 } + \alpha [ \{x\}-\frac{1}{2}]_{\beta } \,=0\,\ ] ] where @xmath39 denotes the friction coefficient , @xmath40 is the memory amplitude and @xmath41 is related to the inertia whereby the system keeps memory of the past evolution .
note that @xmath39 and @xmath41 play a symmetrical role in the dynamics : the solution of eq .
is indeed left invariant if one changes @xmath41 with @xmath39 , while keeping @xmath42 constant . inside each potential cell eq .
can be recasted in a third - order linear differential form . indeed , by applying the operator @xmath43 to each side of eq .
, one obtains ( for @xmath44 ) @xmath45 it is worth observing that the nonlinearity of the original model is simply reduced to a change of sign of the forcing term @xmath46 when @xmath7 crosses the cell boundaries ( in our case the integer values ) .
as we will see , chaotic dynamics essentially results from the unpredictability of the crossing times .
+ partial solutions of the third order nonlinear differential equation can be easily calculated inside each open interval @xmath47 . to obtain a global solution ,
such partial solutions should be matched at @xmath48 by assuming that the position @xmath7 , the velocity @xmath49 and the memory @xmath50_\beta$ ] are continuous , whereas the acceleration @xmath51 turns out to be undefined .
however , it is easily shown that each pair of acceleration values `` immediately '' before and after the crossing times are related by @xmath52 .
+ the fixed points of eq .
are of course @xmath53 , @xmath54 , and their local stability depends on the roots of the characteristic equation @xmath55 all the fixed points are unstable when @xmath40 is larger than the critical value @xmath56 . in this case
. has one real negative root @xmath57 and a complex conjugate pair of roots with positive real part @xmath58 @xmath59 . for @xmath60 , the partial solution in the generic open interval @xmath47
can be finally written in the explicit form @xmath61 where the constants @xmath62 , @xmath63 , @xmath64 are linearly related to the ( cell by cell ) initial conditions .
the motion inside each cell consists of an amplified oscillation around a central point which translates towards the center of the cell .
a change of cell yields a discontinuous variation of the acceleration @xmath65 and consequently of the coefficients @xmath62 , @xmath63 , @xmath64 .
suppose that , at a given time , say @xmath66 , the particle enters the @xmath67th cell at its left boundary with positive velocity ( the reverse case can be symmetrically analyzed ) . in this case @xmath68 .
the question is now on whether the particle leaves the cell either from the left side ( i.e. @xmath48 ) or from the right side of the cell ( i.e. @xmath69 ) . assigned the model parameters @xmath70 , @xmath71 and @xmath72 , the minimum positive time @xmath73 such that @xmath74 depends of course on @xmath75 , and therefore on the initial conditions @xmath76
unfortunately , the direct problem is transcendent .
moreover , as shown in fig .
[ fig : t ] , its solution is strongly sensitive to the initial conditions .
this fact is a direct consequence of the crossing time definition : @xmath77 is indeed determined by the intersection of an amplified oscillation and a decreasing exponential .
a small change in the initial conditions may thus cause a discontinuous variation of the crossing time .
this is the very origin of the system chaotic dynamics which suggests a stochastic treatment of the lorenz diffusion equation .
despite the fact we have a three - dimensional space of parameters to investigate the model behaviors , the interesting region is actually a limited portion .
this easily follows from the following simple considerations .
for @xmath39 large enough in eq . , the motion rapidly stops in one of the lattice fixed points . in order to have non trivial solutions ,
the viscous coefficient must be smaller than a maximum value which can be explicitly derived from the condition @xmath78 @xmath79\ , .\ ] ] in the opposite limit , if @xmath39 is too small , the friction term is negligible with respect to the memory term , and the resulting motion is `` ballistic '' , _
i.e. _ , @xmath80 ^ 2\right > \sim
analogous considerations can be repeated for @xmath40 and @xmath41 . inside this region of parameters the lorenz diffusion equation generates a wide variety of behaviours : as we will see , the observed regimens are strongly sensitive to the control parameters , and furthermore this dependence is often in contrast with the intuitive meaning of @xmath40 , @xmath41 , @xmath39 . in fig .
[ regimi ] a few numerical simulation of eq . are shown , corresponding to different values of the parameters .
these examples suggest that the variety of motion regimens ranges from `` ballistic '' ones to clearly `` diffusive '' , and even `` trapped '' in one cell or in groups of nearby cells .
the analysis of the motion in the elementary cell shows that the system rapidly reaches a steady - state . in
the diffusive regimens the points corresponding to subsequent cell entering conditions @xmath81 , @xmath82 are quickly attracted on a particular locus of the plane ( fig .
[ fig : mappe ] ) . in the general situation
this attracting set does not define a univocal map between the entering velocity and acceleration .
however , it is not difficult to see that , for large @xmath73 , @xmath83 and @xmath84 satisfy the piecewise linear relation @xcite @xmath85 if the diffusive regimen admits large enough times of permanence , the couple of straight lines defined by eq
. therefore belongs to the entering condition attracting set ( fig .
[ fig : mappe ] ) .
moreover , in the limiting case of very large @xmath70 the attracting set exactly reduces to such lines , and in the steady state eq .
is satisfied by all entering conditions @xcite .
the macroscopic evolution of the system can be statistically analyzed by introducing the diffusion coefficient @xmath86 ^ 2\right>/(2t)\,.$ ] diffusive regimens are identified by a finite value of @xmath87 . as shown in fig .
[ regimi ] , different values of the control parameters can lead to qualitatively very different diffusive motions .
what is surprising is that the change of regimen is very irregular and abrupt with @xmath87 , and sometimes it does not reflect our intuition on the role played by the parameters in eq . .
for instance , @xmath87 shows a non - monotonic dependence on the viscosity @xmath39 ( fig .
[ rison ] ) , and it is possible to observe transitions from trapped ( or diffusive ) motion to ballistic ones even by increasing the viscosity itself ( table [ table : d ] ) . despite the somehow strange aspect of some system trajectories , it should be stressed that no anomalies ( _ i.e. _ superdiffusive transport ) have been encountered in the statistical analysis of diffusion , the distribution of long flies having always exponential tails .
@xmath88 & ballistic&1.425 - 1.725 & trapped + @xmath89 & trapped&1.725 - 1.975 & diffusive + @xmath90 & ballistic&1.975 - 2.050 & trapped + @xmath91 & diffusive&2.050 - 2.325 & diffusive + @xmath92 & trapped&@xmath93 & trapped + @xmath94 & diffusive & & + to conclude , we have presented a deterministic system which can generate large - scale diffusive transport induced by a lorenz - like microscopic chaotic dynamics .
two main problems should still be tackled .
first , the diffusion coefficient seems to have a fractal dependence on the control parameters , as observed for other discrete chaotic maps @xcite .
this fact requires more investigations , possibly using the analytical tools developed in ref .
second , it would be interesting to consider a two - dimensional extension of the lorenz diffusion equation , mainly to investigate the possible emergence of anomalous diffusion induced by the increased spatial dimension .
the authors would like to thank m. la camera for useful suggestions .
d. v. was supported by grants of the university of nice and of the university of genova .
this work has been partially supported by the infm project gepaigg01 and cofin 2001 , prot .
|
the lorenz 1963 dynamical system is known to reduce in the steady state to a one - dimensional motion of a classical particle subjected to viscous damping in a past history - dependent potential field .
if the potential field is substituted by a periodic function of the position , the resulting system shows a rich dynamics where ( standard ) diffusive behaviours , ballistc motions and trapping take place by varying the model control parameters .
this system permits to highlight the intimate relation between chaos and long - time deterministic diffusion .
|
experimental techniques for manipulating ultracold atoms have made great progress since the successful realization of atomic bose - einstein condensates ( becs ) @xcite .
optical lattices , formed by a standing wave of laser light , are providing the ideal stages for an experimental investigation of the fundamental many - body problems in condensed matter physics via ultracold atomic gases @xcite .
a recent series of experimental studies on bosonic mott - insulators @xcite have clearly demonstrated this feature , namely the realization of quantum simulators .
a noteworthy advantage of atomic gases over condensed matter such as solids or liquids is that their experimental parameters are highly controllable .
both density and temperature of atomic gases are fully controlled in the process of evaporative cooling @xcite .
the depth of the optical lattice potential is controlled by the light intensity , which leads to the precise alternation of atom tunneling between adjacent sites @xcite .
furthermore , we can create one- , two- , and three - dimensional optical lattices depending on the laser beam configuration with the use of interference effects @xcite .
even the interactions between the atoms can be widely tuned with the help of feshbach resonances @xcite .
therefore , in the experiments , it is possible to precisely change the important system parameters such as density , temperature , tunneling , dimensionality of lattices , interaction strength and so forth .
ultracold atoms in optical lattices allow us to investigate the quantum many - body effects in a well - controlled manner .
experiments with quantum degenerate fermionic gases have already clarified the intriguing physics of such phenomena as a molecular bec @xcite and the crossover between a bec and a bardeen - cooper - schrieffer ( bcs ) superfluid @xcite .
ultracold fermionic atoms trapped in optical lattices are now stimulating both experimental and theoretical interest .
this system is well captured by the hubbard model hamiltonian @xcite , which is widely discussed with respect to strongly correlated electron systems .
several theoretical analyses have predicted the possibility of the mott metal - insulator transition ( mmit ) in this system @xcite . although experimental investigations are still under way @xcite , the recent demonstration of a fermionic band insulator in three - dimensional optical lattices @xcite has convinced us that mmit will be observed in near future
. a precise comparison of the theory and experiments will provide us with a deeper understanding of the quantum phase transition in fermionic systems .
the _ superlattice _ geometry of optical lattices , i.e. , optical superlattice , was successfully demonstrated as an advanced manipulation technique in a recent report @xcite .
this technique has been introduced to control the distribution of atoms on lattices more functionally and will play a key role in the implementation of quantum computing with ultracold atoms @xcite . on the other hand , in condensed matter physics , it is known that lattice distortions generate a variety of many - body effects in strongly correlated electron systems such as the charge - transfer organic materials and ferroelectric perovskites @xcite .
the ultracold fermionic atoms in optical superlattices @xcite can be regarded as a quantum simulator for these materials .
our present study is highly motivated by these important potentialities of optical superlattices . in this paper
, we investigate the ground state properties of s=1/2 fermions trapped in a one - dimensional optical superlattice with 2-site periodicity . the external harmonic confinement inherent in the real experiments @xcite is further considered in our model .
we show that a new coexistence phase appears consisting of three distinct insulating regions caused by the superlattice potentials , which is in contrast to the ordinary hubbard model in the optical lattice .
this is elucidated via a detailed analysis of the local density of atoms and spin correlations by using the density matrix renormalization group method .
furthermore , we find unique spin correlations between the insulating regions when these regions are separated by a metallic region .
this paper is organized as follows . in the next section we briefly mention the model and the method .
we then show the density profiles of fermions and several density / spin correlation functions to clarify the zero temperature properties .
the final section provides a brief summary .
[ sec : mod ] is the difference between the potential energies of odd and even sites . ]
we consider a gas of fermionic atoms embedded in an optical superlattice with an alternating superlattice potential and a confining parabolic potential ( see fig .
1(b ) ) , which is described by the hubbard hamiltonian , @xmath0 where @xmath1 , @xmath2 and @xmath3 are respectively the fermion creation , annihilation and number operators relevant to the site labeled @xmath4 with spin @xmath5 . here
, @xmath6 is the on - site repulsion ( @xmath7 ) , @xmath8 the curvature of the parabolic confining potential , @xmath9 the local potential at site @xmath4 and @xmath10 the hopping amplitude between nearest neighbor sites .
we define the local potential @xmath9 with 2-site periodicity as @xmath11 where @xmath12 is the difference between the potential energies of odd and even sites ( see fig .
we denote the total number of fermions ( lattice sites ) as @xmath13 ( @xmath14 ) .
there are several kinds of numerical methods that can be employed with our fermionic superlattice hubbard model .
the numerical diagonalization method can be applied straightforwardly to fermionic models , but it restricts our analysis to rather small systems with a typical lattice size of 20 sites , which is not large enough for our purpose .
the quantum monte carlo simulations method is an efficient way to deal with quantum systems at finite temperatures , and this approach has been applied successfully to bosonic hubbard models to elucidate the transition between the superfluid and mott insulating phases in optical lattices @xcite .
this method , however , suffers from negative - sign problems in fermionic cases , which makes it difficult to obtain reliable results at low temperatures .
the density matrix renormalization group ( dmrg)@xcite is another powerful numerical method with which we investigate quantum systems such as fermionic models with extremely high accuracy , although it is restricted to one - dimensional cases .
this technique allows us to deal with a very large system as required for the analysis of optical lattices . in this paper
we therefore exploit a variant of the dmrg ( the so - called finite - system dmrg ) @xcite to study our fermionic hubbard model in an optical superlattice at zero temperature .
the essence of the idea is that we first diagonalize a given small system , and then enlarge the system size step by step via a renormalization procedure by retaining relevant low - energy states which have large eigenvalues of the density matrix .
this simple procedure is known to give precise results for one - dimensional quantum systems .
this method has already been reported in detail @xcite . in our nonuniform model with harmonic confinement
, the numerical calculation can be done in parallel to the ordinary case without confinement .
the only point we have to pay attention to is that the superlattice potential should be properly included in each renormalization step .
we restrict our discussions to the singlet ground state with an even number of fermions , which are tractable by the dmrg method @xcite . as far as the local density and its variance
are concerned ( see sec.iii a and b ) , the even - number case is sufficient to discuss the characteristic properties .
a relevant difference between the even and odd cases may appear in the spin correlations ( sec.iii c ) since a free s=1/2 spin remains unscreened in the system for the odd case .
the free spin is not important physically since it does not affect the global nature of the system , but rather acts as a magnetic impurity .
in our optical lattice , the spatial distribution of atoms is inhomogeneous because of the confining harmonic potential .
we thus consider the profile of the local density , i.e. the spatial distribution of @xmath15 , to be @xmath16 in fig .
[ fig : local](a ) , we show the density profile for @xmath17 fermions trapped in a lattice with the linear size @xmath18 , where the other parameters are @xmath19 , @xmath20 and @xmath21 .
it is seen in fig .
[ fig : local](a ) that the amplitude of the local density alternates between even and odd sites reflecting the 2-site periodic potential .
when there is no superlattice structure , namely @xmath12=0 , it is known that a mott insulating region appears characterized by plateau formation at @xmath22 @xcite . in the present case , however , the density profile reveals more complicated sawtoothed structures , from which we can not immediately distinguish metallic and insulating regions . to characterize the insulating region clearly
, we therefore introduce the average number of atoms in the unit cell as , @xmath23 in fig . [
fig : local](b ) , we show the @xmath24 value obtained by averaging the data of ( a ) in the unit cell .
plateaus ( or shoulders ) appear at @xmath24 = 0.5 , 1.0 , and 1.5 , which characterize three distinct insulating regions . , @xmath25 , @xmath26 , @xmath27 and @xmath28 .
the filled circles denote the local density @xmath15 and the solid lines are a guide for the eye .
( b ) shows the density profile , @xmath24 , averaged within the unit cell .
there are plateaus at @xmath29 and 1.5 . ] and @xmath20 : ( a ) density of atoms , @xmath30 , as a function of the chemical potential @xmath31 .
the plateaus , which are formed clearly at @xmath30=0.5 , 1.0 , and 1.5 , imply the existence of well - defined insulating phases .
( b-1 ) and ( b-2 ) are schematic diagrams of the insulating phases for @xmath32 and @xmath33 . ] to understand the above characteristic behavior in @xmath24 , it is instructive to observe the bulk properties of the system without harmonic confinement , which are shown in fig .
figure [ fig : dp](a ) shows the infinite dmrg results for filling factor @xmath30 as a function of chemical potential @xmath34 .
we can see the formation of well - defined plateaus at @xmath350.5 , 1.0 , and 1.5 , which clearly feature the insulating phases separated by the metallic phases .
the insulating phases with @xmath350.5 and 1.5 are caused by the superlattice potential with 2-site periodicity .
the corresponding spatial distribution of atoms is schematically illustrated in fig .
[ fig : dp](b ) . when @xmath32 , as shown in fig .
[ fig : dp](b-1 ) , the system is in a variant of the mott insulating phase where each site with lower ( higher ) potential @xmath9 accommodates one ( no ) atom . on the other hand ,
when @xmath36 , as shown in fig .
[ fig : dp](b-2 ) , the system is another mott type insulator where each site with higher potential accommodates one atom , while that with lower potential is almost fully occupied by two atoms .
the insulating phase with @xmath37 changes its nature depending on the interaction strength @xmath6 in a slightly complicated way , from the band insulator ( small @xmath6 ) to the mott insulator ( large @xmath6 ) .
actually this problem has been extensively discussed so far , which is briefly summarized here .
early studies by the numerical @xcite and weak - coupling bosonization methods @xcite suggested that a single phase transition occurs from the band insulator to the mott insulator when @xmath6 is varied at half filling . on the other hand , fabrizio _ et al .
_ predicted a new intermediate insulating phase so - called bond - charge density wave ( bcdw ) phase " between them @xcite . since their proposal ,
extensive investigations have been done to confirm it .
quantum monte carlo simulations @xcite supported the existence of the intermediate phase . by applying the method of topological transitions in spin and charge berry phases , torio _
et al . _ presented a ground - state phase diagram , which is consistent with the scenario of fabrizio _ et al .
_ @xcite the dmrg calculations @xcite have been performed by several groups .
for example , the calculation of various structure factors supported the existence of intermediate phase @xcite though kampf _
_ @xcite were not able to resolve two transitions clearly .
recent calculations based on the dmrg @xcite and the level - spectroscopy method @xcite with high accuracy have confirmed the existence of the intermediate bcdw phase .
the above results for the homogeneous system show that the plateau regions at @xmath24 = 0.5 , 1.0 , and 1.5 in the system with confining potential have the above insulating properties , which are sandwiched by metallic regions . , along the trap for different fillings .
the flat terraces are the insulating regions .
the parameters are @xmath38 , @xmath39 , @xmath28 , and @xmath40 . in ( a )
, the number of atoms is changed from 20 to 104 , while in ( b ) the number is fixed as @xmath13 = 20 , 40 , 60 , and 90 . ]
let us now observe how the density profile of the system changes when the atom number @xmath13 is changed systematically . in fig .
[ average](a ) , we show the average density of atoms , @xmath24 , when @xmath13 is changed continuously .
it is seen that the system enters several distinct ground states as @xmath13 increases . in fig .
[ average](b ) , for reference , we show cross - sectional views of ( a ) by selecting several typical @xmath13 values . for @xmath41
, there is an insulating region with @xmath42 in the middle of the system .
just bedside this region , there are narrow metallic region windows with @xmath43 , which smoothly continue to the vacant - atom region @xmath44 .
the system is therefore in a phase where metals and insulators coexist , as is the case for an ordinary optical lattice without superlattice potentials @xcite . for @xmath45 ,
a metallic region with @xmath46 appears in the middle of the system , so that the insulating regions with @xmath42 are now sandwiched by the two metallic regions with @xmath43 and @xmath46 . for @xmath17 ,
another insulating phase with @xmath47 emerges in the middle region , and a further increase in the atom number , as seen when @xmath48 , results in an insulating region where @xmath49 .
a brief comment on the finite - size effect is in order here . since we are dealing with the rather small system with 60 fermions , the resulting metallic regions are very narrow , and the density profile changes very sharply there
this is due to the finite - size effect of our small system .
if we consider a system with larger number of fermions , the metallic regions become wider and the density profile changes more smoothly . in order to characterize the above insulating regions , we further investigate local density fluctuations in our superlattice system . to this end
, we introduce the variance of local density fluctuations , @xmath50 , at each site as @xmath51 in ordinary one - dimensional optical lattices , it is known that @xmath50 is suppressed and forms a plateau in the insulating region , while it has larger values in the metallic region @xcite .
this quantity , therefore , can describe the way in which insulator - metal crossover occurs when the on - site repulsion increases . figure [ fig : many ] shows the local density @xmath15 , the average density @xmath24 , the variance @xmath50 and the double - occupation probability @xmath52 for three different choices of @xmath53 with the other parameters unchanged . it is convenient to discuss characteristic properties by specifying each region in terms of the average local density @xmath24 .
let us first observe the insulating region characterized by @xmath47 , which is located around the site @xmath54 ( or equivalently @xmath55 ) in all three cases in fig .
[ fig : many ] . for @xmath56
, we can see the sawtoothed oscillation both in @xmath15 and @xmath52 since the system is in the band insulator where almost doubly - occupied sites and vacant sites are realized alternately .
note that even in this case , the variance @xmath50 forms a plateau without such alternations , implying that density fluctuations occur locally only via a process between two adjacent sites . as the interaction strength @xmath6 increases , @xmath15 ( @xmath52 ) continuously approaches @xmath57 ( @xmath58 ) .
on the other hand , @xmath50 is once enhanced ( @xmath59 , fig .
[ fig : many](b ) ) , and then suppressed ( @xmath60 , fig .
[ fig : many](c ) ) . therefore , we naturally expect that there may be another insulating state , which is characterized by an enhanced @xmath50 , between the band insulating state ( small @xmath6 ) and the mott insulating state ( large @xmath6 ) .
as mentioned above , in a homogeneous system without confinement , at half filling ( corresponding to @xmath61 ) , the bcdw intermediate phase appears between the band - insulating phase and the mott insulating phase @xcite . in the bcdw state
, there is rather large hopping within a coupled dimer , which gives rise to enhanced local density fluctuations .
therefore , we can distinguish the bcdw state from the other insulating states in terms of the variance @xmath50 . summarizing the above results
, we can say that the region with @xmath47 is either in the band insulating state , the bcdw state or the mott insulating state depending on the strength of the on - site repulsion .
these states are smoothly connected to each other with the increase in @xmath6 ( crossover behavior ) in contrast to the phase transitions realized in the homogeneous case .
we next focus on the region around the site @xmath62 , where the average density @xmath47 at @xmath56 is changed to @xmath42 at @xmath59 and @xmath63 . in this case ,
the local density @xmath15 is reduced from 2 to 1 at each odd site with the increase in @xmath6 , and @xmath50 and @xmath52 decrease monotonically to zero ( even sites are always empty ) .
namely , the system is smoothly driven from the band insulator to the mott insulator . finally , we look at the region around @xmath64 , where @xmath24 is slightly larger than 1.5 at @xmath56 without plateau structures , which is characteristic of the metallic region .
in fact , @xmath50 exhibits strong even - odd dependence on the site index . at @xmath59
, we see the plateau formation at @xmath49 , where @xmath65 and @xmath66 for even sites . therefore , the region enters the insulating region with mott ( odd sites ) and band - insulating ( even sites ) characteristics . a further increase in @xmath6
again drives the mott insulating state to the metallic state characterized by @xmath67 .
we have repeated similar calculations for various choices of the model parameters . among them
we have presented only the three selected cases above , since they capture most of the essential properties realized in our optical superlattice with 2-site periodicity . finally , we discuss local and spatially - extended spin correlations in the system . in fig .
[ fig : ss](a)-(c ) , we show the computed results for the variance of local spin fluctuations @xmath68 together with the density profile @xmath69 and the average profile @xmath24 for the system with @xmath70 and @xmath71 . also shown in fig .
[ fig : ss-1](a)-(c ) is the spin correlation function @xmath72 between the @xmath4th and @xmath73th sites with the latter fixed as three typical positions , @xmath74 , 45 , and 55 .
together with ( b ) @xmath75 and ( c ) @xmath24 .
the data are obtained for the system with @xmath70 , @xmath71 .
the other parameters are fixed as @xmath18 , @xmath17 , @xmath21 . ] as clearly seen from fig .
[ fig : ss ] , the local moment is well developed ( @xmath76 ) when a given site is located in the mott insulating region , while it is suppressed ( @xmath77 ) in the band - insulating region as well as in the empty region .
these characteristic properties are in accordance with what we naively expect from the homogeneous case without harmonic confinement .
let us now look briefly at the spin correlation function . in fig .
[ fig : ss-1](a ) , the @xmath4-dependence of the spin correlation @xmath72 is shown where the site @xmath74 is fixed in the insulating region with @xmath42 .
it is seen that the spin correlation exhibits an antiferromagnetic nature within the same region , and quickly decays when the site @xmath4 enters the metallic region .
interestingly , the spin correlation increases slightly again around @xmath78 belonging to another insulating region with @xmath42 .
this unusual behavior is discussed below in detail by using a more significant example . on the other hand , for @xmath79 in the insulating region with @xmath47
, we can not see the long - range spin correlation as in the @xmath74 case , since this mott insulating region is not well stabilized as regards the choice of parameters . for @xmath80 in fig .
[ fig : ss-1](c ) , the spin correlation shows typical behavior inherent in the metallic state : a small spin correlation amplitude , which immediately decreases with distance . th and @xmath73th sites , @xmath72 , where the site @xmath73 is fixed as ( a)@xmath74 , ( b)45 and ( c)55 .
the data are obtained for the system with @xmath70 , @xmath71 .
the other parameters are fixed as @xmath18 , @xmath17 , @xmath21 . ] in figs .
[ fig : ss2 ] and [ fig : ss2 - 1 ] , we show the results obtained when we chose a slightly larger interaction strength @xmath6 . in this case
also , the spin correlation shows similar behavior to that in fig .
[ fig : ss-1 ] .
there are two points to be mentioned here .
since in this case the local moment is well developed in the mott insulating region , as seen from fig .
[ fig : ss2](a ) , the spin correlation shows more significant antiferromagnetic correlations . even for site @xmath73 belonging to the metallic region , we can observe slightly enhanced antiferromagnetic correlations ( fig . [
fig : ss2 - 1](c ) ) .
another important point to be noticed is that the spin correlations in fig .
[ fig : ss2 - 1](a ) are strongly enhanced around @xmath78 , although they are suppressed in the region of @xmath81 ( c.f . fig .
[ fig : ss-1](a ) ) .
this characteristic behavior is a new feature that has not been observed in homogeneous systems without confinement .
this is closely related to the formation of several distinct regions . in the mott insulating region ,
a finite number of local spins are well developed , which are sensitive to effective magnetic fields .
such spins belonging to the different mott regions can have enhanced correlations even if they are separated by the metallic phases .
this is why we have encountered enhanced spin correlations between @xmath74 and @xmath78 in fig .
[ fig : ss2 - 1](a ) .
it should be noted that this effect is particularly significant if there is an odd number of spins in the mott region .
for example , in figs .
[ fig : ss-1](a ) and [ fig : ss2 - 1](a ) , 3 sites around @xmath74 and @xmath78 belong to the well - defined mott regions , which indeed exhibit enhanced spin correlations .
therefore , the above phenomenon is related to the finite - size effect of the mott region , so that such behavior should be somewhat obscured when the mott regions become large .
it is thus interesting to know whether such behavior can really be observed by future experiments in the optical lattices .
we have studied the characteristic properties of fermions trapped in a one - dimensional optical superlattice with 2-site periodicity .
we found that plateau regions appear in the profiles of the average number of atoms in the unit cell @xmath24=0.5 , 1.0 , and 1.5 .
in contrast to the ordinary hubbard model in the optical lattice , new insulating regions appear with @xmath820.5 and 1.5 caused by the superlattice potentials . furthermore , it has been shown by the variance and double - occupation probability results that three different insulating regions of band- , bcdw- and mott - type emerge in the insulating region with @xmath47 , which are smoothly connected to each other via a crossover as the strength of the on - site repulsion @xmath6 is altered .
we have also found that it is possible to enhance spin correlations between different mott regions , even if they are once suppressed in the intermediate metallic region .
this interesting phenomenon is due to the finite - size effect of the insulating regions , which are naturally formed by harmonic confinement .
specifically , it occurs more significantly when an effective number of sites in the mott insulating region is odd . since systematic experimental studies on the one - dimensional optical superlattice
may be soon within our reach , we hope that the characteristic properties discussed for cold fermions in this paper will be observed in the near future .
the numerical computations were carried out at the supercomputer center , the institute for solid state physics , university of tokyo .
this work was supported by grant - in - aid for scientific research on priority areas ( grant no .
18043017 ) from the ministry of education , culture , sports , science and technology of japan . for a review ,
see nature ( london ) * 416 * , 205 - 246 ( 2002 ) . i. bloch and m. greiner , in _ advances in atomic , molecular , and optical physics _ ,
editied by p. berman and c. lin ( academic press , new york , 2005 ) , vol .
52 , p. 1 .
i. bloch , nature phys .
* 1 * , 23 ( 2005 ) .
d. jaksch and p. zoller , ann .
( ny ) * 315 * , 52 ( 2005 ) .
o. morsch and m. oberthaler , rev .
phys . * 78 * , 179 ( 2006 ) .
m. greiner , o. mandel , t. esslinger , t. w. h@xmath83nsch , and i. bloch , nature ( london ) * 415 * , 39 ( 2002 ) .
o. mandel , m. greiner , a. wldera , t. rom , t. w. h@xmath83nsch , and i. bloch , nature ( london ) * 425 * , 937 ( 2003 ) .
t. st@xmath84ferle , h. moritz , c. schori , m. k@xmath84hl , and t. esslinger , phys .
lett . * 92 * , 130403 ( 2004 ) .
b. paredes , a. widera , v. murg , o. mandel , s. f@xmath84lling , i. cirac , g. v. shlyapnikov , t. w. h@xmath83nsch , and i. bloch , nature ( london ) * 429 * , 277 ( 2004 ) . s. f@xmath84lling , f. gerbier , a. widera , o. mandel , t. gericke , and i. bloch , nature ( london ) * 434 * , 481 ( 2005 ) . f. gerbier , a. widera , s. f@xmath84lling , o. mandel , t. gericke , and i. bloch , phys . rev . lett . * 95 * , 050404 ( 2005 ) .
f. gerbier , s. f@xmath84lling , a. widera , o. mandel , and i. bloch , phys .
* 96 * , 090401 ( 2006 ) .
s. inouye , m. r. andrews , j. stenger , h. -j .
miesner , d. m. stamper - kurn , and w. ketterle , nature ( london ) * 392 * , 151 ( 1998 ) .
m. greiner , c. a. regal , and d. s. jin , nature ( london ) * 426 * , 537 ( 2003 )
. s. jochim , m. bartenstein , a. altmeyer , g. hendl , s. riedl , c. chin , j. h. denschlag , and r. grimm , science * 302 * , 2101 ( 2003 ) . m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , s. gupta , z. hadzibabic , and w. ketterle , phys .
lett . * 91 * , 250401 ( 2006 ) . c. a. regal , m. greiner , and d. s. jin , phys .
* 92 * , 040403 ( 2004 ) .
m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
lett . * 92 * , 120403 ( 2004 ) .
m. bartenstein , a. altmeyer , s. riedl , s. jochim , c. chin , j. h. denschlag , and r. grimm , phys .
* 92 * , 120401 ( 2004 ) . t. bourdel , l. khaykovich , j. cubizolles , j. zhang , f. chevy , m. teichmann , l. tarruell , s. j. j. m. f. kokkelmans , and c. salomon , phys .
lett . * 93 * , 050401 ( 2004 ) .
m. rigol , a. muramatsu , g. g. batrouni , and r. t. scalettar , phys .
lett . * 91 * , 130403 ( 2003 ) .
m. rigol and a. muramatsu , phys .
a * 69 * , 053612 ( 2004 ) .
m. machida , s. yamada , y. ohashi , and h. matsumoto , phys .
rev . lett . * 93 * , 200402 ( 2004 ) ; m. rigol , s. r. manmana , a. muramatsu , r. t. scalettar , r. r. p. singh , and s. wessel , phys .
* 95 * , 218901 ( 2005 ) ; m. machida , s. yamada , y. ohashi , and h. matsumoto , phys .
* 95 * , 218902 ( 2005 ) .
liu , p. d. drummond , and h. hu , phys .
* 94 * , 136406 ( 2005 )
. v. l. campo , jr . and k. capelle , phys . rev .
a * 72 * , 061602(r ) ( 2005 ) .
x. gao , m. polini , m. p. tosi , v. l. campo , jr . , k. capelle , and m. rigol , phys .
b * 73 * , 165120 ( 2006 ) .
g. modugno , f. ferlaino , r. heidemann , g. roati , and m. inguscio , phys .
a * 68 * , 011601(r ) ( 2003 ) .
h. ott , e. de mirandes , f. ferlaino , g. roati , g. modugno , and m. inguscio , phys .
lett . * 92 * , 160601 ( 2004 ) .
m. k@xmath84hl , h. moritz , t. st@xmath84ferle , k. g@xmath85nter , and t. esslinger , phys .
lett . * 94 * , 080403 ( 2005 ) .
t. st@xmath84ferle , h. moritz , k. g@xmath85nter , m. k@xmath84hl , and t. esslinger , phys .
lett . * 96 * , 030401 ( 2006 ) .
j. hubbard and j. b. torrance , phys .
* 47 * , 1750 ( 1981 ) .
n. nagaosa and j. takimoto , j. phys .
. jpn . * 55 * , 2735 ( 1986 ) .
t. egami , s. ishihara and m. tachiki , science * 261 * , 1307 ( 1993 ) .
s. ishihara , t. egami and m. tachiki , phys .
b * 49 * , 8944 ( 1994 )
. r. resta and s. sorella , phys .
rev . lett . * 74 * , 4738 ( 1995 ) ; * 82 * , 370 ( 1999 ) .
m. tsuchiizu and y. suzumura , j. phys .
jpn . * 68 * , 3966 ( 1999 ) .
m. fabrizio , a. o. gogolin , and a. a. nersesyan , phys .
lett . * 83 * , 2014 ( 1999 ) .
t. wilkens and r. m. martin , phys .
b * 63 * , 235108 ( 2001 ) .
m. e. torio , a. a. aligia , and h. a. ceccatto , phys .
b * 64 * , 121105 ( 2001 ) .
y. takada and m. kido , j. phys .
. jpn . * 70 * , 21 ( 2001 ) .
j. lou , s. qin , t. xiang , c. chen , g .- s .
tian , and z. su , phys .
b * 68 * , 045110 ( 2003 ) .
y. z. zhang , c. q. wu , and h. q. lin , phys .
b * 67 * , 205109 ( 2003 ) .
a. p. kampf , m. sekania , g. i. japaridze , and p. brune , j. phys . : condens .
matter * 15 * , 5895 ( 2003 ) .
s. r. manmana , v. meden , r. m. noack , and k. sch@xmath84nhammer , phys .
b * 70 * , 155115 ( 2004 ) .
h. otsuka and m. nakamura , phys .
b * 71 * , 155105 ( 2005 ) .
g. g. batrouni , v. rousseau , r. t. scalettar , m. rigol , a. muramatsu , p. j. h. denteneer , and m. troyer , phys .
lett . * 89 * , 117203 ( 2002 ) .
v. a. kashurnikov , n. v. prokofev , and b. v. svistunov , phys .
a * 66 * , 031601(r ) ( 2002 ) .
s. wessel , f. alet , m. troyer , and g. g. batrouni , phys .
a * 70 * , 053615 ( 2004 ) .
|
we study the ground - state properties of spin-1/2 fermionic atoms confined in a one - dimensional optical superlattice with harmonic confinement by using the density matrix renormalization group method . for this purpose
, we consider an ionic hubbard model that has superlattice potentials with 2-site periodicity .
we find that several different types of insulating regimes coexist even if the number of atoms at each site is not an integer , but its average within the unit cell is an integer or half integer .
this is contrasted to the coexisting phase of the metallic and mott - insulating regimes known for the ordinary hubbard model in an optical lattice .
the phase characteristics are elucidated by investigating the profiles of the atom density , the local density / spin fluctuations , the double occupation probability and the spin correlations in detail .
|
an important role in noncommutative geometry @xcite is played by the set of compact operators whose partial sums of singular values are logarithmically divergent .
this set can be adequately described using the terminology of marcinkiewicz spaces .
consider the marcinkiewicz sequence space @xmath0 where we set @xmath1 here , @xmath2 is the sequence @xmath3 rearranged in nonincreasing order .
fix an infinite - dimensional separable complex hilbert space @xmath4 and consider the set @xmath5 of all compact operators @xmath6 on @xmath4 such that the sequence of its singular values @xmath7 falls into the space @xmath8 ( recall that the singular values of a compact operator @xmath9 are the eigenvalues of the operator @xmath10 ) .
we set @xmath11 it is well known that the ideal of compact operators @xmath5 equipped with the norm @xmath12 is a banach space .
we refer to the recent paper @xcite by pietsch for additional references and information on these spaces .
we describe briefly a construction of singular traces on the ideal @xmath5 due to dixmier @xcite and its various modifications which are of importance in noncommutative geometry @xcite . for a more detailed treatment
we refer to @xcite .
let @xmath13 , @xmath14 be the operator on @xmath15 defined by @xmath16 let @xmath17 be a @xmath13-invariant generalised limit on @xmath18 that is , @xmath17 is a positive normalised functional on @xmath15 such that @xmath19 for all @xmath20 and such that @xmath21 where @xmath22 is the subspace of all vanishing sequences .
for an element @xmath23 we set @xmath24 it is well known ( see e.g. 5 in @xcite and additional references therein ) that @xmath25 is an additive functional on the positive part of @xmath26 thus , @xmath25 admits a linear extension to a unitarily invariant functional ( trace ) on @xmath26 this trace vanishes on all finite - dimensional operators from @xmath27 such singular traces are called _ dixmier traces _
( see @xcite ) .
a smaller subclass of dixmier traces was introduced by connes in @xcite by observing that a functional @xmath28 is @xmath13-invariant state on @xmath29 for all @xmath30 here , @xmath31 is an arbitrary generalised limit on the space @xmath32 of all bounded measurable functions and the operator @xmath33 is a cesaro operator defined by the formula @xmath34 referring to @xmath17 above as a functional on @xmath18 we tacitly apply an isometric embedding @xmath35 given by @xmath36 where @xmath37 is the characteristic function of the interval @xmath38 dixmier traces @xmath25 defined such @xmath17 s are termed _ connes - dixmier traces_. we refer to @xcite and @xcite for discussion of their properties . finally , various formulae of noncommutative geometry ( in particular , those involving heat kernel estimates and generalised @xmath39function ) were established in @xcite for yet a smaller subset of connes - dixmier traces , when the functional @xmath17 was assumed to be @xmath33-invariant . this class ( and its further modifications ) was first introduced in @xcite ( see also @xcite ) and further studied and used in @xcite . for brevity
we refer to the latter class ( a proper subclass of connes - dixmier traces ) as a class of _ @xmath33-invariant dixmier traces_. in the case , when we deal with the standard trace @xmath40 and the standard trace class @xmath41 of compact operators from @xmath42 the classical lidskii theorem asserts that the trace @xmath43 for any @xmath44 here , @xmath45 is the sequence of eigenvalues of @xmath46 taken in any order .
this arbitrariness of the order is due to the absolute convergence of the series @xmath47 in particular , we can choose the decreasing order of absolute values of @xmath48 and counting multiplicities .
the core difference of this situation with the setting of dixmier traces living on the ideal @xmath5 consists in the fact that the series @xmath49 for simplicity , we explain the emerging obstacle in the case of a self - adjoint operator @xmath50 for such @xmath46 by the definition , @xmath51 where @xmath52 even in this case , it is not clear why the equality @xmath53 should hold for the special enumeration of the set @xmath45 given by the decreasing order of absolute values of @xmath54 or for that matter for _ any _ enumeration of this set .
the following result from @xcite establishes the equality above under significant additional constraints on @xmath25 and @xmath55 [ formula lidskogo from as ] let @xmath17 be @xmath33-invariant and let @xmath56 satisfy the assumption @xmath57 for some @xmath58 and all @xmath59 .
we have @xmath60 where @xmath61 is the spectrum of @xmath62 in the case when @xmath9 is a _ positive _ arbitrary element from @xmath5 and @xmath17 is taken from a rather special subset of all @xmath33-invariant generalised limits ( termed in @xcite `` maximally invariant dixmier functionals '' ) this result can be already found in ( * ? ? ?
* proposition 2.4 ) . in (
* theorem 1 ) , the assertion from ( * ? ? ?
* proposition 2.4 ) was extended to an arbitrary @xmath33-invariant @xmath63 another modification of the class of @xmath17 s for which the result of ( * ? ? ?
* proposition 2.4 ) and ( * ? ? ? * theorem 1 ) holds is given in ( * ? ? ?
* proposition 4.3 ) .
in this paper we prove significant extensions and generalisations of theorem [ formula lidskogo from as ] from @xcite , ( * ? ? ?
* proposition 2.4 ) , ( * ? ? ?
* theorem 1 ) and ( * ? ? ?
* proposition 4.3 ) .
many of our results are established for a general class of marcinkiewicz ideals . here , for convenience of the reader ,
we restate these results for traces on @xmath26 our first main result shows that the assertion of theorem [ formula lidskogo from as ] holds for an arbitrary connes - dixmier trace @xmath25 . [ formulalidskogo1 ] let @xmath25 be a connes - dixmier trace on @xmath26 we have @xmath64 theorem [ formulalidskogo1 ] follows immediately from theorem [ formula lidskogo dlya l1infty ] below .
our second main result is the answer to a natural question whether formula holds for every dixmier trace .
this question is answered in negative in theorem [ fails for dixmier traces ] .
our third ( and the last ) main result answers in the affirmative the question whether there exists a modification of the summation method used in formula ensuring that it holds for all dixmier traces .
[ formulalidskogo2 ] let @xmath25 be a dixmier trace on @xmath26 we have @xmath65 theorem [ formulalidskogo2 ] follows immediately from theorem [ lidskii formula ] below .
at the end of the paper we also provide an application of our results .
the result proved in the last section concerns heat kernel type formulae from noncommutative geometry ( see @xcite ) and has been already established in @xcite with a rather arcane argument .
we present here a very simple approach to these formulae .
it is convenient to consider the general class of marcinkiewicz spaces since many of our results hold for this class with no extra effort .
we frequently use commutative results as a stepping stone to obtain their noncommutative analogues .
recall that the distribution function @xmath66 of a bounded measurable function @xmath6 is defined by the formula @xmath67 we write @xmath68 for the decreasing rearrangement of the function @xmath6 : @xmath68 is the right continuous non - increasing function whose distribution function coincides with that of @xmath69 ( see @xcite ) .
the following formula is frequently used in the proofs below sometimes without explicit referencing .
@xmath70 here , @xmath71 is any positive number .
marcinkiewicz spaces are a special case of fully symmetric function and sequence spaces , see @xcite .
denote by @xmath72 the class of all concave increasing functions such that @xmath73 @xmath74 as @xmath75 and @xmath76 as @xmath77 for every @xmath78 marcinkiewicz space @xmath79 is a set of all bounded measurable functions @xmath6 on @xmath80 such that @xmath81 marcinkiewicz sequence space @xmath82 is a set of sequences ( see e.g. @xcite ) satisfying the condition @xmath83 in this paper , we mainly work with functions @xmath84 satisfying the following condition .
@xmath85 let @xmath86 be the ideal of all compact operators .
if @xmath82 is a marcinkiewicz sequence space , then the corresponding marcinkiewicz operator space @xmath87 is the set of all @xmath88 such that @xmath89 equipped with the norm @xmath90 let @xmath84 and let @xmath17 be a dilation invariant generalised limit .
the mapping @xmath25 defined by the formula @xmath91 is a subadditive homogeneous functional on @xmath92 if @xmath25 is additive on @xmath93 then @xmath25 is called _ dixmier trace generated by @xmath17_. we refer the reader to @xcite for conditions which guarantee the additivity of @xmath94 it is well known that @xmath25 is additive for any @xmath17 as above when @xmath95 similarly , the definitions of connes - dixmier traces and @xmath33-invariant traces naturally extend to denote corresponding singular traces on marcinkiewicz ideals @xmath87 ( see @xcite ) .
our main result for general dixmier traces on ideals @xmath87 is given in theorem [ lidskii formula ] which asserts that for any dixmier trace @xmath25 on @xmath87 with @xmath84 satisfying condition we have @xmath96 the result of theorem [ formulalidskogo2 ] follows immediately from the formula above ,
if we set @xmath97 for all @xmath98 here , we show that there are dixmier traces @xmath25 on @xmath5 for which formula fails . to this end
we use @xmath17 provided by the lemma below . define a subadditive functional @xmath99 by the formula @xmath100 clearly
, @xmath101 is positive and homogeneous .
the following lemma is routine .
we include the proof for convenience of the reader .
[ pi lemma ] 1 . if @xmath102 such that @xmath103 then @xmath17 is dilation invariant generalised limit .
2 . if @xmath104 then there exists a dilation invariant generalised limit @xmath17 such that @xmath105 we prove the first assertion and then derive the second one from it . 1 . at
first we note that by assumption @xmath106 for every @xmath107 note that @xmath108 for every @xmath109 it follows that @xmath17 is positive . + further , for every @xmath110 we have @xmath111 therefore , @xmath112 hence , @xmath113 thus , @xmath17 is dilation invariant . + if @xmath110 is such that @xmath114 as @xmath115 then @xmath116 it follows from that @xmath117 + noting that @xmath118 we conclude that @xmath17 is dilation invariant generalised limit .
2 . consider linear space @xmath119 spanned by element @xmath120 set @xmath121 for every @xmath122 it follows that @xmath123 on @xmath124 by the hahn - banach theorem , there exists a functional @xmath102 such that @xmath125 and @xmath126 it follows from above that @xmath17 is a dilation invariant generalised limit .
[ fails for dixmier traces ] there exist a positive function @xmath127 and a dixmier trace @xmath25 such that @xmath128 define a function @xmath6 by the formula @xmath129}.\ ] ] if @xmath130,$ ] then @xmath131 thus , @xmath132 we claim that @xmath133 set @xmath134 it is clear that @xmath135 for every @xmath136.$ ] since @xmath137 for every @xmath138 $ ] and every @xmath139,$ ] we can rewrite the expression under the limit in the left - hand side as @xmath140 @xmath141 this proves the claim . thus , @xmath142 the assertion of the theorem now follows from lemma [ pi lemma ] and .
in this section , we extend results of @xcite ( and , partially , those of @xcite ) to a wider class of marcinkiewicz spaces and connes - dixmier traces . to this end
, we need some extra assumptions on @xmath143 the need of such additional conditions is seen from the example below , which shows that analogue of formula for an arbitrary @xmath84 fails .
let @xmath144 and let @xmath145 if @xmath25 is a dixmier trace on @xmath146 then @xmath147 it is clear that @xmath148 we have @xmath149 for all sufficiently large @xmath150 hence , @xmath151 the assertion follows immediately .
thus , some additional restrictions on the function @xmath152 are needed .
we require the following condition @xmath153 it is clear that holds and , therefore , marcinkiewicz space @xmath79 admits nonzero dixmier traces ( see @xcite,@xcite,@xcite ) .
now we show that formula holds for all connes - dixmier traces on @xmath154 [ second dx estimate ] let @xmath84 satisfy condition . if @xmath155 we have @xmath156 for every @xmath157 and every sufficiently large @xmath150 assume the contrary .
hence , there exists a sequence @xmath158 such that @xmath159 for every @xmath160.$ ] by the definition of marcinkiewicz norm , @xmath161 it follows from that @xmath162 the contradiction proves the lemma .
[ 1+o1 remark ] let @xmath163 and let @xmath164 as @xmath77 if @xmath165 we have @xmath166 [ m k nulyu ] let @xmath84 satisfy condition .
we have @xmath167 as @xmath168 for every positive @xmath169 and every @xmath170 the assertion is linear with respect to @xmath120 since the assertion holds for @xmath171 it is sufficient to verify it for @xmath172 instead of @xmath120 hence , we may assume that @xmath173 thus , integral in the right - hand side is unbounded as @xmath174 make a substitution @xmath175 in the left - hand side integral .
it follows from the condition that @xmath176 indeed , by lagrange theorem , we have @xmath177 it follows from remark [ 1+o1 remark ] that @xmath178 evidently , @xmath179 noting that @xmath180 the combination of and yields the assertion .
[ ocenka snizu connes ] let @xmath84 satisfy the condition and let @xmath25 be a connes - dixmier trace on @xmath181 we have @xmath182 for every positive @xmath132 due to and lemma [ second dx estimate ] we have @xmath183 @xmath184 [ prostaya ocenka sverhu ] let @xmath84 and let @xmath25 be a dixmier trace on @xmath181 we have @xmath185 for every positive @xmath186 we claim that @xmath187 the inequality is evident if @xmath188 if @xmath189 then @xmath190 for every @xmath191.$ ] it follows that @xmath192 @xmath193 thus , claim holds in either case .
it follows that @xmath194 the assertion follows immediately .
the next theorem follows immediately from lemma [ ocenka snizu connes ] and lemma [ prostaya ocenka sverhu ] .
[ formula lidskogo dlya l1infty comm ] let @xmath84 satisfy the condition and let @xmath25 be a connes - dixmier trace on @xmath181 we have @xmath195 for every positive @xmath186 consider weak space @xmath196 ( the smallest symmetric ideal containing @xmath197 ) .
suppose that @xmath152 satisfies the condition .
if @xmath25 is an arbitrary dixmier trace on @xmath146 then we have @xmath198 for every positive @xmath199 using lemma [ second dx estimate ] , the equality above follows immediately .
arguing as in the section [ noncomm section ] below , we obtain a noncommutative version of theorem [ formula lidskogo dlya l1infty comm ] , which strengthens ( * ? ? ?
* theorem 1 ) and ( * ? ? ?
* corollary 2.12 ) ( see theorem [ formula lidskogo from as ] ) .
[ formula lidskogo dlya l1infty ] let @xmath84 satisfy the condition and let @xmath25 be a connes - dixmier trace on @xmath154 we have @xmath200 for every operator @xmath201
as we have seen in theorem [ fails for dixmier traces ] , formula does not hold for dixmier traces @xmath94 in this section , we consider a modification of formula which holds for all dixmier traces @xmath25 on a commutative marcinkiewicz space @xmath146 @xmath143 [ dx estimate ] let @xmath84 satisfy condition . if @xmath202 then there exists a constant @xmath203 such that @xmath204 for every sufficiently large @xmath150 set @xmath205 it follows from that there exists a constant @xmath206 and @xmath207 such that @xmath208 for every @xmath209 thus , @xmath210 for @xmath209 consider sets @xmath211 and @xmath212 defined by the formula @xmath213 fix @xmath214 such that @xmath215 it follows that @xmath216 for all @xmath209 therefore , @xmath217 if @xmath209 since @xmath218 is an increasing function ( see @xcite ) , we have @xmath219 for @xmath209 since @xmath212 is an interval , we have @xmath220 provided that @xmath209 thus , for @xmath221 we have @xmath222 [ psi compatible remark ] let @xmath84 and let @xmath25 be a dixmier trace on @xmath181 we have @xmath223 for every @xmath224 indeed , if @xmath25 is linear then ( see @xcite ) @xmath225 this remark is frequently used below together with the following lemma from @xcite .
[ 34 ] let @xmath102 be an arbitrary generalised limit . if @xmath226 are such that @xmath227 then @xmath228 [ ocenka snizu ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath181 we have @xmath229 for every positive @xmath186 let @xmath230 be the constant defined in lemma [ dx estimate ] . clearly , @xmath231 it follows from remark [ psi compatible remark ] and lemma [ 34 ] that @xmath232 it follows from lemma [ dx estimate ] that @xmath233 however , since @xmath17 is dilation invariant , we have @xmath234 [ ocenka sverhu ] let @xmath84 and let @xmath25 be a dixmier trace on @xmath181 we have @xmath235 for every positive @xmath186 fix @xmath236 clearly , @xmath237 it follows from remark [ psi compatible remark ] and lemma [ 34 ] that @xmath238 we claim that @xmath239 the inequality is evident if @xmath240 if @xmath241 then @xmath242 for every @xmath243.$ ] thus , @xmath244 @xmath245 and the claim follows .
hence , @xmath246 it follows from and the dilation - invariance of @xmath17 that @xmath247 since @xmath248 is arbitrary large , we are done .
the following theorem is the principal result of this section .
it follows immediately from lemmas [ ocenka snizu ] and [ ocenka sverhu ] .
[ lidskii comm ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath181 we have @xmath249 for every positive @xmath186 arguing similarly , one can obtain similar assertion for marcinkiewicz sequence spaces .
[ lidskii comm seq ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath250 we have @xmath251 for every positive @xmath252
in this section , we extend preceding results to dixmier traces on marcinkiewicz operator ideals .
the following assertion follows directly from the theorem [ lidskii comm seq ] .
[ self - adjoint ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath154 we have @xmath253 for every self - adjoint operator @xmath254 the following three lemmas are used to extend the formula above to the case of normal operators .
[ useful estimate ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath181 we have @xmath255 for every positive @xmath186 a similar assertion holds for marcinkiewicz sequence space @xmath250 fix @xmath236 it follows from the dilation - invariance of @xmath17 that @xmath256 it is clear that @xmath257 if @xmath241 we have @xmath258 if @xmath259 then @xmath260 the last inequality holds for all sufficiently large @xmath261 by lemma [ dx estimate ] . in either case , @xmath262 it follows now from the that @xmath263 it is clear that @xmath264 @xmath265 it follows from the dilation - invariance of @xmath17 that @xmath266 it follows from remark [ psi compatible remark ] and lemma [ 34 ] that both terms in the right - hand side of the equality above are equal to @xmath267 therefore , @xmath268 since @xmath248 is arbitrary large , we are done .
[ nenujnye chleny ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath154 we have @xmath269 @xmath270 for any normal operator @xmath254 we prove the first assertion only .
proof of the second one is identical .
note that @xmath271 if and only if @xmath272 it follows immediately that @xmath273 @xmath274 the assertion follows now from lemma [ useful estimate ] .
[ eshe nenujnye chleny ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath154 we have @xmath275 for any normal operator @xmath254 it is clear that @xmath276 @xmath277 @xmath278 the assertion follows now from lemma [ useful estimate ] .
the following theorem extends result of lemma [ self - adjoint ] to normal operators from @xmath181 [ lidskiy normal ] let @xmath84 satisfy the condition and let @xmath25 be a dixmier trace on @xmath154 we have @xmath279 for any normal operator @xmath254 it follows from lemma [ self - adjoint ] that @xmath280 by lemma [ nenujnye chleny ] , @xmath281 the same is valid for @xmath282 by the linearity , @xmath283 it follows from lemma [ eshe nenujnye chleny ] that @xmath284 recall the following result of ringrose ( see theorems 1 , 6 and 7 from @xcite ) .
let @xmath285 be a compact operator .
there exists a projection - valued measure @xmath286 such that 1 .
@xmath287 2 .
either @xmath288 or @xmath289 3 .
[ quasinilpotence criterion ] if , in addition , @xmath290 then @xmath9 is quasi - nilpotent .
[ normal plus quasinilpotent ] let @xmath285 be a compact operator .
there exist compact normal operator @xmath291 and compact quasi - nilpotent operator @xmath292 such that @xmath293 and @xmath294 define an operator @xmath291 by the following formula @xmath295 a straightforward computation shows that the operator @xmath296 satisfies the condition [ quasinilpotence criterion ] of the theorem above .
hence , @xmath292 is quasi - nilpotent .
evidently , @xmath291 is a diagonal operator with eigenvalues of @xmath9 on the diagonal .
hence , @xmath294 by the weil theorem , sequence of eigenvalues of @xmath9 is majorized by the sequence of its singular values ( see theorem 3.1 from @xcite ) . hence , for @xmath297 we obtain @xmath298 the following assertion directly follows from the theorem 3.3 from @xcite ) .
if @xmath299 is a quasi - nilpotent operator , then @xmath292 belongs to the commutator @xmath300.$ ] [ quasinilpotent inessential ] if @xmath299 is a quasi - nilpotent operator and @xmath25 is an arbitrary dixmier trace on @xmath146 then @xmath301 indeed , due to @xcite , we have @xmath302)=0 $ ] for every @xmath303 and every @xmath304 the following theorem is the main result of this section .
[ lidskii formula ] let @xmath84 satisfy condition and let @xmath25 be a dixmier trace on @xmath154 we have @xmath305 for any operator @xmath201 let @xmath291 be a normal operator constructed in corollary [ normal plus quasinilpotent ] .
the assertion holds for @xmath291 by theorem [ lidskiy normal ] .
note that @xmath306 by corollaries [ normal plus quasinilpotent ] and [ quasinilpotent inessential ] . since @xmath307 we are done .
in this section , we provide a simple proof of one of the heat semigroup formulae from @xcite ( see also earlier results in @xcite ) .
our hypothesis on @xmath17 is very mild .
let @xmath315 be the rearrangement of @xmath46 that is @xmath316 without loss of generality , @xmath317 since distributions of @xmath9 and @xmath6 coincide , we have @xmath318 setting @xmath319 we obtain @xmath320 it follows from the weak version of karamata theorem ( see @xcite for details ) that @xmath321 it is clear that @xmath322 @xmath323 integrating by parts , we obtain @xmath324 hence , @xmath325 it follows from the lemma [ another m lemma ] that @xmath326 the assertion follows now from the @xmath327invariance of @xmath63 100 azamov n. , sukochev f. _ a lidskii type formula for dixmier traces _ , c.r .
paris 340 ( 2005 ) , no .
2 , 107112 . benameur m. , fack t. _ type ii non - commutative geometry .
i. dixmier trace in von neumann algebras _ , adv .
199 ( 2006 ) , no .
1 , 2987 .
carey a. , phillips j. , sukochev f. _ spectral flow and dixmier traces _ , adv . math .
173 ( 2003 ) , no .
1 . , 68113 .
carey a. , phillips j. , rennie a. , sukochev f. _ the hochschild class of the chern character for semifinite spectral triples _
, j. funct .
213 , ( 2004 ) no . 1
, 111153 .
carey a. , rennie a. , sedaev a. , sukochev f. _ the dixmier trace and asymptotics of zeta functions _ , j. funct .
249 ( 2007 ) , no .
2 , 253283 .
carey a. , sukochev f. _ dixmier traces and some applications in non - commutative geometry _
, russian math .
surveys 61:6 10391099 .
connes a. _ noncommutative geometry _ , academic press , san diego , 1994 .
dixmier j. _ existence de traces non normales _ , c. r. acad .
paris , 262 ( 1966 ) .
dodds p. , de pagter b. , sedaev a. , semenov e. , sukochev f. _ singular symmetric functionals _
, ( russian ) zap .
nauchn . sem .
otdel . mat .
inst . steklov .
( pomi ) 290 ( 2002 ) , issled .
po linein .
30 , 4271 , 178 ; translation in j. math .
( n. y. ) 124 ( 2004 ) , no .
2 , 48674885 dodds p. , de pagter b. , sedaev a. , semenov e. , sukochev f. _ singular symmetric functionals and banach limits with additional invariance properties _ , ( russian ) _ izv . ross .
nauk ser . mat .
_ , vol .
67 , no . 6 ( 2003 )
dodds p. , de pagter b. , semenov e. , sukochev f. _ symmetric functionals and singular traces _
, positivity 2 ( 1998 ) , no .
1 , 4775 . gohberg i. , krein m. _ introduction to the theory of linear nonselfadjoint operators _ , translations of mathematical monographs , vol . 18 american mathematical society , providence , r.i .
1969 kalton n. _ spectral characterization of sums of commutators .
i _ , j. reine angew .
( 1998 ) , 115125 . krein s. , petunin ju . and semenov e. _ interpolation of linear operators _ , nauka , moscow , 1978 ( in russian ) ; english translation in translations of math
. monographs , vol .
* 54 * , amer .
soc . , providence , ri , 1982 .
lord s. , sedaev a. , sukochev f. _ dixmier traces as singular symmetric functionals and applications to measurable operators _ , j. funct .
224 ( 2005 ) , no .
1 , 72106 .
pietsch a. _ about the banach envelope of @xmath328 _ , rev .
22 ( 1 ) ( 2009 ) 209226 .
ringrose j. _ super - diagonal forms for compact linear operators _ , proc .
london math .
( 3 ) 12 ( 1962 ) 367384 .
sedaev a. _ generalized limits and related asymptotic formulas _ , math.notes .
86:4 ( 2009 ) , 612 - 627 .
|
we establish several analogues of the classical lidskii theorem for some special classes of singular traces ( dixmier traces and connes - dixmier traces ) used in noncommutative geometry .
|
the continuous time random walk ( ctrw ) formalism introduced four decades ago by montroll and weiss @xcite has been successfully applied to a wide and diverse variety of physical phenomena over the years @xcite but only recently to finance @xcite . in this latter context , the efforts have been mostly focused on the statistical properties of the waiting time between successive transactions and the asset return at each transaction .
different studies in different markets are conceiving the idea that the empirical distributions of both random variables are compatible with an asymptotic fat tail behavior @xcite . within the ctrw formalism
some of us have recently investigated the mean exit time ( met ) of asset prices out of a given interval of size @xmath0 for financial time series @xcite .
this study shows that the met follows a quadratic growth in terms of @xmath0 for small interval lengths @xmath0 . in the same study ,
this functional form was observed for a specific time series of the foreign exchange ( fx ) market , which is the u.s .
dollar / deutsche mark futures time series @xcite . in this paper
we investigate both theoretically and empirically the met of price returns traded in a stock exchange . in our empirical investigation
we study the met of high frequency return time series of 20 highly capitalized stocks traded in new york stock exchange .
empirical results about this market confirm that the met follows a power law with a pre - factor that depends on the specific stock chosen .
this observation motivates us to first verify and then release some of the assumptions used in ref .
@xcite therefore generalizing the model discussed in that paper .
the theoretical generalization has been performed by introducing and solving a new two - state chain markovian model able to both describe the quadratic scaling property of the met and provide the data collapse of the met stock pre - factor .
we show that a satisfactory data collapse of the met is obtained when some degree of autocorrelation in the stock returns is introduced in the two - state chain markovian model .
we attempt to further improve the accuracy by extending the model to a three - state markov chain for which we are still able to evaluate the met .
nevertheless , empirical data show that the three - state model does not improve the quality of data collapse in the met profiles of the different stocks although the theoretical curve shows a better agreement with the empirical data than that of the two - state markov chain model .
the paper is organized as follows . in sections
[ sect_mean ] and [ sect_discrete ] , we discuss the met behavior under the ctrw formalism and a series of simplifying assumptions . in sect . [ sect_scaling ]
we empirically investigate the scaling and data collapse properties of highly capitalized stock data .
section [ sect_correlation ] relates the time correlation of stock return with the absence of data collapse of met observed in the previous section . in sect .
[ sect_markov ] we introduce and solve a two - state and a three - state markov chain model to describe the empirical met observations .
conclusions are drawn in sect .
[ sect_conclusions ] .
in the most common version of the ctrw formalism a given random process @xmath1 shows a series of random increments or jumps at random times @xmath2 remaining constant between these jumps .
therefore , after a given time interval @xmath3 , the process experiences a random increment @xmath4 and the resulting trajectory consists of a series of steps as shown in fig .
[ model ] .
both waiting times @xmath5 and random jumps @xmath6 are assumed to be independent and identically distributed ( i.i.d . )
random variables described by their probability density functions ( pdfs ) which we denote by @xmath7 and @xmath8 respectively .
however , in the most general representation of the formalism , another function is needed to describe the time evolution of @xmath1 .
we denote this function by @xmath9 which is the joint pdf of waiting times and jumps : @xmath10 note that the functions @xmath7 and @xmath8 are the marginal probability density functions of @xmath9 .
we refer the reader to ref .
@xcite for a more complete account of the ctrw formalism . process along with the corresponding value of the random variable @xmath11.,scaledwidth=45.0% ] in this paper we will apply the ctrw to study some aspects of the exit problem of financial time series
. we will take as underlying random process @xmath1 the logarithmic price @xmath12 , where @xmath13 is the stock price at time @xmath14 .
we specifically consider the problem of obtaining the mean exit time of @xmath1 out of a given interval @xmath15 $ ] .
we assume that at certain reference time @xmath16 , right after an event , the price has a known value @xmath17 , @xmath18 $ ] .
we focus our attention on a particular realization of the process and suppose that at certain time @xmath19 the process first leaves the interval ( see fig .
[ model ] ) .
we call the interval @xmath20 the exit time out of the region @xmath15 $ ] and denote it by @xmath11 .
this quantity is a random variable since it depends on the particular trajectory of @xmath1 chosen and the met is simply the average @xmath21 $ ] .
the standard approach to exit time problems is based on the knowledge of the survival probability which is generally quite involved @xcite .
however , within the ctrw formalism one can assume that the events compose a series of independent and identically distributed two - dimensional random variables .
under such an assumption , some of us @xcite have recently shown that one can obtain the met directly , without making use of the survival probability . in this framework
the met obeys the following integral equation @xcite @xmath22 + \int_a^bh(x - x_0)t(x)dx , \label{met_iid}\ ] ] where @xmath23 = \int_0^{\infty}\psi(\tau ' ) \tau ' d\tau'\ ] ] is the mean waiting time between jumps .
it is worth noticing that eq .
( [ met_iid ] ) is still valid even when @xmath5 and @xmath24 are cross - correlated .
in fact , in the case of an i.i.d .
process the met only depends on the pdfs of waiting times @xmath7 and jumps @xmath8 , but it does not depend on the particular form of the joint pdf @xmath9 . however if we would remove the i.i.d .
hypothesis we should specify a functional form for @xmath9 .
we now assume that returns increments are distributed according to an even pdf , @xmath25 , which also satisfies the following scaling condition , @xmath26 where @xmath27 is the scale of the fluctuations given by the standard deviation @xmath27 of jumps , where @xmath28 @xmath29 ^ 2]$ ] .
the parameter @xmath27 corresponds to the transaction - to - transaction volatility . under these assumptions some of us showed that the met out of a small region of size @xmath30 is @xcite @xmath31 \left[1 + 2 h(0)\left(\frac{l}{2\kappa}\right)\right.\nonumber \\ & + & ( h'(0^{+})+4 h(0)^2)\left(\frac{l}{2\kappa}\right)^2 + \left .
o\left(\frac{l^3}{\kappa^3}\right)\right ] , \label{start}\end{aligned}\ ] ] where it is assumed that the return process is initially in the middle of the interval @xmath15 $ ] . in ref .
@xcite some of us have applied the above result to the fx market of the u.s .
dollar / deutsche mark future price .
one important conclusion there was that the quadratic growth of the met is still a good approximation even for large intervals , i.e. @xmath32 .
it is less clear nonetheless that the coefficients of the polynomial which we have obtained in eq .
( [ start ] ) are those that better reproduce the global behavior of the met .
we refer the reader to ref .
@xcite for a more detailed discussion on this particular aspect of the problem . in this paper
we analyze the scaling properties of the met for @xmath33 highly capitalized stocks traded at the nyse in the four year period 1995 - 1998 and spanning @xmath34 trading days .
table [ summary ] shows the list of stocks and the relevant parameters .
we have measured the met for each stock and compared them in the scaled variables @xmath35 $ ] and @xmath36 .
if all previous hypotheses of the model were correct , and the function @xmath37 is of universal nature , one should observe the same curve for all stocks , i.e. a data collapse , as well as a quadratic growth in @xmath0 at least for small @xmath0 .
however , fig .
[ startscaling ] shows that there is a considerable spread of the curves .
although the parabolic shape is recovered in all cases not only for small intervals , as expected , but for the whole investigated range of @xmath38 . l|ccc ticker & transactions & @xmath27 ( @xmath39 ) & @xmath40 $ ] ( seconds ) ahp & 521,639 & 9.18 & 44.2 + aig & 472,393 & 6.98 & 49.2 + bmy & 573,397 & 7.30 & 40.3 + chv & 449,328 & 8.47 & 51.5 + dd & 645,164 & 8.80 & 36.0 + ge & 1,319,145 & 7.17 & 18.7 + gte & 512,581 & 13.10 & 45.2 + hwp & 930,003 & 9.20 & 25.0 + ibm & 1,072,395 & 5.42 & 22.7 + jnj & 728,686 & 8.75 & 32.1 + ko & 784,357 & 9.11 & 29.7 + mo & 971,700 & 10.11 & 23.9 + mob & 461,669 & 6.82 & 50.0 + mrk & 971,842 & 7.34 & 23.8 + pep & 767,929 & 15.48 & 30.3 + pfe & 1,003,518 & 7.46 & 23.1 + pg & 679,601 & 7.36 & 34.2 + t & 1,030,761 & 11.13 & 22.6 + wmt & 565,946 & 15.80 & 40.1 + xon & 674,412 & 7.08 & 34.5 for the @xmath33 investigated stocks .
the @xmath41 axis is scaled by @xmath42 , whereas the @xmath43 axis is scaled by the mean waiting time @xmath40 $ ] .
the inset shows the same curves in a double logarithmic plot.,scaledwidth=45.0% ]
our first objective is to understand why the quadratic term governs both the long and short range behavior of the met , without a drastic change of the general model and its assumptions . in the present approach
we develop a model for @xmath8 based on the small - scale properties of the system .
it is worth noticing that this approach is also used in the context of option pricing , when the fair price of a derivative product is obtained by making use of the binomial trees methodology , where it is assumed that the stock price makes a jump up or down with some probability @xcite .
here we introduce the following symmetrical three - state discrete model : @xmath44 .
\label{2s_h}\ ] ] where @xmath45 represents the probability that the price remains unchanged , and @xmath46 is the basic jump size @xcite . by substituting this expression of @xmath8 into eq .
( [ met_iid ] ) we obtain @xmath22 + q\ t(x_0 ) + \frac{1-q}{2}\left[t(x_0+c)+t(x_0-c)\right],\ ] ] with the convention that the term @xmath47 only counts if @xmath48 , and similarly that @xmath49 only appears when @xmath50 .
let us analyze these two boundary conditions in greater detail .
in general , the limits of our interval can be expressed in the following form : @xmath51 with @xmath52 and @xmath53 .
however it is easy to conclude that @xmath54 can depend neither on @xmath55 nor on @xmath56 .
the only way of leaving the interval is by reaching the points @xmath57 or @xmath58 , because @xmath59 , and @xmath60 lay always inside the interval
. therefore we will not loose generality by setting @xmath61 .
after that , the length of the interval in the natural scale units of the problem is @xmath62 , and the use of the following notation @xmath63 , with @xmath64 , arises in a natural way . summing up , for the discrete model given by eq .
( [ 2s_h ] ) the met out of the interval @xmath15 $ ] obeys the following set of difference equations @xmath65}{1-q}+\frac{1}{2 } \left(t_{n+1}+t_{n-1 } \right ) \label{2s_met}\ ] ] @xmath66 with boundary conditions : @xmath67 the solution to eqs .
( [ 2s_met])-([2s_bc ] ) is given by @xmath68(n+1)(n+1-n ) , \label{2s_solution}\ ] ] where the random variable @xmath69 is related to @xmath70 in such a way that @xmath71=\frac{\text{e}[\tau]}{1-q}. \label{tilde_tau}\ ] ] by repeating the above derivation leading to eq .
( [ 2s_met ] ) , one can show that the random variable @xmath69 represents the waiting time between jumps if one neglects zero - return trades , i.e. if one identifies the occurrence of a jump when @xmath1 actually changes its value .
thus , for instance , if @xmath72 , @xmath73 , and @xmath74 with corresponding waiting times @xmath75 , @xmath76 , and @xmath77 , we can replace the pair of events @xmath78 with a single transaction of size @xmath79 , taking a waiting time @xmath80 . from eq .
( [ 2s_solution ] ) we see that , for even values of @xmath81 , the met starting from the middle of the interval reads @xmath82\left(1+\frac{n}{2}\right)^2 . \label{middle_even}\ ] ] looking at the general solution given by eq .
( [ 2s_solution ] ) and also at eq .
( [ middle_even ] ) we clearly observe a quadratic behavior of the met as a function of @xmath81 , that is to say , as a function of the length of the interval . indeed , from eq .
( [ middle_even ] ) we have : @xmath83}= \left(1+\frac{l}{2\tilde\kappa}\right)^2 . \label{met_iid_sol_exact_a}\ ] ] where @xmath84 and @xmath85 @xmath86 ^ 2]$ ] @xmath87 .
the same kind of scaling also holds in terms of the parameters of the three - state model .
in fact , from eq .
( [ tilde_tau ] ) we get @xmath88}= \frac{1}{1-q}\left(1+\frac{l\sqrt{1-q}}{2\kappa}\right)^2 . \label{met_iid_sol_exact}\ ] ] where @xmath89 @xmath90 ^ 2]$ ] @xmath91 .
hence , for large values of @xmath92 , @xmath88 } \sim \left(\frac{l}{2\kappa}\right)^2 . \label{met_iid_sol_b}\ ] ]
in the previous section we have shown that a simple discrete model for the jump distribution results in a quadratic growth of the met valid for arbitrary values of the length of the interval and not only for small values of @xmath92 , as was the case of eq .
( [ start ] ) .
unfortunately the discrete model does not properly account for the spread of the met curves observed in fig .
[ startscaling ] when we consider different stocks .
one could argue that this spread can be controlled through the parameter @xmath45 appearing in eq .
( [ met_iid_sol_exact ] ) since @xmath45 may distinguish one stock from another .
however , as eq .
( [ met_iid_sol_b ] ) shows the met is practically independent of @xmath45 for large values of @xmath0 and the difference between stocks would disappear in this range of lengths .
nevertheless , we clearly see in fig .
[ startscaling ] that the spread between stocks does not tend to vanish but , even in some cases , it increases with @xmath0 . in this section
we try to identify the possible reasons of the failure of the data collapse of our previous model .
we revisit some of our assumptions and derive consequences with the aim of finding the most important feature that we are leaving aside .
the final goal is to improve our description in the simplest possible way .
let us first summarize some potential causes for the lack of data collapse in the previous models : 1 .
the probability density @xmath8 is different for different stocks .
this would imply different mean exit time curves .
2 . there is some dependency on the cross - correlation between waiting times and price returns .
3 . the time auto - correlation of waiting times should be included in the model .
the time auto - correlation of returns should be included in the model .
we analyze the impact of these hypotheses on the empirical outcomes by performing shuffling experiments .
thus , in order to test the first hypothesis we shuffle independently the time series of @xmath24 and @xmath5 and we perform the mean exit time analysis on the shuffled time series .
this shuffling destroys all the time- and cross- correlations but it preserves the shape of the pdfs @xmath8 and @xmath7 and therefore the values of the scaling parameter @xmath27 and the average waiting time @xmath40 $ ] .
figure [ shuffling2 ] shows a very good collapse indicating that the assumption of a master pdf for the returns of all the stocks is good working hypothesis and it is not the reason for the lack of collapse of fig .
[ startscaling ]
. moreover the met curves are well fitted by the functional form @xmath93 . for the @xmath33
investigated time series after shuffling both waiting times and return increments .
the @xmath41 axis is scaled by @xmath42 , whereas the @xmath43 axis is scaled by the mean waiting time @xmath94 $ ] .
the inset shows the same curves in a double logarithmic plot .
the dashed curve represents the functional form @xmath93.,scaledwidth=45.0% ] the other three hypotheses can be similarly tested by performing three different shuffling experiments . specifically , hypothesis 2 ) can be tested by shuffling simultaneously the two series and preserving the cross - correlation between @xmath24 and @xmath5 .
notice that even if ref .
@xcite has shown that in the absence of autocorrelation of waiting times and returns ( i.i.d .
model ) the met is independent of the cross - correlations @xmath9 , in the general case @xmath9 may play a role .
hypothesis 3 ) can be tested by shuffling only the series of waiting times and preserving the order in the series of returns .
finally , hypothesis 4 ) can be tested by shuffling only the time series of returns and preserving the order of the series of waiting times .
figure [ 4tests ] shows the results of these shuffling experiments for the general electric ( ge ) stock .
the figure shows that neglecting the autocorrelation function of waiting times does not change the met curve . on the other hand , when one destroys the auto correlation of returns the met ( star ) changes dramatically and becomes close to the one predicted by the i.i.d . model . for general electric obtained by performing four different shuffling experiments .
the dashed ( black ) curve is the met of the original data .
the solid ( red ) line is the met for data shuffled both in the waiting times and in the return increments .
the squares ( green ) give the met when one shuffles both @xmath24 and @xmath5 but preserving the association ( cross correlation ) between the two and circles ( blue ) give the met when one shuffles only the waiting times .
the star ( magenta ) give the met when one shuffles only the returns.,scaledwidth=45.0% ] by summarizing , in order to have a good model of the mean exit time one can not neglect the correlation properties of price return , whereas the other correlations can be neglected as a first approximation .
what is the origin of correlation of returns ? there are , in principle , two possible answers .
one contribution comes from the linear autocorrelation of the increments of returns given by @xmath95 $ ] .
a second contribution is related to nonlinear properties which can be exemplified by the nonlinear correlation @xmath96 $ ] .
the first contribution can be easily evaluated by taking the linear autocorrelation function of the increments of price returns .
figure [ correlation ] shows this quantity for ge ( solid line ) . in the figure it is clear that for a lag of one trade the linear autocorrelation function is negative and significantly different from zero
this is a known effect of transaction prices which is due to the presence of a spread between the best bid and the best ask ( the bid - ask bounce " , see for example @xcite ) .
it is reasonable to assume that this short range correlation can be included in the model by using a markov process .
the nonlinear correlation is related to the volatility and can be quantified by plotting the autocorrelation function of @xmath97 .
the result ( dashed line ) in fig .
[ correlation ] is a slowly decaying function indicating a long range correlation , which is probably not compatible with a markovian model .
we can test the relative importance of the two contributions to the correlation of returns by performing another shuffling experiment . we can obtain a surrogate time series with the same linear autocorrelation function but with an uncorrelated volatility ( absolute value of return increments ) . the method
( see for example chapter 7 of ref .
@xcite ) consists in taking the fourier transform of the original time series and then randomize its phases . because of the wiener - kinchine theorem the linear autocorrelation of the surrogate time series is the same as the original , but the nonlinear correlation will be zero .
therefore , we have a time series with the same bid - ask bounce properties but with an uncorrelated volatility .
one difficulty of the method is the fact that the pdf of the surrogate series will be in general different from the original one ( unless the time series is gaussian ) .
since we know that , in general , the met depends on the return pdf , we should control that the distortion of the pdf introduced by the phase randomization is not critical in changing the properties of the met . to this end
we first compare the met curve of the shuffled original time series to the shuffled phase - randomized time series .
these two series are both i.i.d . but with different pdfs .
figure [ surrogate ] shows that the two mets ( dashed line and squares ) are very close , thus indicating that the distortion of the pdf introduced by the phase randomization is not critical for the met properties .
we can now compare the mets of the original data to the ones for the phase - randomized time series .
the two series have the same linear autocorrelation function , but the first one displays a clustered volatility whereas the randomized series does not .
the two mets ( see solid line and circles in fig . [ surrogate ] ) are again very close , thus showing that the most important time correlation contribution to the met is the linear autocorrelation ( bid - ask bounce ) , while the clustered volatility plays a minor role . ) .
the squares ( blue ) refer to the shuffled phase - randomized time series .
the solid ( black ) curve refers to the original data ( the curve is also shown in fig . [ startscaling ] ) and the circles ( red ) refer to the phase - randomized time series.,scaledwidth=45.0% ]
in order to incorporate the linear correlations of returns , we derive an integral equation for the mean exit time out of a given interval when the driving process is a continuous time random walk with memory .
in particular , we will consider markov processes within the ctrw framework .
such models lead to the following joint conditional probability density function : @xmath98 in this markovian case the conditional met , @xmath99 , will also depend on both the magnitude of the previous jump @xmath100 and its sojourn time @xmath101 ( see fig .
[ model ] ) . now , and
contrary to eq .
( [ met_iid ] ) , the integral equation for the conditional met depends on the complete joint probability density function .
it reads : @xmath102 + \int_0^{\infty } d\tau \int_{a}^{b } \rho(x - x_0,\tau|\delta x_0,\tau_0)t(x|\delta x,\tau)dx , \label{met_markovian0}\ ] ] where @xmath103 and @xmath104 . the level of complexity of eq .
( [ met_markovian0 ] ) can be considerably reduced by noting that , as we have shown in sect .
[ sect_correlation ] , it is possible to remove the correlation between consecutive waiting times without affecting the met .
therefore , we will assume that the correlation involving waiting times is negligible , and that all relevant information we have to consider when dealing with the @xmath105-th event is the magnitude of the previous change .
in such a case instead of eq .
( [ rho_corr1 ] ) we write @xmath106 hence , the integral equation for the met is simpler , because on the right hand side of eq .
( [ met_markovian0 ] ) we can perform the integral over time .
we thus obtain @xmath107 \label{met_markovian } \nonumber \\ & + & \int_{a}^{b } h(x - x_0|\delta x_0)t(x|\delta x ) dx.\end{aligned}\ ] ] in this case the met only depends on the marginal probability density function of the return increments , @xmath108 , @xmath109 and on the conditional expectation of waiting times @xmath110 $ ] which has to be evaluated through the marginal pdf , @xmath111 , @xmath112 we finally observe that although @xmath18 $ ] , we let @xmath113 to be any real number . in order to solve eq .
( [ met_markovian ] ) and obtain explicit expressions for the met that can be compared with empirical data , we follow the same approach of sect . [ sect_discrete ] and choose a discrete model for @xmath108 . at this point
we can opt for a two - state model in which , at any time step , returns can only go up and down a fixed quantity @xmath46 , or for a three - state model where in addition the return increment can be zero .
we have shown in sect .
[ sect_discrete ] that for an i.i.d .
process both alternatives are equivalent . in the case of a markovian process
the equivalence is not complete .
as we will see below , the final expressions obtained for the unconditional met are slightly different although , for large values of @xmath0 , the leading term is the same in both cases .
let us start with a two - state markov chain model .
in the symmetrical case in which up and down movements are equally likely , the conditional pdf for return increments is @xmath114 where @xmath115 is the correlation between the magnitude of two consecutive jumps : @xmath116 } { \sqrt{\text{var}[\delta \tilde{x}_n]\text{var}[\delta \tilde{x}_{n-1}]}}. \label{r}\ ] ] from eq .
( [ h(x|y ) ] ) we see that the squared volatility , @xmath117 is independent of @xmath43 . by substituting eq .
( [ h(x|y ) ] ) into eq . ( [ met_markovian ] )
we get the following difference equation for the met : @xmath118+\frac{c+r\delta \tilde{x_0}}{2c } t(x_0+c|c ) \nonumber\\ & + & \frac{c - r\delta \tilde{x_0}}{2c } t(x_0-c|-c ) , \label{met_discrete}\end{aligned}\ ] ] where @xmath119 , @xmath120 if @xmath121 and @xmath122 if @xmath123 .
we extend the notation introduced in sect .
[ sect_discrete ] and define @xmath124 now , eq .
( [ met_discrete ] ) is equivalent to the following set of recurrence equations : @xmath125+\frac{1+r}{2 } t_{n+1,n}+\frac{1-r}{2 } t_{n-1,n } \label{set_1},\\
n + 1}&=&\text{e}[\tilde{\tau}]+\frac{1-r}{2 } t_{n+1,n}+\frac{1+r}{2 } t_{n-1,n},\label{set_2}\end{aligned}\ ] ] ( @xmath126 ) with boundary conditions : @xmath127 note that in writing eqs .
( [ set_1])-([set_2 ] ) we have set @xmath128=\text{e}[\tilde{\tau}]$ ] which is consistent with the assumed symmetry between up and down movements ( see also eq .
( [ uncond_met_def ] ) below ) .
the solution to problem ( [ set_1])-([bc_set ] ) reads @xmath125(n+1-n)\left[1+n \frac{1-r}{1+r}\right],\\ t_{n , n + 1}&=&\text{e}[\tilde{\tau}](n+1)\left[1+(n - n ) \frac{1-r}{1+r}\right].\end{aligned}\ ] ] the quantity of interest for our analysis is the unconditional met @xmath129 , which is related to @xmath130 by @xmath131 that is , @xmath132\left[1+\frac{n}{1+r}+\frac{1-r}{1+r}n(n - n)\right ] .
\label{uncond_met}\ ] ] thus the met starting from the middle of the interval reads @xmath133\left[\frac{2r}{1+r}\left(1+\frac{n}{2}\right)+\frac{1-r}{1+r}\left(1+\frac{n}{2}\right)^2\right ] .
\label{met_middle}\ ] ] in this case , and contrary to the i.i.d .
case given in eq .
( [ middle_even ] ) , the met @xmath134 is not a perfect quadratic expression .
however , taking into account that @xmath84 and @xmath135 we have @xmath136}=\frac{2r}{1+r } \left(1+\frac{l}{2\tilde{\kappa}}\right)+\frac{1-r}{1+r } \left(1+\frac{l}{2\tilde{\kappa}}\right)^2 , \label{met_middle_2}\ ] ] and for large values of @xmath137 we recover the expected quadratic behavior in the leading term : @xmath136}\sim\frac{1-r}{1+r}\left(1+\frac{l}{2\tilde{\kappa}}\right)^2 .
\label{twostates}\ ] ] note that the value of @xmath115 depends on each particular stock .
therefore , the scaled met defined as : @xmath138 } , \label{t_rs}\ ] ] tends , for increasing values of @xmath0 , to a quadratic function of the interval length which is independent of the particular stock chosen . ) of the two - state markovian model .
the dashed ( black ) line is the parabolic curve @xmath93 .
the inset is the same shown in fig .
[ startscaling ] for the i.i.d .
process.,scaledwidth=45.0% ] in fig .
[ twostatesfig ] we compare the met curves for the data scaled in the original way ( inset ) , as given in eq .
( [ met_iid_sol_exact_a ] ) , and for the data scaled according to eq .
( [ t_rs ] ) .
the inset of figure [ twostatesfig ] is essentially the same as fig .
[ startscaling ] .
figure [ twostatesfig ] shows that the scaling of eq .
( [ twostates ] ) gives a significant improvement with respect to the original one .
we note that the scaled curves are systematically below the curve @xmath93 , probably because some correlation is not taken correctly into account by the two - state model .
we will now attempt to correct this bias by introducing a three - state model .
one can argue that the two - state model just developed would need an improvement in order to include zero - return transactions , i.e. those with @xmath139 . in sect .
[ sect_discrete ] we have shown that for the i.i.d .
process the inclusion of a third possible state is completely equivalent to a two - state ( up and down ) model after redefining the mean waiting time and the volatility @xmath27 by including the probability @xmath45 of zero - return transactions .
however , when memory is present , as is now the case , this equivalence is not complete .
we thus outline a discrete three - state markov chain model .
the markov - chain model is now characterized by the following transition matrix : @xmath140 where @xmath141 and similar definitions for the rest of the matrix elements .
since we are also assuming that the process is symmetrical for positive and negative returns , the transition matrix @xmath142 can be written in the following form : @xmath143 where @xmath144 is the probability for trapping , @xmath145 and @xmath115 measures the strength of the persistence : @xmath146 note that the first order autocorrelation coefficient , defined as on the right - hand side of eq . ( [ r ] )
is now given by ( _ cf .
_ eq . ( [ r ] ) ) @xmath147 } { \sqrt{\text{var}[\delta { x}_n]\text{var}[\delta { x}_{n-1}]}}=pr .
\label{pr_estimate}\end{aligned}\ ] ] we also need to specify the unconditional probabilities of each state : @xmath148 however , in this case @xmath45 is not an independent parameter as was the case of the i.i.d . model ( _ cf _ sect . [ sect_discrete ] ) . indeed , using the total probability formula and taking into account the values of the transition matrix @xmath142 and the vector @xmath149
, we get @xmath150 observe that when @xmath151 we have @xmath152 and there is no trapping in the value of the random process . now
the pdf of the returns reads @xmath153 , \nonumber\end{aligned}\ ] ] where @xmath154 in this case the integral equation ( [ met_markovian ] ) is equivalent to the following set of difference equations for @xmath155 and @xmath130 : @xmath156 + ( 1-p)t_{n , n}\nonumber\\ & + & p\left[\frac{1+r}{2 } t_{n+1,n}+\frac{1-r}{2 } t_{n-1,n}\right ] \label{3set_1}\end{aligned}\ ] ] @xmath157&+ & q t_{n , n}\nonumber\\ & + & \frac{1-q}{2}\left[t_{n+1,n}+t_{n-1,n}\right ] \label{3set_2}\end{aligned}\ ] ] @xmath158 + ( 1-p)t_{n , n}\nonumber\\ & + & p\left[\frac{1-r}{2 } t_{n+1,n}+\frac{1+r}{2 } t_{n-1,n}\right ] \label{3set_3}\end{aligned}\ ] ] where @xmath130 are defined as in eq .
( [ t_n_def ] ) and @xmath159 in writing eqs .
( [ 3set_1 ] ) and ( [ 3set_3 ] ) we have taken into account the symmetry @xmath160=\text{e}[\tau|-c]\equiv\text{e}[\tau|c].\ ] ] finally the solution to eqs .
( [ 3set_1])-([3set_3 ] ) with boundary conditions @xmath161 ( _ cf _ eq .
( [ bc_set ] ) ) reads @xmath162}{1-q}(n+1-n)\left[1+n\frac{1-pr}{1+pr}\right ] \label{3solution_1}\ ] ] @xmath163}{1-q } & + & \frac{\text{e}[\tau]}{2(1-q ) } \biggl[\frac{1-pr}{1+pr}\bigl[(n - n)(n+1)\nonumber\\ & + & n(n - n+1)\bigr]+n\biggr ] \label{3solution_2}\end{aligned}\ ] ] @xmath164}{1-q}(n+1)\left[1+(n - n)\frac{1-pr}{1+pr}\right ] , \label{3solution_3}\ ] ] where @xmath40 $ ] is the ( unconditional ) mean waiting time which is related to @xmath165 $ ] and @xmath166 $ ] by @xmath23=q\text{e}[\tau|0]+(1-q)\text{e}[\tau|c].\ ] ] in terms of @xmath167 and @xmath155 the unconditional met @xmath129 is given by @xmath168 and starting from the center of the interval we explicitly have @xmath169 } { 1-q}\left[\frac{2pr}{1+pr}\left(1+\frac{n}{2}\right)+\frac{1-pr}{1+pr}\left(1+\frac{n}{2}\right)^2\right]\nonumber\\ & + & q\left[\frac{e[\tau|0]}{1-q}-\frac{e[\tau]}{1-q}\right ] .
\label{met_three_states}\end{aligned}\ ] ] the main difference between this expression and eq .
( [ met_middle ] ) is the final constant term that accounts for the two kind of trapping that the system may experience : the probabilistic one , @xmath170 , and the temporal one , @xmath171\neq e[\tau]$ ] . by using the fact that @xmath172 , the leading term is again of the form : @xmath173}\sim \frac{1-pr}{1+pr}\left(\frac{l}{2\kappa}\right)^2 . \label{first_manuscript}\end{aligned}\ ] ] in fig .
[ threestatesfig ] we show the met curves for the data scaled according to eq .
( [ first_manuscript ] ) . it is worth noting that the quantity @xmath174 appearing in eq .
( [ first_manuscript ] ) can be estimated in two different ways .
one could separately compute @xmath175 , the probability of a change in the price return provided a previous change , and @xmath115 which is the strength of the persistence .
alternatively one can estimate directly the quantity @xmath174 by using eq .
( [ pr_estimate ] ) .
this second approach has the advantage of reducing the dependence of the estimates from the specific details of the model .
this is the reason why in the curves shown in fig .
[ threestatesfig ] the quantity @xmath174 has been estimated by using eq .
( [ pr_estimate ] ) .
the scaling shown in figure [ threestatesfig ] is not satisfactory , we would even say that it is less satisfactory than the scaling corresponding to the two - state model which has been shown in fig .
[ twostatesfig ] .
however , it is worth noting that the rescaled curves are not systematically below the parabolic curve @xmath176 as it was in the case of the two - state model . ) of the three - state markovian model .
the quantity @xmath174 has been estimated by using eq .
( [ pr_estimate ] ) .
the dashed ( black ) line represents the curve @xmath93.,scaledwidth=50.0% ]
this paper presents theoretical and empirical results about the met of financial time series .
specifically the scaling property of the met as a function of the size @xmath0 has been confirmed to follow a quadratic law for a number of stock price time series .
we empirically verify that the quadratic scaling law has associated a pre - factor which is specific to the analyzed stock .
we have performed a series of tests to determine which kind of correlation are responsible for this dependence .
it turned out that the main contribution is associated with the linear autocorrelation property of stock returns .
we have therefore introduced and solved analytically both a two - state and a three - state markov chain models .
the analytical results obtained through the two - state model allow us to get a quite satisfactory data collapse of the 20 met profiles into a single parabolic curve as predicted by the model .
however , this parabolic curve appears to be systematically above real data , that is , the model overstimates the mean exit time .
we have been able to solve a three - state markov chain model as well .
unfortunately this more detailed model does not provide an improvement on data collapse .
the main advantage of this generalization is that the met provided by the model lies close to the empirical curves . in other words
, the three - state model does not overestimate the mets .
we do not have a convincing explanation for this observation but only some indications .
specifically , we have seen that the symmetries assumed in the three state model are not present in some empirical transition matrices @xcite .
perhaps this assumption prevents the data from a convincing data collapse and the system would perform a better collapse in a model taking into account a certain degree of asymmetry in the markovian transition matrix .
however , obtaining the analytical solution for this more general case seems to be very involved and it has been left for future research . in conclusion ,
the met and the search of a data collapse in the met curves of stock prices provide a good occasion for testing the underlying hypothesis characterizing the return dynamics and for the improvement of the ctrw models describing this phenomenon .
we believe we have detected the essential ingredients to be accounted for a feasible model within the ctrw .
we hope that the results may provide a certain insight on the way to bridge the gap between the description of the stochastic dynamics at very short time horizons with that of longer time scales @xcite .
the authors acknowledge support from the esf project `` cost action p10 physics of risk '' .
j.p . and j.m .
acknowledge partial support from direccin general de investigacin under contract no .
bfm2003 - 04574 and by generalitat de catalunya under contract no .
2001 sgr-00061 .
and r.n.m acknowledge support from the research project miur - firb rbne01cw3 m `` cellular self - organizing nets and chaotic nonlinear dynamics to model and control complex system '' , from the research project miur 449/97 high frequency dynamics in financial markets " and from the european union strep project n. 012911 `` human behavior through dynamics of complex social networks : an interdisciplinary approach . '' .
weiss and r.j .
rubin , adv .
* 52 * , 363 ( 1983 ) ; c.w .
gardiner , _ handbook of stochastic methods _ ( springer - verlarg , berlin , 1983 ) ; k. lindenberg and b.j . west , j. stat .
phys . * 42 * , 201 ( 1986 ) .
this choice for @xmath8 implies that the flat levels in every particular trajectory will be in a regular grid of size @xmath46 centered at the starting point @xmath177 .
_ a priori _ we could expect that @xmath178 is related with @xmath45 and , by means of a naive finite - difference approximation , that @xmath179 .
for instance , the general electric transition matrix has the following elements : @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 , @xmath187 , and @xmath188 .
note that some symmetry assumptions on the transition matrix of eqs .
( [ transmat1])-([transmat2 ] ) are approximately satisfied : @xmath189 and @xmath190 while others are not : @xmath191 and @xmath192 .
|
we study theoretical and empirical aspects of the mean exit time of financial time series .
the theoretical modeling is done within the framework of continuous time random walk .
we empirically verify that the mean exit time follows a quadratic scaling law and it has associated a pre - factor which is specific to the analyzed stock .
we perform a series of statistical tests to determine which kind of correlation are responsible for this specificity .
the main contribution is associated with the autocorrelation property of stock returns .
we introduce and solve analytically both a two - state and a three - state markov chain models .
the analytical results obtained with the two - state markov chain model allows us to obtain a data collapse of the 20 measured met profiles in a single master curve .
|
the field of extrasolar planet research has recently made a leap forward with the direct detection of extrasolar giant planets ( egps ) . using spitzer space telescope ,
@xcite and @xcite have detected infrared photons from two transiting planets , tres-1b and hd209458b , respectively .
@xcite have reported the infrared imaging of an egp orbiting the nearby young brown dwarf 2m1207 with vlt / naco , whereas @xcite have collected evidence for an egp companion to the t - tauri star gq lup using vlt / naco as well .
although there are claims that the direct detection of terrestrial planets could be performed from the ground with yet to come extremely large telescopes @xcite , it is widely believed that success will be more likely in space .
direct detection is the key to spectroscopy of planetary atmospheres and discovery of biomarkers , namely indirect evidence of life developed at the planetary scale ( e.g. * ? ? ?
both nasa and esa have space mission studies well underway to achieve this task .
darwin , the european mission to be launched in 2015 , will be a thermal infrared nulling interferometer with three 3.5-m free - flying telescopes @xcite .
terrestrial planet finder , the american counterpart , will feature two missions : a @xmath2 m - monolithic visible telescope equipped with a coronagraph ( tpf - c ) to be launched in 2015 , and an analog to darwin ( tpf - i ) to be launched in the 20152019 range @xcite . the direct detection of the photons emitted by a terrestrial planet is made very challenging by the angular proximity of the parent star , and by the very high contrast ( i.e. luminosity ratio ) between the planet and its star : about @xmath3 in the thermal infrared and about @xmath4 in the visible .
both wavelength ranges have their scientific merits and technical difficulties , and both of them are thought to be necessary for an unambiguous detection of habitability and signs of life ( e.g. * ? ? ?
* ) . in this paper
, we deal with the visible range only . in the visible
, planet detection faces two fundamental noise sources : ( i ) quantum noise of the diffracted star light , and ( ii ) speckle noise due to the scattering of the star light by optical defects .
@xcite proposed a technique based on _ dark speckles _ to overcome speckle noise : random fluctuations of the atmosphere cause the speckles to interfere destructively and disappear at certain locations in the image , thus creating localized dark spots suitable for planet detection .
the statistical analysis of a large number of images then reveals the planet as a spot persistently brighter than the background .
@xcite proposed to use a deformable mirror ( dm ) instead of the atmosphere to make speckles interfere destructively in a targeted region of the image called _ search area _ or _ dark hole _ ( dh or @xmath5 ) .
following the tracks of these authors , this paper discusses methods to reduce the speckle noise below the planet level by using a dm and an ideal coronagraph .
however , unlike @xcite , we propose non - iterative algorithms , in order to limit the number of long exposures needed for terrestrial planet detection .
we will refer to these methods as _ speckle nulling _ techniques , as @xcite call them .
technical aspects of this work are inspired by the high contrast imaging testbed ( hcit ; * ? ? ?
* ) , a speckle - nulling experiment hosted at the jet propulsion laboratory , specifically designed to test tpf - c related technology .
after reviewing the process of speckle formation to establish our notations ( [ sec : speckle_formation ] ) , we derive two speckle nulling methods in the case of small aberrations ( [ sec : speckle_nulling ] ) .
the speckle nulling phase is preceded by the measurement of the electric field in the image plane ( [ sub : measurement ] ) . the performance of both methods
are then evaluated with one- and two - dimensional simulations ( [ sec : simulations ] ) , first with white speckle noise ( [ sub : sim_white ] ) , then with non - white speckle noise ( [ sub : sim_real ] ) .
various effects and instrumental noises are considered in [ sec : discussion ] .
finally , we conclude and discuss some future work ( [ sec : conclusion ] ) .
this paper is written in the framework of fourier optics considering a single wavelength , knowing that a more sophisticated theory ( scalar or vectorial ) in polychromatic light will eventually be needed .
fourier transforms ( fts ) are signaled by a hat .
let us consider a simple telescope with an entrance pupil @xmath6 .
in the pupil plane , we use the reduced coordinates @xmath7 , where @xmath8 are distances in meters and @xmath9 is the wavelength .
we define the pupil function by @xmath10 even in space , i.e. when not observing through a turbulent medium like the atmosphere , the optical train of the telescope is affected by phase and amplitude aberrations .
phase aberrations are wavefront corrugations that typically originate in mirror roughness caused by imperfect polishing , while amplitude aberrations are typically the result of a heterogeneous transmission or reflectivity
. moreover , fresnel propagation turns phase aberrations into amplitude aberrations , and the reverse ( e.g. * ? ? ?
regardless of where they originate physically , all phase and amplitude aberrations can be represented by a complex aberration function @xmath11 in a re - imaged pupil plane , so that the aberrated pupil function is now @xmath12 .
the electric field associated with an incident plane wave of amplitude unity is then @xmath13 exoplanet detection requires that we work in a regime where aberrations are reduced to a small fraction of the wavelength . once in this regime , we can replace @xmath14 by its first order expansion @xmath15 ( we will discuss in [ sub : discuss_linearity ] the validity of this approximation ) . the electric field in the image plane being the ft of ( [ eq : e_pu ] ) , we get @xmath16 where @xmath17 are angular coordinates in the image plane .
the physical picture is as follows .
the first term ( @xmath18 ) is the direct image of the star .
the second term ( @xmath19 ) is the field of speckles surrounding the central star image , where each speckle is generated by the equivalent of first - order scattering from one of the sinusoidal components of the complex aberration @xmath11 .
each speckle is essentially a ghost of the central psf . in the remainder of this paper , we focus on means to measure and correct the speckles in a coronagraphic image . following @xcite we will leave out the unaberrated psf term by assuming that it is was canceled out by a coronagraph of some sort ( see @xcite for a review on coronagraphs ) .
thus we clearly separate the gain in contrast that can be obtained by reducing the diffracted light with the coronagraph on one hand , and by fighting the scattered light with the speckle nulling technique on the other hand .
the purpose of speckle nulling is to reduce the speckle noise in a central region of the image plane .
this region , the dark hole , then becomes dark enough to enable the detection of companions much fainter than the original speckles .
speckle nulling is achieved by way of a servo system that has a deformable mirror as actuator .
because our sensing method requires dm actuation and is better understood with the knowledge of the command control theory , we first model the deformable mirror ( [ sub : deformable_mirror ] ) , then present two algorithms for the command control ( [ sub : field_nulling ] & [ sub : energy_min ] ) , and conclude with the sensing method ( [ sub : measurement ] ) .
the deformable mirror ( dm ) in @xcite consists of a continuous facesheet supported by @xmath0 actuators arranged in a square pattern of constant spacing .
this dm format is well adapted to either square or circular pupils , the only pupil shapes that we consider in this paper .
we assume that the dm is physically located in a plane that is conjugate to the entrance pupil .
however , what we call dm in the following is the projection of this real dm in the entrance pupil plane .
the projected spacing between actuators is denoted by @xmath20 .
we assume that the optical magnification is such that the dm projected size is matched to the entrance pupil , i.e. @xmath21 , where @xmath22 is either the pupil side length or its diameter .
the dm surface deformation in response to the actuation of actuator @xmath23 is described by an _ influence function _ , denoted by @xmath24 . the total phase change introduced by the dm ( dm phase function ) is @xmath25 where @xmath26 are actuator strokes ( measured in radians ) .
note that contrary to the complex aberration function @xmath11 , the dm phase function is purely real . with an ideal coronagraph and a dm ,
the image - plane electric field formerly given by ( [ eq : e_im ] ) becomes @xmath27 in the next two sections , we explore two approaches for speckle nulling . in
[ sub : field_nulling ] , we begin naively by trying to cancel @xmath28 . because there is a maximum spatial frequency that the dm can correct for , the dh has necessarily a limited extension .
any energy at higher spatial frequencies will be aliased in the dh and limit its depth . therefore
, the dm can not be driven to cancel @xmath28 , unless @xmath19 is equal to zero outside the dh ( i.e. unless there are already no speckles outside the dh ) .
with this in mind , we start over in [ sub : energy_min ] with the idea that speckle nulling is better approached by minimizing the field energy .
the speckle field nulling approach consists in trying to null out @xmath28 in the dh region ( @xmath5 ) , meaning we seek a solution to the equation @xmath29 although , as we shall show , this equation has no exact solution unless @xmath19 happens to be a band - limited function within the controllable band of the dm . by replacing @xmath30 with its expression ( [ eq : psi ] ) , we obtain @xmath31 we recognize in ( [ eq : field2 ] ) a linear system in the @xmath26 that could be solved using various techniques such as singular value decomposition ( svd ; * ? ? ? * ) .
although general , this solution does not provide much insight in the problem of speckle nulling .
for this reason , let us examine now a different solution , less general but with more explanatory power .
we will comment on the use of svd at the end of this section .
we consider a square pupil . in this case , all dm actuators receive light and the pupil function has no limiting effect on the dm phase function , i.e. @xmath32 . moreover , we assume all influence functions to be identical in shape , and write @xmath33 . under these hypotheses , @xmath34 where @xmath35 is dirac s bidimensional distribution , and @xmath36 denotes the convolution . substituting @xmath37 by the ft of ( [ eq : psi2 ] ) in ( [ eq : field1 ] ) yields @xmath38 we recognize in the left - hand side of ( [ eq : field3 ] ) a truncated fourier series .
if we choose the @xmath26 to be the first @xmath39 fourier coefficients of @xmath40 , i.e. @xmath41}^2 } -\frac{\widehat{p\phi}(\alpha,\beta)}{\hat{f}(\alpha,\beta ) } \ : e^{i \frac{2\pi d}{\lambda } ( k \alpha + l \beta ) } \ , \mathrm{d}\alpha \ , \mathrm{d}\beta,\ ] ] then according to fourier theory , we minimize the mean - square error between both sides of the equation ( see e.g. * ? ? ?
this error can not be reduced to zero unless the fourier coefficients of @xmath40 happen to vanish for @xmath42 and @xmath43 . at this point
, we have reached the important conclusion that _ perfect speckle cancellation can not be achieved with a finite - size dm unless the wavefront aberrations are band - limited_. moreover , we can assert that the maximum dh extension is the square domain @xmath44}^2 = { [ -\frac{n}{2}\frac{\lambda}{d } , \frac{n}{2}\frac{\lambda}{d}]}^2 $ ] .
solution ( [ eq : coef1 ] ) is physically acceptable only if the fourier coefficients are real numbers , which means mathematically that @xmath45 should be hermitian is said to be hermitian if @xmath46 .
the ft of a real function is hermitian and vice versa . ] .
if there are phase aberrations only , @xmath47 is real , @xmath45 is hermitian , and the @xmath26 are real .
this is no longer true if there are amplitude aberrations as well , reflecting the fact that the dm alone can not correct both phase and amplitude aberrations in @xmath5 .
however , by considering the hermitian function that is equal to @xmath45 in one half of the dh , say @xmath48 \times [ -\frac{\lambda}{2d},\frac{\lambda}{2d}]$ ] , we obtain the real coefficients @xmath49 \mathrm{d}\alpha \ , \mathrm{d}\beta,\ ] ] that correct both amplitude and phase aberrations in @xmath50 . as we have @xmath51
, the dh has a size of @xmath0 resolution elements ( resels ) with phase aberrations only , and of @xmath52 resels with phase and amplitude aberrations .
therefore , _ a dm can correct both amplitude and phase aberrations in the image plane , albeit in a region that is either the left , right , upper , or lower half of the phase - corrected region . _
as @xcite pointed out , let us remind the reader that @xmath53 is equal to the nyquist frequency for a sampling interval @xmath54 .
therefore , we find that maximum extension for the dh corresponds to the range where the sampling theorem applies to the wavefront at the dm actuator scale . indeed , taking the inverse ft of ( [ eq : field3 ] ) leads to the wavefront reconstruction formula @xmath55 again , this reconstruction can not be perfect unless the spectrum of @xmath47 is contained in @xmath5 .
note that because @xmath56 is generally not a flat function ( as it would be the case if influence functions were for instance 2d sinc functions ) , actuator strokes are not equal to the negative of wavefront values sampled at the actuator locations .
our fourier solution was derived by assuming that ( a ) all influence functions are identical in shape , and ( b ) that the pupil has a square shape .
hypothesis ( a ) appears to be reasonable at least for the dm in use on the hcit ( joseph green , personal communication ) , but this remains to be precisely measured .
if hypothesis ( b ) is relaxed then ( i ) some actuators do not receive any light and play no role , so there are effectively fewer terms in the summation in ( [ eq : field3 ] ) , and ( ii ) the fact that influence functions on the pupil boundary are only partly illuminated is ignored .
now that we have two methods to solve ( [ eq : field1 ] ) , fourier expansion and svd , let us compare their solutions .
we deal here with functions belonging to the hilbert space of square integrable functions @xmath57 .
this space has @xmath58 for dot product , and @xmath59 for norm .
as mentioned earlier , fourier expansion minimizes the mean - square error between both sides of ( [ eq : field3 ] ) , i.e. @xmath60 . by contrast , svd has the built - in property of minimizing the norm of the residuals of ( [ eq : field2 ] ) , i.e. @xmath61 . in other words ,
svd minimizes @xmath62 , the speckle field energy , which seems more satisfactory from a physical point of view . to find out what is best , we have performed one - dimensional numerical simulations .
it turns out that svd yields dark holes 50% deeper ( median value ) than fourier expansion .
in addition , svd does not require all influence functions to have the same shape .
however , considering four detector pixels per resel in two dimensions ( critical sampling ) , svd would require us to manipulate matrices as large as @xmath63 ( or even @xmath64 when real and imaginary parts are separated ) .
such matrices would occupy 537 mb of memory space for @xmath65 actuators and single - precision floating - point numbers .
by contrast , fourier expansion would be straightforwardly obtained with ffts of @xmath66 arrays at critical sampling , but again at the cost of a 50% shallower dark hole and a strong hypothesis on the influence functions . in the next section
, we seek to find a computationally less intensive solution that still minimizes the speckle energy in the dark hole , but does not require any hypothesis on the influence functions .
let us start with the idea that the best solution is defined as the one _ minimizing the total energy of the speckle field in the dh_. for the sake of simplicity , we assume once again a square pupil , but not necessarily a common shape for the influence functions .
the total energy in the speckle field reads @xmath67 using the same notation as in [ sub : field_nulling ] . given that @xmath68 , the energy is minimized when @xmath69 where @xmath70 stands for the real part .
note that this is less demanding than ( [ eq : field1 ] ) , as ( [ eq : field1 ] ) implies ( [ eq : energy2 ] ) but the reverse is not true . using the definition ( [ eq : psi ] ) for @xmath30 and realizing that @xmath71 is a real number together with the symmetry of @xmath5 . ] , we get finally @xmath72 as in ( [ eq : field2 ] ) , we find a system that is linear in the actuator strokes . by replacing double indices with single ones ,
e.g. @xmath73 becomes @xmath74 , ( [ eq : energy3 ] ) can be solved in matrix format by inverting a @xmath1 real matrix .
this is already an improvement with respect to the @xmath63 complex matrix required by svd in the previous section .
it appears that the same solution can be obtained with a much less demanding @xmath75 matrix inversion , provided two - dimensional influence functions can be written as the tensor product of two one - dimensional functions ( separation of variables ) , i.e. @xmath76 .
this would be the case for box functions or bidimensional gaussians , and is good at the 5% level for the dm in use on the hcit .
this property also holds in the image plane since the ft of the previous equation yields @xmath77 . by separating variables , ( [ eq : energy3 ] ) becomes @xmath78 as the left - hand side happens to be the product of three @xmath0 matrices , ( [ eq : energy4 ] ) can be rewritten as an equality between square matrices .
@xmath79 for square - box and actual hcit influence functions , numerical calculations show that @xmath80 is diagonally dominant$ ] is said to be diagonally dominant if @xmath81 .
] and therefore invertible by regular gaussian elimination .
the solution to ( [ eq : energy5 ] ) is then @xmath82 note that @xmath83 can be precomputed and stored , so that computing the strokes effectively requires only two matrix multiplications .
as shown in appendix [ app : global ] , an equivalent result can be obtained by working with pupil plane quantities . as for the field nulling approach , correcting amplitude errors as well implies restricting the dark hole to either @xmath84 \times [ -\frac{n}{2}\frac{\lambda}{d},\frac{n}{2}\frac{\lambda}{d}]$ ] or @xmath85 \times [ -\frac{n}{2}\frac{\lambda}{d},\frac{n}{2}\frac{\lambda}{d}]$ ] .
to account for amplitude errors and keep the formalism we have presented so far , it is sufficient to replace @xmath19 by a function equal to @xmath86 in either @xmath50 or @xmath87 ( depending on the half where one wishes to create the dark hole ) , and equal to @xmath88 in the other half ( hermitian symmetry ) . because its ft is hermitian
, the new aberration function in the pupil plane is real , and thus the algorithm processes amplitude and phase errors at the same time as if there were phase errors only . let us derive the residual total energy in the dh after the correction has been applied .
starting from definition ( [ eq : energy1 ] ) and rewriting condition ( [ eq : energy2 ] ) as @xmath89 , we find @xmath90 the former term is the initial speckle energy in the dh , while the latter is the speckle energy decrease gained with the dm .
mathematically , @xmath91 measures the distance ( according to the norm we have defined ) between the speckle field and its approximation with the dm inside the dh . because there is no exact solution to ( @xmath92 ) , the residual energy can not be made equal to zero in @xmath50 or @xmath87 .
however , the energy approach offers an additional degree of freedom : by reducing concentrically the domain over which the energy is minimized , the speckle energy can be further decreased ( see [ sec : simulations ] ) .
so far , our speckle nulling theory has presupposed the knowledge of the speckle field @xmath19 , or equivalently of the phase and amplitude aberrations across the pupil , embodied in the complex phase function @xmath47 . in this section ,
we show how the speckle field can be measured directly in the image plane .
as the detector measures an intensity , a single image yields only the modulus of the speckle field .
the phase of the speckle field can be retrieved by perturbing the phase function @xmath47 in a controlled way , and by recording the corresponding images , a process analogous to _ phase diversity _ ( e.g. * ? ? ?
* ) . in our system
, the dm provides the natural means for creating this controlled perturbation .
as we will see , exactly three images obtained with well chosen dm settings provide enough information to measure @xmath19 .
let us call image 0 the original image recorded with a setting @xmath93 , whereas images 1 and 2 are recorded with settings @xmath94 and @xmath95 .
to be general , we consider in the field of view the presence of an exoplanet and an exozodiacal cloud ( exozodi for short ) , in addition to the star itself .
the electric fields of these objects are incoherent with that of the star , so their intensities should be added to the star s intensity .
because they are much fainter than the star , the speckles they produce are negligible with respect to the star speckles , and their intensities can be considered as independent of @xmath11 and @xmath30 .
the total intensity of every image pixel @xmath17 takes then the successive values @xmath96 where @xmath97 and @xmath98 are the exoplanet and exozodi intensities , respectively .
system ( [ eq : i_system1 ] ) can be reduced to the linear system @xmath99 where the exponent @xmath36 denotes the complex conjugate .
notice how the exoplanet and exozodi intensities have disappeared from the equations , demonstrating that faint objects do not affect the measurement process of stellar speckles .
however , note that because of quantum noise , the planet detection can still be problematic if the exozodi is much brighter than the planet .
now , system ( [ eq : i_system2 ] ) admits a unique solution if its determinant , @xmath100 is not zero , that is to say if @xmath101 ) tells us that the dm setting changes , @xmath102 and @xmath103 , should modify the speckles differently in any given pixel , otherwise not enough information is secured to measure unambiguously @xmath19 in this pixel .
it should be expected for this method to work practically that the magnitude of the speckle modification be greater than the photon noise level .
we have not yet found a rigorous derivation of the optimum values for the amplitude @xmath104 and @xmath105 , but a heuristic argument suggests to us that the optimum perturbations may be that @xmath106 and @xmath107 .
that is to say , the dm - induced speckle instensity pattern , taken by itself , should be approximately the same as the original speckle intensity pattern .
thus at each pixel we choose @xmath108 , with the caveat that neither should be zero to keep ( [ eq : i_condition ] ) valid .
the phase of @xmath109 does not matter , but the phase difference between @xmath109 and @xmath110 should be made as close to @xmath111 as possible to keep @xmath112 from zero .
practically , this can be realized as follows : 1 .
compute @xmath102 stroke changes from ( [ eq : coef2 ] ) or ( [ eq : energy6 ] ) by replacing @xmath19 by @xmath113 , where @xmath114 is a random phase ; 2 .
compute @xmath103 stroke changes from ( [ eq : coef2 ] ) or ( [ eq : energy6 ] ) by replacing @xmath19 by @xmath115 .
now that we have made sure that @xmath116 , we derive finally @xmath117 equation ( [ eq : i_solution ] ) shows that the initially unknown speckle field ( @xmath19 ) can be experimentally measured in just three exposures taken under identical circumstances but with different shapes imposed on the dm .
in this section , we perform one- and two - dimensional simulations for the theoretical case of white speckle noise caused by phase aberrations only .
the dm has 64 actuators and top - hat influence functions .
smoother influence functions have been tested and do not lead to qualitatively different results .
a simulation with actual hcit influence functions will be presented in the next section .
the simulated portion of the pupil plane is made twice as big as the pupil by zero padding , so that every element of resolution in the image plane would be sampled by two detector pixels .
this corresponds to the realistic case of a photon - starved exoplanet detection where read - out noise must be minimized .
figure [ fig : f1 ] shows a complete one - dimensional simulation including speckle field measurement ( [ sub : measurement ] ) and speckle nulling with field nulling ( [ sub : field_nulling ] ) and energy minimization ( [ sub : energy_min ] ) .
the standard deviation of the phase aberrations is set to @xmath118 .
intensities are scaled with respect to the maximum of the star psf in the absence of a coronagraph .
ideal conditions are assumed : no photon noise , noiseless detector , and perfect precision in the control of dm actuators . under these conditions , the speckle field
is perfectly estimated , and the mean intensity in the dh is @xmath119 with field nulling and @xmath120 with energy minimization , i.e. about 1500 and 6500 times lower than the mean intensity outside the dh , respectively .
repeated simulations with different noise sequences show that energy minimization performs always better than field nulling by a factor of a few .
field nulling solved with svd yields the same numerical solution as energy minimization ( they differ by the last digit only ) , in agreement with the idea that they both minimize speckle energy . in the one - dimensional case , it is easy to predict roughly the shape and the depth of the dh .
the function @xmath121 is band - limited since the pupil has a finite size . as the pupil linear dimension is @xmath122 , the maximum spatial frequency of @xmath121 is @xmath123 .
let us apply the sampling theorem at the nyquist sampling frequency @xmath122 , and write @xmath124 \mbox{sinc } \ !
\left ( \frac{\alpha d}{\lambda } - n \right),\ ] ] where the subscript @xmath125 denotes the function value for @xmath126 . substituting @xmath127 by @xmath128 and @xmath20 by @xmath129 leads to @xmath130 the field nulling equation ( [ eq : field1 ] ) takes here the discrete form @xmath131 i.e. actuator strokes are computed thanks to an inverse fft .
let us now turn to the residual speckle field @xmath132 because the sinc function decreases rapidly with @xmath127 , the terms flanking the dh ( @xmath133 and @xmath134 ) should by themselves give the order of magnitude of the residual speckle field in the dh . in case of phase aberrations only and
white noise , we have @xmath135 , where @xmath136 is the mean intensity in the image plane prior to the dh creation .
therefore , a crude estimate of the intensity profile in the dh should be @xmath137}^2.\ ] ] we have superimposed this approximation as a thick line in fig . [
fig : f1 ] . in this case
the match is remarkable , but more simulations show that it is generally good within a factor of 10 only .
nevertheless , it demonstrates that the dh depth depends critically on the residual speckle field at its edges , hence on the decreasing rate of the complex aberration spectrum with spatial frequency . in that respect ,
a white spectrum is certainly the worst case .
equation ( [ eq:1d5 ] ) further indicates that the dh depth depends on the number of actuators : as @xmath138 is increased , the dh widens and gets deeper . with 8 , 16 , 32 , and 64 actuators , ( [ eq:1d5 ] ) predicts @xmath139 to reach about 100 , 300 , 1000 , and 4500 . as mentioned in
[ sub : energy_min ] , speckle nulling by energy minimization can be performed in a region narrower than the maximum dh .
figure [ fig : f2 ] illustrates this point : by reducing the search area from 64 to 44 resels ( 31% reduction ) , the dh floor was decreased from @xmath120 to @xmath140 , i.e. a gain of about 5200 in contrast ( further reducing the search area does not bring any significant gain ) . by giving up search space ,
one frees the degrees of freedom corresponding to the highest spatial frequency components on the dm pattern .
these can be used to improve the dh depth at lower spatial frequency because of the psf angular extension ( this is essentially the same reason why high spatial frequency speckles limit the dh depth ) .
as the search space is reduced , the leverage of these highest spatial frequency components decreases ( psf wings falling off ) .
the energy minimization algorithm compensates by putting more energy at high frequency ( see lower panel in fig . [
fig : f2 ] ) , which produces increasingly oscillatory dm patterns ( see top panel in fig . [
fig : f2 ] ) and increasingly brighter spots in the image ( around @xmath141 and @xmath142 in bottom panel of fig .
[ fig : f2 ] ) .
thus the trade - off range might be limited in practice by the maximum actuator stroke ( currently @xmath143 m on the hcit ) , and/or by the detector s dynamic range . in two - dimensions ,
the trade - off limits are well illustrated by the following example : considering a @xmath65 dm and a random wavefront , we find that the dh floor can be decreased from @xmath144 to @xmath145 ( a factor of 17 ) if the search area is reduced from @xmath65 to @xmath146 resels ( 12% reduction in area ) .
this implies a maximum actuator stroke of 10 nm and a detector dynamic range of @xmath3 . a further reduction to @xmath147 resels
does not feature a lower dh floor ( @xmath148 ) , and would imply a maximum actuator stroke of @xmath149 m and a detector dynamic range of @xmath150 . in this case , the leverage of the additionally freed high - spatial frequency components is so weak that the algorithm starts diverging . in figs .
[ fig : f3][fig : f4 ] , we show an example of two - dimensional speckle nulling with phase and amplitude aberrations for a square pupil .
to reflect the fact that phase aberrations dominate amplitude aberrations in real experiments ( see * ? ? ?
* ) , the rms amplitude of amplitude aberrations is made ten times smaller than that of phase aberrations ( the choice of a factor ten is arbitrary ) .
the dh is split into two regions : in the right one ( @xmath50 ) , amplitude and phase aberrations are corrected , whereas in the left one ( @xmath87 ) , phase aberrations are corrected and amplitude aberrations are made worse by a factor of four in intensity . with the @xmath151-m tpf - c primary mirror in mind , we have studied the phase aberration map of an actual 8-m mirror : the primary mirror of antu , the first 8.2-m unit telescope of eso s very large telescope ( vlt ) .
this phase map was obtained with the active optics system on , and is characteristic of zonal errors ( aberrations which can not be fitted by low - order zernike - type polynomials ) .
it can be seen in fig .
[ fig : f5 ] that the azimuthally averaged power spectral density ( psd ) of such a map is well represented by @xmath152 where @xmath153 .
values for psd@xmath154 , @xmath155 and @xmath156 are listed in table [ tab : t1 ] . for comparison ,
the same treatment has been applied to the hubble space telescope ( hst ) zonal error map from @xcite .
we conclude from this study that _ a realistic phase aberration psd for an 8-m mirror decreases as the third power of the spatial frequency_. the standard deviation of the vlt phase map is 20.9 nm ( 18.5 nm for hst )
. the square root of the power of phase aberrations in the 0.54 m@xmath157 spatial frequency range ( 432 @xmath158 for an 8-m mirror ) is 19.4 nm , i.e. about @xmath159 at 500 nm , clearly not in the validity domain of our linear approximation .
figure [ fig : f6 ] shows a simulation performed in the same conditions as fig . [ fig : f1 ] , but with a vlt - like psd .
the psd is scaled so that the standard deviation of phase aberrations is equal to @xmath118 .
the average dh floor is now @xmath160 , six orders of magnitude below the intensity peak in the original image ! in agreement with [ sub : sim_white ] , we find that _ the dh s depth depends critically on the magnitude of the speckle field at the edge of the dh , hence on the decrease of the phase aberration psd with spatial frequency_. for the two - dimensional simulation in figs .
[ fig : f7][fig : f8 ] , we have kept the original vlt phase map and circular pupil , but scaled the standard deviation of phase aberrations to @xmath118 .
in addition , we have used the actual hcit influence functions from @xcite .
the average dh floor is then @xmath161 with field nulling ( case shown ) , and @xmath162 with energy minimization .
the worse performance of the second method reflects the cost of the variable separation hypothesis , only accurate to within 5% for the hcit .
note that the dh retains its square shape with a circular pupil , as the dh shape is fixed by the actuator grid geometry on the dm ( a square grid of constant spacing in our case ) .
in [ sec : simulations ] , we presented noise - free simulations .
to give an idea of the effect of quantum and read - out noises , let us consider a sun - like star at 10 pc observed by a @xmath151 m space telescope with a 5% overall efficiency . in a 100 nm bandwith centered at 600 nm , the telescope collects about @xmath163 photo - electrons in one - hour exposures . considering the quantum noise , a 1 e@xmath164 read - noise and ignoring chromatic effects , simulations of sequences of four one - hour exposures
show that the average dh floor in fig .
[ fig : f1 ] would jump from @xmath120 to @xmath165 , whereas the average dh floor in fig .
[ fig : f6 ] would jump from @xmath166 to @xmath167 . in practice
, our speckle nulling process will work as stated provided eq .
( [ eq : e_im ] ) holds , that is to say if @xmath168 .
if @xmath169 is the improvement in contrast with respect to the speckle floor and @xmath170 the standard deviation of wavefront aberrations in radians , this condition translates into @xmath171 , or @xmath172 . in terms of optical path difference ,
the standard deviation should then be much less than @xmath173 for @xmath174 .
this is why we considered @xmath118 rms wavefronts in our simulations .
as the wavefront will probably not be of this quality at the start , the speckle nulling method presented here is intended to be used in the course of observations , after a first phase where the bulk of the aberrations have been taken out .
when the linear approximation breaks down , three images with different dm settings still provide enough information about the aberrations , so that a dh could be created thanks to a global non - linear analysis of these images @xcite .
@xcite also explored non - linear solutions , but with many more iterations ( @xmath175 ) .
dwelling on the validity of eq .
( [ eq : e_im ] ) , real coronagraphs would not only remove the direct image of the star ( @xmath18 ) , they would also modify the speckle field ( @xmath19 ) and the dm phase function ( @xmath37 ) .
this can be easily incorporated in the theory .
a more delicate point is that real coronagraphs are not translation - invariant systems . as a consequence ,
effective influence functions as seen from behind the coronagraph will vary over the pupil . for image - plane coronagraphs with band - limited sinc masks
@xcite , we estimate this variation to be of the order of 10% , assuming @xmath176 and 64 actuators
. only energy minimization , not field nulling ( unless solved with svd ) , can accomodate this effect .
what about the precision at which actuators should be controlled ? as a consequence of the linearity of ( [ eq : energy3 ] ) , the dh depth depends quadratically on the precision on the actuator strokes .
we deduce and this is confirmed by simulations that a four orders of magnitude deep dh can only be obtained if the strokes are controlled at a 1% precision , i.e. 6 pm rms with @xmath118 aberrations at 600 nm .
this precision corresponds to the current resolution of the actuator drivers on the hcit .
regarding instrumental stability , we assumed that the instrument would remain perfectly stable during the four - step process . however , despite the foreseen thermal and mechanical controls of the spacecraft , very slow drifts during the few hours of single exposures should be expected .
therefore we intend to study in a subsequent paper how to incorporate a model of the drifts in our method .
the exact parameters of this model would be derived from a learning phase after the launch of the spacecraft .
we have not considered the effect of chromaticity .
let us point out that phase aberrations due to mirror surface errors scale with wavelength , so the correction derived from one wavelength would apply to all wavelengths .
this is unfortunately not the case for amplitude aberrations .
although these are weaker than phase aberrations , a degradation of the correction should be expected in polychromatic light . moreover , polychromatic wavefront sensing will require a revised theory as speckles will move out radially in proportion to the wavelength .
in this paper , we presented two techniques to optimally null out speckles in the central field of an image behind an ideal coronagraph in space .
the measurement phase necessitates only three images , the fourth image being fully corrected .
depending on the number of actuators and the desired search area , the gain in contrast can reach several orders of magnitude .
these techniques are intended to work in a low aberration regime , such as in the course of observations after an initial correction phase .
they are primarily meant to be used in space but could be implemented in a second - stage ao system on ground - based telescopes . out of these two methods , the speckle energy minimization approach seems to be the more powerful and flexible : ( i ) it offers the possibility to trade off some search area against an improved contrast , and ( ii ) it can accomodate influence function variations over the pupil ( necessary with real coronagraphs )
. if influence functions feature the required symmetry ( variable separation ) , it is computationally very effective , but is otherwise still better than svd . since the principles underlying these speckle nulling techniques are general , it should be possible to use them in conjunction with most coronagraph designs , including those with band - limited masks @xcite , pupil - mapping @xcite , and shaped pupils @xcite .
it is our intent to complete our work by integrating in our simulations models of these coronagraphs , and to carry out experiments with the hcit .
in addition , we will seek to incorporate in the measurement theory a linear model for the evolution of aberrations , and we will work toward a theory accommodating the spectral bandwidth needed for the detection and spectroscopy of terrestrial planets .
we wish to thank the anonymous referee for his insightful comments that helped to improve greatly the content of this paper .
we acknowledge many helpful discussions with chris burrows , john trauger , joe green , and stuart shaklan .
this work was performed in part under contract 1256791 with the jet propulsion laboratory ( jpl ) , funded by nasa through the michelson fellowship program , and in part under contract 1260535 from jpl .
jpl is managed for nasa by the california institute of technology .
this research has made use of nasa s astrophysics data system .
in this appendix , we show that energy minimization can be formulated in the pupil plane as well . note that no measurement takes place in the pupil plane : the aberration function @xmath47 is obtained by the inverse ft of @xmath19 which is still measured as described in [ sub : measurement ] .
although we present here a solution for phase aberrations , amplitude aberrations can be corrected in half of the domain without changing the formalism , exactly as explained in
[ sub : energy_min ] . by replacing @xmath177 with its expression as a ft ,
the dark hole energy reads @xmath178 now we invert the integration order and integrate over @xmath5 to get @xmath179 where @xmath180 $ ] .
the energy is minimized when @xmath181 \ , h(u'-u)\ , h(v'-v ) \ , \mathrm{d}u \ , \mathrm{d}v \ , \mathrm{d}u ' \mathrm{d}v ' = 0.\ ] ] in the next steps , we first replace @xmath30 with ( [ eq : psi ] ) , then @xmath182 with its tensor product @xmath183 in order to integrate separately the variables @xmath184 and @xmath185 .
the final result reads @xmath186 system ( [ eq : app4 ] ) has a form identical to ( [ eq : energy5 ] ) and can be solved with the same technique .
bord , p. , traub , w. , & trauger , j. 2004 , proceedings of the second tpf / darwin conference , meeting held in san diego , usa , july 2629 , 2004 , http://planetquest1.jpl.nasa.gov/tpfdarwinconf/confproceedings.cfm
|
high - contrast imaging from space must overcome two major noise sources to successfully detect a terrestrial planet angularly close to its parent star : photon noise from diffracted star light , and speckle noise from star light scattered by instrumentally - generated wavefront perturbation .
coronagraphs tackle only the photon noise contribution by reducing diffracted star light at the location of a planet .
speckle noise should be addressed with adaptative - optics systems . following the tracks of @xcite
, we develop in this paper two analytical methods for wavefront sensing and control that aims at creating _ dark holes _ , i.e. areas of the image plane cleared out of speckles , assuming an ideal coronagraph and small aberrations .
the first method , _ speckle field nulling _ , is a fast fft - based algorithm that requires the deformable - mirror influence functions to have identical shapes .
the second method , _ speckle energy minimization _
, is more general and provides the optimal deformable mirror shape via matrix inversion . with a @xmath0 deformable mirror ,
the size of matrix to be inverted is either @xmath1 in the general case , or only @xmath0 if influence functions can be written as the tensor product of two one - dimensional functions .
moreover , speckle energy minimization makes it possible to trade off some of the dark hole area against an improved contrast . for both methods , complex wavefront aberrations ( amplitude and phase )
are measured using just three images taken with the science camera ( no dedicated wavefront sensing channel is used ) , therefore there are no non - common path errors .
we assess the theoretical performance of both methods with numerical simulations including realistic speckle noise and experimental influence functions .
we find that these speckle nulling techniques should be able to improve the contrast by several orders of magnitude .
|
stochastic perturbations are typical for any physical realization of a given dynamical system .
also round - off errors , inevitable in numerical investigation of any dynamics , may be considered as a random noise .
quantitative characterization of dynamical systems with external stochastic noise is a subject of several recent studies @xcite . on the other hand ,
the influence of noise on various low dimensional dynamical systems and the properties of random dynamical systems have been extensively studied for many years @xcite .
consider a discrete dynamical system generated by @xmath0 , where @xmath1 is a subset of @xmath2 , in the presence of an additive noise @xmath3 where @xmath4 are independent random vectors fulfilling @xmath5 and @xmath6 .
the case with vanishing noise strength @xmath7 will be called the deterministic limit of the model .
properties of such stochastic systems have recently been analyzed by means of the periodic orbit theory @xcite .
convergence of invariant measures of the noisy system in the deterministic limit has been broadly discussed in the mathematical literature ( see for instance @xcite ) . a dynamical system generated by @xmath8
is called chaotic if its kolmogorov - sinai ( ks ) entropy is positive @xcite .
such a definition is not applicable for stochastic systems , characterized by infinite entropy . in this case
the partition dependent entropy diverges if the partition @xmath9 of the space @xmath1 is made increasingly finer . in this paper
we propose a generalization of the ks - entropy for systems with additive noise ( [ langev ] ) . since entropy diverges also for the pure noise ( with the trivial deterministic limit @xmath10 , for @xmath11 )
, we study the difference between the total entropy of the system with noise and the entropy of the noise itself .
firstly , we set the partition fixed , and then we take the supremum over all finite partitions with regular cell boundaries . in this way
our definition resembles the _ coherent states dynamical entropy _ , two of us proposed several years ago @xcite as a measure of quantum chaos .
the entropy of the noise , discussed in this paper , plays the role of entropy of quantum measurement , connected with the overlap of coherent states and linked to the heisenberg uncertainty relation .
even though our definition is suitable for @xmath12dimensional systems with an arbitrary additive noise , we demonstrate its usefulness on simple one dimensional systems .
we choose a specific kind of distribution defining the noise , which can be expanded in a finite basis of @xmath13 functions in both variables @xmath14 and @xmath15 .
this condition allows us to express the @xmath16-steps probabilities , required to compute the entropy , as a product of certain matrices .
moreover , we represent the frobenius perron operator of the system with noise by an @xmath17 matrix , and obtain its spectrum by numerical diagonalization .
the deterministic limit @xmath7 requires @xmath18 , which resembles the classical limit of quantum mechanics .
this paper is organized as follows . in sect .
ii the dynamical entropy for noisy systems is defined and some of their properties are analyzed .
one dimensional systems with expandable noise and their invariant measures are analyzed in sect .
iii , while different methods of computing the entropy are presented in sect .
the entropy for some exemplary systems with noise ( rnyi map , logistic map ) is studied in sect . v. the paper
is concluded by sect .
vi , while an illustrative iterated function system , used for computation of the entropy of noise , is provided in appendix a.
let us consider a partition @xmath19 of @xmath1 into @xmath20 disjoint cells .
the partition generates the symbolic dynamics in the @xmath20-symbol code space .
every @xmath16steps trajectory can be represented by a string of @xmath16 symbols , @xmath21 , where each letter @xmath22 denotes one of the @xmath20 cells . assuming that initial conditions are taken from @xmath1 with the distribution @xmath23 invariant with respect to the map @xmath8 , let us denote by @xmath24 the probability that the trajectory of the system can be encoded by a given string of symbols , i.e. , @xmath25 the _ partial entropies _ @xmath26 are given by the sum over all @xmath27 strings of length @xmath16 @xmath28 while the _ dynamical entropy of the system @xmath8 with respect to the partition _ @xmath9 reads @xmath29 the above sequence is decreasing and the quantity @xmath30 , \label{h1}\ ] ] which depends on @xmath8 only via @xmath23 , is just the _ entropy of the partition _ @xmath9 .
we denote it by @xmath31 .
entropy _ of the system @xmath8 is defined by the supremum over all possible partitions @xmath9
@xcite @xmath32 a partition for which the supremum is achieved is called a _ generating _ partition .
knowledge of a @xmath33elements generating partition for a given map allows one to represent the time evolution of the system in @xmath33letters symbolic dynamics and to find the upper bound for the ks - entropy : @xmath34 . in the general case it is difficult to find a generating partition and
one usually performs another limit , tending to zero with the diameter of the largest cell of a partition , which implies the limit @xmath35 .
we shall denote this limit by @xmath36 . for simplicity
we consider one - dimensional case taking @xmath37 $ ] , imposing periodic boundary conditions and joining the interval into a circle .
we denote the lebesgue measure on @xmath1 by @xmath38 , setting @xmath39 ( clearly @xmath40 . the noisy system introduced in ( [ langev ] )
will be denoted by @xmath41 . from now on we assume that all the random vectors @xmath42 @xmath43 in ( [ langev ] ) have the same distribution with the density @xmath44
. then the probability density of transition from @xmath14 to @xmath15 under the combined action of the deterministic map @xmath8 and the noise is given by @xmath45 , where @xmath46 and the difference is taken @xmath47 . in the pure noise case ( @xmath48 ) this density depends only on the length of the jump and equals to @xmath49 .
we assume that @xmath41 has a unique invariant measure @xmath50 , which is absolutely continuous with respect to the lebesgue measure @xmath38 ( i.e. it has a density @xmath51 ) . clearly , @xmath52 , and so @xmath53 .
moreover , we assume that the measure @xmath50 tends weakly to @xmath23 respectively , for @xmath7 , where @xmath23 is some invariant measure for the deterministic system @xmath8 . in sect .
we discuss the situation , where the above assumptions are fulfilled .
now , let us fix a partition @xmath9 of @xmath1 .
we define the _ total entropy _ @xmath54 of the noisy system @xmath41 by formulae ( [ part ] ) and ( [ padep ] ) , analogously to the deterministic case . note , however , that in this case the initial conditions should be taken from @xmath1 with the measure @xmath50 . as we shall see below this entropy grows unboundedly with @xmath20 .
hence , we can not define partition independent entropy of the noisy system using formula ( [ parks ] ) , as the supremum in ( [ parks ] ) is equal to the infinity . on the other hand , there are two kinds of randomness in our model : the first is connected with the deterministic dynamics ; the second comes from the stochastic perturbation .
accordingly , we split the total partition dependent entropy @xmath55 of a noisy system @xmath56 given by ( [ part ] ) and ( [ padep ] ) into two components : the _ noise entropy _ and the _ dynamical entropy_. the latter quantity characterizes the underlying dynamics @xmath41 and is defined by @xmath57 where the entropy of the noise @xmath58 reads @xmath59 and @xmath60 is a stochastic system given by ( [ langev ] ) with @xmath48 ( _ pure noise _ ) .
although the both quantities @xmath55 and @xmath61 may diverge in the limit of fine partition @xmath62 ( @xmath63 ) for a nonzero noise strength , one can make their difference @xmath64 bounded , taking an appropriate sequence of partitions , as we shall see in the next subsection . in order to keep away from the ambiguity in the choice of a partition , we eventually define the _ dynamical entropy of _ @xmath41 as @xmath65 the supremum being taken over all finite partitions @xmath66 such that @xmath67 and @xmath68 for each @xmath69 , @xmath70
. we will call such partitions _ uniform_. the restriction to uniform partitions is necessary , since otherwise we may encounter various `` pathologies '' in the deterministic limit @xcite . note , that the uniformity assumption may be omitted in the case when all the measures @xmath50 , @xmath71 , and @xmath38 coincide .
it seems that in many cases the entropy of the noise ( [ noi - ent ] ) tends to zero in the deterministic limit @xmath72 , and the dynamical entropy of @xmath41 converges to the ks - entropy of the corresponding deterministic system @xmath8 ( for partial results in this direction see @xcite and @xcite , for numerical evidence see sect .
vd ) . in this section
we discuss the behavior of the dynamical entropy in the another limit , @xmath62 ( @xmath63 ) .
next , we introduce the _ boltzmann gibbs ( bg ) entropy _ of the noise : @xmath73 for interpretation and generalizations of this quantity , sometimes called _ continuous entropy _
, consult the monographs of guiau @xcite , martin and england @xcite , jumarie @xcite , or kapur @xcite . in the simplest case of the rectangular noise given by @xmath74 , for @xmath75 and @xmath11 , the bg - entropy is equal to ln @xmath76 .
note , that this quantity vanishes for the noise uniformly spread over the entire space ( @xmath77 ) , becomes negative for @xmath78 and diverges to minus infinity in the deterministic limit @xmath79 . for the system
@xmath80 combining the deterministic evolution @xmath8 and the stochastic perturbation , the probability density of transition from @xmath14 to @xmath15 during one time step is given by @xmath81 for @xmath82 .
the bg - entropy for this system can be defined as @xmath83 due to the homogeneity of the noise and due to the periodic boundary conditions the integral over @xmath15 in ( [ hbol ] ) does not depend on @xmath14 .
therefore for any system @xmath8 perturbed by a nonzero noise ( @xmath84 ) one obtains @xmath85 applying this equality and using the same methods as in @xcite we can prove that the total entropy fulfills the following inequalities ( the first inequality can be deduced from the lower bound for the variation of information obtained in theorem 2.3 from @xcite ; the second inequality comes from the definition ) @xmath86 for @xmath48 we get @xmath87 hence , both the total entropy @xmath88 and the noise entropy @xmath58 diverges logarithmically in the limit @xmath62 .
let us now study how does the dynamical entropy , which is the difference of these quantities , depend on the partition @xmath89 .
if the partition @xmath90 consists of one cell only ( @xmath91 ) , we have @xmath92 .
thus , the dynamical entropy with respect to this trivial partition equals zero for any system , which guarantees that the dynamical entropy given by the supremum over all partitions ( [ hdynsup ] ) is non negative .
let us now investigate the behavior of the total entropy in the opposite case @xmath62 ( @xmath63 ) for a non - zero noise strength @xmath93 . performing the time limit ( [ padep ] ) we find , as in @xcite , that for very fine partitions the total entropy of the system is given approximately by the sum of the boltzmann - gibbs entropy and the entropy of the partition ( this statement is again based on the theorem 2.3 from @xcite )
@xmath94 observe that due to the property ( [ hceqhcf ] ) the right - hand side does not depend on the dynamical system @xmath8 and the approximate equality ( [ totapprox ] ) holds also for the entropy of the noise @xmath58 .
therefore dynamical entropy tends to zero for both limiting cases @xmath95 let us now discuss , what are the minimal and maximal dynamical entropies admissible for a certain kind of stochastic noise . from ( [ bgineq ] ) and ( [ bgineqnoise ] ) we get @xmath96
thus the dynamical entropy is bounded from above by @xmath97 .
combining this with ( [ limdyn ] ) one obtain @xmath98 this relation provides a valuable interpretation of the boltzmann gibbs entropy .
this quantity , determined by the given probability distribution of the noise @xmath99 , tells us whether the character of the dynamics of a specific deterministic system @xmath8 can be resolved under the influence of this noise .
for example , the rectangular noise of width @xmath77 may be called _ disruptive _ , since the corresponding bg
entropy is equal to zero , and consequently @xmath100 for every uniform partition @xmath9 . under the influence of such a noise
we have no information , whatsoever , concerning the underlying dynamics @xmath8 .
furthermore , it is unlikely to distinguish between two systems , both having ks entropies larger than @xmath97 of the noise present .
evidently , in the deterministic limit the maximal entropy tends to infinity . on the other hand , in this case
, one obtains in ( [ limdyn ] ) the ks entropy .
this apparent paradox consists in the order of the two limits : the number of cells in coarse - graining to infinity and the noise strength to zero .
these two limits do not commute .
note that several authors proposed different approaches to the notion of dynamical entropy of noisy system .
crutchfield and packard @xcite introduced the _ excess entropy _ to analyze the difference between partial entropies of a noisy system and the corresponding deterministic system , and investigated its dependence on the noise strength @xmath93 and the number of the time steps @xmath16 . in order to avoid problems with the unbounded growth of the total entropy for sufficiently fine partitions gaspard and wang studied @xmath101entropy @xcite , where the supremum is taken only over the class of the partitions , for which the minimal diameter of a cell is larger than @xmath101 .
this quantity can be numerically approximated by the algorithm of cohen and procaccia @xcite .
the @xmath101-entropy diverges logarithmically in the limit @xmath102 ; the character of this divergence may be used to classify various kinds of random processes @xcite .
the dependence of the dynamical entropy on time yields another interesting problem . for discrete deterministic systems
the ks - entropy is additive in time : @xmath103 . on the other hand it follows from ( [ hmax ] ) that the dynamical entropy of a noisy system fulfills @xmath104 , for each time @xmath105 .
thus for a nonzero @xmath93 the ratio @xmath106 tends to zero in the limit @xmath107 , while for the deterministic dynamics @xmath108/t
= h_{ks}(f)$ ] .
the symbol @xmath109 represents the same deterministic system @xmath8 , subjected to the stochastic perturbation only once for @xmath105 time steps .
the related issue has been recently raised by fox @xcite in the context of deterministic evolution of gaussian densities .
the discontinuity of @xmath106 in the limit @xmath7 is a consequence of the fact that the another limits : time to infinity and noise strength to zero do not commute . in some sense
this resembles the noncommutativity of the limits _ time _
@xmath110 and @xmath111 in quantum mechanics , crucial for investigation of the so called _ quantum chaos _
( see e.g. @xcite ) .
continuing this analogy even further , the entropy of noise corresponds to the entropy of quantum measurement @xcite , while the boltzmann - gibbs entropy @xmath112 plays the role of the wehrl entropy @xcite , recently used by two of us ( ws , k ) to estimate the coherent states dynamical entropy @xcite .
we now discuss the computation of the entropy of noise for the rectangular noise @xmath113 ( see sect .
iic ) , with the periodic boundary conditions imposed .
computation of the transition probabilities in ( [ part ] ) reduces to simple convolutions of the rectangular noise and is straightforward for the first few time steps . for larger @xmath16
the calculations become tedious , and the convergence in the definition of entropy ( [ padep ] ) is rather slow ( not faster than @xmath114 ) .
it is hence advantageous to consider the sequence of _ relative entropies _ which converge much faster to the same limit @xmath115 @xcite .
for some systems the exponential convergence of this quantity was reported @xcite . in our analytical and numerical computations we used relative entropies @xmath116 . in all of the cases
studied , the term @xmath117 gives the entropy @xmath118 with a relative error smaller than @xmath119 . for a partition consisting of two equal cells ( @xmath120 ) and the rectangular noise @xmath121 we obtained an explicit expression for @xmath122 , as a function of the noise width @xmath76 .
analytical result obtained in @xcite are too lengthy to reproduce here gives an approximation of the entropy of noise with precision @xmath123 .
we analyzed the dynamical entropy of the rnyi map @xmath124_{\bmod 1}$ ] ( with integer parameter @xmath125 ) subjected to the rectangular noise .
independently of the noise strength , the uniform distribution remains the invariant density of this system .
for a large noise @xmath126 the dynamical entropy is close to zero , since the difference between the noise and the system with noise is hardly perceptible .
the dynamical entropy grows with the decreasing noise width @xmath76 and in the deterministic case seems to tend to the ks - entropy of the rnyi map @xmath127 . for more involved systems
the computation of the dynamical entropy becomes rather difficult even for this simple rectangular noise . in order to avoid calculating @xmath27 different integrals in ( [ part ] ) , in the subsequent section we introduce the class of noises for which computing of probabilities @xmath128 for any dynamical system reduces to multiplication of matrices .
in this section we define the particular discrete family of the probability distributions @xmath129 representing the noise and study properties of dynamical system subjected to this noise . as above ,
we consider one - dimensional space @xmath130 and impose periodic boundary conditions .
we shall look for a kernel @xmath131 homogeneous , periodic , and being decomposable in a finite basis @xmath132 for @xmath133 , where @xmath134 is a real matrix of expansion coefficients .
we assume that the functions @xmath135 and @xmath136 are continuous in @xmath130 and linearly independent .
consequently , we can uniquely express @xmath137 as their linear combinations .
both sets of base functions form an @xmath138-dimensional hilbert space .
the last property in ( [ expand ] ) is necessary in order to proceed with the matrix method of computation of the probabilities in ( [ part ] ) .
all these conditions are satisfied by the _ trigonometric noise _ @xmath139 where @xmath13 is even ( @xmath140 ) .
the normalization constant @xmath141 can be expressed in terms of the euler beta function @xmath142 or the double factorial @xmath143 we use basis functions given by @xmath144 where @xmath11 and @xmath145 .
we do not require their orthonormality .
expanding cosine as a sum to the @xmath13-th power in ( [ trignoise ] ) we find that the @xmath17 matrix @xmath146 , defined in ( [ expand ] ) , is diagonal for this noise @xmath147 the parameter @xmath13 controls the strength of the noise measured by its variance @xmath148 where @xmath149 stands for the derivative of the digamma function @xcite .
[ figtrig ] presents the densities of the noise for @xmath150 and @xmath151 .
the deterministic limit is obtained by letting @xmath13 tend to infinity .
since @xmath13 determines the size of the hilbert space , in which the evolution of the densities takes place , it can be compared with the quantum number @xmath152 used in quantum mechanics .
note that for every value of the parameter @xmath13 the probability function @xmath153 ( mod @xmath154 ) , so the analyzed perturbation is not _
local _ in the sense of blank @xcite .
it is worthwhile to mention that the properties ( [ expand ] ) are preserved for the kernel @xmath155 describing the dynamics of the system with noise ( [ langev ] ) .
the expansion matrix @xmath146 is the same , if one use the modified basis functions defined by @xmath156 , for @xmath11 , which explicitly depend on the deterministic dynamics @xmath8 . to illustrate some features of our model we plot in fig .
[ logdens ] the transition kernel @xmath155 for the logistic map perturbed by the noise defined in ( [ trignoise ] ) with @xmath157 .
the density of the invariant measure @xmath158 of the system ( [ langev ] ) is given as the eigenstate of the frobenius
perron operator @xmath159 corresponding to the largest eigenvalue equal @xmath154 .
for the deterministic system , the invariant density @xmath160 fulfills the formal equation @xmath161 in the presence of a stochastic perturbation this equation becomes @xmath162 where @xmath159 is the _ frobenius - perron ( fp ) operator _ connected with the noisy system ( [ langev ] ) .
let us assume that the kernel satisfies the conditions ( [ expand ] ) listed in the preceding subsection and it can be expanded as @xmath163 .
then we have @xmath164 v_l(y ) \\ & = & \displaystyle\sum_{r=0}^n \bigl[\int_0 ^ 1 u_r(f(x ) ) \rho(x ) dx \bigr ] \tilde{v}_{r}(y ) \end{array } \label{fp1}\ ] ] for @xmath165 , where @xmath166 .
thus , any initial density is projected by the fp
operator @xmath159 into the vector space spanned by the functions @xmath167 , and so its image may be expanded in the basis @xmath168 .
this statement concerns also the invariant density @xmath158 .
expanding @xmath158 @xmath169 with unknown coefficients @xmath170 and inserting this into ( [ fp1 ] ) we obtain the eigenequation for the vector of the coefficients @xmath171 @xmath172 the fp - operator is represented by the @xmath138 dimensional matrix @xmath173 , where @xmath146 is given by ( [ alr ] ) , and the entries of the matrix @xmath174 are given by @xmath175 for @xmath176 .
observe that @xmath146 does not depend on the deterministic dynamics @xmath8 , while @xmath174 depends on the noise via the basis functions @xmath177 and @xmath178 .
furthermore , note that @xmath179 for @xmath180 , and @xmath181 for each vector @xmath182 .
it follows from ( [ fp1 ] ) and ( [ fp3 ] ) that there is a one - to - one correspondence between the eigenvectors of @xmath183 and the eigenfunctions of the fp
operator @xmath159 .
the latter has a one dimensional eigenspace corresponding to the eigenvalue @xmath154 , since the kernel @xmath131 vanishes only for @xmath184 ( mod @xmath154 ) , which implies that the two - step probability @xmath185 for @xmath82 ( see @xcite , th .
thus the equation ( [ fp2 ] ) has the unique solution @xmath186 fulfilling @xmath187 or equivalently @xmath188 , where @xmath189 .
we find it diagonalizing numerically the matrix @xmath183 .
the function @xmath158 given by ( [ invmes ] ) is then the invariant density for the system with noise @xmath41 .
this technique was used to find the invariant measure for the logistic map given by @xmath190 , for @xmath11 in the presence of noise .
[ rolog ] presents the invariant density for the logistic map with noise parameter @xmath157 .
it can be compared to the histogram showing the density of the @xmath191-th iteration of one million uniformly distributed random initial points . in the deterministic limit
@xmath7 the size of the matrix @xmath192 grows to infinity .
we believe that our approach can be used to approximate the invariant measure of the deterministic system by decreasing the noise strength . fig .
[ roforlog ] presents a plot of invariant densities for the logistic map perturbed with the trigonometric noise for @xmath150 and @xmath151 , compared with the invariant measure for the deterministic case @xmath193 given by @xcite @xmath194 for @xmath165 .
the spectral decomposition of the frobenuis
perron operators corresponding to classical maps is a subject of an intensive current research @xcite .
the spectrum of a fp - operator is contained in the unit disk on the complex plane and depends on the choice of a function space , in which acts the fp - operator . if the dynamical system has an invariant density exists , the largest eigenvalue is equal to the unity
the radius of the second largest eigenvalue determines the rate of convergence to the invariant measure . to characterize the spectrum one defines _
essential spectral radius _ @xmath195 .
it is the smallest non - negative number , for which the elements of the spectrum outside the disk of radius @xmath195 , centered at the origin , are isolated eigenvalues of finite multiplicity .
it was shown @xcite that for one - dimensional piecewise @xmath196 expanding maps and the fp - operator defined on the space of functions of bounded variations , the spectral radius is related to the expanding constant .
we analyzed the spectral properties of the fp - operator of the perturbation of the logistic map , for which the lyapunov exponents equals to ln(2 ) and @xmath197 .
the interval @xmath198 $ ] is joined into a circle to keep the system conservative in the presence of noise .
fp - operator of the system subjected to the shift - invariant additive perturbation @xmath199 is represented by the matrix @xmath183 of the size @xmath192 .
we obtained its spectrum by the numerical diagonalization .
the largest eigenvalue @xmath200 of @xmath183 was equal to the unity up to a numerical error of order @xmath201 .
the second eigenvalue @xmath202 was found to approach the essential spectral radius @xmath195 in the deterministic limit @xmath193 .
since the matrix @xmath183 has real entries , its eigenvalues are real or appear in conjugate pairs .
[ spect ] presents the largest eigenvalues of this system for @xmath150 and @xmath151 .
all other eigenvalues are so small that they coincide with the origin in the picture .
observe , that eigenvalues do not tend to the values @xmath203 for @xmath204 found for the deterministic system in @xcite .
our results show that the structure of the spectrum of the fp - operator of a deterministic system depends on the character of the method used to approximate it . introducing a random noise
may be considered as a possible approximation , since it enables us to represent the fp - operator by a matrix of a finite dimension . in other words ,
the presence of the noise predetermines a certain space , in which the eigenstates live .
this numerical finding corresponds to the recent results of blank and keller @xcite , who showed the instability of the spectrum for some maps subjected to certain perturbations .
our aim is to compute probabilities entering the definition of the total dynamical entropy of a noisy system @xmath128 @xmath205 introducing @xmath206 times the expansion ( [ expand ] ) applied to the kernel @xmath207 and interchanging the order of summing and integration we arrive at @xmath208 q(p)\ , , \label{mat1}\ ] ] where @xmath146 , @xmath209 and @xmath186 are defined in sect .
iiib , @xmath210 , and matrices @xmath211 are given by the integral over the cell @xmath212 , i.e. @xmath213 for @xmath69 ; @xmath214 in the analogy to ( [ brm ] ) .
the above formula provides a significant simplification in the computation of entropy . instead of performing multidimensional integrals in ( [ probab ] )
, we start from computing the matrices @xmath215 for any cell @xmath69 , and receive the desired probabilities by matrix multiplications . by this method
the probabilities may be efficiently obtained even for larger numbers of the time steps @xmath16 .
the only problem consists in the number of terms in ( [ part ] ) , equal to @xmath27 , which for larger number of cells @xmath20 becomes prohibitively high . to overcome this difficulty we apply in this case the technique of iterated functions systems presented below . in this section
we present a method of computing the dynamical entropy ( [ hdyn ] ) , which is especially useful when the number of cells @xmath20 of the partition of the space @xmath1 is large .
we use the concept of iterated function systems ( ifss ) , discussed in details in the book of barnsley @xcite . consider the set of @xmath20 functions @xmath216 and maps @xmath217 defined as @xcite @xmath218 where the vector @xmath209 and the matrices @xmath215 are defined in sects .
iiib and iva , respectively , and @xmath219 is a convex closure of the set of all vectors of the form @xmath220 for @xmath221 $ ] .
let us stress that the spaces @xmath37 $ ] and the @xmath192 dimensional space @xmath222 are different .
the normalization of the kernel @xmath223 for @xmath224 , leads to the condition @xmath225 for any @xmath226 .
therefore the functions @xmath227 can be interpreted as place dependent probabilities and together with the functions @xmath228 form an ifs .
it is uniquely determined by the dynamical system @xmath8 with the noise given by the density @xmath229 and a specific @xmath20elements partition @xmath9 .
thus , the number of cells @xmath20 determines the size of ifs .
it can be shown @xcite that the entropy of the considered dynamical system with noise is equal to the entropy of the associated ifs .
the ifs generates a markov operator @xmath230 acting on the space of all probability measures on @xmath222 . for any measurable set
@xmath231 the following equality holds @xmath232 it describes the evolution of the measure @xmath233 transformed by @xmath230 .
if the functions @xmath228 fulfill the strong contraction conditions @xcite , there exists a unique attracting invariant measure @xmath234 for this ifs @xmath235 which , in general , displays multifractal properties @xcite .
the total entropy can be computed as the shannon entropy @xmath236 averaged over the invariant measure @xcite @xmath237 the calculation of such an integral from the definition corresponds to the matrix method presented previously . however , the existence of the attracting invariant measure @xmath234 and the kaijser
elton ergodic theorem @xcite assures that @xmath238 where @xmath239 is a generic random sequence produced by the ifs .
such a method of computing of an integral is often called random iterated algorithm @xcite .
we start computations from an arbitrary initial point @xmath240 , iterate it by the ifs , and compute the average ( [ keiser ] ) along a random trajectory . to avoid transient dependence on the initial point @xmath240
one should not take into account a certain number of initial iterations .
note that the computing time grows only linearly with the number of cells @xmath20 and one does not need to perform the burdensome time limit ( [ padep ] ) .
we used a similar method to compute the quantum coherent states entropy @xcite and the rnyi entropies for certain classical deterministic maps @xcite .
in this section we will study the entropy of the rnyi map and the logistic map perturbed by the trigonometric noise given by ( [ trignoise ] ) .
we will consider the partitions @xmath241 of the interval @xmath198 $ ] into @xmath20 equal subintervals .
we put @xmath242 , @xmath243 , @xmath244 , @xmath245 and @xmath246 .
a simple integration allows us to obtain the bg
entropy @xmath247 for this kind of the noise @xmath248 = { n\over 2 } \left[\psi(\textstyle\frac{n}{2}\displaystyle)- \psi(\textstyle\frac{n+1}{2}\displaystyle)\right]+1-\ln { \rm c_n } , \label{bgtrig}\ ] ] where @xmath249 denotes the digamma function @xcite and the normalization constant @xmath141 is given by ( [ cnorm ] ) .
it follows from ( [ bgtrig ] ) that in the deterministic limit ( @xmath250 ) the bg entropy diverges to minus infinity , namely @xmath251 this relation shows , how the maximal dynamical entropy @xmath252 ( see ( [ bgineq2 ] ) ) , admissible by a certain level of the noise , grows logarithmically in the deterministic limit .
we used the matrix method of computing probabilities , which lead to partial entropies @xmath26 and the relative entropies @xmath116 . fast ( presumably exponential ) convergence of the sequence @xmath116 allowed us to approximate the entropy by @xmath117 with accuracy of order @xmath253 .
[ hszifs ] presents the dependence of the entropy of the noise @xmath254 on the number of cells @xmath20 in the partition @xmath241 for two different amplitudes of noise ( @xmath255 and @xmath256 ) . the data for large number of cells @xmath257 are obtained by the technique of ifs .
the results are compared with the upper and lower bounds for the entropy of the noise which occurred in ( [ bgineqnoise ] ) .
it follows from ( [ bgineqnoise ] ) that the entropy diverges logarithmically with the number of cells @xmath20 in the partition . for a fixed partition it decreases to zero with decreasing strength of the stochastic perturbation ( increasing parameter @xmath13 ) .
the rnyi map @xmath258_{\rm mod 1}$ ] ( @xmath259 ) , with explicitly known metric entropy @xmath260 , is particularly suitable to test changes of the dynamical entropy with stochastic perturbation .
results obtained for the trigonometric noise ( [ trignoise ] ) are much more accurate than these obtained for rectangular noise and reviewed briefly in sec .
data presented below are received for the rnyi map with @xmath261 ( we put @xmath262 ) .
dependence of the total entropy @xmath263 on the number of cells @xmath20 is presented in fig .
[ 6xtot]a for four levels of noise ( @xmath150 and @xmath151 ) .
the solid line represents the entropy of the partition @xmath264 ( upper bound ) and the dashed line provides the @xmath13dependent lower bound given by @xmath265 ( for @xmath255 ) , while the stars denote the partition dependent entropy of the deterministic system given by @xmath8 .
it saturates at the generating partition @xmath266 and achieves the value @xmath267 .
it seems that this value gives an another lower bound for the total entropy @xmath55 .
the total entropy and the entropy of the noise diverge in the limit of fine partition @xmath62 ( @xmath268 ) , but their difference remains bounded .
[ 6xtot]b shows the difference @xmath269 necessary for computation the dynamical entropy ( [ hdynsup ] ) .
this quantity tends to zero for @xmath91 and @xmath270 ( [ limdyn ] ) and achieves its maximum - giving a lower bound for the dynamical entropy @xmath271 - close to the number of cells @xmath33 in the generating partition .
dynamical entropy is equal to zero for @xmath272 and increases with the decreasing noise strength . in the limit @xmath18
it is conjectured to tend to the ks - entropy of the deterministic system @xmath273 represented by a horizontal line .
a similar study was performed for the logistic map given by @xmath190 for @xmath274 $ ] perturbed by the trigonometric noise ( [ trignoise ] ) .
as before we treat the interval @xmath1 as a circle setting @xmath275 .
numerical data produce pictures analogous to those obtained for the rnyi map with @xmath261 . instead of presenting them here
, we supply a compilation of the results for both systems .
computing total entropy and entropy of the noise for several partitions we took the largest difference between them as an approximation of the dynamical entropy ( [ hdynsup ] ) .
[ hdynlim ] shows how the dynamical entropy changes with the noise parameter @xmath13 for both systems .
it is conjectured to tend to the corresponding values of the ks - entropy ( ln(2 ) for the logistic map and ln(6 ) for the rnyi map ) in the deterministic limit @xmath276 .
the standard definition of the kolmogorov
sinai entropy is not applicable for systems in the presence of a continuous random noise , since the partition dependent entropy diverges in the limit of a fine partition .
we generalize the notion of the ks - entropy for dynamical systems perturbed by an uncorrelated additive noise .
the total entropy of a random system is split into two parts : the _ dynamical entropy _ and the _ entropy of the noise_. in the deterministic limit ( the variance of the noise tends to zero ) the entropy of the noise vanishes , while the dynamical entropy of the stochastically perturbed system is conjectured to tend to the ks - entropy of the deterministic system .
the continuous boltzmann - gibbs entropy characterizes the density of the distribution of the noise .
it provides an upper bound for the maximal dynamical entropy observable under the presence of this noise .
if the bg entropy is equal to zero such a noise may be called disruptive , because one can not draw out any information concerning the underlying deterministic dynamics . investigating properties of the dynamical entropy
we find that the two limits : the diameter of the partition to zero and the noise strength to zero do not commute , and point out some consequences of this fact . computation of the dynamical entropy becomes easier , if one assumes that the density of the noise can be expanded in a finite basis consisting of continuous base functions . in this case
we find a simple way of computing the probabilities of trajectories passing through a given sequence of the cells in the partition .
the calculations are based on multiplication of matrices of size @xmath192 and the computing time grows linearly with the length of a trajectory @xmath16 .
on the other hand , diminishing the noise strength causes an increase of the matrix dimension .
for each dynamical system perturbed by this kind of noise and for a given @xmath20-element partition of the phase space we construct an associated iterated function system , which consists of @xmath20 functions with place dependent probabilities and acts in a certain @xmath192 dimensional auxiliary space .
entropy of the dynamical system with noise is shown to be equal to the entropy of ifs , which can be easily computed by the random iterated algorithm .
this method is particularly suitable for large number of cells @xmath20 , for which the number of possible trajectories grows in time as @xmath27 .
we study some one dimensional maps perturbed by trigonometric noise , for which the basis functions are given by trigonometric functions . in this case
we can represent the frobenius - perron operator for the noisy system by a matrix of size @xmath192 .
diagonalizing this matrix numerically we find the spectrum of this operator .
analysing the logistic map subjected by such a random perturbation we indicate that the invariant measure tends to the invariant measure of the deterministic system in the limit @xmath193 . on the other hand , the spectrum of the frobenius
perron operator describing the noisy system need not to tend to the corresponding characteristics of the deterministic system . the deterministic limit @xmath193 resembles in a sense the semiclassical limit of quantum mechanics @xmath111 .
for example , if one discuss the quantum analogues of classical maps on the sphere @xcite , the size of the hilbert space @xmath277 behaves as @xmath278 , where @xmath279 is the spin quantum number .
therefore , it would be interesting to analyze such two dimensional classical systems in the presence of noise ( in the case of two dimensional `` trigonometric '' noise the fp - operator can be represented by a matrix of the size @xmath280 ) and to compare , how the spectrum of a given classical map is approached in two complementary limits : the semiclassical limit @xmath281 of the corresponding quantum map and the deterministic limit @xmath282 of a classical noisy system .
some preliminary results on related issue of truncating the infinite matrix which represents the fp - operator of a deterministic system have been achieved very recently @xcite .
one of us ( k ) would like to thank ed ott for the hospitality at the university of maryland , where a part of this work has been done and gratefully acknowledges the fulbright fellowship .
this work was also supported by the polish kbn grant p03b 060 13 .
to illustrate the ifs method we discuss the computation of the entropy of the noise given by ( [ trignoise ] ) for @xmath283 and for the partition of the interval @xmath198 $ ] into @xmath284 equal cells . in this case
( [ ifs1 ] ) gives the ifs consisting of @xmath284 functions acting in a @xmath285-dimensional space @xmath286 ^ 3 $ ] .
the probabilities @xmath227 are place dependent @xmath287 while the functions read @xmath288\\ f_2(x , y , z)=\frac{1}{p_2(x , y , z ) } [ \frac{x(3\pi-8)}{32\pi}+\frac{y}{16\pi}+\frac{z}{32}\ , ; \frac{x}{8\pi}+\frac{y}{16}+\frac{3z}{8\pi}\ , ; \frac{x}{32}+\frac{3y}{16\pi}+\frac{z(3\pi+8)}{32\pi}]\\ f_3(x , y , z)=\frac{1}{p_3(x , y , z ) } [ \frac{x(3\pi-8)}{32\pi}-\frac{y}{16\pi}+\frac{z}{32}\ , ; -\frac{x}{8\pi}+\frac{y}{16}-\frac{3z}{8\pi}\ , ; \frac{x}{32}-\frac{3y}{16\pi}+\frac{z(3\pi+8)}{32\pi}]\\ f_4(x , y , z)=\frac{1}{p_4(x , y , z ) } [ \frac{x(3\pi+8)}{32\pi}-\frac{3y}{16\pi}+\frac{z}{32}\ , ; -\frac{3x}{8\pi}+\frac{y}{16}-\frac{z}{8\pi}\ , ; \frac{x}{32}-\frac{y}{16\pi}+\frac{z(3\pi-8)}{32\pi } ] \\ \end{array}\ ] ] for @xmath289 .
[ ifsdense ] presents the support of the invariant measure @xmath234 for this ifs . applying the random iteration algorithm we obtain in this case the entropy of the noise @xmath290 .
ironically , less interesting ( more contracting ) fractal picture leads to a faster convergence of the sum ( [ keiser ] ) .
|
dynamics of deterministic systems perturbed by random additive noise is characterized quantitatively . since for such systems the kolmogorov - sinai ( ks )
entropy diverges if the diameter of the partition tends to zero , we analyze the difference between the total entropy of a noisy system and the entropy of the noise itself . we show that this quantity is finite and non negative and call it the dynamical entropy of the noisy system . in the weak noise limit
this quantity is conjectured to tend to the ks - entropy of the deterministic system . in particular , we consider one - dimensional systems with noise described by a finite - dimensional kernel for which the frobenius - perron operator can be represented by a finite matrix .
|
observational evidence has emerged over the past few decades indicating that carbon is ubiquitous in the early universe . the class of carbon - enhanced metal - poor ( cemp ; @xmath3$]@xmath11 , e.g. , * ? ? ?
* ; * ? ? ?
* ) stars are found with increasing fractions at lower metallicities , and account for at least 80% of all ultra metal - poor ( ump ; [ fe / h]$ ] = @xmath12 , where @xmath13 is the number density of atoms of elements @xmath14 and @xmath15 in the star ( @xmath16 ) and the sun ( @xmath17 ) , respectively . ]
@xmath18 ) stars observed to date @xcite .
in particular , the so - called cemp - no stars ( which exhibit sub - solar abundances of neutron - capture elements ; e.g. , @xmath5 $ ] @xmath19 0.0 ) are believed to be direct descendants from the very first stellar generations formed after the big bang @xcite .
in addition , the discovery of high - redshift carbon - enhanced damped ly@xmath20
systems @xcite , which present qualitatively similar light - element ( from c to si ) abundance patterns as the cemp - no stars , provides additional evidence that carbon is an important contributor to the earliest chemical evolution .
one of the main open questions is whether the presence of carbon is required for the formation of low - mass second - generation stars @xcite .
this idea can be tested with low - mass long - lived ump stars that are thought to have formed from an ism polluted by the nucleosynthesis products of massive metal - free population iii ( pop iii ) stars .
these massive stars could have formed as early as several hundred million years after the big bang , at redshift @xmath21 @xcite .
the recently discovered quasar ulas j1120@xmath40641 at redshift @xmath22 @xcite , with an overall metal abundance ( defined as elements heavier then helium ) of @xmath23$]@xmath24 , could be an example of a viable site for the formation of the first stars .
the lack of heavy metals may prevent the formation of low - mass stars ( due to inefficient cooling ; * ? ? ?
* ) , supporting the suggestion that the early - universe initial mass function was strongly biased toward high - mass stars . in the picture that has developed , it is these high - mass stars that would quickly evolve and enrich the primordial ism with elements heavier than helium , including carbon @xcite . even though the relevance of cemp - no stars as probes of the first stellar generations in the universe is well - established , the exact conditions that led to their formation remain an active area of inquiry . there have been a number of advances in the theoretical description of the likely stellar progenitors of cemp - no stars over the last decade ( see * ? ?
* for a recent review on the subject ) .
the suggested scenarios include the `` spinstar '' model ( rapidly - rotating , near zero - metallicity , massive stars ; * ? ? ?
* ; * ? ? ?
* ) , the `` faint sne '' that undergo mixing and fallback @xcite , and the metal - free massive stars from @xcite . in the spinstar model , it is assumed that the chemical composition of the observed ump stars is a combination of the evolution of the massive star itself mixed with some amount of interstellar material @xcite .
it follows that the source of heavy metals in the ump stars could arise from a different set of progenitors . for the faint sne and metal - free massive stars , the initial chemical abundances of the progenitor mimic the primordial big bang nucleosynthesis composition : 76% hydrogen , about 24% helium , and a trace amount of lithium . in both cases ,
the observed chemical elements in ump stars were formed during the progenitor stellar evolution , either by internal burning and/or explosive events , and their abundance is the result of mixing between the sne ejecta with surrounding primordial gas .
these two models differ in terms of the treatment of the mixing and fallback of processed materials in the progenitor , which varies with mass and explosion energy ( see * ? ? ?
* for details ) .
the @xcite models compute explosion energy and fallback self - consistently based on a hydrodynamic model , considering that the sn explosion is spherical .
all of the aforementioned models are able to reproduce a subset of ( but not all ) of the observed elemental - abundance patterns of ump stars reasonably well .
nevertheless , the question of whether one or more classes of progenitors were present ( and their relative frequencies ) in the primordial universe is still under discussion . in paper
i of this series , @xcite present evidence based on the morphology of the relationship between the absolute abundance of carbon , @xmath7(c ) @xmath25(c ) , and [ fe / h ] , coupled with clear differences in the absolute abundances of the light elements na and mg among cemp - no stars , that at least two classes of progenitors are likely to be required .
it appears that one class ( spinstars ) dominates at the very lowest metallicities , [ fe / h ] @xmath26 , whereas the other ( faint sne ) dominates over the range @xmath27 [ fe / h ] @xmath28 . in the metallicity range @xmath29
[ fe / h ] @xmath30 , there are examples of stars that are associated with either . for the above reasons
, we suggest that the very best stars to place constraints on the nature of the cemp - no progenitors are the ump stars with metallicities between @xmath31 and @xmath32 .
unfortunately , such stars are still exceedingly rare @xcite . even though their numbers have increased considerably over the last decade ( 21 stars according to * ? ? ?
* ) , many more are needed , in order to fully understand the nature of their stellar progenitors and the associated nucleosynthesis processes . in this paper
, we report on a high - resolution spectroscopic abundance analysis , a relatively bright ( @xmath33 ) cemp - no ( @xmath1 $ ] @xmath34 , @xmath3 $ ] = @xmath41.74 , and @xmath5 $ ] = @xmath35 ) star , which was first identified as the metal - poor candidate cs 30324 - 0063 in the hk survey @xcite , and later re - identified in the hamburg / eso survey ( he s ; * ? ? ?
* ) and the radial velocity experiment ( rave ; * ? ? ?
was also studied by @xcite , focusing on long - term radial - velocity monitoring of cemp - no stars .
we use the newly derived abundance pattern of , along with literature data for 11 other ump and hyper metal - poor ( hmp ; [ fe / h ] @xmath36 ) stars , to assess evidence in support of the conclusion in paper i that cemp - no stars require more than one class of stellar progenitors .
this paper is outlined as follows : section [ secobs ] describes the medium - resolution spectroscopic target selection and high - resolution follow - up observations , followed by the determinations of the stellar parameters and chemical abundances in section [ secatm ] .
section [ disc ] details our analysis of ump and hmp stars from the literature with a grid of supernova yields , and evaluates the impact of these data on current hypotheses for chemical evolution in the early universe .
our conclusions are provided in section [ final ] .
the star was selected as a metal - poor candidate by @xcite , based on its weak caii k feature ( 3933 , used as the primary metallicity indicator ) in the he s objective - prism spectrum .
this star was also selected by @xcite , based on its strong ch @xmath37-band ( 4300 , the primary carbon - abundance indicator ) .
figure [ highres ] is a comparison between the low - resolution he s objective - prism spectrum ( @xmath38 , upper panels ) , the medium - resolution ( @xmath39 , middle panels ) spectrum , and the high - resolution ( @xmath40 , lower panels ) spectrum of .
the left panels show a zoom - in of the region surrounding the caii k line , and the right panels show a zoom - in of the region near the ch @xmath37-band .
note the lack of measurable metallic features ( other than ca ) in the he s spectra , which is an indication of the low metallicity of the target .
the balmer lines of hydrogen are also quite weak , suggesting that the target has a cool effective temperature .
however , as can be seen from inspection of the medium- and high - resolution spectra , the caii lines , as well as hydrogen lines from the balmer series , are clearly identifiable , as labeled in figure [ highres ] . in addition , the lower panels show a number of ch features in both regions , which are used below to determine the carbon abundance from the high - resolution spectrum .
lr[!ht ] & + hk survey & + hamburg / eso survey & + 2mass & + rave & + & + + & + @xmath20 ( j2000 ) & 00:22:44.86 + @xmath41 ( j2000 ) & @xmath217:24:29.07 + @xmath42 ( mag ) & 12.89 + @xmath43 & 0.94 + & + + & + @xmath44 & @xmath458,000 + v@xmath46(km / s ) & 90.49 + & + + & + date & 2011 10 16 + ut & 03:40:27 + exptime ( s ) & 120 + @xmath44 & @xmath452,000 + & + + & + date & 2015 06 17 + ut & 08:26:20 + exptime ( s ) & 1800 + @xmath44 & @xmath4535,000 + v@xmath46(km / s ) & 94.91 the medium - resolution spectrum of was obtained as part of an effort to follow - up cemp candidates from the he s , as described in @xcite .
the observations were carried out in semester 2011b using the efosc-2 spectrograph @xcite mounted on the 3.5 m eso new technology telescope . the setup made use of grism7 ( 600 gr mm@xmath47 ) , and a 1@xmath480 slit with @xmath49 on - chip binning , resulting in a wavelength coverage of 3550 - 5500 , resolving power of @xmath39 , dispersion of 0.96 / pixel ( with @xmath50 pixels per resolution element ) , and signal - to - noise ratios s / n@xmath51 per pixel at 4300 .
the calibration frames included fear exposures ( taken following the science observation ) , quartz - lamp flatfields , and bias frames .
all reduction tasks were performed using standard iraf packages .
table [ candlist ] lists basic information on and details of the spectroscopic observations at medium and high resolution .
a high - resolution spectrum was obtained during the 2015a semester using the magellan inamori kyocera echelle ( mike ; * ? ? ?
* ) spectrograph mounted on the 6.5 m magellan - clay telescope at las campanas observatory .
the observing setup included a @xmath52 slit with @xmath53 on - chip binning , yielding a resolving power of @xmath54 ( blue spectral range ) and @xmath55 ( red spectral range ) , with @xmath56 pixels per resolution element .
the s / n at @xmath57 is @xmath58 per pixel .
mike spectra have nearly full optical wavelength coverage ( @xmath59 ) .
the data were reduced using the data reduction pipeline developed for mike spectra , first described by @xcite .
the stellar atmospheric parameters were first obtained from the medium - resolution eso / ntt spectrum using the n - sspp @xcite , a modified version of the segue stellar parameter pipeline ( sspp ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the values for @xmath60 , @xmath61 , and @xmath1 $ ] determined from this analysis were used as first estimates for the high - resolution analysis . using the high - resolution mike spectrum , we determined the stellar parameters spectroscopically , using software developed by @xcite .
equivalent - width measurements were obtained by fitting gaussian profiles to the observed absorption lines .
table [ eqw ] lists the lines used in this work , their measured equivalent widths , and the derived abundance from each line .
we employed one - dimensional plane - parallel model atmospheres with no overshooting @xcite , computed under the assumption of local thermodynamic equilibrium ( lte ) .
lrrrrr[!ht ] + c ch & 4246.000 & & & syn & 5.83 + c ch & 4313.000 & & & syn & 5.73 + n cn & 3883.000 & & & syn & 6.13 + na i & 5889.950 & 0.00 & 0.108 & 97.06 & 2.77 + na i & 5895.924 & 0.00 & @xmath20.194 & 83.35 & 2.79
+ mg i & 3829.355 & 2.71 & @xmath20.208 & syn & 4.60
+ mg i & 3832.304 & 2.71 & 0.270 & syn & 4.60
+ mg i & 3838.292 & 2.72 & 0.490 & syn & 4.60
+ mg i & 4571.096 & 0.00 & @xmath25.688 & syn & 4.90 + mg i & 4702.990 & 4.33 & @xmath20.380 & syn & 4.90 + the effective temperature of was determined by minimizing trends between the abundances of 77 lines and their excitation potentials , and applying the temperature corrections suggested by @xcite .
the microturbulent velocity was determined by minimizing the trend between the abundances of lines and their reduced equivalent widths .
the surface gravity was determined from the balance of the two ionization stages of iron , and .
also had its stellar atmospheric parameters determined from the moderate - resolution ( @xmath62 ) rave spectrum by @xcite .
these values , together with our determinations from the medium- and high - resolution spectra , are listed in table [ obstable ] .
there is very good agreement between the temperatures derived from the medium- and high - reslution spectra used in this work ; the rave value is about @xmath63 k warmer .
the surface gravities are all within 2@xmath64 , and the eso / ntt and rave values agree with each other exactly .
this is expected , since both of these estimates come from isochrone matching , while the high - resolution @xmath61 was determined from the balance of and lines . for @xmath1 $ ] ,
the rave value is 0.8 dex higher than the other two results . according to the rave data release 4 @xcite , the spectrum for has a s / n=40 , which is below the s / n=50 limit set by @xcite for reliable parameter estimates for halo giants , and their pipeline analysis converged without warnings for this star .
since the difference in temperature can not alone account for such a large discrepancy in @xmath1 $ ] , this most likely arises from a combination of the lower s / n and the methods used to calibrate the rave metallicity scale for giants , which have a 0.40 dex dispersion in their residuals .
lcccc[!ht ] + rave & 4974 ( 101 ) & 0.95 ( 0.35 ) & @xmath23.20 ( 0.10 ) & + eso / ntt & 4792 ( 150 ) & 0.98 ( 0.35 ) & @xmath24.06 ( 0.20 ) & + magellan & 4765 ( 100 ) & 1.55 ( 0.20 ) & @xmath24.05 ( 0.05 ) & 1.50 ( 0.20 ) + lrrrrcr[!ht ] + c ( ch ) & @xmath48.43 & @xmath45.78 & @xmath22.65 & @xmath41.40 & 0.15 & 2 + c ( ch ) & @xmath48.43 & @xmath46.12 & @xmath22.31 & @xmath41.74 & 0.15 & 2 + n ( cn ) & @xmath47.83 & @xmath46.13 & @xmath21.70 & @xmath42.35 & 0.20 & 1 + na i & @xmath46.24 & @xmath42.78 & @xmath23.46 & @xmath40.59 & 0.05 & 2 + mg i & @xmath47.60 & @xmath44.58 & @xmath23.02 & @xmath41.03 & 0.10 & 8 + al i & @xmath46.45 & @xmath42.05 & @xmath24.40 & @xmath20.26 & 0.10 & 1 + si i & @xmath47.51 & @xmath44.36 & @xmath23.19 & @xmath40.90 & 0.10 & 1 + ca i & @xmath46.34 & @xmath42.70 & @xmath23.64 & @xmath40.40 & 0.05 & 4 + sc ii & @xmath43.15 & @xmath20.72 & @xmath23.87 & @xmath40.18 & 0.05 & 3 + ti i & @xmath44.95 & @xmath41.17 & @xmath23.78 & @xmath40.27 & 0.05 & 4 + ti ii & @xmath44.95 & @xmath41.19 & @xmath23.76 & @xmath40.28 & 0.10 & 11 + cr i & @xmath45.64 & @xmath41.51 & @xmath24.13 & @xmath20.09 & 0.05 & 5 + mn i & @xmath45.43 & @xmath41.43 & @xmath24.00 & @xmath40.05 & 0.05 & 4 + fe i & @xmath47.50 & @xmath43.45 & @xmath24.05 & 0.00 & 0.05 & 77 + fe ii & @xmath47.50 & @xmath43.45 & @xmath24.05 & 0.00 & 0.08 & 3 + co i & @xmath44.99 & @xmath41.29 & @xmath23.70 & @xmath40.35 & 0.05 & 4 + ni i & @xmath46.22 & @xmath41.94 & @xmath24.28 & @xmath20.23 & 0.05 & 7 + sr ii & @xmath42.87 & @xmath21.91 & @xmath24.78 & @xmath20.73 & 0.07 & 2 + ba ii & @xmath42.18 & @xmath22.97 & @xmath25.15 & @xmath21.11 & 0.05 & 2 + eu ii & @xmath40.52 & @xmath653.28 & @xmath653.80 & @xmath660.25 & & 1 + pb i & @xmath41.75 & @xmath650.25 & @xmath652.00 & @xmath662.05 & & 1 + elemental - abundance ratios , @xmath67 $ ] , are calculated adopting solar photospheric abundances from @xcite . the average measurements ( or upper limits ) for @xmath68 elements , derived from the magellan / mike spectrum , are listed in table [ abund ] .
the @xmath64 values are the standard error of the mean .
abundances were calculated by both equivalent - width analysis and spectral synthesis .
the 2014 version of the moog code @xcite , which includes a more realistic treatment of scattering ( see * ? ? ?
* for further details ) , is used for the spectral synthesis .
carbon abundance was derived from ch molecular features at @xmath694246 ( @xmath70=5.83 ) and @xmath694313 ( @xmath70=5.73 ) , with an average value of @xmath70=5.78 ( @xmath3$]@xmath71 ) .
the left panels of figure [ csyn ] show the spectral synthesis of the ch @xmath37-band for .
the dots represent the observed spectra , the solid line is the best abundance fit , and the dotted and dashed lines are the lower and upper abundances , used to estimate the uncertainty .
the gray line shows the synthesized spectrum in the absence of carbon .
the lower panel shows the residuals ( in % ) between the observed data and the best abundance fit , which are all below 3% for the synthesized region . since is on the upper red - giant branch
, the observed carbon abundance does not reflect the chemical composition of its natal gas cloud . by using the procedure described in @xcite , which interpolates the observed @xmath61 , @xmath1 $ ] , and @xmath3 $ ] of 01 with a grid of theoretical stellar evolution models for low - mass stars , we determine the carbon depletion due to cn processing for to be 0.34 dex .
the @xmath72c/@xmath73c isotopic ratio is a sensitive indicator of the extent of mixing processes in stars on the red - giant branch .
using a fixed elemental carbon abundance ( @xmath70=5.78 ) for the ch features around @xmath694217 , we derived @xmath72c/@xmath73c = 4 , which suggests that substantial processing of @xmath72c into @xmath73c has taken place in the star .
we note that this ratio also indicates that considerable processing of carbon into nitrogen has occured in 01 .
the right panels of figure [ csyn ] show the determination of the @xmath72c/@xmath73c isotopic ratio .
the dots represent the observed spectrum , the solid gray line is the best fit , with the two other values taken to be lower and upper limits .
the lower panel shows that the residuals between the observed data and @xmath72c/@xmath73c = 4 are all below 2% .
the nitrogen abundance was determined from spectral synthesis of the cn band at @xmath693883 .
the nh band at @xmath693360 did not have sufficiently high s / n to allow for a proper spectral synthesis .
for the cn band , we used a fixed carbon abundance of @xmath70=5.78 ( average of carbon abundances determined from the ch band ) , and derived a value of @xmath74=6.13 , with an uncertainty of @xmath75 dex .
abundances of na , sc , ti , cr , co , and ni were determined by equivalent - width analysis only .
for ti , where transitions from two different ionization stages were measured , the abundances agree within 0.02 dex .
spectral synthesis was used to determine abundances for mg , al , si , ca , and mn ( accounting for hyperfine splitting from the linelists from * ? ? ?
the chemical abundances for sr and ba , as well as upper limits for eu and pb , were determined via spectral synthesis .
we used the compilation of linelists by @xcite , based on lines from @xcite , @xcite , @xcite , and the vald database @xcite .
the neutron - capture absorption lines in the blue spectral region , particularly close to strong ch or cn features , need to be carefully synthesized , since these are intrinsically weak in cemp - no stars .
because of that , we included the observed carbon and nitrogen abundances for all the syntheses , as well as the @xmath72c/@xmath73c isotope ratio .
the sr abundance was determined from the @xmath694077 ( @xmath76 = @xmath21.93 ) and @xmath694215 ( @xmath76 = @xmath21.88 ) lines , with an average value of @xmath77 $ ] = @xmath20.73 . for ba ,
both @xmath694554 and @xmath694934 features were successfully synthesized with @xmath78 = @xmath22.97 ( @xmath5 $ ] = @xmath21.11 ) . both sr and
ba abundances are within typical ranges for cemp - no stars @xcite .
upper limits were determined for eu ( @xmath694129 ) and pb ( @xmath694057 ) .
the pb upper limit ( @xmath79$]@xmath802.05 ) is similar to the one for the cemp - no @xcite , and it adds further evidence that the origin of the neutron - capture abundances in is unlikely to be from an unseen evolved companion .
the @xmath79 $ ] should be higher by at least a factor of ten to agree with theoretical predictions for the @xmath81-process @xcite .
furthermore , the radial - velocity monitoring of reported by @xcite revealed no significant variation ( @xmath82 kms@xmath83 ) over a temporal window of 1066 days .
lrrrrr + na i & 0.14 & @xmath20.04 & @xmath20.11 & 0.07 & 0.20 + mg i & 0.13 & @xmath20.12 & @xmath20.07 & 0.04 & 0.19 + al i & 0.11 & @xmath20.07 & @xmath20.11 & 0.10 & 0.20 + si i & 0.14 & @xmath20.02 & @xmath20.01 & 0.10 & 0.17 + ca i & 0.10 & @xmath20.02 & @xmath20.01 & 0.05 & 0.11 + sc ii & 0.09 & 0.07 & @xmath20.04 & 0.06 & 0.13 + ti i & 0.17 & @xmath20.03 & @xmath20.01 & 0.05 & 0.18 + ti ii & 0.06 & 0.06 & @xmath20.08 & 0.03 & 0.12 + cr i & 0.17 & @xmath20.04 & @xmath20.06 & 0.04 & 0.19 + mn i & 0.19 & @xmath20.05 & @xmath20.07 & 0.07 & 0.22 + fe i & 0.16 & @xmath20.05 & @xmath20.09 & 0.01 & 0.19 + fe ii & 0.01 & 0.09 & @xmath20.02 & 0.08 & 0.12 + co i & 0.18 & @xmath20.03 & @xmath20.02 & 0.05 & 0.19 + ni i & 0.17 & @xmath20.07 & @xmath20.09 & 0.04 & 0.21 + sr ii & 0.09 & 0.06 & @xmath20.10 & 0.07 & 0.16 + ba ii & 0.12 & 0.08 & @xmath20.01 & 0.07 & 0.16 + uncertainties in the elemental - abundance determinations , as well as the systematic uncertainties due to changes in the atmospheric parameters , were treated in the same way as described in @xcite .
table [ sys ] shows how variations within the quoted uncertainties in each atmospheric parameter affect the derived chemical abundances .
also listed is the total uncertainty for each element , which is calculated from the quadratic sum of the individual error estimates . even though @xmath60 , @xmath61 , and @xmath84are correlated , we assume complete knowledge of two of them to assess how changes in the third parameter would affect the abundance calculation .
for example , a change in @xmath4150 k in @xmath60 for 01 requires a change of about 0.1 dex in @xmath61 to maintain the balance between and .
however , this change in @xmath61 translates to a 0.01 dex change in abundance , which is 16 times smaller than the change due to @xmath60 .
then , for simplicity , we assume that the variables are independent , and use the quadratic sum . for this calculation , we used spectral features with abundances determined by equivalent - width analysis only .
the variations for the parameters are @xmath4150 k for @xmath60 , @xmath40.3 dex for @xmath61 , and @xmath40.3 kms@xmath83 for @xmath84 .
in this section , we assess the main properties of the possible stellar progenitors of selected ump stars from the literature , by comparing their abundance patterns with theoretical model predictions for pop iii stars .
we employ the non - rotating massive - star models from @xcite , for which the free parameters are the mass ( @xmath85 ) , explosion energies ( @xmath86 ) , and the amount of mixing in the sne ejecta .
the grid used in the work has 16,800 models , and the @xmath87 matching algorithm is described in @xcite .
an online tool with the model database can be accessed at ` starfit ` . for the present application ,
we adopt a similar procedure as the one described in @xcite and @xcite , as described below .
we collected ump stars from the literature for which at least carbon , nitrogen , sodium , magnesium , calcium , and iron were measured . with @xmath1$]@xmath88 ,
only 11 stars have abundances for these elements determined from high - resolution spectroscopy ( including ) .
we also added to the sample , since its metallicity is @xmath1 $ ] = @xmath23.8 , and it has a number of abundances measured with small uncertainties .
table [ litdata ] lists the abundances and references for the literature sample .
the stars are divided according to the cemp - no groups ii and iii from @xcite ; two of our sample stars are classified as carbon - normal stars .
the carbon abundances were corrected for the c depleted by cn processing , following the procedures described in @xcite , and briefly summarized in section [ cnsec ] .
in addition , we calculated the inverse corrections for the nitrogen abundances , by imposing that the [ ( c@xmath4n)/fe ] ratio remains constant throughout the evolution of the star on the giant branch .
the model grid we employed is the same as that in @xcite .
even though there may also be on processing that occurs in the star , the effect is expected to be negligible compared to the cn processing ( see bottom panel of figure 2 in * ? ? ?
* ) . for the model matching
, we have used the following abundances ( where available ) : c , n , na , mg , al , si , ca , sc , ti , v , cr , mn , fe , co , and ni . for consistency , no upper limits were used . for each star
, we assembled @xmath89 unique abundance patterns , based on the observed values from the literature .
for each element , we generated @xmath89 random numbers from a normal distribution , with the measured abundances as the central value , and the uncertainties as the dispersion .
then , we randomly selected individual abundances from each distribution , and generated new abundance patterns . for each of these
, we found the progenitor mass , explosion energy , and the mean squared residuals from the ` starfit ` code .
figure [ group1 ] shows the results of this exercise for the 12 stars in table [ litdata ] . for each star
, the left panel shows the best model fits for the @xmath89 abundance patterns .
the masses and explosion energies of the models are given in the legend at the upper right in each panel , color - coded by their fractional occurence .
the right - hand panel for each star shows the distribution of the mean squared residuals for the simulations .
the colored bars overlaying the upper density distributions on the right panels mark the median residual value , and its value is shown in the legend of these panel at the top right , along with the @xmath90 ( median absolute deviation ) , a robust estimator of the dispersion .
the stars are ordered by increasing median values . as an example , consider the top left panel of figure [ group1 ] , for the star 01 .
we first ran the ` starfit ` code to find the best model fit for each of the @xmath89 abundance patterns generated from the observed abundances ( blue filled circles ) . in about 70% ( 7,041 ) of the resampled abundance patterns , the best fit was the model with @xmath91 and @xmath92 ( black solid line ) . in 2,849 ( 28% ) of the patterns , the model with @xmath93 and
@xmath92 gave the best fit ( dark gray solid line ) , and for the remaining 110 abundance patterns ( light gray solid lines ) , the best - fit models had masses between @xmath94 and explosion energies between @xmath95 .
in total , 6 unique models ( out of the 16,800 models on the grid ) were able to account for the best fits for the @xmath89 resampled abundance patterns .
for each best fit , the `
starfit ` code calculates the mean squared residual , and the distribution of these values is given on the panel to the right side of the abundance patterns . above the distribution
is the density plot ( black stripes ) , with the median value ( blue stripe ) shown on the upper right part of the panel , together with the @xmath90 .
for the first five stars in the left - hand column of panels shown in figure [ group1 ] , the mass and explosion energy of the progenitor are within @xmath94 and @xmath95
. these values are consistent with predictions from @xcite for the faint supernova progenitor scenario .
considering the two most - frequent models for each star , the c - n abundances are well - reproduced in most cases , within @xmath96 of the theoretical values .
the na - si abundances agree within @xmath96 with the best model fits ( except for al in 01 and mg in 02 ) .
the ca and ti abundances also agree within @xmath96 , except for ti in 04 .
the large over - abundances of sc for 01 and 03 are expected , since sc may have contributions from other nucleosynthesis processes ( see * ? ? ? * for further details ) . among the cr - fe abundances , there is also overall good agreement , with fe values agreeing within @xmath97 with the best model fits .
the enhanced cobalt abundances for 01 , 03 , and 05 can not be accounted for by the models , but we note that @xmath98 is often under - predicted by theoretical models ( e.g. , * ? ?
* ; * ? ? ?
* ) . for stars in the right - hand panels of figure [ group1 ] ( red filled squares ) ,
the progenitor masses range from @xmath99 for 09 to @xmath100 for 12 , all with explosions energies of @xmath92 . for these stars ,
as reflected by their residual distributions , there is a clear mismatch between the model predictions and the observed abundances .
the stars 06 , 07 , and 08 all have best fits with a @xmath91 and @xmath92 model .
this is likely a numerical artifact , since the c and n abundances for these three stars are in complete disagreement with the models . for 10 ,
only the ca abundance matches the best model yields , and most values are at least 0.5 dex away from the theoretical values . for 11 ,
the only observed abundance within @xmath97 of the model is fe . in this case , it is clear that no models can reproduce the large difference between the c - n and na - al abundances . for 09 and 12 ,
the models are not able to simultaneously predict the c - n abundances and the na - mg or ca - ti abundances . based on these results
, we infer that the first five stars shown in the left - hand column of panels of figure [ group1 ] have abundance patterns that are consistent with the progenitor stellar population described by the models of @xcite . for the remaining seven stars ,
the poor agreement between observations and models suggests the presence of at least one additional class of stellar progenitors for ump stars .
figure [ residuals ] shows the behavior of the median residual values for and the data for other ump stars from the literature , where the stars are ordered by increasing median values ( panel @xmath101 ) , and as a function of @xmath1$](panel @xmath102 ) , @xmath103 ( panel @xmath104 ) , and @xmath105 ( panel @xmath106 ) . in panel
@xmath101 , the gray shaded areas encompass @xmath107 and @xmath108 for the five stars ( blue filled circles ) in the left - hand column of figure [ group1 ] .
the numbers shown in the legend are the progenitor masses ( in m@xmath109 ) and the explosion energy ( @xmath110 ) from the best model fit for each star .
the gray regions highlighted on panels @xmath104 and @xmath106 are explained below .
distinctions between the stars with low ( blue symbols ) and high ( red symbols ) residuals become more evident when inspecting the median residuals in figure [ residuals ] .
even though this separation does not appear to be correlated with @xmath1 $ ] ( panel @xmath102 ) , observations of additional stars are needed to cover the gap between @xmath111 @xmath1 $ ] @xmath112 , in order to uncover any possible trend . nevertheless , 04 ( @xmath1
$ ] = @xmath113 ) has a median residual value consistent with that of other stars at @xmath1 $ ] @xmath114 , and it does not agree with the values for 09 ( @xmath1 $ ] = @xmath115 ) and 12(@xmath1
$ ] = @xmath116 ) . from inspection of panel @xmath104
, it can be seen that the low - residual stars exhibit @xmath103 @xmath117 , and are mostly concentrated at @xmath103 @xmath118 .
the only exception is 05 , with @xmath103 @xmath119 .
the trend for na can be an indication that another stellar progenitor , producing ( but not limited to ) @xmath103 @xmath120 ( gray shaded area on panel @xmath104 ) , is needed to account for the diferences in the observed abundances . in panel
@xmath106 , even though the low - residual stars exhibit @xmath105 @xmath121 , there is no obvious separation between these and the high - residual stars .
the two high - residual stars with low @xmath103 in panel @xmath104 also have low @xmath105 .
however , this correlation does not hold for the other stars in the same group .
lcccccccccccccccc + bd@xmath444@xmath122493 & 5.83 & 4.40 & 2.71 & 4.23 & 2.06 & 3.78 & 2.82 & @xmath20.39 & 1.37 & 1.99 & 0.97 & 3.62 & 1.76 & 2.21 & 1 & + cs 22949@xmath2037 & 5.66 & 6.01 & 4.07 & 5.06 & 2.93 & 4.13 & 2.70 & @xmath20.62 & 1.33 & 1.18 & 0.78 & 3.57 & 1.50 & 2.00 & 2 & a + he 0020@xmath21741 & 6.12 & 5.79 & 2.78 & 4.58 & 2.05 & 4.36 & 2.70 & @xmath20.72 & 1.18 & 1.51 & 1.43 & 3.45 & 1.29 & 1.94 & 3 & b + he 0057@xmath25959 & 5.21 & 5.90 & 4.14 & 4.03 & & & 2.91 & @xmath20.76 & 1.27 & 1.06 & & 3.42 & & 2.30 & 4 & + + g77@xmath261 & 7.00 & 6.40 & 2.90 & 4.00 & & & 2.90 & & & 2.00 & 1.30 & 3.42 & & & 5 & + he 0107@xmath25240 & 6.96 & 5.57 & 1.86 & 2.41 & & & 0.99 & & @xmath20.62 & & & 2.06 & & 0.60 & 6 & c + he 1310@xmath20536 & 6.64 & 6.88 & 2.28 & 3.87 & 1.91 & & 2.19 & @xmath21.07 & 1.15 & 1.00 & 0.49 & 3.35 & 1.12 & 1.95 & 7 & d + he 1327@xmath22326 & 6.90 & 6.79 & 2.99 & 3.54 & 1.90 & & 1.34 & & @xmath20.09 & & & 1.79 & & 0.73 & 8 & + he 2139@xmath25432 & 7.00 & 5.89 & 4.37 & 5.19 & 2.79 & 4.49 & & & 1.24 & 1.96 & & 3.48 & 1.59 & 2.37 & 9 & + sdss j1313@xmath20019 & 6.39 & 6.34 & 1.61 & 3.04 & 1.34 & & 1.59 & @xmath21.54 & 0.33 & & & 2.50 & & 1.65 & 10 & + + cd@xmath238@xmath122245 & 4.48 & 4.76 & 2.19 & 3.85 & 1.63 & 3.65 & 2.56 & @xmath20.82 & 1.19 & 1.25 & 0.31 & 3.43 & 1.25 & 2.05 & 11 & e + cs 30336@xmath2049 & 4.40 & 4.56 & 2.13 & 3.64 & 1.58 & 3.66 & 2.39 & @xmath20.83 & 1.06 & 0.86 & 0.57 & 3.46 & 1.42 & 2.13 & 12 & f to further investigate the differences among the twelve ump stars , we compared the individual abundances as a function of @xmath1 $ ] , @xmath103 , and @xmath105 with the yields from the 16,800 models used for the matching procedure .
figure [ abunda ] shows the result of this exercise .
the colored symbols are the observed abundances from the literature sample ( including ) , and the gray symbols are yields from the supernova models .
the open symbols show the cemp - no groups proposed in paper i. for the theoretical values , the progenitor mass is proportional to the size of the symbol .
models with abundance values outside the ranges shown in figure [ abunda ] were suppressed for simplicity .
overall , the twelve stars have abundances consistent with the models for the elements ca , ti , and ni . this also holds true for other elements not shown in figure [ abunda ] , such as cr , mn , and co. abundances for the five low - residual stars ( blue filled circles ) present , in general , good agreement with the values predicted by models with m @xmath123 .
some exceptions include : @xmath74 for 02 and 04 ( however , @xmath124 agrees well in both cases ) ; @xmath125 for 01 and 03 is about 0.5 dex lower than the model values .
even though the al abundance was determined from spectral synthesis , that region ( @xmath693961 ) presents strong ch and cn absorption features , which could compromise the determination .
for the seven stars with higher median residual values , there are clear mismatches between the observations and the theoretical predictions . for 07and 08 , carbon and nitrogen are at least 1 dex lower than the model values , and na , mg , and al are also below the model ranges .
even though 06 exhibits better agreement for c and n , the two bottom panels show that four stars are below the model expectations for @xmath70 vs. @xmath103 and @xmath70 vs. @xmath105 .
there is also no agreement for na , mg , and al ( 09 and 12 ) as a function of metallicity .
inspection of the two bottom panels of figure [ abunda ] reveals that c , na , and mg abundances may suggest some deficiencies in the current models , which could exclude these as possible progenitors for four of the seven high - residual stars . in both cases ,
the carbon abundances are over - predicted by the models when compared to the observations .
this suggests that the current models do not adequately capture the progenitors of the full set of ump stars .
the combination of carbon , sodium , and magnesium was used in paper i as part of the justification for the division of the cemp - no stars in groups ii and iii . when comparing the low- and high - residual stars with their cemp - no group classifications , it can be seen that half of the group ii stars ( 09 and 12 ) are outside the model ranges for @xmath70 vs. @xmath103 and @xmath70 vs. @xmath105 .
the two other group ii stars ( 01 and 03 ) have low median residuals , and are close to the limits explored by the models . another interesting case is 04 ( group iii ) , which is the most metal - poor star in this sample ; it presents a low median residual , unlike the other stars with @xmath1$]@xmath126 .
its chemical abundances are in regions where the model grid is sparse , specially for c , n , na , and mg .
regardless , the models used in this work can not account for the abundance patterns of 09 and 12 , which are also group iii stars based on paper i. this could be further evidence that at least one additional , different progenitor population operates at the lowest metallicities , such as the spinstars described by @xcite . for completeness , we also tested the robustness of our fitting results with respect to the recent study of ezzeddine et al .
( 2016 , in preparation ) .
this study suggests the presence of large positive corrections to the fe abundances of the 18 most iron - poor stars , following line formation computations in a non local thermodynamic equilibrium ( nlte ) .
while it would be inconsistent to mix lte and nlte abundances in a given stellar pattern , we tested how trial corrections for @xmath127 of 0.75dex to 04 , 0.70dex to 09 , and 0.3dex to 01 would affect our conclusions . as a result , no significant changes in the progenitor mass , explosion energy , and mean squared residuals were found . to properly interpret these new nlte fe abundances
, we thus have to await the full nlte abundance patterns and compare those with the supernova yields , once available .
we have presented the first high - resolution spectroscopic study of the ultra metal - poor , cemp - no star .
this star adds to the small number of stars with @xmath1 $ ] @xmath88 identified to date , and presents the same behavior as most ( more than 80% ) of ump stars : @xmath3$]@xmath128 and @xmath5$]@xmath129 .
we have attempted to characterize the progenitor stellar population of ump stars by comparing the yields from a grid of metal - free , massive - star models with the observed elemental abundances for and 11 other ump stars from the literature for which at least abundances of c , n , na , mg , al , and fe have been reported .
based on our residual analysis , we find that 42% ( 5 of 12 ) of the sample stars have elemental - abundance patterns that are well - reproduced by the sne models of @xcite .
we conclude that this class of sne explosions and their associated nucleosynthesis can not alone account for the observed abundance patterns of the entire set of ump stars we have considered . in particular , pop iii , massive metal - free star models can not reproduce abundance patterns of stars with @xmath70 @xmath130 , @xmath103 @xmath131 , and @xmath105 @xmath132 .
carbon and sodium could potentially be used as tracers of the stellar progenitor , in addition to the metallicity .
this could be evidence of different progenitor signatures , such as fast - rotating massive pop iii stars from @xcite , or the faint sne from @xcite .
if we assume that the models used in this work are a viable candidate for the stellar progenitors of ump stars , there must exist _ at least _ one additional class of progenitor operating at @xmath1$]@xmath10 , or a single process , yet to be identified , that could account for the abundances of all ump stars .
we should also acknowledge the possibility that more than one progenitor could contribute to the observed abundance patterns of the high - residual stars presented in this work .
recent evidence presented by @xcite suggests that the carbon abundance may be key to differentiate between different stellar progenitors of ump stars .
the classifications based on absolute carbon abundances , @xmath7(c ) , are further confirmed when looking at the sodium and magnesium abundances .
the analysis presented in this paper supports this hypothesis , and suggests that the progenitor population(s ) for ump stars may be even more rich and complex than previously thought .
we emphasize that abundances for additional ump stars , in particular those in the metallicity regime @xmath133 [ fe / h ] @xmath88 , are needed to better constrain the nature of the possible stellar progenitors .
nitrogen abundances are an important constraint to the theoretical models , and currently more than half of the observed ump stars have only limits for this element reported .
we thank chris sneden , who kindly provided part of the linelists used for the spectral synthesis , nozomu tominaga , for clarifying specific characteristics of the faint sne models , richard stancliffe , for clarifications on the stellar evolution models used for the cn corrections , and the anonymous referee , who made valuable suggestions that helped improve the paper . v.m.p .
, t.c.b . , and j.y
. acknowledge partial support for this work from the national science foundation under grant no .
phy-1430152 ( jina center for the evolution of the elements ) .
a.f . and a.c .
are supported by nsf career grant ast-1255160 .
a.h . was supported by a future fellowship of the australian research council ( ft120100363 ) .
this research has made use of nasa s astrophysics data system bibliographic services ; the arxiv pre - print server operated by cornell university ; the simbad database hosted by the strasbourg astronomical data center ; the iraf software packages distributed by the national optical astronomy observatories , which are operated by aura , under cooperative agreement with the nsf ; the saga database ( stellar abundances for galactic archeology ; * ? ? ? * ) ; the ` r - project ` software package @xcite ; the ` gnuplot ` command - line plotting program @xcite ; and the online q&a platform ` stackoverflow ` ( http://stackoverflow.com/ ) .
lrrrrr c ch & 4246.000 & & & syn & 5.83 + c ch & 4313.000 & & & syn & 5.73 + n cn & 3883.000 & & & syn & 6.13 + na i & 5889.950 & 0.00 & 0.108 & 97.06 & 2.77 + na i & 5895.924 & 0.00 & @xmath20.194 & 83.35 & 2.79
+ mg i & 3829.355 & 2.71 & @xmath20.208 & syn & 4.60
+ mg i & 3832.304 & 2.71 & 0.270 & syn & 4.60
+ mg i & 3838.292 & 2.72 & 0.490 & syn & 4.60
+ mg i & 4571.096 & 0.00 & @xmath25.688 & syn & 4.90
+ mg i & 4702.990 & 4.33 & @xmath20.380 & syn & 4.90
+ mg i & 5172.684 & 2.71 & @xmath20.450 & syn & 4.25
+ mg i & 5183.604 & 2.72 & @xmath20.239 & syn & 4.25
+ mg i & 5528.405 & 4.34 & @xmath20.498 & syn & 4.85 + al i & 3961.520 & 0.01 & @xmath20.340 & syn & 2.05
+ si i & 4102.936 & 1.91 & @xmath23.140 & syn & 4.36 + ca i & 4454.780 & 1.90 & 0.260 & syn & 2.64 + ca i & 5588.760 & 2.52 & 0.210 & 4.87 & 2.71 + ca i & 6122.220 & 1.89 & @xmath20.315 & 8.78 & 2.75 + ca
i & 6162.170 & 1.90 & @xmath20.089 & 11.92 & 2.68 + sc ii & 4246.820 & 0.32 & 0.240 & 57.45 & @xmath20.75 + sc ii & 4400.389 & 0.61 & @xmath20.540 & 13.70 & @xmath20.69 + sc ii & 4415.544 & 0.59 & @xmath20.670 & 10.63 & @xmath20.72 + ti i & 3989.760 & 0.02 & @xmath20.062 & 10.92 & 1.14 + ti i & 4533.249 & 0.85 & 0.532 & 6.36 & 1.19 + ti i & 4981.730 & 0.84 & 0.560 & 7.55 & 1.17 + ti
i & 4991.070 & 0.84 & 0.436 & 6.31 & 1.20 +
ti ii & 3759.291 & 0.61 & 0.280 & 85.34 & 1.10 + ti ii & 3761.320 & 0.57 & 0.180 & 84.26 & 1.12 + ti ii & 4012.396 & 0.57 & @xmath21.750 & 26.82 & 1.30 + ti ii & 4417.714 & 1.17 & @xmath21.190 & 23.80 & 1.31 + ti ii & 4443.801 & 1.08 & @xmath20.720 & 39.21 & 1.07 + ti ii & 4450.482 & 1.08 & @xmath21.520 & 17.08 & 1.34 + ti ii & 4468.517 & 1.13 & @xmath20.600 & 42.03 & 1.07 + ti ii & 4501.270 & 1.12 & @xmath20.770 & 37.73 & 1.13 + ti ii & 4533.960 & 1.24 & @xmath20.530 & 42.09 & 1.12 + ti ii & 4563.770 & 1.22 & @xmath20.960 & 29.43 & 1.25 + ti ii & 4571.971 & 1.57 & @xmath20.320 & 38.54 & 1.23 + cr i & 4254.332 & 0.00 & @xmath20.114 & 57.03 & 1.50 + cr i & 4274.800 & 0.00 & @xmath20.220 & 52.38 & 1.48 + cr i & 4289.720 & 0.00 & @xmath20.370 & 49.12 & 1.56 + cr i & 5206.040 & 0.94 & 0.020 & 19.37 & 1.48 + cr i & 5208.419 & 0.94 & 0.160 & 26.73 & 1.52
+ mn i & 4041.380 & 2.11 & @xmath20.350 & syn & 1.43
+ mn i & 4754.021 & 2.28 & @xmath20.647 & syn & 1.43
+ mn i & 4783.424 & 2.30 & @xmath20.736 & syn & 1.43 + mn i & 4823.514 & 2.32 & @xmath20.466 & syn & 1.43 + fe i & 3727.619 & 0.96 & @xmath20.609 & 70.55 & 3.28 + fe i & 3743.362 & 0.99 & @xmath20.790 & 65.97 & 3.32 + fe i & 3753.611 & 2.18 & @xmath20.890 & 16.00 & 3.41 + fe i & 3758.233 & 0.96 & @xmath20.005 & 97.02 & 3.50 + fe i & 3763.789 & 0.99 & @xmath20.221 & 80.31 & 3.23 + fe i & 3765.539 & 3.24 & 0.482 & 13.97 & 3.20 + fe i & 3767.192 & 1.01 & @xmath20.390 & 78.19 & 3.35 + fe i & 3786.677 & 1.01 & @xmath22.185 & 15.44 & 3.30 + fe
i & 3820.425 & 0.86 & 0.157 & 105.31 & 3.33 + fe i & 3824.444 & 0.00 & @xmath21.360 & 82.76 & 3.28 + fe i & 3886.282 & 0.05 & @xmath21.080 & 96.13 & 3.48 + fe i & 3887.048 & 0.91 & @xmath21.140 & 57.68 & 3.22 + fe i & 3899.707 & 0.09 & @xmath21.515 & 80.32 & 3.39 + fe i & 3902.946 & 1.56 & @xmath20.442 & 57.99 & 3.30 + fe i & 3917.181 & 0.99 & @xmath22.155 & 25.67 & 3.51 + fe i & 3920.258 & 0.12 & @xmath21.734 & 72.13 & 3.33 + fe i & 3922.912 & 0.05 & @xmath21.626 & 83.66 & 3.57 + fe i & 3940.878 & 0.96 & @xmath22.600 & 9.80 & 3.39 + fe i & 3949.953 & 2.18 & @xmath21.251 & 9.97 & 3.49 + fe
i & 3977.741 & 2.20 & @xmath21.120 & 14.87 & 3.58 + fe i & 4005.242 & 1.56 & @xmath20.583 & 57.44 & 3.38 + fe i & 4045.812 & 1.49 & 0.284 & 91.47 & 3.49 + fe i & 4063.594 & 1.56 & 0.062 & 80.91 & 3.45 + fe i & 4071.738 & 1.61 & @xmath20.008 & 71.90 & 3.28 + fe i & 4132.058 & 1.61 & @xmath20.675 & 56.36 & 3.47 + fe i & 4143.414 & 3.05 & @xmath20.200 & 8.23 & 3.32 + fe i & 4143.868 & 1.56 & @xmath20.511 & 64.30 & 3.46 + fe i & 4147.669 & 1.48 & @xmath22.071 & 12.23 & 3.55 + fe i & 4181.755 & 2.83 & @xmath20.371 & 14.24 & 3.51 + fe i & 4187.039 & 2.45 & @xmath20.514 & 20.15 & 3.40 + fe i & 4187.795 & 2.42 & @xmath20.510 & 25.89 & 3.51 + fe i & 4191.430 & 2.47 & @xmath20.666 & 13.35 & 3.35 + fe i & 4202.029 & 1.49 & @xmath20.689 & 60.77 & 3.43 + fe
i & 4216.184 & 0.00 & @xmath23.357 & 31.11 & 3.60 + fe i & 4227.427 & 3.33 & 0.266 & 11.13 & 3.31 + fe i & 4233.603 & 2.48 & @xmath20.579 & 18.59 & 3.44 + fe i & 4250.787 & 1.56 & @xmath20.713 & 65.76 & 3.67 + fe i & 4260.474 & 2.40 & 0.077 & 42.71 & 3.27 + fe i & 4337.046 & 1.56 & @xmath21.695 & 20.27 & 3.51 + fe i & 4375.930 & 0.00 & @xmath23.005 & 36.62 & 3.33
+ fe i & 4383.545 & 1.48 & 0.200 & 93.96 & 3.46 + fe i & 4404.750 & 1.56 & @xmath20.147 & 84.25 & 3.60 + fe i & 4415.122 & 1.61 & @xmath20.621 & 62.06 & 3.48 + fe i & 4427.310 & 0.05 & @xmath22.924 & 50.84 & 3.61 + fe i & 4447.717 & 2.22 & @xmath21.339 & 10.82 & 3.58 + fe i & 4459.118 & 2.18 & @xmath21.279 & 14.36 & 3.62 + fe i & 4461.653 & 0.09 & @xmath23.194 & 31.59 & 3.50 + fe i & 4494.563 & 2.20 & @xmath21.143 & 14.62 & 3.51 + fe
i & 4528.614 & 2.18 & @xmath20.822 & 23.98 & 3.44 + fe i & 4531.148 & 1.48 & @xmath22.101 & 14.00 & 3.59 + fe i & 4602.941 & 1.49 & @xmath22.208 & 13.52 & 3.68 + fe i & 4871.318 & 2.87 & @xmath20.362 & 13.81 & 3.44 + fe
i & 4872.137 & 2.88 & @xmath20.567 & 8.68 & 3.43 + fe i & 4890.755 & 2.88 & @xmath20.394 & 11.25 & 3.37 + fe i & 4891.492 & 2.85 & @xmath20.111 & 19.51 & 3.35 + fe i & 4918.994 & 2.85 & @xmath20.342 & 13.04 & 3.36 + fe i & 4920.503 & 2.83 & 0.068 & 24.34 & 3.28 + fe i & 5012.068 & 0.86 & @xmath22.642 & 22.92 & 3.60 + fe i & 5041.756 & 1.49 & @xmath22.200 & 9.41 & 3.44 + fe i & 5051.634 & 0.92 & @xmath22.764 & 16.69 & 3.61 + fe i & 5083.339 & 0.96 & @xmath22.842 & 12.69 & 3.59 + fe i & 5142.929 & 0.96 & @xmath23.080 & 9.26 & 3.67 + fe i & 5171.596 & 1.49 & @xmath21.721 & 24.59 & 3.46 + fe i & 5194.942 & 1.56 & @xmath22.021 & 14.64 & 3.55 + fe i & 5216.274 & 1.61 & @xmath22.082 & 10.02 & 3.48 + fe i & 5232.940 & 2.94 & @xmath20.057 & 17.98 & 3.33 + fe i & 5269.537 & 0.86 & @xmath21.333 & 80.63 & 3.60 + fe i & 5328.039 & 0.92 & @xmath21.466 & 74.91 & 3.64 + fe i & 5328.531 & 1.56 & @xmath21.850 & 21.81 & 3.59 + fe i & 5371.489 & 0.96 & @xmath21.644 & 65.45 & 3.61 + fe i & 5397.128 & 0.92 & @xmath21.982 & 44.81 & 3.44 + fe i & 5405.775 & 0.99 & @xmath21.852 & 45.96 & 3.42 + fe i & 5429.696 & 0.96 & @xmath21.881 & 45.71 & 3.40 + fe i & 5434.524 & 1.01 & @xmath22.126 & 37.72 & 3.56 + fe i & 5446.917 & 0.99 & @xmath21.910 & 45.31 & 3.46 + fe i & 5455.609 & 1.01 & @xmath22.090 & 38.32 & 3.53 + fe i & 5615.644 & 3.33 & 0.050 & 10.91 & 3.39 + fe ii & 4583.840 & 2.81 & @xmath21.930 & 15.01 & 3.63 + fe ii & 4923.930 & 2.89 & @xmath21.320 & 25.42 & 3.38 + fe ii & 5018.450 & 2.89 & @xmath21.220 & 28.24 & 3.34 + co i & 3845.468 & 0.92 & 0.010 & 33.28 & 1.33 + co i & 3995.306 & 0.92 & @xmath20.220 & 21.90 & 1.24 + co i & 4118.767 & 1.05 & @xmath20.490 & 12.50 & 1.33 + co i & 4121.318 & 0.92 & @xmath20.320 & 21.33 & 1.29 + ni i & 3500.850 & 0.17 & @xmath21.294 & 38.99 & 1.92 + ni i & 3519.770 & 0.28 & @xmath21.422 & 32.90 & 2.00 + ni i & 3566.370 & 0.42 & @xmath20.251 & 61.60 & 1.90
+ ni i & 3597.710 & 0.21 & @xmath21.115 & 44.05 & 1.88
+ ni i & 3783.520 & 0.42 & @xmath21.420 & 32.18 & 2.01
+ ni i & 3807.140 & 0.42 & @xmath21.220 & 36.58 & 1.91
+ ni i & 3858.301 & 0.42 & @xmath20.951 & 51.19 & 1.99 + sr ii & 4077.714 & 0.00 & 0.150 & syn & @xmath21.93 + sr ii & 4215.524 & 0.00 & @xmath20.180 & syn & @xmath21.88 + ba ii & 4554.033 & 0.00 & 0.163 & syn & @xmath22.97 + ba ii & 4934.086 & 0.00 & @xmath20.160 & syn & @xmath22.97 + eu ii & 4129.720 & 0.00 & 0.220 & syn & @xmath653.28 + pb
i & 4057.810 & 1.32 & @xmath20.170 & syn & @xmath650.25 +
|
we report on the first high - resolution spectroscopic analysis of , a bright ( @xmath0 ) , ultra metal - poor ( @xmath1 $ ] = @xmath24.1 ) , carbon - enhanced ( @xmath3 $ ] = @xmath41.7 ) star selected from the hamburg / eso survey .
this star exhibits low abundances of neutron - capture elements ( @xmath5 $ ] = @xmath6 ) , and an absolute carbon abundance @xmath7(c ) = 6.1 ; based on either criterion , is sub - classified as a cemp - no star .
we show that the light - element abundance pattern of is consistent with predicted yields from a massive ( m = @xmath8 ) , primordial composition , supernova ( sn ) progenitor .
we also compare the abundance patterns of other ultra metal - poor stars from the literature with available measures of c , n , na , mg , and fe abundances with an extensive grid of sn models ( covering the mass range @xmath9 ) , in order to probe the nature of their likely stellar progenitors .
our results suggest that at least two classes of progenitors are required at @xmath1 $ ] @xmath10 , as the abundance patterns for more than half of the sample studied in this work ( 7 out of 12 stars ) can not be easily reproduced by the predicted yields .
|
it is now generally agreed that type ia supernovae originate from accreting white dwarfs , and widely accepted that their occurrence signals arrival at the chandrasekhar mass @xmath4 .
however there is no consensus as to how the white dwarf gains mass ( see e.g. livio , 2001 , for a review ) .
two possibilities are currently discussed . in the first
, two white dwarfs merge as their relative orbit shrinks under gravitational radiation ( the double degenerate scenario ) . in the second possibility
, a white dwarf accretes from a non degenerate companion ( the single degenerate scenario ) .
observational evidence is currently too sparse to give a strong preference to either picture .
however both have serious theoretical drawbacks . for white dwarf mergers , many authors ( saio & nomoto , 1985 , 1998 ; kawai , saio & nomoto , 1987 ; mochkovitch & livio , 1990 ; timmes , woosley & taam , 1994 ; mochkovitch , guerrero & segretain , 1997 ) argue that no explosion takes place and that the result is instead a quiet , accretion
induced collapse ( aic ) , forming a neutron star .
the single degenerate picture suffers from two main problems .
first , growth of the white dwarf mass @xmath5 requires efficient nuclear burning of most of the accreted hydrogen .
this happens only for accretion rates @xmath6 satisfying @xmath7 where @xmath8 ( nomoto et al .
, 1979 ; fujimoto , 1982 ) . for lower @xmath9
the burning occurs in nova explosions .
these expel matter and leave the white dwarf mass essentially unchanged . for @xmath10
not all of the hydrogen is burnt . in some pictures
much of the excess is expelled in a radiatively driven wind : in e.g. the strong wind solution of hachisu et al .
( 1996 ) the burning rate is limited by @xmath11 .
this allows mass growth to continue , but for @xmath12 this is clearly an inefficient process , as only a small fraction of the transferred mass is gained by the white dwarf .
if there is no significant wind things may be still worse , as the excess accretion will probably form a common envelope around the binary , causing it to merge and become a massive red giant .
the second difficulty with this single degenerate scenario results from the proximity of a large h
rich companion at the time the supernova explodes .
this may contaminate the supernova ejecta ( marietta , burrows & fryxell , 2000 ) , in conflict with the defining characteristic of type ia sne as having no detectable hydrogen ( see , however , lentz et al .
2002 ) . the @xmath9 constraint rules out most candidate progenitor binaries for the single degenerate scenario .
cataclysmic variables ( cvs ) generally have mass transfer rates @xmath13 and so are prone to losing the transferred mass in nova explosions .
supersoft x ray binaries ( sss ) are more promising . here in contrast
to cvs the companion / wd mass ratio @xmath14 is @xmath15 , and mass transfer shrinks the companion s roche lobe relative to the stellar surface . for a companion with a radiative envelope the result is mass transfer on a thermal timescale , i.e. @xmath16 .
this is close to the required range if the companion has mass @xmath17 and thus a kelvin
helmholtz timescale @xmath18 yr ( van den heuvel et al . , 1992 ) .
the white dwarf in such sss binaries can gain mass .
however , systematic calculations ( e.g. langer et al . , 2000 ) show that it may be difficult to grow @xmath5 from a typical value @xmath2 all the way to @xmath19 .
thus only systems starting with rather more massive white dwarfs give sne ia .
this in turn means that the main sequence progenitor of the white dwarf must itself have been rather massive , significantly reducing the number of systems which can give sne ia by this channel .
in addition , this type of evolution is vulnerable to the hydrogen contamination constraint .
not only is the companion fairly massive and hydrogen rich , its roche lobe offers a large target for the expanding sn shell as the mass ratio @xmath20 must be @xmath15 when the explosion occurs . in this paper
we reconsider the single degenerate scenario , and offer an alternative that avoids these difficulties .
we have seen that sss binaries do not readily produce sne ia because the white dwarf mass does not grow enough .
the accretion rate is sensitive to the mass ratio @xmath20 , and moves out of the narrow band ( [ burn ] ) in which efficient mass gain occurs before much mass is transferred .
an accretion rate which remains almost constant as @xmath20 decreases is clearly a major advantage for a plausible progenitor wd binary class .
only one type of wd binary satisfies this requirement : long period ( @xmath21 d ) systems where the wd accretes from a low mass red giant .
these systems are driven by the nuclear expansion of the red giant .
the helium core mass @xmath22 of this star increases slightly through shell burning during the binary evolution , causing considerable envelope expansion .
the envelope mass is depleted by shell burning but much more by mass transfer to the white dwarf .
calculations ( webbink , rappaport & savonije , 1983 ; ritter , 1999 ) show that the mass transfer rate varies remarkably little while most of the companion s envelope is transferred , the value of the rate being fixed by the core mass or equivalently the binary period .
eq ( 36 ) of ritter ( 1999 ) relates @xmath23 directly to the companion mass @xmath24 for the case where all the transferred mass is accreted by the white dwarf .
[ fig : mdot ] given by eq ( 36 ) of ritter ( 1999 ) plotted as a function of @xmath24 for initial wd and donor masses @xmath25 and various initial donor core masses @xmath26 . ]
must avoid the upper hatched region for the disc to be unstable to dwarf nova outbursts .
the narrow hatched band is the efficient accretion region ( [ burn ] ) .
the assumed duty cycle is 0.004 and the masses of donor , core and white dwarf at the start of this phase are 0.53@xmath27 , 0.2@xmath27 and 1.1@xmath27 respectively . ] .
evolution of the donor core and envelope masses , and the white dwarf mass .
after @xmath28 years the white dwarf reaches the chandrasekhar mass and a type ia supernova ensues . ] shows that for a wd starting at @xmath29 the mass transfer rate @xmath23 changes by no more than factors @xmath30 as @xmath24 decreases from @xmath31 to @xmath32 . while this is encouraging for a candidate sne ia progenitor there appears at first sight to be a major difficulty .
the mass transfer rates in such binaries are typically only @xmath33 , more than a factor 10 short of the steady burning regime ( [ burn ] )
. however we must distinguish the mass _ transfer _ rate @xmath23 from the mass _ accretion _ rate @xmath9 .
all these long period systems are so wide that their accretion discs have ionization zones ; they are dwarf novae ( king et al . , 1997 ) .
this point was previously neglected but is crucial for modeling these long period systems correctly .
it is probably the shortness of the duty cycle @xmath34 , together with the low space density resulting from their relatively rapid evolution that makes discovery of long period systems of this type difficult .
accretion on to the white dwarf occurs almost entirely during outbursts , with accretion rates @xmath35 which can therefore be close to the efficient burning regime .
the mean duty cycle @xmath34 is presumably itself a function of the evolutionary state of the binary , although the lack of understanding of disc viscosity in current theory does not allow one to predict it with any certainty .
however for typical values @xmath36 to a few times @xmath37 one can always find initial white dwarf , donor and core masses which keep the accretion rate during outbursts inside the efficient burning regime ( [ burn ] ) through most of the evolution .
this kind of mass gain therefore works even if wind loss is inefficient for @xmath38 . figs .
[ fig : rgbtracks1 ] , [ fig : rgbtracks2 ] show a case where @xmath5 is able to grow to @xmath19 .
a further advantage for this picture is that the sn ia explosion must always occur for mass ratios @xmath39 , and indeed generally with very little of the hydrogen envelope of the secondary remaining .
this then avoids the second difficulty ( hydrogen contamination ) afflicting the single degenerate scenario .
the secondary is now a rather smaller target for the expanding supernova shell , and in any case contains little hydrogen .
we have seen from the previous section that long period dwarf novae offer a route for making sne ia .
[ fig : progenitors ] .
the precise value depends on the core mass at this stage , which must be in the approximate range 0.20.3@xmath27 , and the details of mass loss for @xmath40 . ]
gives the regions of @xmath41 space where growth to @xmath19 may be possible .
( this figure assumes conservative mass transfer , but we will show in a future paper that similar results are obtained when this assumption is relaxed . ) the required @xmath42 is somewhat higher than typically results from single star evolution .
this might seem to lead to the same need for massive main sequence progenitors and thus the same restriction on progenitor numbers as for the sss channel .
however the nuclear driven evolution discussed here is of course also open to the outcomes of thermal timescale mass transfer , i.e. the sss systems we discussed earlier . many of these , starting from typical white dwarf masses @xmath43 , have already increased them to values of the order required in fig . [ fig : progenitors ] . we therefore envisage a two stage process in the growth of @xmath5 towards @xmath19 ( fig . [
fig : ttmt ] ) .
first , a sss system with @xmath44 and @xmath45 undergoes thermal timescale mass transfer and increases the white dwarf mass to values @xmath46 .
timescale mass transfer stops once the mass ratio @xmath20 drops below a value @xmath47 .
what happens next to the binary depends on the competition between orbital angular momentum loss ( presumably through magnetic stellar wind braking ) and nuclear expansion of the secondary ( cf schenker et al . , 2002 ) .
if the angular momentum loss timescale is shorter than that for nuclear evolution the system becomes a short period cv , often with some signs of chemical evolution as in ae aquarii ( schenker et al . 2002 ) . if instead the nuclear evolution timescale is shorter , the system will expand and eventually become the kind of long period cv we discuss here .
for suitable masses
@xmath41 and orbital separation at the end of thermal timescale mass transfer the white dwarf mass can grow to @xmath19 and produce a sn ia . assuming mass transfer starts at or near the end of the main sequence and that the thermal timescale phase has little effect on the core mass of the donor when it reaches the base of the rgb , the minimum donor mass which reaches the rgb with core mass @xmath48 is @xmath49 . for massive donors ( mass ratios @xmath5034 ) sustained thermal
timescale mass transfer leads to dynamical instability ( hjellming 1989 ; kalogera & webbink 1996 ; kolb et al .
. this _ delayed dynamical instability _ limits the maximum donor mass before the thermal timescale phase to @xmath51 .
thus the donor mass at the start of mass transfer is limited to the approximate range 2 3@xmath27
. future work involving accurate numerical modelling of the thermal timescale evolution will refine this simple picture .
we have shown that long period dwarf novae offer a promising channel for making type ia supernovae .
in particular the white dwarf mass can be raised to the chandrasekhar value without the need to invoke fairly massive and thus rare main sequence progenitors .
moreover there is very little danger of hydrogen contamination of the supernova ejecta . clearly more work is needed .
in particular we need to calculate the birthrate of such systems and compare this with the inferred rate @xmath52 of sne ia per galaxy .
this in turn requires an extensive grid of models .
we shall return to this problem in a future paper .
theoretical astrophysics research at leicester is supported by a pparc rolling grant .
ark gratefully acknowledges a royal society wolfson research merit award .
we thank the referee , david branch , for helpful comments .
|
we show that long period dwarf novae offer a promising route for making type ia supernovae . for typical dwarf
nova duty cycles @xmath0 , mass is accreted by the white dwarf mainly during dwarf nova outbursts at rates allowing steady nuclear burning of most of the accreted matter .
mass gains up to @xmath1 are possible in this way .
although these are too small to allow a @xmath2 wd to reach the chandrasekhar mass , they are sufficient if the wd grew to @xmath3 in a previous episode of thermal timescale mass transfer , i.e. for those long period dwarf novae which descend from supersoft binaries .
a further advantage of this picture is that the supernova always occurs in a binary of small secondary / primary mass ratio , with the secondary having very little remaining hydrogen .
both features greatly reduce the possibility of hydrogen contamination of the supernova ejecta . accretion , accretion discs binaries : general x rays : binaries stars : dwarf novae
supernovae : general galaxies : stellar content cosmology : distance scale
|
despite nearly 40 years of work on accretion disk physics , the simple @xcite thin disk model and its relativistic cousins ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) remain the standard model despite many theoretical alternatives ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and some observational reservations ( see * ? ? ?
quasar accretion disks can not be spatially resolved with ordinary telescopes , so we have been forced to test accretion physics through time variability ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and spectral modeling ( e.g. * ? ? ?
* ; * ? ? ?
one notable success is the use of reverberation mapping @xcite of quasar broad line emission to calibrate the relation between emission line widths and black hole masses .
gravitational telescopes do , however , provide the necessary resolution to study the structure of the quasar continuum source .
each gravitationally lensed quasar image is observed through a magnifying screen created by the stars in the lens galaxy .
sources that are smaller than the einstein radius of the stars , typically @xmath9 cm , show time variable fluxes whose amplitude is determined by the source size ( see the review by * ? ? ?
smaller sources have larger variability amplitudes than larger sources . in this investigation , we exploit the optical microlensing variability in nine gravitationally lensed quasar systems to measure the size of their accretion disks , and we find that disk sizes are strongly correlated with the masses of their central black holes .
lccccc[b ] he0435 - 1223 & 0.50 & @xmath10 & 0.260 & @xmath11 & @xmath12 + sdss0924 + 0219 & 0.11 & @xmath13 & 0.277 & @xmath14 & @xmath15 + fbq0951 + 2635 & 0.89 & @xmath16 & 0.313 & @xmath17 & @xmath18 + he1104 - 1805 & 2.37 & @xmath19 & 0.211 & @xmath20 & @xmath21 + pg1115 + 080 & 0.92 & @xmath22 & 0.257 & @xmath23 & @xmath24 + rxj1131 - 1231 & 0.06 & @xmath25 & 0.422 & @xmath26 & @xmath15 + sdss1138 + 0314 & 0.04 & @xmath27 & 0.203 & @xmath28 & @xmath29 + sbs1520 + 530 & 0.88 & @xmath30 & 0.245 & @xmath31 & @xmath32 + q2237 + 030 & 1.3 & @xmath33 & 0.208 & @xmath34 & @xmath35 + [ tab : rs ] in [ sec : observations ] we describe the monitoring data , the lens models we use based on hubble space telescope ( hst ) images of each system and our microlensing analysis method . in [ sec : results ] we describe our accretion disk model and our results for the relationship between disk size and black hole mass . in
[ sec : discussion ] , we discuss the results and their implications for thin accretion disk theory .
all calculations in this paper assume a flat @xmath36cdm cosmology with @xmath37 , @xmath38 and @xmath39 .
we monitored the gravitationally lensed quasars he 04351223 , sdss 0924 + 0219 , fbq 0951 + 2635 , he 11041805 , pg 1115 + 080 , rxj 11311231 , sdss 1138 + 0314 , sbs 1520 + 530 and q 2237 + 030 in the @xmath40- and @xmath41-bands on the smarts 1.3 m using the andicam optical / infrared camera @xcite , the wisconsin - yale - indiana ( wiyn ) observatory using the wiyn tip tilt module ( wttm ) , the 2.4 m telescope at the mdm observatory using the mdm eight - k , echelle and retrocam @xcite imagers and the 6.5 m magellan baade telescope using imacs @xcite .
we supplemented our monitoring data with published quasar light curves from @xcite , @xcite , @xcite , @xcite and @xcite .
we measured the flux of each image by comparison to the flux from reference stars in the field of each frame .
our analysis of the monitoring data is described in detail by @xcite .
all nine lenses have been observed in the @xmath41- ( f555w ) , @xmath42- ( f814w ) and @xmath43-bands ( f160w ) using the wfpc2 , acs / wfc and nicmos instruments on _ hst_. we fit these images as combinations of point sources for the quasars and ( generally ) de vaucouleurs models for the lenses as described in @xcite .
these provided the astrometry used for lens models and defined a constant mass - to - light ( @xmath44 ) ratio model for the mass distribution in the lens models .
we modeled each system using the _ gravlens _ software package @xcite to generate a series of ten models for each lens starting from a constant @xmath44 model and then adding an nfw @xcite halo .
the sequence is parametrized by @xmath45 , the mass fraction represented by the visible lens galaxy relative to a constant @xmath44 model .
we start with the constant @xmath44 model , @xmath46 , and then reduce its mass in increments of @xmath47 with the nfw halo s mass rising to compensate .
these lens models then provide the convergence @xmath48 , shear @xmath49 and stellar surface density @xmath50 needed to define the microlensing magnification patterns .
we assume a lens galaxy stellar mass function @xmath51 with a dynamic range of a factor of 50 that approximates the galactic disk mass function of @xcite .
for the typical lens we generated 4 magnification patterns for each image in each of the 10 lens models .
we gave the magnification patterns an outer scale of @xmath52 , where @xmath53 is the einstein radius for the mean stellar mass @xmath54 .
this outer dimension is large enough to fairly sample the magnification pattern while the pixel scale is small enough to resolve the accretion disk .
we determined the properties of the accretion disk by modeling the observed light curves using the bayesian monte carlo method of @xcite ( also see * ? ? ? * ) . for a given disk model
we randomly generate light curves , fit them to the observations and then use bayesian methods to compute probability distributions for the disk size averaged over the lens models , the likely velocities of the observer , lens , source , stars , and mass .
we use a prior on the mean microlens mass of @xmath55 , but the disk size estimates are relatively insensitive to this assumption ( see * ? ? ? * ) .
we use black hole mass estimates for the quasars that are based on observed quasar emission line widths and the locally calibrated virial relations for black hole masses @xcite . for most systems we simply adopted the black hole mass estimates from @xcite based on the ( @xmath56 ) , ( @xmath57 ) and h@xmath58 ( @xmath59 ) mass - linewidth relations . for sdss 1138 + 0314 , we measured the width of the ( 1549 ) line in optical spectra from the sloan digital sky survey @xcite and estimated the black hole mass using the normalizations of @xcite .
these mass estimates are reliable to approximately 0.3 dex ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we model the surface brightness profile of the accretion disk as a power law temperature profile , @xmath8 , matching the outer regions of a @xcite thin disk model .
we neglect the central depression of the temperature due to the inner edge of the disk and corrections from general relativity to avoid extra parameters .
the effect of this simplification on our size estimates is small compared to our measurement uncertainties provided the disk size we obtain is several times larger than the radius of the inner disk edge .
we assume that the disk radiates as a black body , so the surface brightness at rest wavelength @xmath60 is @xmath61^{-1}\ ] ] where the scale length @xmath62^{1/3 } \nonumber \\ & = & 9.7 \times 10^{15 } \left ( { \lambda_{rest } \over { \rm \micron } } \right)^{4/3}\nonumber \\ & & \times \left ( { m_{bh } \over 10 ^ 9 { \rm m_{\sun } } } \right)^{2/3 } \left({l \over \eta l_e } \right)^{1/3 } { \rm cm}\end{aligned}\ ] ] is the radius at which the disk temperature matches the wavelength , @xmath63 , @xmath64 is the planck constant , @xmath65 is the boltzmann constant , @xmath66 is the black hole mass , @xmath67 is the mass accretion rate , @xmath68 is the luminosity in units of the eddington luminosity , and @xmath5 is the accretion efficiency . we can also compute the size under the same model assumptions based on the magnification - corrected @xmath42-band quasar fluxes measured in hst observations as @xmath69 where @xmath70 is the angular diameter distance to the quasar in units of the hubble radius , @xmath42 is the magnification - corrected magnitude and @xmath71 is the disk inclination angle .
our results are shown in figures [ fig : r2500 ] through [ fig : rflux ] and summarized in table [ tab : rs ] . for the comparison with theory and the figures
, we corrected the measured sizes to @xmath72 assuming the @xmath73 scaling of thin disk theory and the mean inclination @xmath74 .
there are two striking facts illustrated by the figures .
first , we clearly see from fig . [
fig : r2500 ] that the microlensing sizes are well correlated with the black hole mass .
a power - law fit between @xmath75 and @xmath66 yields : @xmath76 which is consistent with the predicted slope from thin disk theory ( @xmath3 ) and implies a typical eddington factor of @xmath77 if we fix the slope with mass to 2/3 ( see fig .
2 ) . @xcite estimate that the typical quasar has @xmath78 , which would indicate a radiative efficiency of @xmath79 .
this efficiency is low compared to standard models ( e.g. * ? ? ?
second , we find that microlensing sizes are well correlated with sizes estimated from thin disk theory and from the observed flux ( figs . [
fig : r2500 ] and [ fig : rflux ] ) , but the three size estimates show systematic offsets in scale . most of the offset between the microlensing disk size measurements and thin disk theory size estimates could be explained by the existing uncertainties , but the offset from the estimate based on the quasar flux is much more significant ( fig . [
fig : r2500 ] ) .
while the microlensing and flux sizes are well - correlated , the measured disk sizes are @xmath80 dex larger than predicted from the observed flux .
simply put , the quasars are not sufficiently luminous to be radiating as black bodies with a @xmath8 temperature profile . @xcite also noticed this problem in their more qualitative study of lensed quasars with x - ray observations .
are the discrepancies between our measurements and the size estimated from the disk flux due to a problem in the measurements , an oversimplification of the disk model or a fundamental problem in the thin disk model ?
we have tested our approach using monte carlo simulations of light curves and verified that we recover the input disk sizes .
our results are also only weakly sensitive to the assumed prior on the microlens masses ( see * ? ? ?
* for a discussion ) .
we will overestimate the source size if a significant fraction of the observed flux comes not from the continuum emission of the disk but from the larger and minimally microlensed line emitting regions ( e.g. * ? ? ?
this includes not only the obvious broad lines but also the broad and balmer continuum emission that can represent @xmath81 of the apparent continuum flux at some wavelengths @xcite .
we have experimented with adding a fraction of unmicrolensed light and found that 30% contamination does not lead to sufficiently large size changes to resolve the problem .
the sizes shrink by approximately 20% .
conservatively , black hole mass estimates from the virial technique have a scatter of a factor of @xmath6 @xcite , which contributes only 0.3 dex of scatter to the disk size estimates .
the size estimates from the flux could be affected by misestimating the magnification or failing to correct for extinction in the lens galaxy , but the uncertainties in the magnifications are only a factor of @xmath82 at worst and none of these lenses shows significant extinction @xcite relative to the magnitude of the discrepancy .
the problem is also not a consequence of the obvious flaws in our simplified disk model the neglect of the inner edge and the different temperature profile of a relativistic disk ( e.g. * ? ? ?
* ) . for observations at a fixed wavelength , neglecting the inner edge does not have a dramatic effect on the effective source size and the effects of relativity on the temperature profile are modest .
studies of the temperature profile in the disk with microlensing are best done by measuring the variation in source size with wavelength because size ratios can be measured much more accurately than absolute sizes . @xcite
have used the wavelength dependence of the microlensing in he 1104 - 1805 to derive a slope @xmath83 or @xmath84 of @xmath85 that is consistent with thin disk theory but would also allow a shallower temperature profile that would reduce the differences between the microlensing and flux size estimates .
the next step is clearly to use more sophisticated disk models including relativity and model atmospheres such as @xcite or @xcite .
@xcite have argued that microlensing essentially measures the half - light radius ( @xmath86 for our model ) , so the first step should be to try to simultaneously match our size estimates and the observed fluxes assuming this to be the case , since the many additional parameters of the full disk models will make their direct inclusion in the microlensing calculations a major computational challenge .
we thank e. agol , m. dietrich , c. onken , b. peterson , m. pinsonneault , r. pogge and p. osmer for discussions on quasar structure , and m. mortonson , s. poindexter , s. rappaport , and p. schechter for discussions on microlensing .
this research made extensive use of a beowulf computer cluster obtained through the cluster ohio program of the ohio supercomputer center .
support for program hst - go-9744 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas-5 - 26666 .
|
we use the microlensing variability observed for nine gravitationally lensed quasars to show that the accretion disk size at 2500 is related to the black hole mass by @xmath0@xmath1m@xmath2 .
this scaling is consistent with the expectation from thin disk theory ( @xmath3 ) , but it implies that black holes radiate with relatively low efficiency , @xmath4 where @xmath5 .
these sizes are also larger , by a factor of @xmath6 , than the size needed to produce the observed @xmath7 quasar flux by thermal radiation from a thin disk with the same @xmath8 temperature profile .
more sophisticated disk models are clearly required , particularly as our continuing observations improve the precision of the measurements and yield estimates of the scaling with wavelength and accretion rate .
|
massive star - forming regions have traditionally been prime targets for astrochemistry owing to their bright molecular lines ( e.g. , johansson et al .
1984 , cummins et al .
1986 , irvine et al . 1987 , ohishi 1997 )
. massive young stellar objects ( ysos ) have luminosities of @xmath1 l@xmath2 and involve young o- and b - type stars . because their formation proceeds more rapidly than that of low - mass stars and involves ionizing radiation , substantial chemical differences may be expected .
the formation of high mass stars is much less well understood than that of low - mass stars .
for example , observational phenomena such as ultracompact h ii regions , hot cores , masers and outflows have not yet been linked into a single evolutionary picture .
chemistry may well be an important diagnostic tool in establishing such a sequence .
most of the early work on massive star - forming regions has centered on two sources , orion kl and sgrb2 . numerous line surveys at millimeter ( e.g. , blake et al .
1987 , turner 1991 ) and submillimeter ( jewell et al . 1989 , sutton et al .
1991 , 1995 , schilke et al . 1997 ) wavelengths have led to an extensive inventory of molecules through identification of thousands of lines .
in addition , the surveys have shown strong chemical variations between different sources . in recent years , new observational tools have allowed a more detailed and systematic study of the envelopes of massive ysos .
submillimeter observations routinely sample smaller beams ( typically 15@xmath3 vs. 30@xmath31@xmath4 ) and higher critical densities ( @xmath5 vs. @xmath6 @xmath7 ) than the earlier work .
moreover , interferometers at 3 and 1 millimeter provide maps with resolutions of @xmath85@xmath3 . finally , ground- and space - based infrared observations allow both the gas and the ices to be sampled ( e.g. , evans et al . 1991 , van dishoeck et al . 1999 ) .
these observational developments have led to a revival of the study of massive star formation within the last few years .
recent overviews of the physical aspects of high - mass star formation are found in churchwell ( 1999 ) and garay & lizano ( 1999 ) . in this brief review
, we will first summarize available observational diagnostics to study the different phases and physical components associated with massive star formation .
subsequently , an overview of recent results on intermediate mass ysos is given , which are often better characterized than their high - mass counterparts because of their closer distance .
subsequently , we will discuss a specific sample of embedded massive ysos which have been studied through a combination of infrared and submillimeter data .
after illustrating the modeling techniques , we address the question how the observed chemical variations are related to evolutionary effects , different conditions in the envelope ( e.g. , @xmath9 , mass ) or different luminosities of the ysos .
more extensive overviews of the chemical evolution of star - forming regions are given by van dishoeck & blake ( 1998 ) , hartquist et al .
( 1998 ) , van dishoeck & hogerheijde ( 1999 ) and langer et al .
schilke et al .
( this volume ) present high spatial resolution interferometer studies , whereas macdonald & thompson ( this volume ) focus on submillimeter data of hot core / ultracompact h ii regions .
ices are discussed by ehrenfreund & schutte ( this volume ) .
the majority of molecules are detected at ( sub-)millimeter wavelengths , and line surveys highlight the large variations in chemical composition between different ysos , both within the same parent molecular cloud and between different clouds .
the recent 13 mm surveys of sgr b2 ( nummelin et al . 1998 , ohishi & kaifu 1999 ) dramatically illustrate the strong variations between various positions ( see figure 1 ) .
the north position is typical of ` hot core'-type spectra , which are rich in lines of saturated organic molecules .
this position has also been named the ` large molecule heimat ' ( e.g. , kuan & snyder 1994 , liu & snyder 1999 ) .
the middle position has strong so and so@xmath10 lines , whereas the northwest position has a less - crowded spectrum with lines of ions and long carbon chains .
a similar differentiation has been observed for three positions in the w 3 giant molecular cloud by helmich & van dishoeck ( 1997 ) , who suggested an evolutionary sequence based on the chemistry .
the availability of complete infrared spectra from 2.4200 @xmath11 m with the _ infrared space observatory _ ( iso ) allows complementary variations in infrared features to be studied .
figure 2 shows an example of iso sws and lws spectra of two objects : cep a ( @xmath12 l@xmath13 ) and s 106 ( @xmath14 l@xmath13 ) .
the cep a spectrum is characteristic of the deeply embedded phase , in which the silicates and ices in the cold envelope are seen in absorption .
the s 106 spectrum is typical of a more evolved massive yso , with strong atomic and ionic lines in emission and prominent pah features .
a similar sequence has been shown by ehrenfreund et al .
( 1998 ) for a set of southern massive young stars with luminosities up to @xmath15 l@xmath13 . the most successful models for explaining these different chemical characteristics involve accretion of species in an icy mantle during the ( pre-)collapse phase , followed by grain - surface chemistry and evaporation of ices once the yso has started to heat its surroundings ( e.g. , millar 1997 ) .
the evaporated molecules subsequently drive a rapid high - temperature gas - phase chemistry for a period of @xmath16 yr , resulting in complex , saturated organic molecules ( e.g. , charnley et al . 1992 , 1995 ;
charnley 1997 ; caselli et al .
1993 , viti & williams 1999 ) . the abundance ratios of species such as ch@xmath17och@xmath17/ch@xmath17oh and so@xmath10/h@xmath10s show strong variations with time , and may be used as ` chemical clocks ' for a period of 500030,000 yr since evaporation .
once most of the envelope has cleared , the ultraviolet radiation can escape and forms a photon - dominated region ( pdr ) at the surrounding cloud material , in which molecules are dissociated into radicals ( e.g. , hcn @xmath18 cn ) and pah molecules excited to produce infrared emission . the ( ultra-)compact h ii region gives rise to strong ionic lines due to photoionization .
lllll component & chemical & submillimeter & infrared & examples + & characteristics & diagnostics & diagnostics + dense cloud & low - t chemistry & ions , long - chains & simple ices & sgrb2 ( nw ) + & & ( hc@xmath19n , ... ) & ( h@xmath10o , co@xmath10 ) + cold envelope & low - t chemistry , & simple species & ices & n7538 irs9 , + & heavy depletions & ( cs , h@xmath10co ) & ( h@xmath10o , co@xmath10 , ch@xmath17oh ) & w 33a + inner warm & evaporation & high t@xmath20 & high gas / solid , high & gl 2591 , + envelope & & ( ch@xmath17oh ) & t@xmath20 , heated ices & gl 2136 + & & & ( c@xmath10h@xmath10 , h@xmath10o , co@xmath10 ) + hot core & high - t chemistry & complex organics & hydrides & orion hot core , + & & ( ch@xmath17och@xmath17 , ch@xmath17cn , & ( oh , h@xmath10o ) & sgrb2(n),g34.3 + & & vib .
excited mol . ) & & w 3(h@xmath10o ) + outflow : & shock chemistry , & si- and s - species & atomic lines , hydrides & w 3 irs5 , + direct impact & sputtering & ( sio , so@xmath10 ) & ( [ s i ] , h@xmath10o ) & sgrb2(m ) + pdr , compact & photodissociation , & ions , radicals & ionic lines , pahs & s 140 , + h ii regions & photoionization & ( cn / hcn , co@xmath21 ) & ( [ neii ] , [ cii ] ) & w 3 irs4 + table 1 summarizes the chemical characteristics of the various physical components , together with the observational diagnostics at submillimeter and infrared wavelengths . within the single - dish submillimeter and iso beams ,
many of these components are blended together and interferometer observations will be essential to disentangle them .
nevertheless , the single - dish data are useful because they encompass the entire envelope and highlight the dominant component in the beam . combined with the above chemical scenario
, one may then attempt to establish an evolutionary sequence of the sources .
the physical distinction between the ` hot core ' and the warm inner envelope listed in table 1 is currently not clear : does the ` hot core ' represent a separate physical component or is it simply the inner warm envelope at a different stage of chemical evolution ? even from an observational point of view , there appear to be different types of ` hot cores ' : some of them are internally heated by the young star ( e.g. , w 3(h@xmath10o ) ) , whereas others may just be dense clumps of gas heated externally ( e.g. , the orion compact ridge ) .
this point will be further discussed in
4 and 5 .
disks are not included in table 1 , because little is known about their chemical characteristics , or even their existence , around high - mass ysos ( see norris , this volume ) .
intermediate - mass pre - main sequence stars , in particular the so - called herbig ae / be stars , have received increased observational attention in recent years ( see waters & waelkens 1998 for a review ) .
these stars have spectral type a or b and show infrared excesses due to circumstellar dust .
typical luminosities are in the range @xmath22 l@xmath13 , and several objects have been located within 1 kpc distance .
systematic mapping of co and the submillimeter continuum of a sample of objects has been performed by fuente et al .
( 1998 ) and henning et al . (
the data show the dispersion of the envelope with time starting from the deeply embedded phase ( e.g. , lkh@xmath23234 ) to the intermediate stage of a pdr ( e.g. , hd 200775 illuminating the reflection nebula ngc 7023 ) to the more evolved stage where the molecular gas has disappeared completely ( e.g. , hd 52721 ) .
the increasing importance of photodissociation in the chemistry is probed by the increase in the cn / hcn abundance ratio .
this ratio has been shown in other high - mass sources to be an excellent tracer of pdrs ( e.g. , simon et al .
1997 , jansen et al .
line surveys of these objects in selected frequency ranges would be useful to investigate their chemical complexity , especially in the embedded phase .
iso - sws observations of a large sample of herbig ae / be stars have been performed by van den ancker et al .
( 2000b , c ) . in the embedded phase ,
shock indicators such as [ s i ] 25.2 @xmath11 m are strong , whereas in the later phases pdr indicators such as pahs are prominent .
an excellent example of this evolutionary sequence is provided by three herbig ae stars in the bd+40@xmath244124 region ( @xmath25 kpc ) .
the data suggest that in the early phases , the heating of the envelope is dominated by shocks , whereas in later phases it is controlled by ultraviolet photons .
iso - lws data have been obtained for a similar sample of herbig ae / be stars by lorenzetti et al .
( 1999 ) and giannini et al .
( 2000 ) , and are summarized by saraceno et al .
the [ c ii ] 158 @xmath11 m and [ o i ] 63 and 145 @xmath11 m lines are prominent in many objects and are due primarily to the pdr component in the large lws beam ( @xmath26 ) .
co and oh far - infrared lines have been detected in some objects and indicate the presence of a compact , high temperature and density region of @xmath281000 au in size , presumably tracing the inner warm envelope ( see figure 3 ) .
far - infrared lines of h@xmath10o are seen in low - mass yso spectra , but are weak or absent in those of intermediate- and high - mass ysos , with the exception of orion - kl and sgrb2 ( e.g. , harwit et al .
1998 , cernicharo et al .
1997 , wright et al .
. the absence of h@xmath10o lines in higher - mass objects may be partly due to the larger distance of these objects , resulting in substantial dilution in the lws beam .
however , photodissociation of h@xmath10o to oh and o by the enhanced ultraviolet radiation may also play a role . in summary , both the submillimeter and infrared diagnostics reveal an evolutionary sequence from the youngest ` group i ' objects to ` group iii ' objects ( cf . classification by fuente et al .
1998 ) , in which the envelope is gradually dispersed .
such a sequence is analogous to the transition from embedded class 0/i objects to more evolved class ii / iii objects in the case of low - mass stars ( adams et al .
the iso data provide insight into the relative importance of the heating and removal mechanisms of the envelope . at the early stages of intermediate - mass star formation
, shocks due to outflows appear to dominate whereas at later stages radiation is more important .
the availability of complete , high quality iso spectra for a significant sample of massive young stars provides a unique opportunity to study these sources through a combination of infrared and submillimeter spectroscopy , and further develop these diagnostics .
van der tak et al .
( 2000a ) have selected a set of @xmath2810 deeply embedded massive ysos which are bright at mid - infrared wavelengths ( 12 @xmath11 m flux @xmath29 100 jy ) , have luminosities of @xmath30 l@xmath13 and distances @xmath314 kpc .
the sources are all in an early evolutionary state ( comparable to the ` class 0/i ' or ` group i ' stages of low- and intermediate - mass stars ) , as indicated by their weak radio continuum emission and absence of ionic lines and pah features .
in addition to iso spectra , jcmt submillimeter data and ovro interferometer observations have been obtained . for most of the objects
high spectral resolution ground - based infrared data of co , @xmath32co and h@xmath33 are available ( mitchell et al . 1990 , geballe & oka 1996 , mccall et al .
1999 ) , and occasionally h@xmath10 ( lacy et al .
1994 , kulesa et al .
1999 ) . for comparison ,
5 infrared - weak sources with similar luminosities are studied at submillimeter wavelengths only .
this latter set includes hot cores and ultracompact h ii regions such as w 3(h@xmath10o ) , iras 20126 + 4104 ( cesaroni et al .
1997 , 1999 ) , and ngc 6334 irs1 . in order to derive molecular abundances from the observations , a good physical model of the envelope is a prerequisite .
van der tak et al .
( 1999 , 2000a ) outline the techniques used to constrain the temperature and density structure ( figure 4 ) .
the total mass within the beam is derived from submillimeter photometry , whereas the size scale of the envelope is constrained from line and continuum maps .
the dust opacity has been taken from ossenkopf & henning ( 1994 ) and yields values for the mass which are consistent with those derived from c@xmath34o for warm sources where co is not depleted onto grains .
the temperature structure of the dust is calculated taking the observed luminosity of the source , given a power - law density structure ( see below ) . at large distances from the star ,
the temperature follows the optically thin relation @xmath35 , whereas at smaller distances the dust becomes optically thick at infrared wavelengths and the temperature increases more steeply ( see figure 5 ) .
it is assumed that @xmath36 , consistent with explicit calculations of the gas and dust temperatures by , e.g. , doty & neufeld ( 1997 ) for these high densities .
the continuum data are sensitive to temperature and column density , but not to density .
observations of a molecule with a large dipole moment are needed to subsequently constrain the density structure .
one of the best choices is cs and its isotope c@xmath37s , for which lines ranging from @xmath27=21 to 109 have been observed . assuming a power - law density profile @xmath38 , values of @xmath23 can be determined from minimizing @xmath39 between the cs line data and excitation models .
the radiative transfer in the lines is treated through a monte - carlo method .
the best fit to the data on the infrared - bright sources is obtained for @xmath40 , whereas the hot core / compact h ii region sample requires higher values , @xmath41 .
this derivation assumes that the cs abundance is constant through the envelope ; if it increases with higher temperatures , such as may be the case for hot cores , the values of @xmath23 are lowered .
note that the derived values of @xmath42 are lower than those found for deeply embedded low - mass objects , where @xmath43 ( e.g. , motte et al . 1998 , hogerheijde et al .
1999 ) .
figure 5 displays the derived temperature and density structure for the source gl 2591 , together with the sizes of the jcmt and ovro beams .
while the submillimeter data are weighted toward the colder , outer envelope , the infrared absorption line observations sample a pencil - beam line of sight toward the yso and are more sensitive to the inner warm ( @xmath44 k ) region . on these small scales ,
the envelope structure deviates from a radial power law , which decreases the optical depth at near - infrared wavelengths by a factor of @xmath45 ( van der tak et al.1999 ) . for sources for which interferometer data are available ,
unresolved compact continuum emission is detected on scales of a few thousand au or less .
this emission is clearly enhanced compared with that expected from the inner `` tip '' of the power - law envelope , and its spectral index indicates optically thick warm dust , most likely in a dense circumstellar shell or disk .
the presence of this shell or disk is also indicated by the prevalence of blue - shifted outflowing dense gas without a red - shifted counterpart on @xmath46 scales .
the iso - sws spectra of the infrared - bright sources show absorption by various gas - phase molecules , in addition to strong features by ices .
molecules such as co@xmath10 ( van dishoeck et al .
1996 , boonman et al . 1999 , 2000a )
, h@xmath10o ( van dishoeck & helmich 1996 , boonman et al .
2000b ) , ch@xmath47 ( boogert et al . 1998 ) , hcn and c@xmath10h@xmath10 ( lahuis & van dishoeck 2000 ) have been detected ( see also van dishoeck 1998 , dartois et al .
1998 ) . in the infrared , absorption out of all @xmath27-levels is observed in a single spectrum .
the excitation temperatures @xmath48 of the various molecules , calculated assuming lte , range from @xmath49 to @xmath44 k between sources , giving direct information on the physical component in which the molecules reside . while co is well - mixed throughout the envelopes , h@xmath10o , hcn and c@xmath10h@xmath10 are enhanced at high temperatures .
in contrast , co@xmath10 seems to avoid the hottest gas .
high spectral resolution ground - based data of hcn and c@xmath10h@xmath10 by carr et al .
( 1995 ) and lacy et al .
( 1989 ) for a few objects suggest line widths of at most a few km s@xmath50 , excluding an origin in outflowing gas .
the abundances of h@xmath10o , hcn and c@xmath10h@xmath10 increase by factors of @xmath51 with increasing @xmath48 ( see figure 7 ) .
the warm h@xmath10o must be limited to a @xmath52 au region , since the pure rotational lines are generally not detected in the @xmath53 iso - lws beam ( wright et al .
1997 ) . for co@xmath10 ,
the abundance variation between sources is less than a factor of 10 , and no clear trend with @xmath54 is found .
for the same sources , the h@xmath10o and co@xmath10 ice abundances show a decrease by an order of magnitude , consistent with evaporation of the ices . however , the gas - phase h@xmath10o and co@xmath10 abundances are factors of @xmath55 lower than expected if all evaporated molecules stayed in the gas phase , indicating that significant chemical processing occurs after evaporation .
more detailed modeling using the source structures derived from submillimeter data is in progress .
the jcmt data of the infrared - bright objects show strong lines , but lack the typical crowded ` hot core ' spectra observed for objects such as w 3(h@xmath10o ) and ngc 6334 irs1 .
complex organics such as ch@xmath17och@xmath17 and ch@xmath17ocho are detected in some sources ( e.g. , gl 2591 , ngc 7538 irs1 ) , but are not as prominent as in the comparison sources .
yet warm gas is clearly present in these objects . is the ` hot core ' still too small to be picked up by the single dish beams , or
are the abundances of these molecules not ( yet ) enhanced ? to investigate this question , van der tak et al .
( 2000b ) consider the analysis of two species , h@xmath10co and ch@xmath17oh .
both species have many lines throughout the submillimeter originating from low- and high - lying energy levels .
given the physical structure determined in 4.2 , abundance _
profiles _ can be constrained .
two extreme , but chemically plausible models are considered : ( i ) a model with a constant abundance throughout the envelope .
this model is motivated by the fact that pure gas - phase reaction schemes do not show large variations in calculated abundances between 20 and 100 k ; ( ii ) a model in which the abundance ` jumps ' to a higher value at the ice evaporation temperature , @xmath56 k. in this model , the abundances in the outer envelope are set at those observed in cold clouds , so that the only free parameter is the amount of abundance increase .
it is found that the h@xmath10co data can be well fit with a constant abundance of a few @xmath57 throughout the envelope . however , the high @xmath58 data for ch@xmath17oh require a jump in its abundance from @xmath59 to @xmath60 for the warmer sources .
this is consistent with the derived excitation temperatures : h@xmath10co has a rather narrow range of @xmath48=5090 k , whereas ch@xmath17oh shows @xmath48=30200 k. moreover , the interferometer maps of ch@xmath17oh rule out constant abundance models .
the jump observed for ch@xmath17oh is chemically plausible since this molecule is known to be present in icy grain mantles with abundances of 540% with respect to h@xmath10o ice , i.e. , @xmath61 w.r.t .
similar increases in the abundances of organic molecules ( e.g. , ch@xmath17oh , c@xmath10h@xmath17cn , .... ) are found with increasing @xmath62 for a set of ` hot core ' objects by ikeda & ohishi ( 1999 ) .
both the infrared and submillimeter data show increases in the abundances of various molecules with increasing temperature .
four types of species can be distinguished : ( i ) ` passive ' molecules which are formed in the gas phase , freeze out onto grains during the cold ( pre-)collapse phase and are released during warm - up without chemical modification ( e.g. , co , c@xmath10h@xmath10 ) ; ( ii ) molecules which are formed on the grains during the cold phase by surface reactions and are subsequently released into the warm gas ( e.g. , ch@xmath17oh ) ; ( iii ) molecules which are formed in the warm gas by gas - phase reactions with evaporated molecules ( e.g. , ch@xmath17och@xmath17 ) ; ( iv ) molecules which are formed in the hot gas by high temperature reactions ( e.g. , hcn ) .
these types of molecules are associated with characteristic temperatures of ( a ) @xmath63 k , where co is frozen out ; the presence of co ice is thought to be essential for the formation of ch@xmath17oh ; ( b ) @xmath64 k , where all ices evaporate on a time scale of @xmath65 yr ; and ( c ) @xmath66 k , where gas - phase reactions drive the available atomic oxygen into water through the reactions o + h@xmath10 @xmath18 oh @xmath18 h@xmath10o ( ceccarelli et al . 1996 , charnley 1997 ) .
atomic oxygen is one of the main destroyers of radicals and carbon chains , so that its absence leads to enhanced abundances of species like hcn and hc@xmath17n in hot gas .
water is abundantly formed on the grains , but the fact that the h@xmath10o abundance in the hot gas is not as large as that of the ices suggests that h@xmath10o is broken down to o and oh after evaporation by reactions with h@xmath33 .
h@xmath10o can subsequently be reformed in warm gas at temperatures above @xmath28230 k , but figure 7 indicates that not all available gas - phase oxygen is driven into h@xmath10o , as the models suggest .
the low abundance of co@xmath10 in the warm gas is still a puzzle , since evaporation of abundant co@xmath10 ice is observed .
the molecule must be broken down rapidly in the warm gas , with no reformation through the co + oh @xmath18 co@xmath10 reaction ( see question by minh ) .
the objects studied by van der tak et al .
( 2000a ) are all in an early stage of evolution , when the young stars are still deeply embedded in their collapsing envelope . nevertheless , even within this narrow evolutionary range ,
there is ample evidence for physical and chemical differentiation of the sources .
this is clearly traced by the increase in the gas / solid ratios , the increase in abundances of several molecules , the decrease in the ice abundances , and the increase of the amount of crystalline ice with increasing temperature ( boogert et al .
2000a ) .
the fact that the various indicators involve different characteristic temperatures ranging from @xmath6750 k ( evaporation of apolar ices ) to 1000 k ( @xmath48 of gas - phase molecules ) indicates that the heating is not a local effect , but that ` global warming ' occurs throughout the envelope .
moreover , it can not be a geometrical line - of - sight effect in the mid - infrared data , since the far infrared continuum ( 45/100 @xmath11 m ) and submillimeter line data ( ch@xmath17oh ) show the same trend .
shocks with different filling factors are excluded for the same reason .
can we relate this ` global warming ' of the envelope to an evolutionary effect , or is it determined by other factors ?
the absence of a correlation of the above indicators with luminosity or mass of the source argues against them being the sole controlling factor . the only significant trend is found with the ratio of envelope mass over stellar mass .
the physical interpretation of such a relation would be that with time , the envelope is dispersed by the star , resulting in a higher temperature throughout the envelope .
the results discussed here suggest that the observed chemical abundance and temperature variations can indeed be used to trace the evolution of the sources , and that , as in the case of low- and intermediate - mass stars , the dispersion of the envelope plays a crucial role .
the combination of infrared and submillimeter diagnostics is very important in the analysis .
an important next step would be to use these diagnostics to probe a much wider range of evolutionary stages for high - mass stars , especially in the hot core and ( ultra-)compact h ii region stages , to develop a more complete scenario of high - mass star formation .
the relation between the inner warm envelope and the ` hot core ' is still uncertain : several objects have been observed which clearly have hot gas and evaporated ices ( including ch@xmath17oh ) in their inner regions , but which do not show the typical crowded ` hot core ' submillimeter spectra .
are these objects just on their way to the ` hot core ' chemical phase ? or is the ` hot core ' a separate physical component , e.g. , a dense shell at the edge of the expanding hyper - compact h ii region due to the pressure from the ionized gas , which is still too small to be picked up by the single - dish beams ? in either case , time or evolution plays a role and would constrain the ages of the infrared - bright sources to less than a few @xmath68 yr since evaporation .
interferometer data provide evidence for the presence of a separate physical component in the inner 1000 au , but lack the spatial resolution to distinguish a shell from any remnant disk , for example on kinematic grounds
. an important difference between high- and low - mass objects may be the mechanism for the heating and dispersion of their envelopes . for low - mass ysos , entrainment of material in outflows
is the dominant process ( lada 1999 ) . for intermediate - mass stars ,
outflows are important in the early phase , but ultraviolet radiation becomes dominant in the later stages ( see 3 ) .
the situation for high - mass stars is still unclear .
the systematic increase in gas / solid ratios and gas - phase abundances point to global heating of the gas and dust , consistent with a radiative mechanism .
however , a clear chemical signature of ultraviolet radiation on gas - phase species and ices in the embedded phase has not yet been identified , making it difficult to calibrate its effect . on the other hand ,
high - mass stars are known to have powerful outflows and winds , but a quantitative comparison between their effectiveness in heating an extended part of the envelope and removing material is still lacking .
geometrical effects are more important in less embedded systems , as is the case for low - mass stars , where the circumstellar disk may shield part of the envelope from heating ( boogert et al .
. to what extent does the chemical evolution picture also apply to low - mass stars ?
many of the chemical processes and characteristics listed in table 1 are also known to occur for low - mass ysos , but several important diagnostic tools are still lacking .
in particular , sensitive mid - infrared spectroscopy is urgently needed to trace the evolution of the ices for low - mass ysos and determine gas / solid ratios . also , molecules as complex as ch@xmath17och@xmath17 and c@xmath10h@xmath19cn have not yet been detected toward low - mass ysos , although the limits are not very stringent ( e.g. , van dishoeck et al .
evaporation of ices clearly occurs in low - mass environments as evidenced by enhanced abundances of grain - surface molecules in shocks ( e.g. , bachiller & prez - gutirrez 1997 ) , but whether a similar ` hot core ' chemistry ensues is not yet known . differences in the h / h@xmath10 ratio and temperature structure in the ( pre-)collapse phase may affect the grain - surface chemistry and the ice composition , leading to different abundances of solid ch@xmath17oh , which is an essential ingredient for building complex molecules .
future instrumentation with high spatial resolution ( @xmath69 ) and high sensitivity will be essential to make progress in our understanding of the earliest phase of massive star formation , in particular the sma and alma at submillimeter wavelengths , and sirtf , sofia , first and ultimately ngst at mid- and far - infrared wavelengths .
the authors are grateful to g.a .
blake , a.c.a .
boogert , a.m.s .
boonman , p. ehrenfreund , n.j .
evans , t. giannini , f. lahuis , l.g .
mundy , a. nummelin , w.a .
schutte , a.g.g.m .
tielens , and m.e .
van den ancker for discussions , collaborations and figures .
this work was supported by nwo grant 614.41.003 m. ohishi i agree with the point you mentioned , that the chemical differences among hot cores is due to a difference of evolutionary stage .
now we have several well - known hot cores such as orion kl / s , w 3 irs5/h@xmath10o / irs4 , sgrb2
n / m / nw etc .
can you give us your personal view on the evolutionary differences of these sources ?
. van dishoeck van der tak et al .
( 2000a ) argue that the infrared - bright objects such as w 3 irs5 represent an earlier evolutionary phase than the hot cores , on the basis of an anti - correlation with the radio continuum .
the physical picture is that the ionizing uv radiation and stellar winds push the hottest dust in the inner regions further out , decreasing the temperature of the dust and thus the near - infrared continuum . at the same time
, the size of the region which can be ionized is increased .
the ` erosion ' of the envelopes thus occurs from the inside out . for other sources ,
infrared diagnostics are lacking , so that the situation is less clear .
it would be great if chemistry could help to tie down the time scales of the various phases .
van dishoeck the absence of pah features in the early embedded stage can be due either to a lack of ultraviolet radiation to excite the features or to an absence of the carriers .
manske & henning ( 1999 , a&a 349 , 907 ) have argued for the case of herbig ae / be stars that there should be sufficient radiation to excite pahs in the envelope / disk system , so that the lack is likely due to the absence of the pahs themselves .
perhaps the pahs have accreted into the icy mantles at the high densities in the inner envelope and do not evaporate and/or are chemically transformed into other more refractory species on grains . alternatively , the region producing ultraviolet radiation ( h ii region ) may be very small in these massive objects , and the photons may not reach the pah - rich material or have a very small beam filling factor .
once the envelope breaks up and ultraviolet radiation can escape to the less dense outer envelope , the pah features from those regions will appear in spectra taken with large beams .
van dishoeck the ` xcn ' feature is indeed the best candidate for tracing the ultraviolet processing of ices . if ascribed to ocn@xmath70 , it likely involves hnco as a precursor . in the laboratory
, hnco is produced by photochemical reactions of co and nh@xmath17 , but in the interstellar medium grain surface chemistry is an alternative possibility which does not necessarily involve ultraviolet radiation ( see ehrenfreund & schutte , this volume ) .
van dishoeck charnley & kaufman ( 2000 , apj , 529 , l111 ) argue that the evaporated co@xmath10 is destroyed by reactions with atomic hydrogen at high temperatures in shocks .
this is an interesting suggestion , but needs to be tested against other species such as h@xmath10o and h@xmath10s which can be destroyed by reactions with atomic hydrogen as well .
also , the amount of material in the envelope that can be affected by shocks is not clear .
|
recent observational studies of intermediate- and high - mass star - forming regions at submillimeter and infrared wavelengths are reviewed , and chemical diagnostics of the different physical components associated with young stellar objects are summarized .
procedures for determining the temperature , density and abundance profiles in the envelopes are outlined .
a detailed study of a set of infrared - bright massive young stars reveals systematic increases in the gas / solid ratios , the abundances of evaporated molecules , and the fraction of heated ices with increasing temperature .
since these diverse phenomena involve a range of temperatures from @xmath0k to 1000 k , the enhanced temperatures must be communicated to both the inner and outer parts of the envelopes .
this ` global heating ' plausibly results from the gradual dispersion of the envelopes with time .
similarities and differences with low - mass ysos are discussed .
the availability of accurate physical models will allow chemical models of ice evaporation followed by ` hot core ' chemistry to be tested in detail .
|
the research in language dynamics has arose an increasing interest of the complex systems community in the last years .
most of the researchers focus their investigations on issues like rise , competition , extinction risk and death of languages @xcite .
furthermore , recent advances in archeology , genetics and linguistics have provided relevant contributions to a better comprehension of the linguistic diversification @xcite .
some investigations have demonstrated that distinct causes have greatly affected the evolution of the linguistic diversity . among the main elements are geographic factors , economic features , complexity of the language , to cite just a few .
for instance , sutherland @xcite has shown that beside country area , forest area and maximum altitude contribute to increase diversity , whereas the diversity decreases for a larger latitude . according to bellwood @xcite and renfrew @xcite
the occurrence of agricultural expansion was the responsible for the massive population replacements initiated about 10,000 years ago and caused the disappearance of many of the old world languages .
in a recent work , we investigated the evolution of the linguistic diversity by introducing a spatial computer simulation model that considers a diffusive process which is able to generate and sustain the diversity @xcite .
the model describes the occupation of a given area by populations speaking several languages . to each language
was assigned a fitness value @xmath3 which is proportional to the number of sites colonized by populations that speak that language . in the process of colonization , language mutation or differentiation and
language substitution can take place , which affords the linguistic diversity .
this simple model gives rise to scaling laws in close resemblance with those reported in @xcite . in the current contribution
, we study the dynamics of the linguistic diversity but now we assume that the fitness of each language is bounded by a given maximum ( saturation ) value which is randomly chosen from an uniform distribution .
the saturation hypothesis mimics factors like the difficulty / ease of learning the languages and economy that permit some languages to propagate more easily than others .
the paper is organized as follows . in section 2
we introduce the model . in section 3
we discuss the results . and finally , in section 4 we present the conclusions .
our model is defined on a two - dimensional lattice of linear size @xmath4 , and composed of @xmath5 sites with periodic boundary conditions . each lattice site
@xmath6 represents a given region , which can be occupied by a single population speaking just one language .
we ascribe to each site a given capability @xmath7 , whose value we estimate from a uniform distribution , defined in the interval 0 - 1 .
the capability means the amount of resources available to the population which will colonize that place .
it is implicit that the population size in each cell @xmath6 is proportional to its capability @xmath7 . in the first step of the dynamics
, we randomly choose one site of the lattice to be colonized by a single population that speaks the ancestor language .
each language is labeled by an integer number .
as soon as a new language arises , it is labeled by the next upper integer . to each language ,
we assign a fitness value @xmath3 , which is calculated as the sum of the capabilities of the sites which speak that specific language .
but now differently from reference @xcite , the fitness can not exceed an integer value @xmath8 which we have chosen to be in the range 1 - 2000 .
this saturation term @xmath8 is randomly chosen when the language @xmath9 appears .
thus , the initial fitness of the ancestor language is the capability of the initial site . in the second step ,
one of the four nearest neighbors of the site containing the ancestor language will be chosen to be colonized with probability proportional to its capability .
we assume that regions containing larger amount of resources are most likely to be colonized faster than poor regions . the referred site is then occupied by a population speaking the ancestor language or a mutant version of it .
mutations are the mechanisms responsible for generating diversity , and together with the natural selection mantains the standing level of diversity on the system .
the probability of occurrence of a mutation in the process of propagation is @xmath10 , where @xmath11 is a constant , and so the mutation probability is inversely proportional to the fitness of the language .
the form of the mutation probability @xmath12 is inspired by population genetics , where the most adapted organisms are less likely to mutate than poorly adapted organisms @xcite .
the probability of producing reverse mutations is zero , that is , the language generated by a mutation is always different of the previous ones .
are randomly chosen in the interval 1 - 2000 .
the lattice size is @xmath13 and @xmath14 see text for detail.,width=415,height=415 ] in the subsequent steps , we check the empty sites which are located on the boundary of the colonized cluster , and we then choose one of those empty sites according to their capabilities . again , those sites with higher capabilities enjoy of a greater likelihood to be occupied .
after that , we choose the language to be incorporated in the chosen cell among those languages occupying the neighboring sites . languages with higher fitness have higher chance to expand .
the process continues while there are empty sites in the network .
after completion , we count the total number of languages @xmath15 . in order to give to the reader some insight about our model , in figure 1
we present the snapshot for a typical realization of the dynamics at the first moment of colonization of all sites ( in this figure the gray scale represents different languages ) .
the striated linguistic domains presenting very small territories occupied by different languages shown in figure 1 remind us the actual distribution of languages observed in the caucasus region between black and caspian seas , a relatively small area of 300,000 km@xmath16 where languages of the caucasic , indo - european and altaic families coexist distributed within a large variety of peoples @xcite .
in figure 2 , we show the diversity @xmath15 as a function of the area @xmath17 ( total number of sites in the lattice ) for mutation parameter @xmath18 and saturation values defined in the interval 1 - 2000 .
the points are averages over 100 independent simulations when @xmath19 and over 20 simulations when @xmath13 .
we observe that the curve presents just one scaling region which extends over five decades .
the exponent @xmath20 is in quite satisfactory agreement with the exponent observed for the actual distribution of languages on earth . for sake of completeness
, we also exhibit in figure 2 the observed values ( @xmath21 ) of diversity versus area obtained in reference @xcite for all languages spoken on earth ( the ten data points are associated with the interval from @xmath22 km@xmath16 to @xmath23 km@xmath16 of the actual distribution ) .
we notice in passing that although there is not a perfect scaling relationship between diversity and area along five decades in area , both the simulation and the actual data of @xmath24 curiously seem to be modulated by a similar tendency to oscillate in respect to the main scaling behavior ( the deviations from perfect scaling in the actual data have no connection with the choice of the bins ) .
we have also investigated the situation at which the saturation value is the same for all languages . we have noticed a linear growth of the diversity with area when the maximum @xmath25 is very small .
for large values of @xmath25 we notice the existence of two scaling regions . for very large values of @xmath25
we recover the result obtained for the case where the fitness are not limited @xcite . as a function of the area @xmath17 for @xmath18 .
the exponent is @xmath20 .
the asterisks represent data from the actual distribution of languages on earth .
see text and figure 1 of reference @xcite for detail.,width=453,height=453 ] main plot - number of languages with population greater than @xmath26 , @xmath27 , as a function of @xmath26 .
@xmath28 with @xmath29 for @xmath30 and @xmath31 for @xmath32 .
inset - corresponding differential distribution @xmath33 with lognormal best fit ( continuous line ) .
see text for detail.,width=453,height=453 ] figure 3 displays the number of languages with population size greater than @xmath26 , @xmath27 , as a function of @xmath26 . in order to obtain the curves , we have assumed that the population in a given site is proportional to the capability in the site .
we have considered that the population in a given site is its capability multiplied by a factor 100 . in the plot ,
the values of the parameters are @xmath13 and @xmath18 . in close analogy with the distribution of languages on earth @xcite , we find two distinct scaling regimes @xmath28 : @xmath29 for @xmath30 , and @xmath31 for @xmath32 .
the inset exhibits the differential distribution of languages spoken by a population of size @xmath26 , @xmath33 .
this distribution also agrees with the one observed for languages on earth @xcite ; in particular it is well described by the lognormal function @xmath34 $ ] , with @xmath35 and @xmath36 ( continuous curve in the inset ) .
we have introduced a model for evolution of linguistic diversity that considers a bounded fitness value for languages .
we have considered a random chosen value of saturation of the fitness for each language in order to mimic the fact that different languages have different conditions to propagate .
we have noticed a considerable improvement of the results when compared to the earlier approach @xcite .
now , the relationship between diversity and area presents just one scaling regime .
for @xmath18 we obtain @xmath37 , which is in very good agreement with the exponent observed for the languages on the earth @xcite , along five decades of variability in area .
we have also observed that the exponents @xmath38 for the two power law regimes in @xmath27 as a function of @xmath26 are closer to those obtained by empirical observations @xcite . in order to compare other kinds of saturation conditions ,
we have also studied the case where the saturation values are the same for all the languages . with this condition
, we could not reproduce the basic relationship between diversity and area observed for the actual distribution of languages , although for the very particular and unrealistic case where @xmath39 and @xmath40 , we can perfectly reproduce the differential distribution of languages spoken by a population of size @xmath26 , @xmath33 , as well as the number of languages with population size greater than @xmath26 , @xmath27 , as a function of @xmath26 .
our results seem to demonstrate that different assumptions on the behavior of the fitness function have very important consequences on the characteristics of the language spreading .
v. m. de oliveira and m. a. f. gomes are supported by conselho nacional de desenvolvimento cientfico e tecnolgico and programa de ncleos de excelncia ( brazilian agencies ) .
p. r. a. campos is supported by cnpq .
|
a simple spatial computer simulation model was recently introduced to study the evolution of the linguistic diversity @xcite .
the model considers processes of selective geographic colonization , linguistic anomalous diffusion and mutation . in the approach , we ascribe to each language a fitness function which depends on the number of people that speak that language . here
we extend the aforementioned model to examine the role of saturation of the fitness on the language dynamics .
we found that the dependence of the linguistic diversity on the area after colonization displays a power law regime with a nontrivial exponent in very good agreement with the measured exponent associated with the actual distribution of languages on the earth .
@xmath0departamento de fsica , universidade federal de pernambuco , 50670 - 901 , recife , pe , brazil + @xmath1departamento de fsica e matemtica , universidade federal rural de pernambuco 52171 - 900 , dois irmos , recife - pe , brazil + @xmath2centro de informtica , universidade federal de pernambuco , 50670 - 901 , recife , pe , brazil
|
it is now widely accepted that the first order fermi acceleration , which is driven in various astrophysical shocks , provides a factory of non - thermal energetic particles .
for example , recent observations discovered synchrotron x - rays from energetic electrons with energies above @xmath0ev in the shell of the supernova remnant 1006 @xcite . since the basic theory was proposed more than 20 years ago by 78 , krymski ( 1977 ) , axford et al .
( 1977 ) , and blandford and ostriker ( 1978 ) , many papers have dealt with fundamental mechanisms of shock acceleration .
the mechanism has been applied to many situations ranging from interplanetary shocks to relativistic shocks in active galactic nuclei or gamma - ray bursts ( see for review blandford & eichler 1987 and kirk & duffy 1999 ) .
although most of the recent theoretical interests have been concentrated on non - linear problems where the shock structure is modified by accelerated particles ( ellison , baring & jones 1996 , berezhko & ellison 1999 ) , some of the linear problems in the test particle approximation still remain to be clarified and need more close examinations , especially for relativistic shocks , oblique shocks and injection of seed particles .
the first order fermi acceleration in shock waves with test particle approximation is the most basic problem of the shock acceleration . as a particle gains energy whenever the particle repeats shock crossing and recrossing between the upstream and downstream of the shock front
, it is essential to calculate the average energy gain per this cycle and the number distribution of the repeated cycles before the particles escape to far downstream .
these are determined by the shock speed , the compression ratio of the shock , the pitch angle distribution of particles at the shock crossing and the return probability of particles , i.e. , the probability of recrossing the shock front for particles that cross the front from the upstream to the downstream .
as is well known , the energy spectrum of particles accelerated by this mechanism takes a power law form , @xmath1 . for the non - relativistic shocks ,
the conventional diffusion approximation can be used to obtain the universal power law spectrum with an index of @xmath2 , where @xmath3 is the compression ratio of the shock ( bell 1978a ) . however , for relativistic shocks , oblique shocks and acceleration of low energy particles ,
in which the fluid speed is not negligible compared with the particle velocity , anisotropy of particle distribution at the shock front becomes large and the diffusion approximation is not relied on .
thus , one should develop more sophisticated treatments beyond the diffusion approximation even in the linear regime .
there are three ways to do this .
one is to solve the transport equation with appropriate collision operators , as was developed by kirk & schneider ( 1987 , 1988 ) for relativistic shocks , and by malkov & vlk for the injection problem .
they solved for eigenvalues and eigenfunctions of the transport equation both for upstream and downstream and by using the matching condition at the shock front they obtained the spectral index @xmath4 and the distribution function .
this method works well when the pitch angle diffusion is a main process of the scattering , while it seems to have met some difficulties when the simpler case of pure large angle scattering is considered .
another approach is the single - particle approach , which was originally used by 78 for non - relativistic shocks and was extended by 81 for relativistic shocks .
although this method is in principle equivalent to the former method , it is more intuitive and more tractable as will be described in this paper . to determine the pitch angle distribution at the shock front ,
78 and 81 basically utilized the distribution functions obtained by macroscopic methods such as the diffusion equation .
81 also used an upstream distribution function , which is realized far upstream when particles suffer large angle scattering ; these functions may be different from real ones at the shock front as was shown in kirk & schneider ( 1987 , 1988 ) .
the third one is the monte carlo simulation in which the motion of each particle is traced faithfully ( ellison , jones & reynolds 1990 ; ostrowski 1991 ; bednarz & ostrowski 1998 ) .
while monte carlo simulation can make a direct estimation of the distribution function and the spectral index , a large scale simulation is needed to obtain sufficiently accurate results .
physical interpretations of simulation results in terms of analytical models are desirable .
the results of acceleration in such situations in fact depend on the nature of scattering mechanisms of particles by magnetic irregularity existing in the background plasma .
this is because the anisotropy of particle distribution at the shock front is strongly subject to the detail of the scattering process .
especially , two conventional scattering models , large angle scattering and pitch angle diffusion , lead to significantly different spectral index in relativistic or highly oblique shocks ( kirk & schneider 1988 ; ellison , jones & reynolds 1990 ; naito & takahara 1995 ; ellison , baring & jones 1996 ) .
it also may be the case for acceleration of low energy particles .
the pitch angle diffusion is derived by the quasi - linear theory of plasma turbulence ( e.g. , melrose 1980 ) but this description is adequate only for weak turbulence while it is often supposed that the turbulence can be strong in astrophysical shock environments ( see ellison , jones & reynolds 1990 and references therein ) . on the other hand , the large angle scattering , which isotropizes the particle direction in a single scattering independently of the direction before scattering ,
may mimic some effect of strong turbulence , though there is no theoretical justification for this .
theory of particle transport in strongly turbulent field required for realistic investigation of this problem is very difficult and not yet established , to our knowledge . in this paper
, we adopt the large angle scattering model regarding that it mimics the scattering under strongly turbulent field , and investigate the nature of the shock acceleration in highly anisotropic situation , basically taking the second approach mentioned above .
the specific large angle scattering model used in the paper is prescribed in section2 , and assumes that the particle direction after scattering does not depend on the direction before scattering ( although this feature is not always true for general large angle scattering , we use the term ` large angle scattering ' for the specific model in this paper ) .
this mathematical simplicity allows us to treat the problem of particle transport as a random walk .
thus , our approach is based on the theory of random walk in a moving medium not on the diffusion theory to examine the pitch angle distribution at the shock front and the return probability of the particles from the downstream of the shock .
therefore , it can be applied to any value of the shock velocity .
the paper is organized as follows . in
, we show that the motion of a particle suffering from the large angle scattering can be treated as a random walk in a moving medium , and formulate the random walk using the probability theory . in
, we derive analytical approximate solutions , and devote to a specific case of isotropic scattering in .
finally , we apply these results to the shock acceleration in .
in this paper , we treat the first order fermi acceleration in parallel shocks for any value of the shock speed .
we assume the test particle approximation and adopt large angle scattering as the scattering process of particles . in order to know how particles are accelerated ,
the properties of particle trajectories are to be investigated in detail .
when particles move around on each side of the shock front , their motion can be regarded as one - dimensional because , in plane - parallel shocks , only the position along the field line is relevant .
if the scattering process is assumed to be described by large angle scattering , the motion of particles on each side of the shock front can be treated as a random walk in a moving medium as is described in this section .
thus , the problem will be reduced to the random walk of particles in a moving medium with an absorbing boundary ; the shock front becomes such a boundary both for upstream and for downstream . in this section ,
first , we formulate a random walk treatment of the motion of particles following the large angle scattering and examine the properties of the random walk in a moving medium .
although the method described in this section may also be used for general problems of particle motion in a moving medium if this scattering model is suitable , we confine our attention to the fermi acceleration in shock waves in this paper . in the following argument
, we assume that the velocity of scattering centres is uniform on each side of the shock front and is be equal to the velocity of fluid ( therefore , we call the velocity of the scattering centres the fluid velocity in the following ) .
it is also assumed that energy loss mechanisms of particles ( e.g. synchrotron loss ) is negligible , and that the scattering of particles is elastic in the fluid frame , i.e. , the particle energy does not change upon scattering .
hereafter , we take the unit @xmath5 ( @xmath6 is the speed of light ) and take the spatial coordinate @xmath7 along the magnetic field line or the shock normal toward downstream direction so that the fluid velocity , @xmath8 , is positive on each side of the shock front .
we use subscript ( + ) and ( - ) to denote the downstream direction ( @xmath9 ) and upstream direction ( @xmath10 ) , respectively .
physical quantities measured in the fluid frame are expressed with a prime .
we prescribe the large angle scattering model in the fluid frame by the following three conditions . 1 .
the energy of particle measured in the fluid frame is conserved at scattering .
thus , while particles stay in either the upstream or the downstream region , particle speed @xmath11 does not change in the fluid frame . 2 .
the probability density of displacement of the particle along the magnetic field line between successive scatterings is assumed to obey an exponential distribution with mean free path @xmath12 .
( here , the term ` the mean free path ' means the displacement along a magnetic field line and _ not _ a length on the orbit of its gyro motion . )
then , the probability density of the displacement @xmath13 is written as @xmath14 where the signs of @xmath13 and @xmath15 must be the same .
the pitch angle cosine of particle @xmath15 after scattering is determined according to the probability density function @xmath16 independently of the pitch angle before scattering .
below , for simplicity , we represent this function as @xmath17 .
clearly , it must satisfy @xmath18 .
the functional form of @xmath19 and @xmath16 still is free and various scattering processes can be adopted .
an often used case is that the mean free time is independent of @xmath15 and the scattering is isotropic , which we consider in .
because the problems of particle transport with a boundary are to be considered in the following , it is better to describe the position of the particle in the reference frame in which the coordinate of the boundary position takes a fixed value , and scattering centres are flowing with fluid speed @xmath8 .
we call this reference frame the boundary rest frame .
( for the shock acceleration , this frame is of course the shock rest frame for each side of the shock front .
but we consider general case here . ) therefore , below , we describe the speed @xmath11 and pitch angle cosine @xmath15 of particles in the fluid frame and describe the position of particle @xmath7 and the time @xmath20 in the boundary rest frame .
( since we consider only steady states below , we need not consider about @xmath20 in the following in fact . )
this treatment is widely used for various transport equations such as diffusion convection equation or radiation hydrodynamics ( kirk , schlickeiser & schneider 1988 ) .
thus , we describe the scattering law in terms of the displacement measured in the boundary rest frame , @xmath21 , which is given through a lorentz transformation of @xmath13 as @xmath22 where @xmath23 is defined as @xmath24 and @xmath25 .
here , @xmath23 represents the degree of advection effects to the particle motion in the boundary rest frame and is a fundamental parameter in the random walks in a moving medium described below . clearly ,
if @xmath26 , a particle can not advance against the flow of scattering centres , which is not of interest for the shock acceleration .
hence , we consider only the case of @xmath27 below .
it can be shown easily that when a particle moves towards ( + ) direction @xmath28 in the boundary rest frame , the range of the pitch angle cosine in the fluid frame is @xmath29 , and for ( - ) direction @xmath30 , the range becomes @xmath31 .
then , the probability distribution of @xmath21 becomes @xmath32 where @xmath33 here , @xmath34 denotes the mean free path of the particle measured in the boundary rest frame and can depend on @xmath15 even if @xmath11 is the same . for later argument
, we define the following two scale lengths , @xmath35 where @xmath36 and @xmath37 denote the scale of scattering length in the boundary rest frame for ( + ) direction and for ( - ) direction , respectively .
it should be noted that , even if both @xmath38 and @xmath17 depend on @xmath11 , when the dependences are separable into @xmath11 and @xmath15 , the properties of particle motion is determined only by @xmath23 except that the temporal and spatial scales can be different .
as far as only the probability of return and the pitch angle distribution at the shock front are to be considered , these scales are not relevant at all .
then , one can choose the scale of the time and space in order for the computation to become the simplest . in the large angle scattering model described in the previous subsection , since the pitch angle of a particle after scattering
is determined independently of that of before scattering and @xmath11 is conserved , the displacement of the particle between successive scatterings is determined independently and equally on each scattering . because of this property , the particle motion can be treated as a well - defined random walk . in this subsection
, we formulate such a random walk treatment .
our arguments as to the random walk in this subsection are based on the chapter 2 of cox & miller ( 1965 ) . in the random
walk considered here , the displacement of the particle @xmath21 takes a continuous value rather than a discrete value .
for such a random walk , one of the fundamental concepts is the probability density function ( p.d.f . ) of the displacement for each step of random walk , @xmath39 . for the large angle scattering model , we can write this function in terms of @xmath40 and @xmath17 , which introduced in the previous subsection . using , we obtain @xmath41 in this description , the renewal of @xmath15 at the scattering and the translation to the following scattering point are treated as _ one step _ of the random walk .
although this p.d.f .
have no information about @xmath15 , if we combine the information of the random walk with the scattering law , the information about @xmath15 can be taken out as described later .
corresponding to the p.d.f .
@xmath39 , the moment generating function ( m.g.f . )
is defined as @xmath42 which is the two - sided laplace transform of the p.d.f .
@xmath39 . clearly , @xmath43 . in general , at some values of @xmath44 this integral diverges .
for example , for the large angle scattering model , the value of @xmath45 and @xmath46 are such divergence points . however , in the following arguments , we need to consider only the value of @xmath44 in the range , @xmath47 and so we need not worry about such a divergence
. we can derive the following formulas about the derivatives of @xmath48 and the moments of @xmath39 from m.g.f .
as @xmath49 @xmath50 in the range of @xmath44 mentioned above , @xmath51 always holds and then @xmath48 is a downward - convex function in this region .
therefore , the value of @xmath52 at which @xmath53 always exists .
this quantity @xmath54 , which is determined only by the functional form of @xmath39 , includes the most important characteristics of the random walk in a moving medium .
in fact , this @xmath54 together with @xmath23 play an important role in the following description .
clearly , the sign of @xmath54 is equal to the sign of @xmath23 and @xmath55 , i.e. , it is positive throughout this paper .
the problem of random walk in a moving medium with boundaries can be formulated as the problem of _ the random walk with absorbing barriers _ in the probability theory .
the procedure is as follows .
first , consider that a particle begins a random walk from the origin @xmath56 under the situation in which there are two absorbing barriers at the position @xmath57 and @xmath58 @xmath59 .
then the particle moves according to the p.d.f .
@xmath60 at each step .
if the particle advances beyond one of the barriers , it is absorbed by the barrier and the random walk is terminated then . to make the argument clear
, we introduce the probability density function of scattering points at the @xmath61-th step , @xmath62 .
the integration of @xmath62 outside of the boundaries ( @xmath63 or @xmath64 ) denotes the probability that the particle is absorbed at the @xmath61-th step .
the integral of @xmath62 over the inside region is _ not _ normalized to unity but denotes the probability that the particle is not absorbed until or at the @xmath61-th step and can carry out the @xmath65 step .
the function @xmath62 can be written as a convolution of @xmath66 by @xmath60 over the non - absorption region , @xmath67 because the particle starts from @xmath56 , it can be seen easily @xmath68 and @xmath69 .
all @xmath62 can be calculated in principle using iteratively .
next , we introduce the density of scattering points summed over all steps , @xmath70 which constitutes one of the key concepts in the following description . if one is not discussing the number of steps before absorption , the only relevant quantity is the density of scattering points @xmath71 .
we limit our considerations to those using @xmath71 in this paper , although the distribution of the number of steps before absorption is important to discuss the dispersion of acceleration time of particles .
it is noted that @xmath71 is not the same as the physical number density of particles and their mutual relation is explained in the appendix a. from and , we can derive a fredholm integral equation of the second kind for @xmath71 : @xmath72 especially , for no - boundary case ( @xmath73 ) , the solution @xmath74 obeys @xmath75 it is noted that @xmath71 satisfies certain integral conditions . to simplify the following description ,
we introduce the probability density of absorption positions , @xmath76 here @xmath77 instead of @xmath7 is used for the spatial coordinate in the absorbing regions .
@xmath78 satisfies the following two conditions .
since the particle is eventually absorbed by either of the absorbing barriers , @xmath79 in addition , in the problem of the random walk with absorbing barriers , a theorem which is called the wald s identity holds .
one of the simplest versions of this theorem is the following identity @xmath80 where @xmath54 was defined in the previous subsection . in the shock acceleration ,
the barrier exists only at one side , i.e. , @xmath81 for downstream region and @xmath82 for upstream region . in the following section , we use one of these conditions to normalize the approximate solutions . for @xmath83 , we use , i.e. , @xmath84 on the other hand , for @xmath85 , we use , i.e. , @xmath86 in order to know the properties of random walk after large numbers of steps , it is helpful to consider the asymptotic form of @xmath74 for @xmath87 or @xmath88 . using the convolution theorem to ,
the two - sided laplace transform of @xmath62 for the no - boundary case turns out to be @xmath89 therefore , from , the two - sided laplace transform of the scattering point density for no boundary case @xmath74 within the range @xmath90 becomes @xmath91 thus , @xmath74 is calculated by the bromwich integral as @xmath92 where @xmath4 is real and @xmath93 . for @xmath87 , the integral path is taken on the @xmath94 side and the residue at @xmath95 is dominant in the integral .
the residue is calculated easily as @xmath96 then , we obtain the asymptotic form of @xmath74 for @xmath87 as @xmath97 the asymptotic form for @xmath88 is calculated in a similar way .
the integral path is taken on the @xmath98 side and the residue at @xmath99 is dominant in the integral .
the residue is calculated as @xmath100 then , we obtain @xmath101 for later convenience , we define the ratio of these two constants as @xmath102 if @xmath103 holds , @xmath104 is almost unity ( see ) . however , it is to be noted that @xmath104 is not equal to unity , i.e. , @xmath105 in general .
equation indicates that the scale length of density of scattering points far upstream is equal to the inverse of @xmath54 .
we denotes this length as @xmath106 while @xmath36 and @xmath37 are the scale lengths of single scattering , @xmath107 is the scale length of multiple scatterings .
if @xmath23 is small enough and the scattering is isotropic , this scale length @xmath107 should coincide with the diffusion length derived by the conventional diffusion equation , @xmath108 , where @xmath109 is the diffusion coefficient ( @xmath110 ) , which is explicitly shown in .
in contrast , when @xmath23 is large , we can not safely use the conventional diffusion equation and the length @xmath111 .
however , even in such a case , we can regard @xmath107 as a diffusion length , since the distribution is an outcome of multiple scatterings in the random walk . here and hereafter
, we call this @xmath107 simply ` the diffusion length ' and call @xmath111 ` the conventional diffusion length ' .
we also show in that this diffusion length @xmath107 indeed coincides with the scale length of particle distribution derived by peacock ( 1981 ) and kirk & schneider ( 1988 ) in the far upstream region for relativistic shocks . for later convenience
, we also define the ratio of scale lengths of single scattering to the diffusion length as @xmath112 as already mentioned , although the density of scattering points does not give direct information about pitch angle cosine of the particle @xmath15 , we can derive the probability density of @xmath15 at absorption with an aid of the scattering law .
the probability density of @xmath15 at the absorption by the boundary @xmath57 is given by @xmath113 and that at the boundary @xmath58 is given by @xmath114 the integral conditions for @xmath71 can be rewritten as the conditions for @xmath115 or @xmath116 . for @xmath85
, the condition can be rewritten in terms of @xmath115 as @xmath117 similarly , for @xmath83 , the wald s identity can be rewritten as @xmath118 here , we define the total absorption probability by the boundary at @xmath119 @xmath120 this function also means the probability that the particle , which is injected in downstream side at a distance of @xmath121 from the boundary , reaches the boundary against the flow of scattering centres .
if @xmath122 holds , also means @xmath123 for _ any _ models .
since , for isotropic scattering and for @xmath124 , @xmath125 approaches the conventional diffusion length @xmath111 as already mentioned , agrees with the result of the conventional diffusion approximation obtained by drury ( 1983 ) . in order to apply the random walk theory described above to the shock acceleration , it is useful to consider the probability density of the pitch angle cosine @xmath15 when the particle which crossed the shock front with @xmath126 returns to the front crossing the reverse direction . for downstream , random walk with a finite @xmath121 and @xmath81
is applied while for upstream that with a finite @xmath127 and @xmath82 is applied . as will be shown in later sections , all the quantities associated with the shock acceleration
can be represented with this probability density .
hereafter , we denote the spatial coordinate by @xmath128 with the shock front at @xmath129 in order to retain @xmath7 coordinate for the random walk problem treated so far where @xmath56 corresponds to the injection point of the particle .
first , let us consider the return probability for the upstream .
the particle first crosses the boundary toward ( - ) direction , i.e. , from the downstream to the upstream with a pitch angle cosine @xmath130 . in order to calculate the probability density of the pitch angle cosine at the return @xmath131 , @xmath132 , it is useful to introduce the density of scattering points before return , @xmath133 , which is defined on @xmath134 .
this density can be calculated in the following way .
the probability density of the first scattering point for the particle which crossed the boundary with pitch angle cosine @xmath135 is determined by the scattering law .
particles which have the first scattering point at @xmath136 make the scattering point density @xmath137 , where @xmath138 denotes the scattering point density of particles injected at @xmath56 in the random walk with absorbing barriers at @xmath57 and @xmath58 .
therefore , we obtain @xmath139 where @xmath140 and @xmath141 . using this ,
@xmath132 is calculated as @xmath142 where @xmath143 . for this case , the total return probability is always unity ( ) , i.e. , @xmath144 we can rewrite this as an integral condition for @xmath133 , @xmath145 for the downstream , the probability density of pitch angle cosine @xmath146 at return for the particle which initially crossed the shock front with @xmath147 ( ( + ) direction ) , @xmath148 , is calculated in a similar way . the scattering point density @xmath149
is written as @xmath150 which is defined on @xmath151 .
using this density , @xmath152 is calculated as @xmath153 the total return probability @xmath154 is not unity and is given as @xmath155 the integral condition becomes a condition for @xmath156 as @xmath157 as is shown in . in the following sections , we derive approximate analytic solutions of @xmath158 and @xmath159 where we use the normalization condition for the scattering point density @xmath160 and @xmath161 .
in this section , we derive approximate analytic solutions for the probability densities by obtaining approximate solutions for the integral equation .
we start from the asymptotic solution of and for the no - boundary problem .
first we approximate the no - boundary solution @xmath74 as @xmath162 note that this approximation neglects a peak around @xmath56 arising from the injection ( see in for monte carlo results ) . to obtain approximate solutions for problems with finite boundaries from those for no - boundary problems , we transform the original integral equation to an alternative integral equation @xmath163 where we define the kernel @xmath164 ( see for its derivation ) .
this integral equation is , of course , equivalent to .
first , we consider approximate solution for the upstream @xmath165 taking @xmath85 and a finite value of @xmath127 .
starting from the approximation , the kernel of the integral equation for @xmath71 becomes separable into @xmath7 , @xmath166 as @xmath167 where @xmath168 substituting this into , we obtain @xmath169 where @xmath170 and @xmath171 the term proportional to @xmath172 represents the effects of the existence of the absorbing boundary at @xmath57 .
although we do not need to explicitly evaluate @xmath173 in our procedures described below , @xmath173 can be solved if we substitute into the definition of @xmath173 as @xmath174 where @xmath175 if we use the approximation again and remembering @xmath176 , the becomes @xmath177 where @xmath178 and @xmath179 in , the second term with the scale length @xmath107 represents the distribution formed by the diffusion process after a large number of scattering , similar to the asymptotic form of @xmath74 for far upstream .
the first term represents the correction to the diffusive term by the following three non - diffusive effects ; the density of scattering points contributed by particles which take only a few steps , the effect of finite initial mean free path @xmath180 ( finite distance from the boundary to the first scattering point ) , and the absorbing effect near the boundary .
when we take the limit @xmath181 , this term disappears and is just reduced to the diffusive solution . depending on @xmath180 ,
@xmath182 can take a positive or negative value . for small @xmath180
the correction term is positive and it mainly reflect the scattering points of the few step particles , while for large @xmath180 it is negative accounting for the effects of finite mean free path and absorption at the boundary .
these features are mentioned again in for a specific scattering model . substituting this to
, we obtain an approximate solution for @xmath183 @xmath184 however , this approximate solution does not satisfy the condition in general because we do not use the exact solution of @xmath71 .
although there may be various ways of remedying this problem , here we adopt a simple and practical way to renormalize the amplitude of @xmath185 so as to satisfy for a given @xmath180 .
thus , instead of @xmath185 we use the coefficient @xmath186 defined by @xmath187 where we define @xmath188 this method also has a merit in avoiding to calculate a complicated integral in the original definition of @xmath185 .
next , we consider approximate solution for the downstream @xmath189 in a similar way by taking @xmath83 and a finite value of @xmath121 . starting from the approximation ,
the kernel becomes @xmath190 where @xmath191 @xmath71 is obtained as @xmath192 where @xmath193 substituting this into and using the approximation again , we obtain @xmath194 where @xmath195 @xmath196 similarly to the case of @xmath160 , the second term in denotes the diffusive term , which corresponds to the asymptotic form for far downstream , and the first term represents the correction to it .
@xmath197 takes positive value for small @xmath180 and negative value for large @xmath180 . using this to , we obtain approximate solution of @xmath198 @xmath199 we again renormalize @xmath200 so as to satisfy the integral condition for the same reason as in the upstream .
thus , instead of @xmath200 we use @xmath201 defined by @xmath202 where we define @xmath203 and we use the relation @xmath204 which is derived by together with the definition of @xmath54 and @xmath18 .
the total return probability @xmath205 is calculated by easily , @xmath206 where we define @xmath207 to compare our approximation with more conventional diffusive approximation , for reference we consider also the latter approximation which we call multi - step approximation .
this is done by ignoring the first term in or , i.e. , by setting @xmath208 . in this approximation
, the density profile takes the same form to the results of the diffusive approximation .
but , since we do not take the limit @xmath181 , this approximation includes partly the effect of the finite mean free path .
( although , the effect of the first step return is lost . ) thus , in the multi - step approximation , the probability density functions become @xmath209 @xmath210 and @xmath211 where we define @xmath212 in the multi - step approximation , the pitch angle distribution at return becomes independent of @xmath213 and @xmath214 .
further @xmath183 is independent of @xmath126 and @xmath198 is separable with respect to @xmath126 and @xmath15 .
although in general the multi - step approximation is not satisfactory , there are cases where this makes some sense as shown in .
the absorption probabilities of the particle injected at a distance @xmath121 in the downstream from the shock front , @xmath116 and @xmath215 , are of some theoretical interest , although it is not explicitly used in later considerations . to obtain their approximate expression , we start with together with the approximation . we renormalize the amplitude of @xmath216
so as to satisfy the wald s identity as @xmath217 using @xmath218 , we obtain @xmath219 @xmath220 it is to be noted that @xmath215 has the scale length @xmath107 for large @xmath121 .
the absorption probability of the particle injected in the upstream at the distance of @xmath127 from the shock front and the pitch angle cosine @xmath15 , @xmath115 , can be also calculated in a similar manner .
using with we renormalize @xmath221 so as to satisfy , although we do not present the result here . in principle it is possible to calculate @xmath183 and @xmath198 by using these @xmath222 .
however , since the expressions are too cumbersome and impractical to denote @xmath183 and @xmath198 in an analytical form , we use the and for @xmath183 and @xmath198 respectively in the following argument .
we make further approximation such as the multi - step approximation .
if we set @xmath223 for all @xmath7 , which is the diffusive solution , and if we ignore the first term in , using wald s identity , we obtain @xmath224 @xmath225 in the limit of @xmath124 , in which the conventional diffusion approximation holds , the constant @xmath226 approaches unity and this expression of @xmath215 agrees with .
peacock(1981 ) derived the absorption probability in a similar form ( see his equation ( 24 ) ; he did not give a concrete expression of the constant @xmath226 ) , but did it in the downstream fluid frame under non - relativistic fluid speed condition and the diffusion approximation .
in the following section , we will make comparisons with his expression for the isotropic scattering .
corresponding approximation in the upstream can be made in a similar way .
if we set the diffusive solution @xmath227 , @xmath71 becomes @xmath228 , where @xmath229 is some constant .
ignoring the first term in and determining @xmath229 through , we obtain @xmath230 if we use these expressions for @xmath116 and @xmath115 to calculate @xmath198 and @xmath183 , the results coincide with the results of the multi - step approximation , obtained in the previous subsection as should be since only diffusive return is taken into account .
we call these expressions for @xmath116 , @xmath215 and @xmath115 ` the multi - step approximation ' , too .
in this section , we apply the random walk theory developed so far to a specific model of scattering .
we consider here the model in which the mean free time @xmath231 is independent of @xmath15 ( although the energy dependence may be allowed , it is not relevant here ) , i.e. , @xmath232 , and the scattering is isotropic in the fluid frame , @xmath233 this model has been widely used both in analytic works ( peacock 1981 ) and in the monte carlo simulations ( ellison , jones & reynolds 1990 ) .
to check the validity of our model , we also perform the monte carlo simulation in which the particle position is traced step by step faithfully according to the adopted scattering law . in the simulation ,
the escaping boundary is set at @xmath234 , which is far enough not to influence the return probability of the particle ( see e.g. ) . in this model ,
@xmath34 is given by @xmath235 for the reason mentioned in section 2.1 , we can take the unit of length as @xmath236 in the following argument .
thus , @xmath34 becomes @xmath237 and @xmath238 the p.d.f .
of the random walk is given by @xmath239 where @xmath240 is the exponential integral defined by ( see abramowitz & stegun ) @xmath241 the m.g.f .
is given by @xmath242 and the moments of @xmath60 are calculated as @xmath243 shows the p.d.f .
@xmath60 ( a ) , and the m.g.f .
@xmath48 ( b ) for @xmath244 .
it is seen that the p.d.f .
becomes more asymmetric for larger @xmath23 because of the advection effect .
the equation to determine @xmath54 ( or @xmath245 ) is given by @xmath246 presents the scale lengths of the random walk , @xmath36 , @xmath37 and @xmath107 .
we also show the conventional diffusion length @xmath111 mentioned in section 2.2.3 extrapolating it to high @xmath23 .
since it depends on the lorentz factor of fluid speed @xmath247 in the unit of length used here ( @xmath248 ) , we show the two cases ; the case of a relativistic particle in a relativistic flow ( @xmath249 ; lower dotted curve ) , and the case of a non - relativistic particle in a non - relativistic flow ( @xmath250 ; upper dotted curve , @xmath251 ) . the diffusive scale length for a pitch angle diffusion model @xmath252 ( kirk & schneider 1987a ) is represented , too .
( here , we choose the pitch angle diffusion coefficient in their model so that it has the same spatial diffsion coefficient as our model . ) for @xmath253 , @xmath107 can be approximated as @xmath254 and for @xmath255 , @xmath256 where @xmath257 therefore , if the ratio @xmath23 is small enough ( @xmath258 ) , our diffusion length @xmath107 coincides with the conventional one @xmath111 , as expected . on the other hand ,
if @xmath23 becomes larger , the difference between @xmath107 and @xmath111 becomes larger . as mentioned in section 2.2.3 , for large @xmath23
, @xmath111 becomes meaningless because the conventional diffusion approximation is no longer valid .
however , @xmath107 still has a meaning of ` the diffusion length ' and this is explicitly shown later in .
in fact , this scale length has been derived by 81 and kirk & schneider ( 1988 ) independently as the scale length of the particle distribution on the far upstream region of relativistic parallel shocks for large angle scattering models . in 81 ,
the equation that is equivalent to was derived from the conservation law of particles in the upstream of the shock using ` the relativistic diffusion approximation ' ( which was named by 87 ) , where @xmath259 corresponds to our @xmath54 in his equation ( 16 ) . in kirk & schneider ( 1988 ) ,
such equivalent equation to was also derived as the equation to determine the eigenvalues for large - angle scattering operator . in their equation
( 34 ) , @xmath260 corresponds to @xmath54 and @xmath261 was taken to unity .
the upstream density profile obtained in a previous monte carlo work for oblique shocks ( naito & takahara 1995 ) deviates from the diffusion approximation , which can be interpreted using @xmath107 obtained here if we consider in the de hoffmann - teller frame . also shows that , when @xmath23 becomes larger , the difference between @xmath36 and @xmath37 is larger and @xmath107 approaches to @xmath37 .
behaviour of @xmath252 is similar to that of @xmath107 but takes a smaller value .
it indicates that the diffusive scale length depends on the scattering model explicitly .
the constants @xmath213 and @xmath214 , which determine the density of scattering points in the no - boundary case , becomes @xmath262 shows the ratio @xmath263 as a function of @xmath23 .
for the limit @xmath124 , @xmath104 approaches to unity , which agrees with the results of the conventional diffusion approximation ( drury 1983 ) .
however , as @xmath23 becomes larger , @xmath104 becomes smaller and approaches @xmath264 as @xmath265 when @xmath255 .
shows an example of the no - boundary solution @xmath74 ( solid histogram ) for @xmath266 , which is obtained by the monte carlo simulation , together with the asymptotic approximation @xmath267 (; dashed curve ) .
it confirms that @xmath267 approaches @xmath74 asymptotically for @xmath88 or @xmath87 .
although @xmath74 has a peak around @xmath56 , our approximation ignores this peak as described in .
shows an example of the results when the single boundary is located at @xmath268 .
the density of scattering points @xmath71 ( solid histogram ) for @xmath266 is obtained by a monte carlo simulation , along with our approximate solutions and that of the conventional diffusion equation ( drury 1983 ; his equation ( 3.41 ) ) @xmath269 ( dotted curve ) .
it is clearly seen that the conventional diffusion equation does not give a sufficiently good approximation to the real solution for high @xmath23 .
especially , the random walk result shows that @xmath71 has a finite value at the absorption barriers , while @xmath269 becomes @xmath264 there . the small difference between the monte carlo result and the our approximation near the absorbing boundary is mainly caused by omitting the peak in @xmath74 around @xmath56 in making our approximation . here , we present the return probabilities of particles for the isotropic large angle scattering model under the approximation made in the previous section .
this model gives @xmath270 @xmath271 @xmath272 and @xmath273 where we define @xmath274 from these quantities , the probability densities of pitch angle at return @xmath183 and @xmath198 are calculated by and , while those for the multi - step approximation are calculated by and , respectively .
the total return probability @xmath205 is also calculated by or .
shows the probability density of pitch angle at return @xmath275 in ( a ) and @xmath276 in ( b ) for @xmath277 and for several values of the initial pitch angle cosine @xmath126 .
our approximation ( solid curves ) gives a fairly good fit to the monte carlo simulation ( grey histograms ) for all initial pitch angle cosine @xmath126 .
for small @xmath126 , the fit deviates a little because the effects of the peak in @xmath74 around @xmath56 , which is ignored in our approximation , affect the probability density .
the multi - step approximation ( dotted curves ) does not fit well when the initial mean free path @xmath180 is small because effects of the return after only a few steps of scattering becomes important .
but , for large @xmath180 and for @xmath198 , the multi - step approximation can fit to the monte carlo simulation .
the expression of the absorption probability @xmath215 by our approximation is a little cumbersome and is presented in .
the absorption probability by the multi - step approximation is given by together with .
presents @xmath215 as a function of the initial position @xmath121 for @xmath278 .
the result of our approximation ( solid curve ) gives a good fit to the monte carlo results ( filled circles ) ( small difference is in part caused by omitting the peak in @xmath74 ) .
the multi - step approximation ( dashed curve ) deviates from the monte carlo results , but gives a correct scale length ( @xmath107 ) .
this result directly confirms that the diffusion length for such large @xmath23 is given by our diffusion length @xmath107 .
results using the conventional diffusion length @xmath111 instead of @xmath107 in the expression of the multi - step approximation ( and ) for two case as in are also represented ; lower dotted curve for @xmath279 and upper dotted curve for @xmath280 ) .
it should be noted that if we extrapolate the expression ( 24 ) in peacock ( 1981 ) , which was derived by adopting the diffusion approximation for particles in downstream of the shock front , to relativistic fluid speed in fluid frame and then transform it to the expression for the boundary rest frame , it turns out to correspond to the multi - step approximation except that it has the scale length of return @xmath280 ( in our unit ) .
these results do not give a satisfactory fit .
presents the total return probability @xmath281 from the downstream for mildly high @xmath23 ( @xmath282 ) as a function of initial pitch angle cosine @xmath283 , which is measured in the boundary rest frame .
the results of our approximation ( solid curve ) excellently agree with the monte carlo simulation ( filled circles ) .
the dotted curve shows the approximate solution of 81 ( shown in his fig.2 ) .
the result of the multi - step approximation ( dashed curve ) , which uses the correct length @xmath107 , is similar to peacock s one , but slightly better than it .
these two diffusive approximations do not agree with the monte carlo results for small @xmath283 .
the reason is similar to that for @xmath183 and @xmath198 ; in diffusive approximations , effects of return after only a few steps of scattering are not accounted for . in general , for higher @xmath23 or smaller @xmath180 , such effects become important and these diffusive approximations become worse . here , we show the results of the density of the scattering points for the particles which cross the shock front with the pitch angle cosine @xmath126 .
shows @xmath284 in ( a ) and @xmath285 in ( b ) calculated by the monte carlo simulation ( histograms ) for several values of the initial pitch angle @xmath126 along with our approximate solutions . it is seen that our approximation ( solid curves ) coincides well with monte carlo simulation .
but , a discrepancy is recognized for @xmath286 in ( b ) . because the normalization conditions used in our approximation ( the , ) employ only the information of finally absorbed particles and do not care about escaping particles to the far downstream , our approximation fails to give a good fit in a distant region where escaping particles dominate , while it gives a satisfactory fit in a near region where almost all the last scattering points exist .
this also means that our approximation works well for absorbed particles , therefore the return probability densities themselves are represented by our approximations quite well as shown in and .
another important feature of @xmath161 and @xmath160 is that the density of the scattering points near the boundary depends on the initial pitch angle @xmath126 . here
, we explain this feature for @xmath161 ( that for @xmath160 is similar ) .
( b ) shows that @xmath189 increases when @xmath287 for small @xmath180 but that it decreases when @xmath287 for large @xmath180 .
this can be explained as already mentioned in section 3 ; for small @xmath180 the scattering point of few step particles are important , while for large @xmath180 the effect of finite initial mean free path is important . in our approximate expression of , while @xmath200 is always positive , @xmath197 becomes negative for large @xmath180 , as shown in ( b ) .
( this feature also appears for @xmath160 , as shown in ( a ) . ) because of this feature , when we average the return probability over an incoming particle distribution , the first term in , which denotes the non - diffusive effects , will become small enough compared with the diffusive term which is associated with @xmath200 .
then , there appears a possibility to use the multi - step approximation to determine the _ averaged _ pitch angle distribution over all incident particles even if @xmath23 is large .
the same consideration applies for @xmath160 , too . as will be shown in the following section , at least for the isotropic large angle scattering model and for typical incoming @xmath126 distributions ,
_ this is actually the case_. in practice , for this reason , we can use the multi - step approximation to determine the pitch angle distribution at the shock front for _ arbitrary _ shock speed as shown in the following section .
in this section , we apply the random walk theory developed so far to the shock acceleration and calculate the spectral index of the accelerated particles for the isotropic large angle scattering model described in the previous section . in the following ,
we consider the acceleration in parallel shocks where the fluid speed measured in the shock rest frame are uniform on each side of the shock front ; @xmath288 for the upstream and @xmath289 for downstream respectively .
we also consider only relativistic particles ( @xmath290 , i.e. , @xmath291 ) because , in this case , the ratio @xmath23 does not change on each side of the shock front even if the particle gains energy in the process of multiple crossing at the shock front .
in addition , for such particles , the lorentz transformation of @xmath292 to @xmath15 becomes as simple as @xmath293 where @xmath292 is the pitch angle cosine measured in the shock rest frame .
hereafter , we use subscripts u , d or superscripts ( u ) , ( d ) to indicate quantities in the upstream and downstream regions , respectively .
the pitch angle cosine measured in the upstream fluid frame is denoted by @xmath294 and that measured in the downstream fluid frame by @xmath295 . in the following
, we present the results only for two simple jump conditions @xmath296 and @xmath297 ( i.e. , @xmath298 ) .
but , since the results are expressed in terms of @xmath288 and @xmath289 , they are applicable to any jump conditions .
( jump conditions for relativistic shocks are discussed in blandford & mckee , 81 , appl & camenzind and heavens & drury . )
the spectral index of the accelerated particles @xmath4 is given by @xmath299 where @xmath300 is the averaged return probability for the particles crossing the shock front toward downstream and @xmath301 is the logarithm of the energy gain factor ( see peacock 1981 ) .
( here , @xmath302 denotes to average over the pitch angle distribution crossing the shock front . )
@xmath301 is given as @xmath303 where @xmath304 @xmath305 and @xmath306 is the relative velocity of the upstream fluid with respect to the downstream fluid , @xmath307 if the dependences on @xmath15 and @xmath11 in @xmath38 and @xmath17 are separable , these quantities are independent of the particle energy , i.e. , @xmath11 .
introducing the normalized pitch angle distribution of particles crossing the shock front from upstream toward downstream by @xmath308 , and that from downstream toward upstream by @xmath309 , i.e. , @xmath310 we can write relevant quantities as follows . using @xmath311 and @xmath312 ,
@xmath300 is written as @xmath313 the logarithm of the energy gain factor is given by @xmath314 and @xmath315 the factor @xmath316 in means that only particles which return again to the shock front are to be considered . the pitch angle distribution @xmath308 and @xmath309 in a steady state
are determined as follows .
if @xmath308 is known , @xmath309 is determined by @xmath308 , @xmath317 on the other hand , @xmath308 is also determined by @xmath309 , @xmath318 therefore , in a steady state , @xmath309 satisfies an integral equation ( a fredholm equation of second kind ) @xmath319 where the kernel of this integral equation is given by @xmath320 this kernel denotes the probability density of the particle which crosses the shock front from downstream to upstream with @xmath321 crosses the shock front again from downstream to upstream with @xmath295 after one cycle .
if @xmath309 can be solved , @xmath308 is calculated through .
although in the above we have made the integral equation for @xmath309 , one can alternatively make the integral equation for @xmath308 .
it is generally difficult to solve the integral equation analytically even if our approximation are used .
then , we numerically evaluate the kernel for our approximation . on the other hand , in the multi - step approximation , @xmath322 and @xmath323
can be calculated analytically without solving the integral equation . because the dependence of @xmath324 in are separable in @xmath15 and @xmath135 in this approximation
, it becomes @xmath325 because @xmath326 in is independent of @xmath135 , we obtain @xmath327 these results agree with the results obtained by peacock(1981 ) , but @xmath107 was replaced by @xmath111 in . shows the kernel of for @xmath328 and @xmath297 .
the numerical evaluation based on our approximation for @xmath329 ( solid curve ) , @xmath330 ( dashed curve ) and @xmath331 ( dotted curve ) together with that of the multi - step approximation are shown .
it is seen that the kernel is almost independent of the initial pitch angle cosine @xmath283 .
the reason for this coincidence is considered to be due to compensation effects of several non - diffusive effects when averaged over the initial pitch angle as was mentioned in the previous section . although this is partly due to our assumption of the isotropic large angle scattering , it is of a great importance in applications to the shock acceleration . although the multi - step approximation is a poor approximation to the detailed particle transport , it can be a good approximation when we average over the pitch angle distribution .
thus , this result gives a justification for using the multi - step approximation even for relativistic shocks .
presents the pitch angle distribution at the shock front in various frames for various shock velocities .
the results of the multi - step approximation ( dotted curve ) and the monte carlo simulation ( solid histogram ) are shown . as is seen ,
the results of the multi - step approximation agree quite well with the monte carlo results even for highly relativistic shocks .
this is not surprising once we admit that the kernel is well approximated by the multi - step approximation .
thus , we have confirmed the point mentioned above that the multi - step approximation gives a good fit to the steady state pitch angle distribution at the shock for any shock speed . also , it is to be noted that the assumptions used in peacock ( 1981 ) are adequate for arbitrary shock speed if the correct diffusion length @xmath107 is used in the downstream as in the multi - step approximation .
= 7.95 cm in the previous subsection , we have shown that the pitch angle distributions at the shock crossing @xmath322 and @xmath323 can be well approximated by the multi - step approximation . here , using this approximation , we show analytical expressions of the quantities associated with the shock acceleration , @xmath300 , @xmath301 and @xmath4 , for the isotropic large angle scattering .
first , using @xmath332 ( ) and @xmath322 by the multi - step approximation , @xmath300 is calculated as @xmath333 where @xmath334 with @xmath335 @xmath336 @xmath337 @xmath338 @xmath339 is defined in .
then , the logarithm of the energy gain factor is given by @xmath340 and @xmath341 where @xmath342 with dilogarithm @xmath343 defined by ( see abramowitz & stegun 1965 ) @xmath344 using these results , gives the spectral index @xmath4 . in ( a )
are presented the quantities associated with the shock accelerations as functions of shock speed @xmath345 for the compression ratio @xmath346 . in ( b ) are shown the results for different jump condition ( @xmath298 ) . as is seen in these two figures ,
the result of the multi - step approximation excellently agree with the results of the monte carlo simulation even if shock speed becomes highly relativistic .
presents the spectral index @xmath4 as functions of the compression ratio @xmath3 similar to fig . 4 of 90 .
our results ( solid curves ) basically coincide with their approximate expression ( dashed curves ; their equation(29 ) ) .
small discrepancy between these two results may be caused by the difference of the method to determine the spectral index @xmath4 ; we calculate @xmath4 using , while @xmath4 was determined by fitting the calculated particle spectrum directly in 90 .
although we have considered the particle acceleration in parallel shocks with the large angle scattering model , it is of some interest to discuss whether our formulation can be extended to other scattering regime .
recently , the acceleration in ultra - relativistic shocks has attracted some attention in relation to the ultra - high - energy cosmic rays ( uhecr ) . however , in such highly relativistic shocks , the large angle scattering model adopted here can not be used safely in a physical sense ; because the residence time of particles in upstream is quite short ( it is shorter than the gyro - period of particles ) , its point - like feature may lead to unsuitable results ( bednarz & ostrowski 1996 ) .
this is also problematic for the pitch angle diffusion model based on the quasi - linear theory ; because it assumes the resonance between the magnetohydrodynamical waves and particle gyro - motion , it requires time much longer than the gyro - period .
gallant & achterberg ( 1999 ) proposed two deflection mechanisms of particles in upstream ; regular deflection owing to the gyro - motion in uniform magnetic field , and ` direction angle ' diffusion by randomly oriented small magnetic cells , where the direction angle is defined as the angle between the particle velocity and the shock normal .
gallant , achterberg & kirk ( 1999 ) and kirk et al . (
2000 ) calculated the spectral index using such direction angle diffusion model .
bednarz & ostrowski ( 1998 ) also calculated the index adopting the small angle scattering model , which is approximately equivalent to the direction angle diffusion model in a mathematical sense .
these works showed the spectral index @xmath4 typically becomes @xmath347 for ultra - relativistic shock limit , while our large angle scattering model gives @xmath348 .
this index of 2.2 may be universal when the direction angle diffusion ( or equivalent process ) is proper for the scattering / deflection process in both upstream and downstream , though it is not yet settled whether the real scattering / deflection process during such quite short time can be treated well as a diffusion in direction angle . because our probability function based description of the shock acceleration described in section 5.1 needs only two probability functions @xmath349 and @xmath350 to obtain the spectral index , it can be extended to include other scattering or deflection mechanisms if corresponding probability functions can be defined .
for example , for the model of regular deflection in upstream the probability function @xmath349 can be defined and expressed in an analytical form under the condition that the direction of velocity perpendicular to the shock normal is randomly distributed at the shock crossing from downstream .
thus , if the large angle scattering model is adopted in downstream ( assuming the bohm diffusion ) , one can calculate the spectral index .
we have given a new formulation of the first order fermi acceleration in shock waves with any shock speed in the point of view of the random walk of single particles suffering from large angle scattering .
first , we have investigated the properties of particle trajectories in a moving medium .
we showed that the problem could be formulated in terms of the random walk with absorbing boundaries based on the probability theory and derived an integral equation for the density of scattering points . by approximately solving it in an analytical form ,
we have given approximate analytic expressions for the probability density of pitch angle at return for particles which cross the shock front at a given pitch angle .
we have confirmed that our approximate solutions agree with the monte carlo results quite well for the isotropic scattering .
we have also shown that they are quite different from those based on the diffusion approximation for relativistic and mildly relativistic shocks .
it is seen that the non - diffusive effects such as the return after only a few steps of scattering or the finite mean free path , which are not included in the diffusion approximation , are very important . using these results
we have applied our formulation to the shock acceleration and compared our results with those in the literature such as ` the relativistic diffusion approximation ' , which was developed by 81 , together with the conventional diffusion approximation .
we have found that the multi - step approximation , in which the effects of the return after only a few steps of scattering are neglected , gives a quite good fit to the monte carlo results on the pitch angle distributions at the shock crossing and the spectral index of accelerated particles .
this is somewhat surprising because the multi - step approximation gives a poor fit to the monte carlo results when we fix the initial pitch angle at the shock crossing .
one reason for this is considered that when averaging over the initial pitch angle , the effects of a few steps of return and the finite mean free path tend to cancel out and as a result the diffusive return term becomes dominant .
thus , our approach gives a theoretical base to use the multi - step approximation in the shock acceleration with any shock speed .
we have also given a correct expression for the diffusive length scale for relativistic shocks , which equals to the scale lengths previously derived by 81 and kirk & schneider ( 1988 ) for far upstream distribution in relativistic shocks , instead of a naive diffusion length based on the conventional diffusion equation .
this work is supported in part by the grant - in - aid for scientific research of the ministry of education , science , sports and culture ( no .
11640236 ) .
abramowitz m. , stegun i.a .
, 1965 , handbook of mathematical functions .
dover , new york appl s. , camenzind m. , 1988 , a&a , 206 , 258 axford w.i .
, leer e. , skadron g. , 1977 , proc.15th int .
cosmic ray conf.(plovdiv ) , 11 , 132 bell a.r . , 1978a , mnras , 182 , 147 blandford r.d . , eichler d. , 1987 , phys.rep . , 154 , 1 blandford r.d . ,
mckee c.f . , 1976 , phys.fluids , 19 , 1130 blandford r.d . , ostriker j.p . , 1978 , apj , 221 , l229 bednarz j. , ostrowski m. , 1996 , mnras , 283 , 447 bednarz j. , ostrowski m. , 1998 , phys.rev.lett . , 80 , 3911 berezhko e.g. , ellison d.c . , 1999 ,
apj , 526 , 385 cox d.r . , miller h.d . , 1965 ,
the theory of stochastic processes .
methuen , london drury l.oc . , 1983 , rep.prog.phys , 46 , 973 ellison d.c . , jones f.c .
, reynolds s.p . , 1990 , apj , 360 , 702 ellison d.c . , baring m.g .
, jones f.c . , 1996 ,
apj , 473 , 1029 gallant y.a .
, achterberg a. , 1999 , mnras , 305 , l6 gallant y.a .
, achterberg a. , kirk j.g . , 1999 ,
a&as , 138 , 549 heavens a.f . ,
. , 1988 , mnras , 235 , 997 kirk j.g . , duffy p. , 1999
, j.phys.g .
nucl.part.phys . , 25 , r163 kirk j.g . , guthmann w. , gallant y.a . ,
achterberg a. , 2000 , apj , in press kirk j.g . , schlickeiser r. , schneider p. , 1988
, apj , 328 , 269 kirk j.g . , schneider p. , 1987a
, apj , 315 , 425 kirk j.g . , schneider p. , 1988 ,
a&a , 201 , 177 koyama k. , petre r. , gotthelf e.v . , hawng u. , matsuura m. , ozaki m. , holt s.s . , 1995 , nat , 378 , 255 krymsky g.f . , 1977 , soviet phys.dokl . , 22 , 327 malkov m.a . , vlk h.j . , 1995 ,
a&a , 300 , 605 melrose d.b . , 1980 , plasma astrophysics .
gordon and breach , new york naito t. , takahara f. , 1995 , mnras , 275 , 1077 ostrowski m. , 1991 , mnras , 249 , 551 peacock j.a . , 1981 , mnras , 196 , 135
in order to consider the connection of the density of scattering points in random walk to the physical number density , let us introduce the concept of _ the staying time density_. we define the staying time density @xmath351 as the total staying time in the region between @xmath7 and @xmath352 for the particle which is injected at the origin @xmath56 and absorbed eventually by one of the absorbing barriers .
the total staying time in the non - absorbing region before absorption is clearly given by @xmath353 then , the particle - averaged staying time density @xmath354 ( i.e. , the ensemble average of @xmath355 ) is given as follows .
consider the contribution to the staying time of a particle between the successive scatterings .
since the probability of the particles to move beyond @xmath21 without scattering is @xmath356 and since the staying time density is @xmath357 , the pitch angle averaged staying time density made by the particle during one step , @xmath358 , is given by @xmath359 where @xmath360 is measured in the boundary rest frame . using the probability density of the @xmath61-th scattering point @xmath62 , the staying time density made by the particles during the @xmath61-th step to @xmath65-th step is written as @xmath361 therefore , summing up all steps from @xmath362 to @xmath363 , the total staying time density is written as @xmath364 if we consider a steady state problem , in which particles are injected at the origin at a constant rate ( with pitch angle distribution @xmath17 ) , @xmath354 gives the physical number density for unit injection rate .
these relations also have been used in the monte carlo simulations ( e.g. naito & takahara 1995 ) .
the problem of the random walk of single particles , which is discussed in , is equivalent to a steady state problem in which one particle is injected at the origin per one step . in the latter problem , @xmath62 means the number density of scattering points of particles injected before @xmath61-steps and @xmath71 correspond to the number density of scattering points made by all particles existing that time . here
, we consider the integral equation for @xmath71 in this equivalent steady problem .
let us consider the random walk of particles in steady state with no boundary but with _ imaginary _ boundaries at @xmath57 and @xmath58 .
the number density of scattering points clearly equal to @xmath74 .
then , we can distinguish the particles that cross the boundaries at least once ( ` passed particles ' ) from those that never crossed the boundaries ( ` non - passed particles ' ) .
we write densities of scattering points of the former and latter particles as @xmath365 and @xmath71 , respectively .
this @xmath71 obviously coincides with the density of scattering points of the random walk of single particle with the absorbing barriers at @xmath57 and @xmath58 .
the sum of these two densities should be the solution for the no - boundary problem , @xmath74 , @xmath366 non - passed particles are injected at the origin steadily at one particle per step . on the other hand , passed particles are _ created _ when non - passed particles cross one of the imaginary boundaries and this can be regarded as the passed particles are injected at the first scattering point after they first cross the boundary . if we denote density of the first scattering point as @xmath367 , this is calculated by the distribution of non - passed particles @xmath71 as , @xmath368 each passed particle which is injected at each first scattering point @xmath369 makes the density of scattering point @xmath370 .
thus , @xmath365 is written as @xmath371 using and , we obtain @xmath372 by the definition of @xmath367 , @xmath373 @xmath374 thus , defining the kernel as @xmath375 we get a fredholm integral equation of the second kind , @xmath376 this provides an alternative form of the integral equation for @xmath71 .
this is the case of @xmath83 .
we first rewrite the wald s identity as @xmath378 this equation can be rewritten further as @xmath379 on the other hand , @xmath380 in the can be rewritten as @xmath381 thus , combining these two , we obtain the condition to be satisfied for @xmath380 : @xmath382 using the representation of p.d.f . for large angle scattering model , this condition becomes .
the absorption probability from initial position @xmath121 , @xmath215 , by our approximation is calculated as follows .
@xmath383 where we define functions @xmath384 @xmath385 and a constant @xmath386
|
we formulate the first order fermi acceleration in shock waves in terms of the random walk theory .
the formulation is applicable to any value of the shock speed and the particle speed , in particular , to the acceleration in relativistic shocks and to the injection problem , where the particle speed is comparable to the fluid speed , as long as large angle scattering is suitable for the scattering process of particles .
we first show that the trajectory of a particle suffering from large angle scattering can be treated as a random walk in a moving medium with an absorbing boundary ( e.g. , the shock front ) .
we derive an integral equation to determine the density of scattering points of the random walk , and by solving it approximately we obtain approximate solutions of the probability density of pitch angle at and the return probability after the shock crossing in analytic form .
these approximate solutions include corrections of several non - diffusive effects to the conventional diffusion approximation and we show that they agree well with the monte carlo results for isotropic scattering model for any shock speed and particle speed . when we neglect effects of ` a few step return ' , we obtain ` the multi - step approximation ' which includes only the effect of finite mean free path and which is equivalent to ` the relativistic diffusion approximation ' used by peacock ( 1981 ) if the correct diffusion length is used in his expression .
we find that the multi - step approximation is not appropriate to describe the probability densities of individual particles for relativistic shocks , but that the pitch angle distribution at the shock front in steady state is in practice quite well approximated by that given by the multi - step approximation because the effects of finite mean free path and a few step return compensate each other when averaged over pitch angle distribution .
finally , we give an analytical expression of the spectral index of accelerated particles in parallel shocks valid for arbitrary shock speed using this approximation . # 1#2 # 1section [ sec:#1 ]
# 1appendix [ apn:#1 ] # 1equation ( [ eq:#1 ] ) # 1([eq:#1 ] ) # 1fig .
[ fig:#1 ]
81peacock 87kirk & schneider 90ellison et al .
88kirk 78bell # 1#1_u # 1#1_d # 1#1_f # 1#1_rel # 1#1_+ # 1#1_- # 1#1_d # 1#1_d0 # 1#1_ud # 1#1_du # 1#1_uw # 1#1_dw # 1#1^(u ) # 1#1^(d ) 0_0 # 1dilog ( # 1 ) acceleration of particles methods : analytical shock waves cosmic rays .
|
the advances on cavity quantum electrodynamics ( cqed ) and on semi - conductor quantum electrodynamics ( scqed ) , reached in the last three decades , allowed developing devices that approach the predictions of the quantum theory of the matter - radiation interaction to the experimental results . in this context , the manipulation of the quantum states of light is a subject of growing interest , with topics as the production of only - one photon sources @xcite , the analysis of the quantum interference mechanisms like the schrdinger cat states of light @xcite and those derived from the matter - radiation interaction @xcite for instance .
they are promising topics for technological applications in quantum computation and information processing @xcite , quantum teleportation @xcite and quantum opto - electronic system @xcite-@xcite .
+ in spite of the accepted quantum nature of topics as the schrdinger cat states , a classical approach to them seems to be possible by adding novel considerations to the phase - space representation of the optical wave - field , like the spatial coherence wavelets emitted by both radiant and virtual point sources @xcite .
it could revaluate the actual limits of the classical theories and the real grounds of the physical behavior of light .
for instance , the analysis presented in this paper suggest that the schrdinger cat states are actually originated by the spatial coherence state of the light .
+ the more recent description of the properties and behavior of optical fields in arbitrary state of spatial coherence has been proposed in the framework of their phase - space representation @xcite . in this context , the spatial coherence wavelets and the marginal power spectrum @xcite have been introduced as field descriptors , in such a way that the propagation of power and correlation features of the field is depicted in terms of wavelets emitted by source pairs gathered in classes determined by the pair separations @xcite . +
this description is compatible to the second order spatial coherence theory of the optical field @xcite , and provides new insight about the interference and the diffraction phenomena , as discussed in the present work .
for instance , the marginal power spectrum provides a ray - map with two kinds of rays , i.e._the carrier rays _ along which the radiant energy of the field is transported , and the _ modulating @xmath0 rays _ which propagate positive and negative energies , responsible for the interference phenomenon @xcite . additionally to the conventional geometrical definition of the rays @xcite , the carrier rays are radiometric quantities , whose energy does not depend on the spatial coherence state of the field but it is recordable by squared modulus detectors , while the positive and negative energies associated to the modulating @xmath0 rays are determined by the spatial coherence state of the field but they are not detectable by such type of detectors , i.e. their existence is inferred from the redistribution of the power across the interference and the diffraction patterns @xcite .
+ carrier and modulating @xmath0 rays are emitted by two sets of point sources of different nature@xcite , i.e. _ radiant point sources _ for the first kind of rays , and _ virtual point sources _ responsible for the non - recordable modulating @xmath0 rays .
these sets of point sources are optically disjoint , so that they can be regarded as allocated over two distinct layers of the space , which eventually are bringing together onto the same plane in specific arrangements . in such cases ,
dual point sources are induced by the coincidence of radiant and virtual sources at the same point on the plane .
the union of the sets of radiant and virtual point sources provides a unified structure of point sources for the field .
consequently , carrier and modulating @xmath0 rays emitted by the dual sources share the same paths in space . +
this model is successfully applied in the description of diffraction of light with arbitrary states of spatial coherence and the young s experiment with diffraction effects @xcite .
two potential uses make it attractive , i.e. the developing of the robust simulation algorithms and of new strategies for optical information processing based on the manipulation of the virtual sources of the second layer of the space , instead of the radiant sources of the first layer as conventional
. both uses will be explored in next papers .
furthermore , this model reveals some unexpected features of the field , morphologically similar to the quantum squeezed states and the schrdinger cat states , without appealing to explicit quantum constrains .
let us suppose a stationary scalar wave field in any state of spatial coherence , on propagation between the aperture plane ( ap ) and the observation plane ( op ) , at a distance @xmath1 to each other in the fresnel - fraunhofer domain .
let us assume that the centre and difference coordinates @xmath2 and @xmath3 univocally specify the positions of any pair of points at the ap and the op respectively , i.e. @xmath4 and @xmath5 @xcite .
the phase space representation of this field is provided by the spatial coherence wavelets @xcite , denoted as @xmath6 + where @xmath7 , with @xmath8 the wavelength of the field , and @xmath9 + is a wigner distribution function with energy units , called the marginal power spectrum @xcite , with @xmath10 + the cross - spectral density @xcite of the field at the ap , @xmath11 $ ] the complex degree of spatial coherence @xcite , @xmath12 the power distribution of the field at the ap and @xmath13 $ ] the complex transmission of this plane .
the marginal power spectrum provides the phase - space diagram or ray - map of the scalar wave field in any state of spatial coherence @xcite .
+ the superposition of the spatial coherence wavelets yields the cross - spectral density of the field at the op @xcite , i.e. @xmath14 + whose evaluation for @xmath15 is the power spectrum recorded at this plane , i.e. @xmath16 + because of the definition of the marginal power spectrum in equation ( [ eq2 ] ) , the superposition of spatial coherence wavelets in equation ( [ eq4a ] ) can explain the interference and diffraction phenomena of light , which are close related to the lowest order of spatial coherence and are observed as modulations on the power spectrum in equation ( [ eq4b ] ) recorded at the op @xcite .
however , it is not enough to predict phenomena due to spatial coherence properties of higher order .
it is useful to introduce the dimensionless function @xmath17 $ ] @xcite with @xmath18 the dirac s delta and @xmath19 a constant that assures the dimensionless character of the function and the accomplishment of the conservation law of the total energy , into the integral in equation ( [ eq2 ] ) in order to express the marginal power spectrum as @xmath20 , with @xmath21 + and @xmath22 + where equation ( [ eq3 ] ) was taken into account with @xmath23 the power distribution that emerges from the ap , @xmath24 and @xmath25 .
the cosine function in equation ( [ eq5b ] ) results from considering the two degrees of freedom in direction of each separation vector @xmath26 .
accordingly , @xmath27 determine the flux of radiant energy provided by each individual centre of secondary disturbance at the ap onto any point at the op , while @xmath28 determines positive and negative energies due to the correlation between the pairs of centres of secondary disturbance , placed within the spatial coherence support centred at @xmath29 on the ap , onto any point at the op . the term _ structured spatial coherence support _ denotes the region around any point on the ap determined by the spatial coherence state of the wave , where pairs of emitters are included depending on specific correlation properties established in detail in @xcite .
the positive and negative energies of @xmath28 can not be interpreted as a radiant flux @xcite .
nevertheless , they play a crucial role in describing interference and diffraction from the point of view of the energy transport between the ap and the op @xcite .
indeed , equations ( [ eq4b]),([eq5a ] ) and ( [ eq5b ] ) yield the power spectrum at any point on the op , i.e. @xmath30d^{2}\xi_{a}\end{aligned}\ ] ] + with @xmath31 the radiant power provided by the individual centres of secondary disturbance , and @xmath32 the modulating power , which can take on positive and negative values , and therefore has a different physical meaning as the radiant power . the condition @xmath33 imposes @xmath34 , and therefore @xmath35 stands , i.e. @xmath32 modulates @xmath36 determining the power spectrum at any point of the op , a behavior presented in interference and diffraction light .
furthermore , the conservation law of the total energy of the field can be expressed of as @xcite @xmath37 + where @xmath38d^{2}r_{a } \nonumber\\\hspace{0.3cm}= \left(\frac{1}{\lambda z}\right)\int_{op}\int_{ap}[\textsf{\textbf{s}}_{ind}(\boldsymbol{\xi}_{a},\mathbf{r}_{a})+\textsf{\textbf{s}}_{pairs } ( \boldsymbol{\xi}_{a},\mathbf{r}_{a } ) ] d^{2}\xi_{a}d^{2}r_{a},\end{aligned}\ ] ] + with @xmath39 + and @xmath40 + equations ( [ eq8b ] ) and ( [ eq8c ] ) imply that @xmath41 and @xmath42 + accordingly , the modulating power @xmath32 is not recordable by squared modulus detectors placed at the op , but its modulation effects are revealed as a redistribution of the radiant power emitted at the ap onto the op .
this is the meaning of interference and diffraction from the point of view of the energy transport between the ap and the op @xcite .
+ the results above suggest that an optical field in any state of spatial coherence can be thought as emitted by the following types of point sources , placed at positions @xmath29 on two distinct layers of the ordinary space , as conceptually depicted in fig . [ fig1 ] : * _ radiant emitters
_ gathered on the first layer , which are responsible for the emission of the radiant energy of the field , i.e. the energy which is recorded by detectors at the op .
they are pure individual sources in the sense that their emissions , described by equation ( [ eq5a ] ) , do not depend from correlations with neighbour emitters . + * _ virtual emitters
_ gathered on the second layer , they emit the positive and negative modulating energies .
they are virtual in the sense that their contributions are not directly recordable by square modulus detectors but through the redistribution of the power spectrum at the op that they produce . for such detectors
, the second layer is actually vacuumed .
the emission of the point virtual emitters is described by equation ( [ eq5b ] ) and depends on the correlation between all the pairs of centres of secondary disturbance placed within the spatial coherence support centred at the position of the corresponding point virtual emitter . in this two - layer structure of the ordinary space , two correlated radiant emitters of the first layer turn on a virtual emitter in the second layer , located in the middle point of the distance between them . a consequence : * the distribution of radiant emitters on the first layer must be discrete , even regarding the set of maximal density of such emitters , because there should be place between consecutive correlated radiant emitters for allocating the turned - on virtual ones . *
in contrast , the set of maximal density of virtual emitters of a scalar wave field approaches to continuum by the lowest - order of spatial coherence .
* only radiant emitters will be placed just at the edge of apertures at the ap . *
radiant and virtual emitters can be placed at the same positions @xmath29 of the ordinary space ( although in separate layers ) if the correlation extends beyond the consecutive pairs .
* the strengths provided by all the radiant sources are identical , except if the diffracted wavefront is no uniform at the ap ; while those provided by the virtual sources depend of the size of the corresponding supports ( fig . [ fig2 ] ) , because each virtual source emits the addition of the contributions due to all the pairs of secondary disturbance within its support .
so , only the virtual sources associated to supports of the same shape and size emit the same strength .
thus , the complete distribution of radiant and virtual point sources of the scalar wave field will be obtained when projecting the two layers of the space onto the same plane , as depicted in [ fig1 ] .
such distribution will contain a further type of point sources , i.e. _ dual emitters _
conformed by radiant and virtual point sources placed at the same position . in arbitrary units )
in which the number of pairs within any support diminish and the diminishing rate .
there are no pairs within the support centred at the surrounding of the aperture edge no matter the spatial coherence state of the field.,scaledwidth=60.0% ] from this point of view , fully spatially incoherent scalar wave fields propagate only through the first layer of the ordinary space and are emitted only by radiant point sources ( i.e. such field are unable to turn on virtual sources , and therefore they neither contain dual sources ) . partially ( or completely ) spatial coherent fields use the both layers of the ordinary space in their propagation , because they are emitted by both the radiant and the virtual point sources ( and then , by virtual emitters ) .
+ such structure of radiant and virtual point sources for the optical field , distributed onto two different layers of the ordinary space constitutes a novel description of the spatial coherence properties of the lowest - order of the field .
it allows understanding the diffraction and interference of scalar optical fields in any state of spatial coherence not as a result of geometrical features ( i.e. the effect of the aperture edges at the ap and the phase difference between superimposed wavefronts at the op respectively ) but as a result of the same radiative feature , i.e. the emission of positive and negative modulating energies by virtual point emitters distributed within the whole aperture at the ap .
it is one - dimensionally performed by arranging a double slit mask at the ap , whose parameters are depicted in fig . [ fig3 ] . for mathematical simplicity and without lack of generality ,
let us assume diffraction of an uniform scalar wave field in fraunhofer domain by each slit , i.e. negligible phase term @xmath43 and constant power @xmath44 over the slits .
the ray - map for this experiment can be expressed as @xmath45 , with
|
in the framework of the spatial coherence wavelets , different features of the first - order spatial coherence ( young s interference ) are analyzed by calculating the corresponding marginal power spectrum , a close related quantity to the classical wigner distribution function ( wdf ) of the optical field .
the consideration of the radiant and virtual point sources evidences some effects , conventionally attributed to non - classical correlations of light , although such type of the correlations are not explicitly included in the model .
specifically , a light state is produced that has similar morphology to the wdf of the well - known quantum schdinger cat state .
|
the calculation of real - time quantum dynamics for large molecular systems is a longstanding problem in chemical physics . of particular interest
is the case of a small subsystem embedded in a surrounding thermal bath , which forms the basis for the investigation of condensed phase energy and electron transfer as well as spin and charge transport in nanoscale devices@xcite . in such situations ,
one is typically interested in the calculation of the system s _ reduced _ dynamics , averaged over the bath degrees of freedom .
due to the importance of these problems and the absence of a general solution , a variety of methods have been developed which vary as to the regimes in which they are accurate and their applicability to large systems .
many of these approaches rely on either averaging over semi - classical trajectories of bath degrees of freedom or propagating the system s reduced density matrix ( rdm ) directly without explicit treatment of the bath , for example by master equations or path integral techniques .
trajectory based methods include the mean - field ehrenfest@xcite , surface - hopping@xcite , semi - classical initial value representation@xcite , linearization@xcite , and classical mapping@xcite approaches .
these offer the appeal of a transparent link to the classical limit and simple application to complex condensed phase systems and to slow baths .
however , these methods typically fail in regimes with strong system - bath coupling or high - frequency baths where quantum effects are known to be important .
techniques based on direct propagation of the reduced density matrix include the markovian and non - markovian bloch - redfield equations@xcite and the noninteracting blip approximation@xcite which serves as a time - dependent generalization of frster theory@xcite .
to make such approaches manageable , perturbation theory in the system - bath coupling ( bloch - redfield ) and in the non - adiabatic coupling ( niba , frster ) are typically carried out only to second - order and hence performance degrades when these terms become large .
in addition , path integral approaches , such as quantum monte carlo @xcite and the quasi - adiabatic propagator path integral ( quapi)@xcite have also been developed to propagate the rdm . while these approaches can be made numerically exact , they are frequently difficult to converge in practice when the system - bath coupling is strong or the bath is slow . furthermore , many variants scale poorly with the number of system states , drastically limiting the size of treatable systems .
hence when the bath dynamics are `` fast '' compared to those of the system ( the so - called _ non - adiabatic _ regime ) , both perturbative quantum master equations and numerically exact path integral approaches are typically accurate since the correlation time of the bath is short , rendering non - markovian memory effects less significant .
when the bath dynamics are `` slow '' compared to those of the system ( the _ adiabatic _ regime ) , these methods fail but it becomes reasonable to treat the bath degrees of freedom classically and hence trajectory - based schemes generally yield accurate results . given this complementarity , it would clearly be desirable to develop a scheme which can accurately treat both regimes by naturally tuning between the above approaches in a consistent manner .
in addition , such an approach might effectively treat intermediate regimes where no obvious time - scale separation exists and hence neither rdm nor trajectory - based approaches are accurate .
the `` dynamical hybrid '' scheme of wang , thoss and miller @xcite offers such an approach in which bath degrees of freedom are partitioned into `` core '' modes to be treated with quantum mechanics and `` reservoir '' modes to be treated with classical mechanics . in this scheme , the wavefunction - based multi - configurational time - dependent hartree ( mctdh)@xcite method is used for the system and core while a mean - field ehrenfest treatment is applied to the reservoir .
when applied to the spin - boson problem , this dynamical hybrid scheme frequently converges to the exact result with only a small percentage of modes included in the core .
however , to achieve convergence in the presence of fast baths , up to 80% of the bath modes need to be included in the core .
due to the high cost of treating modes using mctdh , this severely reduces the efficiency of the dynamical hybrid approach and hence limits the ease with which it can be applied to larger systems . in this work
, we present a methodology which avoids the use of wavefunctions , replacing them with a single reduced density matrix averaged over the quantum core modes .
the reservoir modes exert a classical time - dependent driving force on this system rdm , which in turn drives the classical modes via a back - reaction force .
this reduced density matrix hybrid formalism ( rdm - hybrid ) allows one to use an appropriately chosen partitioning of the bath into core and reservoir modes such that the system rdm can be propagated either by an exact reduced dynamics algorithm _ with a decreased cost _ or by an inexpensive approximate master equation _ with improved accuracy_. in other words , by using physical intuition , one can choose the core - reservoir partitioning such that different methods are applied only to the set of bath modes for which they are expected to work .
we demonstrate our approach by applying it to the spin - boson model and treating the system and core using the noninteracting blip approximation ( niba)@xcite this approach yields a methodology which is numerically cheap while obtaining impressive quantitative accuracy over all regimes investigated as well as offering excellent scaling with the number of system states .
the outline of the paper is as follows .
section [ ssec : ehren ] briefly reviews the ehrenfest methodology .
this background is then used in sec .
[ ssec : hybrid ] , to derive the rdm - hybrid approach .
section [ sec : spinboson ] then describes how this scheme can be applied to the spin - boson model .
the results of the application are presented in sec .
[ sec : results ] , and sec .
[ sec : conc ] concludes .
while there are many routes to the mean - field ehrenfest equations of motion and subsequently the rdm - hybrid method to be presented below , we proceed via the quantum - classical liouville equation , which provides a particularly simple derivation .
we begin by considering the generic coupled system - bath hamiltonian , @xmath0 where the free bath hamiltonian is @xmath1\ ] ] and the coupling is assumed to be of the form @xmath2 the dynamics of the total density matrix , @xmath3 , is given by the liouville equation , @xmath4,\ ] ] where henceforth we set @xmath5 . by employing the partial wigner transform of the density matrix with respect to the @xmath6 environmental degrees of freedom , @xmath7 and taking the classical limit of the corresponding partial wigner transformed liouville equation ( for more details , see e.g. ref . ) , one obtains the so - called quantum - classical liouville equation @xmath8 \nonumber\\ & \hspace{4em } + \frac{1}{2 } \left ( \big\ { h^w,\rho^w(t ) \big\ } + \big\ { \rho^w(t ) , h^w \big\ } \right),\end{aligned}\ ] ] where @xmath9 $ ] denotes a commutator and @xmath10 is the poisson bracket .
now , one makes the approximation that the system and bath reduced density matrices ( rdms ) decouple at all times , @xmath11 with the system rdm @xmath12 and bath rdm @xmath13 $ ] . inserting this product form into the quantum - classical liouville equation and noting that the bath reduced density matrix is simply the classical phase space distribution , @xmath14 one obtains the mean - field ehrenfest equations of motion@xcite for the system rdm , @xmath15 \nonumber\\ & = -i \big [ h_s + \sum_i f_i(s ) g_i\big(\mathbf{q}(t)\big ) , \rho_s(t ) \big ] \nonumber\\ & \equiv -i \big [ \tilde{h}_{s}(t ) , \rho_s(t ) \big],\end{aligned}\ ] ] and the classical bath degrees of freedom , @xmath16 \nonumber\\ & = -\frac{\partial v(q_k)}{\partial q_k } - \sum_i \frac{\partial g_i\big(\mathbf{q}(t)\big)}{\partial q_k } { \rm tr}_{s } \big\ { f_i \rho_s(t ) \big\}.\end{aligned}\ ] ] the ehrenfest approximation to an observable of the reduced system is given by @xmath17 in accordance with the classical treatment of the bath , this latter integral is evaluated by molecular dynamics obeying the equations of motion given above . to generate the canonical ensemble average consistent with the initial density matrix @xmath18 one averages over classical trajectories with bath initial conditions sampled from either the wigner - transformed bath density operator , @xmath19 or more simply from the classical boltzmann distribution , @xmath20
although the wigner distribution gives initial conditions consistent with the exact quantum distribution , this property is not conserved by the classical dynamics .
in contrast to the approach taken above , in which the full hamiltonian is split into a system and bath with coupling between the two , we instead divide the bath degrees of freedom into two sets : the reservoir modes ( @xmath21 ) to be treated classically and the core modes ( @xmath22 ) to be treated quantum mechanically . hence , we partition the hamiltonian as @xmath23 where the system - core hamiltonian is @xmath24 and the system - core and system - reservoir couplings are defined as @xmath25 since nothing in our above discussion of ehrenfest dynamics utilized the specific properties of the system , we could just as well re - label the _ combined system and core _ as the _ system _ of the previous section . likewise , we associate the _ reservoir _ with the _ bath _ from before . thus we assume the total ( wigner - transformed ) density matrix factorizes at all times into a system - core rdm and reservoir rdm , @xmath26 taking the classical limit we arrive at the ehrenfest - like rdm - hybrid equations of motion for the system - core density matrix , @xmath27 \nonumber\\ & = -i \big [ h_s + \sum_i f_i(s ) g_i\big(\mathbf{q}(t)\big ) , \rho_{sc}(t ) \big ] \nonumber\\ & \equiv -i \big [ \tilde{h}_{s}(t ) , \rho_{sc}(t ) \big ] , \label{eq : rhosc}\end{aligned}\ ] ] and the classical bath degrees of freedom , @xmath28 \nonumber\\ & = -\frac{\partial v(q_k)}{\partial q_k } - \sum_i \frac{\partial g_i\big(\mathbf{q}(t)\big)}{\partial q_k } { \rm tr}_s { \rm tr}_c \big\ { f_i \rho_{sc}(t ) \big\}. \label{eq : batheom2a}\end{aligned}\ ] ] although the exact calculation of the system - core density matrix dynamics given by eq .
( [ eq : rhosc ] ) is extremely demanding , this was essentially the approach taken by wang , thoss , and miller ( wtm)@xcite who averaged high - dimensional wavefunction trajectories with initial conditions of the core wavefunction sampled from the exact quantum mechanical core density matrix and classical reservoir degrees of freedom sampled from the quantum - classical wigner distribution .
the important point we seek to make in this work is that @xmath29 is a pure system operator , i.e. the core and reservoir are not directly coupled , such that the classical reservoir equation of motion above may be written @xmath30 where we have introduced the system rdm , @xmath31 , averaged over the quantum core degrees of freedom .
furthermore , the calculation of any dynamical system variable is given by an expression analogous to the pure ehrenfest approximation above , @xmath32 which again only requires the system rdm ( and _ not _ the much higher - dimensional system - core rdm ) , with classical reservoir trajectories to be sampled from an appropriate distribution .
we have thus arrived at a self - consistent dynamical scheme which treats the core bath modes quantum mechanically yet only requires the propagation of a typical system rdm and classical reservoir degrees of freedom .
this rdm - hybrid formulation presented above constitutes the major theoretical result of this work .
we point out that a similar approach appears to have been investigated by golosov et al.@xcite , though it was restricted to their previously developed memory - equation algorithm@xcite and only minimally pursued .
when the system rdm averaged over the quantum core modes , @xmath33 , is treated fully quantum mechanically , our methodology yields the exact result when all bath modes are included in the core .
however , in practice one expects to achieve numerical convergence before this limit is reached .
of course , the exact quantum dynamical treatment of the system rdm is in general no less numerically challenging than the original problem ( a system coupled to a quantum bath ) .
however , the advantage of this approach is that one can tailor the partitioning in eq .
( [ eq : partition ] ) either to alleviate the computational expenses of a numerically exact reduced quantum dynamics method ( such as the quapi - based iterative tensor propagation scheme of makri et al.@xcite ) or to improve the accuracy of approximated quantum dynamics ( such as a perturbative quantum master equation ) for the system rdm . in this sense ,
the above presented formalism outlines a _ framework _ for the efficient partitioning and calculation of real - time reduced quantum dynamics where one can employ any exact or approximate dynamical scheme for the system rdm depending on the complexity of the problem . as discussed in the following section , here we adopt an approximate treatment of the system rdm using a perturbative master equation to yield a very cheap hybrid methodology with impressive quantitative accuracy .
while the above formalism is indeed quite general here we demonstrate its effectiveness at treating the spin - boson hamiltonian , @xmath34/2,\ ] ] which describes a two - level system with energy bias @xmath35 and tunneling matrix element @xmath36 , linearly coupled to the coordinates of a bath of harmonic oscillators . the spin - boson hamiltonian has been used as a model for a variety of physically distinct quantum relaxation and transport processes .
a principal observable of interest is the population variable , @xmath37 \equiv \langle \sigma_z(t ) \rangle,\ ] ] which measures the difference in population of sites 1 and 2 .
furthermore , we will henceforth work in reduced units with respect to the tunneling matrix element , @xmath36 , in addition setting @xmath38 . to begin our application
, we first discuss the separation of modes into the core and the reservoir .
the harmonic bath in the spin boson problem is fully specified by its spectral density @xmath39 when separating the bath modes into a core and reservoir , we will require that @xmath40 intending to treat high frequency modes in the quantum mechanical core and low frequency modes in the classical reservoir , we use a switching function , @xmath41 , which switches from 1 to 0 as @xmath42 goes from 0 to @xmath43 .
this allows the core and reservoir spectral densities to be defined as , @xmath44 , \\
j_{\rm res}(\omega ) & = j(\omega)s(\omega,\omega^*).\end{aligned}\ ] ] an infinite number of switching functions satisfy such a criteria and the best form will depend on the hamiltonian adopted , the methodology used to treat the core , and the required balance been accuracy and efficiency . the simplest approach would be to use a step function , @xmath45 however in many cases , such a sharp switch will leave a core which is expensive to treat using quantum mechanics . the reason for this difficulty can be seen by considering the bath correlation function , given by @xmath46,\end{aligned}\ ] ] whose correlation time determines the range of memory ( non - markovian ) effects .
it is straightforward to see that using a step function switching like that in eq .
( [ eq : step ] ) will result in a long - time oscillatory tail in the bath correlation function , @xmath47 , and thus extensive non - markovian effects which are difficult to treat quantum mechanically .
it is therefore advantageous to separate the spectral density using a smooth switching function which can yield a short bath correlation time for the core modes and allow for efficient quantum mechanical treatment . for our purposes a switching function of the form , @xmath48 ^ 2 & \omega < \omega^ * \\ 0 & \omega >
\omega^ * \end{cases}\ ] ] was chosen . in fig .
[ fig : split ] , we show the bath correlation function when this function is applied to the debye spectral density , @xmath49 where @xmath50 is the reorganization energy . as can be seen
, this smooth switching does not introduce any long - time tails and allows the bath correlation time for the core modes to be decreased as @xmath43 is increased and more of the slow modes , which give rise to temporally non - local effects , are moved into the reservoir . in addition , removal of slow modes from the core encourages the use of approximate master equations , such as the noninteracting blip approximation discussed in the next section , which complement the ehrenfest treatment of the slow modes .
( bottom ) . ]
although the rdm - hybrid framework allows for the application of many approaches to treat the system and core modes , in this work we employ the noninteracting blip approximation ( niba)@xcite . to see why such a choice is computationally simple , note that because the fluctuating classical degrees of freedom provide an effective time - dependent bias , the system hamiltonian no longer commutes with itself at different times , and thus its propagator is a cumbersome time - ordered exponential ( dyson series ) .
however , since niba only treats the diagonal part of this hamiltonian exactly , its propagator is trivially calculated .
another important reason for our choice is the complementarity between niba and ehrenfest . while ehrenfest dynamics work best in the adiabatic regime , @xmath51 , niba is perturbative in @xmath52 , thus working best in the non - adiabatic regime , @xmath53 . to exemplify this complementarity , in fig .
[ fig : crossover ] we show a coherent to incoherent crossover , as a function of @xmath54 in the high - temperature , strong - coupling regime . as can be seen in fig .
[ fig : crossover ] , the ehrenfest method yields quantitatively exact population dynamics for an adiabatic choice of parameters , @xmath51 [ panels ( a ) and ( b ) ] , whereas niba gives the exact result upon crossing over into the non - adiabatic regime , @xmath53 [ panels ( c ) and ( d ) ] .
with this in mind we are strongly encouraged to consider a niba - ehrenfest hybridization , the details of which are described next .
, for the spin - boson model with a debye spectral density .
the hamiltonian parameters are @xmath55 , @xmath56 , and @xmath57 .
ehrenfest and niba are compared to the numerically exact results of thoss , wang , and miller@xcite in ( a)-(c ) and to our own converged quapi calculation in ( d ) . ]
the total bath was discretized by the relation , @xmath58 where @xmath59 is a density of frequencies chosen to reproduce the reorganization energy for any number of modes , @xmath6 .
for an ohmic spectral density @xmath60 with cutoff function @xmath61 , we thus require @xmath62 which is clearly satisfied by the choice @xmath63 for the debye spectral density considered here , the cutoff function is given by @xmath64 $ ] . having obtained a discrete distribution of frequencies , we then take into account the switching function and _ recalculate _ the reservoir couplings as @xmath65 in the process removing all bath modes with @xmath66 from the reservoir . in a complimentary fashion , the core spectral density , @xmath67 , is used in the niba equations . in all results presented ,
@xmath68 discrete bath modes were found to be sufficient for convergence and averages were performed over @xmath69 reservoir initial conditions . for pure ehrenfest results , we sample from the wigner distribution , which for the spin - boson hamiltonian s harmonic bath takes the form , @xmath70.\end{aligned}\ ] ]
as alluded to above , when high frequency baths are present , wigner sampling can improve the accuracy of ehrenfest dynamics since the initial conditions are consistent with the exact quantum mechanical distribution .
however , since the zero - point energy inserted by wigner sampling is not conserved by the classical dynamics , long - time predictions can be unreliable ( zero - point energy leakage)@xcite .
hence , in regimes with strong coupling or high bath frequencies , sampling from either the classical boltzmann or quantum wigner distributions can produce inaccurate results .
in contrast , for our rdm - hybrid approach we find that the results are largely insensitive to the choice of reservoir initial conditions , since most quantum mechanical modes are included in the core leading to only low frequencies being present in the reservoir , for which classical and wigner sampling give identical results .
this effect is shown later , in fig . [
fig : nonadiabatic ] , where pure ehrenfest dynamics using classical sampling is seen to underestimate the decay of the system population variable while using wigner sampling overestimates the decay .
the rdm - hybrid approach , however , produces graphically identical results using either sampling scheme .
the time evolution of the coupled classical dynamics and niba equations were solved with a second - order runge - kutta scheme with a timestep of @xmath71 . for the spin - boson hamiltonian considered here , the reservoir equations of motion , eqs .
( [ eq : batheom1])-([eq : batheom2b ] ) , take the form @xmath72 with the system population variable @xmath73
. taking @xmath74 to be constant over a half - timestep as called for by the runge - kutta scheme used here , one has @xmath75 \cos ( \omega_k \delta t/2 ) \nonumber\\ & \hspace{1em } + \frac{p_k(t)}{\omega_k } \sin ( \omega_k \delta t/2 ) - \frac{c_k}{\omega_k^2}\tilde{p}(t),\\ p_k(t+\delta t/2 ) & = -\left [ q_k(t ) + \frac{c_k}{\omega_k^2 } \tilde{p}(t ) \right ] \omega_k \sin ( \omega_k \delta t/2 ) \nonumber\\ & \hspace{1em } + \frac{p_k(t)}{\omega_k } \cos ( \omega_k \delta t/2).\end{aligned}\ ] ] for a two - level system coupled to a common bath , the niba master equation for the population difference with a time - dependent bias is given simply as@xcite @xmath76,\ ] ] with the kernels @xmath77\cos\left[q_1(t-\tau)\right ] \nonumber\\ & \hspace{1em}\cos\left\ { \zeta(t,\tau ) + ( 1+\delta)\left[q_1(t)-q_1(\tau)\right ] \right\ } , \label{eq : kernel1}\\ k_-(t,\tau ) & = 4\delta^2 \exp\left[-q_2(t-\tau)\right]\sin\left[q_1(t-\tau)\right ] \nonumber\\ & \hspace{1em}\sin\left\ { \zeta(t,\tau ) + ( 1+\delta)\left[q_1(t)-q_1(\tau)\right ] \right\}. \label{eq : kernel2}\end{aligned}\ ] ] the parameter @xmath78 originates from the bath initial condition , with @xmath79 if the system and bath are initially uncoupled and @xmath80 if the bath is initially solvating the donor .
the twice - integrated bath correlation function , @xmath81 , is given by @xmath82\right .
\nonumber\\ & \left.+\ i\sin(\omega t ) \right\}.\end{aligned}\ ] ] the effect of a general time - dependent bias @xmath83 is captured in the function @xmath84 which for the core - reservoir splitting considered here is defined as @xmath85.\ ] ] the true population difference , @xmath86 is finally given by the average of @xmath74 over classical trajectories with initial conditions sampled from the appropriate distribution .
importantly , the scheme described here yields the correct population dynamics of a two - level system driven by a classical external field , so that by taking the limit @xmath87 , we remove all bath modes from the core , and recover the ehrenfest solution . in the opposite limit , @xmath88
, we recover the full niba treatment . because we are not treating the core exactly
, it is important to emphasize that @xmath43 is no longer a convergence parameter . instead , @xmath43 tunes the balance between the two approximate methods , here taken to be ehrenfest dynamics and niba .
the optimal _ a priori _ choice of this switching frequency is an interesting problem for future work , but here we present a physically motivated prescription for its determination . we again recall that `` adiabaticity '' is inherently a problem of timescale separation .
the adiabatic regime is physically realized when the timescale of the bath is much greater than that of the system , and _
vice versa _ for the non - adiabatic regime .
a given bath mode @xmath89 could be classified into `` core '' or `` reservoir '' by comparing its characteristic timescale , @xmath90 with that of the system , @xmath91 . for the spin - boson hamiltonian , eq .
( [ eq : spinboson ] ) , the system frequency @xmath92 is given by the rabi frequency , @xmath93 in practice we use the smooth switching function , eq .
( [ eq : smoothswitch ] ) , and following the line of argument above , set the switching frequency equal to the rabi frequency , i.e. @xmath94 .
this procedure effectively partitions the bath modes into those that are faster than the system dynamics ( comprising the core ) and those that are slower ( comprising the reservoir ) .
while comparison with existing exact results has shown that this choice , @xmath94 , is not always truly optimal , we will show in the next section that it nevertheless provides a very robust methodology , with quantitative predictive ability .
in all our results we compare to the numerically exact population dynamics computed by thoss , wang , and miller@xcite for the spin - boson hamiltonian , eq .
( [ eq : spinboson ] ) , with a debye spectral density , eq .
( [ eq : debye ] ) and initial condition @xmath95 , i.e. @xmath79 in eqs .
( [ eq : kernel1])-([eq : kernel2 ] ) .
population dynamics of the spin - boson model in the adiabatic regime ( @xmath96 ) without an energetic bias .
the bath has a debye spectral density and parameters as given in the figure .
note the difference in the time axis . ]
we begin in the adiabatic regime , @xmath96 , where ehrenfest dynamics are known to be relatively accurate .
[ fig : adiabatic](a ) shows that this is indeed the case for high temperature , @xmath97 . though qualitatively good , the very strong coupling , @xmath98 , degrades the accuracy of the ehrenfest approach . however , our rdm - hybrid method yields a long - time population decay in much better agreement with exact results . in fig .
[ fig : adiabatic](b ) , we investigate another system with strong coupling , @xmath56 , but with a temperature one order of magnitude lower .
the classical mechanical approximation made by using ehrenfest dynamics intrinsically worsens for lower temperature , where quantum mechanical effects are expected to play a more dominant role .
we see that the rdm - hybrid result is again in better quantitative agreement with the exact long - time limit . both panels ( a ) and ( b ) in fig .
[ fig : adiabatic ] underscore the important point that our rdm - hybrid approach is not simply an `` average '' of the two methods , but can in fact generate qualitatively different dynamics which are not situated `` in between '' those of the individual methods .
the system trapping seen above leading to a quasi - stationary state is nearly impossible to capture with perturbative quantum master equation approaches , as exemplified by the poor niba results . for comparison ,
the wtm mctdh - ehrenfest scheme required 25% of the bath - modes to be treated quantum mechanically , supporting our observation of largely classical bath dynamics .
population dynamics of the spin - boson model in the non - adiabatic regime ( @xmath99 ) without an energetic bias .
the bath has a debye spectral density and parameters as given in the figure .
also shown are the effects of classical vs. wigner sampling of the bath ( reservoir ) initial conditions for ehrenfest dynamics ( rdm - hybrid ) . ] in fig .
[ fig : nonadiabatic ] , we investigate the non - adiabatic regime , @xmath99 , with strong coupling @xmath98 , where niba is quantitatively accurate . here
we show the effects of sampling from a classical distribution or a wigner distribution .
the ehrenfest approximation with wigner initial conditions yields a decay which is too rapid , while that with classical initial conditions yields a decay which is too slow .
both completely miss the two - step relaxation dynamics at very short times .
rdm - hybrid captures these short time dynamics exactly and shows a robust insensitivity to the initial sampling employed .
although our hybrid method exhibits a slight discrepancy at longer times , it is much more accurate than the ehrenfest approach .
population dynamics of the spin - boson model in the intermediate regime ( @xmath100 ) without an energetic bias .
the bath has a debye spectral density and parameters as given in the figure .
note the difference in the time axis . ]
having demonstrated that our rdm - hybrid approach can yield very good agreement with the exact results when only one of its composite methods ( ehrenfest dynamics or niba ) is successful , we now consider in fig .
[ fig : intermediate ] the intermediate regime , @xmath100 , where neither method is particularly accurate by itself . fig .
[ fig : intermediate](a ) depicts the population dynamics at the high temperature @xmath97 for relatively strong coupling , @xmath56 . here , both niba and ehrenfest dynamics are in very poor numerical agreement with the exact result , though the ehrenfest dynamics correctly predicts a two - step relaxation , albeit only qualitatively . on the contrary ,
rdm - hybrid correctly exhibits a rebound in the population dynamics near @xmath101 ( although one that is somewhat overpronounced ) and yields dynamics at all later times in excellent agreement with the exact result . in the lower panel , fig .
[ fig : intermediate](b ) , we consider a weaker coupling and a temperature _ two orders of magnitude _ lower . here , niba does quite poorly and ehrenfest dynamics are mildly better , exhibiting slightly overdamped oscillations and a weak phase - shift at long times . once again , rdm - hybrid is nearly exact , despite this severely low temperature . in both examples above , i.e. figs .
[ fig : intermediate](a ) and ( b ) , thoss , wang , and miller found it necessary to treat over 50% of the bath modes quantum mechanically with mctdh .
it is remarkable that our computationally inexpensive rdm - hybrid method employing niba for the core modes is able to yield such similarly accurate results for these problems , which serves as a testament to its robustness .
population dynamics of the spin - boson model in the adiabatic regime ( @xmath96 ) without an energetic bias ( a ) and with an energetic bias ( b ) .
the bath has a debye spectral density and parameters as given in the figure .
note the difference in the time axis . ]
population dynamics of the spin - boson model in the non - adiabatic regime ( @xmath99 ) without an energetic bias ( a ) and with an energetic bias ( b ) .
the bath has a debye spectral density and parameters as given in the figure . ]
we conclude this section by briefly considering the effect of an energetic bias on the performance of our rdm - hybrid method . in figs .
[ fig : adiabaticbias](a ) and ( b ) , we show two sets of population dynamics for the same high - temperature , adiabatic regime , but with an energetic bias @xmath102 in panel ( b ) . as expected , niba performs poorly due to the adiabaticity of the conditions under consideration in both panels . as for the ehrenfest method , while it gives excellent results in the unbiased case [ panel ( a ) ] , it strongly deviates from the correct dynamics in the presence of a bias [ panel ( b ) ] .
this deficiency of the ehrenfest approach and similar quantum - classical methods for biased system is well - known and typically connected to unphysical treatment of zero - point energy in classical treatments .
attempts to correct this behavior have included an adjustable zero - point energy parameter for sampling the initial conditions @xcite and its determination by comparison with exact short - time moments.@xcite in spite of this potential concern , we see in fig .
[ fig : adiabaticbias ] that the rdm - hybrid approach performs incredibly well and equally so for both unbiased and biased systems .
unfortunately , this success is not universal , as seen in fig .
[ fig : nonadiabaticbias ] , which considers the effect of a bias on non - adiabatic dynamics .
while rdm - hybrid is very successful for the unbiased case , it performs more poorly in the presence of a bias .
this error appears to be attributable to the ehrenfest dynamics , which rdm - hybrid tracks quite closely at short times .
nonetheless , the rdm - hybrid population does appear to approach the correct long - time value , which both ehrenfest and niba incorrectly predict .
to summarize , we have formulated a quantum - classical methodology for the quantum dynamics of a coupled system - bath hamiltonian , averaged over the bath degrees of freedom , by employing a core - reservoir partitioning of the bath degrees of freedom .
unlike the technique of wang , thoss , and miller@xcite , our rdm - hybrid approach avoids the use of expensive wavefunction - based quantum mechanics , replacing it by a reduced density matrix calculation .
this allows for flexibility to use a variety of inexpensive approximate approaches for different parts of the computation , allowing for combined accuracy and efficiency . within the above framework
, we then used the noninteracting blip approximation ( niba ) for the reduced dynamics , yielding an efficient , scalable quantum dynamics methodology with excellent accuracy for both adiabatic and non - adiabatic parameter regimes . specifically , for @xmath103 discrete states , a numerical implementation of niba scales only as @xmath104 ( see eq .
( [ eq : niba ] ) ) , which allows for very efficient investigation of bigger systems .
the largest discrepancy with exact results was found in the non - adiabatic regime with an energetic bias . to alleviate such deficiencies , we enumerate a number of avenues for future work .
one could imagine pursuing alternative approximate treatments of the quantum core - averaged system rdm .
for example , the accuracy of niba ( employed here ) is known to degrade at low temperature , in particular for biased systems .
the redfield equations , on the other hand , have been very successfully applied to such systems , being restricted , however , to weak - coupling and non - adiabatic baths .
thus , an rdm - hybrid approach employing redfield and ehrenfest dynamics may be one way to extend the validity of the redfield equations into these regimes where it would otherwise fail .
ideally , one would like an exact treatment of the quantum system rdm , such that the method becomes numerically exact when all modes are included in the core . as alluded to at the end of sec .
[ ssec : hybrid ] , we believe the tensor propagation scheme of makri et al .
@xcite is especially promising . because the quapi algorithm scales exponentially with the memory time required to span the bath correlation function ,
the partitioning discussed here leading to a reduced bath correlation time would constitute an enormous reduction in computational expense .
work along this direction is currently in progress .
lastly , for a chosen system - core dynamics method , a more rigorous investigation of the form of the switching function , @xmath41 , and switching frequency , @xmath43 , should be pursued . ideally these choices should be automated and not left up to physical considerations as was done here .
for example , one might try comparing the short - time moments of the population dynamics to those obtained exactly from quantum perturbation theory .
the investigation of this point will be a subject of future work . in spite of these desired improvements
, the rdm - hybrid method employing niba has been shown to be very accurate , often correcting the long - time populations of the otherwise accurate ehrenfest dynamics .
future work will include the application and investigation of both of these methods to models of electronic energy transfer in molecular aggregates , including the multi - site frenkel exciton hamiltonian describing the fenna - matthews - olsen complex , a prototypical photosynthetic system .
the authors gratefully thank aaron kelly for a critical reading of the manuscript and helpful comments .
t.c.b . was supported by the department of energy office of science graduate fellowship program ( doe scgf ) , administered by orise - orau under contract no .
de - ac05 - 06or23100 .
d.r.r . was supported by the national science foundation under grant no .
che-0719089 .
40ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ _
( , ) @noop _ _ ( , ) @noop _ _ , edited by , , and ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , edited by , , , and ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
|
we present a new approach to calculate real - time quantum dynamics in complex systems .
the formalism is based on the partitioning of a system s environment into `` core '' and `` reservoir '' modes , with the former to be treated quantum mechanically and the latter classically .
the presented method only requires the calculation of the system s reduced density matrix averaged over the quantum core degrees of freedom which is then coupled to a classically evolved reservoir to treat the remaining modes .
we demonstrate our approach by applying it to the spin - boson problem using the noninteracting blip approximation to treat the system and core , and ehrenfest dynamics to treat the reservoir .
the resulting hybrid methodology is accurate for both fast and slow baths , since it naturally reduces to its composite methods in their respective regimes of validity .
in addition , our combined method is shown to yield good results in intermediate regimes where neither approximation alone is accurate and to perform equally well for both strong and weak system - bath coupling .
our approach therefore provides an accurate and efficient methodology for calculating quantum dynamics in complex systems .
|
dynamical systems with delays play an important role in modeling natural processes which include the influence of past effects for a better description of the evolution . in classical physics , life sciences , physiology , engineering system , neural networks , epidemiology , and economics
, realistic models must take in account the time - delays due to the finite propagation speed to determine the future evolution .
many examples of real phenomena with delay effects can be found in @xcite .
furthermore of importance in applications , the delay system is , in general , described by delay differential equations which have several distinct mathematical properties of ordinary and partial differential equations , and also provides them with a purely mathematical interest . in this work , we consider the differential equation with discrete delays .
let @xmath0 , @xmath1 , not all zero @xmath2 , @xmath3 , and @xmath4 is a continuous preshape function defined on @xmath5 $ ] .
consider the following delay differential equation : @xmath6 we allow the discontinuity of the preshape function @xmath4 at @xmath7 , i.e. , @xmath8 is not necessary equal to @xmath9 .
we want to find analytically the formula to describe the effect of input function @xmath10 to the system s output @xmath11 instead of just the output due to a specified input via numerical method , like ` dde23 ` in matlab . for those delay systems with different lengths of delay - time
, it can also be transferred into the form given by ( [ eqn : mdde ] ) .
thus our result can be applied without any difficulty . for the single delay differential equation , i.e. , set @xmath12 , @xmath13 , @xmath14 , and @xmath15 into ( [ eqn : mdde ] ) ) : @xmath16 refer to the paper by coreless _
al . _ , 1996 @xcite ,
it is suggested by using lambert w function to describe the corresponding solution as @xmath17 for any reasonable choice of @xmath18 ( i.e. such that the sum makes sense ) . here
@xmath19 denotes the @xmath20-th branch of lambert w function
. introduced by lambert and euler in 1700 s , the lambert w function is defined to be any function @xmath21 satisfies @xmath22 taking the logarithms on both sides of ( [ eqn : lambertw ] ) leads to @xmath23 after some arrangement , and it is obviously that there are infinite many solution due to the property of logarithmic functions in the complex plane . when @xmath24 and @xmath25 , there are two possible real values of @xmath21 ( see figure [ fig : lambertw]a ) .
denote the branch satisfying @xmath26 by @xmath27 , or just @xmath21 when there is no confusion , and the branch satisfying @xmath28 by @xmath29 .
the @xmath21 is referred to as the _ principal branch _ of the lambert w function .
when @xmath30 is complex , we use @xmath31 to denote the @xmath20-th branch of the lambert w function whose ranges are shown in figure [ fig : lambertw]b .
we can verify that @xmath32 for all integers @xmath20 .
applications of lambert w function , we refer to coreless _ et . al _ , 1996 @xcite , and references there in .
[ cols="^,^ " , ]
lambert w function has been widely applied in the research of science and engineering problems . in this paper
, we construct the time response of scalar systems with multiple discrete delays based on the laplace transform .
we first establish the state transition function as the sum of infinite series of exponentials acting on eigenvalues inside countable branches of the lambert w functions . and the time response function
is then obtained in terms of initial condition , preshape function , and the convolution between state transition function and input function .
eigenvalues in each branch of lambert w function are computed by a numerical iteration .
once the eigenvalues are obtained , the approximation of time responses is calculated .
numerical examples are presented to conform the successful application of lambert w function in establishing the time response of multi - delay systems .
our main contribution is to construct an analytical formula in representing the time response which can act as the concise foundation of designing the feedback controller for multi - delay systems .
|
in this work , we establish the response of scalar systems with multiple discrete delays based on the laplace transform . the time response function is expressed as the sum of infinite series of exponentials acting on eigenvalues inside countable branches of the lambert w functions .
eigenvalues in each branch of lambert w function are computed by a numerical iteration .
numerical examples are presented to illustrate the results obtained .
delay system , lambert w function , multiply delays , time response , 34a45 , 34k06 , 34k07 , 34k35 , 44a10
|
the failure of antibiotic treatments to eliminate bacterial infections has become both more evident and better understood in the past several decades . although there is evidence that the over use of antibiotics has amplified the number of chromosomal - resistant bacteria @xcite , it is becoming increasingly clear that there are other mechanisms that protect populations of bacteria
. many of these mechanisms depend on whether the bacteria exist in a biofilm or not @xcite .
in particular , the notion that small sub - populations of bacteria may display innate tolerance to various biocides has been proposed as a possible reason for the failure of treatment for bacterial infections @xcite .
bacteria within a biofilm are enmeshed in a physical gel that provides a secondary boundary that may allow small numbers of bacteria to evade the antibiotic ; therefore , the failure to eliminate the entire population can allow the population to regrow . it should be noted that populations of planktonic bacteria also contain these highly tolerant or persister cells @xcite . thus understanding the process of persister formation and the response of the population to biocidal application is fundamental to developing dosing protocols and treatments in both batch culture and biofilm populations . as in many areas of biology
, mathematical modeling has been used as a counterpart to experimental observations . because there are several hypotheses regarding the mechanism of persister formation
, mathematical modeling can be used to provide insight into the success of failure of treatment protocols as well as the consistency of various hypotheses .
currently , there are at least two distinct hypotheses concerning persister formation - one of which is senescence . in this case ,
persister cells are assumed to be those that have undergone many division cycles .
it is known that asymmetric division leads to degradation of parts of the cellular machinery that may be the underlying cause of persistence @xcite .
mathematical analysis of a model of senescence has been described in both chemostat and biofilm settings @xcite .
a different hypothesis argues that persisters are a phenotype that is expressed at a rate that depends on the external environment .
although the biological details are not well understood , it is thought that this might be due to toxin / antitoxin interaction or some other stress response @xcite .
this has been investigated mathematically as well @xcite . in @xcite , a very simple model of persister formation
was developed and optimal dosing protocols , that entail alternating application and resting , were described . in @xcite ,
toxin / antitoxin interaction was explicitly included and the resulting model was analyzed in a chemostat . in both of these investigations
only one particular form of the kinetics was analyzed .
one of the goals of the current investigation is to extend these results to a more general form for kinetics .
this is an important process , because it has been shown that not only is there a successful dosing protocol , there is an optimal protocol . without knowing how robust the model is
, there is no way to determine how robust the conclusions are .
here we take a model of the form proposed in @xcite and give a condition that yields successful dosing protocols , at least for some initial makeups of the population .
the manuscript is organized as follows : we begin by describing the model for the dynamics of the bacterial population in response to antibiotic challenge .
we then develop the theory by analyzing two extreme cases ( no dosing and constant dosing ) and the intermediate case .
this leads to a local sufficient condition for treatment success .
next we give a condition for global treatment failure , supported by numerical simulations of the model .
we also show that the speed of eradication does not necessarily depend monotonically on the duration of the administration of the antibiotic .
consider the following chemostat model : @xmath0b_s+k_g(t)b_p\label{s1}\\ { \dot b_p}&=&k_l(t)f(s)b_s-[k_g(t)+d]b_p\label{s2}\\ { \dot s}&=&d(s^0-s)-\frac{f(s)b_s}{y}\label{s3}\end{aligned}\ ] ] where @xmath1 is the concentration of the cells which are susceptible to antibiotics , @xmath2 is the concentration of the persister cells which are not affected by the antibiotic and @xmath3 is the concentration of the nutrient .
this model deviates from the one in @xcite because it is a chemostat model , which is reflected in the additional loss terms at rate @xmath4 ( called the dilution rate or washout rate ) , and the inflow ( at the same rate @xmath4 ) of nutrient with an input concentration @xmath5 .
the per capita growth rate of the susceptible cells is denoted by @xmath6 , for which we assume the following throughout the rest of the paper : @xmath7 the persister cells do not consume nutrient , hence the lack of a corresponding growth term in the @xmath2-equation .
the conversion of nutrient into new biomass occurs with a yield of @xmath8 .
the remaining functions @xmath9 and @xmath10 are non - negative , time - varying functions which describe the effect of antibiotics on the population .
first , @xmath11 is the killing rate of the susceptible population .
note in particular that the killing rate is proportional to the growth rate of the cells .
it is positive when both antibiotic and nutrient are present , but zero when either one is missing .
secondly , @xmath12 is the rate at which susceptible cells switch to persister cells .
it is also positive when antibiotic is present , but zero when this is not the case .
finally , @xmath10 is the rate at which persister cells revert to the susceptible state when antibiotic is absent ( and zero when antibiotic is present ) .
since antibiotics are administered to the reactor vessel in a controlled ( lab ) environment , we will make the simplifying assumption that the functions @xmath9 and @xmath10 are @xmath13-periodic ( for some given @xmath14 ) , and of the bang - bang type with simultaneous switching instances : for some @xmath15 $ ] , and for positive parameters @xmath16 and @xmath17 , there holds that @xmath18 thus , antibiotics are present during a fraction @xmath19 of the period @xmath13 , and absent during the remaining fraction @xmath20 of the period .
clearly , this is a simplification of reality because the concentration of an antibiotic is not expected to be of the bang - bang type . in a more realistic model
, the functions @xmath21 and @xmath22 would be replaced by functions depending on ( at least ) a new state variable for the concentration of the antibiotic , and the periodicity would arise through a periodic forcing term in the equation for this new variable .
we leave the study of such a model to the future . throughout the rest of this paper
we assume that the net effect of the antibiotic alone is lethal to the susceptible population : @xmath23 note that this assumption is valid for the parameter values related to the experiments described in @xcite .
the main purpose of this paper is to investigate how the behavior of system @xmath24 with @xmath25 changes qualitatively in terms of @xmath19 .
in this section we collect a couple of basic results concerning the dynamical behavior of @xmath24 with @xmath25 .
for a real - valued function @xmath26 , we denote the extended real numbers @xmath27 and @xmath28 by @xmath29 and @xmath30 respectively .
[ dissipative ] system @xmath24 with @xmath25 has @xmath31 as a forward invariant set , and it is dissipative .
the first assertion is obvious .
dissipativity follows by considering the dynamics of @xmath32 given by @xmath33 and hence @xmath34 not only are all state components of every solution ultimately bounded from above by some constant which does not depend on initial conditions , we also notice in the following lemma , that @xmath35 is ultimately bounded from below by some positive constant which is independent of initial conditions .
[ lower ] there is a constant @xmath36 such that for all solutions of @xmath24 with @xmath25 , there holds that @xmath37 .
consider the function @xmath38 .
then @xmath39 is increasing with @xmath40 and @xmath41 , hence by the intermediate value theorem , there is a unique @xmath42 such that @xmath43 .
we will show that @xmath37 .
if not , then since @xmath44 by the proof of lemma @xmath45 , it follows from corollary @xmath46 in @xcite -a consequence of the famous fluctuation lemma- applied to @xmath47 that @xmath48\\ & \geq & d(s^0-s_{\infty})-f(s_{\infty})s^0\\ & > & d(s^0-\theta)-f(\theta)s^0,\end{aligned}\ ] ] which contradicts that @xmath43 .
first we study the cases where antibiotic is either present or absent for all times .
our conclusions are the expected ones : when antibiotic is present continuously , all susceptible cells are killed , and consequently the persisters die out as well since they can only survive if susceptible cells become persisters .
when the population is never exposed to antibiotics , then all persisters revert to the susceptible state , and ultimately the population will consist entirely of susceptible cells ( provided the dilution rate is not too high ) . [ extreme ] 1 . case @xmath50 ( * continuous antibiotic dosing * ) .
the steady state @xmath51 of system @xmath24 with @xmath25 is globally asymptotically stable .
2 . case @xmath49 ( * never antibiotic dosing * ) . if @xmath52 , then all solutions @xmath53 with @xmath54 converge to the steady state @xmath55 , where @xmath56 is the unique positive value satisfying @xmath57 , and @xmath58 . 1 . if @xmath50 , then @xmath59 , @xmath60 and @xmath61 for all @xmath62 .
by @xmath63 it follows that for all solutions , @xmath64 , and hence @xmath65 as @xmath66 .
this suggests that we should study the linear limiting system @xmath67 whose solutions clearly converge to @xmath68 .
the conclusion now follows immediately by applying theorem f.1 in @xcite .
2 . if @xmath49 , then @xmath69 and @xmath70 for all @xmath62 .
then the restriction on the initial condition implies that @xmath71 for all @xmath72 .
notice that @xmath73 as @xmath66 , suggesting we should study the limiting system @xmath74b_s\\ { \dot s}&=&d(s^0-s)-\frac{f(s)b_s}{y}\end{aligned}\ ] ] this is the classical chemostat model of a single cell population growing on a single nutrient , described in @xcite : all solutions with @xmath75 converge to @xmath76 under the assumptions of the lemma .
the conclusion follows again by theorem f.1 in @xcite .
in this section we deal with the @xmath13-periodic model @xmath24 with @xmath25 , assuming that @xmath78 .
we will also assume that the dilution rate is not too large , as in lemma @xmath79 : @xmath80 notice that system @xmath24 with @xmath25 has a steady state @xmath81 , regardless of the value of @xmath19 . to determine its stability properties , we determine the @xmath13-periodic variational equation at @xmath82 : @xmath83 and the floquet multipliers of this system are @xmath84 ( which is of course inside the unit circle of the complex plane ) and the floquet multipliers of @xmath85 using @xmath25 , these floquet multipliers are the eigenvalues of the following matrix : @xmath86 where @xmath87 are triangular , quasimonotone matrices .
notice also that @xmath88 is hurwitz by @xmath63 , and that @xmath89 has one negative and one positive eigenvalue by @xmath90 .
since @xmath88 and @xmath89 are quasi - monotone ( i.e. their off - diagonal entries are all non - negative ) , it follows that their matrix exponentials are ( entry - wise ) non - negative , triangular matrices and then their product @xmath91 is a ( entry - wise ) positive matrix whose spectral radius @xmath92 is an eigenvalue by the perron - frobenius theorem@xcite .
consequently , to determine stability of @xmath82 , we need to establish whether or not @xmath92 is inside the unit circle : if @xmath93 , then @xmath82 is locally asymptotically stable .
if @xmath94 , then @xmath82 is unstable .
summarizing , we have established [ loc - stab ] let @xmath77 , and assume that @xmath90 holds .
then the steady state @xmath95 is locally stable for @xmath24 with @xmath25 if @xmath93 , but unstable if @xmath94 .
our main concern is knowing how @xmath92 varies as a continuous function of @xmath19 ( this variation is continuous since eigenvalues of a matrix are continuous functions of its entries , and clearly the entries of @xmath91 are continuous in @xmath19 ) .
for @xmath49 ( never using antibiotic ) , and hence also for @xmath19 near @xmath96 by continuity of @xmath92 , we have that @xmath97 .
this is in accordance with lemma @xmath79 , where it was shown that all solutions with @xmath54 converge to @xmath98 , and thus @xmath95 must be unstable . for @xmath50
( using antibiotic continuously ) , and hence also for @xmath19 near @xmath99 , we have that @xmath100 .
this is in accordance with lemma @xmath79 as well because it was shown there that all solutions converge to @xmath82 in this case .
we can actually determine the dependence of @xmath92 on @xmath19 explicitely because fortunately , both @xmath88 and @xmath89 are diagonalizable ( their eigenvalues are distinct ) , which simplifies the computation of their matrix exponentials somewhat .
it is easily verified that @xmath101 where @xmath102 and @xmath103 and thus using @xmath104 that @xmath105 a lengthy algebraic calculation shows that the positive matrix @xmath106 is given by @xmath107}\\ & & + k_gk_l\left({\operatorname{e}}^{(1-p)\tau ( f(s^0)-d)}- { \operatorname{e}}^{-(1-p)\tau ( k_g+d ) } \right)\left({\operatorname{e}}^{-p\tau d}-{\operatorname{e}}^{p\tau [ ( 1-k_d - k_l)f(s^0)-d ] } \right),\end{aligned}\ ] ] @xmath108 @xmath109 } \right)\ ] ] and @xmath110 where @xmath111 is a positive constant , independent of @xmath19 . since @xmath91 is a positive matrix , its spectral radius
can now be calculated explicitly in terms of its entries : @xmath112
in this section we show that the spectral radius @xmath92 also plays a key role in the global behavior of system @xmath24 with @xmath25 and @xmath77 . we will show that if @xmath94 , then not only is @xmath82 unstable as we have shown in theorem @xmath113 , but treatment fails globally , because both cell populations persist uniformly .
in addition we will show that there are positive periodic solutions .
[ main - persist ] let @xmath77 , and assume that @xmath90 holds .
if @xmath94 , then treatment fails and the population is uniformly persistent , i.e. there is some @xmath114 ( independent of initial conditions ) , such that all solutions of @xmath24 with @xmath25 and @xmath75 , have the property that : @xmath115 moreover , there are @xmath13-periodic solutions @xmath116 with @xmath117 for all @xmath62
. define the following matrix : @xmath118 where @xmath119 notice that @xmath120 , and thus since @xmath94 , it follows that @xmath121 as well , because the spectral radius of any matrix is continuous with respect to its entries .
we fix some @xmath122 such that @xmath123 holds .
we will first show that @xmath1 is uniformly weakly persistent , i.e. that there is some @xmath124 such that if @xmath75 , then @xmath125 . by contradiction , if @xmath1 is not uniformly weakly persistent , then there is some solution @xmath53 with @xmath75 such that @xmath126 by corollary @xmath46 in @xcite applied to @xmath127 , and since @xmath128 by @xmath127 , we have that @xmath129\\ & \geq&d(s^0-s_{\infty})- \frac{f(s^0)b_s^{\infty}}{y},\\\end{aligned}\ ] ] and hence by @xmath130 that @xmath131 thus , for some @xmath132 , there holds that @xmath133 for all @xmath134 .
it follows from @xmath135 , that for all @xmath134 : @xmath136 where the vector inequalities should be interpreted componentwise .
notice that the vector field on the right - hand side of @xmath137 is that of a @xmath13-periodic , cooperative linear system whose principal fundamental matrix solution evaluated over one period @xmath13 equals @xmath138 . by kamke s comparison theorem ( see e.g. theorem b.1 in appendix b of @xcite ) it follows that for all @xmath134 , the vector @xmath139 is not smaller ( component - wise ) than the solution starting in @xmath140 of the @xmath13-periodic , cooperative linear system with vector field given in the right - hand side of @xmath137 . but all non - zero , non - negative solutions of the linear system diverge because @xmath141 . then so does @xmath142 , and this contradicts @xmath130 .
we have thus shown that @xmath1 is uniformly weakly persistent .
next we establish that @xmath1 is in fact uniformly strongly persistent .
this follows from theorem @xmath143 in @xcite , applied to the map @xmath144 which maps @xmath145 to @xmath146 , where @xmath147 , @xmath148 and @xmath149 .
the map @xmath144 is continuous and maps @xmath150 into itself , and it has a global attractor because it is compact and dissipative .
it follows that there is some @xmath151 , independent of initial conditions , such that if @xmath75 , then @xmath152 , and also that @xmath153 by theorem @xmath154 in @xcite .
next we show that uniform strong persistence of @xmath1 , implies uniform strong persistence of @xmath2 .
consider equation @xmath47 and notice that for all sufficiently large @xmath62 : @xmath155 where @xmath156 is the positive constant from lemma @xmath157 .
it is not hard to show that the linear equation @xmath158 has a positive @xmath13-periodic solution @xmath159 and that all non - negative solutions converge to it .
therefore , it follows that for all sufficiently large @xmath62 , @xmath160 establishing uniform strong persistence for @xmath2 , since @xmath161 is independent of initial conditions .
we conclude the proof of uniform strong persistence of @xmath1 and @xmath2 by setting @xmath162 . finally , to show that there are @xmath13-periodic solutions with @xmath163 , we apply theorem @xmath164 from @xcite applied to the continuous map @xmath144 defined above . we have already remarked that this map is continuous , maps @xmath150 into itself , is dissipative and compact , and we have just proved that it is uniformly strongly persistent with respect to @xmath165 .
observe also that @xmath150 is relatively open in @xmath166 , and that @xmath150 is convex .
then by theorem @xmath164 from @xcite , the map @xmath144 has a fixed point in @xmath150 , and this in turn implies the existence of @xmath13-periodic solutions with @xmath71 for @xmath24 . the same argument as the one used above to establish uniform strong persistence of @xmath167 , shows that these @xmath13-periodic solutions are such that @xmath168 as well .
we use the following numerical values : the per capita growth rate is of michaelis - menten type : @xmath169 and the parameters @xmath170 the chemostat setting requires that we specify two additional parameters : @xmath171 it can be easily verified that these choices satisfy the conditions @xmath63 and @xmath90 .
note that there is no need to specify the yield coefficient @xmath172 in order to calculate @xmath92 .
indeed , the matrix @xmath91 in @xmath104 does not depend on @xmath172 .
the graph of the spectral radius @xmath92 in terms of @xmath19 , determined using formula @xmath173 , is given in figure @xmath174 .
we see that @xmath175 when @xmath19 is approximately equal to @xmath176 . clearly , @xmath92 is not monotone , and it has a global mimimum of approximately @xmath177 which is achieved at @xmath178 ( determined numerically using mathematica ) .
thus , the optimal strategy in a dosing experiment with a period of @xmath179 hours occurs for a dosing duration of @xmath180 hours .
here , optimality means that eradication happens as quickly as possible . for @xmath181 $ ] ( @xmath182 , other parameters in text.),width=453 ] let us also illustrate what happens if @xmath19 equals approximately @xmath183 .
then @xmath92 equals approximately @xmath184 , implying that treatment fails .
it appears that the solution of @xmath24 with @xmath185 ( and all other parameters as above ) starting from the initial condition @xmath186 converges to a @xmath13-periodic solution , see figures @xmath187 and @xmath188 .
these observations are in accordance with theorem @xmath189 . [ cols="^,^ " , ] finally we remark that the spectral radius of @xmath91 may be monotone : if we change the value of @xmath190 from @xmath191 to @xmath99 , and leave all other parameters unchanged , then @xmath92 is a decreasing function of @xmath181 $ ] , and achieves its minimum at @xmath50 .
this indicates that for this case , the optimal strategy is to use antibiotics continuously , see figure @xmath192 . for @xmath181 $ ] ( @xmath193 , other parameters in text).,width=453 ]
bacterial infections are a source of problems in a wide variety of situations including industrial , environmental and clinical settings .
growing understanding of the inability of antibiotics and biocides to treat these infections has driven investigations into the cause of the failure of treatments .
it is becoming increasingly evident that persister cells must play an important role in protecting populations of bacteria .
it has been observed that other protective mechanisms including physiological and physical processes are not sufficient to explain the observed failures @xcite . moreover , because it is very difficult to investigate and classify persister cells experimentally , mathematical modeling can play an important role in supporting hypotheses as well as generating useful predictions .
we have described and analyzed a general model for the dynamics of persister formation in response to antibiotic challenge .
we have been able to provide a condition for the success / failure of antibiotic challenge in a chemostat .
we have found that there is an optimal strategy , that is one that kills the bacteria the fastest .
these results indicate that periodic dosing is an effective treatment protocol for a variety of bacteria , substantially strengthening the results in @xcite .
the theoretical results were also confirmed by direct numerical simulations . , _ increasing resistance of planktonic and biofilm cultures of _ burkholderia cepeciax to ciproflaxacin and ceftazidime during exponential growth _ _ , journal of antimicrobial chemotherapy , 42 ( 1998 ) , pp .
|
the tolerance of bacterial populations to biocidal or antibiotic treatment has been well documented in both biofilm and planktonic settings . however , there is still very little known about the mechanisms that produce this tolerance .
evidence that small , non - mutant subpopulations of bacteria are not affected by antibiotic challenge has been accumulating and provides an attractive explanation for the failure of typical dosing protocols .
although a dosing challenge can kill all the susceptible bacteria , the remaining persister cells can serve as a source of population regrowth .
we give a robust condition for the failure of a periodic dosing protocol for a general chemostat model , which supports the mathematical conclusions and simulations of an earlier , more specialized batch model .
our condition implies that the treatment protocol fails globally , in the sense that a mixed bacterial population will ultimately persist above a level that is independent of the initial composition of the population .
we also give a sufficient condition for treatment success , at least for initial population compositions near the steady state of interest , corresponding to bacterial washout .
finally , we investigate how the speed at which the bacteria are wiped out depends on the duration of administration of the antibiotic .
we find that this dependence is not necessarily monotone , implying that optimal dosing does not necessarily correspond to continuous administration of the antibiotic .
thus , genuine periodic protocols can be more advantageous in treating a wide variety of bacterial infections . * keywords : persister , biofilm , model , chemostat , tolerance *
|
the measurement of ultra high energy cosmic rays ( uhecr ) flux is the goal of a wide class of past , present and future detectors @xcite-@xcite .
uhe neutrinos are expected to be produced by the interaction of hadronic matter with the surrounding radiation / matter . a search for this signal is currently performed by several neutrino telescopes @xcite-@xcite .
neutrinos with energy above @xmath6 ev are expected to originate from the interaction of uhe cosmic rays with the cosmic microwave background ( cmb ) _ via _ the @xmath7-photoproduction , @xmath8 , the so - called _ cosmogenic neutrinos _
the prediction for such a flux is however affected by several uncertain physical quantities , namely the spatial distribution of astrophysical sources , the ejected proton fluxes ( if proton ) and the way of modelling the diffuse extragalactic electromagnetic background in the different frequency regions .
one can assume a reasonable ansatz for all these quantities combined with the measurement of the diffuse photon flux in the gev region by egret @xcite , and the agasa / hires data .
these models and the predictions of more exotic scenarios have been exhaustively discussed in several papers ( see for example ref.s @xcite ) .
high energy neutrinos are hardly detected , as they are almost completely shadowed by earth and rarely interact with the atmosphere .
an eev neutrino has an interaction length of the order of 500 km water equivalent in rock and , even crossing horizontally the atmosphere ( 360 meters water equivalent ) , only one neutrino out of thousand will be interacting . due to the very low expected flux and the small neutrino - nucleon cross section , km@xmath9-neutrino telescopes and giant surface arrays have very few chances of detection . in this framework ,
an interesting strategy for @xmath10 detection is described in ref.s @xcite@xcite . as shown for example in figure 1 of ref .
@xcite , for energy between @xmath11 and @xmath12 ev the @xmath13 decay length is not much larger than the corresponding interaction range .
thus , an energetic @xmath13 , produced by charged current ( cc ) @xmath10 interaction not too deep under the surface of the earth , has a chance to emerge in the atmosphere as an upgoing particle . unlike @xmath13 s , muons crossing the rock rapidly loose energy and decay .
almost horizontal @xmath10 , just skimming the earth surface , will cross an amount of rock of the order of their interaction length and thus will be able to produce a corresponding @xmath13 , which might shower in the atmosphere and be detected . in order to estimate the number of upgoing @xmath13 expected in the pierre auger observatory ( pao )
, one needs to know the value of the neutrino - nucleon cross section for cc interaction , @xmath14 .
the aim of this paper is to estimate the number of possible upgoing @xmath13 showers which the fluorescence detector ( fd ) of pao could detect .
the predictions are analyzed with respect to their dependence on different neutrino fluxes and by using a new estimate of @xmath15 cross section .
the paper is organized as follows . in section [ sec : nuflux ] different models for neutrino flux predictions are discussed . in section
[ sec : nucross ] , the general features of deep inelastic neutrino cross sections are outlined for neutrino energies up to 10@xmath16 ev . in section [ sec : tauloss ] , high energy @xmath13 propagation through matter is illustrated by considering all the relevant interaction mechanisms .
average values for @xmath13 energy loss are provided by taking into account the recent calculations of photonuclear interaction given in ref .
@xcite . in section [ sec : earthskimming ] the number of expected upgoing @xmath13 showers is derived and discussed .
finally in section [ sec : conclusions ] , we give our conclusions and remarks .
ultra high energy protons , with energy above @xmath17 ev , travelling through the universe mostly loose their energy _ via _ the interaction with cmb radiation .
the large amount of charged and neutral pions produced will eventually decay in charged leptons , neutrinos ( _ cosmogenic neutrinos _
@xcite ) , and high energy gamma rays . at the gev energy range
the extragalactic diffuse gamma - ray background was measured by the egret experiment @xcite .
this measurement provides an upper bound for possible neutrino fluxes from pion production .
in particular , it gives the expected maximum flux of cosmogenic neutrinos from an initial spectrum of measured uhe protons @xcite .
it is worth noticing that , since at least part of uhecr are protons , the existence of cosmogenic neutrinos is guaranteed , even if their flux is very uncertain . in figure [ nu_gzk ] the gzk neutrino flux for three possible scenarios is plotted .
the thick solid line gives the case of an initial proton flux @xmath18 , by assuming in addition that the egret flux is entirely due to @xmath7-photoproduction ( gzk - h ) .
the thin solid line shows the neutrino flux when the associated photons contribute only up to 20% in the egret flux ( gzk - l ) .
the dashed line stands for the conservative scenario of an initial proton flux @xmath19 ( gzk - wb ) . in this case
the neutrino flux is compatible with the so called waxman - bahcall limit @xcite . note
that no lower bounds can be set for the cosmogenic neutrino flux .
in particular , in the most conservative but rather unrealistic case , the astrophysical sources can not accelerate protons up to energies above gzk cutoff , and thus the secondary neutrinos will be produced in negligible quantities .
all neutrino fluxes presented in this section were calculated by a propagation code @xcite which takes into account the neutrino production via @xmath7-photoproduction with microwave , infrared , optical and radio photon backgrounds , as well as in neutron decay .
most of the models trying to explain highest energy cosmic rays ( @xmath20 ev ) in terms of exotic particles , predict a large associated flux of neutrinos . in figure [ nu_exotic ] , the expected neutrino flux for two of such scenarios
is plotted .
one of them is the model of _ new hadrons _
( nh ) @xcite , with mass @xmath21 gev , capable of generating uhecr events above gzk cutoff . in susy theories , for example , the new hadrons are bound states of light bottom squarks or gluinos and , once produced in suitable astrophysical environments , can reach the earth without significant energy losses . in spite
the production of new hadrons is a subdominant process , it generates a large number of neutrinos ( see figure [ nu_exotic ] ) .
the dashed line in figure [ nu_exotic ] shows the neutrino flux for a topological defects model ( td ) ( for a review see @xcite ) . in this case
uhecr events with energy @xmath20 ev are explained in terms of @xmath22 s which are produced in the decay of heavy particles with mass of the order of @xmath23 ev . as in the previous case ,
the associated neutrino flux for this kind of models is extremely large .
we do not discuss here the so called z - burst scenarios , which attempts to explain uhecr as products of z - boson decay .
these models are in fact strongly disfavored @xcite by the upper bounds on uhe neutrino flux put by forte @xcite and glue @xcite and by the cosmological limits on neutrino mass set by wmap and lss data @xcite . in the following sections we will estimate the sensitivity of pao to the uhe tau neutrino flux both in the case of cosmogenic neutrinos ( figure [ nu_gzk ] ) and of exotic models ( figure [ nu_exotic ] ) .
at energy above 1 gev neutrino - atoms interaction is dominated by the process of _ deep inelastic scattering _ ( dis ) on nucleons , since the contributions of both elastic and quasi - elastic interactions become negligible .
the effect of the neutrino scattering with atomic electrons will not be taken into account here , since the cross section for this process is , at each energy , about three orders of magnitude lower than the neutrino - nucleon cross section production , occurring at @xmath24 pev , whose contribution to the total event rate remains nevertheless negligible @xcite . ] .
detectable leptons are produced through charged current interaction , @xmath25 whereas neutral current ( nc ) interaction causes a modulation in the spectrum of the interacting neutrinos , @xmath26 these total cross sections can be written in terms of differential ones as follows @xmath27 where @xmath28 is the energy of the incoming neutrino , @xmath29 is the mass of the outgoing charged lepton and @xmath30 is the inelasticity parameter , defined as @xmath31 with @xmath32 the energy of the outgoing charged ( for cc ) or neutral ( for nc ) lepton .
.average inelasticity parameter for cc ( @xmath33 ) and nc ( @xmath34 ) interaction for different incoming neutrino energy .
[ cols="^,^,^",options="header " , ]
ultra high energy @xmath10 s ( @xmath35 ) could have real detection chance at giant surface apparatus like pao .
almost horizontal tau neutrinos crossing distance in the earth of the order of their interaction length might produce @xmath13-shower earth skimming events potentially detectable by the auger fd . in this paper for a representative sample of @xmath10 fluxes either produced as _ cosmogenic neutrinos _ or in more exotic scenarios , the number of @xmath13-shower earth skimming events per year expected at the fd of the pierre auger observatory has been computed . for this calculation
we have used the cteq6 @xcite parton distribution functions in the dis factorization scheme and recent estimate of radiative @xmath13 energy losses in matter .
a decrease of @xmath36 allows @xmath13 to emerge with a smaller nadir angle ( less horizontal ) and thus increases the number of earth skimming events . in the relevant region of @xmath37
, we essentially have @xmath38 ; thus a factor @xmath39 in the neutrino - nucleon cross section leads to an almost double number of events at fd . as a further remark
, we point out that the level of theoretical uncertainties on cross - sections at extremely high energy is very large ( like in case of new physics above the tev energy region ) , due to the poor experimental knowledge of parton density functions for very small @xmath40 .
the number of events per year at the fd of the pierre auger observatory is presented in table [ table::events ] for the different neutrino models reported in figure [ nu_gzk ] and [ nu_exotic ] . for a five years pao detection campaign ( south plus north )
we essentially have to scale these numbers by a factor ten and thus at least the exotic models could produce detectable events . however , it is worth noticing that this estimate has been performed under a rather conservative assumption on the possibility of @xmath13 showering on the apparatus : we have used an average value for the parameter @xmath41 which essentially allows only one particle out of three to decay inside the detection fiducial volume . in a paper in progress
a more sophisticated simulation is going to be performed by taking into account the real morphology of the pao site .
this analysis should slightly increase the numbers presented in table [ table::events ] , which however give a reasonable estimate of the number of events at fd .
we would like to thank i. a. sokalski and e. bugaev for the extremely helpful suggestions on the matter of @xmath13 photonuclear interaction .
the authors would also like to thank m. ambrosio , f. guarino and s. pastor for valuable comments .
work of d.s . supported in part by nasa atp grant nag5 - 13399 .
efimov et al . , proc .
international symposium on _ astrophysical aspects of the most energetic cosmic rays _ , eds . m. nagano and f. takahara ( world scientific singapore , 1991 ) p.20 ; b.n .
afnasiev , proc . of the international symposium on
_ extremely high energy cosmic rays : astrophysics and future observatories _ , ed .
m. nagano ( institute for cosmic ray research , tokyo , 1996 ) , p.32 . v. s. beresinsky and g. t. zatsepin , phys . lett .
b * 28 * , 423 ( 1969 ) ; v. s. berezinsky and g. t. zatsepin , sov . j. nucl .
11 ( 1970 )
111 [ yad . fiz .
11 ( 1970 ) 200 ] .
p. sreekumar _ et al .
_ , astrophys .
j. * 494 * , 523 ( 1998 ) [ astro - ph/9709257 ] ; a. w. strong , i. v. moskalenko and o. reimer , astro - ph/0306345 ; astro - ph/0405441 .
o. e. kalashev , v. a. kuzmin , d. v. semikoz and g. sigl , phys .
d * 66 * , 063004 ( 2002 ) .
e. waxman and j. n. bahcall , phys .
d * 59 * , 023002 ( 1999 ) [ hep - ph/9807282 ] .
o. e. kalashev , v. a. kuzmin and d. v. semikoz , astro - ph/9911035 ; mod .
a * 16 * , 2505 ( 2001 ) [ astro - ph/0006349 ] .
m. kachelriess , d. v. semikoz and m. a. tortola , phys .
d * 68 * , 043005 ( 2003 ) [ hep - ph/0302161 ] .
p. bhattacharjee and g. sigl , phys .
rept . * 327 * , 109 ( 2000 ) [ astro - ph/9811011 ] .
n. g. lehtinen , p. w. gorham , a. r. jacobson and r. a. roussel - dupre , phys .
d * 69 * , 013008 ( 2004 ) [ astro - ph/0309656 ] .
p. w. gorham , c. l. hebert , k. m. liewer ,
c. j. naudet , d. saltzberg and d. williams , astro - ph/0310232 .
s. hannestad , jcap * 0305 * , 004 ( 2003 ) [ arxiv : astro - ph/0303076 ] .
p. crotty , j. lesgourgues and s. pastor , phys .
d * 69 * ( 2004 ) 123007 [ arxiv : hep - ph/0402049 ] .
g. l. fogli , e. lisi , a. marrone , a. melchiorri , a. palazzo , p. serra and j. silk , arxiv : hep - ph/0408045 .
r. gandhi , c. quigg , m. reno , i. sarcevic , phys .
d * 58 * , 093009 ( 1998 ) .
v. n. gribov and l. n. lipatov , yad .
fiz . * 15 * , 1218 ( 1972 ) [ sov .
j. nucl .
* 15 * , 675 ( 1972 ) ] .
l. n. lipatov , sov .
j. nucl .
phys . * 20 * , 94 ( 1975 ) [ yad .
fiz . * 20 * , 181 ( 1974 ) ] .
y. l. dokshitzer , sov .
jetp * 46 * , 641 ( 1977 ) [ zh . eksp .
1216 ( 1977 ) ] . g. altarelli and g. parisi , nucl .
b * 126 * , 298 ( 1977 ) .
s. i. dutta , m. h. reno , i. sarcevic , phys .
d * 63 * , 094020 ( 2001 ) .
|
ultra high energy neutrinos are produced by the interaction of hadronic cosmic rays with the cosmic radiation background .
more exotic scenarios like _ topological defects _ or _ new hadrons _ predict even larger fluxes .
in particular , earth skimming tau neutrinos could be detected by the fluorescence detector ( fd ) of pierre auger observatory .
a detailed evaluation of the expected number of events has been performed for a wide class of neutrino flux models .
an updated computation of the neutrino nucleon cross section and of the tau energy losses has been carried out . for the most optimistic theoretical models , about one earth skimming neutrino event is expected in several years at fd .
* earth skimming uhe tau neutrinos at the fluorescence detector of pierre auger observatory * * c. aramo*@xmath0 , * a. insolia*@xmath1 , * a. leonardi*@xmath1 , * g. miele*@xmath0 , * l. perrone*@xmath2 , * o. pisanti*@xmath0 , * d.v .
semikoz*@xmath3 _ @xmath0 dipartimento di scienze fisiche , universit di napoli `` federico ii '' and istituto nazionale di fisica nucleare sezione di napoli , complesso universitario di monte s. angelo , via cinthia , i-80126 napoli , italy .
+ @xmath1 dipartimento di fisica e astronomia , universit di catania and istituto nazionale di fisica nucleare sezione di catania , via s. sofia 64 , i-95123 catania , italy .
+ @xmath2 fachbereich c , sektion physik , universitt wuppertal , d-42097 wuppertal , germany .
+ @xmath4 department of physics and astronomy , ucla , los angeles , ca 90095 - 1547 usa .
+ @xmath5 inr ras , 60th october anniversary prospect 7a , 117312 moscow , russia . _ _ pacs numbers : 95.85.ry , 13.15.+g , 96.40.tv , 95.55.vj , 13.35.dx;_ = .8 cm
|
the yang mills gradient flow or the wilson flow @xcite has attracted much attention in recent years in the context of lattice gauge theory .
its known applications include , scale setting @xcite , definition of the topological charge @xcite , definition of non - perturbative gauge coupling @xcite , chiral condensation @xcite , improvement of step scaling @xcite , etc .
even its application to supersymmetric theory @xcite and to the operator product expansion @xcite is considered .
reference @xcite is a review of this notion and further related works can be found in a review @xcite and in a most recent paper on the non - perturbative beta function @xcite . a crucial property of the yang
mills gradient flow , underlying the above applications , is its `` ultraviolet ( uv ) finiteness '' @xcite . the gradient flow is a one - parameter ( called the flow - time ) evolution of the gauge field , according to a `` heat diffusion equation '' ( called the flow equation ) . a remarkable fact that can be rigorously proven @xcite in all orders of perturbation theory
is that any correlation function of the evolved ( or flowed ) gauge field becomes uv finite _ without _ the wave function renormalization , as long as the parameters of the theory are properly renormalized .
moreover , _ any _ local product of the flowed gauge field remains uv finite without further ( multiplicative as well as subtractive ) renormalization .
this remarkable property of the gradient flow facilitates , in particular , the construction of renormalized composite operators of the gauge field .
that is , any simple product of the flowed ( bare ) gauge field as it stands is a renormalized ( i.e. , uv - finite ) quantity . in ref .
@xcite , as possible application of the gradient flow , one of us ( h.s . ) considered the construction of the energy
momentum tensor in lattice gauge theory .
this application of the gradient flow to the energy
momentum tensor was further developed from a somewhat different perspective in ref .
the construction was then generalized to gauge theories including the fermion field @xcite . the genuine energy
momentum tensor can not be defined on the lattice because the lattice structure breaks the translational invariance explicitly .
even the construction of a lattice operator that reduces to the correctly normalized conserved energy
momentum tensor in the continuum limit is quite non - trivial as investigated in refs .
reference @xcite is a pioneering work on this issue .
the basic idea of refs .
@xcite , which uses the uv finiteness of the gradient flow in an essential way , is recapitulated in sect .
[ sec:6 ] of the present paper .
the aim of refs .
@xcite is to construct a lattice operator that automatically reduces to the correctly normalized conserved energy
momentum tensor in the continuum limit .
theoretically , there is only little room for doubt on the reasoning in refs .
practically , however , it is not a priori clear whether presently available lattice parameters are sufficient to extract physical information by using the construction . on this issue , the promising result in ref .
@xcite for thermodynamical quantities in quenched qcd is quite encouraging .
still , it is indispensable to numerically demonstrate the conservation law of the energy
momentum tensor by using lattice monte carlo simulations . under these situations
, it seems useful to study a simpler system that would allow a similar construction of the lattice energy
momentum tensor using the gradient flow .
one of the basic assumptions in refs .
@xcite is that the theory is asymptotically free .
not so many field theories exhibit asymptotic freedom , however .
this was our original motivation for the present study on the gradient flow in the 2d @xmath0 non - linear sigma model @xcite .
it is well known @xcite that the physics of this systems possesses many similarities with the 4d non - abelian gauge theory .
these include asymptotic freedom , dynamical generation of the mass gap , and , for @xmath3 , the topological term and associated @xmath4-parameter .
see also refs .
this system is also advantageous from a computational perspective ( and thus from our original motivation ) , because there exists a very efficient cluster simulation algorithm @xcite .
the state of the art in non - perturbative lattice study of the 2d @xmath0 non - linear sigma model can be found in ref .
@xcite . in the present paper
, we will show that there exists another surprising similarity between the 2d @xmath0 non - linear sigma model and the 4d gauge theory : any correlation function of the flowed @xmath1-vector field in the former becomes uv finite _ without _ the wave function renormalization , as long as the parameters of the theory are renormalized .
this uv finiteness also persists for _ any _ local product of the flowed @xmath1-vector field .
this similarity is surprising , because the uv finiteness of the flowed gauge field is a non - trivial consequence @xcite of the gauge brs symmetry that acts _ non - linearly _ on the gauge field .
in fact , matter fields such as the fermion field transform linearly under the gauge brs symmetry and they _ do _ require wave function renormalisation even after the flow @xcite . in the 2d @xmath0 non - linear sigma model , however , it is not clear at first glance what plays the same role as this gauge brs symmetry in the 4d gauge theory .
our proof clarifies this point . on the other hand , happily , because of the uv finiteness of the gradient flow in the 2d @xmath0 non - linear sigma model , we can repeat the construction of the lattice energy
momentum tensor in refs .
@xcite .
the following describes the organization of the present paper and gives a summary of the contents of each section . in sect .
[ sec:2 ] , we introduce the flow equation in the 2d @xmath0 non - linear sigma model . if one considers the application in lattice numerical simulations , this is the equation that should be solved numerically in conjunction with the conventional monte carlo simulations .
we then formulate the perturbative expansion for the system defined by the combination of the 2d @xmath0 non - linear sigma model and the flow equation ( the flowed system ) . in sect .
[ sec:3 ] , on the basis of the perturbative expansion developed in sect .
[ sec:2 ] , we explicitly compute the two - point function of the flowed bare @xmath1-vector field to the one - loop order .
this explicit calculation shows that the two - point function is made uv finite by the conventional parameter renormalization in the non - linear sigma model @xcite , but without the wave function renormalization .
we carry out the computation in dimensional regularization and in lattice regularization and arrive at the same conclusion .
although this computation is only in the one - loop level , it strongly indicates that the gradient flow in the non - linear sigma model has a similar uv property as the gauge theory . as the proof for the 4d gauge theory in ref .
@xcite and the renormalizability proof in the stochastic quantization @xcite , our proof in all orders of perturbation theory uses a local field theory with one spacetime dimension higher : we use a @xmath2-dimensional field theoretical representation of the flowed system . in sect .
[ sec:4 ] , we define this @xmath2-dimensional local field theory . then we show that the system defined through the flow equation in sect .
[ sec:2 ] and the @xmath2-dimensional field theory have equivalent perturbative expansions .
it is easy to see the rough equivalence .
however , a closer look reveals that there are discrepancies between the two systems ; the measure term in the former is missing in the latter , while the former does not have the flow - line loop diagrams of the latter . presumably , the step to show that these two apparently different elements are actually equivalent ( sect .
[ sec:4.4 ] ) is the hardest part in our argument .
we will find that , to address this very subtle problem in a convincing manner , it is necessary to first discretize the flow - time derivative and then take the continuum limit for this discretization ; this necessity of discretization is also counterintuitive .
once having obtained a local field theory that is ( perturbatively ) equivalent to the flowed system , a possible way to proceed is to write down a ward takahashi relation or a zinn - justin equation @xcite ( see , e.g. , ref .
@xcite ) for the 1pi generating functional , which restricts the possible form of counterterms , on the basis of a certain symmetry in the @xmath2-dimensional system .
this is the content of sect .
[ sec:5 ] . here
, we encounter another surprise : the @xmath2-dimensional field theory possesses _ local _ gauge symmetries , although it does not contain any gauge field .
note that the unique internal symmetry in the original 2d @xmath0 non - linear sigma model is the global @xmath0 symmetry .
because of these gauge symmetries , we have to fix the gauge . even under the gauge fixing , there still remains a residual symmetry that acts _ non - linearly _ on various fields .
we will find that the zinn - justin equation associated with this non - linear symmetry does the job .
then , by listing possible counterterms ( by borrowing the information obtained in sect . [ sec:4.4 ] ) and examining the restriction implied by the zinn - justin equation
, we finally show that the only counterterms required are those of the original 2d @xmath0 non - linear sigma model .
in particular , the flowed @xmath1-vector field ( and its composite operators ) is not renormalized .
this completes our proof for the uv finiteness of the gradient flow . in sect .
[ sec:6 ] , on the basis of the uv finiteness established in sect .
[ sec:5 ] , we construct the energy momentum tensor in a lattice formulation of the non - linear sigma model , following the line of reasoning of refs . @xcite . in summary
, we have found another example in which the gradient flow exhibits a remarkable uv finiteness : in the 2d @xmath0 non - linear sigma model , any correlation function of the flowed @xmath1-vector field and its composite operators is uv finite without multiplicative ( as well as subtractive ) renormalization .
our proof in the present paper also clarifies subtle but very interesting technical issues arising in the theoretical analysis of the gradient flow , such as the necessity of the discretization of the flow - time derivative and the emergence of gauge and/or non - linear symmetries in the corresponding local field theory with one dimension higher .
the knowledge obtained here will be useful in considering the application of the gradient flow to a wider range of systems . also ,
going back to our original motivation , we hope to numerically test the idea of refs .
@xcite by using the energy
momentum tensor constructed in sect .
[ sec:6 ] in the near future .
the 2d @xmath0 non - linear sigma model is a field theory of an @xmath1 component vector with the unit length .
its partition function is given by is used for the functional integral over functions on the @xmath5-dimensional spacetime . ]
@xmath6 \left[\prod_x\delta(n(x)^2 - 1)\right ] \exp\left [ -\frac{1}{2g_0 ^ 2}\int\mathrm{d}^dx\,\sum_{i=1}^n \partial_\mu n^i(x)\partial_\mu n^i(x ) \right ] , \label{eq:(2.1)}\ ] ] where @xmath7 and @xmath8 is the bare coupling constant .
although the spacetime dimension @xmath5 is @xmath9 for our target theory , expressions for generic @xmath5 are useful because we will extensively use dimensional regularization in what follows . in the present paper , as an analogue of the yang
mills gradient flow @xcite , we consider the following @xmath10-evolution of the @xmath1-vector field ( the flow equation ) : @xmath11 where the initial value is given by the @xmath1-vector field in the @xmath0 non - linear sigma model , @xmath12 which is subject to the functional integral . the projection operator @xmath13 in the right - hand side of the flow equation
is defined by @xmath14 ( in eq . and
in what follows , the sum over the repeated index is understood ) .
the projection operator is introduced so that the flow is consistent with the constraint @xmath15 , where @xmath16 , i.e. , @xmath17 .
the latter would be a natural requirement for the flow equation for the @xmath0 non - linear sigma model .
in fact , a flow equation identical to eq . has also been advocated in appendix b of ref .
@xcite from the perspective of the symmetry of the present system . as usual , for the perturbative treatment of the @xmath0 non - linear sigma model
, we parametrize the constraint @xmath18 in eq . in terms of @xmath19 independent components ( the @xmath20-field ) as @xmath21 and
then expand expressions regarding @xmath22 as a small fluctuation . in this perturbative treatment
, the partition function becomes @xmath23 \left[\prod_x\frac{1}{\sqrt{1-\pi(x)^2}}\right ] \notag\\ & \qquad { } \times\exp\left ( -\frac{1}{2g_0 ^ 2}\int\mathrm{d}^dx\ , \left\ { \left[\partial_\mu\pi(x)\right]^2 + \left[\partial_\mu\sqrt{1-\pi(x)^2}\right]^2 \right\ } \right ) .
\label{eq:(2.7)}\end{aligned}\ ] ] the above arbitrary choice of the perturbative branch , eq . with small @xmath22
, however , induces infrared ( ir ) divergences in the perturbative expansion of @xmath0 non - invariant quantities @xcite . to regularize the ir divergences , we introduce the mass term @xmath24 \notag\\ & = \frac{m_0 ^ 2}{g_0 ^ 2}\int\mathrm{d}^dx\ , \left\{\frac{1}{2}\pi(x)^2+\frac{1}{8}\left[\pi(x)^2\right]^2 + \dotsb\right\ } , \label{eq:(2.8)}\end{aligned}\ ] ] and take the massless limit @xmath25 in the very end of the calculation . with this mass term ,
the particular perturbative branch is favored for a weak coupling .
also for the flowed field @xmath26 , since @xmath15 holds along the flow evolution , we set @xmath27 then the perturbative expansion of the flow equation is obtained from the integral representation , @xmath28 , \label{eq:(2.11)}\ ] ] where @xmath29 is the heat kernel , ] ] @xmath30 and @xmath31 .
\label{eq:(2.14)}\ ] ] noting that @xmath32 and @xmath33 , we see that eq .
solves eq . with the initial condition . by iteratively solving eq . in terms of the initial value @xmath34 , therefore , we have a perturbative solution of the flow equation .
this expansion can be represented diagrammatically ( the flow feynman diagram @xcite ) and , throughout this paper , we represent the heat kernel by a double wavy line in fig .
[ fig:1 ] .
this line is also called the `` flow - line propagator '' or simply the `` flow line '' . on the other hand , the combination @xmath35 in eq .
represents the effect of non - linear terms in the flow equation and , in what follows , this interaction will be denoted by an open circle ( the flow vertex ) ; see fig .
[ fig:4 ] for an example . the initial value of the flow , @xmath34 in eq . , is a quantum field subject to the functional integral . from eq . and eq . ( with the mass term )
, one then sees that the quantum free propagator of the flowed field is given by @xmath36 note that in this propagator , the flow times at the end points appear in the sum ( not the difference ) . throughout this paper , this free propagator will be denoted by a single wavy line ( fig .
[ fig:2 ] ) . finally , the functional integral generates interaction vertices among the @xmath37 .
the interaction vertices in the action integral will be denoted by a filled circle ( see fig .
[ fig:3 ] for an example ) . on the other hand , the interaction vertices arising from the functional measure in eq . ,
the `` measure term '' , @xmath38\right\ } , \label{eq:(2.16)}\ ] ] will be represented by a cross as in fig .
[ fig:5 ] .
an explicit one - loop calculation of the correlation functions of the flowed field is quite instructive , because it shows a remarkable uv property of the gradient flow . as the uv regularization
, we first adopt dimensional regularization , setting @xmath39 let us compute the two - point function of the flowed @xmath20-field . the lowest - order ( tree - level ) two - point function
is given by the free propagator in fig .
[ fig:2 ] . in the one - loop level ,
diagram 01 in fig .
[ fig:3 ] , which contains the interaction vertex in the original non - linear sigma model only , gives @xmath40 g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2 } \notag\\ & \qquad { } + \frac{g_0 ^ 2}{4\pi}\frac{n-3}{2 } \left[-\frac{2}{\epsilon } + \ln\left(\frac{\mathrm{e}^{\gamma_\mathrm{e}}m_0 ^ 2}{4\pi } \right)\right ] g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\mathrm{e}^{-(t+s)p^2}\ , \frac{m_0 ^ 2}{(p^2+m_0 ^ 2)^2 } , \label{eq:(3.2)}\end{aligned}\ ] ] where @xmath41 is euler s constant . on the other hand , the contribution of another one - loop diagram , diagram 02 in fig .
[ fig:4 ] , that contains the flow vertex is @xmath42 \notag\\ & \qquad { } \times g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2}. \label{eq:(3.3)}\end{aligned}\ ] ] we note that the measure term in eq . vanishes identically in dimensional regularization with which @xmath43 .
thus , in total , we have @xmath44\biggr\ } \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad{}\times g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2 } \notag\\ & \qquad { } + \frac{g_0 ^ 2}{4\pi}\frac{n-3}{2 } \left[-\frac{2}{\epsilon } + \ln\left(\frac{\mathrm{e}^{\gamma_\mathrm{e}}m_0 ^ 2}{4\pi } \right)\right ] g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\mathrm{e}^{-(t+s)p^2}\ , \frac{m_0 ^ 2}{(p^2+m_0 ^ 2)^2}+o(g_0 ^ 4 ) . \label{eq:(3.4)}\end{aligned}\ ] ] now , the parameter renormalization in the original @xmath0 non - linear sigma model with the mass term is known to be ( in the minimal subtraction ( ms ) scheme ) @xmath45 and @xmath46m^2,\qquad z_3=1-\frac{g^2}{4\pi}2(n-1)\frac{1}{\epsilon}+o(g^4 ) , \label{eq:(3.6)}\ ] ] where @xmath47 is the wave function renormalization factor for the unflowed @xmath20-field , @xmath48 .
, as a byproduct of our analysis , we will have a proof for these renormalization rules . ]
if we make these substitutions in eqs . , we obtain the following completely uv - finite expression : @xmath49\biggr\ } \notag\\ & \qquad\qquad\qquad\qquad{}\times g^2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+z_mm^2}+o(g^4 ) , \label{eq:(3.7)}\end{aligned}\ ] ] where @xmath50 remarkably , when expressed in terms of renormalized parameters , the two - point function of the flowed @xmath20-field is uv finite _ without multiplicative wave function renormalization_. this uv finiteness of the flowed field is similar to that of the 4d gauge field flowed by the yang
mills gradient flow , a property first observed in ref .
@xcite in lower - order perturbative computations and then proven in all orders of perturbation theory in ref .
the above result indicates that by a similar mechanism to the 4d gauge theory , the @xmath1-vector field flowed to positive flow times is uv finite only with parameter renormalization .
it is also instructive to repeat the above calculation by using lattice regularization instead of dimensional regularization .
we adopt the prescription that in eq .
@xmath51 , where @xmath52 denotes the lattice spacing , and the derivative @xmath53 is replaced by the forward difference operator .
the laplacian in the flow equation is replaced by @xmath54 , where @xmath53 and @xmath55 are the forward and backward difference operators , respectively . then the contribution of fig .
[ fig:3 ] is @xmath56 g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2 } \notag\\ & \qquad { } + \frac{g_0 ^ 2}{4\pi } \left\{\frac{n-3}{2 } \left[\ln(am_0)^2 - 5\ln2\right ] -\pi\right\ } g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\mathrm{e}^{-(t+s)p^2}\ , \frac{m_0 ^ 2}{(p^2+m_0 ^ 2)^2 } \notag\\ & \qquad\qquad { } + \frac{g_0 ^ 2}{4\pi}\left(-\frac{4\pi}{a^2}\right ) g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{(p^2+m_0 ^ 2)^2 } , \label{eq:(3.9)}\end{aligned}\ ] ] which is quadratically divergent .
the quadratic divergence in the last term is canceled by the measure term with @xmath57 for lattice regularization .
in fact , the contribution of the measure term to the two - point function ( fig . [ fig:5 ] ) is @xmath58 on the other hand , the contribution of fig .
[ fig:4 ] is @xmath59 \notag\\ & \qquad { } \times g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ ,
\frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2}. \label{eq:(3.11)}\end{aligned}\ ] ] thus , we have in total @xmath60+\pi \notag\\ & \qquad\qquad\qquad { } + \frac{1}{2}(n-1)\ln(2\mathrm{e}^{\gamma_{\mathrm{e}}}m_0 ^ 2 t ) + \frac{1}{2}(n-1)\ln(2\mathrm{e}^{\gamma_{\mathrm{e}}}m_0 ^ 2s ) \biggr]\biggr\ } g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\ , \frac{\mathrm{e}^{-(t+s)p^2}}{p^2+m_0 ^ 2 } \notag\\ & \qquad { } + \frac{g_0 ^ 2}{4\pi } \left\{\frac{n-3}{2}\left[\ln(am_0)^2 - 5\ln2\right]-\pi\right\ } g_0 ^ 2\delta^{kl}\int_p\mathrm{e}^{ip(x - y)}\mathrm{e}^{-(t+s)p^2}\ , \frac{m_0 ^ 2}{(p^2+m_0 ^ 2)^2}+o(g_0 ^ 4 ) .
\label{eq:(3.12)}\end{aligned}\ ] ] it is obvious that all uv divergences are removed by the parameter renormalization and with the replacement @xmath61 ; again , remarkably , no wave function renormalization is required .
although the two - point function is uv finite , it contains ir divergences ( i.e. , it diverges for @xmath62 ) because it is not an @xmath0 invariant `` physical '' quantity @xcite . as a simple example of an ir - finite @xmath0-invariant observable , we can consider the `` energy density '' , defined by @xmath63 which is analogous to the energy density introduced in ref .
@xcite for the gauge theory . for the vacuum expectation value , @xmath64 ^ 2 + \left[\partial_\mu\sqrt{1-\pi(t , x)^2}\right]^2 \right\ } \right\rangle , \label{eq:(3.14)}\ ] ] there are four flow feynman diagrams to the next - to - leading order , as depicted in figs
. [ fig:6][fig:9 ] ( the cross denotes the operator @xmath65 ) .
a straightforward calculation using dimensional regularization yields @xmath66 \notag\\ & = \frac{g^2}{4\pi}(n-1)\frac{1}{4 t } \left[1+\frac{g^2}{4\pi}(n-2)\ln(8\pi\mu^2 t)+o(g^4)\right ] .
\label{eq:(3.15)}\end{aligned}\ ] ] this is ir finite as expected and uv finite in terms of the renormalized coupling constant , again indicating the uv finiteness of the flowed field .
if this uv finiteness persists to all orders ( we will prove this in a later section ) , the result shows that the combination @xmath67 provides a possible non - perturbative definition of a renormalized coupling as the gradient flow scheme in the 4d gauge theory ( see , e.g. , refs .
that is , we can set @xmath68 then it must be interesting to investigate the running of this non - perturbative coupling in numerical lattice simulations , in view of the expected conformal and walking behaviors of the @xmath69 non - linear sigma model with non - zero @xmath4-parameters @xcite .
in the next section , we reveal the renormalization structure of the flowed system defined in sect .
[ sec:2 ] .
we prove in particular that the flowed @xmath1-vector field does not require the wave function renormalization .
our strategy is identical to the case of the 4d gauge theory @xcite ; we seek a @xmath70-dimensional local field theory that reproduces the flow feynman rules in the preceding sections and use this to show the renormalizability .
we neglect the ir - regulating mass term in this section , because it complicates the argument destroying the @xmath0 symmetry .
we will consider the effect of the mass term at the very end of the next section .
as in refs .
@xcite , we consider a @xmath70-dimensional ( @xmath71 for our target theory ) field theory defined in the half space , @xmath72 , that ( at least perturbatively ) is equivalent to the gradient flow in the 2d @xmath0 non - linear sigma model .
we will find that , to resolve subtleties associated with the measure term and the flow - line loop ( see sect . [ sec:4.4 ] ) , it is necessary to specify a prescription for the flow - time derivative .
we will use the forward difference prescription ( with the discretization length @xmath73 ) for this .- dimensional system that assumes a particular forward difference for the flow - time derivative .
we do not mean , however , that the time evolution in the gradient flow must be defined by the forward time difference ; any sound discretization of the flow - time derivative can be used to implement the flow equation in numerical simulations .
the @xmath70-dimensional system below is merely an intermediate tool for the renormalization proof and , in our present context , is not an object to be simulated . ]
the regularization for the @xmath5-dimensional `` spacetime '' direction is , on the other hand , arbitrary and we may assume , for instance , dimensional regularization or lattice regularization .
the partition function of the @xmath70-dimensional field theory that we consider is defined by @xmath74 \left[\prod_{i=1}^n\mathcal{d}n^i\right ] \left[\prod_x\delta(n(x)^2 - 1)\right ] \notag\\ & \qquad { } \times \left[\prod_{t=0}^\infty\prod_{i=1}^n\mathcal{d}\lambda^i(t)\right ] \left[\prod_{t=0}^\infty\prod_{i=1}^n\mathcal{d}n^i(t)\right ] \left[\prod_{t=0}^\infty\prod_x \delta(n(t , x)^2 - 1)\sqrt{1-n_\bot(t+\epsilon , x)^2}\right]\ , \mathrm{e}^{-s } , \label{eq:(4.1)}\end{aligned}\ ] ] where @xmath75 , @xmath73 , @xmath76 , , and @xmath77 -\partial_\mu\partial_\mu n^j(t , x ) \right\ } \notag\\ & \qquad\qquad { } -i\int\mathrm{d}^dx\,\xi^i(x)\left[n^i(0,x)-n^i(x)\right ] .
\label{eq:(4.2)}\end{aligned}\ ] ] in these expressions , @xmath78 corresponds to the @xmath1-vector field in the @xmath5-dimensional @xmath0 non - linear sigma model and @xmath26 corresponds to the @xmath1-vector field evolved by the flow equation .
the basic idea is that the functional integral over the lagrange multiplier @xmath79 imposes the flow equation with the discretized flow time .
note that the left - hand side of eq . can equivalently be written as @xmath80 with the projection operator @xmath13 in eq . .
the integration over another lagrange multiplier @xmath81 in eq .
, on the other hand , imposes the initial condition . in eq . , @xmath82 , and @xmath83
it can be shown that , with the factor @xmath84 in the integration measure , the partition function @xmath85 can be obtained from the original partition function @xmath86 by inserting unity ( up to infinite gauge volume ; see below ) .
however , since @xmath87 for @xmath88 , this factor can be neglected in the @xmath88 limit and we do not explicitly include this factor in what follows .
the above @xmath70-dimensional system possesses the following symmetries .
one is the global @xmath0 symmetry that is inherited from the original @xmath0 non - linear sigma model : @xmath89 where @xmath90 are infinitesimal constant parameters .
other , somewhat unexpected ones are _ local _ gauge symmetries : @xmath91 where @xmath92 and @xmath93 and @xmath94 are _ local _ parameters that can depend on their arguments .
these local symmetries , which exist even with the discretized flow - time and @xmath5-dimensional regularization , follow from the constraints @xmath95 in the functional integral and the property @xmath96 . because of these gauge symmetries , the partition function itself is infinite .
this is not a problem in our present context , because what we need at this moment is a generating functional of the perturbative expansion of the flowed system . to formulate perturbation theory in the above @xmath70-dimensional field theory
, we thus have to first fix the gauge symmetries . for this
, we adopt the following gauge fixing conditions , @xmath97 and follow the faddeev
popov procedure .
thus we insert unity @xmath98 \notag\\ & \qquad\qquad{}\times \left[\prod_{t=0}^\infty\mathcal{d}h(t)\right ] \left[\prod_{t=0}^\infty\prod_x\delta(\lambda^n(t , x)-h(t , x)n^n(t , x ) ) \left|n^n(t , x)\right|\right]=1 \label{eq:(4.8)}\end{aligned}\ ] ] into the functional integral .
then , using the invariance of the action and the functional measure under the transformations , we can factor out the gauge volume @xmath99 \label{eq:(4.9)}\ ] ] from the partition function .
we further solve the constraints @xmath95 in terms of @xmath19 independent components , as eqs . and and eqs . and
then , after the gauge volume is factored out , the partition function is given by @xmath100 \left[\prod_{k=1}^{n-1}\mathcal{d}\pi^k\right ] \notag\\ & \qquad{}\times \left[\prod_{t=0}^\infty\prod_{k=1}^{n-1}\mathcal{d}\lambda^k(t)\right ] \left[\prod_{t=0}^\infty\prod_{k=1}^{n-1}\mathcal{d}\pi^k(t)\right]\ , \prod_x\frac{\overline{\sqrt{1-\pi(x)^2}}}{\sqrt{1-\pi(x)^2}}\ , \mathrm{e}^{-s } , \label{eq:(4.10)}\end{aligned}\ ] ] where @xmath101 and @xmath102 ^ 2 + \left[\partial_\mu\sqrt{1-\pi(x)^2}\right]^2\right\ } \notag\\ & \qquad{}-i\epsilon\sum_{t=0}^\infty\int\mathrm{d}^dx\,\lambda^k(t , x ) \left ( \frac{1}{\epsilon}\left[\pi^k(t+\epsilon , x)-\pi^k(t , x)\right ] -\partial_\mu\partial_\mu\pi^k(t , x ) -r^k(t , x)\right ) + \mathcal{e } \notag\\ & \qquad\qquad{}-i\int\mathrm{d}^dx\,\xi^k(x ) \left[\pi^k(0,x)-\pi^k(x)\right ] , \label{eq:(4.12)}\end{aligned}\ ] ] where the combination @xmath103 is defined by eq . and
@xmath104 \notag\\ & \qquad\qquad\qquad{}+\sqrt{1-\pi(t , x)^2 } \frac{1}{\epsilon } \left[\sqrt{1-\pi(t+\epsilon , x)^2}-\sqrt{1-\pi(t , x)^2}\right ] \biggr\}. \label{eq:(4.13)}\end{aligned}\ ] ] next we derive the feynman rules in the above system . to write down the free propagator , we introduce the heat kernel with the discretized flow time , by @xmath105 which fulfills @xmath106 -\partial_\mu\partial_\mu k_t^\epsilon(x)=0,\qquad k_0^\epsilon(x)=\delta^d(x ) . \label{eq:(4.15)}\ ] ] clearly , @xmath107 reduces to the heat kernel in the continuum flow - time limit , @xmath108 . by using this object ,
we change the integration variables from @xmath109 to @xmath110 as @xcite @xmath111 then the action becomes @xmath102 ^ 2+\dotsb\right\ } \notag\\ & \qquad{}-i\epsilon\sum_{t=0}^\infty\int\mathrm{d}^dx\,\lambda^k(t , x ) \left\ { \frac{1}{\epsilon}\left[p^k(t+\epsilon , x)-p^k(t , x)\right ] -\partial_\mu\partial_\mu p^k(t , x)+\dotsb \right\}+\dotsb \notag\\ & \qquad\qquad { } -i\int\mathrm{d}^dx\,\xi^k(x)p^k(0,x ) , \label{eq:(4.17)}\end{aligned}\ ] ] where abbreviated terms are cubic or higher in fields .
it is then straightforward to find free propagators and the result is @xmath112 where @xmath113 is a `` regularized '' step function , @xmath114 note that @xmath115 ( not , e.g. , @xmath116 ) . since other free propagators among @xmath37 , @xmath110 , @xmath117 , and @xmath118 vanish , eqs . in conjunction with eq .
show , @xmath119 in passing , we note @xmath120 and @xmath121 this completes our derivation of free propagators . in the continuum flow - time limit
@xmath88 , the @xmath122-propagator reproduces the @xmath122-propagator in eq .
and the @xmath123-propagator reproduces the flow - line propagator @xmath124 in eq .
; the step function @xmath125 is implicitly implied in eq . through the retarded time - ordering , @xmath126 .
the interaction terms in the present @xmath70-dimensional system are given by terms in eq .
being cubic or higher in fields .
the first line of eq .
of course reproduces the interaction terms in the action of the @xmath0 non - linear sigma model , eq . .
on the other hand , the term @xmath127 in the limit @xmath88 , combined with the above @xmath123-propagator , precisely reproduces the last term of the integral equation ( i.e. , the flow vertex ) .
thus , we have observed that our present @xmath70-dimensional system basically reproduces the perturbative expansion of the flowed system defined in sect .
[ sec:2 ] ; they seem to be basically equivalent .
nevertheless , we should note that the equivalence appears not quite complete .
the measure term is missing in eqs .
( the factor @xmath128 becomes unity under the integration over @xmath129 and this is not the measure term ) .
although the measure term identically vanishes when one uses dimensional regularization , it plays an important role in other regularizations , such as lattice regularization . if the equivalence including the measure term does not hold , then the renormalizability proof in the next section , which is based on the present @xmath70-dimensional field theory , does not apply to the gradient flow with , e.g. , lattice regularization
then , the uv finiteness of the gradient flow with lattice regularization , which we observed through an explicit calculation in sect . [ sec:3 ] , is not explained by the proof .
we will find that , rather surprisingly , the measure term is generated from naively @xmath130 terms in the action .
the aim of the next subsection is to clarify this point and to establish the perturbative equivalence between the above @xmath70-dimensional system and the flowed system in sect .
[ sec:2 ] .
we first integrate over the lagrange multiplier @xmath118 in the partition function .
then @xmath131 is identified with @xmath37 and we have @xmath132 in eq . .
next we note that the perturbative expansion of eq . generates loop diagrams consisting solely of the flow - line propagator .
such `` flow - line loop diagrams '' are depicted in figs .
[ fig:10 ] and [ fig:11 ] .. ] it is now very important to recognize that there is no counterpart to the above flow - line loop diagrams in the perturbative expansion of the flowed system in sect . [ sec:2 ] .
this is a consequence of the retarded nature of the flow equation and one can confirm this by drawing flow - line diagrams starting from eq . .
thus , there appears some ( apparent ; see below ) discrepancy between the perturbative expansions of the above two systems .
let us begin our investigation from the flow - line loop diagram in fig .
[ fig:10 ] which starts and ends at the same flow vertex .
first note that the @xmath123-propagator vanishes when the flow time of @xmath133 is greater or _
equal _ to the flow time of @xmath20 . therefore , in eq .
, the genuine flow vertex containing the non - linear term @xmath103 does not contribute to the flow - line loop diagram in fig .
[ fig:10 ] .
what contributes is the self - contraction in the combination @xmath134 .
the self - contraction of @xmath135 yields @xmath136\right\}. \label{eq:(4.27)}\ ] ] if we taylor expand @xmath137 in this expression with respect to @xmath73 , we find @xmath138+o(\epsilon ) .
\label{eq:(4.28)}\ ] ] since this is a total derivative , only the boundary field @xmath139 is contained .
then , remarkably , eq . coincides with the measure term for @xmath88 . the above result can be obtained in a somewhat different manner .
we first taylor expand @xmath135 , which yields @xmath140 ^ 2 + \frac{\left[\pi^l(t , x)\partial_t\pi^l(t , x)\right]^2}{1-\pi(t , x)^2 } \right\}\epsilon+o(\epsilon^2)\right ) , \label{eq:(4.29)}\ ] ] which is @xmath130 .
this @xmath130 term becomes @xmath141 under the self - contraction , because in the @xmath88 limit of the @xmath123-propagator behaves as @xmath142 and the delta function at the equal flow - time is interpreted as @xmath143 .
the self - contraction in eq . thus cancels the factor @xmath73 and leaves the @xmath141 result , eq . .
next , we see that a flow - line loop diagram that contains a plurality of flow vertices , such as the diagram in fig .
[ fig:11 ] , vanishes as @xmath88 .
a little thought shows that _ all _ vertices in such a flow - line loop diagram must be the vertex arising from @xmath134 .
this is again because the @xmath123-propagator vanishes when the flow time of @xmath133 is greater than or equal to that of @xmath20 .
the vertex is @xmath130 as in eq . .
the integration of the flow time of each vertex eliminates one delta function and finally one is left with an overall integration and @xmath143 . in the present case of a plurality of flow vertices ,
however , the power of @xmath73 coming from the vertices is always greater than or equal to two ; thus the flow - line loop diagram vanishes for @xmath88 .
the conclusion is that eq .
is the unique contribution of the flow - line loop diagrams for @xmath88 . by similar reasoning
, it can be confirmed that eq . is the unique place in which an apparent @xmath130 term in the action contributes in the @xmath88 limit .
the integration of the flow time of each vertex eliminates one delta function and the singularity @xmath143 can arise only from the flow - line loop diagrams , the case already considered above .
this observation justifies the taylor expansion with respect to @xmath73 and the neglect of the @xmath130 terms besides that particular term in eq . .
thus , we have observed that the perturbative expansions in the above two systems are equivalent by a remarkable mechanism : a flow - line loop diagram in the @xmath70-dimensional system , which is not generated in the perturbative expansion of the original flow equation , reproduces the measure term which is absent in the original partition function of the @xmath70-dimensional system , eq . .
the mechanism is remarkable , because an apparent @xmath130 term in the action , i.e. , @xmath134 , plays the crucial role through the flow - line loop .
now , having established the equivalence between the @xmath70-dimensional field theory and the flowed system in sect .
[ sec:2 ] , we are ready to prove the uv finiteness of the gradient flow in sect .
[ sec:2 ] .
in this section , on the basis of the @xmath70-dimensional field theory in the preceding section , we show that any correlation function of the flowed @xmath1-vector field in terms of the renormalized coupling is uv finite , without the wave function renormalization .
we first note that even with the gauge fixing , there remains a residual symmetry that is a particular combination of the global @xmath0 symmetry and the local symmetries .
it is given by the requirement that it does not affect the gauge fixing conditions .
that is , @xmath144 from these , we have @xmath145 under this residual symmetry , other field components transform as @xmath146 where indices @xmath147 and @xmath148 run over only from @xmath149 to @xmath19 .
the interesting part in the above residual symmetry is the @xmath150 part corresponding to the choice of parameters @xmath151 .
writing @xmath152 , it induces the following non - linear transformations @xmath153 it can be directly confirmed that the integration measure and the action in eqs . are invariant under this non - linear transformation ; this is expected , because the original partition function with the discrete flow time , eq . with eq .
, is invariant under eqs . and .
we can express the invariance of the system under the non - linear transformation as an identity for the generating functional of 1pi correlation functions .
first , we introduce the source terms for elementary fields , @xmath154 -\epsilon\sum_{t=0}^\infty\int\mathrm{d}^dx\ , \left[j_\pi^k(t , x)\pi^k(t , x)+j_\lambda^k(t , x)\lambda^k(t , x)\right ] , \label{eq:(5.6)}\ ] ] _ except _ for the lagrange multiplier field @xmath118 .
it turns out that this omission of the @xmath129-source greatly simplifies the discussion of the renormalization .
this implies that we omit correlation functions including @xmath118 from our consideration .
however , since the @xmath129-field appears only in the quadratic ( i.e. , free ) part of the action @xmath155 only linearly , if 1pi correlation functions of other elementary fields turn out to be uv finite after renormalization , any correlation functions including the elementary @xmath129-field are also uv finite .
hence nothing is lost by the omission of the @xmath129-source for our present purpose . to write down the ward
takahashi relation associated with the symmetry , we also supplement additional terms to the action , as @xmath156 where @xmath157 and @xmath158 where the source @xmath159 is symmetric in indices @xmath160 by definition .
we now consider the variation of integration variables of the form of eq . in the partition function :
\left[\prod_{k=1}^{n-1}\mathcal{d}\pi^k\right ] \notag\\ & \qquad{}\times \left[\prod_{t=0}^\infty\prod_{k=1}^{n-1}\mathcal{d}\lambda^k(t)\right ] \left[\prod_{t=0}^\infty\prod_{k=1}^{n-1}\mathcal{d}\pi^k(t)\right]\ , \prod_x\frac{\overline{\sqrt{1-\pi(x)^2}}}{\sqrt{1-\pi(x)^2}}\ , \mathrm{e}^{-s_{\text{tot}}-s_j}. \label{eq:(5.11)}\end{aligned}\ ] ] we note @xmath162 and @xmath163 + \sum_{i=1}^n\epsilon^{l_i } \mathcal{o}^{k , l_1\dotsc\not{l}_i\dots l_n}(t , x ) .
\label{eq:(5.13)}\ ] ] then , by the standard argument , the invariance of the integration measure and of @xmath155 imply that the generating functional of 1pi functions , defined by the legendre transformation , @xmath164 + \epsilon\sum_{t=0}^\infty\int\mathrm{d}^dx\ , \left[j_\pi^k(t , x)\pi^k(t , x ) + j_\lambda^k(t , x)\lambda^k(t , x)\right ] , \label{eq:(5.14)}\ ] ] where @xmath37 , @xmath109 , and @xmath117 denote expectation values of elementary fields , follows an identity @xmath165 \notag\\ & \qquad { } + \int\mathrm{d}^dx\,h(x)\pi^m(x ) + \int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\,h(t , x)\pi^m(t , x ) \notag\\ & \qquad{}+\int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , k^{k , l}(t , x ) \left [ \frac{\delta{\mit\gamma}}{\delta k^{m , kl}(t , x ) } + \frac{\delta{\mit\gamma}}{\delta k^{k , ml}(t , x ) } -\delta^{lm}\lambda^k(t , x ) \right ] \notag\\ & \qquad { } + \int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , \sum_{n=2}^\infty k^{k , l_1\dots l_n}(t , x ) \notag\\ & \qquad\qquad { } \times \left [ \frac{\delta{\mit\gamma}}{\delta k^{m , kl_1\dots l_n}(t , x ) } +
n \frac{\delta{\mit\gamma}}{\delta k^{k , ml_1\dots l_n}(t , x ) } + \sum_{i=1}^n\delta^{l_im } \frac{\delta{\mit\gamma}}{\delta k^{k , l_1\dotsc\not{l}_i\dots l_n}(t , x ) } \right]=0 . \label{eq:(5.15)}\end{aligned}\ ] ] in writing down this identity , we have taken the continuum flow - time limit @xmath88 .
this is justified because we have observed that the symmetry is preserved by the flow - time discretization .
also , we have observed that the @xmath70-dimensional system eqs . with @xmath88 reproduces the perturbative expansion of the flow equation .
thus we can study the renormalizability of the flowed system in sect .
[ sec:2 ] by using the identity .
our statement of the renormalizability is that the 1pi generating functional @xmath166 can be made uv finite in terms of renormalized quantities , by appropriately choosing the constants @xmath167 and @xmath47 in @xmath168 order by order in perturbation theory .
in particular , we claim that the flowed or `` bulk '' fields , @xmath109 and @xmath117 , do not require multiplicative renormalization .
our argument proceeds by mathematical induction based on the loop expansion .
we set @xmath169 where @xmath170 is the generating functional in the @xmath171 th loop order .
the above assertion is certainly true for @xmath172 ( tree - level approximation ) for which @xmath173 is sufficient .
then suppose that , in perturbation theory with renormalized quantities fixed , the constants @xmath167 and @xmath47 in eq . can be chosen so that @xmath174 , , @xmath170 , are uv finite in terms of renormalized quantities
. then consider the @xmath175 th loop order calculation on the basis of the above chosen @xmath167 and @xmath47 .
since @xmath167 and @xmath47 have already been chosen so that @xmath174 , , @xmath170 are finite , by considering uv - divergent part of the identity in the @xmath175 th loop order , we have @xmath176 where @xmath177 denotes uv - divergent part of @xmath178 and @xmath179 \notag\\ & \qquad { } + \int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , \left [ \frac{\delta{\mit\gamma}^{(0)}}{\delta\pi^m(t , x ) } \frac{\delta}{\delta h(t , x ) } + \frac{\delta{\mit\gamma}^{(0)}}{\delta h(t , x ) } \frac{\delta}{\delta\pi^m(t , x ) } \right ] \notag\\ & \qquad { } + \int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , \left [ \frac{\delta{\mit\gamma}^{(0)}}{\delta\lambda^k(t , x ) } \frac{\delta}{\delta k^{m , k}(t , x ) } + \frac{\delta{\mit\gamma}^{(0)}}{\delta k^{m , k}(t , x ) } \frac{\delta}{\delta\lambda^k(t , x ) } \right ] \notag\\ & \qquad{}+\int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , k^{k , l}(t , x ) \left [ \frac{\delta}{\delta k^{m , kl}(t , x ) } + \frac{\delta}{\delta k^{k , ml}(t , x ) } \right ] \notag\\ & \qquad { } + \int_0^\infty\mathrm{d}t\int\mathrm{d}^dx\ , \sum_{n=2}^\infty k^{k , l_1\dots l_n}(t , x ) \notag\\ & \qquad\qquad { } \times \left [ \frac{\delta}{\delta k^{m , kl_1\dots l_n}(t , x ) } +
n \frac{\delta}{\delta k^{k , ml_1\dots l_n}(t , x ) } + \sum_{i=1}^n\delta^{l_im } \frac{\delta}{\delta k^{k , l_1\dotsc\not{l}_i\dots l_n}(t , x ) } \right ] , \label{eq:(5.19)}\end{aligned}\ ] ] and @xmath180 we next study the most general form of the divergent part @xmath177 .
first of all , by a general theorem , the divergent part must be an integral of a local polynomial of fields and their derivatives .
we then note that there is no divergence corresponding to a local term in the `` bulk '' @xmath181 , a term that is written as an @xmath182 integral of a local polynomial of fields and their derivatives : as we explained in detail in sect [ sec:4.4 ] , there is no loop diagram consisting solely of the `` flow - line '' @xmath123-propagator , other than the diagram in fig .
[ fig:10 ] , which reduces to the measure term at the boundary @xmath75 , eq . .
then , since the @xmath122-propagator possesses the gaussian damping factor @xmath183 ( for @xmath88 ) , any loop diagram in which the flow times of the vertices ( they must be the same for the divergent part ) are positive is uv finite .
therefore , there is no divergence that is written as the bulk integral . any divergent part is thus written as the integral on the boundary @xmath75 .
noting that for @xmath71 the fields @xmath184 and @xmath109 possess the the mass dimension @xmath185 , @xmath186 , @xmath117 , and @xmath159 possess @xmath9 , and @xmath187 possesses @xmath188 , the most general possible form of the divergent part is @xmath189 , \label{eq:(5.21)}\end{aligned}\ ] ] where @xmath190 contains at most two derivatives and @xmath191 is symmetric in indices @xmath160 .
note that we have not included the flow field at zero flow time , @xmath131 , in the possible form of the divergent part .
the redundancy to use this field variable in addition to @xmath184 follows from the relation @xmath192 i.e. , the expectation value of the variation of the action with respect to the @xmath129-field .
note that here the field variables denote the expectation values in the presence of source fields and _ not _ the integration variables in the functional integral .
this identity shows that as the arguments of the 1pi generating functional , the variables @xmath131 and @xmath37 can not be independent , because they can not take different values for any configuration of the source fields .
we note also that the combination @xmath193 does not appear in eq . :
an external @xmath109 line in a 1pi diagram can arise only from a flow vertex that inevitably contains the lagrange multiplier field @xmath117 . since there is no flow - line loop ( other than the diagram in fig .
[ fig:10 ] which reduces to a boundary term ) , the flow - line propagator starting from @xmath117 can end only at _ another _ flow vertex that contains another @xmath194 .
this shows that any 1pi diagram containing @xmath109 must accomplish at least one @xmath133 .
the combination does not match this rule .
now , having obtained the general form of the divergent part , eq .
, we examine the implication of the identity with eq . .
first of all , examining the coefficient of @xmath195 in eq . that arises from @xmath196 in eq .
, we have @xmath197 then , from various terms in eq . , we have @xmath198c , \label{eq:(5.26)}\\ & \frac{\partial d^k}{\partial\pi_r^m(x ) } + \delta^{mk}\frac{\pi_r^l(x)}{1-\pi_r(x)^2}d^l = 0 , \label{eq:(5.27)}\end{aligned}\ ] ] and @xmath199 the above conditions for @xmath190 and @xmath200 , eqs . and , are completely identical to the conditions on the divergent part in the original 2d @xmath0 non - linear sigma model @xcite .
the general solution to these is given by @xcite @xmath201 ^ 2 + \left[\partial_\mu\sqrt{1-\pi_r(x)^2}\right]^2\right\ } \notag\\ & \qquad { } -\delta z_3\frac{1}{2\mu^\epsilon g^2 } \left\ { \left[\partial_\mu\pi_r(x)\right]^2 -\partial_\mu\sqrt{1-\pi_r(x)^2 } \partial_\mu\frac{\pi_r(x)^2}{\sqrt{1-\pi_r(x)^2}}\right\ } , \label{eq:(5.31)}\end{aligned}\ ] ] where @xmath202 and @xmath203 are constants .
next , from the linearly realized @xmath204 symmetry ( that is preserved in our all steps ) , one can set @xmath205 .
. immediately shows that @xmath206 and @xmath207 next , from eqs . and , and
the fact that @xmath208 is symmetric under the exchange @xmath209 , we have @xmath210 finally , we note that , if the right - hand side of eq .
vanishes , then @xmath211 and thus @xmath212 this is actually the case by mathematical induction because the right - hand side of eq .
vanishes for @xmath213 from eqs . and and then for @xmath214 again from eq . ; we see that @xmath215 for all @xmath216 . in summary
, we observed that possible divergent part in the present system is given by eq . with eqs . and and @xmath217 .
this is identical to the divergent part in the 2d @xmath0 non - linear sigma model .
one can see that the divergent part with eqs . and is canceled by the variation of the total action @xmath218 under the change of the renormalization constants in eq . by @xmath175 th loop order quantities : @xmath219 the 1pi generating functional in the @xmath175 th loop order , @xmath178 ,
is thus made uv finite .
this completes the mathematical induction for the renormalizability .
in particular , we showed that _ there is no need of the wave function renormalization for the flowed fields _
, @xmath109 and @xmath117 .
we have shown that any correlation function of the flowed fields is uv finite under the conventional parameter renormalization , without multiplicative wave function renormalization .
then , it is easy to see that , because of gaussian damping factors in propagators , this uv finiteness holds even when some spacetime coordinates of the correlation function coincide , i.e. , even in the equal - point limit , as long as all flow - time coordinates of flowed fields are strictly positive .
the local product of any number of flowed fields does not contain uv divergences .
this robust uv finiteness , that makes the construction of renormalized composite operators straightforward , is the key property in application of the gradient flow in lattice field theory . in renormalized perturbation theory
, one uses the propagators and the vertices in terms of renormalized parameters and renormalized fields .
this renormalized feynman rule is obtained by making the substitution in the action .
the part including the @xmath129-field becomes @xmath220 + i\int\mathrm{d}^dx\,(z_3^{1/2}-1 ) \xi^k(x)\pi_r^k(x ) , \label{eq:(5.37)}\ ] ] and the second term is regarded as the perturbation . in this renormalized perturbation theory , from the first term , the free propagator is given by @xmath221 while the second term is regarded as a counterterm . in this way
, we can also use eq . for @xmath184 by identifying @xmath222 .
as the @xmath223-propagator shows , the second term in eq . acts as a two - point vertex at the boundary @xmath75 that connects between @xmath109 and @xmath224 .
this counterterm thus plays the same role as the boundary counterterm @xmath225 in the gauge theory ( sect .
7.1 of ref .
@xcite ) .
finally , the ir - regulating mass term can readily be incorporated in the above argument by the substitution @xmath226 in particular , from eq . , we see that the generating functional becomes uv finite in terms of @xmath227 = h_r(x)+\frac{1}{\mu^\epsilon g^2}\frac{z_3^{1/2}m_0 ^ 2}{z}. \label{eq:(5.40)}\ ] ] this shows that the mass parameter is renormalized as @xmath228 , as we already noted in eq . .
in the preceding section , we have shown that any local product ( the composite operator ) of the bare flowed @xmath1-vector field becomes uv finite under the conventional parameter renormalization in the 2d @xmath0 non - linear sigma model .
as application of this fact , in the present section , we consider the construction of the energy
momentum tensor , the noether current associated with the translational invariance , in a lattice formulation of the non - linear sigma model .
the idea is the same as that in refs .
@xcite and @xcite : since lattice regularization explicitly breaks the translational invariance , the construction of the energy
momentum tensor is awkward .
instead of considering this construction directly , we construct a composite operator of the flowed field which , under dimensional regularization , becomes the energy momentum tensor . since dimensional regularization preserves the translational invariance , the description of the energy momentum tensor that fulfills the correct ward
takahashi relation is straightforward . on the other hand , since the composite operator of the flowed field is uv finite under the parameter renormalization , it must become independent of the regularization in the limit that the regulator is removed ( after the renormalization , as long as the same renormalization conditions are adopted ) . in this way ,
low - energy correlation functions of the energy momentum tensor may be computed by using lattice regularization .
this construction in ref .
@xcite has been applied to the thermodynamics of quenched qcd in ref .
@xcite and promising results have been obtained . the energy
momentum tensor @xmath229 can be obtained from the variation of the action under the infinitesimal translation with a localized parameter , @xmath230 as @xmath231 the explicit form is given by @xmath232 .
\label{eq:(6.3)}\ ] ] assuming that we are using dimensional regularization , which preserves the translational invariance , the above classical expression as it stands fulfills the correct ward
takahashi relation associated with the translational invariance : @xmath233 in this expression , @xmath234 is a bounded integration region , @xmath235 is an operator outside the region @xmath234 , and @xmath236 is an operator inside the region .
we defined the renormalized energy
momentum tensor by subtracting the vacuum expectation value , @xmath237 .
takahashi relation ensures that the bare quantity @xmath229 is not multiplicatively renormalized .
although naively the energy momentum tensor is traceless for @xmath71 , uv divergences in the composite operator @xmath238 being proportional to @xmath239 makes this expectation invalid even for @xmath88 .
instead , we have the the trace anomaly , @xmath240 where the @xmath241 scheme is assumed in the renormalized composite operator in the right - hand side and the coefficient is given by the @xmath242 function , @xmath243 ( here , the derivative with respect to the renormalization scale @xmath244 is taken while bare quantities are kept fixed ) , and @xcite @xmath245 and @xmath246 . \label{eq:(6.8)}\ ] ] we construct a composite operator of the flowed field which reduces to the 2d composite operator by using the small flow - time expansion introduced in ref . @xcite . for this
, we take an @xmath0-invariant dimension-@xmath9 second - rank composite operator of the flowed field : @xmath247 according to the argument in ref .
@xcite , for @xmath248 , this composite operator of the flowed field can be expressed as a series of 2d local operators with increasing mass dimensions , as @xmath249 \notag\\ & \qquad\qquad { } + \zeta_{12}(t)\left [ \delta_{\mu\nu}\partial_\rho n^i(x)\partial_\rho
n^i(x ) -\left\langle\delta_{\mu\nu}\partial_\rho n^i(x)\partial_\rho n^i(x ) \right\rangle \right ] + o(t ) .
\label{eq:(6.10)}\end{aligned}\ ] ] similarly , we have @xmath250+o(t ) .
\label{eq:(6.11)}\end{aligned}\ ] ] inverting these relations with respect to the 2d operators and substituting them into eq .
, we have @xmath251 \notag\\ & \qquad{}+c_2(t)\left [ \frac{1}{2 } \delta_{\mu\nu}\partial_\rho n^i(t , x)\partial_\rho n^i(t , x ) -\left\langle \frac{1}{2 } \delta_{\mu\nu}\partial_\rho n^i(t , x)\partial_\rho n^i(t , x ) \right\rangle \right]+o(t ) , \label{eq:(6.12)}\end{aligned}\ ] ] where @xmath252 \zeta_{22}(t)^{-1}+\zeta_{11}(t)^{-1}\right\}. \label{eq:(6.14)}\end{aligned}\ ] ] hence , if we know the @xmath248 behavior of the coefficients @xmath253 in eqs . and
, then the energy momentum tensor can be obtained by the @xmath248 limit of the right - hand side of eq . .
thus , we are interested in the @xmath248 behavior of the coefficients @xmath253
. since all the composite operators in the above expansions are bare ones , by the standard renormalization group argument , the expansion coefficients are independent of the renormalization scale @xmath254 , if they are expressed in terms of the running parameter @xmath255 .
in particular , we may take @xmath256 . then
because of asymptotic freedom , the running coupling behaves as @xmath257 for @xmath248 and @xmath253 for @xmath248 can be evaluated by perturbation theory . to find the coefficients @xmath253 in eq . , we consider correlation functions , @xmath258 and @xmath259 and compute the coefficients of combinations , @xmath260 in tensors @xmath261 and @xmath262 .
then , we determine @xmath253 so that the relation holds in correlation functions in view of the combinations .
set @xmath263 , where superscripts denote the number of loops . in the tree level , @xmath264 . from this
, it follows that the one - loop correction @xmath265 ( @xmath266 ) is given by the difference of one - loop coefficients of the former ( the latter ) combination in eq . between eqs . and .
in the one - loop level , for eq . , we have four diagrams in figs . [ fig:12][fig:15 ] . for eq . , we have only two diagrams in figs . [ fig:12 ] and [ fig:15 ] .
we use dimensional regularization to regularize the uv divergences and the mass term to regularize the ir divergences .
.one - loop coefficients of the combinations in @xmath261 in eq . in units of @xmath267 .
[ cols="^,>,>",options="header " , ]
we would like to thank koji harada for helpful discussion on perturbation theory in the non - linear sigma model and kengo kikuchi for exposition on his work , ref .
we are grateful to martin lscher for explaining the hard part of the proof in ref .
@xcite patiently to us . the work of h. s. is supported in part by a grant - in - aid for scientific research 23540330 .
in recent papers @xcite , the solution to the flow equation in the @xmath268 expansion is studied . in the former work , the expectation value of eq . at finite temperature
is computed for the large-@xmath1 limit and it is shown that the expectation value correctly reproduces thermodynamic quantities in the presence of a non - perturbative mass gap .
m. lscher , jhep * 1008 * , 071 ( 2010 ) [ arxiv:1006.4518 [ hep - lat ] ] .
s. borsnyi , s. drr , z. fodor , c. hoelbling , s. d. katz , s. krieg , t. kurth and l. lellouch _ et al .
_ , jhep * 1209 * , 010 ( 2012 ) [ arxiv:1203.4469 [ hep - lat ] ] . j. noaki , s. aoki , g. cossu , h. fukaya , s. hashimoto and t. kaneko , pos lattice * 2013 * , 263 ( 2013 ) .
z. fodor , k. holland , j. kuti , d. nogradi and c. h. wong , jhep * 1211 * , 007 ( 2012 ) [ arxiv:1208.1051 [ hep - lat ] ] .
p. fritzsch and a. ramos , jhep * 1310 * , 008 ( 2013 ) [ arxiv:1301.4388 [ hep - lat ] ] .
m. lscher , jhep * 1304 * , 123 ( 2013 ) [ arxiv:1302.5246 [ hep - lat ] ] .
m. lscher , jhep * 1406 * , 105 ( 2014 ) [ arxiv:1404.5930 [ hep - lat ] ] . k. kikuchi and t. onogi , arxiv:1408.2185 [ hep - th ] . c. monahan and k. orginos , pos lattice * 2014 * , 330 ( 2014 ) [ arxiv:1410.3393 [ hep - lat ] ] .
m. lscher , arxiv:1308.5598 [ hep - lat ] .
s. aoki , y. aoki , c. bernard , t. blum , g. colangelo , m. della morte , s. drr and a. x. el khadra _ et al .
_ , eur .
j. c * 74 * , no .
9 , 2890 ( 2014 ) [ arxiv:1310.8555 [ hep - lat ] ] .
a. hasenfratz , d. schaich and a. veernala , arxiv:1410.5886 [ hep - lat ] .
m. lscher and p. weisz , jhep * 1102 * , 051 ( 2011 ) [ arxiv:1101.0963 [ hep - th ] ] .
h. suzuki , ptep * 2013 * , no .
8 , 083b03 ( 2013 ) [ arxiv:1304.0533 [ hep - lat ] ] .
l. del debbio , a. patella and a. rago , jhep * 1311 * , 212 ( 2013 ) [ arxiv:1306.1173 [ hep - th ] ] . h. makino and h. suzuki , ptep * 2014 * , no .
6 , 063b02 ( 2014 ) [ arxiv:1403.4772 [ hep - lat ] ] .
s. caracciolo , g. curci , p. menotti and a. pelissetto , nucl .
b * 309 * , 612 ( 1988 )
. s. caracciolo , g. curci , p. menotti and a. pelissetto , annals phys . *
197 * , 119 ( 1990 ) .
k. fujikawa , z. phys .
c * 25 * , 179 ( 1984 ) .
m. asakawa _ et al . _
[ flowqcd collaboration ] , phys .
d * 90 * , 011501 ( 2014 ) [ arxiv:1312.7492 [ hep - lat ] ] .
a. m. polyakov , phys .
b * 59 * , 79 ( 1975 ) .
a. a. migdal , sov . phys .
jetp * 42 * , 743 ( 1975 ) [ zh .
fiz . * 69 * , 1457 ( 1975 ) ] .
e. brzin and j. zinn - justin , phys .
lett . * 36 * , 691 ( 1976 ) .
m. e. peskin and d. v. schroeder , _ an introduction to quantum field theory _
( addison - wesley , reading , ma , 1995 ) . m. bgli , f. niedermayer , m. pepe and u. j. wiese , jhep * 1204 * , 117 ( 2012 ) [ arxiv:1112.1873 [ hep - lat ] ] .
d. ngrdi , jhep * 1205 * , 089 ( 2012 ) [ arxiv:1202.4616 [ hep - lat ] ] .
u. wolff , phys .
lett . * 62 * , 361 ( 1989 ) .
u. wolff , nucl .
b * 334 * , 581 ( 1990 ) .
j. balog , f. niedermayer , m. pepe , p. weisz and u .- j .
wiese , jhep * 1211 * , 140 ( 2012 ) [ arxiv:1208.6232 [ hep - lat ] ] .
e. brzin , j. zinn - justin and j. c. le guillou , phys .
d * 14 * , 2615 ( 1976 ) .
j. zinn - justin , nucl .
b * 275 * , 135 ( 1986 ) .
j. zinn - justin and d. zwanziger , nucl .
b * 295 * , 297 ( 1988 )
. j. zinn - justin , `` renormalization of gauge theories , '' saclay-d.ph-t-74-88 lectures given at int .
summer inst . for theoretical physics ,
jul 29aug 9 , 1974 , bonn , west germany .
j. c. taylor , _ gauge theories of weak interactions _ ( cambridge university press , cambridge , uk , 1976 ) .
f. david , nucl .
b * 209 * , 433 ( 1982 ) .
s. hikami and e. brzin , j. phys .
a * 11 * , 1141 ( 1978 ) .
s. hikami , nucl .
b * 215 * , 555 ( 1983 ) .
w. bernreuther and f. j. wegner , phys .
lett . * 57 * , 1383 ( 1986 ) .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
h. makino and h. suzuki , arxiv:1403.4772v4 [ hep - lat ] , appendix d.
|
it is known that the gauge field and its composite operators evolved by the yang
mills gradient flow are ultraviolet ( uv ) finite without any multiplicative wave function renormalization . in this paper , we prove that the gradient flow in the 2d @xmath0 non - linear sigma model possesses a similar property : the flowed @xmath1-vector field and its composite operators are uv finite without multiplicative wave function renormalization .
our proof in all orders of perturbation theory uses a @xmath2-dimensional field theoretical representation of the gradient flow , which possesses local gauge invariance without gauge field . as application of the uv finiteness of the gradient flow
, we construct the energy
momentum tensor in the lattice formulation of the @xmath0 non - linear sigma model that automatically restores the correct normalization and the conservation law in the continuum limit .
|
scalar - tensor ( st ) theories of gravitation have a long history . the first ones to appear were those of jordan@xcite , fierz@xcite , and brans - dicke@xcite .
these are the simplest theories in the sense that they consist of just one massless scalar field and its coupling strength to matter is constant .
later bergmann@xcite , nordtvedt@xcite , and wagoner@xcite generalized the theory . in their versions
the scalar field has a dynamic ( scalar field dependant ) coupling to matter and/or an arbitrary self - interaction .
more recently , in a very comprehensive work , damour and esposito - farse@xcite considered the theory with multiple scalar fields .
one could say that scalar - tensor theories of gravity are the simplest alternative to einstein s general theroy of relativity ( gr ) .
most recent attempts at unified models of fundamental interactions , i.e.,string theories @xcite , predict the existence of scalar partners to the tensor gravity of gr , and would have st gravity as their low energy limit . in addition , these theories are important in inflationary cosmology where they provide a way of exiting the inflationary epoch in a non - fine - tuned way@xcite .
scalar - tensor theories provide an important framework for comparison with results of gravitational experiments that are going to take place in the near future , e.g gravity probe - b@xcite , ligos@xcite , geo@xcite , and virgo@xcite .
in the present paper we study the cosmological evolution of the metric and scalar gravitational fields in st gravity .
we consider the theories where the scalar field is massless , i.e. , no self interaction .
although it seems natural that without a gauge invariance the scalar would acquire a mass , certain results in string theory suggest that massless scalars arise naturally@xcite .
furthermore , it has been shown@xcite that when the scalar field is massless , the cosmological evolution drives the scalar field to a value that minimizes the coupling function , towards making the contribution of the scalar field very small .
this effect makes the theory practically indistinguishable from gr at the present epoch .
solar system tests can put limits on the post - newtonian parameters @xmath2 and @xmath3@xcite , which in st gravity translate into limits on @xmath6 and @xmath7 : the first two coefficients of the taylor series expansion about today of @xmath8 , the coupling function of the scalar field @xmath9 to matter .
we have the relations@xcite : @xmath10 from analysis of the shapiro time delay in the viking mars data@xcite , and from vlbi radio wave deflection@xcite we have @xmath11 while lunar laser ranging@xcite gives the limit @xmath12 these limits ( the most restrictive known to the authors ) imply that at the present epoch in the weak field regime gravity s behavior does not differ much from that predicted by gr .
we will show that the cosmological constraint places stronger limits on @xmath6 if @xmath7 is positive .
in addition it implies that if there is a scalar field contribution to gravity , it must be very small from the epochs of @xmath0 annihilation and nucleosynthesis up to the present .
the cosmological constraint has been analyzed less generally by casas _
al.__@xcite , kalligas _ _ et .
al.__@xcite , and serna and alimi@xcite .
a preliminary version of our results appears in @xcite . throughout this paper
we follow the convetions of mtw@xcite(@xmath13 ) .
we consider a theory of gravitation in which the gravitational interaction is carried by the ususal spin-2 field , the metric @xmath14 , plus a spin-0 field , a scalar field @xmath9 .
we work in the einstein conformal frame of @xcite , which we call the spin - frame , because the spin-2 and spin-0 excitations are then decoupled from each other .
the main advantage of the spin - frame is that , in general , calculations of the behavior of the fields ( as opposed to the motion of matter ) are more readily manageable .
another advantage of the spin - frame is that the cauchy problem is well posed@xcite . the field equations for our theory follow from extremizing the action @xmath15\;\ ;
+ \;\ ; s_{m}. \label{eq : s}\ ] ] here @xmath16 , @xmath17 is the ricci scalar , and @xmath18 is the matter action , which includes all the matter fields and their interactions including those with the gravitational fields .
the matter fields couple to the gravitational fields through the physical metric @xmath19 from the previous equation we see that the scalar field couples conformally to matter via the function @xmath20 .
a priori @xmath20 is a completely unspecified function of the scalar field .
solar system experiments constrain this function somewhat , and we will show that cosmology can give additional strong restrictions on this function . the three functions @xmath21 will be very useful because they are related to observable quantities
. it will be convenient to express these functions as their taylor series expansion about the asymptotic cosmological value of the scalar field in the present epoch , @xmath22 : @xmath23 where we scale our coordinates so that @xmath24 .
when we vary the action with respect to @xmath25 and @xmath9 , we obtain the field equations @xmath26 where @xmath27 are the spin - frame stress - energy tensor , its trace , and the divergence of the gradient of @xmath9 .
the bianchi identity leads to the equations of motion @xmath28 in all these formulae we follow the usual convention that comas represent partial derivatives and semi - colons represent covariant derivatives . from the field equations we see that
if we set @xmath29 with @xmath30 a constant such that @xmath31 and @xmath32 a solution to einstein s equations of gr , we have a solution to our field equations .
it follows that all general relativistic solutions are also solutions in st gravity with a appropriate choice of constant scalar field .
moreover since @xmath33 , @xmath30 locates an extremum of our coupling function @xmath1 .
when @xmath30 produces a minimum of @xmath1 , the gr solutions are stable solutions of st gravity , while if @xmath30 produces a maximum or a inflection point the solutions are unstable .
it has been demonstrated@xcite that if @xmath1 has a minimum or approaches one symptotycally , then the cosmological evolution of during the matter era drives the scalar field towards the minimum value of @xmath34 , and therefore the universe evolves towards general relativity .
we can express@xcite the measured gravitational constant far from all matter,@xmath35 , and the ppn parameters @xmath2 and @xmath3 in terms of the taylor expansion coefficients of @xmath36 @xmath37 from these last formulae we see how solar system tests give us information about the coupling function @xmath20 , i.e. , by constraining the possible values of @xmath38 and @xmath39 .
our cosmological constraint will give us information on these same two taylor series coefficients which , in turn , can be translated into a restriction on @xmath2 .
we now turn to the study of the evolution of the universe as a whole in st gravity .
we assume the universe is homogeneous and isotropic , so that the spacetime has a friedmann - robertson - walker metric @xmath40\ , , \ ] ] with @xmath41 being the element of solid angle and @xmath42 according to whether the universe is open , flat , or closed . in order to compare with observations we need to express all the physically meaurable quantities ( proper time , expansion rate , density , pressure , etc . ) in terms of spin - frame quantities ( the ones we use for our analysis ) .
because of the relation ( [ eq : physg ] ) between the physical metric and the spin - frame metric , the proper time measured by ideal clocks is not @xmath43 ; it is @xmath44 in particular , for comoving observers @xmath45 . we shall call @xmath46 the spin - frame `` proper '' time , but rememeber that it is not the time measured by ideal comoving clocks .
the physical scale factor of the universe is @xmath47 instead of @xmath48 , the spin - frame scale factor .
the condition of homogeneity implies that our scalar field is only a function of the the spin - frame time coordinate , @xmath49 .
the matter in the universe behaves to a large degree like a perfect fluid @xmath50 in the previous equation @xmath51 is the spin - frame 4-velovity normalized to @xmath52 for massive particles . since @xmath53
, we have @xmath54 with @xmath55 being the physical 4-velocity , i.e. , the actual 4-velocity particles have .
the physical stress - energy tensor is defined by @xmath56 analogous to equation ( [ eq : t ] ) for the spin - frame stress - energy tensor . using the relation ( [ eq : physg ] ) between the physical metric and the spin - frame metric
, we get @xmath57 which leads us to the following expressions for the spin - frame density and pressure of the matter in terms of the physical quantities , @xmath58 we remind readers that quantities with a tilde are the physical ones , which are the usual measured values . substituting the metric for our isotropic and homogeneous cosmological model and the perfect fluid spin - frame stress - energy tensor into our field equations ( [ eq : g ] ) and ( [ eq : f ] )
, we obtain @xmath59 where the overdots denote derivatives with respect to spin - frame coordinate time , @xmath60 . from the bianchi identity ( [ eq : bianchi ] ) we obtain the `` conservation '' equation @xmath61 which , of course , is a consequence of the field equations . just as is the case in gr , in st gravity we need an equation of state in order to solve our field equations . in general ,
equations of state relating density and pressure are valid only for physical quantities , @xmath62 , and not spin - frame quantities .
all the cosmological epochs we consider , except the @xmath63 era , have simple equations of state of the type @xmath64 .
we will have a similar type of relation between the spin - frame quantities because of equation ( [ eq : p ] ) and the simple form of the equation of state .
the @xmath63 era is handled by applying first order perturbation theory in the effect of the scalar field ( arguing why it is not necessary to go to higher orders ) . to first order
the spin - frame density and pressure are equal to the physical ones , enabling us to use the same equation of state for the spin - frame quantities .
we now turn our attention to the question of what cosmology says about st gravity . as expected , because of the presence of the scalar field
, the expansion rate of the universe will be different than the one predicted from gr .
nucleosynthesis is sensitive to the speed - up factor @xmath65 ( the ratio of the expansion rate at nucleosynthesis to that predicted by gr ) .
thus we see that the production of primordial light elements , especially @xmath4he , will provide us with limits on st gravity . in order to obtain these constraints
we will integrate semi - analytically ( in perturbation theory ) our cosmological equations from the present era back to the epoch of @xmath63 annihilation .
this last era is handled numerically , performing the appropriate matching at the transitions between eras .
the next step is to establish the relation between the speed - up factor , the scalar field , and the other quantities of our theory .
the following relation between the speed - up factor , @xmath65 , and the conformal factor , @xmath20 , has been derived @xcite from quite general considerations , @xmath66 with @xmath67 being the value of the scalar field at the time of nucleosynthesis and @xmath22 its value today . since @xmath65 can not differ from unity by more than @xmath68% without over- or under producing@xmath4he @xcite we can taylor expand our coupling function about @xmath22 and consider terms beyond the second order as unimportant for all times between the present and nucleosynthesis : @xmath69 combining the two previous equations with equation ( [ eq : a ] ) we arrive at @xmath70 to the desired order .
the equations of motion can be integrated to yield @xmath71 .
when this quantity is substituted in our last equation we will have the speed - up factor as a function of the the taylor coefficients @xmath38 and @xmath72 .
then we can use the abundances of primordial light elements ( functions of @xmath65 ) to obtain the allowed values of @xmath38 as a function of @xmath39 .
we will restrict ourselves to the study of those st cosmologies that evolve towards gr .
this is equivalent to considering only coupling functions @xmath1 that at least have a minimum or approach one asymptotically . in terms of the taylor expansion coefficients of the coupling function , this implies @xmath73 .
excluded is the brans - dicke case [ @xmath1 linear in @xmath9 ] , which is not that interesting because the deviations from gr are fixed , i.e. , @xmath75 does not depend on the scalar field .
excluded also are the cases where the coupling function has no minimum , has points of inflection and at most one maximum . in this case solar system constraints place us very near a point of inflection or the maximum . in either case
the coupling function will evolve away from such a point and , in the future , the deviations from general relativity will be more pronounced
. we will consider both flat and open universes .
for each universe we will have the following eras : @xmath0 annihilation , radiation dominated , and matter dominated . for the open universe
we also have a curvature dominated epoch .
we approximate the transitions between eras as sudden .
scalar - tensor theories are very general because the coupling function @xmath1 could be any function of the scalar field .
this generality prevents one from performing any calculation unless one chooses a particular coupling function .
thus it seems that any calculations will be dependent on the choice among the uncountably many coupling functions .
this appears quite unnatural especially since the coupling function @xmath1 should be a fundamental property of nature ( @xmath34 constant corresponding to gr ) .
therefore @xmath1 should be derived from fundamental principles or somehow be obtained from experimental facts .
we try to do the latter in the present work using the light element abundances .
the function @xmath76 acts as a measure of the deviations from gr .
when @xmath77 the scalar contribution to gravity is small and the theory is very close to gr , while if @xmath78 there will be large deviations from gr .
the solar systems limits on the ppn parameters @xmath2 and @xmath3 mentioned in equations ( [ eq : ppngam ] ) and ( [ eq : ppnbet ] ) imply that the value of @xmath75 today , @xmath38 , is small .
so at the present time the universe has small deviations from a general relativistic behavior .
this suggests that we can treat the effects of the scalar contribution as a first order perturbation on gr .
since we are very near a minimum of @xmath1 , @xmath79 acts as a `` measure '' of how far we are from that minimum , i.e. , the strength of the scalar contribution to gravity . for @xmath80
, we have no scalar contribution ( i.e. , gr ) while for @xmath81 we have small scalar contributions . at present , since the effect of the scalar field is small , we expect @xmath81 and the cosmological field today , @xmath22 , to be different by order @xmath38 from a value that minimizes @xmath1 , i.e. , @xmath82 . moreover , as long as we can apply first order perturbation theory we expect the deviations of the scalar field from its value today to also be proportional to @xmath38 : @xmath83 therefore we can consider @xmath38 as the small parameter for a first order perturbation theory expansion about general relativity .
besides @xmath9 itself , @xmath75 is the only other scalar field dependent quantity to appear in the field equations ( [ eq : g ] ) and ( [ eq : f ] ) .
hence to do first order perturbation theory in the effect of the scalar field we need @xmath75 to linear order in @xmath9 : @xmath84 since this is the derivative with respect to @xmath9 of equation ( [ eq : aexp ] ) , we see that we need @xmath85 , to second order .
but if higher order deviations are important , we will over- or underproduce @xmath4he .
therefore our first order perturbation theory approach is valid at least up to the time of nucleosynthesis .
if we look at the field equations ( [ eq : g ] ) for the metric , we see that they depend on derivatives of @xmath9 squared .
since the physical metric as given by equation ( [ eq : physg ] ) differs from the spin - frame metric by quantities of second order ( @xmath86 ) , it must differ from the gr metric by quantities of at least second order .
moreover , the matter lagrangian , and hence , the stress - energy tensor , will be equal to the gr quantities to at least zeroth and first order . therefore the metric will be unaffected to first order while to zeroth order we just have a gr solution , @xmath32 : @xmath87 in general , the physical metric is of more interest .
the first order correction in the scalar field affects the physical metric to second order . the second order correction to the
metric affects ideal clocks and measuring sticks to the same order as the first order scalar field correction .
therefore we would need to find this metric correction to understand the full physics of a problem to the desired order .
for the nucleosynthesis constraint we are fortunate because we have formula ( [ eq : xi ] ) or ( [ eq : lxi ] ) which relates the speed - up factor to the coupling function independently of the metric .
we see that in this particular case we can neglect the back reaction of the metric in the presence of the scalar field .
there will also be corrections to the spin - frame stress - energy tensor , but as we previously argued they will be second order : @xmath88 with @xmath89 the general relativistic stress - energy tensor . to this order
the spin - frame stress - energy tensor is equal to the physical one .
now we consider the st cosmological and `` conservation '' equations ( [ eq : acc]-[eq : cons ] ) in perturbation theory .
we note that these equations are not independent ( they are related through the bianchi identity ) .
hence we do not use equation ( [ eq : acc ] ) . for the spin - frame expansion factor , scalar field , density , and pressure
we have the following relations to the desired order : @xmath90 when we substitute these in the cosmological equations we obtain to zeroth ( gr ) order : @xmath91 where we have dropped the @xmath92 superscripts .
to first order scalar field equation becomes @xmath93 for later use we define @xmath94 so that @xmath95 satisfies @xmath96 for the nucleosynthesis constraint we will need the speed - up factor . substituting the perturbation expansion ( [ eq : exphi ] ) of @xmath9 into the relation ( [ eq : lxi ] ) for the speed - up factor we get @xmath97 = a_{1}^{2}(-\frac{1}{2 } + \frac{\psi ^{2}(t_{n})-1}{2a_{2 } } ) \ , , \label{eq : lxiexp}\ ] ] where @xmath98 is the time at nucleosynthesis , and we used equation ( [ eq : psi ] ) to get the last equality .
next we find solutions to the field equations in the different epochs that occur in the two types of universe we consider : flat and open .
we consider an era in which the evolution of the universe is mainly driven by the curvature term , i.e. , the curvture term is large compared to the density term .
this describes to a fairly good aproximation the present epoch for a sufficiently open universe ( @xmath99 ) .
when we neglect the density term in our zeroth order field equation ( [ eq : h0 ] ) we have @xmath100 which trivially integrates to @xmath101 with @xmath102 a constant of integration .
we must consider how the matter density depends on the scale factor since we should include the matter density term in equation ( [ eq : dpsi ] ) for the scalar field because although it is small at each time it builds up during the evolution .
taking @xmath103 in the zeroth order conservation equation ( [ eq : cons0 ] ) gives us the well known simple result @xmath104 where @xmath105 is a constant of normalization and the subscript @xmath106 on @xmath107 is to specify that it is the matter density .
to first order we use the auxiliary scalar field @xmath95 and its corresponding equation ( [ eq : dpsi ] ) . using our last result the @xmath95 equation becomes @xmath108 this is a transformed bessel s equation@xcite , which has solution @xmath109 where the @xmath110 are constants of integration , and @xmath111 and @xmath112 are bessel functions of the first and second kind respectively .
it is interesting that @xmath95 in equation ( [ eq : psicd ] ) does not converge to zero for large times , rather it approaches a constant .
to first order @xmath113 corresponds to the value of @xmath9 for which @xmath114 .
therefore the universe does not evolve towards a state in which the deviations from gr disappear , but as it evolves the deviation from gr converges to a finite , though small , value .
the attractor mechanism of @xcite is not too effective because for large times the hubble damping makes the scalar field `` stop '' before reaching the value which produces @xmath115 .
the full attractor mechanism is present only for flat scalar - tensor cosmologies because those will have gr as a limit for large times . in a matter
dominated epoch the evolution of the universe is driven by the density of the matter ( described as a fluid of nonrelativistic particles ) .
hence the matter density is much larger than either the curvature term or the density and pressure of radiation .
such an epoch corresponds to the present for a flat universe ( @xmath116 ) , or to the epoch previous to the curvature dominated one in an open universe ( @xmath99 ) .
the zeroth order field equation ( [ eq : h0 ] ) becomes @xmath117 this equation is easily integrated to yield @xmath118 with @xmath119 a constant of integration .
we need the scalar field evolution to first order .
just as we did in the previous section ( [ ssec - cd ] ) , we use the auxiliary field @xmath95 .
when the results ( [ eq : rhom ] ) and ( [ eq : rmd ] ) are substituted in the @xmath95 equation ( [ eq : dpsi ] ) we have @xmath120 by making the variable change @xmath121 we end up with an equation for a damped harmonic oscillator : @xmath122 where primes denote derivatives with respect to @xmath123 .
this equation has three types of solution ( under , over , or critically - damped ) , according to the value of @xmath39 .
( underdamped : @xmath124 ) : @xmath125 \ , .
\label { eq : psimdu}\ ] ] ( criticallydamped : @xmath126 ) : @xmath127 ( overdamped : @xmath128 ) : @xmath129 \ , .
\label { eq : psimdo}\ ] ] in all three cases @xmath130 are constants of integration .
previous to the matter era , there is an epoch in which the density and pressure of relativistic particles ( radiation ) is large compared to the density of the matter ( nonrelativistic fluid ) as well as the curvature term .
this epoch describes the universe fairly well prior to matter - radiation equality . remembering that the equation of state for radiation is @xmath131 , our zeroth order conservation equation ( [ eq : cons0 ] ) becomes @xmath132 which trivially gives @xmath133 with @xmath134 a integration constant .
the subscript @xmath135 specifies that these are the density and pressure of radiation .
taking into account this last result , the expansion factor equation ( [ eq : h0 ] ) to zeroth order is @xmath136 we solve this equation to find @xmath137 to first order we must solve equation ( [ eq : dpsi ] ) for @xmath95 .
the last term in this equation is proportional to @xmath138 , which vanishes if we consider radiation . on the other hand the matter density ( @xmath139 )
also contributes to the total density .
this is a small correction at any particular time in the radiation epoch , but it builds up as we integrate the @xmath95 equation over the whole radiation era
. therefore we must include the effect of the matter density in the scalar field equation .
we find @xmath140 where we have introduced @xmath141 this is a transformed bessel s equation@xcite , which has solution @xmath142 + \frac{b_{3}}{(t + b_{1})^{\frac{1}{2}}}y_{1 } [ 4(a_{2 } \chi)^{\frac{1}{2 } } ( t + b_{1})^{\frac{1}{4 } } ] \ , , \label{eq : psirint}\ ] ] where @xmath143 is a bessel function of the first kind and @xmath144 is a bessel function of the second kind . in addition @xmath145 are constants of integration . without loss of generality we can set @xmath146 . at the time when we match the radiation era solution for @xmath95 to the matter epoch ,
the argument of the bessel functions in equation ( [ eq : psirint ] ) is large .
for @xmath147 we have @xmath148 so setting @xmath146 is consistent at the time of matching . to see whether our assumption is correct we note that the radiation era starts at a time a lot smaller that when it ends .
therefore the matching at the other end might be troublesome because for @xmath149 we have @xmath150 . to see that we do not run into problems we first let @xmath151 so that ( redefining @xmath152 by multiplication by an appropriate constant ) we get @xmath153 if we let @xmath154 be the values of @xmath155 at the point of matching ( call it @xmath156 ) to the epoch previous to the radiation era , we have @xmath157 therefore even if the matching is done at @xmath158 we have @xmath159 at worst .
hence we are justified in taking @xmath146 throughout the radiation epoch .
finally we obtain @xmath142 \ , .
\label{eq : psird}\ ] ] the last era we consider is the one when the electron - positron pairs annhilate . during the early evolution of the universe , as we have mentioned , the cosmological fluid can be approximated as a perfect fluid with a radiation equation of state .
consider some species @xmath160 of particles and antiparticles with mass @xmath161 .
when the temperature of the universe passes through the threshold @xmath162 , the pairs of @xmath160 particles and antiparticles annihilate , creating important though small deviations from the radiation equation of state .
these deviations act as a source for our scalar field , giving it an extra kick .
in particular , we are interested in the thershold for @xmath163 annihilation , @xmath164 k , because it is the only annihilation that occurs after the neutron - proton ratio has frozen out of equilibrium , occuring around the same time that we have synthesis of primordial elements
. we will start the @xmath163 epoch at the time of neutrino decoupling , @xmath165 k. around such time the universe can be safely described as radiation including the behavior of the @xmath163 pairs because they are still relativistic .
we will terminate it when the temperature drops to @xmath166 . at this point
we can once again approximate the cosmological fluid as radiation since , even though the leftover electrons are not relativistic their density and pressure are very small .
therefore we match with the radiation era solution at this time .
as derived in appendix [ app - thrm ] we have the following densities and pressures as functions of temperature : @xmath167 \label { eq : epe } \\
\rho_{e}(t ) \simeq 3p_{e}(t ) + \frac{2 m_{e}^{4}}{\pi ^{4}\hbar ^{3 } } \times [ ( \frac{kt}{m_{e}})k_{1}(\frac{m_{e}}{kt } ) - ( \frac{kt}{2m_{e}})k_{1}(\frac{2 m _ { e}}{kt } ) ] \label { eq : erhoe } \ , .\end{aligned}\ ] ] here @xmath168 is planck s constant divided by @xmath169 , @xmath170 is boltzmann s constant , and @xmath171 is the electron rest mass .
the subscripts @xmath172 refer to photons , neutrino - antineutrino pairs , and electron - positron pairs respectively .
the functions @xmath173 are hyperbolic bessel functions . in appendix [ app - thrm ]
we obtain the following differential equations for the scale factor and time as functions of temperature : @xmath174}{t } \right ) = 0 \label{eq : ert } \\
\frac{dt}{dt } = \frac{\frac{d}{dt } \ln \left[\frac{t}{\rho_{1}(t ) + p_{1}(t ) } \right]}{\sqrt{24 \pi [ \rho_{1}(t ) + \rho_{\nu}(t ) ] } } \label{eq : ett } \
, , \end{aligned}\ ] ] where @xmath175 .
writing equation ( [ eq : dpsi ] ) for @xmath95 using the temperature as the independent variable , we have @xmath176 where @xmath177 are defined by : @xmath178}{\frac{dt}{dt } } \\
f_{2}(t ) \equiv \frac{\rho_{1 } - 3p_{1}}{\left(\frac{dt}{dt } \right)^{2 } } \ , .\end{aligned}\ ] ] the @xmath95 equation was integrated numerically using a fourth order runge - kutta method with adaptive stepsize obtained from @xcite .
the initial conditions for the integration were obtained by matching to the radiation epoch solution at the lower temperature ( @xmath179 ) .
there was no need to solve for the scale factor in the @xmath0 era because the speed - up factor depends only on the scalar field and everywhere the scale factor appears in the differential equation for @xmath95 , it can be replaced by functions of the densities and temperature .
the production of light elements in the early universe depends on the baryon density , the number of light(@xmath180 mev ) neutrinos @xmath181 , the neutrino lifetime @xmath182 , and the speed - up factor @xmath65 in such a way that upper limits on the mass fractions @xmath183 and @xmath184 provide an upper limit on @xmath65 . to be more specific ,
the primordial @xmath185he mass fraction , @xmath183 , is given by@xcite : @xmath186 with @xmath187 .
for the primordial mass fraction of @xmath188he we adopt the conservative upper limit@xcite @xmath189 .
this limit implies the inequality@xcite : @xmath190 combining equations([eq : xhe ] ) and ( [ eq : eta ] ) , we find @xmath191 we use the values @xmath192 and @xmath193 sec .
, which are consistent with experimental constraints .
we take the reasonable limit @xmath194 , which implies @xmath195 we now turn our attention to a flat universe ( zero spatial curvature , @xmath116 ) . for each of these universes we calculated the speed - up factor , which from equation ( [ eq : lxiexp ] ) we know depends only on the value of the scalar field at nucleosynthesis . in a flat universe
, the present epoch can be approximated very well as a matter dominated era .
so our solutions for the scale factor @xmath48 and the auxiliary scalar field @xmath95 are the ones given in subsection [ ssec - md ] . to evaluate some of the constants of integration we impose the appropriate boundary conditions : @xmath196 these conditions determine two of the constants @xmath130 in terms of the remaining one .
moving back in time we arrive at the radiation dominated epoch , which we analyzed in subsection [ ssec - rd ] .
we can take our radiation era scale factor , equation ( [ eq : rrd ] ) , to describe the behavior of the universe up to much higher densities , i.e. , @xmath197 , because deviations from radiation dominance ( such as annihilation epochs ) are short lived and only contribute very small corrections . once again to determine the constants of integration we do the approproate matching at the matter epoch radiation era interface , i.e. , continuity of @xmath198 .
going farther into the past we get to the era of @xmath0 annihilation .
we have the auxiliary field @xmath95 as a function of @xmath39 from the solution in the radiation era .
therefore for different values of @xmath39 , we have different values of @xmath95 at nucleosynthesis after the appropriate matching with the radiation era solution . combining equation ( [ eq : lxiexp ] ) , which expresses @xmath199 in terms of @xmath200 , with the nucleosynthesis constraint equation ( [ eq : xicons ] ) for the speed - up factor we obtain allowed values of @xmath201 which we plot in figure [ fig1 ] along with the solar system constraint ( dotted line ) .
the allowed ranges are the regions under the curves .
we see from equation ( [ eq : lxiexp ] ) that for a given @xmath39 , @xmath202 .
the first thing to note is that throughout most of the range the constraint from nucleosynthesis is more restrictive than the solar system constraint .
one thing that catches the eye are the peaks at @xmath203 .
although they appear finite in the graph , they are infinite ; meaning that at the peaks @xmath38 could have any value ( no constraint at all ) .
this is so because , for these values of @xmath39 , the speed - up factor at nucleosynthesis is unity ( the expansion rate at nucleosynthesis is the same in gr and in st gravity with these values of @xmath72 ) .
hence at the peaks the abundances of primordial light elements do not place any restrictions on @xmath38 .
there will be very small ranges around the peaks where the nucleosynthesis constraint is less restrictive the solar system one ( the first peak is in the region where the solar system constraint is more restrictive anyway ) .
there are values of @xmath39 that produce zeros of @xmath79 .
if there are any deviations ( no matter how small ) from gr at these values in the present , the bounds on primordial light elements will be violated .
most values of @xmath39 are consistent with an upper limit of @xmath204 which implies an upper limit on the ppn parameter @xmath2 of @xmath205 .
the analysis of the open universe ( negative spatial curvature , @xmath99 ) is very similar to that of the flat universe .
the difference between the two is that the present epoch in the open universe is a curvature dominated one , which was analyzed in subsection [ ssec - cd ] .
the epochs previous to the present one are matter dominated , radiation dominated , and a @xmath0 annihilation era , in that order .
we take our open universe to have @xmath206 of the critical density , i.e. , @xmath207 . for the open universe we plot in figure [ fig2 ] the @xmath38 vs @xmath39 constraints .
the same discussion as for the flat case applies here .
we have the upper limits @xmath208 and @xmath209 . the main difference with the flat universe results is that the limits on @xmath38 are 1 to 2 orders of magnitude higher , but still more restrictive than solar system limits over most of the range of @xmath39 .
as mentioned in the introduction , there is motivation to study scalar - tensor theories of gravity from the theoretical point of view as well as a means for comparison with experimental gravity . after noting the difficulty posed by the arbitrariness in the scalar coupling function @xmath1
, we noted that its slope must be small at present in order to satisfy solar system limits on ppn parameters . in order to remove the freedom in the choice of @xmath1
, we used perturbation theory about the value of the the scalar field today ( @xmath22 ) .
as long as @xmath22 is close to a value that minimizes @xmath1 , the universe will have general relativistic behavior at present and peturbation theory will be equivalent to perturbing about gr , having the slope of @xmath1 as the perturbation parameter . in order to find nucleosynthesis constraints on @xmath38 and @xmath39 we used first order perturbation theory because , if higher order correction are important , they would lead to violation of present bounds on primordial light elements . moreover , because of equation ( [ eq : lxi ] ) for the speed - up factor @xmath210 indictes that @xmath211 , we can neglect the back reaction on the metric .
we analyzed the scalar field evolution in flat ( @xmath212)and open universes ( @xmath207 ) to obtain the speed - up factor at nucleosynthesis as a function of @xmath38 and @xmath39 . using the constraint that the abundances of light elements place on @xmath213 , allowed values of @xmath38 as a function of @xmath39
were obtained . in both universes
the constraint was more restrictive ( for @xmath214 ) than the one from solar system experiments , the most restrictive one coming from a flat universe .
this means that if st gravity is the true theory of gravity , it must be practically indistinguishable from gr through a significant period of the universe s history at least for these cosmologies .
d. i. santiago was supported by nasa grant nas8 - 39225 to gravity probe b. d. kalligas was partially supported by the 1995 penned program ( no . 512 ) .
very special thanks to the friday gravity sewing circle at gp - b .
the purpose of the present appendix is to review and rederive some of the well known results@xcite on the densities and pressures of different species as functions of temperature .
differential equations for the temperature and the expansion factor are also obtained . in order to calculate thermodynamic quantities for the different species we need the number density of particles in a energy range @xmath215 .
this quantity is given by @xmath216 where @xmath217 is the number of spin states , @xmath218 is the chemical potential , @xmath219 is the momentum , @xmath220 is the energy , and the @xmath221(@xmath222 ) sign corresponds to fermions(bosons ) . for photons and for neutrinos ( assuming no neutrino degeneracy ) we have @xmath223 . for electrons and positrons
we assume that the difference between their numbers is small compared to their sum , so that we can set @xmath223 for them too ( this approximation breaks down when the density and pressure of electrons and positrons is negligible ) . for photons and neutrino - antineutrino pairs ( assuming three neutrino species )
we obtain @xmath225 where the subscripts @xmath2 and @xmath226 refer to photons and neutrino - antineutrino pairs respectively . for the combined densities of electrons and positrons we have @xmath227 where @xmath228 are hyperbolic bessel functions .
the following will prove useful in some of our next derivations : @xmath229 with @xmath230 . in the epoch of @xmath0 annihilation
the neutrinos have just deocoupled , hence they do not participate in the thermodynamic equilibrium , but are separetely conserved .
the very small baryon contribution is also separetely conserved because at these times they have long been decoupled from the equlibrium .
photons and @xmath0 pairs interact and are in thermodynamic equilibrium .
we have the conservation equation ( [ eq : cons0])(which is valid to first order in the perturbation expansion of the present paper ) : @xmath231 for neutrinos and for baryons ( which we take to be pressureless ) the conservation equation leads to @xmath232 with the subscripts @xmath233 refering to neutrinos and baryons respectevely . for the interacting species ,
@xmath0 pairs and photons , we define the following densities and pressures : @xmath234 substituting these last expressions into the conservation equation and using the equality @xmath235 we obtain @xmath236 the last equation can be written in the much more compact form @xmath237}{t}\right ) \ , .\ ] ] if we combine this result with the field equation ( [ eq : h0 ] ) for the scale factor we end up with @xmath238}{\sqrt{24 \pi \rho } } \ , , \ ] ] where @xmath239 is the total density .
99 p. jordan , nature * 164 * , 637 ( 1949 ) ; z. phys . * 157 * , 112 ( 1959 ) .
m. fierz , helv .
acta * 29 * , 128 ( 1956 ) . c. brans and r. h. dicke , phys . rev . *
124 * , 925 ( 1961 ) .
p. g. bergmann , int .
. phys . * 1 * , 25 ( 1968 ) .
k. nordtvedt , astrophys .
j. * 161 * , 1059 ( 1970 ) .
r. v. wagoner , phys . rev .
d * 1 * , 3209 ( 1970 ) . t. damour and esposito - farse , class .
quantum grav .
* 9 * , 2093 ( 1992 ) .
m. b. green , j. h. shwarz , and e. witten , _ superstring theory vol.1 and 2 _ ( cambridge , 1987 ) p. j. steinhardt and f. s. acceta , phys . rev
. lett . * 64 * , 2740 ( 1990 )
. s. buchmann , _ et al .
_ , in _ proccedings of the seventh marcel grossmann meeting on general relativity _
( world scientific , 1996 ) .
a. abramovici , _ et al .
_ , science * 256 * , 325 ( 1992 ) .
j. hough , in _ proceedings of the sixth marcel grossmann meeting on general relativity _
( world scientific , 1992 ) . c. bradaschia , _ et al .
. instrum . and methods a * 289 * , 518 ( 1990 ) .
c. callan _ et al .
b * 262 * , 593 ( 1985 ) .
t. damour and a. m. polyakov , nucl .
b * 423 * , 532 ( 1994 ) ; gen .
* 26 * , 1171 ( 1994 ) . t. damour and k. nordtvedt , phys .
d * 48 * , 3436 ( 1993 ) . c. m.
will , _ theory and experiment in gravitational physics , _ revised edition _ _ ( cambridge , 1993 ) .
r. d. rosenberg _ et al .
_ , astrophys
. j. * 234 * , l209 ( 1979 ) .
e. b. fomalot and r. a. sramek , phys .
lett . * 36 * , 1475 ( 1976 ) .
k. nordtvedt , in _ proccedings of the seventh marcel grossmann meeting on general relativity _
( world scientific , 1996 ) .
j. a. casas , j. garca - bellido , and m. quiros , mod .
lett . a * 7*,447 ( 1985 )
d. kalligas , k. nordtvedt , and r. v. wagoner , in _ proccedings of the seventh marcel grossmann meeting on general relativity _ ( world scientific , 1996 ) .
a. serna and j. m. alimi , phys .
d * 50 * , 7304 ( 1996 ) .
r. v. wagoner and d. kalligas , in _ gravitation and gravitational radiation , les houches 1995 _ ( cambridge , 1996 ) . c. w. misner , k. s. thorne , and j. a. wheeler , _ gravitation _ ( freeman , 1973 ) .
r. v. wagoner , astrophys . j. * 179 * , 343 ( 1973 ) .
r. v. wagoner in _ experimental gravitational physics _ ( world scientific , 1988 ) .
w. h. beyer _ standard mathematical tables , _
25th edition _ _ ( crc press , 1978 ) .
w. h. press _ et .
_ , _ numerical recipes in fortran _ second edition _ _ ( cambridge , 1992 )
. k. a. olive _ et .
b * 36 * , 454 ( 1990 ) .
r. a. alpher , j. w. follin jr . , and r. c. herman , phys . rev . * 92 * , 1347 ( 1953 )
r. v. wagoner , w. a. fowler , and f. hoyle , astrophys . j. * 148 * , 3 ( 1967 ) .
|
we study the cosmological evolution of massless single - field scalar - tensor theories of gravitation from the time before the onset of @xmath0 annihilation and nucleosynthesis up to the present . the cosmological evolution together with the observational bounds on the abundances of the lightest elements ( those mostly produced in the early universe ) place constraints on the coefficients of the taylor series expansion of @xmath1 , which specifies the coupling of the scalar field to matter and is the only free function in the theory . in the case when @xmath1 has a minimum ( i.e. , when the theory evolves towards general relativity ) these constraints translate into a stronger limit on the post - newtonian parameters @xmath2 and @xmath3 than any other observational test .
moreover , our bounds imply that , even at the epoch of annihilation and nucleosynthesis , the evolution of the universe must be very close to that predicted by general relativity if we do not want to over- or underproduce @xmath4he .
thus the amount of scalar field contribution to gravity is very small even at such an early epoch . #
|
sgr a east is an elliptical shell structure elongated along the galactic plane with a major axis of length @xmath9 pc and a center displaced from the apparent dynamical nucleus , sgr a west , by @xmath10 pc in projection toward negative galactic latitudes .
the latter is a ( @xmath11 pc ) three - arm spiral configuration of ionized gas ( ekers et al . 1983 ; lo & claussen 1983 ) that engulfs a compact nonthermal radio source , sgr a * , having a dimension between 1.1 au and 0.1 au ( morris & serabyn 1996 ) .
the bulk of the thermally emitting sgr a west appears to be located in front of sgr a east ( yusef - zadeh & morris 1987 ; pedlar et al . 1989 ) , whose exact distance behind the nucleus is not known , but a number of arguments suggest that it is positioned very close to the galactic center ( e.g. , gsten & downes 1980 ; goss et al .
1989 ; yusef - zadeh et al . 1998 ) .
thus , if present , the massive black hole ( presumably associated with sgr a * ) lies within the sgr a east shell , but at one extreme end of its elongated structure .
sgr a east may be a supernova remnant ( snr ) , but its inferred energetics ( @xmath0 ergs ; mezger et al .
1989 ) and size appear to be extreme and have generated some uncertainty regarding this interpretation .
the explosion that produced sgr a east may have been the tidal disruption of a main sequence star whose trajectory took it within ten schwarzschild radii of the central object ( khokhlov & melia 1996 ) .
however , regardless of what actually caused the initial explosion , recent observations of this region at 1720 mhz ( the transition frequency of oh maser emission ) have revealed the presence of several maser spots on the remnant s boundary ( yusef - zadeh et al .
this is important because oh masers can be effective tracers of shocks produced at the interface with molecular clouds , particularly for supernova remnants ( frail et al .
1994 ; wardle , yusef - zadeh & geballe 1998 ) .
radio continuum observations of this object at @xmath12 and @xmath13 cm ( pedlar et al .
1989 ) reveal that these sites are also nonthermal emitters , but if this is due to synchrotron , then a relativistic electron distribution with @xmath14 is required , where @xmath15 .
this is rather peculiar for several reasons .
first , such a particle spectrum is unlike that ( @xmath16 ; jones & ellison 1991 ) thought to be produced directly in shock acceleration .
second , this situation appears to be different from that of other snrs in which @xmath17 ( e.g. , gaisser et al .
1998 ) .
this latter distinction is very noticeable in view of the kinship between sgr a east and other snrs suggested by egret observations ( see e.g. , green 1995 ; esposito et al .
1996 ; saken et al . 1992 ; lessard et al .
the galactic center is very complex and the @xmath3-rays from this region may be produced by other means .
for example , the egret @xmath3-ray source 2egj1746 - 2852 may be the black hole candidate sgr a * ( markoff , melia & sarcevic 1997 ; mahadevan et al .
1997 ) or it may be due to a process different than what is producing the @xmath3-ray emission in the snrs ( melia , yusef - zadeh & fatuzzo 1998 ; see below ) . our goal in this _
letter _ is to demonstrate in a self - consistent manner how the radio through tev spectrum of sgr a east may be produced with a single process , and how it can be placed in context with the other egret snrs when one takes into account the differences between the ambient medium surrounding sgr a east and that of the other remnants .
egret has identified a central ( @xmath18 ) @xmath19 mev - @xmath20 gev continuum source with luminosity @xmath21 ergs s@xmath22 ( mayer - hasselwander et al .
its @xmath3-ray flux does not vary down to the instrument sensitivity ( roughly a @xmath23 amplitude ) , suggesting that it may be diffuse ( see , however , markoff , melia & sarcevic 1997 ) . in a recent study ( melia , yusef - zadeh & fatuzzo 1998 )
, we showed that the @xmath3-ray characteristics of 2egj1746 - 2852 may be understood as arising from inverse compton scatterings between the power - law e@xmath24 ( inferred from the radio data ) and the uv and ir photons from the nucleus .
this is facilitated by the unique environment at the galactic center , where the inner @xmath25 pc region is a source of intense ( stellar ) uv and ( dust - reprocessed ) far - ir radiation that bathes the extended synchrotron - emitting structures .
however , this analysis left several questions unanswered .
for example , it is rather suspicious that the galactic center @xmath3-ray source has a similarly distinctive bump at @xmath26 gev as the five snrs detected by egret ( see , e.g. , esposito et al . 1996 ) . in their detailed calculations of the emission in the snr shells of ic 443 and @xmath3 cygni , gaisser , protheroe and stanev ( 1998 )
included the contribution to ic from microwave background radiation , the galactic ir / optical background radiation and the ir resulting from shock - heated dust .
at least in these snrs , the dominant contributions to the @xmath3-ray spectrum come from bremsstrahlung ( by relativistic electrons colliding with the ambient proton distribution ) and pion decay , the latter constituting the aforementioned @xmath27 mev characteristic bump .
thus , if 2egj1746 - 2852 is the @xmath3-ray counterpart to the radio emitting shell in sgr a east , one might expect that its @xmath3-ray bump could also be due to pion decays . at the same time , an observationally significant pion production rate in sgr a east also offers the exciting possibility that the @xmath28 pion decay byproducts may produce the @xmath29 power - law distribution inferred from sgr a east s radio spectrum , and these particles then also self - consistently account for the underlying @xmath3-ray spectrum via either bremsstrahlung , or the ic scattering of the central ir and uv photons .
we work with the hypothesis that the egret @xmath3-ray bump is due primarily to @xmath4 decays ( @xmath30 ) , these being produced in @xmath31 scatterings between ambient protons and a relativistic population accelerated by shocks within the sgr a east shell . in this process ,
a multiplicity of pions is produced , the charged members of which decay leptonically ( @xmath32 , with @xmath33 ) .
we first find the physical conditions required to produce this @xmath3-ray bump , and then use the associated electrons and positrons produced in the pion decays to calculate the radio spectrum . fitting the synchrotron emission to the vla data ( see fig .
3 ) is then a matter of adjusting just one unknown , viz .
the average magnetic field @xmath34 in the gyration region , though this is expected to be close to equipartition .
a full description of our @xmath31 scattering algorithm is given elsewhere ( markoff , melia & sarcevic 1997 , 1998 ) .
briefly , we assume that the accelerated protons are accelerated at the shock with a rate of @xmath35 @xmath36 s@xmath22 gev@xmath22 .
the steady - state proton spectral index is thought to lie in the range of @xmath37 ( jones & ellison 1991 ) , and is determined from a diffusion loss treatment . because the protons are ultrarelativistic , the leading order nucleons produced will continue contributing to the spectrum via multiple collisions in an ensuing cascade , until they lose enough of their energy to rejoin the ambient plasma .
the procedure for calculating the pion production rate is complicated by the energy dependence of the pion multiplicity and the @xmath31 scattering cross section .
fortunately , uncertainties in the @xmath31 scattering are likely to be smaller than @xmath38 , within the possible errors of the astrophysical components , such as the source size .
we calculate the steady - state proton distribution resulting from the balance of injection from the shock , cascade collisions producing secondary particles which feed back into the system , and synchrotron and inverse compton losses .
we note that proton - photoproduction processes are here relatively insignificant , since the maximum energy attained by the relativistic protons in the face of energy losses due to @xmath31 scatterings and other processes is about @xmath39 gev , close to the pair creation threshold for protons scattering off uv radiation ( see below ) . in figure 1
, we show the @xmath3-ray spectrum produced from @xmath4 decays for two different cases , characterized by their respective values of @xmath40 and the ambient proton density @xmath41 .
of course , the full spectrum will be the sum of this component and those arising from bremsstrahlung and inverse compton scattering , which we consider below .
however , we can already see that the fit to the @xmath3-ray data between @xmath42 mev and @xmath20 gev is rather robust over a range of conditions .
what does change from case to case is the energy content in the relativistic proton population , which is related to the efficiency introduced above .
so , for example , when the ambient proton density @xmath41 is low ( case 2 ) , the removal rate of protons from the relativistic distribution is smaller than that in case 1 , resulting in a higher equilibrium density of relativistic protons .
this is why the energy content of these particles is much higher in case 2 than in case 1 , accounting for roughly @xmath43 and @xmath44 of a typical sn burst of @xmath45 ergs for these two situations , respectively .
the electron distribution resulting from these two cases is shown in figure 2 .
the @xmath46 and @xmath47 distributions can mimic a power - law with a spectral index @xmath7 very well , consistent with the value implied by the radio spectrum of sgr a east .
apparently , the leptons produced in @xmath5 decays in this source are numerically dominant over those accelerated directly within shocks .
in fact , assuming the same spectral index for the electrons and protons , the shock would have to accelerate roughly @xmath48 as many of the former compared to the latter at any given energy in order to account for the synchrotron emissivity without the decay leptons .
this imbalance comes about because ( 1 ) each relativistic proton produces an average of @xmath49 electrons and ( 2 ) a shock accelerated electron distribution would contain a much larger contribution from low energy electrons than the distribution shown in figure 2 , which shows a deficiency at small @xmath3 .
these relativistic electrons are immersed in an intense source of uv and ir radiation from the central @xmath25 parsecs .
the circumnuclear disk ( cnd ) is a powerful source ( @xmath50 ) of mid to far - infrared continuum emission with a dust temperature of @xmath51 k ( e.g. , telesco et al .
1996 ; davidson et al .
this radiation is due to reprocessing by warm dust that has absorbed the same power in the uv ( becklin , gatley and werner 1982 ; davidson et al .
models of the photodissociation regions in the cnd require an incident flux ( @xmath13 ev @xmath52 ev ) of @xmath53@xmath54 erg @xmath55 s@xmath22 ( wolfire , tielens & hollenbach , 1990 ) , implying a total uv luminosity of about @xmath56 .
this radiation field has a specific photon number density per solid angle @xmath57 , where @xmath58 , and @xmath59 . here , @xmath60 is the lab - frame photon energy and @xmath61 and @xmath62 are , respectively , the temperature ( assumed to be @xmath63 k ) of the stellar uv component and of the reprocessed cnd radiation , which is assumed to peak at @xmath64 m , corresponding to a characteristic temperature @xmath65 k. the normalization constants @xmath66 and @xmath67 incorporate the dilution in photon number density as the radiation propagates outwards from the central core .
this is calculated assuming that the radiation emanates from a sphere of radius @xmath68 pc for the uv , and that the ir luminosity is @xmath69 from a disk with radius @xmath70 pc .
the calculation of the inverse compton ( x - ray and @xmath3-ray ) emissivity for this known @xmath28 population bathed by the ir and uv radiation is carried out according to what is now a standard procedure ( see , for example , melia , yusef - zadeh & fatuzzo 1998 ) .
the upscattered radiation is emitted isotropically from within a volume @xmath71 pc@xmath72 in sgr a east , corresponding to a shell with radius @xmath73 pc and thickness @xmath74 pc .
the centroid of this structure is assumed to lie at @xmath75 pc behind the nucleus .
the remaining continuum component of importance is bremsstrahlung radiation resulting from the interaction between the relativistic leptons and the ambient ( fixed target ) nuclei ( see koch & motz 1959 ) .
thus , the overall spectrum from sgr a east is a superposition of the @xmath3-rays from @xmath4 decays , synchrotron radiation by the relativistic leptons produced during the decay of the charged pions , bremsstrahlung emission by these electrons and positrons , and their comptonization of the ir and uv radiation from the central @xmath25 pc .
the key result of our calculations is displayed in figure 3 for case 2 above .
the inferred magnetic field is @xmath76 gauss , which is within a factor of @xmath77 of the equipartition value of @xmath34 in sgr a east , assuming a radio luminosity of @xmath78 ergs s@xmath22 arising from relativistic leptons gyrating in a magnetic field over a volume of @xmath79 pc@xmath80 .
the asca measurement is an upper limit , since it does not distinguish between thermal and nonthermal emission from this region .
our predicted hard x - ray spectrum is also consistent with ginga observations , which show evidence of a hard tail in the diffuse x - ray spectrum from the inner region of the galaxy ( yamasaki et al .
1996 ) , and with the current osse upper limits ( purcell et al . 1997 ) . at the highest energy measurement to date , a 270 minute exposure by the whipple observatory produced a small excess ( @xmath8 ) of @xmath81 tev emission from within the egret error box of 2eg j1746 - 2852 , yielding an integrated flux upper limit of @xmath82 @xmath55 s@xmath22 ( buckley et al .
although this signal is at best only @xmath8 above the noise , our calculated spectrum would be consistent with a detection at this level .
case 1 would not produce an acceptable spectrum because the larger ambient proton density ( @xmath83 @xmath36 ) results in a significantly higher lepton bremsstrahlung flux .
the spectral shape would then also not be a good fit to the data since the @xmath3-rays from @xmath4 decays would be swamped by the other radiative components . in fact , we found it very difficult to fit this broadband spectrum with any ambient density in excess of @xmath84 @xmath36 .
this raises an interesting issue , since the energy content in the relativistic protons ( see above ) is then a large fraction ( @xmath85 ) of the energy in a `` standard '' sn burst ( @xmath86 ergs ) . on the other hand
, this relativistic energy content would constitute only @xmath87 of the available energy if the burst that produced the sgr a east remnant was @xmath0 ergs , as suggested by other observations ( see above ) .
our results may therefore be somewhat consistent with the supposition that sgr a east was produced not by a single , standard supernova explosion , but rather by several such events , or a single more catastrophic incident , such as the tidal disruption of a star . in both cases 1 and 2
we adopted a value of @xmath40 within a narrow range close to @xmath77 .
we found our fit to be very sensitive to @xmath40 .
even with the large error bars associated with the vla points , a proton distribution steeper than this produces a steeper lepton distribution , whose synchrotron emissivity does not fit the radio data .
a shallower proton index is , of course , uncomfortably below the theoretical range .
if the @xmath3-rays detected by egret are indeed produced in this fashion , it seems inevitable to us that the upscattered ir and uv photons should result in a significant intensity at intermediate ( i.e. , @xmath88 kev ) energies , as indicated in figure 3 . this flux density ( @xmath89 photons
@xmath55 s@xmath22 mev@xmath22 ) may be above the sensitivity limit ( at @xmath90 photons @xmath55 s@xmath22 mev@xmath22 for a point source ) of unex , a proposed balloon flight instrument ( rothschild 1998 ) .
other than the overall energetics , it appears that the main distinction between sgr a east and the snrs is the intensity of the external radiation upscattered by the relativistic electrons and positrons . whereas the ambient soft photon intensity in the latter is too low to produce an observationally significant spectral component compared to bremsstrahlung ( gaisser , protheroe & stanev 1998 ) ,
the ir and uv radiation from the galactic center does produce a dominant broadband continuum when scattering with the relativistic leptons in sgr a east .
|
sgr a east is a very prominent elongated shell structure surrounding ( though off - centered from ) the galactic nucleus .
its energetics ( @xmath0 ergs ) , based on the power required to carve out the radio synchrotron remnant within the surrounding dense molecular cloud , appear to be extreme compared to the total energy ( @xmath1 ergs ) released in a typical supernova ( sn ) explosion .
yet it shares several characteristics in common with sn remnants ( snrs ) , the most significant of which is the @xmath2 gev @xmath3-ray spectrum measured by egret , if we associate the galactic center source 2egj1746 - 2852 with this nonthermal shell . we here show that the highest - energy component in sgr a east s spectrum , like that of snrs , can be fitted with the @xmath3-rays produced in @xmath4 decays .
further , we demonstrate in a self - consistent manner that the leptons released in the associated @xmath5 decays produce an @xmath6 distribution that can mimic a power - law with index @xmath7 , like that inferred from the vla data for this source .
these relativistic electrons and positrons also radiate by bremsstrahlung , and inverse compton scattering with the intense ir and uv fields from the nucleus .
we show that the overall emissivity calculated in this way may account for sgr a east s broadband spectrum ranging from ghz frequencies all the way to tev energies , where whipple has thus far set an upper limit to the flux corresponding to a @xmath8 signal above the noise .
3gcm^-3 3cm^-3 # 1#1 #1#2#3#1 , a&a , * # 2 * , # 3 .
# 1#2#3#1 , a&a _ ( letters ) _ , * # 2 * , # 3 .
# 1#2#3#1 , a&as , # 2 , # 3 # 1#2#3#1 , aj , # 2 , # 3 # 1#2#3#1 , _ apj _ , * # 2 * , # 3 .
# 1#2#3#1 , _ apj ( letters ) _ , * # 2 * , # 3 .
# 1#2#3#1 , apjs , # 2 , # 3 # 1#2#3#1 , ara&a , # 2 , # 3 # 1#2#3#1 , baas , # 2 , # 3 # 1#2#3#1 , icarus , # 2 , # 3 # 1#2#3#1 , _ mnras _ , * # 2 * , # 3 .
# 1#2#3#1 , _ nature _ , * # 2 * , # 3 .
# 1#2#3#1 , pasj , # 2 , # 3 # 1#2#3#1 , pasp , # 2 , # 3 # 1#2#3#1 , qjras , # 2 , # 3 # 1#2#3#1 , science , # 2 , # 3 # 1#2#3#1 , soviet astr .
, # 2 , # 3 # 1#2#3#1 , soviet astr .
lett . , # 2 , # 3 # 1#2#3#4#1 , # 2 , # 3 , # 4 submitted to the editor of the astrophysical journal letters . 0.5 in
|
following gai and kapadia@xcite , a _ financial crisis _ is defined as a situation in which at least 5% of banks go bankrupt .
fig.1 illustrates the frequency and the conditional average of the size of crises .
some interesting properties are observed .
first , fig.1a reveals that the uniform immunization strategy based on pagerank is most successful in terms of reducing the frequency of crises .
the least efficient one is random immunization .
the reason for this is that a bank with a high pagerank score generally a ) has many interbank liabilities ( i.e. , high in - degree ) and b ) borrows from lenders that have few interbank assets ( i.e. , low out - degree ) . in this model
, the fact that a bank borrows a lot means that the bank borrows from many other banks ( see methods ) .
such a bank is likely to be systemically important because its failure would undermine the balance sheets of many lenders .
in addition , the fact that the lenders of bank @xmath6 have few interbank assets indicates that the lenders solvency depends largely on the solvency of bank @xmath6 since the lenders portfolios of interbank assets are not so diversified .
secondly , fig.1b states that the average number of banks that fail during a crisis increases as more banks become immunized .
this is because a larger number of banks begin to hold a common low - risk external asset , which reduces the frequency of medium - scale simultaneous defaults ( fig.2a ) .
moreover , fig.1b also shows that a crisis is likely to become more severe under the pagerank - based strategy than under the other immunization strategies .
immunizing banks in the order based on pagerank efficiently increases the probability that systemically important banks will fail simultaneously , which would lead to a crisis .
in other words , the pagerank - based strategy will rapidly shift the financial market toward a _ robust - yet - fragile _ system .
histogram of the number of defaulted banks . * ( a ) uniform immunization based on pagerank .
( b ) counteractive immunization based on edge - betweenness centrality .
the figures show that i ) uniform immunization decreases the frequency of medium - scale defaults while increasing the likelihood of extremely large - scale defaults .
ii ) counteractive immunization slightly reduces the frequency of medium - scale cascades while increasing the likelihood of small - scale defaults .
see the caption of fig.1 for parameter values.,width=566 ] the corresponding figures for counteractive immunization are shown in figs.1c , d .
counteractive immunization introduces anti - correlations between two assets that were previously independent of each other without changing the riskiness of individual assets .
in contrast to the case of uniform immunization , counteractive immunization decreases the frequency of crises while having an unambiguous influence on the average number of bank defaults at the time of a crisis .
assigning negatively correlated assets to pairs of connected banks will avoid medium - scale default cascades by preventing the balance sheets of the immunized banks from moving in sync ( fig.2b ) .
however , the speed at which the frequency of a crisis decreases is much slower under counteractive immunization than under the pagerank - based uniform immunization .
this implies a trade - off between frequency and severity .
the desirability of an immunization strategy thus depends on the extent of social risk aversion , represented by parameter @xmath0 . in the following ,
we clarify how the level of systemic risk under various types of immunization strategies depends on the degree of risk aversion , @xmath0 .
the relationship between the fraction of immunized banks and the expected costs . *
( a)-(d ) : the case of uniform immunization .
the increases in the expected costs observed in ( a ) reflect the de - diversification effect .
the expected costs tend to become upward - sloping as the value of @xmath0 increases .
this reflects the fact that the society is becoming more risk - averse .
( e)-(h ) : the case of counteractive immunization . the outcome of uniform immunization based on a random vaccination order is also shown for comparison .
the shape of the curve is little affected by the value of @xmath0 under counteractive immunization .
see the caption of fig.1 for parameter values.,width=585 ] fig.3 depicts the relationship between the fraction of immunized banks and the expected costs under four different values of @xmath0 .
some important features should be noted .
first , under uniform immunization , the expected costs may increase even if @xmath7 , where the expected cost is equal to the average number of defaults .
an intuition into why the average number of defaults increases is that a homogenization of banks external assets effectively makes the portfolio of interbank assets less diversified .
if every bank has an idiosyncratic external asset , then the holdings of multiple interbank assets contribute to portfolio diversification .
however , if borrowers begin to hold the same external assets as their lenders , then the effective degree of the lenders portfolio diversification will be lowered .
such an effect , what we call the _ de - diversification effect _ , may increase the average size of default cascades even when the risk of external assets is diminishing . on the other hand
, asset harmonization itself also has an effect of decreasing contagious defaults .
this is because asset harmonization increases the average number of banks that would fail simultaneously due to the loss of external assets only .
thus , if the riskiness of the common asset is sufficiently low , then asset harmonization would reduce the expected costs .
secondly , the expected costs are more likely to become upward - sloping as @xmath0 increases .
this is because the degree of convexity of the cost function increases as @xmath0 goes up , reflecting the fact that the society is becoming more risk - averse . for large values of @xmath0 , the possibility of a large - scale simultaneous default turns out to be an important source of risk even though its probability of occurrence is quite low .
in contrast , figs.3e - h show that counteractive immunization reduces systemic risk virtually monotonically .
this is because counteractive immunization can lower the likelihood of medium - scale default cascades without causing increased asset commonality .
a limitation of this strategy is that it does not change the riskiness of a single external asset itself .
in fact , assigning anti - correlated assets to a pair of isolated banks can slightly lower the probability that both of the banks will survive . to demonstrate this ,
suppose that the probability of default is @xmath8 for each bank and that the correlation coefficient of the asset returns is -1 .
the probability that both banks will be solvent is then @xmath9 .
if asset returns are independent , then the corresponding probability is given by @xmath10 . since @xmath11 @xmath12
, it follows that counteractive immunization reduces the probability that both banks will survive .
this is why counteractive immunization may slightly increase the expected costs in some circumstances .
the frequency with which the gscc of the financial network disintegrates . *
( a)-(c ) : uniform immunization .
( d ) and ( e ) : counteractive immunization .
restrepo et al.s method is used to judge whether the gscc is present or not.@xcite the frequency of network disintegration is correlated closely with the expected costs under @xmath13 .
see the caption of fig.1 for parameter values.,width=604 ] fig.4 depicts the frequency with which the gscc of the interbank network disintegrates due to bank defaults ( supplementary figure s1 shows the simulated percolation threshold ) .
it turns out that uniform immunization tends to decrease the frequency of disintegration , but there is a region in which the likelihood of disintegration is increased .
this phenomenon is observed when the de - diversification effect is most profound .
under counteractive immunization , on the other hand , the frequency of disintegration decreases virtually monotonically as vaccination proceeds .
this is because counteractive immunization makes it possible to reduce the frequency of medium - scale default cascades without causing the de - diversification effects ( figs.1c , d ) .
we also found that the frequency of disintegration follows a trajectory of the expected costs in the case of @xmath14 .
this suggests that the risk of network disintegration could be well captured by the third moment of the number of failed banks .
an intuitive explanation for this coincidence is as follows .
recall that the value of @xmath0 represents the degree of risk aversion , which indicates to what extent simultaneous defaults are more costly than a sequence of single defaults .
if @xmath0= 1 , then there is no difference between a simultaneous default of multiple banks and a sequence of single defaults as long as the total number of defaulted banks is the same .
this implies that the society puts an identical weight on various sizes of collective defaults , which means that the small - scale collective defaults that would not matter for the existence of the gscc are cared too much . when @xmath0 = 10 , on the other hand , the society is so risk averse that it cares about infrequent but extremely large - scale simultaneous defaults .
the society thus undervalues the costs of the medium - scale collective defaults that would still exceed the percolation threshold .
only if the value of @xmath0 is just around 3 , the society puts appropriate weights on the size of collective defaults that would be larger than the percolation threshold .
the expected cost under uniform immunization relative to that under counteractive immunization . * for each immunization strategy , expected costs are minimized with respect to the fraction of immunized banks . (
a ) , ( b ) and ( c ) : asset returns follow a normal distribution .
( d ) , ( e ) and ( f ) : asset returns follow the t - distribution with degree of freedom 5 . unless otherwise noted , uniform immunization is based on pagerank while counteractive immunization is based on a random order .
baseline parameters are : @xmath15 , @xmath16 and @xmath17 .
the optimal immunization strategy is to employ uniform immunization when @xmath0 is small and counteractive immunization when it is large .
the threshold value of @xmath0 depends on model parameters.,width=585 ] the next question is how to choose an immunization strategy .
fig.5 illustrates the expected costs under uniform immunization relative to that under counteractive immunization .
the relative cost depends on various aspects of the model , such as distribution of asset returns , immunization strategy , the degree of asset correlation and the riskiness of the common low - risk asset .
the main results can be summarized as follows .
first , it turns out that uniform immunization is more likely to be preferable as the tail of asset - return distribution becomes thicker .
second , a decrease in @xmath3 will move the threshold value of @xmath0 rightward , which means that a counteractive immunization becomes less attractive .
third , the degree of correlation between anti - correlated assets may have an ambiguous impacts on the efficiency of counteractive immunization .
fig.5e illustrates that counteractive immunization is most attractive when @xmath15 , while this is not the case if asset returns follow a normal distribution .
a general and robust result is that uniform immunization becomes more efficient than counteractive immunization when @xmath0 is sufficiently small , where the degree of social risk aversion is moderate .
counteractive immunization becomes more advantageous when @xmath0 is large , where the possibility of large - scale simultaneous defaults is of great concern to the society .
since systemic risk can be thought of as a fear of default contagion that might occur in financial markets , the previously proposed immunization strategies seem useful for preventing financial contagion
. however , a straightforward application of the previous immunization strategies would not work if there is some possibility of serious side effects that would prevail among vaccinated individuals . in the case of uniform immunization
, there is a good possibility that a large number of banks will fail simultaneously due to the commonality of assets .
we showed that counteractive immunization can be more useful than uniform immunization as long as the society has a strong aversion to such serious side effects .
the highest priority for the highly risk - averse society is to avoid severe financial crises that would rarely occur , rather than a few bank failures that would occur with high probability .
negatively correlated assets can become useful vaccines even if the efficacy of each vaccine is lower than that of the common vaccine that is used under uniform immunization .
these results reinforce the idea that micro - prudential policies do not necessarily add up to a macro - prudential policy.@xcite from a micro point of view , it is efficient to use a common low - risk asset when immunizing banks . from a macro point of view , however , the best immunization strategy may be to prevent a large - scale contagion at the sacrifice of individual failures .
the effectiveness of counteractive immunization suggests that financial networks should be _ fragile - yet - robust _ rather than _ robust - yet - fragile_.
there is no wide agreement about how to calculate the level of systemic risk .
there are many candidates , such as the expected number of defaulted banks , the expected economic losses that would be caused by bank failures , and the likelihood of a large - scale default.@xcite we show two intrinsically different risk measures in order to better understand what is happening in the financial network .
the first one is based on the costs that would result from bank failures . following beale et al.@xcite and kobayashi@xcite , we assume that social costs take the form @xmath18 , where @xmath19 is the number of failed banks . the expected cost is thus given as @xmath20 = \sum_{n=1}^n q(n ) n^\chi , \ ] ] where @xmath21 ( = 1000 ) is the total number of banks in the financial market and @xmath22 is the probability function .
@xmath0 can be interpreted as the degree of social risk aversion .
notice that if @xmath23 , this is simply the average number of defaulted banks .
if @xmath24 , however , simultaneous defaults of multiple banks become more costly than a sequence of single bank failures .
another measure is the connectivity of the network . as a measure of connectivity
, we use the simulated probability that the gscc of the interbank network disintegrates . in each simulation
, we check whether the gscc is kept connected after removing the failed banks from the network .
we use the method proposed by restrepo et al.@xcite , who showed that the largest eigenvalue of the adjacency matrix becomes smaller than 1 if the gscc is absent as long as the network has locally treelike structure . in both of the risk measures ,
the size of failed banks is not explicitly taken into account .
however , this does not necessarily mean that systemic importance is the same across banks .
the extent of infectivity generally varies from bank to bank , depending not only on the amount of borrowings ( i.e. , in - degree ) , but also on the topological location in the network and the whole network structure.@xcite systemic importance of individual banks will differ if their infectivity differs even when their balance - sheet sizes are the same .
the systemic risk measures used in this study allow us to treat the systemic importance of each bank in an agnostic manner .
interbank markets are directed networks .
many studies show that the degree distribution of an interbank network follows a power law.@xcite in the present study , the algorithm that generates an interbank network follows from chung et al.@xcite .
their algorithm is also used by restrepo et al.@xcite .
let @xmath25 and @xmath26 denote the in - degree and out - degree of a bank , respectively .
the in - degree of bank @xmath6 represents the number of banks from which bank @xmath6 borrows funds . the degree distribution derived by chung
et al.s algorithm takes the form @xmath27 we assume that there is no degree - degree correlation .
chung et al.s algorithm proceeds as follows .
i ) gven the parameters @xmath28 , @xmath29 and @xmath30 , express the bank @xmath31s in - degree as @xmath32 for @xmath33 $ ] .
ii ) the out - degree distribution is obtained as a random permutation of the sequence of @xmath34 .
iii ) nodes @xmath35 and @xmath36 are linked with probability @xmath37 , where @xmath38 is the average in- or out - degree given by the researcher .
iv ) by using the condition that the total number of edges is equal to @xmath39 , it can be shown that @xmath40dn^{1/(\beta -1)}$ ] and @xmath41^{\beta -1}$ ] for a sufficiently large @xmath21 , where @xmath42 denotes the maximum in - degree given by @xmath43 .
@xmath42 and @xmath38 must be determined in such a way that the probability of placing an edge is less than 1 .
the parameter values follow from restrepo et al.@xcite ( except for @xmath42 ) : @xmath44 , @xmath45 and @xmath46 . to eliminate the possibility that differences in the obtained results might be due to differences in the network structure , we use one particular network structure in all simulations . prior to immunization ,
the asset side of bank @xmath6 s balance sheet consists of interbank assets , @xmath47 , a riskless asset , @xmath48 , and a risky external asset , @xmath49 , where @xmath50 for @xmath51 the liability side of bank @xmath6 s balance sheet consists of interbank liability , @xmath52 , deposits , @xmath53 , and net worth , @xmath54 .
the balance sheet condition implies that @xmath55 = @xmath56 , @xmath57 . the amount of bank @xmath6 s borrowings from bank @xmath58 is expressed as @xmath59 , where @xmath60 denotes the relative weight of bank @xmath6 s borrowings from @xmath58 , and thereby @xmath61 .
the amount of bank @xmath6 s total interbank assets , @xmath62 , is given by @xmath63 bank @xmath6 will default if @xmath64 where @xmath65 and @xmath66 stand for the ex - post values of external assets and interbank liabilities , respectively . it should be pointed out that deposits , @xmath53 , are reserved because deposits are senior to interbank assets .
provided that there is no loss in the interbank assets , those banks that hold a common low - risk asset will fail with probability @xmath3 while the others will default with probability @xmath1 .
following beale et al.,@xcite asset returns are assumed to follow students t - distribution with degree of freedom @xmath67 for the low - risk asset and @xmath68 for the other external assets .
for those banks with out - degree 1 , the size of net worth , @xmath69 , is determined such that @xmath70 or @xmath71 , where @xmath72 is the cdf of student - t distribution with degree of freedom @xmath68 .
the unit size of the interbank asset , @xmath73 , is given as @xmath74 , where @xmath75 is the predetermined ratio of interbank assets to net worth .
more generally , the amount of net worth is determined as @xmath76 for @xmath77 .
given @xmath69 , @xmath67 is obtained by solving a nonlinear problem , @xmath78 . in the case where asset returns follow normal distributions ( i.e. , @xmath79 ) , their variances
are directly adjusted according to the probability of default , taking as given the size of net worth . given the unit size of interbank lending , the sizes of interbank assets and liabilities , @xmath62 and @xmath80 , are given by the structure of the interbank network . to ensure that the probability of fundamental defaults , i.e. , defaults due to the loss of external assets only , is common across banks , the size of external assets relative to the net worth , @xmath81 ,
if bank @xmath6 has so many incoming edges that its liability side would be bigger than the sum of interbank assets and external assets , then riskless asset , @xmath48 , is imposed to adjust the asset side .
otherwise , deposits , @xmath53 , is imposed to adjust the liability side .
the parameter values are as follows : @xmath82 , @xmath16 , @xmath83 and @xmath84 .
we use three alternative values of @xmath3 : @xmath85 , @xmath86 and @xmath4 .
the ratio of interbank assets to capital is roughly consistent with the data shown by upper.@xcite these parameters indicate that the ratio of capital to total risky assets is .1 for those banks that have a positive amount of interbank assets and @xmath87 for the other banks . solving the above - mentioned nonlinear problem yields @xmath88 for @xmath89 , 15.687 for @xmath17 , and 47.296 for @xmath90 .
random asset returns are generated @xmath91 times for each case .
if bank @xmath6 failed , then bank @xmath6s creditors lose @xmath92 of their credits extended to bank @xmath6
. some of these creditors may fail due to the loss of their interbank assets .
accordingly , the creditors of the creditors of bank @xmath6 may fail as well , because they lose @xmath92 of their credits extended to the failed banks .
the spread of contagious defaults stops if it turns out that no more banks are going to fail due to the loss of interbank assets . following the literature ,
we assume that @xmath93 . under uniform immunization
, the vaccination order is based on either pagerank@xcite or in - degree .
pagerank , @xmath94 , is given by @xmath95 where @xmath96 is the @xmath97-th element of the adjacency matrix , which takes 1 if @xmath98 and 0 otherwise .
@xmath99 is the out - degree of node @xmath58 .
the parameter values are @xmath100 and @xmath101 .
when two or more nodes have a tie score , we randomize the order among them . under counteractive immunization , we pick up a pair of banks sequentially . in the case of the edge - betweenness - based strategy ,
we first select a pair of banks whose edge spanned between the two banks takes the highest value in terms of the edge - betweenness measure.@xcite we remove the edges coming from and going to the selected nodes , as well as the selected nodes themselves . in the following step , edge - betweenness centrality is recalculated , and then the new highest - ranked edge is selected .
if there are multiple edges that have the same edge - betweenness , we randomly choose one edge among those edges . as for random strategy , we basically randomize the vaccination order obtained by the edge - betweenness - based strategy .
the only difference is that we randomly choose a new pair of banks when there are multiple edges that exhibit the same edge - betweenness . in both strategies
, we randomly pick up a pair of banks from the remaining banks if there are no more edges in the network .
the selected banks are required to hold negatively correlated external assets whose correlation coefficient is @xmath102 .
any two correlated returns , @xmath103 , are expressed as @xmath104 where @xmath105 has mean zero and variance @xmath106 , while @xmath107 has mean zero and variance @xmath108 .
there is no correlation between @xmath109 and @xmath110 , @xmath111 .
notice that the variance of each asset return is independent of the value of @xmath5 , which means that the riskiness of individual assets remains unchanged .
kobayashi gratefully acknowledges financial support from kakenhi 25780203 and 24243044 .
t.k . designed research , performed research and wrote the paper .
t.k . and k.h .
wrote the matlab codes , and k.h . surveyed the literature and prepared the figures
|
many immunization strategies have been proposed to prevent infectious viruses from spreading through a network . in this work ,
we study efficient immunization strategies to prevent a default contagion that might occur in a financial network .
an essential difference from the previous studies on immunization strategy is that we take into account the possibility of serious side effects . _ uniform immunization _ refers to a situation in which banks are vaccinated with a common low - risk asset .
the riskiness of immunized banks will decrease significantly , but the level of systemic risk may increase due to the de - diversification effect . to overcome this side effect
, we propose another immunization strategy , called _ counteractive immunization _ , which prevents pairs of banks from failing simultaneously .
we find that counteractive immunization can efficiently reduce systemic risk without altering the riskiness of individual banks . since the global financial crisis of 2007 - 2009
, many researchers in various fields of natural and social sciences , such as physicists , ecologists and economists , have been tackling the question of how to reduce financial systemic risk.@xcite many of those studies are based on network theory , which has been used extensively to explore complex systems since the early 2000s .
percolation is one of the most frequently used concepts in thinking about the fragility of networks .
percolation theory provides a way to reveal the threshold of the number of removed nodes above which the giant component disintegrates.@xcite applying this theory , researchers examined the fragility of various types of actual complex networks , such as the internet , road networks , and power grids.@xcite percolation theory also gives us some useful insights into the question of how to control the way in which infectious viruses spread through a network .
many studies proposed various immunization strategies to make the percolation threshold as small as possible.@xcite systemic risk can be viewed as a fear of default contagion .
bank defaults can spread through a financial network in a manner similar to the spread of infectious diseases through a social network .
the idea of efficient immunization would therefore give us a clue to the question of how to reduce the likelihood of default contagion .
in this work , we explore efficient bank - immunization strategies by using a simple model of the interbank network ( see methods ) .
we consider two different types of vaccine .
one is a very - low - risk asset , possibly government bonds or cash .
since only a few of all the types of widely traded assets are categorized as very - low - risk assets , we assume that one such asset is used as a common vaccine .
the other vaccines are pairs of negatively correlated assets , which could be interpreted as opposite positions in risky assets .
we assume that the riskiness of each of the negatively correlated assets is the same as that of the uncorrelated assets that banks originally held before immunization .
we call the former type _
uniform immunization _ and the latter _ counteractive immunization _
( see methods ) .
counteractive immunization does not change the riskiness of individual banks , but kobayashi@xcite showed that assigning negatively correlated assets to highly infective banks would reduce systemic risk . even if a common vaccine is used
, the effects of immunization will differ if the order of vaccination varies .
we show that the likelihood of a large - scale contagion , what is called a _ financial crisis _ , can be reduced very efficiently if the order of immunization is based on pagerank . at the same time , however , uniform immunization leads to an undesirable situation in which the number of bank defaults during a crisis tends to increase as immunization proceeds . in other words ,
uniform immunization will make the financial market _ robust - yet - fragile_. we also examined some different orders of vaccination that are based on eigenvector centrality , degree centrality , node - betweenness centrality , and the eigenvector - based index proposed by restrepo et al.@xcite it turns out that the pagerank - based vaccination strategy is most efficient in terms of reducing the likelihood of financial crises .
it is shown that the undesirable trade - off between the likelihood and the size of a crisis does not arise under counteractive immunization . as an objective measure of systemic risk ,
we employ the @xmath0-th moment of the number of defaulted banks.@xcite in addition , to see the effects of immunization on the connectivity of the financial network , we also show the likelihood that the giant strongly connected component ( gscc ) disappears due to the removal of defaulted banks ( see methods ) .
these two measures are supposed to capture different aspects of the undesirable influence of bank defaults .
nevertheless , we find that the two measures are actually closely related and virtually coincide under a certain value of @xmath0 .
finally , we examine under what circumstances counteractive immunization becomes more effective than uniform immunization .
we derive the threshold of the degree of social risk aversion , @xmath0 , above which counteractive immunization becomes more advantageous than uniform immunization .
the frequency and size of financial crises under alternative immunization strategies . *
financial crisis is defined as a situation in which at least 5% of banks go bankrupt .
the average of crisis size is taken conditional on crisis event .
( a ) and ( b ) : the case of uniform immunization .
( c ) and ( d ) : the case of counteractive immunization .
the outcomes of uniform immunization based on a random vaccination order are also shown for comparison .
returns of external risky assets follow the student t - distribution with degree of freedom 5 .
the probability of fundamental default , @xmath1 , is @xmath2 prior to immunization .
the probability of fundamental default after uniform immunization , @xmath3 , is @xmath4 .
the degree of negative correlation under counteractive immunization , @xmath5 , is .6.,width=491 ]
|
the controversial discussions about the nature of the electron dynamics in quantum dots @xcite ( qds ) show that this topic is still not fully understood .
different electron - phonon coupling regimes leading to a multi - phonon process @xcite or a polaron decay @xcite have been proposed . especially for a multi - phonon process a so - called bottleneck effect @xcite is expected due to the discrete qd density of states . on the other hand it has been shown theoretically @xcite as well as experimentally @xcite that electron capture and relaxation by coulomb scattering ( auger effect ) with electrons in the wetting layer can be very efficient for large enough densities .
furthermore , an auger - type process can overcome the bottleneck due to the continuous energy dispersion in the wetting layer .
the depletion of the wetting layer density by an electric field @xcite has been shown to reduce the relaxation rates .
similarly , the electron density in the vicinity of the qds is reduced in a pn diode by the depletion layer within a dlts experiment @xcite .
electrons which are candidates for a coulomb scattering process are not located at the qds anymore , but at some distance .
therefore we are interested in coulomb scattering with remote continuum electrons of a quantum well and bulk contact regions as a function of the distance from the qds .
we propose a theory to calculate the transition probabilities between two qd and two continuum states .
debye screening effects are included according to the three - dimensional poisson equation . from the two - particle scattering probability we calculate the total scattering probability of the electron relaxation within a qd ( auger process ) and the excitation from a lower state into an excited state ( impact ionization or impact excitation ) by a numerical integration over the fermi distributed continuum states .
the band diagram of a qd and a remote quantum well and a bulk region defined by the doping profile is sketched in fig .
we assume that the 2deg and the 3deg are confined by infinitely high barriers . the transition probability per unit time between the qd states @xmath0 and @xmath1 , and the continuum states @xmath2 and @xmath3 is calculated by fermi s golden rule @xmath4 where @xmath5 and @xmath6 are the initial and final energies .
the matrix element is given by @xmath7 where @xmath8 , @xmath9 , @xmath10 and @xmath11 symbolize the respective wave functions @xcite .
@xmath12 represents the effective coulomb potential , which can be written in a two - dimensional fourier representation as @xmath13 with @xmath14 here , @xmath15 and @xmath16 are the relative and absolute permittivities , @xmath17 is the elementary charge , @xmath18 is the normalization area , @xmath19 , @xmath20 are the coordinates in growth direction , @xmath21 , @xmath22 are the in - plane coordinates , and @xmath23 are two - dimensional wave vectors with @xmath24 .
the dielectric function @xmath25 accounts for static screening effects of free electron densities .
the evaluation of the matrix elements ( [ eqmatrixelement ] ) depends on the qd wave functions , which themselves depend crucially on the qd shape . in order to estimate the integral
, we note that the wave functions @xmath26 and @xmath27 are orthogonal and thus the lowest order contribution results from the dipole moment @xmath28 of the effective charge density @xmath29 .
we approximate this dipole moment by two localized charges of opposite sign and thus obtain @xmath30 where the center position of the qd is chosen as . the total transition probability per unit time from a qd state @xmath0 into a qd state @xmath1
is then obtained from an integration over the continuum states @xmath31\text{,}\ ] ] where @xmath32 and @xmath33 are the fermi distribution functions of the initial and final continuum states .
for a two - dimensional electron gas ( 2deg ) we substitute @xmath34 and @xmath35 , and @xmath36 and @xmath37 for a three - dimensional electron gas ( 3deg ) , where @xmath38 is a ( large ) normalization length . in order to find an expression for the dielectric function @xmath25 for a two - dimensional electron gas positioned at @xmath39
, we consider an elementary charge positioned at @xmath40 , corresponding to an external potential with the fourier components @xmath41 for the induced potential @xmath42 originating from the induced electron density @xmath43 in the quantum well layer the fourier components of the three - dimensional poisson equation @xcite read as @xmath44 with the solution @xmath45 within the debye screening approach , @xmath43 is given by @xmath46\text{.}\ ] ] assuming a single 2deg quantization energy @xmath47 , @xmath48 is given by @xmath49 where @xmath50 is the fermi energy , @xmath51 is the effective mass , @xmath52 is the temperature and @xmath53 is boltzmann s constant , leading to the dielectric function for a 2deg @xmath54^{-1}\text{,}\ ] ] with the inverse screening length @xmath55^{-1}\text{.}\ ] ] inserting the two - dimensional fourier expansion of the coulomb potential into eq .
( [ eqn4 ] ) and using plane waves for the 2deg states , we find @xmath56\frac{e^{i\boldsymbol{k}\boldsymbol{r}_{\shortparallel}}}{\sqrt{a } } { { \rm d}}^2r_{\shortparallel}\right|^2 \text{. } \end{split}\ ] ] @xmath57 and @xmath58 are the @xmath19- and parallel components of @xmath59 , @xmath2 and @xmath3 are the two - dimensional wave vectors of the initial and final 2deg states .
integration in real space gives @xmath60}{2 a^2 ( \epsilon_0\epsilon_r)^2(q+\lambda_{\text{2d}})^2}\text{,}\ ] ] with the momentum transfer @xmath61 .
coulomb scattering with semi - confined bulk electrons located at @xmath62 is considered next .
the fourier components of the three - dimensional poisson equation for this problem are @xmath63 where the induced bulk electron density is given by @xmath64\text{.}\ ] ] with the definition of the debye screening wave vector @xmath65 and the expression for the bulk electron density @xmath66 the debye screening wave vector is given by @xmath67 here , @xmath68 denotes the fermi integral of order @xmath69 . solving eqn .
( [ poisson3d ] ) leads to @xmath70}\text{.}\ ] ] here , we have used that @xmath42 is continuously differentiable at @xmath39 .
analogously to eqn .
( [ defeps ] ) we find the dielectric function for the inhomogeneous bulk electron distribution for @xmath62 as @xmath71 using continuum wave functions , which are semi - confined in the @xmath19-direction and plane waves in the radial directions then leads to @xmath72\\ & \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\times \left.\sqrt{\frac{2}{l}}\sin(k_z(z - d))\right|^2\text { , } \end{split } \label{eqmquadrat3d}\ ] ] where @xmath73 and @xmath74 denote the @xmath19-components of the wave vectors .
integration over real space results in @xmath75}{(al)^2 ( \epsilon_0\epsilon_r)^2}\\ & \frac{\lambda_{\text{3d}}^2+q^2}{(\sqrt{\lambda_{\text{3d}}^2+q^2}+q)^2 } \left[\frac{1}{p_z^2+\lambda_{\text{3d}}^2+q^2}-\frac{1}{q_z^2+\lambda_{\text{3d}}^2+q^2}\right]^2 \end{split}\ ] ] with @xmath76 and @xmath77 .
analogously to the 2deg , the transition probability is obtained by integration over occupied continuum states , which is performed numerically .
0 nm ( e.g. , for a wetting layer surrounding the qds ) for different energy differences @xmath78 and temperatures @xmath52 .
in addition the corresponding result without screening is shown for the auger rate at @xmath7980 mev and 77 k ( dash - double - dotted line ) . ] according to the geometry and size of inas qds which have a typical base length of about 10 nm and a strong quantization in z - direction , we use @xmath80=5 nm and @xmath81=0 . for small dipole moments a taylor expansion in eqn .
( [ m2d ] ) and ( [ m3d ] ) leads to a quadratic dependence on the transition dipole moments , and the transition probabilities presented here can be rescaled for specifically shaped qds .
first we performed calculations of the auger and impact excitation probabilities per unit time as a function of the electron density per unit area in the wetting layer , using @xmath82 0 nm and @xmath83 0.023 @xmath84 for different qd level spacings @xmath85 .
[ fig2 ] shows the resulting probabilities for @xmath7930 mev and 80 mev for t=77 k and 300 k. the auger probabilities hardly depend on @xmath78 and the temperature . for densities
less than @xmath86 @xmath87 the auger probability is a linear function of the density . from a fit @xmath88 for low densities we obtain an auger relaxation coefficient of @xmath8960 @xmath90s@xmath91 .
this value is about one order of magnitude larger than the corresponding value calculated in ref .
@xcite ( 1 @xmath90s@xmath91 ) .
we explain this deviation by the different values for the transition dipole moment and the effective mass used in this work .
note that the auger capture coefficients evaluated in ref .
@xcite ( 2@xmath9210@xmath93 @xmath94s@xmath91 ) are also much smaller than those observed experimentally in ref . @xcite ( 10@xmath95 @xmath94s@xmath91 ) . for large densities ,
i.e. , when the 2deg becomes degenerate , the density dependence becomes nonlinear , and the probabilities are even reduced .
[ fig2 ] also shows that the impact excitation probabilities strongly depend on @xmath78 and @xmath52 , and the ratio between the impact excitation and auger probabilities is given by the boltzmann factor @xmath96 . the auger probability for @xmath7980 mev and t=77 k but neglecting screening effects
is also plotted in fig .
[ fig2 ] , demonstrating that screening effects play only a role for densities typically larger than @xmath86 @xmath87 .
10@xmath97 @xmath87 , @xmath7930 mev , @xmath7950 mev and @xmath7980 mev at 77 k ] fig .
[ fig3 ] shows the remote auger probabilities as a function of the distance @xmath98 between the qds and the 2deg , for @xmath99 10@xmath97 @xmath87 , @xmath100 0.06 @xmath84 , @xmath101 77 k and @xmath7930 mev , @xmath7950 mev and @xmath7980 mev .
the auger probabilities decrease exponentially with @xmath98 , depending on the energy transfer
. therefore the coulomb scattering of this type has a local character . as a consequence , a 2deg at @xmath82200 nm as in the memory structure of ref .
@xcite does not contribute to the lifetime limitation of charged qds . at 77k and with @xmath820
nm . in addition
the corresponding result without screening is shown for the auger rate at @xmath7980 mev ( dash - dotted line ) . ]
next , we calculated the coulomb scattering probabilities as a function of the bulk density in the vicinity of the qds .
the result is plotted for @xmath10230 mev ( solid line ) , @xmath7950 mev ( dashed line ) and @xmath7980 mev ( dotted line ) in fig .
[ fig4 ] for t=77 k , @xmath820 nm and @xmath100 0.06 @xmath84 .
again , for low densities the probabilities are linear in the density . but in comparison to the 2deg , the fitted auger coefficients @xmath103 depend on the qd level spacings . here we obtain @xmath104=@xmath10510@xmath106 @xmath107s@xmath91 , @xmath104=@xmath10810@xmath106 @xmath107s@xmath91 and @xmath104=@xmath10910@xmath106 @xmath107s@xmath91 for @xmath10230 mev , @xmath7950 mev and @xmath7980 mev , respectively . for high electron densities
we observe a deviation from this linear behavior , since pairs of occupied and empty continuum states can only be found near the fermi level when the electron gas becomes highly degenerate .
the resulting auger probability neglecting screening effects for @xmath7980 mev is plotted as a dash - dotted line in fig .
[ fig4 ] , demonstrating that screening is only important for densities above 10@xmath110 @xmath111 .
10@xmath112 @xmath111 , @xmath1130.06 @xmath84 and the impact excitation probabilities for @xmath11410@xmath115 @xmath111 , @xmath116 as a function of @xmath98 for different level spacings @xmath78 at 77 k. inset : dependence of the auger probability for @xmath11410@xmath112 @xmath111 , @xmath1130.06 @xmath84 and @xmath117 mev on the transition dipole moment for @xmath98=100
nm : @xmath118 , @xmath119 ( solid line ) ; @xmath120 , @xmath121 ( dashed line ) and @xmath122=@xmath121 ( dash - dotted line ) ] the auger probability as a function of the distance from the qds is investigated next .
the resulting curves for @xmath11410@xmath112 @xmath111 , @xmath1000.06 @xmath84 , @xmath7930 mev ( solid line ) , @xmath7950 mev dashed line , @xmath7980 mev ( dotted line ) at 77 k are shown in fig .
compared to the remote 2deg scattering , the dependence on @xmath98 is weaker , since momentum in the 3deg can additionally be changed in the third dimension . an interesting point is how a metal contact with its high electron density limits the lifetime of charged qds .
therefore , the impact excitation probabilities are also displayed in fig .
[ fig5 ] for @xmath11410@xmath115 @xmath111 , which is a typical electron density in metal contacts , and @xmath113@xmath84 . depending on the energy spacing @xmath78
, we obtain transition times of the order of an hour for @xmath7980 mev and @xmath82400 nm .
this goes well with the observation of bistable behavior in qd structures on this time scale @xcite .
the inset of fig .
[ fig5 ] shows the dependence of the auger probability on the transition dipole moment @xmath123 , which can be well approximated by a quadratic law resulting from the taylor expansion of eq .
( [ eqmquadrat3d ] ) .
the taylor expansion also explains that the transition probabilities are enlarged by a factor of 2 for vertical dipole moments ( in growth direction ) compared to the lateral dipole moments studied throughout this work .
we conclude that the direction of the dipole moment does not affect the order of magnitude of the transition rates .
= 90 nm , @xmath1143@xmath9210@xmath112 @xmath111 for @xmath7930 mev ( solid line ) , @xmath7950 mev dashed line , @xmath7980 mev ( dotted line ) ; dash - dotted lines : functions proportional to the boltzmann factor @xmath124 where @xmath125 is chosen such that lines intersect at 1000/@xmath52=10 k@xmath91 ] finally we consider qds embedded in a pn diode , where the qds are positioned within 3@xmath9210@xmath112 n - doped gaas layers .
such a device was investigated in dlts experiments , and a level spacing of @xmath7982 mev with an emission time of 62 ms at t=40 k has been observed @xcite . the emission from the qd ground state level
is dominated by an excitation into the excited state , from which the electron escapes into the continuum by a fast tunneling process . taking into account the boltzmann factor between excitation and relaxation processes we estimate a timescale for the relaxation process from the excited state into the ground state of about 3 ps . from fig .
[ fig2 ] we conclude that for an auger process on that fast time scale a wetting layer density of 10@xmath127 @xmath87 would be necessary .
however , the wetting layer is not occupied due to the electric field in the depletion layer . from the parabolic band bending and a qd ground state energy of 200 mev below the conduction band edge
we estimate that the depletion layer extends to @xmath8290 nm from the qds . for these parameters
, we display the temperature dependence as an arrhenius plot as shown in fig .
[ fig6 ] for @xmath7930 mev ( solid line ) , @xmath7950 mev dashed line , @xmath7980 mev ( dotted line ) .
the botzmann factor ( functions proportional to @xmath128 chosen such that the lines intersect at 1000/@xmath52=10 k@xmath91 ) is also displayed as dash - dotted line for each energy spacing , showing that the slope of the arrhenius plot of the temperature dependence is mainly given by the energy spacing .
nevertheless , we obtain an impact excitation time of about 10@xmath129 s for @xmath10140 k and @xmath7980 mev , which is orders of magnitudes slower than observed in the dlts experiment . only by reducing the level spacing below @xmath7950 mev impact excitation times in the range of tens of ms
can be obtained at this temperature .
this indicates that in addition phonon mechanisms have to be taken into account .
in conclusion , we have evaluated the auger effect and impact excitation ( coulomb scattering ) in qd devices with remote continuum states .
three - dimensional debye screening has been included , which is important for high electron densities .
our results show that , depending on the detailed device and qd geometries , scattering with remote continuum carriers can be of importance for the electron kinetics if the local wetting layer density is reduced .
|
electron capture and emission by coulomb scattering in self - assembled quantum dot ( qd ) devices is studied theoretically . while the dependence of the coulomb scattering ( auger ) rates on the local wetting layer electron density has been a topic of intense research , we put special interest on the remote scattering between qd electrons and continuum electrons originating from a quantum well , doped bulk layers or metal contacts .
numerical effort is made to include all microscopic transitions between the fermi distributed continuum states .
the remote coulomb scattering is investigated as a function of the electron density , the distance from the qds and the temperature .
our results are compared with experimental observations , considering lifetime limitations in qd memory structures as well as the electron emission in pn - diodes .
|
estimating the sparse precision matrix , i.e. the inverse covariance matrix , from data is a very widely used method for exploring the dependence structure of continuous variables .
the motivation for the approach stems from the fact that for a gaussian markov random field model , zeros in the precision matrix translate exactly to absent edges in the corresponding undirected gaussian graphical model , thus being informative about the marginal and conditional independence relationships among the variables .
the full @xmath1-dimensional covariance matrix contains @xmath2 parameters , making its accurate estimation from limited data difficult .
additionally , the structure learning requires the inverse of the covariance , and matrix inversion is in general a very fragile operation . to make the problem tractable
, some form of regularisation is typically needed .
direct optimisation of the sparse structure would easily lead to very difficult combinatorial optimisation problems . to avoid these computational difficulties ,
several convex @xmath0-penalty - based approaches have been proposed .
popular examples include @xmath0-penalised maximum likelihood estimation @xcite , which also forms the basis for the highly popular graphical lasso ( glasso ) algorithm @xcite .
@xmath0 regularisation has also been used for example in a non - probabilistic alternative with linear - programming - based constrained @xmath0 minimisation ( clime ) algorithm of @xcite . at the heart of the optimisation problems considered by all these methods
is a term depending on the @xmath0 norm of the estimated precision matrix .
@xmath0-penalisation - based approaches such as lasso are popular for sparse regression , but they have a known weakness : in addition to promoting sparsity they also push true non - zero elements toward zero @xcite . in the context of precision matrix estimation
this effect would be expected to be especially strong when some elements of the precision matrix are large , which happens for scaled covariance matrices when the covariance matrix becomes ill - conditioned .
this phenomenon occurs frequently under the circumstances where some of the variables are nearly linearly dependent . in this paper
we demonstrate a drastic failure of the @xmath0 penalised sparse covariance estimation methods for a class of models that have a linear latent variable structure where some variables depend linearly on others . for such models even in the limit of infinite data , popular @xmath0 penalised methods can not yield results that are significantly better than based on random guessing on any setting of the regularisation parameter . yet
these models have a very clear sparse structure that becomes obvious from the empirical precision matrix with an increasing @xmath3 . given the huge popularity and success of linear models in modelling data , structures like the one considered in our work are natural for various real world data sets .
motivated by our discovery , we also explore the inconsistency of @xmath0 penalised methods on models derived from real gene expression data and find them poorly suited for such applications .
we start with a quick recap on the basics of gaussian graphical models in order to formulate the problem of structure learning . for a more comprehensive treatment of the subject
, we refer to ( @xcite ; @xcite ) .
let @xmath4 denote a random vector following a multivariate normal distribution with zero mean and a covariance matrix @xmath5 , @xmath6 .
let @xmath7 be an undirected graph , where the @xmath8 is the set of nodes and @xmath9 stands for the set of edges .
the nodes in the graph represent the random variables in the vector @xmath10 and absences of the edges in the graph correspond conditional independence assertions between these variables .
more in detail , we have that @xmath11 and @xmath12 if and only if @xmath13 is conditionally independent of @xmath14 given the remaining variables in @xmath10 . in the multivariate normal
setting , there is a one - to - one correspondence between the missing edges in the graph and the off - diagonal zeros of the precision matrix @xmath15 , that is , @xmath16 ( see , for instance , @xcite , p. 129 ) . given an undirected graph @xmath17 , a gaussian graphical model is defined as the collection of multivariate normal distributions for @xmath10 satisfying the conditional independence assertions implied by the graph @xmath17 .
assume we have a complete ( no missing observations ) i.i.d .
sample @xmath18 from the distribution @xmath19 .
based on the sample @xmath20 , our goal in structure learning is to find the graph @xmath17 , or equivalently , learn the zero - pattern of @xmath21 .
the usual assumption is that the underlying graph is sparse .
a naive estimate for @xmath21 by inverting the sample covariance matrix is practically never truly sparse for any real data .
furthermore , if @xmath22 the sample covariance matrix is rank - deficient and thus not even invertible . one common approach to overcome these problems is to impose an additional @xmath23-penalty on the elements of @xmath21 when estimating it .
this kind of regularisation effectively forces some of the elements of @xmath21 to zero , thus resulting in sparse solutions . in the context of regression models , this method applied on the regression coefficients goes by the name of _ lasso _ @xcite .
there exists a wide variety of methods making use of @xmath23-regularisation in the setting of gaussian graphical model structure learning ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . in this section
we provide a brief review of selected examples of different types of @xmath0-penalised methods .
we begin with the widely used graphical lasso - algorithm ( glasso ) by @xcite .
glasso - method maximises an objective function consisting of the gaussian log - likelihood and an @xmath24-penalty : @xmath25where @xmath26 denotes the sample covariance matrix and @xmath27 is the regularisation parameter controlling the sparsity of the solution .
the @xmath0 penalty , @xmath28 , is applied on all the elements of @xmath29 , but the variant where the diagonal elements are omitted is also common .
the objective function ( [ eq : glassoobj ] ) is maximised over all positive definite matrices @xmath21 and the optimisation is carried out in practice using a block - wise coordinate descent .
@xcite approach the problem of sparse precision matrix estimation from a slightly different perspective .
their clime - method ( constrained @xmath0-minimisation for inverse matrix estimation ) seeks matrices @xmath29 with a minimal @xmath0-norm under the following constraint @xmath30 is the tuning parameter and @xmath31 is the element - wise maximum .
the optimisation problem @xmath32 subject to the constraint ( [ eq : clime ] ) does not explicitly force the solution to be symmetric , which is resolved by picking from estimated values @xmath33 and @xmath34 the one with a smaller magnitude into the final solution . in practice ,
the optimisation problem is decomposed over variables into @xmath1 sub - problems which are then efficiently solved using linear programming .
@xcite introduced recently a method called sparse column - wise inverse operator ( scio ) .
the scio - method decomposes the estimation of @xmath29 into a following smaller problems @xmath35where @xmath26 and @xmath36 are defined as before and @xmath37 is an @xmath38:th standard unit vector .
the regularisation parameter @xmath36 can in general vary with @xmath38 but this is omitted in our notation .
the solutions @xmath39 form the columns for the estimate of @xmath29 .
also for scio , the symmetry of the resulting precision matrix must be forced , and this is done as described in the case of clime .
in addition to the above - mentioned @xmath0-penalised methods , we consider two alternative approaches . in a `` naive '' approach
, we simply take the sample covariance matrix , invert it , and then threshold the resulting matrix to obtain a sparse estimate for the precision matrix .
the threshold value is chosen using the ground truth graph so that the naive estimator will have as many non - zero entries as there are edges in the true graph .
setting the threshold value according to the ground truth is of course unrealistic , however , it is nevertheless interesting to compare the accuracy of this simple procedure to the performance of the more refined @xmath0-methods , when also their tuning parameters are chosen in a similar fashion .
lastly , we consider a bayesian approach which is based on finding a graph with a highest fractional marginal pseudo - likelihood ( fmpl ) by @xcite .
the fractional marginal pseudo - likelihood is an approximation of the marginal likelihood and it has been shown to be a consistent scoring function in the sense that the true graph maximises it as the sample size tends to infinity , under the assumption that data are generated from a multivariate normal distribution .
the fmpl - score decomposes over variables and in practice , the method identifies optimal markov blankets for each of the variables , which are then combined into a proper undirected graph using any of the three different schemes commonly employed in graphical model learning : or , and and hc .
the assumptions required for a consistent model selection with an @xmath24-penalised gaussian log - likelihood have been studied , for instance , in @xcite .
the authors provide a number of conditions in the multivariate normal model that are sufficient for the recovery of the zero pattern of the true precision matrix @xmath40 with a high probability when the sample size is large . for our purposes ,
the most relevant condition is the following : [ assumption1 ] there exists @xmath41,$ ] such that @xmath42 here @xmath43 is a set defining the support of @xmath40 , that is , the non - zero elements of @xmath40 ( diagonal and the elements corresponding to the edges in the graphical model ) and @xmath44 refers to the complement of @xmath45 in @xmath46 .
the @xmath47 term is defined via kronecker product @xmath48 as @xmath49 and @xmath50 refers to the specific rows and columns of @xmath47 indexed by @xmath51 and @xmath52 , respectively .
the norm in the equation is defined as @xmath53 .
the above result applies to glasso .
however , a quite similar result was presented for scio in @xcite : [ assumption2 ] there exists @xmath54 such that @xmath55 here @xmath56 and @xmath57 .
assumption [ assumption2 ] under the multivariate normality guarantees that the support of @xmath40 is recovered by scio with a high probability as the sample size gets large .
methods for sparse precision matrix estimation generally depend on an objective function ( such as log - likelihood ) and a penalty function or regulariser , which in a bayesian setting is usually represented by the prior . the ideal penalty function for many problems
would be the @xmath58 `` norm '' counting the number of non - zero elements : @xmath59 .
this @xmath58 function is not a proper norm , but it provides a very intuitive notion of sparsity . the main problem with its use is computational : using @xmath58-penalisation leads to very difficult non - convex combinatorial optimisation problems .
the most common approach to avoid the computational challenges is to use @xmath0-penalisation as a convex relaxation of @xmath58 .
as mentioned above this works well in many cases but it comes with a price , since in addition to providing the sparsity , @xmath0 also regularises large non - zero values .
depending on the problem , as we demonstrate here , this effect can be substantial and may cause @xmath0-regularised methods to return totally meaningless results . intuitively , @xmath0-regularised methods are expected to fail when some elements of the true precision matrix become so large that their contribution to the penalty completely overwhelms the other parts of the objective and the penalty .
one example where this happens is when some set of variables depends linearly on another set of variables .
in such situation the covariance matrix can become ill - conditioned and the elements of its inverse , the precision matrix , grow .
one example of when this happens is models with a linear latent variable structure .
let us consider a model for @xmath60 , where @xmath61 .
the graphical structure of the model and the corresponding precision matrix structure are illustrated in fig .
[ fig : latent_variable_structure ] . assuming @xmath62 ,
the covariance of the concatenated vectors @xmath63 is given by the block matrix @xmath64 the covariance matrix has an analytical block matrix inverse @xcite @xmath65 this precision matrix recapitulates the conditional independence result for gaussian markov random fields : the lower right block is diagonal because the variables in @xmath66 are conditionally independent of each other given @xmath67 .
the matrix is clearly sparse , so we would intuitively assume sparse precision matrix estimation methods should be able to recover it .
the non - zero elements do , however , depend on @xmath68 which can make them very large if the noise @xmath69 is small .
it is possible to evaluate and bound the different terms of eq .
( [ eq : glassoobj ] ) evaluated at the ground truth for these models : @xmath70 the magnitude of the last penalty term clearly grows very quickly as @xmath69 decreases .
clearly the magnitude of the two first log - likelihood terms grows much more slowly as they only depend on @xmath71 .
thus the total value of eq .
( [ eq : glassoobj ] ) decreases without bound as @xmath69 decreases . forgetting the ground truth , it is easy to see that one can construct an estimate @xmath21 for which the objective remains bounded . if we assume all values of @xmath72 to be @xmath73 ( after normalisation ) , @xmath74 as the other terms only depend on @xmath21 it is easy to choose @xmath21 so that they remain bounded .
the estimate @xmath21 that yields these values will in many cases not have anything to do with @xmath75 , as seen in the experiments below .
we tested the performance of glasso , scio and clime as well as fmpl using the model structure introduced in sec .
[ sec : latent - variable - like ] .
the performance of the methods was investigated by varying the noise variance @xmath69 , and the sample size @xmath3 .
the model matrix @xmath76 was created as a @xmath77-array of independent normal random variables with mean @xmath78 and variance @xmath79 .
the majority of the tests were run using input dimensionality @xmath80 , output dimensionality @xmath81 and noise variance @xmath82 but we also tested varying these settings . for each individual choice of noise and sample
size , @xmath83 different matrices @xmath76 were generated and the results were averaged .
generating @xmath3 samples using model described , data were normalised and analysed using the five different methods .
we calibrated the methods in a way that number of edges in the resulting graph would match the true number .
similarly , we thresholded the naive method by taking inverse matrix directly to output the correct number of edges .
the fmpl method has no direct tuning parameters so we used its or mode results as such .
similar tuning is not possible in a real problem where the true number of edges is now known .
the tuning represents the best possible results the methods could obtain with an oracle that provides an optimal regularisation parameter .
we evaluated the results using the hamming distance between the ground truth and the inferred sparsity pattern , i.e. the number of incorrect edges and non - edges which were treated symmetrically . for methods returning the correct number of edges , this value
is directly related to the precision @xmath84 through @xmath85 or conversely @xmath86 we will nevertheless use the hamming distance as it enables fair comparison with fmpl that sometimes returns a different number of edges .
[ fig : noise_n100 ] and [ fig : noise_n1000 ] show the hamming distance obtained by the different methods as a function of the noise level when using 100 and 1000 samples , respectively .
the results show that especially for low but also for high noise levels , the @xmath0-based methods all perform very poorly with especially glasso and clime performing very close to random guessing level for low noise levels @xmath87 .
the naive inverse and fmpl work much better up to moderate noise levels of @xmath88 after which the noise starts to dominate the signal and the performance of all methods starts to drop .
scio is a little better than the other @xmath0-based methods but clearly worse than fmpl and naive in the low noise regime .
[ fig : outdim_n1000 ] shows the results when changing the output dimensionality @xmath89 from 10 .
the results show that the performance of all @xmath0-based methods is very poor across all @xmath89 .
glasso performance is close to random guessing level across the entire range considered , while clime is slightly better for @xmath90 and scio slightly better across the entire range .
both fmpl and naive are significantly better than any of the @xmath0-based methods .
[ fig : indim_n1000 ] shows the corresponding result when changing the input dimensionality @xmath91 .
the results are now quite different as all methods are better than random especially for larger values .
scio still outperforms clime which outperforms glasso .
fmpl is really accurate for small @xmath91 but degrades for larger @xmath91 while the naive method is the most accurate in almost all cases . ) for latent variable like model with 1000 samples .
the green curves show the contributions of the first two terms of eq .
( [ eq : glassoobj ] ) and the blue curves show the contributions of the last penalty term .
solid lines show the result of the glasso optimal solution while dashed lines show the result for the true solution . ] to further illustrate the behaviour of glasso on these examples , fig .
[ fig : obj_n100 ] shows the contributions of the different parts of the glasso objective function ( [ eq : glassoobj ] ) as a function of the noise level both for the true solution ( `` truth '' ) as well as the glasso solution .
the results show that for low noise levels the penalty incurred by the true solution becomes massive .
the glasso solution has a much lower log - likelihood ( `` logl '' ) than ground truth but this is amply compensated by the significantly smaller penalty . as the noise increases , the penalty of the true solution decreases and the glasso solution converges to similar values .
it can be checked that the norm @xmath92 in assumption 1 and eq .
( [ eq : assumption1 ] ) for latent - variable - like models depends on the scale of @xmath76 .
we took advantage of this by creating examples with different values of @xmath92 and testing the precision of glasso using the true covariance which corresponds to infinite data limit .
the results of this experiment are shown in fig .
[ fig : latent_assumption1 ] .
the results verify that glasso consistently yields perfect results when @xmath93 which is a part of the sufficient conditions for consistency of glasso .
as @xmath92 grows and the sufficient conditions are no longer satisfied , it is clearly seen that the accuracy of glasso starts to deteriorate rapidly .
this suggests that the sufficient condition of assumption 1 is in practice also necessary to ensure consistence . of assumption 1 and eq .
( [ eq : assumption1 ] ) .
values to the left of the green vertical line satisfy this condition while values to the right violate it .
( higher values are better . ) ]
we tested how often the problems presented above appear in real data using the `` tcga breast invasive carcinoma ( brca ) gene expression by rnaseq ( illuminahiseq ) '' data set @xcite downloaded from https://genome - cancer.ucsc.edu / proj / site / hgheatmap/. the data set contains gene expression measurements for 20530 genes for @xmath94 samples . after removing genes with a constant expression across all samples there are @xmath95 genes remaining . in order to test the methods we randomly sampled subsets of @xmath96 genes and considered the correlation matrix @xmath97 over that subset .
we generated sparse models with known ground truth by computing the corresponding precision matrix @xmath98 from the empirical correlation matrix , setting elements with absolute values below chosen cutoff @xmath99 to 0 to obtain @xmath100 and the testing covariance matrix @xmath101 .
the cutoff lead to networks that were sparse with on average 60% zeros in the precision matrix . ) on real gene expression data showing the fraction of random subsets of @xmath96 genes that fulfil the requirement and various relaxations .
the condition ( [ eq : assumption1 ] ) requires @xmath93 , but the figure shows results also for larger @xmath92 cutoffs . ] fig .
[ fig : tcga_assumption1 ] shows the fraction of covariances derived from random subsets of @xmath96 genes that satisfy the assumption 1 of @xcite ( @xmath102 ) as well as the fraction of values below more relaxed bounds .
the figure shows that the assumption is reliably satisfied only for very small @xmath96 while for @xmath103 , the assumption is essentially never satisfied .
based on the results of fig .
[ fig : latent_assumption1 ] it is likely that glasso results will degrade significantly by for @xmath104 and beyond which are very common for large networks .
we further studied how accurately glasso can recover the graphical structures when the data were generated using the precision matrices described above .
we used a similar thresholding with a cut - off value of @xmath105 in order to first form sparse precision matrices for a random subset of genes with given dimension .
these matrices were then inverted to obtain covariance matrices .
we checked that the resulting matrices were positive definite and then used them to sample multivariate normal data with zero mean with different sample sizes .
the obtained data sets were centred and scaled before computing the sample covariance which was used as input to the glasso algorithm .
the regularisation parameter was chosen with the aid of the ground truth graph , so that the the graph identified by glasso would contain as many edges as there were in the real graph .
results are shown in figure [ fig : tcga_precision ] .
the results show that glasso performance decreases as the network size increases and is approaching that of random guessing for the largest networks considered here . ) on real gene expression data over random subsets of @xmath96 genes .
the values are shown for the @xmath0 penalty term as well as the unnormalised log - likelihood , divided by @xmath96 to make them comparable .
solid lines show the values for glasso result while dashed lines show the result for ground truth . ]
[ fig : tcga_objective ] shows the contributions of different parts of the glasso objective function ( [ eq : glassoobj ] ) as a function of the number of genes @xmath96 .
the regularisation parameter @xmath36 of glasso was tuned to return a solution with the same number of edges as in the true solution .
we used the glasso implementation of scikit - learn @xcite , which ignores the diagonal terms of @xmath29 when computing the penalty .
the figure shows clearly how the penalty term for the true solution increases superlinearly as a function of @xmath96 .
( a linear increase would correspond to a horizontal line . )
the result is even more striking given that the optimal @xmath36 decreases slightly as @xmath96 increases .
the penalty contribution for glasso solution increases much more slowly . the excess loss in log - likelihood from glasso solution increases as @xmath96 increases , but this is compensated by a larger saving in the penalty .
together these suggest that glasso solutions are likely to remain further away from ground truth as @xmath96 increases .
the class of latent variable like models presented in sec .
[ sec : latent - variable - like ] is an interesting example of models that have a very clear sparse structure , which all @xmath0-penalisation - based methods seem unable to recover even in the limit of infinite data .
this class complements the previously considered examples of models where glasso is inconsistent including the `` two neighbouring triangles '' model of @xcite and the star graph of @xcite , the latter of which can be seen as a simple special case of our example .
an important question arising from our investigation is how significant the discovered limitation to inferring sparse covariance matrices is in practice , i.e. how common are the latent variable like structures in real data sets .
given the popularity and success of linear models in diverse applications it seems plausible such structures could often exist in real data sets , either as an intrinsic property or as a result of some human intervention , e.g. through inclusion of partly redundant variables .
the gene expression data set is a natural example of an application where graphical model structure learning has been considered .
the original glasso paper @xcite contained an example on learning gene networks , although from proteomics data .
other authors ( e.g. * ? ? ?
* ) have applied gaussian graphical models and even glasso ( e.g. * ? ? ?
* ) to gene network inference from expression data .
our experiments on the tcga gene expression data suggest that in such applications it is advisable to consider the conditions for the consistency of @xmath0 penalised methods very carefully when planning to apply those .
previous publications presenting new methods for sparse precision matrix have typically tested the method on synthetic examples where the true precision matrix is specified to contain mostly small values . specifying the precision matrix provides a convenient way to generate test cases as the sparsity pattern can be defined very naturally through it . at the same time , this excludes any models that have an ill - conditioned covariance . as shown by our example , such ill - conditioned covariances arise very naturally from model structures that are plausible from the application perspective . ultimately , our results suggest that users of the numerous @xmath0 penalised methods should be much more careful about checking whether the conditions of consistency for precision matrix estimation are likely to be fulfilled in the application area of interest
. o. banerjee , l. el ghaoui , and a. daspremont .
model selection through sparse maximum likelihood estimation for multivariate gaussian or binary data .
_ journal of machine learning research _ , 9:0 485516 , june 2008 .
p. menndez , y. a. i. kourmpetis , c. j. f. ter braak , and f. a. van eeuwijk .
gene regulatory networks from multifactorial perturbations using graphical lasso : application to the dream4 challenge . _ plos one _ , 50 ( 12):0 e14147 , 2010 .
f. pedregosa , g. varoquaux , a. gramfort , v. michel , b. thirion , o. grisel , m. blondel , p. prettenhofer , r. weiss , v. dubourg , j. vanderplas , a. passos , d. cournapeau , m. brucher , m. perrot , and e. duchesnay .
scikit - learn : machine learning in python .
_ journal of machine learning research _ , 12:0 28252830 , 2011 .
p. ravikumar , m. j. wainwright , g. raskutti , and b. yu . high - dimensional covariance estimation by minimizing @xmath0-penalized log - determinant divergence .
_ electronic journal of statistics _ , 5:0 935980 , 2011 .
|
various @xmath0-penalised estimation methods such as graphical lasso and clime are widely used for sparse precision matrix estimation .
many of these methods have been shown to be consistent under various quantitative assumptions about the underlying true covariance matrix .
intuitively , these conditions are related to situations where the penalty term will dominate the optimisation . in this paper
, we explore the consistency of @xmath0-based methods for a class of sparse latent variable -like models , which are strongly motivated by several types of applications .
we show that all @xmath0-based methods fail dramatically for models with nearly linear dependencies between the variables .
we also study the consistency on models derived from real gene expression data and note that the assumptions needed for consistency never hold even for modest sized gene networks and @xmath0-based methods also become unreliable in practice for larger networks .
|
one of the most important unsolved problems in astrophysics is the question of how the solar corona is heated .
directly related to this problem is understanding the evolution of the emission in active regions , particularly in the core .
recent observations from _ hinode _ @xcite are providing a comprehensive view of active regions . they appear to be composed of at least two dominant loop populations : core loops that are seen evolving at multi - temperatures , and peripheral cool loops that develop downflows @xcite . the former appear to be rooted in unipolar magnetic concentrations @xcite .
the latter are seen extending to great heights and appear to be associated with coronal rain .
fan structures around the edges of active regions show strong outflows @xcite that may be significant sources of the slow solar wind ( @xcite ) . a key question is what is the time - scale of heating in the cores of active regions ?
is steady heating important , or is everything dynamic ? if the heating is impulsive , is the repetition time between events shorter than a coronal loop cooling time so that the impulsive heating is effectively steady ? hydrostatic models are able to reproduce the emission from high temperature loops seen in active regions , but they have difficulty reproducing the emission from lower temperature ( warm ) overdense loops .
hydrodynamic simulations of active regions or modeling based on low frequency impulsive heating by nanoflares , improve the agreement with the warm loop observations , but discrepancies still remain .
it is not always clear whether these discrepancies are related to the hydrostatic or hydrodynamic modeling , or to the extrapolation and modeling of the morphology of the magnetic field .
one region where the heating could be quasi - steady is in transition region moss .
moss is the reticulated pattern of emission in active regions that typically evolves slowly over long time - scales and shows only low - level variability @xcite .
indeed hydrostatic modeling of moss is able to reproduce the observed intensities well - to within a filling factor @xcite .
though the general pattern evolves slowly , fine - scale dynamics are observed on a local scale , and this has been interpreted as variability in the overlying spicular material rather than in the moss itself @xcite .
this picture of fine - scale dynamics within a slowly evolving large - scale pattern also well describes the unipolar regions at the footpoints of hot loops @xcite .
the fact that moss is thought to be the emission from the footpoints of hot loops @xcite , suggests that there may be a close relationship between these patterns and we are investigating this issue .
@xcite recently presented an example where the moss maps the magnetic flux closely , though @xcite did not find any spatial correlation with the photospheric magnetic field .
it has been argued that moss in highly evolved active regions may be the only regions in the corona where the heating could be steady because these are the only places where the magnetic field is not changing rapidly @xcite .
detection of flows and motions in the moss , however , would reveal dynamic signatures for the first time .
this letter presents an analysis of the time - variability of flows and motions in moss over 16 hours of eis observations of the core of a flaring active region observed in 2007 , june .
we show that the doppler and non - thermal velocities are no larger than typical quiet sun values in the moss , and that they do not vary substantially over the observations period .
these results support the view that moss at the bases of hot loops is heated quasi - steadily .
the eis instrument obtains high resolution spectra in the wavelength ranges 171212 and 245 - 291 and has a spectral resolution of 0.0223 and 1@xmath0 spatial pixels @xcite .
typical exposure times for active regions are 1530s so that it takes tens of mins to several hours for the eis instrument to step its slit over the full extent of a large active region . thus observational cadence or spatial coverage are usually traded off for diagnostic information when making observations with a spectrometer in comparison with using a broad - band filter imager such as trace ( * ? ? ?
* transition region and coronal explorer ) or _ hinode_/xrt ( * ? ? ?
* x - ray telescope ) .
the eis effective area peaks near the strong 195.119 emission line , however , so that very short exposures ( 1s ) can still yield good counts in this line on a relatively bright target .
it is therefore possible to make rapid scans over a wide area in strong lines only , effectively trading off wavelength coverage to retain high cadence and a large fov .
we use data obtained in this observing mode in this letter . _ hinode _ tracked noaa active region 10960 from limb to limb in june 2007 .
eis repeatedly ran an observing sequence chromo_evap_raster which takes 1s exposures with the 2@xmath0 slit and scans an area of 150@xmath0 by 150@xmath0 in a few mins .
figure [ fig1 ] shows xrt and trace images of ar 10960 with the eis fov overlaid .
the movie associated with figure [ fig1 ] shows that ar 10960 evolved slowly in xrt images , and this is confirmed by the x - ray light curve shown in figure [ fig1 ] .
the intensity observed by xrt in a boxed region above the moss varies by less than 15% over the 8 hour period shown .
the trace movie shows that the pattern of moss in the core of the region is also remarkably stable . nevertheless , this region was flare productive : figure [ fig2 ] is a goes plot of the x - ray activity , which shows numerous c- and m- class flares occuring during its passage across the disk , and also a gradual decline in activity .
these characteristics make ar 10960 an ideal candidate for studying whether impulsive or steady heating dominates in the core of the region .
the eis data analyzed here were obtained between 18ut on 2007 , june 7 , and 10ut on 2007 , june 8 .
several c- class flares did occur during these 16 hours though not all were cleanly observed by _
hinode _ or trace because of south atlantic anomaly ( saa ) and night - time passage .
at least two flares are seen , however , in our xrt and trace movies and it is clear that they occurred in the core of the region near the neutral line with the potential to disrupt or interact with the moss .
the eis data were converted to physical units and processed using the default options for the solarsoft routine eis_prep .
the thermal orbital variation in line centroid position was estimated from single gaussian fits to the 195.119 line profiles at every pixel of every raster scan .
this effect and the spectral line tilt were removed from the data prior to further analysis .
the zero - point velocity for this operation was set by measuring the wavelength of the line averaged over the complete dataset .
the _ hinode _
pointing has an orbital variation on the order of 12 pixels in the x- and y- directions @xcite .
this effect was removed by coaligning the single gaussian fit images by cross - correlation .
since the 2@xmath0 slit instrumental width was not measured in the laboratory pre - launch , we estimated it using measurements of the 184.536 line from a large raster scan of the quiet sun taken on 2007 , november 6 . from these data , we found that the 2@xmath0 slit instrumental width was approximately 19% wider than that of the 1@xmath0 slit .
taking the on - orbit value for the 1@xmath0 slit from @xcite we estimated the 2@xmath0 slit instrumental width to be 0.066 .
the non - thermal velocities were calculated from the line widths assuming this measured instrumental width and the appropriate thermal width at a temperature of 1.35mk .
the 195.119 line is blended with a very weak density sensitive line at 195.179 .
and we found that single gaussian fits to our data overestimated the doppler and non - thermal velocities as a result of density variations .
our measurements are therefore derived from double gaussian fits where the separation of the line centroids is fixed at 0.06 and the widths of the two components are forced to be the same .
these fit parameter constraints were suggested by @xcite and we found that their method worked well on our data , even in single pixels .
figure [ fig3 ] shows an example of the intensity , non - thermal , and doppler velocity maps obtained from two component gaussian fits to the eis data .
the moss is only weakly visible in the non - thermal velocity maps , and little variation in the patterns is seen over 16 hours of data ( see the movie associated with figure [ fig3 ] ) .
we analyzed the time variability of these quantities in 25 locations in the moss by visually selecting pixels that are not affected ( in the line of sight ) by any flaring activity , and do not lie close to the edges of the moss .
these edges border dark areas that have been shown to have enhanced non - thermal velocities and outflows in other active regions @xcite .
larger blue - shifts in these areas can also be seen in figure [ fig3 ] .
our aim was to probe the properties of the moss on smaller spatial scales than @xcite they used a minimum box size of 10@xmath0@xmath110@xmath0 and to reduce smoothing of the measurements that results from averaging over large regions .
variations in intensities in single pixels can result from not sampling precisely the same spatial location in consecutive rasters because of the _ hinode _ orbital pointing variation , however .
furthermore , increased signal in the box improves the reliability of measurements of the contribution from the blend .
therefore , the box size was chosen to be @xmath2 2 pixels around the selected pixels as a compromise .
the velocities here were calibrated to the zero point velocity by comparing the time - averaged wavelength from single and double gaussian fits to the data at each location . nevertheless , the absolute values are not well constrained .
the results for the full 16 hour time - period are shown in figure [ fig4 ] .
residual effects from , e.g. , cosmic ray hits that were not completely removed , have been ignored by only quoting results from the 98% of significant values .
consistent with previous results obtained by trace , the high spectral resolution eis obervations show that the 195.119 intensities do not vary significantly over time .
the standard deviation of the intensities from the 25 locations over 16 hours is less than 15% of the average intensity in each region .
this is less than the variation of the average intensity from location to location ( 30 - 35% ) .
our new measurements of doppler and non - thermal velocities show that they are only weak in the moss ; 2.7 km / s ( red - shifted ) and 21.7 km / s , respectively , when averaged over time in all the 25 locations .
given the uncertainties in the absolute values , the doppler velocity is consistent with a measurement of zero .
the average values of the non - thermal velocities in each of the 25 locations range from 1731 km / s .
the upper limit of this range is an extreme case : in all but this location the values are less than 26 km / s .
these values are no more than the typical values measured for non - thermal mass motions in the quiet sun at this temperature .
we confirmed this by performing double gaussian fits to the 195.119 line in 39 quiet sun synoptic datasets that we had processed for a differential emission measure analysis @xcite .
the data were taken with the eis 1@xmath0 slit for the best spectral resolution , and the non - thermal velocity we obtained for the quiet sun was 25 km / s .
the standard deviation in the moss measurements is less than @xmath3 15% of the average values for each location over the 16 hours of observation , which corresponds to a variation of less than @xmath3 4 km / s in all the boxed areas .
the average doppler velocities are also less than @xmath3 6.5 km / s in each of the regions .
this is consistent with the measurements of only a few km / s in the moss by @xcite .
the doppler velocities also vary by less than 3 km / s during the observations period . as with the intensity results ,
this is less than the measured variation from location to location .
active region 10960 evolved slowly when seen in xrt images , and the moss in the core of the region also developed slowly in trace and eis images .
the region was flare productive , however , with many c- and m- class flares occurring throughout its passage across the solar disk .
several of these flares occurred close to the neutral line in the core of the active region during the 16 hours of data we analyzed with the potential to interact with and disrupt the moss .
ar 10960 was therefore a good candidate for studying the relative importance of impulsive and steady heating in the core of an active region .
the data we presented are the highest cadence spatially resolved maps of non - thermal and doppler velocities in moss ever obtained in the corona .
as such they are ideal for detecting any signatures of dynamics in the moss , should they exist .
our spectrally resolved measurements of 195.119 intensities agree with previous results from broad - band filter images in that the intensities do not vary significantly in the moss .
our new measurements of doppler and non - thermal velocities show that they are only weak in the moss , and that they vary by only a few km / s during the observing period .
the results confirm the findings of @xcite that moss at the bases of high temperature coronal loops is heated quasi - steadily ` to an excellent approximation ' . perhaps surprisingly
, we find that this is also true in the core of a flare productive active region , where the magnetic field is clearly changing . it may be that steady heating dominates in the moss in active region cores , with the variability being due to obscuration by overlying spicular material as suggested by @xcite . in any case
, it is clear that the heating does not give rise to strong flows or motions in the moss , and that the time - scale between impulsive heating events has to be short enough to maintain loops at high temperatures , in contrast to models that assume loops have time to cool substantially between events .
we thank ignacio ugarte - urra and the referee for helpful suggestions .
_ hinode _ is a japanese mission developed and launched by isas / jaxa , with naoj as domestic partner and nasa and stfc ( uk ) as international partners .
it is operated by these agencies in co - operation with esa and nsc ( norway ) .
|
we present new measurements of the time variability of intensity , doppler and non - thermal velocities in moss in an active region core observed by the euv imaging spectrometer on hinode in 2007 , june .
the measurements are derived from spectral profiles of the 195 line . using the 2@xmath0 slit
, we repeatedly scanned 150@xmath0 by 150@xmath0 in a few mins .
this is the first time it has been possible to make such velocity measurements in the moss , and the data presented are the highest cadence spatially resolved maps of moss doppler and non - thermal velocities ever obtained in the corona .
the observed active region produced numerous c- and m- class flares with several occurring in the core close to the moss .
the magnetic field was therefore clearly changing in the active region core , so we ought to be able to detect dynamic signatures in the moss if they exist . our measurements of moss intensities agree with previous studies in that a less than 15% variability is seen over a period of 16 hours .
our new measurements of doppler and non - thermal velocities reveal no strong flows or motions in the moss , nor any significant variability in these quantities .
the results confirm that moss at the bases of high temperature coronal loops is heated quasi - steadily .
they also show that quasi - steady heating can contribute significantly even in the core of a flare productive active region .
such heating may be impulsive at high frequency , but if so it does not give rise to large flows or motions .
|
until recently but for a few exceptions the study of quantum dot ( qd ) heterostructures with the staggered ( type - ii ) band alignment had been largely ignored because of the absence of confinement of one of the two types of carriers and their presumed poor radiative efficiency . however , it has come to be recognized that these structures are interesting , especially for their rich physics of excitons.@xcite in contrast to the usual type - i qds ( e.g. inas / gaas ) where the confinement energy scale is far greater than the energy of the coulomb interaction , the role of confinement in type - ii qds is largely limited to defining the geometry of the system .
this in itself has interesting consequences .
the multiply - connected topology can give rise to an oscillatory ground state energy for the magneto - excitons.@xcite secondly , type - ii qds also act as nanocapacitors@xcite which selectively accommodate only one type of particles ; but once charged they can bind the complementary particle to form an exciton . the strength of the coulomb interaction can be modified by screening or magnetic field.@xcite at higher excitation powers , they can be doubly - charged and form four - particle bound states ( biexcitons ) .
the biexcitons in a type - ii qd system are very unlike their counterparts@xcite in type - i qds and quantum wells .
they always have negative ` binding energy'.@xcite in the atomic physics language , while the usual biexciton is structurally analogous to a hydrogen molecule , the biexciton@xcite in type - ii qds is more like a helium atom.@xcite in this article we have probed the nature of the ensemble photoluminescence ( pl ) emission from a sample with inp qds in a gaas matrix . although the band offsets@xcite of _ bulk _ inp and gaas and some previous studies suggest that this material combination forms type - ii structures with electrons localized within the inp quantum dots and free holes in the gaas matrix , the energy gap of inp and gaas is within @xmath0mev of each other , and the conduction band offset is relatively small.@xcite thus alloying and anisotropic strain within the qds,@xcite can modify the energy gap and the relative offsets in way that is dependent on the details of the size and shape of the qds . for example , a comprehensive k.p calculation @xcite does not find a type - ii alignment in this system .
secondly , thick ( @xmath1 nm ) inp heteroepitaxial layers on gaas and even homoepitaxial inp have shown a broad emission peak at @xmath2 ev@xcite due to donor - acceptor - pair ( da ) recombination , which is rather close to the pl energy of inp / gaas qds .
furthermore , the da emission has characteristics@xcite generally used to classify spatially - indirect excitons from type - ii qds excitation power - dependent blue - shift,@xcite spectrally broad emission at a sub - bandgap energy in macro - pl , and narrow emission lines from localized states in micro - pl.@xcite one of the aims of this study is to unambiguously establish the type - ii band alignment in this system and highlight the role of heterostructure boundary conditions on the size and the binding energy of excitons . in this work
we will not discuss the aharonov - bohm - type effects associated with the topology of the wave function as these have already been extensively discussed in literature@xcite and manifest on energy scales an order of magnitude smaller than we are interested in here . secondly , by extending the measurements to over six orders of magnitude of excitation powers , we have been able to change the average electronic occupancy within the qds .
this leads to significant changes in the diamagnetic shift coefficient that are a result of the interesting physics of biexcitons .
the sample was grown by metal - organic vapor phase epitaxy on ( 001 ) gaas substrate.@xcite the qd density was about @xmath3 with an average diameter of @xmath4 nm ( @xmath5 nm ) and average height of about @xmath6 nm ( @xmath7 nm ) as measured by cross - sectional transmission electron microscopy.@xcite but based on experience from other heterostructures , it is possible that the actual size of quantum dots is much smaller.@xcite non - resonant ( excitation wavelength=532 nm ) pl measurements were performed at liquid helium temperature with the excitation powers varied between ( @xmath8w@xmath9 ) .
the light from the excitation laser and the sample pl was fibre - optically coupled in and out of ( i ) a variable - temperature cryostat within the bore of a superconducting magnet ( b@xmath10 ) for magneto - pl measurements at low excitation powers and ( ii ) a liquid helium bath cryostat whose tail was within the 18 mm bore of the pulsed field coil ( @xmath11 ) for measurements at higher powers @xmath12 .
the pl spectra were recorded by an electron multiplying charged - coupled device after being dispersed by a 30 cm imaging spectrograph .
pulsed magnetic fields of up to 50 t were generated using a 5 kv , 500 kj capacitor bank .
the field had a duration of about 20ms , during which several pl spectra were recorded .
1(inset ) shows three representative spectra under different conditions .
fig.[fig : fig1 ] shows the magnetic field - dependent pl peak - shift measured at @xmath13k , at a very low excitation power of about @xmath14 . using the model of a hydrogenic exciton , one
can semi - phenomenologically describe this shift by the following equations:@xcite @xmath15 these equations are derived under the assumption that the magnetic field - induced change in the ground state energy of a hydrogenic exciton from the low - field regime of quadratic ( diamagnetic ) shift to the high field linear shift ( due to transitions between effectively free landau levels ) is adiabatically continuous ( i.e. its functional form is continuous and differentiable ) between two well - defined limits .
note that the high field limitis approximate as it assumes that the transitions are between _ free _ landau levels and ignores the weak ( @xmath16 , as @xmath17 ) dependence@xcite of the excitonic binding energy on magnetic field . here
@xmath18 is the ground state energy without the magnetic field , @xmath19 corresponds to the magnetic field when @xmath20 , @xmath21 is the excitonic bohr radius , @xmath22 is the magnetic length , @xmath23 is the reduced effective mass , and @xmath24 the magnitude of the electronic charge .
the fit to eq.[eqn : fips ] , along with its physical content , is also shown in fig.[fig : fig1 ] .
the second term in eq.[eqn : fips](b ) corresponds to the excitonic binding energy and is the extrapolation of the high - field slope [ third term in eq.[eqn : fips](b ) ] to @xmath25.@xcite the crossover from quadratic to linear slope is found at @xmath26 t .
the values of the exciton radius , the binding energy and the diamagnetic shift coefficient are given in table 1 .
table 1 also compares the effect of the heterostructure boundary conditions on the excitonic parameters in inp .
note the systematic trend in the value of the diamagnetic shift coefficient .
the effective electron - hole interaction in type - ii dots is the weakest followed by bulk inp , and then finally inp / ga@xmath27in@xmath28p qds which show a much stronger binding [ also see fig.[fig : band diagram ] ] .
thus , on physical grounds , the observations are most consistent with type - ii band alignment .
.comparison of exciton parameters in inp under different boundary conditions [ cols="^,^,^,^ " , ] next , we will explore the multi - particle states in these qds by excitation - power - dependent pl measurements .
these measurements will also help us rule out da recombination , as will be discussed in section iii c. in what follows , we will assume that the qds have type - ii alignment .
the pl peak position , measured without the magnetic field [ fig.[fig : combined figure](b ) ] is strongly excitation power - dependent beyond an incident laser intensity of about @xmath29 w-@xmath9 .
this marks the point where multiparticle states start to play a role in pl .
the observed blue - shift is due to the additional energy associated with the capacitive charging of the qds .
the integrated pl intensity ( not shown here for brevity ) also gradually changes its slope from linear to ( slightly ) superlinear beyond @xmath29 w-@xmath9 , indicating that a fraction of emission is from biexcitonic states .
the dependence of the pl peak position as a function of the magnetic field at different excitation powers , measured over almost six orders of magnitude , is shown in fig.[fig : combined figure](a ) .
notice that the curves are all qualitatively similar and that they all can be fitted to equations [ eqn : fips ] [ solid lines in fig[fig : combined figure](a ) ] .
this is because the b - dependent change in the energy of biexcitonic levels is just the sum of single particle energies , but with the important difference that the biexcitonic radius @xmath30 and hence the diamagnetic shift coefficient are significantly different .
however , for an ensemble , the analysis is complicated by the facts that ( i ) the emission is from a mixture of excitonic and biexcitonic states with an unknown biexciton fraction @xmath31 , ( ii ) when a photon is emitted by a biexciton , what is measured in the magneto - pl experiment is the difference in the shifts of biexcitonic and excitonic levels , because the emission process involves @xmath32 . the measured change in the pl emission energy at low magnetic fields [ diamagnetic shift regime , equation 1(a ) ]
will then be @xmath33b^2 .
\;\;\nonumber ( 2)\ ] ] the diamagnetic shift coefficient @xmath34 is thus expected to be strongly excitation - power - dependent .
indeed this is observed in fig.[fig : combined figure](c ) .
@xmath34 changes by factor of two due to the much smaller biexcitonic radius . recall that in type - i qds , confinement usually renders the diamagnetic shift independent of the charge in the dot.@xcite in general , it depends on the relative extents of the spatial spread of the electron and the hole wave functions . for a rough estimate of the biexciton fraction as a function of excitation power
, we use the ratio of the bohr radii for the helium and the hydrogen atom , @xmath35.@xcite fig.[fig : combined figure](d ) is plotted using this value of @xmath36 and we find that the biexciton fraction at the highest excitation power is about 30% . also note that when @xmath37 , equation 2 predicts that the observed diamagnetic shift coefficient would not only reduce but also become negative at very high excitation powers .
the strong dependence of the diamagnetic shift coefficient on the excitation power also explains the anomalous results of godoy , et al.@xcite , who found the excitonic diameter to be smaller in inp / gaas compared to the bulk inp .
they measured a diamagnetic shift coefficient of between 520@xmath38 , the latter of which is equal to what we measure at the highest excitation power .
fig.[fig : combined figure](e ) shows that the exciton mass ( high field slope of the curves in fig.[fig : combined figure](a ) ) stays at approximately @xmath39 , constant within 15% over the whole range of excitation powers . hence the changes in the diamagnetic shift coefficient can be understood as being largely the difference in the excitonic and biexcitonic radii ( equation 2 ) .
this provides consistency to the analysis .
the value of the mass is between the free electron mass in inp ( @xmath40 ) and the free heavy - hole in gaas ( @xmath41 ) .
this is reasonable because the electron is largely immobilized by the quantum dot .
strain and non - parabolicity effects may further contribute to the enhancement of the electron mass .
the excitation power dependence of the diamagnetic shift coefficient [ fig.3(c ) ] also rules out recombination due to overlap between donor and acceptor wave functions .
da recombination has an energy@xcite @xmath42 @xmath43 is the energy gap , @xmath44 and @xmath45 are the binding energies of the donor and acceptor levels , the fourth term is the coulomb repulsion energy of the ionized centres _ after _
recombination , and @xmath46 are the phonon - assisted transitions.@xcite while the da - pair emission shares some characteristics with emission from type - ii qds such as the diamagnetic shifts from da - recombination would be of the same order as for excitons [ fig.1 ] and at higher excitation powers there would be a blue - shift@xcite qualitatively similar to that seen in fig.3(b ) , the diamagnetic shift coefficient itself should not have an excitation power dependence as observed in fig.3(c ) .
secondly one should expect a strong quenching of the pl intensity in magnetic field for da recombination the wavefunctions shrink in the magnetic field and since the electron and hole centres are not at the same point in space , their ( exponentially small ) overlap will be strongly reduced .
this is also not observed .
we studied the pl from inp / gaas qd heterostructures in very high magnetic fields ( @xmath11 ) over almost six orders of magnitude of excitation powers .
magneto - pl measurements at very low excitation powers established that the excitons have an average bohr radius of @xmath47 nm and a binding energy of @xmath48mev .
these values indicate a much weaker binding for the excitons in comparison with bulk inp and provide strong evidence for type - ii band alignment in these qds .
we also studied the evolution of the electron - hole binding as the qd ensemble makes a gradual transition from a regime where the emission is from ( hydrogen - like ) two - particle excitonic states , to a regime where the emission from ( helium - like ) four - particle biexcitonic states also becomes significant .
this was demonstrated by a strong variation of the diamagnetic shift with the excitation power .
the work at the ku leuven is supported by the fwo and goa programmes and by the methusalem funding of the flemish government .
mh acknowledges the support of the research councils uk .
e. ribeiro , a. o. govorov , w. carvalho , and g. medeiros - ribeiro , phys .
. lett . * 92 * , 126402 ( 2004 ) .
i. r. sellers , v. r. whiteside , i. l. kuskovsky , a. o. govorov , and b. d. mccombe , phys .
lett . * 100 * , 136405 ( 2008 ) .
a. b. kalametsev , v.m .
kovalev , and a.o .
govorov , jetp lett . *
68 * , 669 ( 1998 ) .
m. h. degani , m. z. maialle , g. medeiros - ribeiro , and e. ribeiro , phys .
b * 78 * , 075322 ( 2008 )
. m. k. k. nakaema , f. iikawa , m. j. s. p. brasil , e. ribeiro , g. medeiros - ribeiro , w. carvalho , jr .
, m. z. maialle , and m. h. degani , appl .
lett . * 81 * , 2743 ( 2002 ) .
b. bansal , m. hayne , m. geller , d. bimberg and v. v. moshchalkov , phys .
b * 77 * , 241304(r ) ( 2008 ) .
m. hayne , j. maes , s. bersier , v. v. moshchalkov , a. schliwa , l. mller - kirsch , c. kapteyn , r. heitz , and d. bimberg , appl .
lett . * 82 * , 4355 ( 2003 ) .
m. hayne , j. maes , s. bersier , a. schliwa , l. mller - kirsch , c. kapteyn , r. heitz , d. bimberg and v. v. moshchalkov , physica b * 346 - 347 * , 421 ( 2004 ) .
k. matsuda , s. v. nair , h. e. ruda , y. sugimoto , and t. saiki , appl .
* 90 * , 013101 ( 2007 ) .
j. r. madureira , mrcio p. f. de godoy , maria j. s. p. brasil , and f. iikawa appl .
90 * , 212105 ( 2007 ) .
m. geller , c. kapteyn , l. mller - kirsch , r. heitz , and d. bimberg , appl .
lett . * 82 * , 2706 ( 2003 ) .
n. i. cade , h. gotoh , h. kamada , h. nakano , and h. okamoto , phys .
b * 73 * , 115322 ( 2006 ) . c. e. pryor and m .- e . pistol , phys .
b * 72 * , 205311 ( 2005 ) .
r. l. maltez , e. ribeiro , w. carvalho , jr . , d. ugarte , and g. medeiros - ribeiro , j. appl .
phys . * 94 * , 3051 ( 2003 ) .
d. j. olego , y. okuno , t. kawano , and m. tamura j. appl . phys .
71 , 4502 ( 1992 ) ; m. b. derbali , j. meddeb , h. maref , d. buttard , p. abraham , and y. monteil , j. appl . phys . * 84 * , 503 ( 1998 ) . m. b. derbali , j. meddebb , and p. abraham , thin solid films * 364 * , 192 ( 2000 ) .
b. wang and s .- j .
chua , appl .
78 , 628 ( 2001 ) .
r. timm , h. eisele , a. lenz , s. k. becker , j. grabowski , t .- y .
kim , l. mller - kirsch , k. ptschke , u. w. pohl , d. bimberg , and m. dhne , appl .
. lett . * 85 * , 5890 ( 2004 ) .
r. p. seisyan and b.p .
zakharchenya in _
landau level spectroscopy _
, modern problems in condensed matter sciences , vol .
27.1 , ( north - holland , amsterdam , 1991 ) pp 344 - 443 .
h. weman , m. potemski , m. e. lazzouni , m. s. miller , and j. l. merz , phys .
b * 53 * , 6959 ( 1996).m .
p. f. de godoy , et al . ,
b * 73 * , 033309 ( 2006 ) . m. p. f. de godoy , m. k. k. nakaema , f. iikawa , w. carvalho , jr . , e. ribeiro , and a. l. gobbi , rev .
* 75 * , 1947 ( 2004 ) .
s. b. nam , d. c. reynolds , c. w. litton , t. c. collins , p. j. dean , and r. c. clarke , phys .
b * 13 * , 1643 ( 1976 ) .
m. hayne , r. provoost , m. k. zundel , y. m. manz , k. eberl , and v. v. moshchalkov , phys .
b * 62 * , 10324 ( 2000 ) .
m. sugisaki , h .- w .
ren , s. v. nair , k. nishi , and y. masumoto , phys .
b * 66 * , 235309 ( 2002 )
. a proper theoretical calculation would require solving the schrodinger equation for the four - body problem in presence of suitable boundary conditions and magnetic field , which is beyond the scope of this work . c. klingshirn , pp353 , _ semiconductor optics _
( 2nd ed . , springer , berlin , 2005 ) .
|
we have used magneto - photoluminescence measurements to establish that inp / gaas quantum dots have a type - ii ( staggered ) band alignment . the average excitonic bohr radius and the binding energy
are estimated to be 15 nm and 1.5 mev respectively . when compared to bulk inp ,
the excitonic binding is weaker due to the repulsive ( type - ii ) potential at the hetero - interface .
the measurements are extended to over almost six orders of magnitude of laser excitation powers and to magnetic fields of up to 50 tesla .
it is shown that the excitation power can be used to tune the average hole occupancy of the quantum dots , and hence the strength of the electron - hole binding .
the diamagnetic shift coefficient is observed to drastically reduce as the quantum dot ensemble makes a gradual transition from a regime where the emission is from ( hydrogen - like ) two - particle excitonic states to a regime where the emission from ( helium - like ) four - particle biexcitonic states also become significant .
|
the recent discovery of quasars at @xmath5 ( fan et al
. 2001 ; fan et al .
2003 ) and the subsequent observation of two gunn - peterson ( 1965 ) troughs ( djorgovski et al .
2003 ; white et al . 2003 ) established a new era in studies of the end of the reionization epoch .
the absorption of ly@xmath0 photons from these quasars probes the ionization state of the hydrogen in the intergalactic medium ( igm ) near the quasar redshift and has revealed giant ionized regions ( cen & haiman 2000 ; madau & rees 2000 ) .
comparison of the radii of these regions with the ionizing flux implies quasar ages of @xmath6 years , and an ionization front that is expanding with a relativistic speed ( wyithe & loeb 2004 ) .
the generation of planned low frequency instruments ( such as _ lofar _ or the _ ska _ ) will open a new window on the state of hydrogen in the pre - reionization universe through the emission or absorption of 21 cm photons relative to the cosmic microwave background ( cmb ) by neutral hydrogen in the igm ( scott & rees 1990 ; madau , meiksin & rees 1997 ; gnedin & ostriker 1997 ; shaver et al . 1999 ; tozzi et al .
2000 ; iliev et al .
2002 , 2003 ; ciardi & madau 2003 ; furlanetto , sokasian & hernquist 2003 ; zaldarriaga , furlanetto , & hernquist 2003 ; gnedin & shaver 2003 ; loeb & zaldarriaga 2003 ) .
the giant regions surrounding the highest redshift quasars should be visible via rings of redshifted 21 cm emission from warm but neutral gas beyond the ionization front .
the relativistic expansion of the region implies that the radii measured along and transverse to the line - of - sight using the ly@xmath0 and 21 cm techniques respectively , probe different epochs due to relativistic time delay . in this paper
we discuss what will be learned about the state of the igm surrounding the highest redshift quasars by combining ly@xmath0 and 21 cm measurements . in [ ssp ] we discuss the measurements of radii of the region in ly@xmath0 absorption and redshifted 21 cm emission .
the relation between these radii and the neutral fraction of the igm are then discussed in [ comb ] .
some examples of the brightness temperature profiles around high redshift quasars are presented in [ ex ] . in [ fossil ]
we discuss the observation of fossil regions and their utility as a probe of the radiation background at high redshift . finally in
[ global ] we discuss the utility of quasar strmgren spheres in the calibration of the global signature of reionization .
[ disc ] summarizes our main results . throughout the paper
we adopt the set of cosmological parameters determined by the _ wilkinson microwave anisotropy probe _ ( wmap , spergel et al .
2003 ) , namely mass density parameters of @xmath7 in matter , @xmath8 in baryons , @xmath9 in a cosmological constant , and a hubble constant of @xmath10 .
first , we consider the observed signatures of regions ( strmgren spheres ) around the highest redshift quasars during their early relativistic expansion . neglecting recombinations which do not become important until the expansion becomes sub - relativistic , white et al .
( 2003 ) derived the evolution of the physical radius of the strmgren sphere ( @xmath11 ) as a function of the source age ( @xmath12 ) which can be easily extended to include the possibility of a non - zero radius ( @xmath13 ) at the initial time @xmath14 , @xmath15 where @xmath16 is the production rate of ionizing photons by the quasar , @xmath17 is the co - moving hydrogen number density , @xmath18 is the neutral fraction and @xmath19 is the speed of light .
this cubic equation may be solved algebraically ( white et al .
alternatively , @xmath20 may be found by integrating the differential equation @xmath21 with the initial condition @xmath22 . the latter approach ( wyithe & loeb 2004 ) allows for the inclusion of recombinations .
as long as the number of photons emitted per time interval @xmath23 is larger than the number of hydrogens in a shell of thickness @xmath24 at @xmath11 , the region continues to expand at nearly the speed of light .
calculations of @xmath11 in this paper assume that the ionization front is thin .
the spectrum averaged mean - free path for the ionizing quasar photons is @xmath25mpc , which may be compared to the bubble radius @xmath26mpc , to yield the fractional thickness of the ionization front @xmath27^{-3}$ ] .
the assumption of a thin ionization front may be considered reasonable provided that @xmath28 .
a low neutral fraction is unlikely given the size of the observed regions ( @xmath26mpc ) around the @xmath29 quasars ( wyithe & loeb 2004 ) .
x - ray photons have a longer mean free path but are rare and do not contribute significantly to the number of ionizations .
equations ( [ se1 ] ) and ( [ se2 ] ) refer to the strmgren sphere radius at a time @xmath12 after the source turns on
. however measurements of the radius through ly@xmath0 absorption ( @xmath30 ) are made along the line of sight .
ly@xmath0 photons emitted at @xmath12 are not absorbed until a time @xmath31 satisfying the relation @xmath32 ( note that the quasar may no longer be active at @xmath31 ) . when considering the value of @xmath30 measured through ly@xmath0 absorption , the relevant time to consider is therefore the age of the quasar at the time when the photons being absorbed at the edge of the region
were emitted @xmath33 . in the absence of recombinations ( white et al .
2003 ) , the relation between @xmath30 and @xmath34 is equivalent to the infinite speed of light solution for the evolution of a strmgren sphere radius ( haiman & cen 2000 ; madau & rees 2000 ) , @xmath35^{1/3}\end{aligned}\ ] ] where we add the subscript @xmath36 to denote the isotropic equivalent value of @xmath37 for a measurement done parallel to the line - of - sight .
madau , meiksin & rees ( 1997 ) , and tozzi et al .
( 2000 ) have considered the signature of the igm surrounding a high redshift quasar . outside the region
, neutral hydrogen may be heated by soft x - rays and by scattering of ly@xmath0 photons .
the spin temperature of the neutral hydrogen is coupled to the kinetic temperature of the igm through ly@xmath0 scattering of a background of uv photons , and results in 21 cm emission ( in excess of the cmb flux ) around the quasar .
the spin temperature of a warm igm quickly becomes coupled to the cmb radiation , so that it yields no contrast for observations ( madau , meiksin & rees 1997 )
. however scatterings of ly@xmath0 photons couple the spin temperature of the warm igm at radii @xmath38 through the wouthuysen - field effect ( wouthuysen 1952 ; field 1958 ) .
the spin temperature @xmath39 is weighted between the cmb temperature and the igm temperature through the ly@xmath0 pumping efficiency @xmath40 in equation ( [ couple ] ) , @xmath41k is the temperature corresponding to the transition energy , @xmath42s@xmath43 is the spontaneous decay rate of the hyperfine transition , and @xmath44 is the ly@xmath0 scattering rate per hydrogen . the value of @xmath44 may be calculated for a continuum uv source embedded in an expanding igm around a strmgren sphere as follows ( see loeb & rybicki 1999 for more details on the related physics ) . in the absence of scattering ,
photons that start at a frequency separation @xmath45 on the blue side of the ly@xmath0 resonance frequency @xmath46 would redshift into resonance at some distance @xmath47 from the source , where @xmath48 is the hubble parameter at the quasar redshift . at this distance , the hubble velocity of the igm introduces a doppler shift that is equal to the initial blueshift of these photons relative to the ly@xmath0 line center .
when a resonant photon enters the neutral igm near @xmath49 , its mean - free - path becomes very short because of the large cross - section to ly@xmath0 scattering . as the photon frequency redshifts very close to the ly@xmath0 resonance , the mean - free - path diminishes , and the frequency shift per scattering ( which corresponds to the hubble velocity increment across a mean - free - path ) goes to zero .
consequently , the photon spends a very long time around @xmath50 . at this stage ,
the finite temperature of the gas becomes important .
the photons are able to diffuse out of resonance primarily due to the doppler shifts they encounter by the thermal speed of the atoms at @xmath49 .
each scattering near @xmath49 leads to a fractional frequency shift by the thermal doppler effect of @xmath51 , where @xmath52 is the thermal speed of the atoms and @xmath46 is the resonant ly@xmath0 frequency .
the thermal shifts are of random sign and so the photons go through a random walk in frequency .
consider the photons just around the resonant frequency at @xmath49 .
these photons change their frequency by @xmath53 in each scattering , and have a mean free path of @xmath54 $ ] . here
@xmath55 is the ly@xmath0 cross - section at the center of the thermally broadened ly@xmath0 line ( rybicki & lightman 1979 ) @xmath56 where @xmath57 and @xmath58 are the electron charge and mass , and @xmath59 is the oscillator strength for the ly@xmath0 transition .
the mean intensity of nearly resonant photons at @xmath49 is increased by a factor of @xmath60 relative to optically - thin conditions , where @xmath61 is the so - called sobolev length , defined as the distance over which the bulk velocity gradient of the medium equals the thermal speed of an atom , @xmath62 and @xmath63 is the hubble velocity .
the optical - depth factor @xmath64 reflects the fact that the diffusion speed of photons across the resonance region of width @xmath65 is a factor @xmath66 times smaller than the speed of light . in a steady state ,
the mean intensity at all radii should therefore be enhanced relative to the optically - thin case by a factor of @xmath67 . however
, a steady state will not be established immediately when light from the quasar first penetrates the neutral igm because of the slow diffusion speed of the resonant photons . at early times
, scattering will produce an exponential reduction in resonant photon flux by a factor @xmath68 .
even though photons may be sufficiently blue that they do not redshift into the resonance until @xmath49 , they would still be absorbed by the blue wing of the ly@xmath0 resonance ( where the cross - section is still considerable ) at radii smaller than @xmath49 .
let us therefore consider photons of arbitrary frequency on the blue side of the ly@xmath0 resonance and calculate their mean intensity at @xmath49 .
we define the mean intensity as the angular average of the specific ( directional ) intensity @xmath69 , namely @xmath70 where @xmath71 is the direction relative to the radius vector at @xmath49 ( loeb & rybicki 1999 ) .
the ly@xmath0 optical depth for photons with frequency @xmath72 at a radius @xmath49 is @xmath73\right)n^0_{\rm h}(1+z)^3,\ ] ] where @xmath74 is the thermally - broadened cross - section ( rybicki & lightman 1979 ) , and @xmath75 is the path - length traversed by a photon through the neutral igm after it had crossed the boundary of the region at a time @xmath76 . in the strong scattering regime ( @xmath77 ) , the mean intensity is a factor of @xmath78 larger than the specific intensity along the radial direction , because the radial diffusion speed of the photons is @xmath79 .
the initial mean intensity of nearly resonant photons at radius @xmath49 is therefore @xmath80 where @xmath81^{-1}\ ] ] is the redshifted source spectrum @xmath82 in @xmath83 .
the scattering rate @xmath44 may then be calculated from the mean intensity as @xmath84 the directional intensity of resonant ly@xmath0 photons starts high near the boundary of the region at @xmath11 , but is initially suppressed at larger radii .
resonant photons will diffuse to radii larger than @xmath11 at a speed @xmath85 . in the case of a stationary front
, the high intensity region expands with time , extending a distance @xmath86 beyond the edge of the region after a time @xmath23 , and absorption will truncate the intensity at radii larger than @xmath87 . for the quasars under consideration the expansion speed of the region ( @xmath88 ) is well in excess of the diffusion speed transition of the expanding igm .
] of resonant ly@xmath0 photons ( @xmath89 ) . on the timescales of interest here
, there is a deficit of photons near the line center as described by equation ( [ i ] ) .
the critical thermalization rate above which ly@xmath0 scattering couples @xmath90 to @xmath91 is given by ( madau , meiksin & rees 1997 ) @xmath92 we find that @xmath44 only exceeds the thermalization rate @xmath93 within a negligibly thin shell around the region ( @xmath94mpc ) . as a result ,
detection of redshifted 21 cm emission from the regions of quasars will require the presence of a uv background .
background photons between the ly@xmath0 and the lyman - limit frequencies can propagate through the universe and redshift with time to the point where they resonate with the ly@xmath0 line and couple @xmath95 to @xmath96 .
such a background is expected to be present at @xmath97 since the large electron scattering optical depth measured by wmap ( kogut et al .
2003 ) implies that the igm had been significantly reionized by that time ( requiring the production of more than one ionizing photon per baryon above the lyman - limit ) .
our result for the scattering rate should be compared with the result of madau , meiksin & rees ( 1997 ) who estimated @xmath98 in the steady state this expression incorrectly neglects the rise in intensity that follows from the small diffusion speed of photons near the ly@xmath0 resonance and hence underestimates the value of @xmath44 by a factor of @xmath99 which amounts to several orders of magnitude .
however , in the initial state the expression neglects the absorption of resonant photons before they reach a radius @xmath38 , and so overestimates the value of @xmath44 outside the region by several orders of magnitude .
the scattering of ly@xmath0 photons may also provide a source of heating for the gas . as this peak shifts to longer wavelengths , it loses energy to the hydrogen atoms off which the photons scatter ( chen & miralda - escude 2003 ) . the additional energy density due to the scattering is @xmath100 the rate at which energy is lost from the line in an expanding igm equals the heating rate of the gas .
hence we have @xmath101 this heating rate , which is proportional to the ly@xmath0 scattering rate , does not contribute to the heating of gas outside the region owing to the very low ly@xmath0 intensity there . instead , the heating is dominated by x - rays which we discuss next .
gas in the igm outside the ionized region is subject to secondary electron heating resulting from ionizations by soft x - rays that propagate into the neutral igm beyond the region .
the heating rate @xmath102 in @xmath103 at radius @xmath38 due to a source with a spectrum @xmath82 ( in erg / sec / hz ) is @xmath104,\end{aligned}\ ] ] where @xmath105 is the cross - section for photoionization and @xmath106 is the lyman - limit frequency corresponding to the ionization threshold of hydrogen . in equation ( [ heat ] ) @xmath107 is the path - length of a photon through the neutral igm after it crossed the boundary of the region at time @xmath76 .
we have neglected adiabatic cooling due to cosmological expansion in this expression since the quasar lifetime is much shorter than the hubble time .
the fraction of photon energy converted into heat , @xmath108 , may be evaluated using the fitting formula ( shull & van steenberg 1985 ) @xmath109^{1.3163}\right).\ ] ] this equation is valid for photon energies @xmath110ev . at lower energies , the value of @xmath108 may be roughly approximated by multiplying equation ( [ fx ] ) with an additional factor of @xmath111/100\mbox{ev}\right)}$ ] ( with the constraint that @xmath112 ) .
the residual ionization fraction from cosmological recombination of @xmath113 yields @xmath114 ( madau et al .
1997 ) , which sets the minimum heating rate .
the heating rate leads to a temperature rise according to @xmath115 the above discussion has shown that x - rays from the quasar are capable of heating the gas outside the region , and that the spin temperature of this warm gas can be coupled to its kinetic temperature in the presence of a strong uv ( ly@xmath0 ) background . in regions of the igm where the spin temperature is decoupled from the cmb temperature , a differential antenna temperature of @xmath116 is observed . in equation ( [ delt ] ) we have assumed a uniform igm at the mean density with a neutral fraction @xmath117 .
the luminous high redshift quasars reside in rare high - density regions .
however on the scale of several mpc , the overdensity due to infall around these regions is @xmath118 ( barkana & loeb 2003a ) .
the brightness temperature of the redshifted 21 cm signal is small and subject to severe foreground contamination problems .
studies by di matteo et al .
( 2002 ) , and oh & mack ( 2003 ) have found that on scales of several arc - minutes , randomly distributed point sources would have to be removed down to a level of @xmath1191@xmath120jy before brightness temperature fluctuations of order 10 mk could be detected . in the presence of the expected level of clustering of these sources , even the removal of point sources below 1@xmath120jy would leave foreground fluctuations on this scale that are several orders of magnitude in excess of the redshifted 21 cm signal .
however , the spectrum of the foreground contaminants is expected to be smooth ( shaver et al .
the width of the region corresponds to a recessional velocity @xmath121 or a redshift interval of @xmath122 .
the redshifted 21 cm ring will therefore produce a bump ( or a dip ) on top of the smooth foreground spectrum at the redshift of the quasar . at a frequency @xmath123mhz ,
an instrument like _ lofar _ is expected to have a band - pass of @xmath124mhz .
this should be compared with the frequency shift associated with the redshift difference between light emitted at the quasar and the edge of the region along the line of sight @xmath125 since many channels are contained within each band - pass , the structure of the region can therefore be probed along the line - of - sight in frequency space ( madau , meiksin & rees 1997 ) , allowing determination of the full three dimensional shape of the region , including the radius of the region behind the quasar in addition to the radius @xmath30 of the region in front of the quasar .
there are two distinct possibilities for the redshifted 21 cm signature of a high redshift region .
first , the region might be expanding into a cold igm , in which case emission will only be seen out to a radius where the warming x - rays can reach .
alternatively , the igm may have been pre - heated , either by an x - ray background , by gravitationally induced shocks in the igm ( furlanetto & loeb 2003 ) , or by an early reionization ( e.g. wyithe & loeb 2003a ; cen 2003 ) .
the presence of gunn - peterson troughs in quasar spectra at @xmath1 implies an igm that is at least partially neutral , while the relatively small sizes of the strmgren spheres around these quasars suggest a neutral fraction that is of order unity ( wyithe & loeb 2004 ) . on the other hand , observations from wmap ( kogut et al .
2003 ) suggest an optical depth to electron scattering of @xmath126 and therefore significant reionization at @xmath127 .
if the universe were reionized twice , then the strmgren spheres would be expanding into a partially neutral , but warm igm .
the most natural place to search for the redshifted 21 cm signature of quasar regions is around the highest redshift quasars discovered in wide field surveys such as the _ sloan digital sky survey _ ( sdss ) .
such quasars have regions that are also probed via their ly@xmath0 absorption spectra . in this section
we discuss the complimentary nature of the two observations with respect to the determination of the quasar lifetime , @xmath12 , and the neutral fraction of the igm , @xmath117 .
strmgren spheres expand faster into a partially ionized igm . if the igm has a neutral fraction @xmath117 , then the observation of @xmath30 through ly@xmath0 absorption measures the combination @xmath128 ( where here we have neglected recombinations and assumed @xmath129 for simplicity ) .
detection of the redshifted 21 cm ring reflects the state of the region at @xmath34 .
the differential antenna temperature @xmath130 is proportional to @xmath117 due to the small optical depth at 21 cm , @xmath131 measurement of the contrast in the emission and absorption rings , or of the brightness temperature of the emission ring , will facilitate a measurement of @xmath117 independent from @xmath34 . redshifted 21 cm observations may therefore provide an opportunity to break the degeneracy between @xmath34 and @xmath117 .
unfortunately , the heating rate and temperature of the igm are uncertain , and unless @xmath132 , the value of @xmath130 will be sensitive to the unknown value of @xmath133 . below we discuss an alternative geometric method for finding @xmath117 that is not sensitive to the precise value of @xmath134 .
the relativistic expansion and finite light travel time imply that the 21 cm and ly@xmath0 measurements sample two different epochs in the evolution of the region .
the situation is illustrated schematically in figure [ fig1 ] .
suppose that the quasar turns on at @xmath135 , emitting photons isotropically into an existing region of radius @xmath136 .
ly@xmath0 photons emitted at @xmath135 will reach the edge of the region at @xmath137 . at that time
, 21 cm photons with @xmath138 are emitted by the heated gas at @xmath139 and travel towards the observer .
for simplicity consider 21 cm photons that are emitted from the heated igm just beyond the ionizing front and perpendicular to the line - of - sight ( i.e. at the quasar redshift ) .
the sphere continues to expand at nearly the speed of light between @xmath137 and @xmath140 . at @xmath140 ,
the ly@xmath0 photons emitted at @xmath137 reach the edge of the region .
the absorption of these photons encodes a measurement of @xmath141 .
the continuum and 21 cm photons propagate to the observer , who measures a radius @xmath141 along the line - of - sight in ly@xmath0 absorption , and a different radius @xmath136 in redshifted 21 cm photons .
in this paper we define @xmath30 to be the radius measured through ly@xmath0 absorption along the line - of - sight .
this is compared with @xmath142 , defined to be the radius of the ring measured perpendicular to the line of sight through redshifted 21 cm emission .
the radii @xmath143 and @xmath142 are not equal for a spherical region due to finite light travel time effects mentioned in the previous paragraph .
we note that this effect is separate from the super - luminal transverse expansion of the ring of 21 cm emission from the relativistically expanding portions of the region near the line - of sight .
these super - luminally expanding rings would be seen @xmath144km@xmath145s@xmath43 blueward of the quasar redshift ( for @xmath146mpc ) .
the exact redshift of the quasar may be uncertain by up to @xmath147km@xmath145s@xmath43 ( richards et al .
however this uncertainty corresponds to only 30% of the diameter of the region ( @xmath144km@xmath145s@xmath43 ) .
a point on the edge of the region that is blueshifted by @xmath147km@xmath145s@xmath43 relative to the quasar lies in a direction away from the quasar that makes an angle of @xmath148 with the line - of - sight . as a result
, the misidentification of the quasar redshift will lead to an error of @xmath149 in the transverse diameter of the ring at the quasar redshift .
moreover , since @xmath150 , the ring of 21 cm emission at the observed quasar redshift will not be subjected to superluminal expansion , even if the quasar redshift is misidentified by @xmath147km@xmath145s@xmath43 .
more quantitatively , the relativistic expansion and finite light travel time lead to several interesting possibilities : * the coupling of the spin temperature of neutral hydrogen to the temperature of the warm igm surrounding the region results in redshifted 21 cm radiation in excess of the cmb .
@xmath142 is measured transverse to the line - of - sight and so corresponds to the radius at the time the observed uv photons were emitted .
we can directly infer the ionizing photon rate ( @xmath151 ) which drives the evolution of the region along the line of sight .
however , since the quasar emission may not be isotropic , we allow the evolution of the region transverse to the line of sight to be driven by a different ionizing photon rate @xmath152 .
thus , the ratio @xmath153 is a measure of the anisotropy in quasar emission . since high - redshift quasars
are brightness selected , we assume that @xmath154 .
using equation ( [ evinf ] ) , we find the age of the quasar when light crossing the edge of the region was emitted ( @xmath34 ) and substitute it into equation ( [ se1 ] ) for the radius of the region _ at that time_. this yields the following relationship between @xmath30 and @xmath142 @xmath155 or @xmath156.\end{aligned}\ ] ] note that equations ( [ rr1 ] ) and ( [ rr ] ) are independent of @xmath136 when the quasar emission is isotropic , and that the dependence of @xmath157 on @xmath153 is rather mild .
if the quasar emission is isotropic , the transverse radius @xmath142 is always smaller than the line - of - sight radius @xmath30 , particularly during the relativistic expansion phase . when the expansion becomes sub - relativistic , @xmath142 approaches @xmath30 .
* the radii measured through ly@xmath0 absorption and 21 cm emission correspond to different epochs separated by @xmath158 . for a spherical region
we obtain the average expansion velocity of the ionizing front between the times corresponding to @xmath142 and @xmath30 @xmath159 note that this velocity is the physical velocity of the ionizing front .
the apparent transverse velocity of portions of the ring might be superluminal . * given a fixed luminosity , the expansion speed at a radius @xmath143 is proportional to @xmath160 , and so the neutral fraction can be inferred from the two radii . equation ( [ rr ] ) yields @xmath161^{-1}\\ \nonumber & = & \left[\frac{\left({r_{\rm hi}}/{r_{\rm ly}}\right)}{1-\left(\frac{r_{\rm hi}}{r_{\rm ly}}\right)^3\left(\frac{\dot{n}_\parallel } { \dot{n}_{\perp}}\right ) - \left(\frac{r_0}{r_{\rm ly}}\right)^3\left(\frac{\dot{n}_\parallel } { \dot{n}_{\perp } } -1\right)}\right]\\ & & \hspace{-0mm}\times\left(\frac{r_{\rm ly}}{3.25\mbox{mpc}}\right)^{-2 } \left(\frac{\dot{n}_\parallel } { 10^{57}\mbox{s}^{-1}}\right ) \left(\frac{1+z}{7.5}\right)^{-3 } .\end{aligned}\ ] ] some examples of the dependence of @xmath117 on the ratio @xmath162 given @xmath163mpc and @xmath164s@xmath43 , are shown in figure [ fig2 ] . in the case of isotropic quasar emission ( @xmath165 ) , a small ratio of @xmath162 implies relativistic expansion , and hence a small neutral fraction unless @xmath166mpc .
values of @xmath167 which are less than unity lead to larger inferred values of @xmath117 .
similarly , a non - zero value for @xmath136 leads to larger values of @xmath117 .
the assumptions of isotropy and @xmath129 lead to determination of a lower limit for @xmath18 .
the observation of a redshifted 21 cm ring with a radius greater than 2.5mpc ( 0.5mpc ) around the known @xmath29 quasars ( with @xmath168mpc ) would provide a limit on the neutral fraction of @xmath169 ( @xmath170 ) .
in this section we present examples of the redshifted 21 cm signatures of @xmath5 quasars that will be observed by instruments like _ lofar_. the uv continuum blueward of ly@xmath0
is assumed to be that of the median spectrum in the sample of hst quasars studied by telfer , zheng , kriss & davidsen ( 2002 ) .
they found a power - law @xmath171 with @xmath172 .
this slope may be extended into the x - ray regime based on the results of yuan , brinkmann , siebert & voges ( 1998 ) .
we include the ly@xmath0 emission line centered at @xmath173 , with an equivalent width of @xmath174(vanden berk et al .
2001 ; telfer , zheng , kriss & davidsen 2002 ) , and a line width of @xmath175(vanden berk et al .
2001 ) , providing a uv x - ray spectrum @xmath176.\end{aligned}\ ] ] the normalization of this spectrum is appropriate for a @xmath177m@xmath178 supermassive black hole accreting at its eddington rate ( elvis et al .
1994 ) , and it yields an ionizing photon rate of @xmath179s@xmath43 , representative of the @xmath180 sdss quasars ( white et al .
2003 ) . in this section ,
we show examples of the 21 cm emission signal that may be observed around the known very high redshift quasars , with the spectrum described by equation ( [ spec ] ) .
the level of redshifted 21 cm emission is enhanced through heating of the igm by the quasar x - ray emission .
the corresponding heating rates @xmath181 of the above quasar spectrum are plotted as a function of radius in figure [ fig3 ] .
the three curves correspond to radii where photons traversed path lengths through neutral gas of @xmath182 , 0.1@xmath11 and 0.9@xmath11 .
the neutral fraction was set to unity in these examples .
note that lower neutral fractions lead to higher values of @xmath108 and hence larger heating rates .
if the strmgren spheres expand into a pre - heated igm containing a uv background that coupled the spin temperature of the igm to its kinetic temperature , the regions should appear as holes in the sky of redshifted 21 cm emission .
as an example we first consider an igm that is pre - heated to @xmath183k at @xmath184 ( @xmath185 ) .
we follow the evolution of the strmgren sphere , and at each time compute the heating rate of the igm beyond the edge of the region using equation ( [ heat ] ) .
the evolution of the kinetic temperature of the igm is followed on a grid using equation ( [ temp ] ) .
the spin temperature is then set to @xmath91 as appropriate in the presence of a strong ly@xmath0 background .
figure [ fig4 ] shows four examples of the 21 cm emission signature of a stromgren sphere expanding into a warm igm as a function of radius ( top axes ) and apparent angle ( bottom axes ) for different combinations of @xmath186 and @xmath187 .
an initial value of @xmath129 was assumed .
the y - axes are labeled by the differential antenna temperature @xmath130 ( right axes ) , and the flux @xmath188 ( left axes ) assuming the design goal sensitivity of _ lofar _ @xmath189 and a top - hat beam with a diameter of @xmath190 . in each case , four profiles are shown for observing times corresponding to quasar ages of @xmath191yr at the time when the _ optical photons _ observed along the line - of - sight were emitted .
the lower row includes cases with @xmath192 . the smaller neutral fraction results in larger radii , faster relativistic expansion , and therefore thinner emission shells .
the 21 cm emission signal is weaker in this case because @xmath193 .
there is excess emission above the level of the general igm in a ring covering the region where gas has been heated by the quasar .
however the brightness temperature is only slightly higher than the rest of the igm .
this is due to the temperature dependence of @xmath194 which asymptotes to a constant value of @xmath195mk for high temperatures .
the panels on the right show examples with @xmath196 . the lower transverse flux results in an expansion that becomes sub - relativistic at earlier times .
as a result , the emission shell has a smaller radius and a larger thickness . as our second example we consider an igm that is cold in neutral regions [ @xmath197 , corresponding to adiabatic cooling following the thermal decoupling of the baryons from the cmb ] .
the resulting 21 cm signal is shown in figure [ fig5 ] .
the heated portions of the igm beyond the edge of the region show emission , while the cold igm produces absorption of the cmb at larger radii .
the brightness temperature of the emission ring obtains values of @xmath198mk , smaller in amplitude than the absorption signal of @xmath199mk .
a cold igm therefore offers the best contrast for observation of the transverse extent of the region . as in the previous example
, the emission rings get thicker at late times because the expansion speed becomes less relativistic .
since the transverse expansion becomes sub - relativistic sooner than the line - of - sight expansion , the rings become thick at earlier times for a lower value of @xmath153 .
figure [ fig6 ] shows @xmath200 ( solid dark lines ) as well as the inner radius of the 21 cm emission ring ( solid light lines ) for the example presented in the top left panels of figures [ fig4 ] and [ fig5 ] ( @xmath165 , @xmath201 ) .
the four epochs corresponding to the differential antenna temperature profiles plotted in figures [ fig4 ] and [ fig5 ] are indicated by the vertical tick marks .
similarly , values of @xmath30 at @xmath202yr are indicated by tick marks in the panels of figures [ fig4 ] and [ fig5 ] .
the value of @xmath30 is always larger than the radius of 21 cm emission .
the dashed lines show the velocity at which the edge of the region is moving when observed at a time corresponding to @xmath34 .
the 21 cm ring ( light dashed lines ) is observed at an earlier phase of evolution and so is moving at a higher velocity .
the edge of the region is moving at a speed @xmath203 when @xmath204mpc .
the presence of a ly@xmath0 background has very interesting implications for the observation of fossil regions .
the recombination time within an ionized region of the igm at mean density is @xmath205^{-3}$ ] years , where @xmath206@xmath207s@xmath43 is the case - b recombination coefficient at a temperature of @xmath208k .
thus , the recombination time at the mean cosmic density exceeds the hubble time for @xmath209 .
once a quasar has turned off , it therefore leaves behind a fossil region that survives recombination .
this region will therefore remain detectable in redshifted 21 cm radiation in both scenarios discussed in this paper .
since the bright sdss quasars are likely to be active for only @xmath210 of the time ( wyithe & loeb 2003b ) , there should be of order 100 times as many 21 cm emission rings around fossil regions as around active quasars .
furthermore , the hierarchical buildup of the supermassive black holes powering the quasars would result in an even larger number of small fossil regions .
how many regions should an instrument like _ lofar _ detect ?
it is anticipated that near frequencies of @xmath211mhz , the virtual core of the planned instrument _ lofar _ will have a beam - size of @xmath212 , a resolution of several tens of arc - minutes , and a system temperature of @xmath213k .
the instrument will have an effective collecting area of @xmath214m@xmath215 at this frequency , leading to a resolution of @xmath216 for the telescope in a filled aperture mode .
higher resolutions are obtained by spreading the array out over longer baselines , at the cost of decreased sensitivity . from the radiometer equation ( e.g. burke & graham - smith 1997 ) we can find the root - mean - square noise in the brightness temperature as a function of resolution on angular scales @xmath217 of @xmath218 here
@xmath219 is the detector band - pass , and @xmath220 is the integration time .
the noise in the brightness temperature is constant on scales larger than @xmath221 .
the root - mean - square noise for an observation with a resolution @xmath222 therefore becomes comparable to the expected emission signal ( @xmath223mk ) for a modest integration time of @xmath224hr .
the co - moving optical luminosity function of bright quasars at @xmath225 in units of @xmath226gpc@xmath227 ( fan et al .
2003 ) is @xmath228 where @xmath229 is the logarithmic slope at the bright end . given a band - pass @xmath124mhz and a central frequency @xmath230mhz , the number of active quasars per logarithm of luminosity per field of view is @xmath231 where @xmath232 is the co - moving volume per unit redshift per unit solid angle . equation ( [ n ] ) yields a value of @xmath233 bright @xmath225 quasars per _ lofar _ field .
the number of redshifted 21 cm rings or holes in the emission signature of the igm per field as determined by equation ( [ n ] ) applies to active quasars .
however as noted above , the 21 cm emission may persist after the quasar has turned off .
this is because in the absence of other heating sources , gas heated to temperatures far in excess of the cmb will cool adiabatically .
the @xmath234 dependence of the temperature in this regime should ensure that the gas remains at a temperature in excess of the cmb by the time it is re - heated during the next episode of quasar activity .
we might expect to find significant numbers of fossil regions , provided that there is continued coupling of @xmath95 and @xmath91 due to a uv background .
the number of detectable fossil regions would then equal the number in equation ( [ n ] ) multiplied by the inverse of the quasar duty cycle @xmath235 ( wyithe & loeb 2003b ) , or about 200 per field .
_ lofar _ should also be sensitive to regions around less luminous quasars .
the brightness temperature of the emission ring , or of the igm outside the region has a mild dependence on the quasar luminosity , and its diameter scales only with the one third power of the luminosity .
regions around less luminous quasars should be more numerous by a factor proportional to @xmath236 . for example , detection of redshifted 21 cm emission around quasars with luminosities a factor of 10 smaller than those of the bright sdss quasars require a resolution that is better by a factor of @xmath237 , corresponding to an integration time that is longer by a factor of @xmath238 .
however , the number of active high redshift quasars per field powering such emission rings is likely to be of order @xmath239 , while the number of fossils regions could be @xmath240 .
finally , we briefly discuss the example of a redshifted 21 cm signature around a high redshift quasar presented in 5b of tozzi et al .
their example shows a ring of redshifted 21 cm emission inside a ring of redshifted 21 cm absorption .
this morphology is unlikely to be observed because the observation of an absorption ring requires the absence of a uv background , and the provision of coupling of the spin and kinetic temperatures through ly@xmath0 photons from the quasar itself .
as we have shown , ly@xmath0 photons from the quasar are unable to couple the spin and kinetic temperatures .
an emission ring enclosed within an igm that absorbs everywhere , as shown in figure [ fig5 ] is the correct signature of a strmgren sphere expanding into a cold igm ( provided that there is a uv background ) .
a very exciting possibility for instruments like _ lofar _ is to search for the emission ( or absorption ) of redshifted 21 cm radiation from the neutral igm ( averaged over large scales ) before reionization . by finding the frequency at which this emission ( absorption ) cuts off
, one might determine the redshift of reionization ( e.g. shaver et al .
if reionization took place over an extended period of time as suggested by recent work ( barkana & loeb 2003b ) , then this global reionization signature will be difficult to detect for two primary reasons : ( i ) because it will be superimposed on a smooth but varying foreground spectrum , and ( ii ) calibration needs to be performed over a frequency range that is broader than the detector band - pass .
arc - minute resolution observations of quasar strmgren spheres would provide an alternative method to search for a step in the brightness temperature of the igm due to reionization .
this is because the inner part of the region provides a calibration of the brightness temperature of the radio foreground that is unavailable along other lines of sight .
note that 21 cm emission from the quasar itself does not affect this calibration , since it effectively just adds to the foreground .
observations along the line - of - sight to the quasar would include emission within the spectral ranges corresponding to both the ionized region and the neutral gas within the same band - pass . redshifted 21 cm observations of quasars at a series of redshifts
would therefore allow determination of the brightness temperature of the igm at different cosmic times , and a detection of the onset of reionization .
redshifted 21 cm observation of highest - redshift sdss quasars will immediately determine whether the igm was significantly neutral at @xmath241 .
observations of the reionization epoch in redshifted 21 cm radiation with proposed instruments like _ lofar _ will offer a unique opportunity to study the state of the igm at the end of the dark ages .
the warm neutral hydrogen outside the regions surrounding bright high redshift quasars is expected to emit at 21 cm .
the brightness temperature contrast of this emission will probe the neutral fraction ( with a lower contrast implying a lower neutral fraction ) , although the inference will be subject to uncertainties in the temperature of the warm igm .
alternatively , the relativistic expansion of the regions provides a geometric method to determine the neutral fraction as well as the quasar lifetime .
we have combined the observed ly@xmath0 absorption signal with expectations for redshifted 21 cm signatures of the regions around the highest redshift sdss quasars . because the regions around the known @xmath29 quasars are at a stage of relativistic expansion , the line - of - sight and transverse radii ( measured through ly@xmath0 absorption and redshifted 21 cm emission , respectively ) , correspond to two different epochs in the evolution of the regions .
we have found that measurements of these two radii can be used to determine the neutral fraction of the igm with only a mild dependence on the anisotropy of the quasar emission ( see fig .
2 ) . assuming isotropic quasar emission and the absence of a fossil region prior to the quasar activity , the observation of a redshifted 21 cm emission ring with a radius greater than 2.5mpc ( 0.5mpc ) around the known @xmath29 quasars , would provide a lower limit on the neutral fraction of @xmath242 ( @xmath170 ) .
emission concentrated close to the line - of - sight and the existence of fossil regions around the quasar host galaxy would imply even larger neutral fractions . at low redshifts ,
quasar lifetimes are bracketed by a variety of techniques to the range @xmath243@xmath244yr ( e.g. yu & tremaine 2002 ; martini & weinberg 2001 ; haiman & hui 2001 ; jakobsen , jansen , wagner & reimers 2003 ) , with preferred values near @xmath245 years ( e.g. martini 2003 ) .
the lifetime required for the region to expand to its observed size is proportional to @xmath160 .
the geometric determination of the neutral fraction can therefore be directly used to determine also the quasar lifetime measured from the combination of @xmath11 and @xmath117 .
thus , ly@xmath0 and redshifted 21 cm measurements of the expanding regions around the highest redshift quasars will allow a direct determination of quasar ages on a case by case basis and constrain quasar formation models ( e.g. haiman & loeb 2001 ) . based on current number counts of bright high redshift quasars and the preliminary design of _ lofar _ , we forecast that each _ lofar _ field ( of radius @xmath246 ) will contain @xmath247 active quasars as bright as the observed sdss quasars within one band - pass of frequency ( @xmath248mhz ) .
an integration time of @xmath183 hours should be sufficient to detect the rings of redshifted 21 cm emission from the warm igm outside the quasar regions if @xmath249 .
in the presence of a strong cosmic ly@xmath0 background , these rings should remain visible after the quasar has turned off .
this would result in the detection of a large number of fossil regions that exceeds the number of active quasars by the inverse of the quasar duty cycle ( @xmath250 ) .
arc - minute observations of quasar strmgren spheres will provide an alternative method for measuring the spectral step that makes the so - called global signature of reionization ( shaver et al .
the brightness temperature measured in the spectral range corresponding to the ionized gas within the region provides a calibration of the foreground emission at the same redshift .
this calibration will help overcome the difficulties associated with measuring the global signature of an extended reionization epoch ( as suggested by the large electron scattering optical depth measured by _ wmap _ ; kogut et al .
2003 ) on top of a frequency dependent foreground .
detection of the signal for a statistical sample of quasars would help calibrate the expected scatter in the reionization redshift of different regions in the universe ( barkana & loeb 2003b ) . finally , we emphasize that the redshifted 21 cm observation of existing sdss quasars will determine whether the igm was significantly neutral at @xmath241 as suggested by their observed ly@xmath0 spectra ( wyithe & loeb 2004 ) . when combined with the high value of the optical depth for electron scattering measured by
_ wmap _ ( kogut et al . 2003 ) , a significant neutral fraction at @xmath251 would imply a complex reionization history and possibly an early reionization phase driven by population - iii stars .
|
the ly@xmath0 absorption spectrum of the highest redshift quasars indicates that they are surrounded by giant regions , a few mpc in size .
the neutral gas around these regions should emit 21 cm radiation in excess of the cosmic microwave background , and enable future radio telescopes to measure the transverse extent of these regions . at early times , the regions expand with a relativistic speed .
consequently , their measured sizes along the line - of - sight ( via ly@xmath0 absorption ) and transverse to it ( via 21 cm emission ) should have different observed values due to relativistic time - delay .
we show that the combined measurement of these sizes would directly constrain the neutral fraction of the surrounding intergalactic medium ( igm ) as well as the quasar lifetime .
based on current number counts of luminous quasars at @xmath1 , an instrument like _ lofar _ should detect @xmath2 redshifted 21 cm shells per field ( with a radius of @xmath3 ) around active quasars as bright as those already discovered by sdss , and @xmath4 relic shells of inactive quasars per field .
we show that ly@xmath0 photons from the quasar are unable to heat the igm or to couple the spin and kinetic temperatures of atomic hydrogen beyond the edge of the region .
the detection of the igm in 21 cm emission around high redshift quasars would therefore gauge the presence of a cosmic ly@xmath0 background during the reionization epoch .
|
quantum gases , i.e. systems in which the thermal de broglie wavelength is larger than the average particle distance are an ever increasingly important area of theoretical and experimental study with applications as diverse as quantum computing and quantum simulation of general relativity models .
the most important example of such physics is the bose - einstein condensate where a macroscopic number of bosonic particles at low temperature share the same wavefunction .
after early work on liquid helium , the physics of condensates has been experimentally studied in ultracold atomic gases and , more recently , in exciton - polariton fluids in semiconductor microcavities .
building on these latter studies , an ever growing community is now active on the study of the so - called quantum fluids of light , where the many - photons forming the beam are seen as a gas of interacting particles via the optical nonlinearity of the medium @xcite . among the many hydrodynamic effects that are presently being studied in such fluids of light , we can mention turbulence @xcite , where the general physical processes found in a fluid or superfluid can be found in many other systems such as plasmas or in astrophysical systems .
another intriguing direction is so called `` analogue gravity '' where flowing fluids are used to model gravitational spacetime geometries @xcite .
waves propagating on top of a flowing medium behave the same way as light waves in a gravitational field .
hence , by tailoring the properties and flow geometry of such a medium , it is possible to mimic gravitational spacetimes in a laboratory experiment . as an alternative to actual flowing fluids
, optical analogues obtained using laser beams propagating through nonlinear self - defocusing media or confined in nonlinear optical cavities have been proposed to create a photon fluid with the desired flow pattern . in these systems
the fluid properties can be controlled through the phase and intensity of the incident optical field as well as by the refractive index profile and the structural boundaries of the medium @xcite .
for example , a theoretical framework for photon fluids @xcite has shown that sound - like excitations experience a curved spacetime and hence are promising models for analogue gravity experiments . a full
quantum simulation showing evidence for analog hawking radiation from a black hole horizon in a fluid of light was performed in @xcite and the first experimental reports of black hole horizons in fluids of light were reported in @xcite .
the effective photon - photon interaction in the nonlinear medium arises from the third - order kerr nonlinearity @xcite , i.e. the local change in refractive index proportional to the light intensity @xmath0 : a defocusing nonlinearity with @xmath1 giving repulsive interactions is required to observe a dynamically stable photon fluid .
such a nonlinearity may be obtained through the thermo - optic effect , i.e. an intense laser beam locally heats the medium , which reacts by decreasing the refractive index proportionally to the deposited laser energy thus giving rise to an effective repulsive photon - photon interaction @xcite .
here , we study and propose a relatively well known nonlinear medium , i.e. the thermal defocusing medium described above however , revisited within the context of a fluid of light .
the aim is to study the extent to which such systems can indeed represent a novel platform for studying , at room temperature , the properties of quantum gases with an emphasis on ( but not limited to ) the exciting perspective of reproducing superfluid behaviour , phonon propagation according to the bogoliubov dispersion , and particle generation from a black hole horizon and in an analogue gravity set - up . in the propagating geometry considered here ,
the dynamics of the fluid takes place in the transverse plane ( x , y ) of a laser beam propagating through a bulk nonlinear medium , so that the propagation coordinate z plays the role of the time coordinate @xmath2 @xcite .
hence , by imaging the evolution of the beam profile along the @xmath3-coordinate , it is possible to detect the spatio - temporal evolution of effective sound modes in the photon fluid ( i.e. small oscillations on the transverse beam profile ) and extract information on the dispersion relation of its elementary excitations . using a technique inspired from oceanography , we obtain information on the high wavevector part of the dispersion where excitations have a single - particle nature and we detect the effect of global flow of the fluid . then , a pump - and - probe technique is presented , where the probe beam is used to create elementary excitations in the fluid generated by the pump .
information on the elementary excitation dispersion is obtained from the lateral shift of the interference pattern of pump and probe beams : signatures of the collective nature of the excitations and of the non - locality of the photon - photon interactions are identified and discussed .
an important result of this study is the experimental verification of the existence of a ( phonon ) wavelength regime in which , using the terminology introduced by chiao et al .
, photon superfluidity can be observed @xcite .
the propagation of a monochromatic laser beam with a vacuum optical wavelength @xmath4 in a self - defocusing medium can be described at steady state by the non - linear schrdinger equation ( nlse ) : @xmath5 where @xmath3 is the propagation direction , @xmath6 the optical wave vector and @xmath7 is the linear refractive index of the medium .
the laplacian term @xmath8 with respect to the transverse coordinates @xmath9 describes the linear diffraction of the laser beam , where the second , nonlinear , term proportional to @xmath10 refers to the optical kerr nonlinearity , i.e. an intensity - dependent refractive index . in this work
, we will always take @xmath11 , that is a self - defocusing nonlinearity , which guarantees transverse stability of the beams . by writing the field @xmath12 in terms of its amplitude and phase
@xmath13 , equation ( [ eq : nlse ] ) can be re - written as two differential equations @xcite @xmath14 formally identical to the density and phase equations of a two - dimensional bec , the temporal variable @xmath15 being proportional to the propagation distance .
the optical background intensity @xmath16 is identified as the photon fluid density @xmath17 and the phase @xmath18 defines a fluid velocity @xmath19 .
the small amplitude perturbations @xmath20
( @xmath21 is the background field ) can be described within the bogoliubov theory in terms of sound waves on top of the photon fluid @xcite . in the spatially homogeneous case where @xmath21 does not depend on the transverse coordinate @xmath22 , a sound mode of wavevector @xmath23 and amplitude
@xmath24 has a plane - wave form @xmath25 and its angular frequency @xmath26 ( in the temporal @xmath27 variable ) satisfies the dispersion relation @xmath28 here , @xmath29 is the speed of light and @xmath30 the background flow velocity . using the terminology of hydrodynamics , we note that for low frequencies the dispersion of the sound modes has a sonic character @xmath31 , whereas for higher frequencies the second term dominates and the sound modes follow a single - particle quadratic dispersion @xmath32 .
the separation between these two regimes defines a characteristic length @xmath33 , usually called the _ healing length _ in the bec literature @xcite . as a result
, only long - wavelength sound modes with @xmath34 follow a sonic dispersion with a constant sound speed @xmath35 .
this is of particular relevance , since only a linear dispersion guarantees superfluid behaviour @xcite and , in the context of analogue gravity , lorentz invariance in the acoustic metric @xcite . + translating these concepts in to the language of optics @xcite , @xmath36 is the magnitude of the transverse wave vector of the light field and @xmath37 is related to the change @xmath38 in the wave vector along the propagation axis . to experimentally access the low frequency modes
i.e. the @xmath34 regime , the light has to propagate in the nonlinear medium for at least one oscillation period @xmath39 , so with @xmath40 , one arrives at a minimum propagation length given by @xmath41 .
we therefore see that for a reasonable nonlinearity of @xmath42 , long samples with at least @xmath43 cm length are required to observe the propagation of modes that lie in the linear regime of the dispersion .
a natural choice for the nonlinear material is therefore a liquid that exhibits a thermo - optic nonlinearity .
the main advantage is that samples of the required dimensions can easily be built and that they are isotropic . in thermo - optical media
, self - defocusing of the laser beam arises due to partial absorption of the beam , which results in heating effects and thus to a decrease in the refractive index @xcite .
thermal nonlinearity is usually highly nonlocal , i.e. the change in refractive index at any position depends not only on the local intensity , but also on surrounding field intensities @xcite .
this is a result of heat conduction inside the nonlinear medium and has a striking influence on the wave dynamics in nonlinear optics @xcite .
the medium response may be described by a response function @xmath44 so that @xmath45 can be written as @xmath46 with @xmath47 being a normalisation factor and @xmath48 the field intensity . in general
, @xmath49 depends on the nonlocal process in the material and can be of different shapes .
it is also important to note that due to the long - range nature of thermal diffusion , @xmath49 will also depend on the boundary conditions , e.g. on the shape and materials used for the cell containing the liquid @xcite . for thermo - optical media
an isotropic exponential decay was proposed for @xmath50 , where the width of the response function @xmath51 defines the degree of nonlocality @xcite .
+ straightforward extension of the bogoliubov theory to non - local media leads to a modified bogoliubov dispersion of the form @xcite : @xmath52 where @xmath53 is the three - dimensional fourier transform of the response function @xmath54 .
it is easy to see that well within the paraxial approximation @xmath55 , the rescaled bogoliubov frequency @xmath56 , so the main contribution to the non - locality comes from the @xmath57 dependence of @xmath53 , and we can simplify @xmath58 in eq .
( [ eq : bogodr_nl ] ) . + examples of the bogoliubov dispersion in non - local media with different parameters
are shown in fig .
[ fig : nldr ] . for very low wavevectors
@xmath59 the nonlinearity is responsible for the linear , sonic shape of the dispersion @xmath60 . at very large wavevectors @xmath61 , the dispersion recovers the quadratic single - particle dispersion @xmath32 .
the effect of the non - locality is most interesting for @xmath62 : in this case , there is a wide range of @xmath63 where the nonlocality suppresses the effectiveness of the nonlinearity and the dispersion is pushed towards the single - particle dispersion already at @xmath64 . for a sufficiently small @xmath51 and a sufficiently fast decrease of @xmath49 ( faster than the one we are considering here , e.g. gaussian ) , a ring of roton - like minima appears in the dispersion for finite @xmath57 @xcite . as a sonic shape of the dispersion of excitations is essential for gravitational or superfluid analogies to be rigorous , experimental studies in this direction require having control over the nonlinear and nonlocal properties of the material .
+ in the following , we report experimental evidence of a photon fluid in a nonlocal , self - defocusing medium . we adapted a method from oceanography to measure the spatiotemporal dispersion relation and show that by controlling the phase and intensity of the laser beam , we can imprint a flow as well as tune the velocity of acoustic waves .
we then demonstrate a second method , based on tracking the sound speed of waves in the photon fluid as a function of input power , that allows us to measure both the local nonlinear refractive index change , @xmath45 and the nonlocal length of the medium , @xmath51 .
+ and [ eq : bogodr_nl ] ) . local nonlinear media ( @xmath42 ) show a linear part for wavelengths @xmath34 and @xmath65 ( black solid line , @xmath66 @xmath67 m , equivalent to k@xmath68 1/m ) . in nonlocal media
the effectiveness of the nonlinearity is suppressed for wavelengths @xmath69 that results in a parabolic dispersion relation ( the dotted blue lines correspond to a nonlocal nonlinear dispersion with @xmath70 , @xmath71 and @xmath72 @xmath67m).,scaledwidth=48.0% ]
the experiments were carried out by launching a collimated cw - laser beam with @xmath73 nm wavelength through a cylindrical tube with length @xmath74 cm ( 2 cm diameter ) filled with a methanol / graphene solution .
the beam is magnified to a waist diameter of @xmath75 cm by a 4-f telescope and the central portion with a radius of 1 cm is selected with an aperture to ensure a relatively flat intensity profile .
methanol is known to have a negative thermo - optic coefficient of @xmath76 1/k but has a very low absorption coefficient of @xmath77 @xmath78 @xcite .
nanometric graphene flakes ( 7 nm average size ) are therefore dissolved in the medium in order to increase absorption . differently from widely used dyes
, graphene does not exhibit any efficient fluorescence mechanism so that most of the absorbed laser energy is converted directly into heat . by using a very dilute solution of graphene , it was possible to prepare samples with a total absorption that could be controllably chosen in the full 0 to @xmath79% region over the sample length and adjusted in our experiments so as to have 20% absorption .
a camera / lens system is mounted on a computer - controlled , longitudinal translation stage , so that the transverse beam profile can be imaged at different positions along the propagation axis of the beam .
the imaging was performed at a constant image distance to ensure uniform scaling in transverse and z - directions during scanning . since the z - axis maps into a time coordinate , it is possible to measure the spatiotemporal evolution of small intensity fluctuations on top of the beam by scanning along z and measuring the 2d beam profile at defined increments of @xmath80 .
the input beam diameter was large enough so that linear diffraction is negligible along the sample . to directly measure the dispersion relation , a technique adapted from oceanographic studies
was used where surface waves that naturally occur were recorded in both space and time and by calculating the fourier transform one obtains the dispersion of the medium @xcite .
+ in our experiment , the set of images are stored in a 3d ( x , y , z ) dataset : the origin of the observed fluctuations is due to ( static ) classical amplitude noise that is always present on the transverse profile of any laser beam ( unless one invests significant effort to completely spatially filter it out ) . the dispersion @xmath81 can be calculated via the fourier transform of a lineout of the measured intensity signal @xmath82 .
the resulting spectrum can be converted to frequency using @xmath83 . to enhance the signal - to - noise ratio ,
the data was averaged over 200 line - outs along the y direction .
examples of the measured dispersion of the photon fluid are shown in figs .
[ fig : dispersionrelations](a ) , ( b ) and ( c ) .
the spectra show some `` folding '' determined by the lattice of points at which the signal is sampled and then fourier - transformed
. the large signal at ( @xmath84 ) corresponds to the background pump beam .
some of the spectra also show vertical stripes or other features that do not lie on the dispersion curves and which we were able to attribute to small defects present e.g. on the input or output window facets .
the main finding of these measurements is the observation of an apparently purely parabolic dispersion for all spatial k - vectors and for all sets of parameters .
three different measurements are shown in fig .
[ fig : dispersionrelations ] for three different angles of the input beam with respect to the sample ( and imaging ) axis in the @xmath85 plane .
this tilt angle creates a linear phase gradient along the tilt direction and thus , according to the hydrodynamical equations and controls the transverse flow of the photon fluid .
this can clearly be seen in figs .
[ fig : dispersionrelations](b ) and ( c ) for which the increasing input angles lead to increasing flow .
the data is consistent with eq .
( [ eq : bogodr ] ) with @xmath86 and flow speeds @xmath87 0 , 1.3 , 3.0 x @xmath88 m / s [ ( a ) , ( b ) and ( c ) , respectively ] .
we attribute the purely parabolic dispersion to the nonlocal nature of the material that , as discussed above reduces the effectiveness of the nonlinearity for features smaller than the @xmath51 .
a nonlocal length for methanol of @xmath89 @xmath67 m has been reported @xcite , implying that only modes with a much smaller @xmath90 m@xmath91 would appear to have a linear dispersion ( fig .
[ fig : nldr ] ) .
+ we underline that in these measurements we are imaging the various photon fluid planes inside the actual nonlinear medium .
this may cause some issues related to the fact that we can no longer consider this a linear imaging measurement and the nonlinearity may lead to deformations of the various object planes @xcite .
we did however , experimentally verify that objects placed at the cell input could be imaged through the sample and that distortions only appeared to be significant in the presence of sharp transitions from very low to very high light intensity regions .
we therefore expect that the images of the weak and smooth oscillations lying on top of the more intense laser beam background are not significantly distorted and that the overall shape of the measured dispersion curve is indeed correct .
this is corroborated by the fact that the dispersion relation eq .
( [ eq : bogodr ] ) provides an excellent theoretical fit to the data for the zero - velocity flow case and provides a correct estimate of the photon fluid flow when the input beam is placed at an angle with respect to the imaging axis .
+ however , it is clear that a more detailed and precise estimate of the low frequency behaviour and the presence or absence of superfluid behaviour should be assessed by a technique that does not require imaging through the nonlinear medium .
the results obtained with such a technique are discussed in the next section .
in this experiment , a pump - probe configuration was used to imprint a specific wavelength on top of the background beam as proposed in @xcite .
we then study the wave dynamics for different background intensities .
the probe beam was extracted from the pump beam using a beamsplitter and was attenuated with a half - wave plate and a polarising beam splitter .
the probe beam is then recombined with the pump beam inside the sample : by controlling the relative angle of the two beams we create an interference pattern of the desired modulation depth ( @xmath92 ) and relative wavevector @xmath57 .
both the pump and probe beams are loosely focused into the sample using a cylindrical lens pair to create an elliptical beam with a minor axis diameter of @xmath93 @xmath67 m and a rayleigh range longer than the sample ( @xmath94 cm ) . the interference wave pattern ( i.e. the photon fluid sound wave ) is finally imaged at the output facet of the sample and a shift of this pattern along the x direction can be measured as a function of the laser power [ fig .
[ fig : shift](a ) ] .
the final imaging stage is performed with a 1-to-1 imaging f=300 mm lens . in the focal plane
we place a mask to eliminate the four - wave mixed wave , thus suppressing interference between counter - propagating bogoliubov modes and facilitating the measurement of the phase shift . so as to isolate the pump and probe beams .
our pump - probe setup is similar to that used in the past to study periodic soliton formation @xcite albeit with a lower contrast in the pump - probe interference pattern and , as we discuss below , our attention is focused on the dynamics of the photon - fluid sound waves .
+ as the beam propagates through the nonlinear medium , the phase velocity of the sound wave is determined by @xmath95 with @xmath26 given by eq .
( [ eq : bogodr ] ) and hence is a function of the nonlinearity @xmath45 . as the pump beam power
is increased we therefore expect to observe a spatial shift of the sound wave pattern by @xmath96 due to the increase of the phase velocity .
so by using eq .
( [ eq : bogodr ] ) , we obtain an expression for @xmath97 : @xmath98\,z } \label{eq : shift2 } \end{aligned}\ ] ] by measuring @xmath96 for different values of the pump power we may therefore use eq .
( [ eq : shift2 ] ) to estimate both @xmath45 and @xmath51 . as a function of the sound wave wavevector @xmath57
, one can recognize the finite limit @xmath99 determined by the speed of sound in the fluid of light .
the local vs. non - local nature of the nonlinearity is visible for short wavelengths . in the local case
, the phase shift tends to @xmath100 linearly in @xmath101 with a slope proportional to the refractive index change @xmath102 . in the non - local case ,
an extra decay is introduced by the @xmath103 factor : for the case in which which @xmath104 , for low @xmath57 this tends to zero as @xmath105 . as a result , the phase shift tends to @xmath100 much more quickly .
while a non - zero @xmath96 is by itself a consequence of interactions , the nonlocal effect on the bogoliubov dispersion is visible in the deviation of @xmath96 from a linear dependence on @xmath101 .
+ vs. @xmath106 : the solid green line shows the predicted shift for a local nonlinearity of @xmath107 - the corresponding healing length is @xmath108 @xmath67 m .
the solid blue line includes a nonlocal nonlinearity with @xmath107 and nonlocal length @xmath109 @xmath67 m .
the shaded gray area ( @xmath110 ) highlights the region in which , even in the presence of nonlocality , the medium acts as a superfluid .
the black squares indicate the measured shift values at various wavelengths , several of which ( i.e. for @xmath111 mm ) lie in the superfluid region.,title="fig:",scaledwidth=48.0% ] vs. @xmath106 : the solid green line shows the predicted shift for a local nonlinearity of @xmath107 - the corresponding healing length is @xmath108 @xmath67 m .
the solid blue line includes a nonlocal nonlinearity with @xmath107 and nonlocal length @xmath109 @xmath67 m .
the shaded gray area ( @xmath110 ) highlights the region in which , even in the presence of nonlocality , the medium acts as a superfluid .
the black squares indicate the measured shift values at various wavelengths , several of which ( i.e. for @xmath111 mm ) lie in the superfluid region.,title="fig:",scaledwidth=48.0% ] the experimental data of the relative shifts of several transverse wavelengths @xmath112 between @xmath113 @xmath67 m and @xmath114 mm at a fixed laser power of 28 mw ( fluence , 1 w/@xmath115 ) is shown in fig .
[ fig : shift ] ( solid black squares ) . for comparison ,
the predicted shift in a medium with a purely local ( @xmath116 ) nonlinearity ( @xmath107 ) according to eq .
( [ eq : shift2 ] ) is also plotted ( solid green line ) .
in general , we found that it is not possible to obtain a satisfactory agreement with the experimental trend for any value of @xmath45 and a purely local nonlinearity .
+ on the other hand , if we take into account the nonlocal factor @xmath117 in eq .
( [ eq : shift2 ] ) , we obtain a very good agreement with the data with @xmath107 and @xmath109 @xmath67 m ( solid blue line ) .
+ a remarkable feature of these measurements is that for the longer wavelengths , i.e. for @xmath111 mm the observed shift appears to saturate and , within an 8% margin , is the same for all wavelengths .
in other words , the phase velocity of the waves , @xmath118 m / s , does not depend on wavelength in this region ( shaded in gray in the figure ) and the dispersion relation is dominated by the linear behaviour of the bogoliubov term .
this is a clear evidence of the linear dispersion of collective excitations typical of a superfluid or , in the language of general relativity , of lorentz invariant propagation that is crucial for future analogue gravity experiments .
naturally , the the observation of superfluid features in a system at room - temperature is in itself of great interest and future experiments will be directed at deepening our understanding and control over the flow and wave propagation in this nonlocal photon superfluid .
+ a crucial point here is that a standard z - scan measurement would not be adequate for characterising such a nonlocal nonlinearity .
z - scan relies on translating the sample through the focus of a lens , i.e. through a region in which the transverse length scale of the beam is changing .
the nonlinearity analysed here shows a clear dependence on the transverse beam dimensions through the @xmath53 response function . in a z - scan measurement
the material would therefore respond with different effective nonlinearities at different measurement positions , resulting in a distorted z - scan trace .
we explicitly verified that this is indeed the case ( data not shown ) .
we have reported a nonlinear optical experiment that aims at characterizing the physics of a photon fluid realized in a cavity - less nonlinear medium with a nonlocal optical nonlinearity of thermal origin . with a technique used in hydrodynamics
, we measured the parabolic dispersion relation for small amplitude , small wavelength noise excitations and highlighted the evidence of a hydrodynamic flow in the presence of a linear phase term on the transverse beam profile . with a pump - probe technique we investigated the larger wavelength region by measuring the change in the phase velocity of collective excitations as a function of the pump intensity . in this way , we estimated the thermal nonlinearity to @xmath119 and the nonlocal length @xmath120 @xmath67 m .
this measurement technique also highlighted a superfluid behaviour of the photon fluid , which therefore holds promise both for future room - temperature studies of superfluidity and for analogue gravity measurements that require a lorentz invariant ( i.e. dispersionless ) flowing medium .
our measurements indicate that there is a readily attainable wavelength region in which such measurements may be feasible in the near future .
however , the sound speed in the superfluid regime demonstrated is relatively low @xmath121 m / s and will require very long samples in order to observe significant wave propagation .
future work will therefore be directed at increasing the sound speed , e.g. by use of materials with higher nonlinearities . on the other hand , the degree of nonlocality
can also be tuned by modifying the sample geometry @xcite or absorption @xcite and this opens opportunities for performing related experiments over a broader wavelength range . a crucial and natural demonstration of superfluidity
will involve the study of the transition from laminal flow arond an obstacle at subsonic flow speeds to turbulent dissipative flow with the formation of quantised vortices at supersonic flow speeds @xcite .
furthermore , such media show promise for studies of hydrodynamical turbulence in the presence of nonlocal interactions and the study of collective dynamics of incoherent waves . finally ,
extension of our experimental apparatus beyond the cw measurements shown here so as to measure also time - dependent fluctuations could provide access to quantum fluctuation features @xcite such as analogue hawking radiation @xcite .
df acknowledges financial support from the engineering and physical sciences research council epsrc , grant no .
ep / j00443x/1 and from the european research council under the european union s seventh framework programme ( fp/2007 - 2013)/erc grant agreement n.306559 .
ic acknowledges financial support by the erc through the qgbe grant and by the autonomous province of trento , partly through the project `` on silicon chip quantum optics for quantum computing and secure communications '' ( `` siquro '' ) .
+ a. picozzi , j. garnier , t. hanssom , p. suret , s. randoux , g. millot , d.n .
christodoulides , `` optical wave turbulence : towards a unified nonequilibrium thermodynamic formulation of statistical nonlinear optics , '' phys .
rep . * 542 * , 1 - 132 ( 2014 ) .
r. chiao and j. boyce , `` bogoliubov dispersion relation and the possibility of superfluidity for weakly interacting photons in a two - dimensional photon fluid , '' physical review a * 60 * , 41144121 ( 1999 ) .
a. minovich , d. n. neshev , a. dreischuh , w. krolikowski , and y. s. kivshar , `` experimental reconstruction of nonlocal response of thermal nonlinear optical media .
'' optics letters * 32 * , 1599601 ( 2007 ) .
|
quantum gases of atoms and exciton - polaritons are nowadays a well established theoretical and experimental tool for fundamental studies of quantum many - body physics and suggest promising applications to quantum computing . given their technological complexity
, it is of paramount interest to devise other systems where such quantum many - body physics can be investigated at a lesser technological expense .
here we examine a relatively well - known system of laser light propagating through thermo - optical defocusing media : based on a hydrodynamical description of light as a quantum fluid of interacting photons , we investigate such systems as a valid , room temperature alternative to atomic or exciton - polariton condensates for studies of many - body physics .
first , we show that by using a technique traditionally used in oceanography , it is possible to perform a direct measurement of the single - particle part of the dispersion relation of the elementary excitations on top of the photon fluid and to detect its global flow .
then , using a pump - and - probe set - up , we investigate the collective nature of low - wavevector sound modes of the fluid and observe signatures of superfluid behaviour .
|
when studying the statistical mechanics of random interfaces which typically arise between coexisting phases of a ( @xmath6-dimensional ) physical system in equilibrium , one often considers so - called _ effective _ models , which aim at describing the @xmath3-dimensional surface _ itself _ , free from its surroundings .
arguably the most notorious example in this class is the massless harmonic crystal , or gaussian free field ( precise definitions will follow , see below ) .
a natural approach in trying to gain some insight into the geometry of this field is to inquire about its level sets , say , above a given height @xmath7 . in case
the underlying space is the cubic lattice @xmath0 , with @xmath1 , and due to the presence of strong correlations ( the susceptibility is infinite ) , this gives rise to an interesting percolation model , which was originally introduced by lebowitz and saleur in @xcite , and has since then been investigated in @xcite , @xcite , @xcite , @xcite , @xcite and @xcite , see also @xcite , @xcite for related results .
only recently has it been shown in @xcite that the associated phase transition is non - trivial in _ all _ dimensions @xmath1 ( partial results were already obtained in @xcite and @xcite ) .
our main focus in the present work is to examine the limiting behavior of certain critical parameters associated to this transition in high dimension . of fundamental importance in this context
is the heuristic principle by which this percolation model ought to fall in the `` domain of attraction '' of a corresponding model on the @xmath8-regular tree .
our results indicate that this is indeed the case , and in fact , this paradigm permeates more or less explicitly many of the proofs below .
ideally , we would also like to compare these results with corresponding ones in the tree - case _
directly_. we hope to return to this point elsewhere .
a first step in the direction of high - dimensional asymptotics is given by theorem 3.3 of @xcite , which asserts that the critical height @xmath4 for percolation of gaussian free field level sets , defined in below , is strictly positive when @xmath3 is sufficiently large .
in fact , a careful inspection of the proof of this theorem yields that @xmath9 as @xmath10 , which amounts to saying that the critical density for this percolation model converges to @xmath11 as @xmath3 tends to infinity . we will considerably refine this result by providing the leading asymptotic behavior of the critical density , as well as principal asymptotics for @xmath4 and a related critical parameter @xmath12 ( see below ) as @xmath3 becomes large .
our proof follows in its broad lines the general strategy underlying similar results in bernoulli ( see @xcite and @xcite , section 4 ) and interlacement ( see @xcite and @xcite ) percolation .
however , the implementation of this program highly depends on the specific nature of the model , as it crucially relies on a precise understanding of its dependence structure in high dimension . in particular , in the present context , the random walk representation of the gaussian free field , which will lead us to its `` perturbative '' expansion around a suitable _ independent _ field , will play a pivotal role in allowing for a precise understanding of the _ local _ connectivity of the level set ( we will explain this in greater detail below , see the discussion around ) .
moreover , a severe technical obstruction is the absence of a bk - type inequality ( companion to the long - range dependence ) . in the independent case
, this inequality underlies the successful deployment of such elaborate tools as the lace expansion , see @xcite and @xcite , chapter 9 . in our set - up , as a partial substitute , we develop suitable decoupling inequalities , much in the spirit of @xcite , @xcite , with the notable difference that they will need to work `` _ _ uniformly _ _ well '' for all sufficiently large @xmath3 .
these inequalities will typically allow for a certain kind of ( static ) renormalization procedure , to which the dimension @xmath3 will be inextricably tied .
we now describe our results and refer to section [ notation ] for details .
we consider the lattice @xmath0 , @xmath1 , endowed with the usual nearest - neighbor graph structure , and investigate the gaussian free field on @xmath0 , with canonical law @xmath13 on @xmath14 ( equipped with the product @xmath15-algebra ) such that , @xmath16 = g ( x , y)$ } , \text { for all } x , y \in \mathbb{z}^d , \end{split}\ ] ] where @xmath17 denotes the green function of simple random walk on @xmath0 , see .
note that , for fixed @xmath3 , this model exhibits rather strong correlations , ( in particular , the susceptibility is infinite for all @xmath18 , cf .
below ) . however , as @xmath3 grows , @xmath19 `` approaches '' an independent field , in a sense to be made precise below . for any _ level _ @xmath7 , we introduce the random subset of @xmath0 @xmath20 often referred to as the _ excursion set _ ( or _ level set _ ) of the field @xmath19 above height @xmath21 . in order to study its percolative properties , we set @xmath22 $ ] , the probability that the origin lies in an infinite cluster ( i.e. connected component ) of @xmath23 . the function
@xmath24 being decreasing , we define the critical parameter for level - set percolation as @xmath25 ( with the convention @xmath26 ) . following @xmath27 of @xcite , we also introduce a second critical point @xmath28 = 0 \big\},\ ] ] where the event @xmath29 refers to the existence of a nearest - neighbor path in @xmath30 connecting @xmath31 , the ball of radius @xmath32 around @xmath11 in the @xmath33-norm , to @xmath34 , the @xmath33-sphere of radius @xmath35 around @xmath11 ( in fact does not exactly coincide with ( 0.6 ) in @xcite , which requires the relevant probability to decay at least polynomially in @xmath32 ; the two are in fact equivalent , and can be further weakened , see @xcite , theorem 2.1 ) . the definitions and
immediately yield that @xmath36 for all @xmath1 .
it is presently known that @xmath37 which implies that percolation of @xmath23 exhibits a non - trivial phase transition ( see @xcite , corollary 2 for the former and @xcite , theorem 2.6 for the latter result in ; see also theorem 3 in @xcite for a proof of @xmath38 ) .
in particular , for all @xmath39 , @xmath23 only contains finite clusters , and a ( unique , see @xcite , remark 1.6 ) infinite cluster for all @xmath40 .
the parameter @xmath41 is an important quantity because it characterizes a _ strongly _ subcritical regime . for all @xmath42 and @xmath1
, the probability @xmath43 $ ] decays exponentially in @xmath32 as @xmath44 ( with logarithmic corrections when @xmath45 ) , as follows from theorem 2.1 . in @xcite .
moreover , as mentioned in the expository paragraph ( see @xcite , theorem 3.3 ) , @xmath46 our main goal in this paper is to prove the following asymptotic result concerning the critical density of this percolation model .
[ t : asymptotics_density ] @xmath47 = \frac{1}{d^{1+o(1 ) } } , \textnormal { as $ d \to \infty$}.\ ] ] in fact , theorem [ t : asymptotics_density ] will be an easy consequence of the following two results regarding the principal asymptotics of the critical parameters @xmath48 and @xmath41 in high dimension .
for future reference , we let @xmath49 where @xmath50 refers to the green function at the origin , cf . below .
we will show the following .
[ t : asymptotics_ub ] ( upper bound ) @xmath51 [ t : asymptotics_lb ] ( lower bound ) @xmath52 moreover , for all @xmath53 , there exists a finite positive constant @xmath54 such that for all @xmath55 , @xmath56=1,\ ] ] where @xmath57 and @xmath58 is viewed as the subset @xmath59 of @xmath0 .
: the presence of the factor @xmath60 in the definition of @xmath61 is for esthetic purposes only , since @xmath62 as @xmath10 , cf .
below ) . before describing our methods ,
we make a few remarks concerning these results and the heuristics lurking behind them .
as alluded to above , a recurrent theme in the proofs below will be that , in high dimensions , the free field at _ small _ scales ( at which the geometry felt by the random walk is roughly tree - like ) can be viewed as a _ perturbation _ of an independent gaussian field , and a considerable effort will go into understanding the precise effect of this perturbation on the connectivity of the level sets around the asymptotic value @xmath63 ( see the comments below , and the proofs of theorems [ t : local_cb ] and [ p : subst_comp ] ) .
this perturbative behavior was already hinted at in the proof of in @xcite , which was based on a decomposition of the covariance @xmath64 for @xmath65 belonging to a ( lower - dimensional ) subspace @xmath66 , with @xmath67 , into the sum of a dominant _ diagonal _ part with entries close to @xmath68 ( thus inducing a field of independent gaussians ) and a `` small noise '' ( cf .
lemma 3.1 . of @xcite for a precise statement ) . admitting this ( local ) resemblance of @xmath19 to an independent field
, it is reasonable to compare to @xmath69 , the critical parameter for bernoulli site percolation on @xmath0 , which is known to be asymptotically equal to @xmath70 as @xmath71 , see @xcite , @xcite , @xcite , @xcite . regarding theorems [ t : asymptotics_ub ] and [ t : asymptotics_lb ] , using @xmath72 , and
imply that @xmath73 ( we write @xmath74 as @xmath75 if @xmath76 ) .
it is at present an important unresolved question whether both critical parameters are actually equal ( in any dimension ) .
we now comment on the proofs .
the proof of the upper bound follows a strategy inspired by that used in @xcite to prove finiteness of @xmath12 , for arbitrary , but _ fixed _
dimension @xmath3 ( see also @xcite , @xcite , @xcite , @xcite , @xcite for similar ideas in the context of random interlacements ) . in particular , we use a variant of the renormalization scheme developed therein .
however , the present task requires a scheme which works `` uniformly in @xmath3 '' as @xmath3 becomes large . using careful estimates on the behavior of the green function of simple random walk on a high - dimensional lattice developed in @xcite , we obtain a `` decoupling inequality , '' which enables us to propagate bounds on the relevant crossing events in @xmath77 ( cf .
the definition of @xmath41 ) , where @xmath78 for some @xmath79 , at small scale ( the so - called seed estimates ) to controls of such crossing probabilities at arbitrarily large scale in @xmath80 ( this is in fact much more than we need ) . a significant part of the problem is to produce sufficiently sharp seed estimates , in order to initiate the renormalization , see remark [ r : renorm ] , 2 ) below ( note that , in contrast to the proof of the finiteness of @xmath12 in @xcite , which allowed one to look for a corresponding regime at arbitrary large @xmath21 , we are now constrained to remain in the vicinity of @xmath63 ) . obtaining the desired bounds at small scales
involves controlling the probability to see a crossing in @xmath77 from a given point @xmath81 to the boundary of the @xmath82-ball centered at @xmath81 of radius @xmath83 , see and theorem [ t : local_cb ] below ( we emphasize that the use of the @xmath82-norm is essential here , as this distance controls the short - range behavior of @xmath84 in high dimensions , where the tree - like nature of the lattice manifests itself , cf . ) .
this is the first instance where the aforementioned ( local ) perturbative expansion comes into play . in rough terms , the domain markov property for the free field @xmath19 allows us to `` discover '' it along any given path ( in the present case , one joining @xmath81 to @xmath85 ) `` dynamically , '' starting from a suitable _ independent _ gaussian field @xmath86 , and introducing the required dependence at each step .
specifically , if @xmath87 , @xmath88 , denotes the trace of the path in question , we represent @xmath89 where the error term is a suitable linear combination of @xmath90 , cf .
lemma [ l : conds_exps2 ] below for a precise statement .
this procedure enables us to `` pass '' from @xmath19 to the independent field @xmath86 , provided we carefully keep track of the error we make in each step . for the lower bound , theorem [ t : asymptotics_lb ] ,
our method mimics in its broad lines the approach of @xcite and @xcite , section 4 , to the corresponding problem in bernoulli ( bond and site ) percolation on @xmath0 , involving ideas of @xcite from hypercube percolation ( see also @xcite for a completely different solution , and @xcite for a corresponding result in interlacement percolation , following a similar spirit ) .
the proof essentially comprises two parts , which we briefly detail . in the first part , we show the following finite - size criterion .
suppose we partition the set @xmath91 into translates of the hypercube .
roughly speaking , we show in theorem [ t : local_global ] below that if @xmath92 , with @xmath93 , @xmath53 , * contains a `` giant '' connected component in each of the sets @xmath94 and its four neighboring translates , and * all these components are connected in @xmath92 within the union of @xmath94 and its four neighboring translates , with sufficiently high probability , then @xmath95 percolates ( the `` additive '' sprinkling in @xmath21 is more than enough for the sake of proving ) . due to the long - range dependence , this reduction step does not follow from standard stochastic domination arguments ( see for example @xcite ) , and theorem [ t : local_global ] is established by means of a ( two - dimensional ) renormalization argument . in the second ( and more difficult ) part of the proof of theorem [ t : asymptotics_lb
] , we show that this criterion actually holds . in the independent
setting of @xcite and @xcite , the validity of such a criterion is essentially guaranteed by the analysis in @xcite of ( independent ) percolation in the hypercube .
our procedure essentially comprises two steps .
first , drawing inspiration from @xcite , cf .
in particular lemma 1 therein , we grow _ substantial _ connected components in @xmath96 ( with @xmath93 ) , which have cardinality growing polynomially in @xmath3 . to achieve this ,
we embed into @xmath94 a deterministic @xmath97-regular tree @xmath98 , rooted at @xmath11 , with @xmath99 comparable to @xmath3 for every @xmath53 , of depth depending on @xmath100 only . here again , the `` perturbative representation '' of crucially enters in allowing us to compare @xmath101 to a certain ( supercritical ) galton - watson process on the same tree in order to derive a meaningful lower bound for the probability to see a substantial component at the origin .
the precise statement is the object of theorem [ p : subst_comp ] , and an easy consequence is that most vertices in @xmath94 are in fact either neighboring or contained in such a substantial component , see corollary [ c : number_subst_comps ] .
the second step then consists of gluing all substantial components within the hypercube to a giant one ( i.e. a connected component whose closure in @xmath94 contains at least @xmath102 points ) , and subsequently connecting neighboring giant components , which is achieved in theorem [ t : giant_comp ] by two successive sprinkling operations . quantifying that it is highly unlikely for the substantial components not to merge after sprinkling requires isoperimetric considerations in @xmath94 ( as were used in the proofs of @xcite , theorem 1 and @xcite , theorem 4.2 ) .
however , the ( wealth of ) edges which are hereby deduced to be pivotal for a giant component to emerge or not , are not necessarily `` well spread - out '' within @xmath94 and might therefore influence each other rather strongly .
a slightly careful bookkeeping of this mutual influence is required in order to show that it is too costly for many pivotal sites to remain `` closed '' ( cf .
the proof of lemma [ l : giant_event_h ] ) .
this then completes the proof of the lower bound .
let us now describe the organization of this article . in section [ notation ]
, we introduce some notation and review a few known results concerning simple random walk on a high - dimensional lattice and the gaussian free field .
section [ s : upper_bound ] is devoted to the proof of theorem [ t : asymptotics_ub ] .
subsection [ s : ub_rs ] introduces the renormalization scheme .
its main result is proposition [ p : dec_ineq ] , which entails the induction step ( one - step renormalization ) .
subsection [ s : ub_lcb ] contains the local estimates on the connectivity of @xmath103 , @xmath53 .
the main result , theorem [ t : local_cb ] , is of independent interest , but it is needed crucially to establish the required seed estimate which enables us to trigger the renormalization scheme . finally , in subsection [ s : denouement ] , we put these two ingredients together to complete the proof of .
the proof of the lower bound is the object of section [ s : lower_bound ] .
first , we show the abovementioned finite - size criterion , which requires another ( simpler ) renormalization scheme .
the main result is theorem [ t : local_global ] , which can be found in subsection [ s : finite_size_criterion ] .
having established this reduction step , we prove that most points in the hypercube are either contained or neighboring a substantial component in the level set of interest .
this is done in subsection [ s : subst_comp ] , see in particular theorem [ p : subst_comp ] and corollary [ c : number_subst_comps ] therein .
finally , theorem [ t : giant_comp ] in subsection [ s : connecting_subst_comps ] connects these substantial components , thereby completing the proof of the criterion .
the lower bound then follows readily by collecting the pieces , and subsection [ s : connecting_subst_comps ] also contains the proof of theorem [ t : asymptotics_density ] , which follows straightforwardly from and .
we conclude this introduction with a remark concerning our convention regarding constants : we denote by @xmath104 positive constants with values changing from place to place .
numbered constants @xmath105 and @xmath106 are defined at the place they first occur within the text and remain fixed from then on until the end of the article .
all constants are numerical , and their dependence on any additional parameter , including , most importantly , the dimension @xmath3 , will always appear in the notation .
the only exceptions to this rule are the last two sections [ s : subst_comp ] and [ s : connecting_subst_comps ] , in which constants may implicitly depend on a parameter @xmath53 .
the reader will be reminded of this exception in due time .
in this section , we introduce some notation to be used in the sequel , collect some important estimates related to simple random walk on a high - dimensional lattice , and review a few useful facts concerning the gaussian free field .
we denote by @xmath107 the set of natural numbers , and by @xmath108 the set of integers .
we write @xmath109 for the set of real numbers , abbreviate @xmath110 and @xmath111 for any two numbers @xmath112 , denote by @xmath113 the integer part of @xmath81 , for any @xmath114 , and let @xmath115 .
we consider the lattice @xmath0 , and tacitly assume that @xmath1 . given a subset @xmath116 of @xmath0
, @xmath117 stands for the complement of @xmath116 in @xmath0 , and @xmath118 for the cardinality of @xmath116 .
moreover , in writing @xmath119 , we mean that @xmath116 is a finite subset of @xmath0 . finally , we denote by @xmath120 the minkowski sum of arbitrary sets @xmath121 . on @xmath0
, we respectively denote by @xmath122 , @xmath123 and @xmath124 the @xmath82 , euclidean and @xmath33-norms .
the three norms are equivalent , and satisfy the relations @xmath125 , @xmath126 , and @xmath127 , we let @xmath128 and @xmath129 stand for the the ( closed ) @xmath130-ball and @xmath130-sphere of radius @xmath131 centered at @xmath81 . we also note for later purposes the following bound on the cardinality of a @xmath3-dimensional @xmath82-sphere , @xmath132 , see for example @xcite , lemma 3.2 ( i ) for a proof . given two arbitrary sets @xmath121 , we define their @xmath130-distance as @xmath133 , for @xmath134 , and simply write @xmath135 when @xmath136 is a singleton .
moreover , we denote the interior ( @xmath82-)boundary of @xmath116 as @xmath137 , the outer boundary of @xmath116 as @xmath138 , and write @xmath139 for the ( @xmath82-)closure of @xmath116 and @xmath140 for the relative closure of @xmath116 in @xmath141 .
we now introduce the discrete - time simple random walk on @xmath0 .
we endow the lattice @xmath0 with the usual nearest - neighbor graph structure , and we will frequently use @xmath142 instead of @xmath143 to denote two neighboring vertices @xmath144
. moreover , @xmath144 will be called @xmath145-nearest neighbors if @xmath146 .
a @xmath0-valued nearest - neighbor path is a ( finite or infinite ) sequence @xmath147 of vertices in @xmath148 satisfying @xmath149 for all @xmath150 .
we define its length as the number of edges it traverses .
when its length is infinite , we will often use the term _ trajectory _ instead of _ path_. a @xmath145-path is defined accordingly .
let @xmath151 denote the space of nearest - neighbor trajectories , and let @xmath152 , @xmath153 and @xmath154 , stand for the canonical @xmath15-algebra , the canonical process and the canonical shifts on @xmath151 , respectively .
we write @xmath155 for the canonical law of the walk starting at @xmath156 and @xmath157 for the corresponding expectation .
we denote by @xmath17 the green function of the walk , i.e. @xmath158 , \text { for } x , y \in \mathbb{z}^d,\ ] ] which is finite ( since @xmath159 ) and symmetric .
moreover , @xmath160 due to translation invariance .
we further recall that ( see for example @xcite , theorem 1.5.4 ) @xmath161 given @xmath162 , we denote the entrance time in @xmath116 by @xmath163 and the hitting time of @xmath116 by @xmath164 .
this allows us to define the green function @xmath165 killed outside @xmath116 as @xmath166 , \text { for } x , y \in \mathbb{z}^d,\ ] ] which is symmetric and vanishes if @xmath167 or @xmath168 .
the relation between @xmath169 and @xmath170 for any @xmath162 is given by the following formula , the proof of which is a mere application of the strong markov property ( at time @xmath171 ) , @xmath172 , \text { for } x , y \in \mathbb{z}^d.\ ] ] we now turn to a few aspects of potential theory associated to simple random walk . for any finite subset @xmath116 of @xmath0
, we write @xmath173 , \text { $ x \in k$},\ ] ] for the equilibrium measure ( or escape probability ) of @xmath116 , and @xmath174 for its capacity .
it immediately follows from the definitions and that the capacity is subadditive , i.e. @xmath175 and one also easily infers that it is monotonous , i.e. that @xmath176 the latter follows e.g. by observing that @xmath177 $ ] , which follows from by a straightforward reversibility argument ( see for example @xcite , proposition 2.2.1 ( a ) for more details ) .
moreover , the entrance probability in @xmath116 may be expressed in terms of @xmath178 as @xmath179 = \sum_{y \in k } g(x , y ) \cdot e_{k}(y),\ ] ] which is a mere consequence of the simple markov property ( see for example @xcite , theorem 25.1 , p. 300 ) .
we collect some useful estimates on these quantities in high dimension , which will be used repeatedly in the sequel .
we remind the reader of our convention regarding constants at the end of the previous section .
@xmath180 @xmath181 \leq \frac{c(k)}{d } , \text { for $ k \geq 0$}. \label{return_probas1}\\ & g(x ) \leq \bigg ( \frac{c \sqrt{d } } { |x|_2 } \bigg)^{d-2 } , \text { for } |x|_2 \geq d. \label{g1 } \\ & \text{\textnormal{cap}}(b_2(0,l ) ) \leq \bigg(\frac{cl } { \sqrt{d } } \bigg)^{d-2 } , \text { for } l \geq d. \label{g2 } \end{aligned}\ ] ] for , see @xcite , pp .
246247 ; for and , see @xcite , lemma 1.2 ; for and , see @xcite , lemma 1.1 and @xmath182 , respectively .
we now turn to the gaussian free field on @xmath0 , @xmath18 , as defined in , and introduce certain crossing events involving paths of high level . on the space @xmath183 endowed with its canonical @xmath15-algebra @xmath184 ,
let @xmath185 @xmath186 , for @xmath187 , denote the event that there exists a nearest - neighbor path connecting @xmath116 and @xmath141 along which the configuration has value @xmath68 . letting @xmath188 , @xmath189 , for @xmath7 , we introduce @xmath190 ( part of @xmath191 ) . in words
, this is the event that @xmath116 and @xmath141 are connected by a nearest - neighbor path of vertices in the level set @xmath30 , cf . .
note that this event is increasing upon introducing on @xmath191 the usual partial order ( i.e. @xmath192 if and only if @xmath193 for all @xmath156 ) .
moreover , the probability @xmath194 $ ] is a decreasing function of @xmath7 . in general
, we will use the notation @xmath195 we will also need the notion of _ flipping _ events .
let @xmath196 , @xmath156 , denote the canonical coordinates on @xmath183 .
one defines the inversion map @xmath197 such that @xmath198 , for all @xmath156 .
given some event @xmath199 , let @xmath200 , which will be referred to as _ flipped _ event .
note that @xmath201 is decreasing whenever @xmath202 is increasing .
moreover , observing that @xmath203 has the same law as @xmath204 under @xmath13 , we obtain , for all @xmath205 and @xmath7 , @xmath206 & = \mathbb{p } [ ( 1\ { \varphi_x \geq h \ } ) _ { x \in \mathbb{z}^d } \in a ] = \mathbb{p } [ ( 1\ { - \varphi_x \geq h \ } ) _ { x \in \mathbb{z}^d } \in a ] \\ & = \mathbb{p } [ ( 1\ { \varphi_x \geq -h \ } ) _ { x \in \mathbb{z}^d } \in \overline{a } ] = \mathbb{p } [ \overline{a}^{-h } ] .
\end{split}\ ] ] next , we collect some classical results concerning the maximum of the gaussian free field in a finite set .
let @xmath207 .
first , note that for all @xmath18 , @xmath208 \leq \sqrt { 2 g(0 ) \log|k|},\ ] ] see for example @xcite , proposition 1.1.3 ( see also theorem 1.3.3 in @xcite for a much more general result ) .
moreover , setting @xmath209 which is finite due to , the borell - tsirelson - ibragimov - sudakov ( btis ) inequality ( see for example @xcite , theorem 2.1.1 ) , together with , yields the tail estimate , for all @xmath18 , @xmath210 \leq e^{- ( \alpha - \sqrt { \log|k| } ) ^2 } , \text { if } \alpha > \sqrt { \log|k|}.\ ] ] we also recall the following elementary tail estimate of the normal distribution . for @xmath211
( see for example @xcite , ch .
2 , p. 49 ) , @xmath212 \leq \frac{1}{h } e^{- h^2 /2 } , \text { for all $ h>0$}.\ ] ] we proceed with a classical fact concerning conditional distributions for the gaussian free field on @xmath0 , the proof of which can be found in @xcite ( see lemma 1.2 therein ) .
[ l : conds_exps ] @xmath213
@xmath214 let @xmath215 , \text { for $ x \in \mathbb{z}^d$ , $ y \in k$},\ ] ] and @xmath216 = \sum_{y \in k } p_{x , y}^k \cdot \varphi_y , \text { for $ x \in \mathbb{z}^d$,}\ ] ] which is @xmath217-measurable , and define @xmath218 by @xmath219 then , under @xmath13 , @xmath220 = g_{k^c}(x , y)$. } \end{split}\ ] ] ( in particular , @xmath221 , @xmath13-a.s . whenever @xmath222 ) .
lemma [ l : conds_exps ] yields the following choice of regular conditional distributions for @xmath204 conditioned on the variables @xmath223 , which will prove very useful in several instances below .
namely , @xmath13-a.s . ,
@xmath224 = \widetilde{\mathbb{p } } \big [ ( \widetilde{\varphi}_x^k + \mu_x^k)_{x \in \mathbb{z}^d } \in \cdot \ \big],\ ] ] where @xmath225 , @xmath156 , is given by , and @xmath226 is a centered gaussian field under @xmath227 with covariance structure @xmath228 , independent of @xmath229 , @xmath230 .
the following is an immediate corollary of lemma [ l : conds_exps ] , which provides a way to construct @xmath223 inductively from a family of independent gaussian variables .
[ l : conds_exps2 ] @xmath231 let @xmath232 be an enumeration of the elements of @xmath116 , and @xmath233 , for @xmath234 ( in particular , @xmath235 ) .
let @xmath236 , @xmath222 , be an independent family of gaussian random variables , with @xmath237 and define recursively ( with a slight abuse of notation ) @xmath238 then @xmath239 has the law of gaussian free field restricted to @xmath116 .
we proceed by induction over @xmath240 . for @xmath241 , implies that @xmath242 is indeed a centered gaussian variable with @xmath243 = \mathbb{e}[\psi_{x_{1}}^2 ] = g(0)$ ] .
suppose now that @xmath244 has the law of gaussian free field restricted to @xmath245 , for some @xmath246 .
in particular , by , @xmath247 is a ( linear ) map of @xmath248 , it is therefore independent of @xmath249 .
thus , lemma [ l : conds_exps ] ( with @xmath250 ) readily yields that @xmath251 , with @xmath252 as defined in , has the desired law .
in this section , we show theorem [ t : asymptotics_ub ] . as explained in the introduction , this includes setting up an appropriate renormalization scheme , which is done in subsection [ s : ub_rs ] , and deducing suitable recursive bounds for the probability of the relevant crossing events at different scales .
proposition [ p : dec_ineq ] entails the induction step , which is then propagated inductively in proposition [ p : dec_ineq_propagated ] along any increasing sequence of levels @xmath253 satisfying a mild growth condition ( cf . remark [ r : ub_final ] ) , provided a suitable bound for the probability of crossing events at the lowest scale holds .
obtaining the desired seed estimate requires a substantial amount of work ( see also remark [ r : renorm ] , 2 ) below , which details this difficulty more quantitatively ) and is the object of subsection [ s : ub_lcb ] .
the main result is theorem [ t : local_cb ] therein .
finally , subsection [ s : denouement ] brings together the two ingredients to complete the proof of the upper bound .
we start by developing a renormalization scheme , which , in its broad lines , is adapted from the one described in section 2 of @xcite .
we start by introducing an integer parameter @xmath254 and a geometrically increasing sequence of length scales @xmath255 and corresponding renormalized lattices @xmath256 we are interested in the crossing events ( cf . for notation ) @xmath257 where @xmath258 for all @xmath259 . the case @xmath260 requires special treatment due to the following competing interests . on the one hand , the seed estimate we will need to establish forces @xmath261 , @xmath262 , to be not too small ( this is the reason for introducing the additional parameter @xmath263 ) . on the other hand
, @xmath261 should remain sufficiently `` invisible '' to a simple random walk started at @xmath33-distance of order @xmath264 , for some @xmath88 , from @xmath81 ( see the proof of proposition [ p : dec_ineq ] below ) .
similarly to the approach taken in section 2 of @xcite , instead of looking directly at the whole crossing path above level @xmath21 needed for @xmath265 to occur , we will consider its projection onto @xmath266 small and `` well - separated '' sets at scale @xmath260 ( translates of @xmath267 ) .
these sets will be indexed by the leaves of a binary tree of depth @xmath240 .
we now describe this `` graphical cantor set '' construction more precisely .
we write @xmath268 for all @xmath269 ( with the convention @xmath270 ) , and @xmath271 for the canonical dyadic tree of depth @xmath240 . for arbitrary @xmath150 , we call a map @xmath272 a _ proper embedding of @xmath273 in @xmath0 with root at @xmath259 _ if @xmath274 we denote by @xmath275 the set of proper embeddings of @xmath273 in @xmath0 with root at @xmath259 , for @xmath150 .
one easily infers that @xmath276 .
moreover , on account of , and , if @xmath277 for some @xmath88 and @xmath259 , and if @xmath278 for some @xmath279 with descendants @xmath280 , then @xmath281 by elementary geometric considerations , and using , one then deduces that for all @xmath88 , @xmath259 and @xmath7 , @xmath282 see figure [ f : crossing_high_dim ] below , and inductively that @xmath283 accordingly , we introduce the quantity @xmath284 , \text { for $ n \geq 0 $ , $ h \in \mathbb{r}$},\ ] ] which does not depend on @xmath259 due to translation invariance , and is a decreasing function of @xmath285 .
[ h ! ] ( represented by the dashed line ) connecting @xmath286 to @xmath287 , for some @xmath88 , must cross a box @xmath288 , for some @xmath289 , as well as @xmath290 , for some @xmath291 , thus giving rise to the simultaneous occurrence of @xmath292 and @xmath293.,title="fig:",width=680 ] we will later apply a union bound in in order to bound @xmath294 $ ] . thus , estimates for @xmath295 will have to be strong enough to overcome the combinatorial complexity @xmath296 coming from the number of trees one can choose . to begin with
, the following proposition provides recursive bounds for @xmath297 , @xmath298 , along a suitable non - decreasing sequence @xmath253 .
[ p : dec_ineq ] @xmath299 , @xmath254 , @xmath300 , @xmath301 there exists a constant @xmath302 such that , defining @xmath303 given any positive sequence @xmath304 satisfying @xmath305 and any increasing , real - valued sequence @xmath306 satisfying @xmath307 one has @xmath308 the proof of proposition [ p : dec_ineq ] is similar to that of proposition @xmath309 in @xcite ( see also lemma 2.6 in @xcite and proposition 4.1 in @xcite ) , with certain modifications .
in particular , note that the constants appearing in these references all implicitly depend on @xmath3 . in the present situation
however , it is imperative to have a good control over the `` sprinkling '' condition for the sequence @xmath253 , cf .
, in terms of @xmath3 as @xmath3 grows to infinity .
let @xmath310 be a proper embedding in @xmath311 , for some @xmath150 . for @xmath312 ,
we denote by @xmath313 the proper embedding of depth @xmath240 with root at @xmath314 `` induced '' by @xmath310 , i.e. determined by @xmath315 , for all @xmath316 ( where @xmath317 stands for the concatenation of @xmath318 and @xmath319 ) , and define the sets @xmath320 on account of and , we see that @xmath321 , for @xmath312 .
we introduce a parameter @xmath322 , set @xmath323 ( cf . for the definition of @xmath324 ) , and write , abbreviating @xmath325 , @xmath326 & = \mathbb{p}[a_{\mathcal{t}_1}^h \cap a_{\mathcal{t}_2}^h ] \\ & \leq \mathbb{p }
[ a_{\mathcal{t}_1}^h \cap a_{\mathcal{t}_2}^h \cap \ { \text{max}_{_{k_1 } } \varphi \leq \alpha ' \ } ] + \mathbb{p } [ \text{max}_{_{k_1 } } \varphi > \alpha ' ] \\ & = \mathbb{e } [ 1_{a_{\mathcal{t}_1}^h } \cdot 1 _ { \ { \text{max}_{_{k_1 } } \varphi \ \leq \alpha ' \ } } \cdot \mathbb{p } [ a_{\mathcal{t}_2}^h \ , | \ , ( \varphi_{x})_{x\in k_1 } ] ] \ + \ \mathbb{p } [ \text{max}_{_{k_1 } } \varphi > \alpha ' ] . \end{split}\ ] ] by , the conditional probability appearing in the last line of can be rewritten as @xmath327 = \widetilde { \mathbb{p } } [ ( 1\ { \widetilde{\varphi}_x^{k_1 } + \mu_x^{k_1 } \geq h \})_{x\in \mathbb{z}^d } \in a_{\mathcal{t}_2 } ] , \text { $ \mathbb{p}$-a.s.},\ ] ] with @xmath328 $ ] and where @xmath329 , the canonical @xmath15-algebra on @xmath183 , is such that @xmath330 ( recall ) . in particular , @xmath331 is measurable with respect to the canonical coordinates in @xmath332 .
we will now estimate the random shift @xmath333 , for @xmath334 , which will involve the bounds , . to this end
, we first note that , by construction , @xmath335 indeed , @xmath336 ( in particular , we have used that @xmath337 for all @xmath262 in the second line ) , which , together with the constraints @xmath300 and @xmath338 , immediately yields .
thus , for arbitrary @xmath334 and @xmath339 , on the event @xmath340 , @xmath341 \\ \hspace{-1ex } & \stackrel{\eqref{1.13}}{\leq } & \hspace{-1ex } \alpha ' \cdot \text{cap}(k_1 ) \cdot
\displaystyle\sup_{x \in k_2 , \ ; y \in
k_1 } g(x , y)\\ & \stackrel{\eqref{1.12 } , \eqref{1.12bis } } { \leq } & \hspace{-1ex } \alpha ' 2^n \cdot \text{cap}(b_2(0,(\sqrt{d } + n)l_0))\cdot \displaystyle \sup_{x \in k_2 , \ ; y \in k_1 } g(x , y ) \\ & \stackrel{\eqref{g1 } , \eqref{g2}}{\leq } & \hspace{-1ex } \alpha ' 2^n ( c(\sqrt{d } + n)l_0 / \sqrt{d})^{d-2}\cdot \big(c ' \sqrt{d } / l_0^{n+1 } l_0 \big)^{d-2 } \\ & \leq & \hspace{-1ex } \alpha 2^n ( c_3(\sqrt{d}+n))^{d-2 } \cdot \big ( l_0^{-(d-2 ) } \big)^{n+1 } \stackrel{\text{def.}}{= } \gamma /2 , \end{array}\ ] ] where applies due to and because @xmath342 by assumption , and where the last line in defines the constant @xmath302 appearing in the statement of proposition [ p : dec_ineq ] above
. in particular , on the event @xmath340 and for any @xmath334 , the inequality @xmath343 implies @xmath344 hence , on the event @xmath340 , since @xmath331 is increasing , yields @xmath345 & \leq & \hspace{-1ex } \widetilde { \mathbb{p } } [ ( 1\ { \widetilde{\varphi}_x^{k_1 } - \mu_x^{k_1 } \geq h - \gamma \})_{x \in \mathbb{z}^d } \in a_{\mathcal{t}_2 } ] \\ & = & \hspace{-1ex } \widetilde { \mathbb{p } } [ ( 1\ { -\widetilde{\varphi}_x^{k_1 } - \mu_x^{k_1 } \geq
h - \gamma \})_{x \in \mathbb{z}^d } \in a_{\mathcal{t}_2 } ] \\ & = & \hspace{-1ex } \widetilde { \mathbb{p } } [ ( 1\ { \widetilde{\varphi}_x^{k_1 } + \mu_x^{k_1 } < \gamma - h\})_{x \in \mathbb{z}^d } \in a_{\mathcal{t}_2 } ] \\ & = & \hspace{-1ex } \widetilde { \mathbb{p } } [ ( 1\ { \widetilde{\varphi}_x^{k_1 } + \mu_x^{k_1 } \geq \gamma - h \})_{x \in \mathbb{z}^d } \in \overline{a}_{\mathcal{t}_2 } ] \\ & \stackrel{\eqref{phi_cond_exps}}{= } & \hspace{-1ex } \mathbb{p } [ \overline{a}_{\mathcal{t}_2}^ { \gamma - h } \ , | \ ,
( \varphi_{x})_{x\in k_1 } ] , \end{array } \end{split}\ ] ] where @xmath346 denotes the flipped event ( recall the notation from above ) , and where we have used in the second line that @xmath347 and @xmath348 have the same law under @xmath227 , cf . , and that @xmath227 does not act on @xmath349 . inserting this bound into , we obtain @xmath350 & \leq \mathbb{p } [ a_{\mathcal{t}_1}^h \cap \overline{a}_{\mathcal{t}_2}^ { \gamma - h } ] + \mathbb{p } [ \text{max}_{_{k_1 } } \varphi > \alpha ' ] \\ & \leq \mathbb{p } [ a_{\mathcal{t}_1}^h ] \cdot \mathbb{p } [ a_{\mathcal{t}_2}^ { h - \gamma } ] + \mathbb{p } [ \text{max}_{_{k_1 } } \varphi > \alpha ' ] , \end{split}\ ] ] and the second step is due to and the fkg - inequality for the free field ( see for example @xcite , ch .
4 ) , which applies and yields an upper bound , because the event @xmath351 is increasing and @xmath352 is decreasing , cf . , . it remains to bound the error term .
since @xmath353 , we have by that @xmath354 , and , imply that ( recall that @xmath323 ) @xmath355 \leq e^{-(\alpha- m_n(d , l_0,n))^2 } , \text { for all $ \alpha$ satisfying \eqref{alpha_n_cond}}.\ ] ] substituting this into yields , for all @xmath356 fulfilling and all @xmath357 , @xmath358 \ \leq \
\mathbb{p}[a_{\mathcal{t}}^h ] \leq \mathbb{p } [ a_{\mathcal{t}_1}^{h- \gamma } ] \cdot \mathbb{p } [ a_{\mathcal{t}_2}^{h - \gamma } ] + e^{-(\alpha_n - m_n(d , l_0,n))^2}.\ ] ] the claim now readily follows upon taking suprema over all @xmath359 on both sides , letting @xmath360
( @xmath21 was arbitrary ) , @xmath361 , so that requiring @xmath362 , by virtue of , is precisely the condition .
this concludes the proof of proposition [ p : dec_ineq ] .
we will now propagate the bounds inductively along a suitable sequence @xmath306 , and select to this end @xmath363 for some parameter @xmath364 to be specified later , and with @xmath365 as defined in .
in particular , note that satisfies the condition for every choice of @xmath366 .
[ p : dec_ineq_propagated ] @xmath299 , @xmath254 , @xmath300 , @xmath301 assume @xmath367 and @xmath368 are such that @xmath369 and let the sequence @xmath306 satisfy with @xmath370 as defined in .
then , @xmath371 on account of proposition [ p : dec_ineq ] and the choice of @xmath372 , @xmath150 , in , ( compare this to proposition 2.2 and ( 2.51 ) in @xcite , respectively ) , the proof of proposition [ p : dec_ineq_propagated ] is completely analogous to that of proposition 2.4 in @xcite .
we therefore omit it .
[ r : renorm ] \1 ) even though we will not need this below , note that the conclusions of propositions [ p : dec_ineq ] and [ p : dec_ineq_propagated ] continue to hold for arbitrary increasing ( seed ) events @xmath373 , for @xmath262 , which are measurable with respect to the canonical coordinates in @xmath261 as defined in , upon letting @xmath374 , for all @xmath262 and @xmath7 , defining events @xmath375 , with @xmath376 for arbitrary @xmath150 and @xmath377 , as in , and taking an additional supremum over @xmath259 in the definition of @xmath295 ( no translation invariance required ) .
moreover , by symmetry , if instead all events @xmath378 , @xmath262 , are decreasing , under the assumptions of proposition [ p : dec_ineq ] , the conclusion holds for the sequence @xmath379 in place of @xmath306 .
\2 ) the condition is quite strong . indeed , by and
, we have , for all @xmath7 , @xmath150 and @xmath380 , @xmath381 = \mathbb{p}[b_{\infty}(x , l_n ) \stackrel{\geq h}{\longleftrightarrow } \partial_{\text{int}}b_{n , x } ] \leq |\lambda_{n , x}| \cdot p_n(h).\ ] ] if the bound were to hold for some value of @xmath382 , it would yield , together with the estimate on the combinatorial complexity @xmath296 , that @xmath294 \leq ( c_2 l_0^{2(d-1 ) } / e^{k_0 - b})^{2^n}$ ] . thus @xmath383 will have to satisfy the requirement @xmath384 for a certain constant @xmath385 in order for this bound to be of any use . substituting this into yields the condition @xmath386 \leq c ' l_0^{-2(d-1)}$ ] . for the purpose of proving theorem [ t : asymptotics_ub ] , @xmath387 will be at most @xmath388 , for some @xmath389 .
we will therefore essentially have to prove that the crossing event @xmath390 decays roughly like @xmath391 for all @xmath392 ( recall that @xmath393 is required to grow at least like @xmath394 , cf . ) .
@xmath395 in this subsection , we provide an estimate regarding the local connectivity properties of the level set @xmath103 , for arbitrary @xmath53 , in high dimension . here , `` local '' means that we only investigate connectivity within an @xmath82-ball having a radius of order at most @xmath3 around a given point on the lattice .
specifically , we will consider the probability of a nearest - neighbor path in @xmath103 connecting the center of such a ball to its boundary .
the resulting bound , cf . theorem [ t : local_cb ] , is of independent interest , but it will be crucial in establishing the seed estimate , thus enabling us to launch the renormalization .
we begin with the following simple lemma , which will be useful in several instances below .
[ l : zeta_i_bound ] @xmath396 there exists a constant @xmath385 such that , for all @xmath397 , @xmath156 , all sets @xmath398 satisfying @xmath399 and @xmath400 , and all @xmath53 , @xmath401 \leq \exp \big\{- \frac{c \ , \varepsilon^2}{\ell(\ell+1 ) } d \log d \big\},\ ] ] with @xmath402 as defined in .
let @xmath397 , @xmath156 , @xmath53 , and the sets @xmath403 be fixed as to satisfy the above assumptions .
we abbreviate @xmath404 , which is a centered gaussian variable .
we compute , for arbitrary @xmath405 , @xmath406 & = \lambda^2 \sum_{y , z \in u } p_{x , y}^k \ , p_{x , z}^k \ , \mathbb{e}[\varphi_y \varphi_z ] \\ & = \lambda^2 \big\ { g(0 ) \sum_{y \in u } ( p_{x , y}^k)^2 + \sum_{\substack { y , z \in u \\ y\neq z } } p_{x , y}^k \ , p_{x , z}^k \ , g(y - z ) \big\}. \end{split}\ ] ] observe that for all @xmath407 and @xmath18 , @xmath408 \leq p_{x}[h_y < \infty ] \leq c / d,\ ] ] where the last step follows from an elementary application of the strong markov property at time @xmath409 ( recall that @xmath410 ) and .
moreover , for all @xmath411 and @xmath1 , by the strong markov property at time @xmath412 , @xmath413\cdot g(0 ) \stackrel{\eqref{g(0 ) } , \eqref{var_bound2 } } \leq c / d.\ ] ] inserting the bounds , into and using again that @xmath414 as @xmath10 ( cf . ) gives @xmath415 where the second step follows because @xmath400 by assumption .
thus , by markov s inequality , we obtain , for all @xmath416 and @xmath18 , @xmath417 \leq e^{- \lambda \varepsilon h_{\text{as } } } \cdot e^{\text{var } ( \lambda \zeta_x)/2 } \leq \exp \big\{- \frac{1}{2c \ell ( \ell + 1 ) } d ( \varepsilon h_{\text{as}})^2\big\},\ ] ] where the last step follows by optimizing over @xmath418 , which occurs for the choice @xmath419 ( with @xmath420 the constant appearing in the bound for @xmath421 above ) . on account of , and since @xmath422 , this implies , and thus completes the proof of lemma [ l : zeta_i_bound ] .
we proceed to the main result of this section . for convenience ,
we introduce the shorthand @xmath423 [ t : local_cb ] @xmath424 there exist constants @xmath425 and @xmath426 such that for all @xmath427 and @xmath428 , @xmath429 \leq \exp \{- c_5 \ , f(\varepsilon , n ) \cdot d \log d \},\ ] ] where @xmath430 .
we briefly outline the proof of theorem [ t : local_cb ] , which essentially comprises three steps .
first , instead of considering the whole path connecting the origin to @xmath431 directly , we look at the `` local traces '' it leaves after first visiting each of @xmath432 ( for suitable @xmath433 ) concentric @xmath82-annuli around the origin having a width of order @xmath3 ( similarly to what was done in @xcite in the context of interlacement percolation ) . we are led to consider the probability that @xmath434 `` well - separated '' paths in @xmath435 of length @xmath436 each , for some @xmath437 , all occur simultaneously ( which competes against a suitable combinatorial complexity ) .
the parameter @xmath438 will have to be carefully chosen a posteriori . in a second step , we discover the field along a fixed collection of paths `` dynamically , '' using lemma [ l : conds_exps2 ] , as alluded to in the introduction .
finally , lemma [ l : zeta_i_bound ] will provide good controls on the error terms that arise .
we define @xmath439 let @xmath53 be fixed and assume @xmath440 ( stronger conditions on @xmath263 will follow ) . instead of working with @xmath263 directly , it will be convenient to use @xmath441 we also introduce a parameter @xmath442 to be specified later . given a vertex @xmath156 , we denote by @xmath443 the set of all self - avoiding nearest - neighbor paths of length @xmath444 starting in @xmath81 ( by convention , if @xmath445 , the set consists only of the vertex @xmath81 itself ) . by construction
, any path connecting the origin to @xmath431 intersects all spheres @xmath446 , for @xmath447 .
moreover , any self - avoiding path @xmath448 , with @xmath449 for some @xmath450 and @xmath451 , satisfies @xmath452 .
thus , @xmath453 \\ & \qquad \leq \mathbb{p}\big[\bigcup_{\substack{x_n \in s_1(0 , n c_6 d ) \\
0 \leq n < n_0 } } \ ,
\bigcup_{\substack{\pi_n \in \pi_\ell(x_n ) \\ 0
\leq n < n_0 } } \big\ { e_\varphi^ { \geq h_{\text{as}}^{(\varepsilon ) } } \supseteq \text{range}(\pi_n ) , 0 \leq n < n_0 \big\}\big ] \\ & \qquad \leq \sum_{\substack{x_n \in s_1(0 , nc_6 d ) \\ 0 \leq n < n_0 } } \ ,
\sum_{\substack{\pi_n \in \pi_\ell(x_n ) \\ 0
\leq n < n_0 } } \mathbb{p } \big [ e_\varphi^ { \geq h_{\text{as}}^{(\varepsilon ) } } \supseteq \text{range}(\pi_n ) , 0 \leq n
< n_0 \big ] , \end{split}\ ] ] for all @xmath454 ( i.e. @xmath455 ) and @xmath451 ( see also figure [ f : local_trace ] below ) . for later reference , we note that given two paths @xmath456 and @xmath457 for some @xmath458 , with @xmath459 , @xmath460 and @xmath461 , @xmath462 [ h ! ] of `` local traces '' left by a self - avoiding path in @xmath463 connecting the origin to @xmath464 , obtained by considering the first @xmath465 steps after first hitting each of the spheres @xmath466 , for @xmath467 ( when viewed as starting from @xmath11 ) .
the paths @xmath468 , @xmath467 , are in fact `` well - separated , '' as captured by .,title="fig:",width=566 ] since @xmath469 for all @xmath1 , @xmath458 and @xmath156 , and using the estimate on the cardinality of @xmath82-spheres , the number of terms appearing in the double sum of is bounded by @xmath470 for all @xmath454 , @xmath471 and @xmath472 . returning to , this yields @xmath473 \\ & \qquad \leq \mathbf{c}_d(\varepsilon , n_0 , \ell ) \cdot \sup_{\substack{x_n \in s_1(0 , nc_6d ) \\ \pi_n \in \pi_\ell ( x_n )
\\ 0 \leq n < n_0 } } \mathbb{p } [ e_\varphi^ { \geq h_{\text{as}}^{(\varepsilon ) } } \supseteq \text{range}(\pi_n ) , 0 \leq n < n_0 ] , \end{split}\ ] ] for all @xmath454 , @xmath471 and @xmath472 .
we now focus on the probability appearing on the right - hand side of .
thus , we fix @xmath434 self - avoiding paths @xmath474 and define @xmath475 for notational convenience , we introduce the set of labels @xmath476 ( where @xmath477 , @xmath478 , denote the coordinates of @xmath479 , with values in @xmath480 and @xmath481 , respectively ) , so that @xmath482 for all @xmath467 and @xmath483 .
we further denote by @xmath484 the lexicographic order on @xmath485 .
this induces an ordering of the points in @xmath116 .
finally , we also set @xmath486 the event @xmath487 is measurable with respect to the @xmath15-algebra generated by @xmath229 , @xmath222 , and we will now construct the field @xmath239 using lemma [ l : conds_exps2 ] , by adding one variable at a time , according to the ordering @xmath484 ( thus , @xmath488 as defined in denotes the set of points at which the field has been discovered `` before time @xmath479 '' ) . specifically , we proceed as follows .
let @xmath489 , @xmath490 , be an independent family of gaussian random variables , with @xmath491 ( see for the definition of the killed green function ) , and recall from that @xmath492 $ ] , for all @xmath144 .
we define recursively @xmath493 by lemma [ l : conds_exps2 ] , the law of @xmath494 as defined above is precisely that of the gaussian free field restricted to @xmath116 . by , the main contribution to the sum on the right - hand side of will come from the points which belong to the same path as @xmath495 ( i.e. to @xmath496 ) . accordingly
, we introduce the sets @xmath497 and obtain the decomposition , @xmath498 where the above sums are understood to vanish identically whenever the summation is over the empty set .
now , observe that for all @xmath490 , on the event @xmath499 , @xmath500 \cdot \sup_{y \in k_i \setminus u_i } \varphi_y . \end{split}\ ] ] in order to bound the hitting probability @xmath501 $ ] , we note that , by construction , cf . and ,
@xmath502 thus , we can find @xmath503 such
that @xmath504 & \stackrel{\eqref{1.13}}{\leq } \text{cap}(k_i \setminus u_i ) \cdot \sup_{y \in k_i \setminus u_i } g(y - x_i ) \\ & \stackrel{\eqref{c_0}}{\leq } |k_i \setminus u_i | \cdot 2^{-d/2 -2 } \leq e^{-cd } , \end{split}\ ] ] for all @xmath490 with @xmath505 and all @xmath506 , where applies because of , and we have bounded @xmath507 using , and .
inserting into yields that there exists a suitable constant @xmath508 such that for all @xmath490 and @xmath509 , on the event @xmath499 , @xmath510 ( with the convention @xmath511 , which occurs if @xmath512 , i.e. @xmath513 ) .
thus , for all @xmath490 and @xmath509 , @xmath514 and therefore @xmath515 going back to our initial event in , by a union bound , we obtain , for all @xmath454 , @xmath458 and @xmath509 , @xmath516 \\ & \qquad \leq \prod_{i \in i } \mathbb{p}[\psi_i \geq h_{\text{as}}^{(\varepsilon/2 ) } ] + \mathbb{p}\big [ \sup_{y \in k } \varphi_y > e^{c_7 d}\big ] + |i| \cdot \sup_{i \in i } \mathbb{p}[\zeta_i > \varepsilon h_{\text{as } } /4 ] , \end{split}\ ] ] where we also used independence of the variables @xmath489 , @xmath490 , see above .
we consider each of the three terms in separately .
recall the definition of the combinatorial complexity @xmath517 above .
[ l : master_formula_bd ] @xmath518 \leq \mathbf{c}_d^{-1 } ( \varepsilon , n_0 , \ell ) \cdot e^{- c \varepsilon^2 n_0 \lceil \ell d \rceil \log d } , \text { for $ n_0 \geq 1 $ , $ 0 < \ell \leq 1 $ , $ d \geq c'$}.\label{eq : local_conn_final3 } \\ & \mathbb{p}\big [ \sup_{y \in k } \varphi_y > e^{d}\big ] \leq \mathbf{c}_d^{-2}(\varepsilon , n_0 , \ell ) , \text { for $ n_0 \geq 1 $ , $ 0 < \ell \leq 1 $ , $ d \geq c(\varepsilon , n_0 , \ell)$}. \label{eq : local_conn_final2}\end{aligned}\ ] ] moreover , for the choice of @xmath519 one has @xmath520 \stackrel{\eqref{psi_i}}{= } g_{k_i^c}(x_i , x_i ) \stackrel{\eqref{g - gsubk}}{\leq } g(0 ) \stackrel{\eqref{eq : c_1 } } { < }
c_1 ^ 2 , \text { for all $ i \in i$ and $ d \geq 3$}.\ ] ] hence , the elementary tail estimate yields ( note that @xmath521 , cf . ) , @xmath522 \leq \big(c_1 e^{- h_{\text{as}}^2(1 + \varepsilon/2)^2/2g(0 ) } \big)^{n_0 \lceil \ell d \rceil},\ ] ] for all @xmath454 , @xmath458 and @xmath523 , which , upon expanding @xmath524 , immediately implies . the estimate is also a direct consequence of and a union bound ( the ease with which this is obtained merely reflects the fact that for any @xmath495 , @xmath525 , the shift produced by points lying on paths discovered `` before '' @xmath496 ( if any ) , has little influence on @xmath526 , cf . and ) .
it remains to show .
observe that , by definition of @xmath527 , see , lemma [ l : zeta_i_bound ] with @xmath528 and @xmath529 ( which satisfies @xmath530 , cf . ) , applies and yields ( for a suitable @xmath531 ; this defines the constant appearing in ) , @xmath532 \leq \exp\ { -c_8 \varepsilon^2 d \log d / \ell ( \ell + 1)\},\ ] ] for all @xmath490 , @xmath454 , @xmath437 and @xmath533 . selecting @xmath534 as defined in , which , in particular , satisfies the requirements of
, ensures that @xmath535 ( the last term should be read as twice the exponent appearing in the combinatorial complexity @xmath517 ) , for all @xmath454 and @xmath18 , and follows .
@xmath395 substituting the bounds , and into , we see that for the value of @xmath438 in , all @xmath454 and @xmath536 , @xmath537 \leq \mathbf{c}_d^{-1}(\varepsilon , n_0 ) \cdot e^{- c ' \varepsilon^2 n_0 \ell d \log d}\ ] ] ( the contribution from is dominating ) . finally ,
inserting this into , noting that @xmath538 whenever @xmath539 for some constant @xmath540 , cf . , and on account of , we readily obtain , with @xmath541 , and for a suitable value of @xmath426 .
this completes the proof of theorem [ t : local_cb ] .
the only feature of the function @xmath542 which will be of importance below is that @xmath543 for every @xmath53 . for the sole purpose of proving theorem [ t : asymptotics_ub ] , any other function @xmath542 with this property would have sufficed .
moreover , while the term @xmath544 appearing in the bound looks innocent when @xmath545 $ ] , it is crucially used in to ensure that the resulting function @xmath546 indeed grows at least like a positive power of @xmath263 ( in fact one could even have obtained @xmath547 from ) .
@xmath395 finally , we complete the proof of theorem [ t : asymptotics_ub ] .
this will involve the renormalization scheme introduced in section [ s : ub_rs ] . as noted in remark [
r : renorm ] , 2 ) , the seed estimate needed to initiate the renormalization is rather strong , and theorem [ t : local_cb ] will be of crucial use to establish it .
let @xmath53 be fixed . in order to establish
, we have to show that @xmath548 to begin with , we select the parameters @xmath549 , @xmath393 and @xmath263 in the renormalization scheme from section [ s : ub_rs ] , cf . .
we let @xmath550 ( recall the statement of theorem [ t : local_cb ] for the definition of @xmath551 , @xmath552 ) , so that the constraint @xmath553 in holds for all @xmath55 , and choose @xmath383 appearing in as @xmath554 ( see for the definition of @xmath555 ) .
lastly , we define the increasing sequence @xmath253 recursively as @xmath556 with @xmath372 given by . in particular , @xmath557 satisfies the `` sprinkling '' condition . to see that the choices in , and are judicious , we first check that @xmath558 indeed , it follows from , using , and , that for all @xmath55 and @xmath298 , @xmath559 ( the dependence on @xmath100 is due to @xmath263 ) .
hence , @xmath560 for all @xmath561 .
this is more than enough to deduce ( we comment on this in remark [ r : ub_final ] below ) . in order to complete the proof of theorem [ t : asymptotics_ub ] , we apply proposition [ p : dec_ineq_propagated ] . except for the condition ,
all its requirements are clearly satisfied by our choices of parameters in , and , whenever @xmath55 . to deduce the necessary seed estimate
, we use the connectivity bound from theorem [ t : local_cb ] . observing that the choice of @xmath562 in guarantees that @xmath563 , for @xmath564 and @xmath552 as appearing in , it follows that @xmath565 \\ & \stackrel{\eqref{eq : b_n , x } , \
, \eqref{t : ub_h}}{\leq } & 2d ( 2l_0 + 1)^{d-1 } \cdot \mathbb{p } [ 0 \stackrel{\geq h_{\text{as}}^{(\varepsilon/2)}}{\longleftrightarrow } s_1(0 , n l_0 ) ] \\ & \stackrel{\eqref{t : local_cb1 } , \ , \eqref{t : ub_parameters}}{\leq } & e^{d\log3d - 6 d\log d } \leq d^{-4d } , \end{array}\ ] ] for all @xmath392 . on the other hand , impies that @xmath566 for all @xmath523 , hence the condition @xmath567 in holds for all@xmath392 . we may thus apply proposition [ p : dec_ineq_propagated ] , thereby obtaining @xmath568 with @xmath306 as defined in .
hence , for all @xmath88 and @xmath55 , ( n.b . :
the first of the following chain of inequalities is obtained by covering @xmath569 with essentially disjoint @xmath33-boxes of radius @xmath264 , and performing a union bound ) @xmath570 \leq 2^d \ , \mathbb{p}[b_\infty(0,l_n ) \stackrel{\geq h_{\text{as}}^{(\varepsilon)}}{\longleftrightarrow } s_\infty(0,3l_n ) ] \\ & \begin{array}{rcl } \qquad & \stackrel{\eqref{eventa } , \eqref{eq : b_n , x}}{= } & \hspace{-1ex } 2^d \ , \mathbb{p}[a_{n,0}^{h_{\text{as}}^{(\varepsilon ) } } ] \stackrel{\eqref{t : ub_h_infty}}{\leq } 2^d \ , \mathbb{p}[a_{n,0}^{h_n } ] \stackrel{\eqref{cascading_events } , \eqref{p_n}}{\leq } 2^d |\lambda_{n,0}| \cdot p_n(h_n ) \\
\qquad & \stackrel{\eqref{lambda_n , x},\eqref{p_n_bound_final}}{\leq } & \hspace{-1ex } 2^d \big((c_2 d)^{2(d-1)}\big)^{2^n } \cdot \big((2 c_2 d)^{2(d-1)}\big)^{-2^n } \leq 2^{-2(d-1)(2^n-1 ) } \xrightarrow{n \to \infty } 0 .
\end{array } \end{split}\ ] ] in particular , this implies that @xmath571=0 $ ] , for all @xmath55 .
recalling the definition of the critical level @xmath41 , this yields , and thus completes the proof of theorem [ t : asymptotics_ub ] . @xmath395 [ r : ub_final ] our renormalization scheme is somewhat asymmetrical . on the one hand
, the sprinkling condition for the sequence @xmath253 turns out to be very mild , as it only costs an `` additive '' @xmath100 , cf .
( obtaining @xmath572 , for all @xmath573 would have sufficed for the purpose of proving theorem [ t : asymptotics_ub ] ) . on the other hand ,
the scheme relies on the strong seed estimate , and establishing it is what prevents us from obtaining a more precise result than .
we proceed to show the lower bound , theorem [ t : asymptotics_lb ] .
as described in the introduction , the proof comprises two main steps . the first one , which is the subject of subsection [ s : finite_size_criterion ] , reduces the problem of constructing an infinite cluster to a local statement .
the assertion is roughly the following ( see theorem [ t : local_global ] below ) : given @xmath574 , if the level @xmath21 is such that with high probability , the set @xmath575 possesses a ubiquitous component in each of the @xmath3-dimensional hypercubes @xmath576 , for @xmath577 ( @xmath578 ) and @xmath579 , which are all connected , then @xmath30 percolates whenever @xmath3 is sufficiently large ( depending on @xmath580 ) . in order to prove the lower bound
, it then suffices to verify that this criterion holds when @xmath581 , for arbitrary @xmath53 .
this step is split again into two parts , to which subsections [ s : subst_comp ] and [ s : connecting_subst_comps ] are respectively devoted .
first , we construct a wealth of substantial components in the hypercube , with cardinality growing polynomially in @xmath3 , and show that most vertices in @xmath582 are either contained in or neighboring such a component .
the main result is entailed in corollary [ c : number_subst_comps ] below .
the second part consists of patching together these substantial components to form a `` giant '' one , and to then connect the latter to the ubiquitous components contained in the neighboring translates of @xmath583 .
this is achieved in theorem [ t : giant_comp ] .
all ingredients are put together at the end of subsection [ s : connecting_subst_comps ] to complete the proof of .
we begin by establishing the finite - size criterion that guarantees percolation , theorem [ t : local_global ] below . to cope with the long - range interactions , we use a renormalization argument on @xmath58 , which bears some resemblance to the one developed in section [ s : ub_rs ] above , together with a standard duality argument .
we view @xmath58 as a subset of @xmath0 by identifying @xmath584 with @xmath585 , and write @xmath586 , with @xmath587 , @xmath126 , @xmath134 , for two - dimensional balls .
similarly , we let @xmath588 . moreover
, we denote the @xmath3-dimensional hypercube and its translates in @xmath58 by @xmath589 and abbreviate @xmath590 , so that @xmath591 , @xmath587 , forms a partition of @xmath592 .
the following definition is essentially borrowed from @xcite , section 2 ( this will in particular enable us to reinject certain isoperimetric controls which follow from the results in @xcite ( see lemma 4.3 in @xcite ) directly at a later stage ) .
for arbitrary @xmath7 and @xmath587 , we introduce the ( local ) event @xmath593 ( recall that @xmath594 denotes the @xmath82-closure of @xmath595 in @xmath596 ) . a vertex @xmath587 will be called @xmath21-good if @xmath597 occurs , for arbitrary @xmath7 , and @xmath21-bad otherwise .
note that for all @xmath1 , each of the sets @xmath595 in is necessarily unique in @xmath596 : indeed , if @xmath598 satisfies @xmath599 , then in fact @xmath600 ( each vertex has @xmath3 neighbors in @xmath94 ) , and solving for @xmath601 yields @xmath602 . but any other set @xmath603 not connected to @xmath604 must be contained in @xmath605 and thus satisfy @xmath606 , hence @xmath604 is unique .
accordingly , we will henceforth refer to any set @xmath595 appearing in as _ the _ giant component of @xmath607 .
it is then plain from that for all @xmath1 and @xmath7 , @xmath608 i.e. percolation of @xmath21-good sites in @xmath58 implies percolation of @xmath30 ( in @xmath609 ) . to see this ,
observe that if two neighboring vertices @xmath610 ( with @xmath611 a unit vector in @xmath58 ) are both @xmath21-good , then the corresponding giant clusters @xmath612 and @xmath613 ( cf . )
are connected to the _ same _
( by uniqueness ) giant cluster in @xmath614 , thus @xmath615 and @xmath616 belong to the same cluster of @xmath30 .
it then follows inductively that an infinite nearest - neighbor path of @xmath21-good vertices in @xmath58 implies the existence of an infinite cluster in @xmath617 . by and the definition of @xmath48 in , in order to prove , it suffices to show that @xmath622 > 0 , \text { for all $ d\geq c(\varepsilon)$,}\ ] ] for this implies @xmath623>0 $ ] and hence @xmath624 , for all @xmath573 .
the proof of involves a renormalization scheme , which we now describe .
we introduce the integer parameter @xmath625 and define an increasing sequence of length scales @xmath626 recursively as @xmath627 ( this should not be confused with the geometric sequence of length scales appearing in section [ s : ub_rs ] ) , together with corresponding renormalized lattices @xmath628 we also introduce , for arbitrary @xmath150 and @xmath629 , the ( bad ) events @xmath630 ( recall that @xmath631 is called @xmath21-bad if @xmath632 occurs ) . by , the event @xmath633 is decreasing ( in @xmath19 ) . moreover , by translation invariance , the function @xmath634 , \text { for $ n \geq 0 $ , $ x\in \mathbb{l}^{(2)}_n$ , $ h \in \mathbb{r}$},\ ] ] is well - defined ( i.e. independent of @xmath81 ) and non - decreasing in @xmath21 , for every @xmath150 .
the key to establishing will be to show that , if @xmath549 is a suitable ( increasing ) function of @xmath3 , and holds for some @xmath53 , the probability @xmath635 decays sufficiently rapidly to @xmath11 as @xmath636 , for all @xmath573 , see below .
together with a straightforward ( planar ) duality argument , which will be detailed below , this will then yield . to obtain good estimates for @xmath635 , @xmath150
, we first develop `` recursive bounds '' relating the functions @xmath637 and @xmath638 , for arbitrary @xmath150 ( similar in spirit to what was done in proposition [ p : dec_ineq ] above , but simpler ) .
to this end , we let @xmath150 and @xmath639 be fixed , and introduce the sets @xmath640 by geometric considerations similar to those leading to , see also figure [ f : crossing_high_dim ] , and on account of , we deduce that @xmath641 \leq c_0 ' l_n^2 \sup_{\substack{x_i \in s_i \\ i=1,2 } } \mathbb{p}[d_{n , x_1}^h \cap d_{n , x_2}^h],\ ] ] for a suitable constant @xmath642 and all @xmath7 .
we consider the probability on the right - hand side of , for fixed @xmath643 , @xmath312 and @xmath7 .
abbreviating @xmath644 and observing that @xmath645 , for @xmath312 , due to and , we write @xmath646 \leq & \ \mathbb{p}[d_{n , x_1}^h , \ ,
\inf_{y \in k_1 } \varphi_y \geq - \beta_n , \ , \mathbb{p}[d_{n , x_2}^h \ , | \ , ( \varphi_y)_{y \in k_1 } ]
] \\ & + \mathbb{p } [ \inf_{y \in k_1 } \varphi_y < - \beta_n ] \end{split}\ ] ] for some cut - off value @xmath647 to be selected below
. by , @xmath648= \widetilde{\mathbb{p}}[(1\ { \widetilde{\varphi}_x^{k_1 } + \mu_x^{k_1 } \geq h \})_{x \in \mathbb{z}^d } \in d_{n , x_2 } ] , \text { $ \mathbb{p}$-a.s.}\ ] ] ( recall the notation from ) , and we focus on bounding the random shift @xmath649 , for @xmath334 . by construction , cf . , and , @xmath650 for a suitable constant @xmath651 .
hence , setting @xmath652 we obtain , on the event @xmath653 , for all @xmath334 and @xmath654 , @xmath655 \stackrel{\eqref{1.13}}{\leq } \beta_n \ , \text{cap}(k_1 ) \sup_{y \in k_1 } g(y - x ) \\ & \stackrel{\eqref{c_0}}{\leq } \beta_n |k_1| \big(\frac{c_0 d } { c_1 ' l_{n+1}}\big)^{d/2 - 2 } \stackrel{\eqref{eq : local_global_l_0}}{\leq } c \beta_n l_n^2 2^d l_n^{-(d/2 - 2 ) } \stackrel{\text{def.}}{= } \frac{\delta_n}{2 } , \end{split}\ ] ] where in the second line , we have used the crude estimate @xmath656 ( see , ) , the bound @xmath657 , which follows immediately from , and the fact that @xmath658 for all @xmath150 , due to and .
we henceforth tacitly assume that @xmath659 . by , on @xmath653 ,
the inequality @xmath660 implies that @xmath661 , for arbitrary @xmath334 , and therefore @xmath662 & \leq \widetilde{\mathbb{p}}[(1\ { \widetilde{\varphi}_x^{k_1 } - \mu_x^{k_1 } \geq h + \delta_n \})_{x \in \mathbb{z}^d } \in d_{n , x_2 } ] \\ & = \widetilde{\mathbb{p}}[(1\ { - \widetilde{\varphi}_x^{k_1 } - \mu_x^{k_1 } \geq h + \delta_n \})_{x \in \mathbb{z}^d } \in d_{n , x_2}]\\ & = \mathbb{p}[\overline{d}_{n , x_2}^{-(h + \delta_n ) } \ , | \ , ( \varphi_x)_{x \in k_1 } ] , \end{split}\ ] ] on the event @xmath653 , where we have used that @xmath663 is decreasing in the first line , and the symmetry of @xmath664 in the second line ( recall also ) . inserting into , applying the fkg - inequality and ,
we obtain @xmath665 \leq \mathbb{p}[d_{n , x_1}^{h } ] \cdot \mathbb{p } [ d_{n , x_2}^{h+ \delta_n } ] + \mathbb{p } [ \inf_{x \in k_1 } \varphi_x < - \beta_n].\ ] ] finally , substituting this into , taking suprema over @xmath643 , for @xmath312 , and using symmetry yields , in view of , @xmath666,\ ] ] for all @xmath150 , @xmath7 and @xmath647 . this yields the desired recursive bounds . in order to propagate them inductively
, we select @xmath667 with @xmath324 as defined in and @xmath668 in .
the key estimate comes in the following result . by and
, we have @xmath673 , for all @xmath150 . substituting this into and using that @xmath674 , we obtain the bound @xmath675 , for all @xmath1 and @xmath150 . inserting this into the definition of @xmath676 , see , yields @xmath677 for all @xmath523 , where we have used @xmath678 in the last line . due to the choice of @xmath549 in , given @xmath679
it follows that @xmath680 for all @xmath573 .
moreover ( for reasons that will become clear shortly ) we observe that @xmath681 whenever @xmath682 , due to the choice of @xmath549 in and @xmath683 in theorem [ t : local_global ] ( this is why @xmath683 should not be chosen too large ) . given @xmath53 ,
we define @xmath684 appearing in the statement of lemma [ l : local_global_q ] in a way that and simultaneously hold whenever @xmath685 .
we now prove by induction over @xmath240 .
let @xmath7 , @xmath53 and @xmath685 be fixed . in view of , and because the function @xmath638 is non - decreasing for all @xmath150 , it suffices to show that @xmath686 ( with the convention that the sum equals @xmath11 when @xmath260 ) . by assumption , cf .
, we have that holds for @xmath260 .
assume now it holds for some @xmath150 . by
, we have @xmath687 we bound each of the two terms appearing on the right - hand side separately . by the btis - inequality and the choice of @xmath688 in , @xmath689 moreover , @xmath690 substituting and into yields @xmath691 , as desired .
this completes the proof of , and thus of lemma [ l : local_global_q ] .
@xmath395 we now complete the proof of theorem [ t : local_global ] .
let @xmath53 and @xmath692 be a sequence of levels such that holds .
the latter implies that @xmath693 \\ & \stackrel{\eqref{eq : local_global_events}}{\leq } l_0 ^ 3 \ , \mathbb{p } \big [ \bigcup_{y \in s_\infty^{(2 ) } ( 0,l_0 ) } \ { \text{$y$ is $ ( h(d)+ \varepsilon)$-bad } \}\big ] \\ & \stackrel{\eqref{eq : local_global_scales2}}{\leq } c l_0^{2 + 3a } \mathbb{p}\big[\big(g_0^{h(d)+ \varepsilon}\big)^c \big ] \leq 1 , \end{split}\ ] ] for all @xmath55 , where the last step follows from the choice of @xmath549 in and the assumption . by lemma [ l : local_global_q ] ,
we thus obtain @xmath694 with this estimate at hand , we now prove the assertion by a standard planar duality argument .
let us call a @xmath145-_circuit around @xmath11 _ any closed @xmath145-path @xmath695 in @xmath58 such that the origin is contained in a finite connected component of @xmath696 .
denoting by @xmath697 the unit vector in the first coordinate , we have , recalling the definition of the events @xmath698 in , @xmath699e_1 \cap \mathbb{z}$ } \big\ } \\ & \subseteq \big [ \bigcup_{x \in b_\infty^{(2)}(0,3l_0 ) } ( g_y^{h(d ) } ) ^c \big ] \cup \bigcup_{n\geq 0 } \ \bigcup_{y \in \mathbb{l}_n^{(2 ) } \cap ( 3l_n , 3l_{n+1}]e_1 } d_{n , y}^{h(d)}. \end{split}\ ] ] by the choice of scales in and , this yields , for all @xmath573 , @xmath700 \\ & \begin{array}{rcl } \qquad & \hspace{-1ex } \leq & \hspace{-1ex } \displaystyle cl_0 ^ 2 \mathbb{p } [ ( g_0^{h(d ) } ) ^c ] + \sum_{n\geq 0 } 3l_n \cdot q_n(h(d ) ) \\ \qquad & \hspace{-1ex } \stackrel{\eqref{eq : local_global_q_bound_final}}{\leq } & \hspace{-1ex } \displaystyle c ' d^2 \mathbb{p } [ ( g_0^{h(d ) } ) ^c ] + 3 \sum_{n\geq 0 } l_n^{-2 } \\
\qquad & \hspace{-1ex } \leq & \hspace{-1ex } \displaystyle c ' d^2 \mathbb{p } [ ( g_0^{h(d ) + \varepsilon } ) ^c ] + \sum_{n\geq 0 } ( c''d)^{-2a(1+a)^n } \\
\qquad & \hspace{-1ex } \stackrel{\eqref{eq : local_global } } { < } & \hspace{-1ex}1 , \end{array } \end{split}\ ] ] where we have also used in the penultimate step that @xmath701 is decreasing ( in @xmath19 ) and that @xmath702 for all @xmath150 .
this completes the proof of , hence of theorem [ t : local_global ] . by a more careful analysis
, the condition in theorem [ t : local_global ] can be somewhat relaxed . indeed , continues to hold under the weaker assumption that @xmath703 = 0 , \text { for some $ \delta > 0$}\ ] ] ( using a choice of @xmath683 in depending on @xmath580 )
. however , this will not be of importance , as our proof will show that for the relevant choice of @xmath704 , with @xmath53 arbitrary ( the factor of @xmath705 is just for convenience ) , the above probability decays to @xmath11 as @xmath10 faster than any polynomial , see theorem [ t : giant_comp ] below .
@xmath395 with theorem [ t : local_global ] at hand , in order to prove the lower bound , we will check that condition holds at level @xmath704 , for arbitrary @xmath53 ( the factor @xmath705 is immaterial , and merely reflects the fact that a few more sprinkling operations will be performed on the way ) .
this will involve showing that @xmath94 contains a giant component above this level with sufficiently high probability , cf . .
we will construct this component by gluing together smaller building blocks , so - called substantial components , which , by definition , have cardinality growing like a polynomial in @xmath3 of sufficiently high degree . in this subsection , we show that most vertices in @xmath94 are either neighboring or contained in such a substantial component with high probability , see corollary [ c : number_subst_comps ] below for the precise statement .
the substantial component neighboring a given point in @xmath94 will be built as a connected subset of a ( large ) deterministic tree embedded in @xmath94 and rooted at this point .
in particular , the `` perturbative '' representation of lemma [ l : conds_exps2 ] will enable us to show that , conditionally on an event of high probability , the law of the restriction of @xmath706 to this tree dominates a galton - watson process ( with suitable binomial offspring distribution ) on the same tree , cf .
lemma [ l : subt_comp_dom ] below .
let @xmath53 .
for the sake of clarity , in the remainder of this article , the dependence of constants on @xmath100 will be kept implicit .
we introduce a parameter @xmath707 given @xmath162 , we call @xmath604 a _ substantial component of @xmath116 _ if @xmath604 is a connected subset of @xmath116 containing at least @xmath708 points . by symmetry , it suffices to consider the case @xmath711 .
we begin with some notation .
let @xmath712 .
we assume without loss of generality that @xmath713 , with @xmath714 as defined in ( it suffices to show theorem [ p : subst_comp ] for @xmath715 , since the remaining cases can be taken care of by adapting the constant @xmath420 appearing in ) , and introduce @xmath714 consecutive subsets @xmath716 of @xmath717 .
we will interpret a part of @xmath94 as a tree by considering @xmath718 where @xmath719 denotes the canonical unit vector in the @xmath479-th direction , for @xmath720 .
given @xmath721 , we refer to @xmath722 in the ( unique ) decomposition of @xmath723 , with @xmath724 , @xmath725 , as the generation of @xmath81 , and to the set @xmath726 ( understood as @xmath727 if @xmath728 ) as the children of @xmath81 .
thus , every vertex in @xmath729 in generation smaller than @xmath714 has precisely @xmath730 children , and @xmath714 corresponds to the depth of @xmath729 .
furthermore , it will be convenient to set @xmath731 .
we now show that @xmath732 \geq 1 - c e^{-c ' d^\varepsilon } , \end{split}\ ] ] for all @xmath523 , which implies . to this end , we enumerate the elements of @xmath98 as a sequence @xmath733 , @xmath734 in a hierarchical way , i.e. such that @xmath735 for all @xmath736 , and set @xmath737 , for @xmath734 .
we construct the field @xmath738 using lemma [ l : conds_exps2 ] according to this ordering .
thus , introducing a family @xmath739 of independent random variables with @xmath740 we define the field @xmath738 in terms of @xmath739 as in . moreover , letting @xmath741 we introduce the independent events @xmath742 , for @xmath734 , and define @xmath743 ( with the convention @xmath744 ) .
thus , the events @xmath745 decrease towards @xmath746 . observing that @xmath747 , cf . , and that @xmath748 for all @xmath721 , yields that @xmath749 \leq |\mathbb{t}| \cdot \sup_{x \in \mathbb{t } } \mathbb{p}[\psi_{x } > ( 4c_3')^{-1 } \varepsilon d h_{\text{as}}(d ) ] \leq e^{-c d^2 \log d},\ ] ] for all @xmath1 .
hence , by looking separately at the intersection of the event @xmath750 with @xmath751 and its complement , respectively , we deduce that in order to prove , it suffices to show that @xmath752 \geq 1 - c e^{-c'd^\varepsilon}. \end{split}\ ] ] holds for @xmath523 .
we now show that , conditionally on @xmath751 , the collection of variables @xmath753 stochastically dominates a galton - watson chain on @xmath98 with suitable ( binomial ) offspring distribution .
more precisely , we prove the following .
let @xmath755 , @xmath721 , denote a family of independent bernoulli variables under some auxiliary probability measure @xmath756 such that @xmath757= 1- \mathbf{p}[\mathbf{b}_x = 0]= d^{-(1- \frac{3}{2}\varepsilon)}$ ] for @xmath758 .
then , for all @xmath759 , @xmath760 \ ; \geq_{\textnormal{st . } } \ ; ( \mathbf{b}_x)_{x \in \mathbb{t } } \circ \mathbf{p}.\ ] ] we begin with the following remark .
from , one immediately infers that @xmath761 , for @xmath734 , where @xmath762 with @xmath763 , @xmath764 , as defined in .
we claim that @xmath765 indeed , @xmath766 \cdot \sup_{1 \leq l < n } \sum_{k=1}^l \alpha_{l , k},\ ] ] for all @xmath734 .
since the elements of the tree @xmath98 are enumerated in a hierarchical fashion , @xmath767 for all @xmath768 , and therefore @xmath769 \leq p_{x_{n}}[\widetilde{h}_{b_1(0 , |x_{n-1}|_1 ) } < \infty ] \stackrel{\eqref{return_probas1 } , \eqref{eq : subst_comp_const}}{\leq } c_3'/d\ ] ] ( by construction , @xmath770 , for @xmath771 , and the depth @xmath714 does not depend on @xmath3 , cf . ) .
in particular , this probability is less than @xmath772 , uniformly in @xmath240 for all @xmath768 , whenever @xmath523 . inserting this into and a trivial inductive argument yield .
we proceed with the proof of .
let @xmath773 we first aim at showing that for all @xmath523 , @xmath774 $ , stochastically } \\ & \text{dominates the law of $ ( z_x)_{x\in \mathbb{t}}$ under $ \mathbb{p}[\ , \cdot \ , |m]$}. \end{split}\ ] ] to this end , we claim that for all @xmath523 and @xmath775 , @xmath776 indeed , this is trivial for @xmath241 ( recall that @xmath777 , cf . and @xmath778 . by construction , for all @xmath768 , on the event @xmath779 , @xmath780 \cdot \displaystyle \min_{1\leq
n } \varphi_{x_i } \\ & \geq & \hspace{-1ex } \psi_{x_{n } } + p_{x_{n}}[h_{k_{n } } < \infty ] \cdot \displaystyle \min_{1\leq
i < n } \big [ \sum_{k=1}^i \alpha_{i , k } \cdot \min_{1 \leq l < n } \psi_{x_l } \big ] \\ & \stackrel{\eqref{eq : subst_comp_m}}{\geq } & \hspace{-1ex } \displaystyle h_{\text{as}}(1-\frac{3}{2}\varepsilon ) - \frac{\varepsilon}{4 c_3 ' } d h_{\text{as } } \cdot p_{x_{n}}[h_{k_{n } } < \infty ] \cdot \max_{1\leq i < n } \sum_{k=1}^i \alpha_{i , k}. \end{array}\ ] ] inserting the bounds , into the last line immediately yields that @xmath781 , for @xmath759 , and follows .
but since the sets @xmath745 decrease towards @xmath751 , actually implies that , @xmath782 for all @xmath783 and @xmath759 , or , equivalently , that @xmath784=1 , \text { whenever $ d\geq c$}.\ ] ] by a classical theorem of strassen , see @xcite , the existence of this monotone coupling is equivalent to the asserted stochastic domination in . finally , we explain how follows from . first , notice that by definition of @xmath751 , see , the variables @xmath785 , @xmath758 are still ( conditionally ) independent under @xmath786 $ ] .
next , since @xmath787 \leq g(0)$ ] for all @xmath734 , we can arrange that @xmath788 \leq ( 1 - 2\varepsilon)/(1-\frac{3}{2}\varepsilon)^2 $ ] ( note that this last quantity is always greater than @xmath68 for @xmath712 ) .
moreover , recalling the definition of @xmath789 above , we have that @xmath790 & = & \hspace{-1ex}\mathbb{p}[\psi_{x_n } \geq h_{\text{as}}(1-\frac{3}{2}\varepsilon)| \widehat{m}_n ] \\ & \geq & \hspace{-1ex } \mathbb{p}[\psi_{x_n } \geq h_{\text{as}}(1-\frac{3}{2}\varepsilon ) ] \\
\hspace{-1ex } & \stackrel{\eqref{grv_basic_estimate}}{\geq } & \hspace{-1ex } \displaystyle \frac{c}{h_{\text{as}}(1-\frac{3}{2}\varepsilon ) } \cdot d^{-\tilde{g}_n(1-\frac{3}{2}\varepsilon)^2 } \geq d^{-(1-\frac{3}{2}\varepsilon ) } , \end{array}\ ] ] for all @xmath791 .
a standard coupling then yields that the law of @xmath792 under @xmath793 $ ] dominates the law of @xmath794 under @xmath756 , for all sufficiently large @xmath3 .
together with , this implies , and thus completes the proof of lemma [ l : subt_comp_dom ] .
@xmath395 we continue with the proof of .
we introduce , for @xmath795 , the variable @xmath796 which can be interpreted as the number of `` existing '' children of @xmath81 in the galton - watson chain @xmath794 .
we will need the following estimate .
fix some @xmath795 and @xmath799 .
it suffices to show that @xmath800 \leq e^{-c d^\varepsilon } , \text { when $ d \geq c$,}\ ] ] for then follows with a simple union bound , observing that @xmath801 . since @xmath802 is a sum of independent @xmath803-valued random variables , a classical chernov estimate
( see for example @xcite , ch .
5.3 , p.111 ) gives @xmath804\ , ] \leq e^{-\frac{\delta^2 \mathbf{e}[\mathbf{n}(x)]}{2}}.\ ] ] observing that @xmath805\geq \lfloor \frac{d}{b } \rfloor d^{-(1-\frac{3}{2}\varepsilon ) } \geq c d^{\frac{3}{2}\varepsilon}$ ] , for all @xmath758 , this bound ( with , say , @xmath806 ) is more than enough to deduce .
@xmath395 with lemmas [ l : subt_comp_dom ] and [ l : subst_comp_gwbound ] at hand ,
the claim follows readily .
first , observe that if the event @xmath807 occurs , then the origin has a neighbor in @xmath808 ( in fact even @xmath809 such ) which belongs to a connected component of @xmath810 containing at least @xmath811 points , i.e. , a substantial component of @xmath98 ( recall the definition below ) .
hence , we obtain , for all @xmath759 , applying lemmas [ l : subt_comp_dom ] and [ l : subst_comp_gwbound ] , @xmath812 \\ \stackrel{\eqref{l : subt_comp_dom1}}{\geq } & \mathbf{p}[\text{$0 $ has a child in $ \mathbb{t}_0 $ which is contained in a substantial component}\\ & \hspace{-1ex } \text{of $ \{x\in \mathbb{t } ; \ , \mathbf{b}_x = 1 \}$ } ] \\
\geq & \hspace{-1ex } \mathbf{p}[\mathbf{n}(x ) \geq d^\varepsilon / b , \text { for all $ x \in \mathbb{t}_0 \cap b_1(0,b-1)$ } ] \\ \stackrel{\eqref{eq : subst_comp_gwbound}}{\geq } & \hspace{-1ex } 1 - e^{-cd^\varepsilon } , \end{array}\ ] ] which is .
the proof of theorem [ p : subst_comp ] is now complete . for future reference
, we introduce the random set @xmath813 for arbitrary @xmath712 , and think of the points in @xmath814 as _ bad _ points in @xmath94 . theorem [ p : subst_comp ]
has the following immediate on account of , chebyshev s inequality in the form @xmath818
\leq \lambda^{-1 } \sum_{x\in \mathbb{h } } \mathbb{p}[&x \text { has no neighbor in $ \mathbb{h}$ contained in a } \\ & \text{substantial component of $ \mathbb{h } \cap e_{\varphi}^{\geq h_{\text{as}}(1 - 2\varepsilon)}$]},\end{aligned}\ ] ] with @xmath819 and suitable @xmath820 , readily yields . in this subsection
, we complete the proof of theorem [ t : asymptotics_lb ] .
in what follows , and in accordance with our definition in the paragraph following , given a ( random ) set @xmath821 , we call any connected subset @xmath604 of @xmath116 satisfying @xmath822 a _ giant _ component of @xmath116
. we will prove theorem [ t : asymptotics_lb ] by verifying the finite - size criterion when @xmath823 , for all sufficiently small @xmath53 . in order to deduce that the event @xmath824 appearing in this context ( recall ) occurs with sufficiently high probability
, we will use isoperimetry considerations to patch together the substantial components of @xmath825 we have just constructed , see corollary [ c : number_subst_comps ] above , first to form a giant component in @xmath826 , and then to connect the latter to neighboring giant components within @xmath827 .
this is the object of theorem [ t : giant_comp ] .
the lower bound then follows readily , by virtue of theorem [ t : local_global ] .
fix @xmath830 .
we start by showing @xmath831 \geq 1- c e^{-c'd^{\varepsilon}},\ ] ] for all @xmath18 and @xmath577 . by translation invariance
, it suffices to consider the case @xmath711 .
we denote by @xmath832 ( with @xmath833 if @xmath834 ) the collection of substantial components of @xmath835 ( we assume for sake of definiteness that @xmath836 are enumerated according to a specified procedure , e.g. using the lexicographic order induced by the points closest to the origin in each component ) . with a slight abuse of notation , @xmath837
will also be used to denote the set @xmath838 , but the meaning will always be clear from the context . by definition ,
see , the random sets @xmath839 satisfy @xmath840 , for all @xmath841 , whenever @xmath523 , and @xmath842 for all @xmath843 .
we define the following set of partitions of @xmath837 , @xmath844 since the number @xmath845 of substantial components of @xmath846 is bounded by @xmath847 , when @xmath523 , the ( random ) partition set @xmath848 satisfies @xmath849 ( with the convention that @xmath850 is the whole space @xmath191 whenever @xmath851 ) occurs with sufficiently high probability . in words , @xmath852 is the event that for any partition of the substantial components of @xmath835 into two `` sizeable '' classes ( in the sense that the cardinality of the respective aggregate unions of substantial components is at least @xmath853 ) , one can find a substantial component in each class such that the two are connected in @xmath854 .
let @xmath857 denote the set of singletons and nearest - neighbor edges in @xmath94 , i.e. if @xmath858 , then either @xmath859 for some @xmath860 or @xmath861 with @xmath862 and @xmath142 . for @xmath858
, we define @xmath863 with @xmath864 \varphi_z$ ] , for @xmath865 , as in .
since a non - vanishing contribution to the previous sum arises only from the points in @xmath866 , and since @xmath867 , lemma [ l : zeta_i_bound ] applies ( with @xmath868 and @xmath403 appearing therein both equal to @xmath866 here ) , thus yielding @xmath869 \leq e^{-c d \log d } , \text { for all $ u \in \pi$ , $ d \geq 3$}.\ ] ] we also introduce the ( good ) event @xmath870 ( recall for the definition of @xmath814 ) . on account of and ,
a union bound yields , for all @xmath1 , @xmath871 \leq ce^{-c ' d^{\varepsilon } } + ( 1 + d)|\mathbb{h}| \cdot e^{-c '' d \log d } \leq c'''e^{-c ' d^{\varepsilon}}.\ ] ] it will be convenient to specify configurations of the level set above @xmath872 in the hypercube .
thus , given @xmath873 , we abbreviate @xmath874 , and write @xmath875 \hspace{-1ex } & \leq & \hspace{-1ex } \displaystyle \mathbb{p}[\mathcal{g}^c ] + \sum_{k_+ \subset \mathbb{h } } \mathbb{p}[\mathcal{g } , \ ; \mathcal{h}^c , \ ; \mathcal{c}(k_+ ) ] \\ \hspace{-1ex } & \stackrel{\eqref{eq : giant_event_h}}{\leq } & \hspace{-1ex } \mathbb{p}[\mathcal{g}^c ] + \displaystyle \sum_{k_+ \subset \mathbb{h } } |\mathbf{p}(s_{k_+})| \sup_{\{k , k'\ } \in \mathbf{p}(s_{k_+ } ) } \mathbb{p}[\mathcal{g } , k \nleftrightarrow k ' \text { in $ e_\varphi^{\geq h_\text{as}(1 - 5\varepsilon)}$ } , \ , \mathcal{c}(k_+ ) ] \\ \hspace{-1ex } & \stackrel{\eqref{eq : giant_partitions}}{\leq } & \hspace{-1ex } \mathbb{p}[\mathcal{g}^c ] + \displaystyle 2^{d^{-10}|\mathbb{h}| }
\sup_{k_+ \subset \mathbb{h } } \ ; \sup_{\{k , k'\ } \in \mathbf{p}(s_{k_+ } ) } \mathbb{p}[\mathcal{g } , k \nleftrightarrow k ' \text { in $ e_\varphi^{\geq h_\text{as}(1 - 5\varepsilon)}$ } \ , | \ , \mathcal{c}(k_+ ) ] , \end{array}\ ] ] for all @xmath523 , where the set @xmath876 in the second and third line refers to the ( deterministic ) family of substantial components associated to the configuration @xmath877 . in order to bound the conditional probability appearing on the right - hand side of
, we rely on isoperimetry considerations for subsets of the hypercube by bollobs and leader @xcite .
thus , let @xmath873 be such that @xmath878 is non - empty and @xmath879 be fixed .
first , observe that @xmath880 is a deterministic set under @xmath881 $ ] . by construction , @xmath882 ( recall that @xmath883 denotes the @xmath82-closure of @xmath116 in @xmath94 ) .
moreover , by definition of @xmath878 , and on the event @xmath884 , cf .
, latexmath:[\[\label{eq : isoperimetry_h_cond2 } on account of and , lemma 4.3 in @xcite ( itself a consequence of corollary 4 in @xcite ) yields that for all @xmath523 , @xmath886 among the family @xmath887 , @xmath888 , we may select @xmath889 sets @xmath890 satisfying the additional assumption @xmath891 now , if the event @xmath892 occurs ( conditionally on @xmath877 ) , then by construction , the field value in at least one of the sites in each set @xmath893 can not exceed @xmath894 , for otherwise @xmath893 forms a path connecting @xmath116 to @xmath141 in the level set @xmath854 , cf . . thus , setting @xmath895 , for @xmath896 , and @xmath897 , which contains @xmath884 , see , we obtain , for all @xmath523 , @xmath898 \\ & \qquad \leq \mathbb{p } [ \widetilde{\mathcal{g } } , \mathcal{f}_k , 1\leq k \leq m \ , | \ , \mathcal{c}(k_+ ) ] \\ & \qquad = \big ( \prod_{k=1}^m \mathbb{p } [ \mathcal{f}_k \ , | \ ,
\widetilde{\mathcal{g } } , \,\mathcal{f}_l , k < l \leq m , \ , \mathcal{c}(k_+ ) ] \big ) \cdot \mathbb{p}[\widetilde{\mathcal{g } } \ , | \ , \mathcal{c}(k_+ ) ] .
\end{split}\ ] ] next , we consider a single factor @xmath899 $ ] in this product , and show that it does nt converge too rapidly ( in terms of @xmath3 ) to @xmath68 as @xmath71 . by definition of @xmath900 , cf . ,
and on account of , the event @xmath901 is measurable with respect to the @xmath15-algebra generated by @xmath902 , @xmath903 , ( recall that @xmath904 denotes the @xmath82-closure of @xmath94 ) , hence for every @xmath896 , @xmath905 ( here and in what follows , we use the shorthand @xmath906 , for any borel set @xmath907 and @xmath162 ) .
since all elements @xmath908 lie on the exterior boundary of a substantial component of @xmath825 , we necessarily have that @xmath909 .
thus , for all @xmath896 , @xmath910 = 1- \frac{\mathbb{p } [ \ , \mathbb{p } [ \varphi_{|_{u_k } } < h_{\text{as}}(1 - 2\varepsilon ) , \mathcal{f}_k^c\ , | \ , ( \varphi_x)_{x \in \overline{\mathbb{h } } \setminus u_k } ] , \ , \mathcal{d}_k]}{\mathbb{p } [ \ , \mathbb{p } [ \varphi_{|_{u_k } } < h_{\text{as}}(1 - 2\varepsilon ) \ , | \ , ( \varphi_x)_{x \in \overline{\mathbb{h } } \setminus u_k } ] , \ , \mathcal{d}_k]}.\ ] ] recall from that @xmath911 = \widetilde{\mathbb{p}}[\widetilde{\varphi}^{k } + \mu^{u_k^c } \in \cdot \,]$ ] , where we have abbreviated @xmath912 by @xmath913 , which we view as a 2-dimensional gaussian vector with covariance @xmath914 under @xmath227 .
thus , by definition of the event @xmath900 in , we obtain , for all @xmath896 , all _ decreasing _ events @xmath915 and @xmath7 , that @xmath13-a.s . on @xmath900 , @xmath916 \leq \widetilde{\mathbb{p}}[a^h(\widetilde{\varphi}^{k}+ |\mu^{u_k^c}| ) ] \leq \mathbb{p}[a^h \ , | \ , ( \varphi_x ) _ { x \in \mathbb{z}^d \setminus u_k } ] \leq \widetilde{\mathbb{p}}[a^h(\widetilde{\varphi}^{k}- |\mu^{u_k^c}| ) ] \leq \widetilde{\mathbb{p}}[a^{h + \varepsilon h_{\text{as}}}(\widetilde{\varphi}^{k } ) ] \end{split}\ ] ] ( cf . for notation ) .
we apply this separately to the numerator and denominator above . to do this , we note that @xmath917 , for all @xmath896 , and that , despite not being decreasing , the event @xmath918 appearing in the numerator can be rewritten using @xmath919 , for all @xmath920 ( the events appearing on the right - hand side are both decreasing ) .
all in all , we infer , setting @xmath921 and @xmath922 , that @xmath923 & \leq 1-\frac{\widetilde{\mathbb{p}}[\widetilde{\varphi}^{k } \in [ a , b)]\cdot \mathbb{p}[\mathcal{d}_k ] } { \widetilde{\mathbb{p}}[\widetilde{\varphi}^{k } < h_{\text{as}}(1-\varepsilon ) ] \cdot \mathbb{p}[\mathcal{d}_k ] } \\ & \leq 1 - \widetilde{\mathbb{p}}[\widetilde{\varphi}^{k } \in [ a , b ) ] , \\ \end{split}\ ] ] for all @xmath896 ( in writing @xmath924 $ ] , we obviously mean that both components of @xmath925 should lie in @xmath926 ) . moreover , letting @xmath927 , and denoting by @xmath928 the distribution function of a standard gaussian variable , we can bound this probability as @xmath929 & = \widetilde{\mathbb{p}}\big [ \ , \widetilde{\mathbb{p}}[a<\widetilde{\varphi}^{k}_{y_k } < b \ ,
| \ , \widetilde{\varphi}^{k}_{z_k } ] , \,a<\widetilde{\varphi}^{k}_{z_k } < b \big ] \\ & = \widetilde{\mathbb{e}}\big[\big(\phi\big(b - \frac{1}{2d}\widetilde{\varphi}^{k}_{z_k } \big ) -
\phi \big(a - \frac{1}{2d } \widetilde{\varphi}^{k}_{z_k}\big)\big)1\{a<\widetilde{\varphi}^{k}_{z_k } < b \ }
\big]\\ & \geq ( \phi(b)-\phi(a ) ) \cdot \widetilde{\mathbb{p } } [ a<\widetilde{\varphi}^{k}_{z_k } < b ] \geq
c \big((b - a ) e^{- { b}^2/ 2 } \big)^2 \geq c ' h_{\text{as}}^2 \ , d^{-2 } \geq d^{-2 } , \end{split}\ ] ] for all @xmath759 and @xmath896 ( here , the first inequality in the third line follows because the linear shift @xmath930 produced by conditioning on @xmath931 is in fact between , say , @xmath11 and @xmath68 , when @xmath3 is large enough , since @xmath932 , while @xmath933 as @xmath71 ) . inserting this bound into , and in view of , , we obtain , for all @xmath523 , @xmath873 and @xmath934 , @xmath935 \leq ( 1-d^{-2})^{c d^{-7 } |\mathbb{h}| } \leq 2^{-cd^{-9}|\mathbb{h}|},\ ] ] where we have also used that @xmath319 as appearing in is bounded from below by @xmath936 , cf . above .
finally , going back to , and on account of , we see that @xmath937 \leq ce^{-c ' d^\varepsilon}$ ] , for all @xmath938 , and thus for all @xmath1 by adjusting the constant @xmath420 .
this completes the proof of lemma [ l : giant_event_h ] .
@xmath395 as we will now see , follows from lemma [ l : giant_event_h ] by virtue of a counting argument .
specifically , it suffices to show that for all @xmath759 , @xmath939 which , on account of and , implies .
we now show .
recall that @xmath836 are the ( ordered ) substantial components of @xmath940 , and denote their union by @xmath837 . by definition of @xmath814 , cf . , @xmath941 and therefore certainly @xmath942 on the set @xmath943 defined as @xmath944 if and only if @xmath945 , and let @xmath946 be the ( random ) partition of @xmath943 induced by its equivalence classes .
denoting by @xmath947 the equivalence class maximizing the quantity @xmath948 , for @xmath949 , we claim that on the event @xmath950 and for @xmath759 , @xmath951 indeed , if did not hold , i.e. @xmath952 for all @xmath949 , on account of , one could easily construct a set @xmath953 obtained as union of some of the sets @xmath949 with the property @xmath954 . but by , this would yield , setting @xmath955 @xmath956 , that @xmath957 . because @xmath852 is assumed to occur
, this would imply that one could find substantial components @xmath837 and @xmath958 belonging to @xmath116 and @xmath141 respectively , such that @xmath959 , thus contradicting the definition of @xmath960 ( observe that by construction , @xmath837 and @xmath958 belong to different equivalence classes ) . hence holds .
similarly , implies , for the joint occurrence of @xmath852 , and @xmath961 would also violate the definition of @xmath960 . having established and , we observe that on @xmath950 and for @xmath759 , @xmath962 thus , on the event @xmath950 and for @xmath759 , the set @xmath963 forms a connected component of @xmath964 with @xmath965 . by definition , cf . , it is therefore a giant component of @xmath964 .
this yields , which completes the proof of ( for @xmath966 , and thus all @xmath587 by translation invariance ) . with at hand
, we proceed with the proof of , which is similar , but simpler . for @xmath587 , we denote by @xmath615 the giant component of @xmath967 , cf .
, whenever it exists . by , @xmath968 \leq \sum_{\substack{x \in \mathbb{z}^2:}{|x|_1 \leq 1 } } & ( \mathbb{p}[\text{$c_{x}$ does not exist } ] \\ & + \mathbb{p}[\text{$c_0 $ , $ c_{x}$ exist and } c_0 \nleftrightarrow c_{x } \text { in $ e_\varphi^{\geq h_{\text{as}}(1 - 8\varepsilon)}$ } ] ) .
\end{split}\ ] ] on account of , it thus suffices to prove that for all @xmath1 and @xmath587 with @xmath969 , @xmath970 \leq ce^{-c'd^\varepsilon}.\ ] ] for arbitrary @xmath587 with @xmath969 , let @xmath971 denote the singletons and nearest - neighbor edges in @xmath972 , and denote by @xmath973 , with @xmath974 as defined in . as in
, we have @xmath975 \leq c e^{-c ' d^{\varepsilon}}$ ] .
thus , we see that follows at once if we show that @xmath976 \leq ce^{-c'd^\varepsilon},\ ] ] for all @xmath18 , where @xmath977 specifies the configuration of @xmath978 in @xmath979 .
fix @xmath577 with @xmath969 and @xmath980 . letting @xmath981 , with @xmath982 , @xmath983 , cf .
, observe that , whenever @xmath984 and @xmath985 exist , the ( disjoint ) sets @xmath986 and @xmath987 each contain at most @xmath988 elements .
moreover , the joint occurrence of @xmath984 and @xmath985 implies immediately that for all @xmath1 , @xmath989 for @xmath893 not to form a path joining @xmath984 and @xmath985 in the level set @xmath990 , at least one of the two sites in @xmath893 must have a field value smaller than @xmath991 , i.e. setting @xmath992 , we obtain @xmath993 \\ & \qquad \leq \mathbb{p}[\hat{\mathcal{g } } , \hat{\mathcal{f}}_k , 1 \leq k \leq m \ ,
| \ , \hat{\mathcal{c}}(k_+)].\end{aligned}\ ] ] now , on account of , an analysis similar to the one below yields that this last quantity is bounded from above by @xmath994 , for all @xmath759 , which is more than enough to deduce .
this completes the proof of theorem [ t : giant_comp ] .
let @xmath712 .
it follows from theorem [ t : giant_comp ] that @xmath995 \geq 1 - c e^{-c'd^{\varepsilon}}$ ] , for all @xmath18 .
in particular , @xmath996 = 0 $ ] , and theorem [ t : local_global ] yields that @xmath997 , for all @xmath998 .
the claim follows .
moreover , by construction , cf .
, percolation then occurs in @xmath592 , which implies .
@xmath395 \1 ) a thorough review of the proofs reveals that the long - range dependence present in the model is a serious impediment and considerably hinders any efforts to obtain more precise results than the leading exponential order given by theorem [ t : asymptotics_density ] .
\2 ) the heuristic paradigm requiring for this model to exhibit strong similarities to a corresponding one on the @xmath8-regular tree is evidently inherent to many of the above proofs .
specifically , this behavior manifests itself when we investigate the local connectivity of the level set , essentially because the `` tree - like '' structure of the lattice determines the behavior of the random walk at short range , see the proofs of lemma [ l : zeta_i_bound ] and theorem [ t : local_cb ] ( which crucially rely on ) for the upper bound , and the proof of theorem [ p : subst_comp ] for the lower bound .
accordingly , we would like to compare to the critical density of the same model on this tree .
we hope to come back to this point in the future .
@xmath395 _ note added in proof : _ a recent preprint of lupu @xcite , see theorem 3 therein , shows that for any @xmath1001 , there exists a coupling between random interlacements @xmath1002 at level @xmath1003 , see @xcite for the definition , and the free field @xmath19 such that @xmath1004 , where @xmath1005 is the so - called vacant set ( at level @xmath1003 ) .
in particular , this readily implies that @xmath1006 , where @xmath1007 denotes the critical parameter for interlacement percolation .
together with our lower bound , this shows that @xmath1008 , which matches precisely the lower bound obtained by sznitman in @xcite , and thus provides an alternative proof of this result .
similarly , the inequality @xmath1009 and sznitman s asymptotic upper bound on @xmath1010 , cf .
@xcite , theorem 0.1 and remark 4.1 , yield another proof of .
* acknowledgements . * a. d. and p .- f .
r. would like to thank the forschungsinstitut fr mathematik at eth zrich and the columbia university mathematics department , respectively , for their hospitality and financial support .
we are grateful to r. rosenthal , a. sapozhnikov and a .- s .
sznitman for their comments on an earlier version of this paper .
we also thank an anonymous referee for various remarks on the manuscript .
georgii , o. hggstrm and c. maes . the random geometry of equilibrium phases .
_ phase transitions and critical phenomena _
( edited by c. domb and j. lebowitz ) , * 14 * : pp .
1142 . academic press ( 2001 ) .
h. kesten .
asymptotics in high dimension for percolation . in disorder in physical systems , a volume in honor of john hammersley on the occasion of his 70th birthday .
_ oxford sci .
_ : pp.219240 ( 1990 ) .
|
we consider the gaussian free field on @xmath0 , @xmath1 , and prove that the critical density for percolation of its level sets behaves like @xmath2 as @xmath3 tends to infinity .
our proof gives the principal asymptotic behavior of the corresponding critical level @xmath4 .
moreover , it shows that a related parameter @xmath5 introduced by rodriguez and sznitman in @xcite is in fact asymptotically equivalent to @xmath4 .
alexander drewitz and pierre - franois rodriguez
|
mounting evidence indicates that massive molecular cloud cores possess properties which are not adequately explained by models employing an isothermal equation of state . for example , surveys of prestellar cores in a variety of star - forming complexes show that intermediate and high mass cores have nonthermal line widths which are significantly greater than their thermal line widths ( @xcite ) .
models of the collapse of isothermal molecular cloud cores exhibit a number of features which are not borne out by observations , such as a time - invariant rate of accretion onto the central object ( @xcite ) and highly supersonic infall ( @xcite ) . to address some of the shortcomings of isothermal models
, @xcite advanced a model for molecular cloud core collapse which accounts for some degree of turbulent support via a phenomenological equation of state .
this logatropic equation of state reads + @xmath1 where the subscript `` c '' denotes `` central '' values .
the constant @xmath2 is an adjustable parameter whose value for a range of molecular cloud core masses was determined empirically by @xcite to be @xmath3 .
the logatropic equation of state admits two equilibrium solutions .
the features of each solution are developed in @xcite .
the `` singular '' equilibrium solution is analogous to the singular isothermal sphere ( sis ) of @xcite , except that its density profile is @xmath4 .
the `` nonsingular '' solution is analogous to the bonnor - ebert sphere , in that it is a pressure - truncated equilibrium solution with a finite central density and density profile in its outer regions which matches that of the singular solution .
the nonsingular logatrope has the potential to be as useful a model for massive star - forming cores as the bonnor - ebert sphere is for low mass star - forming cores ( see , for example , @xcite , who showed that a bonnor - ebert sphere is an excellent fit to the structure of dark cloud b68 ) .
hence , we undertook a numerical study of the collapse of a nonsingular logatropic sphere from an initially near - equilibrium state until most of its mass had been accreted onto the central condensation .
our fully 3d simulations were conducted using the zeus - mp parallelized hydrodynamics code .
we used a 256@xmath5 cartesian grid with a `` sink cell '' at its centre .
mass falling into the sink cell was presumed to have accreted onto the central protostar and was represented thereafter by a point mass at the grid centre .
the initial configuration for each of our simulations was a spherical molecular cloud core in near hydrodynamic equilibrium , surrounded by an external medium of low and constant pressure .
we conducted extensive tests of our numerical method using the collapse of the singular logatrope ( @xcite ) as our test case .
excellent agreement between the analytic and numerical solutions was found .
( see @xcite for a full description of our numerical method ) .
the most interesting of our results from the collapse of the nonsingular logatrope are presented in figures [ f9]-[f11b ] .
figures [ f9 ] and [ f10 ] show the evolution of the density and infall speed profiles of a collapsing nonsingular logatrope prior to the formation of a singular density profile .
the model has been scaled such that the total mass of the core is 1 m@xmath6 .
the total core mass is determined by the choice of central temperature and external pressure our choices of @xmath7 k and @xmath8 k@xmath9 result in a core of total mass 1 m@xmath6 , but the results can be rescaled and applied to models of higher mass ( for example , by assuming a higher confining pressure ) . indeed , observational evidence has emerged which indicates that the confining pressures on molecular cloud cores are one or two orders of magnitude higher than we assumed ( @xcite ) .
perhaps one of the most notable features of figure [ f9 ] is the behaviour of the density profile at late times . at the moment when the density becomes singular ( labeled @xmath10 ) , the density in the inner parts of the core scales as @xmath11
this is the density profile expected for a collapsing singular logatropic sphere , as well as for a collapsing sis .
this implies that , whereas observations of r@xmath12 density profiles in molecular cloud cores have traditionally been interpreted as evidence for isothermal collapse , this should no longer be the case .
even models which incorporate turbulent support will show such a density profile . because of their extra nonthermal support , nonsingular logatropic spheres collapse more `` gently '' than do their isothermal counterparts . as can be seen by comparing our figure [ f10 ] with figure 1a of @xcite ,
the logatropic collapse is more gentle than that of the bonnor - ebert sphere .
@xcite find that , at the time of singularity formation , 44% of the mass of a bonnor - ebert sphere is in supersonic infall .
we find that less than 5.5% of a nonsingular logatrope is in supersonic infall at the same point in its evolution .
figure [ f11b ] shows the rate of accretion onto the central object in collapsing nonsingular logatropes of masses 1 , 2.5 , and 5 m@xmath6 .
the reader should bear in mind that these are fiducial masses only the models can be rescaled to produce much more massive cores and hence higher accretion rates by increasing the confining pressure .
each curve shows the evolution of the accretion rate from the near - equilibrium starting point and until 95% , 91% , and 29% of the total available mass has been accreted ( for the 1 , 2.5 , and 5 m@xmath6 cases respectively ) .
the accretion profiles shown in figure [ f11b ] bear several important similarities to what is known observationally about accretion in young stellar objects .
a simple picture of the evolution of an accreting molecular cloud core suggests an accretion history of the following form : accretion proceeds slowly through the pre - protostellar ( ppc ) phase , increases markedly at the formation of a central source ( to power the substantial outflows seen in class o sources ) , and finally declines and dissipates as the core passes into the class i stage .
@xcite mark the transition between the class o and class i phase at the point where the mass accreted onto the central object is roughly equal to that remaining in the envelope .
our results show a similar transition at the same point .
as indicated for the 1 m@xmath6 curve in figure [ f11b ] , there is an initial period of slow accretion , which we liken to the ppc stage .
as soon as the density profile of the core becomes singular ( @xmath10 ) , which can be thought of as the moment of protostar formation , the accretion rate increases markedly , launching a phase of vigorous accretion sufficient to power an outflow .
we liken this to the class 0 phase .
once half of the available mass has been accreted , the accretion rate declines steadily to near - zero .
we suggest that our simulations of the collapse of logatropic molecular cloud cores bear significant similarities to the observations .
they offer an alternative method for modelling the structure of intermediate and higher mass molecular cloud cores .
cores which exhibit density power - law indices of between 1 and 1.5 ( see , for example , @xcite ) should be pursued as possible candidate nonsingular logatropes .
our results bolster the argument that models of collapsing protostellar cores _ must _ account for some degree of nonthermal support in order to reproduce the observations .
we are grateful to the organizers of sfchem 2002 for allowing us to present our work .
the work of m. a. r. was supported in part by a scholarship from the natural sciences and engineering research council of canada ( nserc ) .
the work of r. e. p. was supported by nserc .
j. w. is a sharcnet research associate .
|
we have performed fully 3-d simulations of the collapse of molecular cloud cores which obey the logatropic equation of state . by following the collapse of these cores from states of near hydrodynamic equilibrium
, we are able to produce accretion histories which closely resemble those of observed cores .
the accretion proceeds in four distinct stages : an initial period of very slow accretion ; a period of vigorous accretion following the development of a central density singularity , with @xmath0 , as predicted by self - similar models ; a period of relatively stable , vigorous accretion ; and finally a gradual decrease in the accretion rate once about 50% of the available mass of the molecular cloud core has been accreted .
these results may explain the accretion histories of cores as they pass through the pre - protostellar , class 0 , and class i stages .
|
the vela pulsar ( psr b0833 - 45 ) is the strongest @xmath0-ray source in the sky , but it is one of the most difficult pulsars to detect at x - ray energies .
this is in part because it is embedded in a very bright x - ray synchrotron nebula providing a large unpulsed background , but also because its pulsed x - ray emission is comparatively weak .
the first detection of pulsed emission at x - ray energies was made by rosat in the 0.1 - 2 kev band ( ogelman 1993 ) , and the spectrum is consistent with a blackbody .
we detected the pulsar for the first time in hard x - rays ( 2 - 30 kev ) during a 93 ks rxte cycle 1 observation ( strickman , harding & de jager 1999 [ shd99 ] ) and the pulse profile shows two peaks .
the first rxte peak is closely aligned with the first egret @xmath0-ray peak , but the second peak has an energy - dependent phase and is aligned with the second egret peak only at the highest energy ( 16 - 30 kev ) . in our lowest energy rxte band
( 2 - 8 kev ) the second peak is roughly aligned with the second peak of the optical profile ( gouiffes 1998 ) .
the average pulse spectrum joins smoothly with the high - energy spectrum of osse , comptel and egret , although the spectrum of the first peak is significantly harder than that of the second peak , which appeared to be consistent with an extrapolation to the optical flux points .
we ( shd99 ) suggested that the second rxte peak was a blend of separate hard and soft components . in this paper , we report preliminary results of our analysis of a 300 ks rxte cycle 3 observation which confirms this picture .
-0.7 truecm the cycle 3 observations were carried out during april / may and july / august 1999 with a good exposure of 274 ks . for this analysis we epoch - folded the pca data in goodxenon event - by - event mode at the pulsar period using the princeton pulsar database ( arzoumian et al .
1992 ) , to obtain the energy - dependent light curves which we have summed into three broad energy bands shown in figure 1 .
to separate the various possible components of the light curve and compute individual component spectra , we fit a five peak sinusoid model with peaks of the form @xmath1|^{\xi(i ) } , \\ \nonumber \\
\xi(i ) & = & -{0.693\over \log[\cos({\pi\over 2}w(i ) ) ] } \nonumber\end{aligned}\ ] ] to the data in 94 energy channels ( roughly spanning 2 - 30 kev ) , where @xmath2 , @xmath3 and @xmath4 are the amplitude , center phase and width of peak @xmath5 respectively .
the model also includes a constant background level .
we found that @xmath3 and @xmath4 did not vary significantly with energy , except for the phase of peak 3 which has a modest variation ( see table 1 ) , so they were fixed while the values of @xmath2 as a function of energy were determined .
the counts for each peak were then found by integrating each sinusoid curve over phase .
we could then obtain the photon spectrum of each peak separately by fitting a power law with photoelectric absorption of fixed column density @xmath6 .
.center phase and spectrum of light curve peaks [ cols= " < , < , < , > " , ] -1.0 truecm
the rxte light curves in figure 1 show a narrow peak ( peak 1 ) at the phase of the egret first peak and a second peak that is now clearly seen to be a blend of two components .
the first of these components , which we call peak 2 , becomes dominant in the 2 - 8 kev band and is in phase with the optical second peak .
the second component , peak 3 , is harder and is in phase with the egret second peak .
there are two other statistically significant peaks that appear in the rxte light curve : a peak ( peak 5 ) at the radio phase ( 0.0 ) and a weaker peak ( peak4 ) leading the radio peak .
note that there is also a peak in the optical light curve at the radio phase .
the energy spectra have been computed using the sinusoid model fits to each peak .
the resulting power law spectral indices and goodness of fit for each peak are listed in table 1 . the peaks in phase with the egret peaks ( peak 1 and peak 3 ) both have hard , but significantly different , spectra .
the spectra of peak 2 and peak 5 are much softer and both their flux levels and indices are consistent within the uncertainties .
interestingly , an extrapolation of the peak 2 spectrum falls near the optical flux points ( nasuti et al .
these cycle 3 observations have thus independently confirmed the multicomponent nature of pulsed emission from the vela pulsar in the energy range 2 - 30 kev suggested by the cycle 1 observations . with the improved statistics of the cycle 3 data
, we have been able to separate the broad second peak in the rxte x - ray light curve into soft ( peak 2 ) and hard ( peak3 ) spectral components which maintain their phase integrity throughout the rxte energy range .
in addition , we have discovered a new feature in the rxte light curve : a peak ( peak 5 ) at the phase of the radio pulse with an extremely soft spectrum .
there is , in addition , significant emission leading the radio phase ( peak 4 ) .
peaks 1 and 3 make up the hard spectral component whose light curve peaks are in phase with those of the gamma - ray light curve , and whose spectrum smoothly connects to the 100 kev - 5 gev spectrum .
peaks 2 and 5 , whose phases match those of the second optical and radio pulses respectively , make up the soft component .
their spectra , consistent with each other in both flux and spectral index , extrapolate to the optical flux points .
although the rxte hard component spectrum connects to the @xmath0-ray spectrum , the x - ray spectrum is harder , requiring a break around 100 kev .
such a break at the local cyclotron energy , blueshifted by the parallel momentum of the pairs , is predicted by the polar cap cascade model ( harding & daugherty 1999 ) .
the rxte soft component may be either inverse compton scattering radiation of pairs in polar cap cascades ( zhang & harding 1999 ) or synchrotron radiation of backflowing particles from the outer gap ( cheng & zhang 1999 ) .
arzoumian , z. , nice , d. & taylor , j. h. 1992 , gro / radio timing data base , princeton univ .
cheng , k. s. & zhang , l. 1999 , apj , 515 , 337 .
gouiffes , c. 1998 , in neutron stars and pulsars , ed .
n. shibazaki , n. kawai , s. shibata & t. kifune ( univ .
press : toyko ) , p. 363 .
harding , a. k. & daugherty , j. k. 1999 , proc . of 3rd integral workshop , in press .
kanbach , g. et al .
1994 , a & a , 289 , 855 .
nasuti , f. p. et al .
1997 , a & a , 323 , 839 .
ogelman , h. 1993 , in lives of neutron stars , ed .
m. a. alpar et al .
( dordrecht : kluwer ) , p. 101 .
strickman , m. s. , harding , a. k. & dejager , o. c. 1999 , apj , 524 . 373 .
zhang , b. & harding , a. k. 1999 , apj , in press .
|
we report on our analysis of a 300 ks observation of the vela pulsar with the rossi x - ray timing explorer ( rxte ) .
the double - peaked , pulsed emission at 2 - 30 kev , which we had previously detected during a 93 ks observation , is confirmed with much improved statistics .
there is now clear evidence , both in the spectrum and the light curve , that the emission in the rxte band is a blend of two separate components .
the spectrum of the harder component connects smoothly with the osse , comptel and egret spectrum and the peaks in the light curve are in phase coincidence with those of the high - energy light curve .
the spectrum of the softer component is consistent with an extrapolation to the pulsed optical flux , and the second rxte pulse is in phase coincidence with the second optical peak .
in addition , we see a peak in the 2 - 8 kev rxte pulse profile at the radio phase . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
no known solution for a coupled set of duffing oscillators exists , with the exception of specific cases .
the duffing lattice is perhaps more well known as the fpu @xmath0-system , one of the systems studied by fermi , pasta , and ulam in the @xmath2 @xcite to numerically investigate the approach to thermal equilibrium in nonlinear systems .
fpu found that the expected equipartition of energy among the linear normal modes did not occur , and that periodic , non - ergodic motion persisted even in the presence of the nonlinear mixing terms .
subsequent investigations , both numerical and analytic , have shown that different parameter regions specified by the total energy and the strength of the nonlinear terms as specified by the parameter @xmath1 lead to markedly different types of behaviour @xcite , with the expected ergodicity emerging with increasing energy and @xmath1 .
but the observation of the periodic motion in the orginal fpu @xmath0-system computer experiments lends credence to the possibility of finding acceptable pertubative solutions in certain parameter regions . that no exact solution exsists for a large @xmath3 - body lattice modulo some exceptional cases @xcite motivates us to develop pertubative solutions to systems of duffing oscillators .
several pertubation schemes as applied to the duffing oscillator have been investigated @xcite , but the methods are applicable only in the weakly nonlinear regime and also require the time @xmath4 to be small in order to be rigorously trustworthy . we therefore apply a different perturbative procedure called the lindstedt series .
we note that similar perturbation schemes that avoid secular terms have been proposed @xcite in the past . in particular , we note that the method presented here is a more general variation of the so - called shifted frequency perturbation scheme , as described in @xcite .
the lindstedt method is a systematic procedure to compute formal power series expansions of quasi - periodic solutions @xcite .
we consider an anharmonic bath of @xmath3 degrees of freedom in one dimension with fixed endoints @xmath5 , ( we consider these boundary conditions as opposed to periodic boundary conditions so as to avoid zero modes , which are problematic when using the lattice as a bath coupled to a test particle in the standard derivation of the generalized langevin equation ) as described by the hamiltonian : @xmath6 the normal coordinates @xmath7 are related to the site coordinates @xmath8 through the transformations @xmath9 and @xmath10 where @xmath11 by means of this transformation , the harmonic part of the hamiltonian is decomposed into a sum of independent normal modes : @xmath12 where @xmath13 we have the following compact expresssion for @xmath14 expressed in terms of the linear normal modes @xcite : @xmath15 where the coupling coefficients @xmath16 are given by @xmath17 with @xmath18 defined by @xmath19
we seek solutions to the equations of motion for @xmath20 , @xmath21 where @xmath1 is a small parameter @xmath22 .
a hallmark of nonlinear , periodic behaviour is the amplitude dependence of the frequency . to capture this relationship ,
for each normal mode @xmath23 we define a new stretched temporal variable @xmath24 where to first order in @xmath1 we define @xmath25 and the quantity @xmath26 is to be determined .
the lindstedt method can in principle be expanded to all powers in @xmath1 @xcite , although the convergence properties for arbitrary @xmath3 and @xmath1 are not known . in this work
, we will only retain expressions up to first order in @xmath1 in the quantities @xmath27 .
the lindstedt method then makes the assumption that we can express @xmath28 in the form @xmath29 the zeroth order equations are @xmath30 : @xmath31 with solutions @xmath32 the order @xmath1 equations are given by @xmath33 or , in terms of the original time variable @xmath4 : @xmath34 in order to apply the lindstedt method , we need to identify all of the potential resonant driving terms @xcite . terms with the arguments @xmath35 will never provide resonance terms . using the symmetry properties of the coupling coefficients @xmath16
, we arrive at the following resonant forcing terms in the equations of motion for the @xmath36 : @xmath37 \nonumber\\ & & - \frac{\beta_{k}^{2 } \omega^{4}_{k}}{8(n+1 ) } c_{kkkk } \times \bigg\ { \left[3q_{k}^{3}(0)+3\left(\frac{\dot{q}_{k}(0)}{\beta_{k } \omega_{k}}\right)^{2 } q_{k}(0 ) \right ] \cos(\omega_{k } \beta_{k } t ) \nonumber\\ & & \qquad\qquad + \bigg [ 3\left(\frac{\dot{q}_{k}(0)}{\beta_{k } \omega_{k}}\right)^{3 } + 3q_{k}^{2}(0 ) \frac{\dot{q}_{k}(0)}{\beta_{k } \omega_{k } } \bigg ] \sin(\omega_{k } \beta_{k } t ) \bigg\ } \nonumber\\ & & -\frac{3 \beta_{k}^{2 } \omega_{k}^{2}}{8(n+1 ) } \sum_{m \neq k}^{n } \omega_{m}^{2 } c_{kkmm } \bigg\ { \bigg[2q_{k}(0)q_{m}^{2}(0)+2q_{k}(0)\left(\frac{\dot{q}_{m}(0)}{\beta_{m } \omega_{m}}\right)^{2 } \bigg ] \cos(\omega_{k } \beta_{k } t ) \nonumber\\ & & \qquad\qquad + \bigg[2q_{m}^{2}(0)\frac{\dot{q}_{k}(0)}{\beta_{k } \omega_{k } } + 2\frac{\dot{q}_{k}(0)}{\beta_{k } \omega_{k } } \left(\frac{\dot{q}_{m}(0)}{\beta_{m } \omega_{m}}\right)^{2 } \bigg ] \sin(\beta_{k } \omega_{k } t ) \bigg\ } \ , .
\nonumber\\ & & \qquad\qquad + \mbox{non - resonant } \ , \ , \mbox{terms}. \label{wow2}\end{aligned}\ ] ] setting to zero the coefficients mutiplying @xmath38 and @xmath39 yields the following system of coupled nonlinear algebraic equations for the parameters @xmath40 : @xmath41 in principle , one may solve this system of equations for the set @xmath40 .
having obtained @xmath40 , one then substitutes the expressions given by eqn .
( [ zerosol ] ) into the right hand side of eqn .
( [ wow ] ) and uses the harmonic oscillator green s function to solve for the set @xmath42 . in the present work , for the sake of simplicity .
we restrict our attention to corrections to the harmonic frequencies @xmath43 that are first order in @xmath1 ; i.e. , we consider @xmath25 where @xmath26 is independent of @xmath1 .
if one wishes to compute first order corrections to the @xmath43 , then @xmath1 may be set to zero in the right hand side of eqn .
( [ rhokay ] ) , and one obtains : @xmath44 and @xmath25 . therefore the zeroth order normal mode @xmath45 is modified as follows : @xmath46 where @xmath47 with this expression for the parameters @xmath48 , where @xmath26 is independent of @xmath1 , we have not removed resonant driving terms from eqn .
( [ wow ] ) , but rather from the evolution equation for @xmath49 obtained by setting all of the set @xmath27 equal to unity in eqn .
( [ wow ] ) . with the @xmath40 given by eqn .
( [ hewrho ] ) , we have removed resonant driving terms from the equations : @xmath50 therefore , consistency requires that we solve the eqns .
( [ wow2 ] ) to determine the @xmath49 . having solved explicitly for the set @xmath51 in terms of initial conditions , we may now solve for the set @xmath52 .
the equation of motion for @xmath49 then becomes @xmath53 where the brackets @xmath54 denote a restricted sum .
all of the amplitude dependence of the frequencies is carried by the @xmath55 when we only work to first order in @xmath1 in the @xmath27 .
the requirement that no secular terms appear on the right hand side , the same requirement that fixed the parameters @xmath26 and the set @xmath56 , has also guaranteed that there are no resonant driving terms in the above equation for @xmath49 .
this requirement of no resonant terms has thus transformed the orginal sum into a restricted sum . in the restricted sum , @xmath57
, we discard all resonant driving terms .
a fixed set of indices @xmath58 will produce eight different harmonic source terms .
the meaning of the restricted sum is that we only keep the harmonic driving terms where the arguments do not equal @xmath59 .
the resonant terms have already been accounted for in our definition of @xmath26 .
the full solution @xmath20 is parametrized by two arbitrary constants determined from initial conditions , which implies that the solution for @xmath49 will contain no arbitrary constants .
therefore we are only interested in the particular solution for @xmath49 . rewriting the equation of motion as @xmath60 \cos ( ( \omega_{l } + \omega_{m } - \omega_{n})t ) \nonumber\\ & & \qquad\qquad + \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) -\frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad - \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } - \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \cos ( ( \omega_{l } + \omega_{m } + \omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) -\frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad + \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } + \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \cos((\omega_{l } - \omega_{m } - \omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) + \frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad - \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } + \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \cos((\omega_{l}-\omega_{m}+\omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg [ \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & \qquad\qquad + \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } - \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \sin((\omega_{l } + \omega_{m } - \omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg[- \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & \qquad\qquad - \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } - \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \sin((\omega_{l } - \omega_{m } - \omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg [ \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & \qquad\qquad - \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } + \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \sin((\omega_{l } - \omega_{m } + \omega_{n})t ) \nonumber\\ & & \qquad\qquad \bigg[- \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & \qquad\qquad + \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } + \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \sin((\omega_{l } + \omega_{m } + \omega_{n})t ) \bigg\ } \ , , \nonumber\\ & & \end{aligned}\ ] ] and using the results @xmath61 and @xmath62 we see that the solution for @xmath49 is given by the following expression : @xmath63 \left[\frac { \cos(\omega_{k } t ) - \cos ( ( \omega_{l } + \omega_{m } - \omega_{n } ) t ) } { ( \omega_{l } + \omega_{m } - \omega_{n}+\omega_{k})(\omega_{l}+\omega_{m}-\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) -\frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad - \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } - \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \left[\frac { \cos(\omega_{k } t ) - \cos ( ( \omega_{l } + \omega_{m } + \omega_{n } ) t ) } { ( \omega_{l } + \omega_{m } + \omega_{n}+\omega_{k})(\omega_{l}+\omega_{m}+\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) -\frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad + \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } + \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \left[\frac { \cos(\omega_{k } t ) - \cos ( ( \omega_{l } - \omega_{m } - \omega_{n } ) t ) } { ( \omega_{l } - \omega_{m } - \omega_{n}+\omega_{k})(\omega_{l}-\omega_{m}-\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg [ q_{l}(0)q_{m}(0)q_{n}(0 ) + \frac{q_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{m}\omega_{n } } \nonumber\\ & & \qquad\qquad - \frac{q_{m}(0)\dot{q}_{l}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{n } } + \frac{q_{n}(0)\dot{q}_{m}(0)\dot{q}_{l}(0)}{\omega_{m}\omega_{l } } \bigg ] \left[\frac { \cos(\omega_{k } t ) - \cos ( ( \omega_{l } - \omega_{m } + \omega_{n } ) t ) } { ( \omega_{l } - \omega_{m } + \omega_{n}+\omega_{k})(\omega_{l}-\omega_{m}+\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg [ \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & + \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } - \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \left[\frac { \sin(\omega_{k } t)(\omega_{l}+\omega_{m}-\omega_{n } ) -\sin((\omega_{l}+\omega_{m}-\omega_{n } ) t)\omega_{k } } { \omega_{k}(\omega_{l}+\omega_{m}-\omega_{n}+\omega_{k})(\omega_{l}+\omega_{m}-\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg[- \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & - \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } - \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \left[\frac { \sin(\omega_{k } t)(\omega_{l}-\omega_{m}-\omega_{n } ) -\sin((\omega_{l}-\omega_{m}-\omega_{n } ) t)\omega_{k } } { \omega_{k}(\omega_{l}-\omega_{m}-\omega_{n}+\omega_{k})(\omega_{l}-\omega_{m}-\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg [ \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & - \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } + \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \left[\frac { \sin(\omega_{k } t)(\omega_{l}-\omega_{m}+\omega_{n } ) -\sin((\omega_{l}-\omega_{m}+\omega_{n } ) t)\omega_{k } } { \omega_{k}(\omega_{l}-\omega_{m}+\omega_{n}+\omega_{k})(\omega_{l}-\omega_{m}+\omega_{n}-\omega_{k } ) } \right ] \nonumber\\ & & \qquad\qquad + \bigg[- \frac{\dot{q}_{l}(0)\dot{q}_{m}(0)\dot{q}_{n}(0)}{\omega_{l}\omega_{m}\omega_{n } } + \frac{q_{m}(0)q_{n}(0)\dot{q}_{l}(0)}{\omega_{l } } \nonumber\\ & & \hspace*{-1 cm } + \frac{q_{l}(0)q_{n}(0)\dot{q}_{m}(0)}{\omega_{m } } + \frac{q_{l}(0)q_{m}(0)\dot{q}_{n}(0)}{\omega_{n } } \bigg ] \left[\frac { \sin(\omega_{k } t)(\omega_{l}+\omega_{m}+\omega_{n } ) -\sin((\omega_{l}+\omega_{m}+\omega_{n } ) t)\omega_{k } } { \omega_{k}(\omega_{l}+\omega_{m}+\omega_{n}+\omega_{k})(\omega_{l}+\omega_{m}+\omega_{n}-\omega_{k } ) } \right ] \bigg\ } \
, , \label{zwei } \\ & & \end{aligned}\ ] ] the above expression is the principal result of this work . in the following sections , we apply this formalism for specific @xmath3 and specific initial conditions
the major advantage of the pertubative solutions is their generality ; the full set of pertubative solutions to the fpu-@xmath0 chain presented above with @xmath3 degrees of freedom has @xmath65 independent arbritrary constants specified by the initial conditions , as is necessary to represent a fully general solution .
solitons , discrete breathers , and other exotic nonlinear excitations that have been observed in laboratory fpu-@xmath0 -like systems may then presumbably be constructed from the above general pertubative solution with appropriately chosen initial conditions , given that it does in fact represent the fully general solution in the small @xmath1 limit . in this section
we present some general solutions with initial conditions chosen so as to describe accurately these nonlinear excitations . to illustrate the method
, we construct explicit solutions for the simplest non - trivial system , the @xmath64 system .
the equations of motion are : @xmath66 and @xmath67 one easily finds that @xmath68 as an example of random initial conditions , we consider the case @xmath69 and @xmath70 .
we let @xmath71 . these parameter choices result in values for @xmath72 and @xmath73 of @xmath74 and @xmath75 , respectively ( with dimensions of @xmath76 $ ] ) . with these values for the initial conditions and @xmath1 ,
the explicit solutions are : @xmath77 and @xmath78 in fig .
( [ lind ] ) , we plot @xmath79 and @xmath80 and in fig .
( [ zahleric ] ) , we plot the numeric solutions for the same system . in figs .
( [ coord1],[1q200 ] , [ 1q400 ] ) , we plot both the numerical solution and the analytical approximation for @xmath81 for three different time intervals of twenty units , and in figs .
( [ coord2 ] , [ 2q200 ] , [ 2q400 ] ) , we plot both the numerical solution and the analytical approximation for @xmath82 for the same time intervals . in fig .
( [ time80 ] ) we plot @xmath83 and @xmath84 , , where @xmath84 is the numerical solution for @xmath80 of the system of eqns .
( [ imp],[dimp ] ) . in fig .
( [ timeyo ] ) , we plot the quantities @xmath84 ( black curve ) , our perturbative approximation given by eqn .
( [ ander ] ) ( grey curve ) , and the their difference , eqn .
( [ dork ] ) . from fig .
( [ timeyo ] ) , we note that the principal source of the growing disagreement between our approximate solution and the numerical solution arises from the difference in phases between the two functions .
the amplitudes appear to be in strong agreement .
furthermore , from fig .
( [ time80 ] ) , it appears that the phase difference is some linear function of time .
we thus speculate that the full numerical solution has not only an amplitude - dependent frequency , but also a time - dependent frequency .
if we allow for time - dependent frequency shifts @xmath26 , we may achieve greater agreement for much longer times .
we leave this open question for a further investigation .
the fpu-@xmath0 lattice has been a subject of study in relation to fundamental issues in nonlinear dynamical systems since its introduction . because a single duffing oscillator admits of an exact solution , one may be tempted to entertain the notion that a lattice of duffing oscillators with nearest neighbor coupling may also allow fully general analytic solutions . for special initial conditions ,
the full lattice dynamics is described by a single quartic oscillator and thus admits exact analytic solutions @xcite , but these particular solutions correspond to special initial conditions .
while for a small number @xmath3 of degrees of freedom , some of these special solutions are stable over an appreciable range of @xmath1 , they are actually unstable for arbitrarily small @xmath1 in the limit of @xmath3 approaching infinity . for more general initial conditions ,
we have applied the lindstedt method to lowest order in the nonlinearity parameter @xmath1 to obtain analytical , quasiperiodic approximations .
these quasiperiodic solutions are predicted to exist on the basis of the kam ( kolmogorov , arnold and moser ) theorem and is the accepted explanation for the nonchaotic behavior and recurrences that puzzled fermi , pasta and ulam . as noted above , the formal perturbative scheme presented above has unknown convergence properties . for small @xmath1 and early times ,
the solutions of the formalism and numerically obtained solutions are numerically indistinguishable .
for the example in the text , the pertubative and numerical solutions begin to noticeably differ in phase for dimensionless time on the order of forty .
we conclude that the formal perturbation expansions developed here , truncated at first order in @xmath1 , give useful analytic approximations to quasi - periodic solutions of the fpu-@xmath0 lattice for moderately long times .
finally , we remark that the general application of the above perturbative scheme is independent of the spatial dimension of the system
. 99 r.l .
bivins , n. metropolis , j.r .
nonlinear coupled oscillator : modal equation approach _
la-4943 , may 1972 .
m. pettini , l. casetti , m. cerruti - sola , r. franzosi , and e.g.d .
_ weak and strong chaos in fpu models and beyond_. chaos 15 , 015106 ( 2005 ) .
yuriy a. kosevich _ nonlinear sinusoidal waves and their superposition in anharmonic lattices_. physics review letters , vol . 71 , number 13 , p. 2058
, 27 september 1993 swapan mandal , international journal of non - linear mechanics 38 ( 2003 ) 1095 - 1101 , and references thererin .
david s. sholl and b. i. henry , phys .
lett . a 159 ( 1991 ) 21 - 27 .
j. ford , j. math .
phys . 2 ( 1961 ) 387 - 393 .
j. ford and j. waters , j. math .
phys . 4 ( 1963 ) 1293 - 1306 .
e. a. jackson , j. math .
phys . 4 ( 1963 ) 551 - 558 .
e. a. jackson , j. math .
phys . 4 ( 1963 ) 686 - 700 .
sholl and b. i. henry , phys .
a 44 ( 1991 ) 6364 - 6374 .
g. christie and b. i. henry , phys .
e 58 ( 1998 ) 3045 - 3054 .
richard h. rand .
_ lecture notes on nonlinear vibrations_. http://www.tam.cornell.edu/randdocs/ susumu shinohara _ low - dimensional solutions in the quartic fermi - pasta - ulam system_. journal of the physical society of japan , vol .
71 , no . 8 august 2002 , pp .
1802 - 1804 .
susumu shinohara _
low - dimensional subsystems in anharmonic lattices_. progress of theoretical physics supplement no .
150 , 2003 .
p. poggi and s. ruffo _ exact solutions in the fpu oscillator chain_. physica d 103 ( 1997 ) 251 - 272 .
bob rink _ symmetric invariant manifolds in the fermi - pasta - ulam lattice_. physica d 175 ( 2003 ) 31 - 42 .
|
we apply the lindstedt method to the one dimensional fermi - pasta - ulam @xmath0 lattice to find fully general solutions to the complete set of equations of motion .
the pertubative scheme employed uses @xmath1 as the expansion parameter , where @xmath1 is the coefficient of the quartic coupling between nearest neighbors . we compare our non - secular perturbative solutions to numerical solutions and find striking agreement .
|
recent progress in theoretical and experimental studies has clarified many properties of glass transitions @xcite .
however , the mechanism of the glass transition has not yet been fully understood @xcite .
the major issue is the experimental approach to the investigation of the length scale of the glass transition and dynamics of the @xmath0-process such as dynamical heterogeneity @xcite . according to the adam and gibbs theory ,
the dynamics of glass transitions are associated with cooperative motions which are characterized by a so - called _ cooperatively rearranging region _ ( crr ) , in which molecules move cooperatively with each other @xcite .
the size of the crr is assumed to increase as temperature decreases to the glass transition temperature @xmath8 .
the characteristic length scale associated with such cooperative motions has been investigated by multi - dimensional nmr @xcite , dielectric hole burning @xcite and photobleaching @xcite for bulk systems .
another approach to studying the length scale is the investigation of the finite size effect on glass transition dynamics through confined systems such as thin polymer films @xcite or small molecules in nanopores @xcite .
the first direct measurement of the reduction in @xmath8 in thin polymer films has been made by keddie _
et al . _ in 1994 @xcite . in their work
, the film thickness of the thin films of polystyrene ( ps ) supported on a hydrogen passivated silicon wafer has been measured as a function of temperature by an ellipsometer .
the results showed that the observed @xmath8 in such thin films can be described as a function of film thickness , @xmath9 , as follows : @xmath10,\end{aligned}\ ] ] where the best - fit parameters are @xmath11=373.8k , @xmath12=1.3@xmath13 nm , and @[email protected] @xcite . the molecular weight ( @xmath16 ) dependence of @xmath17 has not been observed for @xmath16 range from 1.2@xmath18 to 2.9@xmath19 , and hence it was concluded that the confinement effect of polymer chains within thin layers may be neglected in such systems .
further measurements of @xmath8 in thin films of poly(methyl methacrylate ) ( pmma ) supported on two different kinds of substrate revealed that the strong attractive interactions between substrate and polymers lead to increase in @xmath8 with decreasing film thickness @xcite . in order to remove such interactions between substrate and polymers , forrest
_ et al . _ measured @xmath8 for _ freely standing films _ of polystyrene by brillouin light scattering measurements @xcite .
they obtained the thickness dependence of @xmath8 as follows : @xmath20 where @xmath11 is the glass transition temperature for thick films , @xmath21 is the constant and @xmath22 is the critical thickness .
the critical thickness @xmath22 , below which @xmath8 decreases linearly with decreasing film thickness , depends on the molecular weight , @xmath23 , @xmath22 increases with the molecular weight in a way similar to the radius of gyration of polymer chains . in our previous papers
, we performed electric capacitance measurements for thin films of polystyrene supported on al - deposited glass substrate and determined the @xmath8 as a temperature at which the temperature dependence of capacitance changes discontinuously @xcite . as a result
, we confirmed the reduction of @xmath8 in our system in a way similar to eq.([tg_keddie ] ) .
furthermore , we made volume relaxation measurements on thin polystyrene films and measured the temperature @xmath1 at which the imaginary component of complex thermal expansion coefficient has a peak at a given very low frequency corresponding to the relaxation time about 100 sec during the heating process of a constant rate @xcite .
the temperature @xmath1 thereby obtained can be regarded as the glass transition temperature @xmath8 . in our measurements
it is found that @xmath8 decreases slightly with decreasing film thickness down to a critical thickness @xmath7 , and below @xmath7 it decreases very rapidly .
the value of @xmath7 changes with @xmath16 in the similar way to the radius of gyration of polymer chains .
this behavior of @xmath8 is quite similar to those obtained in freely standing films as shown in eq.([tg_forrest ] ) except the slight change in @xmath8 above @xmath7 .
intensive investigations have been performed so far mainly on polystyrene films @xcite .
it is important to clarify whether the results extracted from the investigations on thin polystyrene films hold also for other polymers . in this paper , therefore , we investigate the dynamics of thin films of poly(vinyl acetate)(pvac ) and pmma by dielectric relaxation spectroscopy .
pvac is a polymer suitable for dielectric relaxation measurements because a monomer unit of pvac has a large dipole moment @xcite . in pmma
, there is a strong @xmath2-process which is due to the hindered rotation of the side group in addition to the @xmath0-process @xcite .
it is indispensable to investigate the size dependence of dynamics of the @xmath0-process and the @xmath2-process when elucidating the nature of glass transition .
this paper consists of five sections . after the introduction , experimental details on preparation method of thin films and dielectric relaxation spectroscopy
are given in sec.ii .
in sec.iii , experimental results on poly(vinyl acetate ) and poly(methyl methacrylate ) by dielectric relaxation spectroscopy are shown .
in particular , thickness dependence of the temperatures @xmath1 and @xmath3 corresponding to the @xmath0- and @xmath2-peaks in the dielectric loss and the width of the distribution of the relaxation time of the @xmath0- and @xmath2-processes are focused in sec.iii .
discussions on the present experimental results compared with the results reported previously are given in sec.iv and a summary is given in sec.v .
the polymer samples used in this study ( pvac and pmma ) are purchased from scientific polymer products , inc .
the molecular weight and radius of gyration of the polymers are listed in table i. the glass transition temperatures of pvac and pmma in the bulk states are 303k and 373k , respectively .
thin polymer films were prepared by spin - coating a toluene solution of pvac ( pmma ) onto al - deposited glass substrate .
film thickness is controlled by changing the concentration of the solution .
the thin films obtained by spin - coat method were annealed _ in vacuo _ for 48 hours at 303k for pvac and 353k for pmma , respectively .
after annealing , al was vacuum - deposited once more onto the thin films to serve an upper electrode .
dielectric measurements were performed by using an lcr meter ( hp4284a ) for the frequency range from 20hz to 1mhz and an impedance analyzer ( solartron instruments
si1260 ) for the frequency range from 0.1hz to 1mhz .
the temperature of a sample cell was changed between 273k and 373k for pvac and between 263k and 423k for pmma at a constant rate of 0.5k / min .
the dielectric measurements during the heating and cooling processes were performed repeatedly several times .
data acquisition was made during the above cycles except the first cycle .
the good reproducibility of dielectric data was obtained after the first cycle . the thickness @xmath9 is related to the electric capacitance @xmath24 of thin films in the following way : @xmath25 , where @xmath26 is the permittivity of the vacuum , @xmath27 is the permittivity of the polymer ( pvac or pmma ) and @xmath28 is the effective area of the electrode ( @xmath28=8.0mm@xmath29 ) .
for the frequency range where there are no contributions due to any dielectric dispersion , @xmath27 can be regarded as constant for any change in frequency , and hence the film thickness is inversely proportional to the electric capacitance .
relative film thickness at a given temperature can be obtained from the electric capacitance of the thin films at a frequency within the above frequency range . the frequency and the temperature we chose for the determination of the relative thickness are 8khz and 298k for pvac , and 1khz and 273k for pmma .
the absolute values of the film thickness for several films were measured directly by an atomic force microscope ( shimadzu spm-9500 ) in order to calibrate the film thickness . as shown in a previous paper @xcite ,
the resistance of the al electrodes can not be neglected for dielectric measurements of very thin films .
this resistance leads to an artifact loss peak on the high frequency side ; this peak results from the fact that the system is equivalent to a series circuit of a capacitor and resistor @xcite .
because the peak shape in the frequency domain is described by a debye - type equation , the `` c - r peak '' can easily be subtracted .
the data thus corrected were used for further analysis in the frequency domain .
figure 1 shows the dependence of the complex electric capacitance ( @xmath30 ) on the logarithm of frequency at various temperatures for thin films of pvac with @xmath16=1.8@xmath510@xmath6 and film thicknesses ( a ) @xmath9=440 nm and ( b ) @xmath9=16 nm . in this figure
, we find that the imaginary component @xmath31 has a peak due to the @xmath0-process .
the peak frequency shifts to the higher frequency side as temperature increases . comparing fig.1(a ) and
fig.1(b ) , we find that the peak position shifts to the higher frequency side at a fixed temperature as the film thickness decreases from 440 nm to 16 nm .
the temperature change in the dielectric loss at 100hz normalized with the peak value for the bulk sample is shown in fig.2 in the case of pvac with @xmath16=1.8@xmath510@xmath6 .
the dielectric peak due to the @xmath0-process possesses a maximum at the temperature @xmath1 .
it is found in fig.2 that the temperature @xmath1 decreases with decreasing film thickness . at the same time
the peak width of the @xmath0-process increases and the height at @xmath1 decreases with decreasing film thickness .
this behavior is related to the change in the distribution of the relaxation times and the relaxation strength with the film thickness .
figure 3 displays the temperature dependence of the frequency @xmath32 of the @xmath0-process for thin films of pvac with three different thicknesses 440 nm , 62 nm , and 18 nm , where @xmath32 is associated with a characteristic time of the @xmath0-process @xmath33 via the relation @xmath34=1 .
each point in fig.3 consists of a data set ( @xmath35 , @xmath36 ) . for each film thickness
, the temperature dependence of @xmath33 can well be reproduced by the vogel - fulcher - tammann ( vft ) law @xcite : @xmath37 $ ] , where @xmath38 is the relaxation times at very high temperatures , @xmath39 is the apparent activation energy , and @xmath40 is the vogel temperature . at a given temperature , the frequency @xmath32 increases with decreasing film thickness , @xmath23 , the relaxation becomes faster in thinner films .
this thickness dependence of @xmath32 ( and @xmath33 ) becomes stronger as the temperature approaches the glass transition temperature .
this behavior is quite similar to that observed in thin films of ps supported on glass substrate @xcite . in order to check how
the @xmath0-process changes with film thickness , the thickness dependence of the temperature @xmath1 at the frequency 100hz is shown in fig.4 for three different molecular weights 1.2@xmath510@xmath6 , 1.8@xmath510@xmath6 and 2.4@xmath510@xmath6 .
the error bars in fig.4 stand for the standard deviation in @xmath1 for the data acquired by measurements done repeatedly several times . because the glass transition temperature @xmath8 can be regarded as the value of @xmath1 obtained for a very low frequency of the applied electric field , for example , @xmath41 sec ) , the absolute value of @xmath8 is lower than that of the @xmath1 for @xmath32=100hz .
however , it is possible to extract qualitative information on the thickness dependence of @xmath8 from the observed thickness dependence of @xmath1 in fig.4 .
the solid curve in fig.4 is obtained by using the same form of eq.([tg_keddie ] ) in which @xmath8 is replaced by @xmath1 with parameters as follows : @[email protected] , @xmath43 nm , @xmath44 . because the single curve can reproduce the observed values of @xmath1 for the three different molecular weights , the temperature @xmath1 seems to have almost no @xmath16 dependence and hence we conjecture that @xmath8 is also independent of the molecular weight for @xmath16=1.2@xmath510@xmath6 - 2.4@xmath510@xmath6 and @xmath4510 nm . in order to discuss the dynamics of the @xmath0-process in thin films of pvac , the observed complex dielectric constant @xmath46 in fig.1 is fitted to the following model function : @xmath47 where @xmath48 and @xmath49 is the permittivity at a very high frequency . the second term in the r.h.s .
is a contribution from space charge @xcite , and it can be attributed to pure dc conductivity if @xmath50 . in the case of pvac ,
the fitted value of @xmath51 was found to be between 0.40 and 0.93 depending on film thickness . the third term in the r.h.s .
comes from the @xmath0-process and its form is empirically proposed by havriliak - negami @xcite , where @xmath52 is the relaxation strength , @xmath53 is the characteristic time , @xmath54 and @xmath55 are the shape parameters . the thickness dependence of the relaxation strength @xmath52 obtained from the observed values of @xmath46 at 333k is shown in fig.5 for thin films of pvac with @xmath16=2.4@xmath510@xmath6 , where @xmath52 is normalized with respect to the value of @xmath52 for the bulk sample ( see open circles ) .
the temperature 333k was chosen because the dielectric loss peak of the @xmath0-process is located at the center of the frequency window in the present measurement at this temperature .
as the film thickness decreases , the normalized relaxation strength decreases from 1 to about 0.2 .
this behavior in relaxation strength corresponds to the decrease in the height at @xmath1 in dielectric loss in fig.2 .
the @xmath9 dependence of @xmath52 will be discussed later .
the fitting parameters @xmath54 and @xmath55 of the hn equation are shown in fig.6 .
it is found that with decreasing film thickness the parameter @xmath54 decreases and @xmath55 slightly increases .
the decrease in @xmath54 with decreasing film thickness implies the broadening of the distribution of the relaxation times .
the experimental observation that the @xmath46 does not obey a simple debye equation but rather a complicated hn equation can be accounted for by assuming that @xmath56 is expressed as the sum of the debye equations with different relaxation times in the following way : @xmath57 where the second term in eq.([eps1 ] ) is omitted .
@xmath58 is a distribution function of the logarithm of the relaxation times of the @xmath0-process ( @xmath59 ) and is normalized as follows : @xmath60 .
if it is assumed that @xmath46 is expressed by the hn equation , the analytical form of @xmath58 can easily be obtained in the following way with the parameters @xmath54 , @xmath55 , and @xmath53 : @xmath61,\end{aligned}\ ] ] where @xmath62 .
figure 7 displays the distribution of @xmath63 calculated in terms of eq.([dist2 ] ) with the parameters @xmath54 , @xmath55 and @xmath53 at 333k for thin films of pvac with @xmath16=2.4@xmath510@xmath6 and @xmath9=1400 nm and 19 nm . in this figure
, it is found that the average relaxation time of the @xmath0-process decreases with decreasing film thickness , and that the width of the distribution of the relaxation times @xmath33 becomes broader in thinner films . in order to quantify
the change in the distribution of @xmath33 , the full width @xmath64 at the half maximum ( fwhm ) at 333k of @xmath58 was evaluated as shown in fig.8(a ) for thin films of pvac with @xmath65=2.4@xmath510@xmath6 ( see open circles ) .
it is found that the width @xmath64 increases with decreasing film thickness .
the thickness dependence of @xmath64 is given by the equation @xmath66 , where @[email protected] and @[email protected] nm for the @xmath0-process of pvac .
this broadening of the distribution of the relaxation times and its functional form have also been observed in thin films of polystyrene by dielectric relaxation spectroscopy @xcite and ellipsometric measurements @xcite .
figure 9 shows the temperature change in the imaginary part of the complex dielectric constant at a fixed frequency for pmma with @xmath16=4.9@xmath510@xmath6 and @xmath9=900 nm and 9.5 nm .
the value of @xmath69 is normalized with the maximum value for the bulk sample ( @xmath9=900 nm ) .
the value @xmath69 at a fixed frequency ( 40hz ) displays an anomalous increase with temperature due to the @xmath0- and @xmath2-processes , and it possesses two maxima at the temperatures @xmath1 and @xmath3 .
the values of @xmath1 and @xmath3 depend not only on the frequency @xmath32 but also on the film thickness , as we can see in fig.9 ; in the case of @xmath32=40hz , @xmath1=401k and @xmath3=320k for @xmath9=900 nm , and @xmath1=383k and @xmath3=301k for @xmath9=9.5 nm . both the height and the width of the @xmath2-peak are larger than those of the @xmath0-peak .
the origin of the @xmath0-process is attributed to the micro - brownian motion of the polymer main chain , while that of the @xmath2-process is to the hindered rotation of the side branch of pmma @xcite . in order to extract the peak temperatures @xmath1 and @xmath3 and the widths @xmath70 and @xmath71 , the observed values @xmath72 as a function of temperature were fitted by the sum of the two lorentzian functions
@xmath73 ^ 2\}$ ] ( @xmath74 ) .
the functional form has no physical meaning , but it is used only to obtain the peak temperature and the width . if necessary , the contribution due to dc conductivity was also taken into account by adding another peak at higher temperature to the fitting function .
the thickness dependence of the widths of the @xmath0- and @xmath2-processes in the temperature domain , normalized with those in the bulk sample , @xmath75 , is shown in fig.8(b ) , where @xmath76 is the width for the bulk sample . the solid lines in fig.8(b )
are given by @xmath77=@xmath78 .
in pmma , @[email protected] and @[email protected] nm for the @xmath0-process and @xmath80 = [email protected] and @[email protected] nm for the @xmath2-process .
the dependence of the frequency @xmath32 on the reciprocal of the temperatures @xmath1 and @xmath3 is shown in the dispersion map of fig.10 for pmma with @xmath16=4.9@xmath510@xmath6 with three different film thicknesses .
the number of data points of the @xmath0-process is limited because the small @xmath0-peak merges with the large @xmath2-peak as the frequency increases , and as a result it is much more difficult to extract @xmath1 at higher frequencies . as shown in fig.10
, the branch of the @xmath2-process is well separated from that of the @xmath0-process for the frequency range below about 500hz . however , the @xmath2-process approaches the @xmath0-process as the frequency or the temperature increases , because the slope of the straight line corresponding to the @xmath2-process is much smaller than that of the @xmath0-process : the corresponding apparent activation energies in this frequency range are 651@xmath81 kj / mol for the @xmath0-process and 32.7@xmath82 kj / mol for the @xmath2-process .
in this limited frequency range investigated here , thickness dependence of the apparent activation energy can not be observed .
( here , it should be noted that the @xmath0-branch can indeed be fitted by a straight line in fig.10 , because of the limitation of the frequency range , but we do not intend to deny the validity of the vft law . ) in fig.10 we find that both the @xmath0- and @xmath2-processes shift to the higher frequency side at a fixed temperature , as the thickness decreases , @xmath23 , both the relaxations become faster . in order to quantify the thickness dependence of @xmath1 and @xmath3 , the frequency is fixed to be 40hz and the temperatures @xmath1 and @xmath3 are evaluated from the observed curve of @xmath72 vs. @xmath83 for various film thicknesses from 9.5 nm to 1000 nm .
the results are summarized in fig.11 , where figs .
11(a ) and ( b ) display the thickness dependence of the temperatures @xmath1 and @xmath3 observed at 40hz for the thin films of pmma with @xmath16=4.9@xmath510@xmath6 , respectively .
figure 11(a ) shows that @xmath1 _ remains almost constant _ within the experimental accuracy as the film thickness decreases down to a critical thickness @xmath84 , at which point @xmath1 begins to decrease rapidly with @xmath9 . the solid curve in fig.11(a ) below @xmath85
is given by a linear relation between @xmath1 and @xmath9 , although it does not appear to be a straight line because of the logarithmic scale of the horizontal axis .
the amount of the depression of @xmath1 in the thin films is about 20k between the bulk sample and the thinnest film with @xmath9=9.5 nm . in this case , the critical thickness @xmath84 is found to be about 20 nm for the @xmath0-process at 40hz . on the other hand ,
thickness dependence of @xmath3 seems to be to some extent different from that of @xmath1 , as shown in fig.11(b ) .
the temperature @xmath3 _ decreases slightly with decreasing film thickness _ down to a critical thickness @xmath86 even for the relatively thick films .
below a critical thickness @xmath86 , @xmath3 decreases more rapidly with decreasing film thickness .
the depression in @xmath3 is about 25k also in this case for the thickness range investigated here , and the critical thickness @xmath86 is about 19 nm . it is quite interesting that the critical thickness of the @xmath2-process almost agrees with that of the @xmath0-process ; @xmath87 .
this result may suggest that there is a strong correlation between the @xmath0-process and the @xmath2-process , although the two processes at 40hz exist quite separately on the dispersion map in fig.10 . in order to investigate the dynamical behavior in thin films of pmma with @xmath9=900 nm and 9.5 nm , we observed the frequency dependence of the complex dielectric constant at a fixed temperature , as shown in fig.12 .
corresponding to fig.9 , the large peak due to the @xmath2-process can be seen in the @xmath31 vs. @xmath32 plot and a small contribution from the @xmath0-process can also be observed at the foot of the large @xmath2-peak at 403k . according to the procedure of data analysis done for pvac , the observed @xmath88 is fitted to a model function consisting of the two hn functions . in fig.12 ,
the calculated values with this model function are shown as solid curves .
for thick films , the model function can very well reproduce the observed results for the temperature and frequency range shown in fig.12 .
for very thin films , the data at higher frequency ( above 10@xmath89 hz ) include relatively large errors .
therefore , in such a case , we used the data below 10@xmath89 hz for the data fitting . in order to evaluate the distribution @xmath58 and the relaxation strength @xmath52
, the fitting parameters were obtained for the @xmath2-process at the temperature at which the @xmath2-peak is located at 300hz for various film thicknesses . in this case , because the temperatures are located between 323 k and 338 k depending on the film thickness , the contribution from the @xmath0-process can safely be neglected . using the parameters @xmath90 , @xmath91 , and @xmath92 obtained by fitting to the hn equation
, we evaluated the distribution function @xmath58 of the relaxation time of the @xmath2-process for pmma , where the subscript @xmath2 means that these parameters are obtained for the @xmath2-process , and then we obtained the thickness dependence of fwhm @xmath64 .
the result of @xmath64 is shown in fig.8(a ) ( see open triangles ) .
for the @xmath2-process of pmma , @[email protected] and @[email protected] nm . in this figure , the result of @xmath64 for the @xmath0-process in pvac is also shown instead of @xmath64 for the @xmath0-process in pmma , because it was difficult to obtain @xmath64 for the @xmath0-process in pmma .
in addition to @xmath64 , the normalized widths of the dielectric loss peak in the temperature domain due to the @xmath0- and @xmath2-processes , @xmath93 ( @xmath94=@xmath0 , @xmath2 ) , are also shown in fig.8(b ) . because the relaxation time @xmath95 is a function of the temperature @xmath83 , the distribution function of relaxation times @xmath96 can be converted into that in the temperature domain , where the distribution function of the relaxation times with respect to temperature is given by @xmath97 .
therefore , the distribution of the relaxation time can be evaluated not only from @xmath96 but also from the width in the temperature domain . comparing these results with each other
, we found that the distribution of the relaxation times of the @xmath0-process becomes much broader with decreasing film thickness , while for the @xmath2-process this broadening is weaker or remains almost constant with decreasing film thickness .
figure 5 shows the dependence of normalized relaxation strength on the reciprocal of film thickness for the @xmath2-process in thin films of pmma ( see open triangles ) .
the relaxation strengths are obtained by fitting the observed frequency dependence of dielectric loss for the @xmath2-process at temperature at which the @xmath2-peaks are located at 300hz .
the dependence of normalized relaxation strength of the @xmath0-process in thin films of pvac with @xmath16=2.4@xmath510@xmath6 is also shown in fig.5 . however , that of the @xmath0-process in pmma has not yet been evaluated because the dielectric loss due to the @xmath0-process in this case is much smaller than that due to the @xmath2-process .
more precise and wider frequency range measurements will be required to extract the relaxation strength of the @xmath0-process in pmma especially in thinner films . in fig.5 , it is found that the dielectric strengths of both the processes have a similar thickness dependence as follows : @xmath98 where @xmath99 is the relaxation strength of the bulk sample , and @xmath100 is a constant independent of @xmath9 .
the characteristic lengths fitted by eq.([de1 ] ) are [email protected] nm and [email protected] nm for the @xmath0-process in pvac and the @xmath2-process in pmma , respectively
we have successfully observed many properties of dynamics of the @xmath0-process in pvac and the @xmath0- and @xmath2-processes in pmma .
here , we will discuss some of the results observed in this study . in the case of pvac , it has been observed that the temperature @xmath1 of the @xmath0-peak in the temperature domain at 100hz decreases with decreasing film thickness in a manner described by eq.([tg_keddie ] ) .
the thickness dependence of @xmath1 has almost no @xmath16 dependence in the @xmath16 range from 1.2@xmath510@xmath6 to 2.4@xmath510@xmath6 . in the case of the freely standing films of polystyrene ,
_ show that in a low @xmath16 regime ( @xmath1013.47@xmath510@xmath6 ) the value of @xmath8 observed by brillouin light scattering has no molecular weight dependence and it decreases gradually with decreasing film thickness in a similar way to eq.([tg_keddie ] ) @xcite . on the other hand , in a high @xmath16 regime ( @xmath1025.14@xmath510@xmath6 )
the thickness dependence of @xmath8 shows a distinct molecular weight dependence given by eq.([tg_forrest ] ) and it depends on the radius of gyration of the polymer chain @xcite .
although the existence of the interfacial interaction may lead to the difference in the @xmath9 and @xmath16 dependence of @xmath8 between the freely standing films and the thin films supported on substrate , our recent study shows that the temperature @xmath1 observed at 100hz and 1khz by dielectric relaxation spectroscopy has a similar thickness and @xmath16 dependence to eq.([tg_forrest ] ) for thin ps films supported on al - deposited glass substrate @xcite .
therefore , it is expected that the thickness and @xmath16 dependence of @xmath1 in thin films supported on substrate can be compared with that of @xmath8 in freely standing thin films .
for this reason , the observed results on @xmath1 in thin films of pvac may correspond to the low @xmath16 regime observed by mattsson _
_ in the freely standing films of ps . in the regime , _ confinement effect of polymer chains
_ within thin films is not dominant for the reduction of @xmath8 or @xmath1 , but _ the finite size effects _ due to a length scale intrinsic to the glass transition may be dominant . if we measure @xmath1 for thin films of pvac with still higher molecular weights , however , @xmath16 dependence of @xmath1 @xmath103 @xmath9 may be observed in a similar way to @xmath8 in the freely standing films of ps . in pmma , it can be expected that the characteristic length scale of the @xmath2-process , if any , should be smaller than that of the @xmath0-process , because the characteristic time of the @xmath2-process at a fixed temperature in the temperature range investigated here is much shorter than that of the @xmath0-process . in this case
, the thickness dependence of @xmath1 should be stronger than that of @xmath3 at a fixed frequency .
however , the observed results do not obey this expectation ; as shown in fig.11 , for @xmath104 ( @xmath105 ) , @xmath1 remains _ almost independent of thickness _ and @xmath3 _ decreases slightly with decreasing film thickness_. according to a layer model , as discussed in the literature @xcite , @xmath8 and the dynamics of thin films are controlled by the competition between the two different effects .
firstly , the existence of a mobile layer near surfaces leads to the speed - up of the dynamics of the @xmath0-process ( and also the @xmath2-process ) . within the mobile layer
molecular motions are enhanced compared with those in the bulk state .
the origin of the enhancement of molecular motion is supposed to be due to the higher density of chain ends of polymer chains @xcite or to the restriction on cooperative motions characteristic of the glass transition within the surface layer @xcite . in the latter case ,
the layer thickness of the mobile layer is of an order of the size of the cooperatively rearranging region ( crr ) @xcite .
secondly , the attractive interaction between the polymers and the substrate leads to slowing - down of the dynamics of the motion @xcite . for example , the adsorption of polymer chains onto the substrate causes @xmath8 of the thin films to increase compared with that in the bulk state . in the case of thin films of pmma
, it has been reported that the glass transition temperature strongly depends on the interaction between the polymers and the substrate .
_ show that _ the @xmath8 increases with decreasing film thickness _ for sio@xmath106 substrate , while _ it decreases with decreasing film thickness _ for au substrate @xcite .
local thermal analysis measurements showed that @xmath8 of thin films of pmma supported on sio@xmath107 is by 7k higher for @xmath9=20 nm than that for the bulk films , while @xmath8 on sio@xmath107 with hexamethyldisilizane is by 10k lower for @xmath9=20 nm @xcite . in thin films of pmma supported on al deposited silicon substrate , it is reported that @xmath8 of thin films with @xmath9=40 nm is higher by 14@xmath154 k than that of bulk pmma @xcite .
these results suggest that the effect of the interaction between the polymer and the substrate is sensitive to the condition of the surface of the substrate in thin films of pmma . in the @xmath0-process in pmma on al - deposited substrate
the mobile layer associated with cooperative length scale of the glass transition tends to reduce @xmath8 and @xmath1 , while the attractive interaction between the polymer and the substrate tends to increase them . as a result ,
the two effects cancel out and the observed value of @xmath1 may _ remain almost constant _ for @xmath104 , as shown in fig.11(a ) . for the @xmath2-process
, it is expected that the two effects no longer cancel out and the mobile layer has dominant effect on @xmath3 . in order to check the validity of the above idea
, we should measure @xmath1 and @xmath3 of pmma thin films for a substrate which has almost no attractive interaction between the polymer and the substrate .
in such a case , it is expected that as the thickness decreases @xmath1 decreases faster than @xmath3 does . in the present measurements ,
it is observed that for @xmath108 a drastic decrease in @xmath1 begins with decreasing film thickness . to our surprise , @xmath3 also begins to decrease drastically with decreasing film thickness at almost the same film thickness @xmath109 .
this indicates that there is a strong correlation between the @xmath0-process and the @xmath2-process in pmma thin films , although both the processes are located away from each other in the dispersion map ( fig.10 ) .
both the values of @xmath85 and @xmath110 agree well with the radius of gyration of the polymer chain ( @xmath11118.6 nm ) , and hence this behavior may be controlled by the polymeric nature of pmma .
the drastic change at @xmath85 implies the crossover from the normal @xmath0-process to other type of dynamics .
recently , de gennes proposed sliding motion , which can be activated via a soft skin layer near the surface instead of normal @xmath0-dynamics @xcite .
below @xmath7 , the sliding motion may be activated in this case .
the measurements of @xmath16 dependence of @xmath1 and @xmath3 for thin films of pmma are highly desirable in order to clarify the nature of the above behavior on @xmath1 and @xmath3 .
the distribution of the relaxation times of the @xmath0-process @xmath33 in thin films of pvac has been found to become broader with decreasing film thickness .
the width @xmath64 of the distribution in pvac with @xmath9=9.5 nm is about 2.5 times as broad as that in bulk pvac .
on the other hand , the width @xmath64 for the @xmath2-process in thin films of pmma depends on film thickness much more weakly , as shown in fig.8(a ) .
the corresponding width @xmath112 in temperature domain has a similar thickness dependence , @xmath23 , the thickness dependence of @xmath113 is much weaker than that of @xmath70 in pmma . from these observations , it is found that the distribution of relaxation times of the @xmath0-process depends on film thickness much stronger than that of the @xmath2-process .
the intrinsic length scale of the @xmath2-process is expected to be much smaller than that of the @xmath0-process , as deduced from fig.10 .
the present observation on the thickness dependence of the distribution of the relaxation times is consistent with this expectation .
as shown in fig.2 , the dielectric loss peak due to the @xmath0-process becomes more depressed with decreasing film thickness .
this is consistent with the result that the relaxation strength @xmath52 decreases with decreasing film thickness , as shown in fig.5 .
the thickness dependence of @xmath52 is found to obey eq.([de1 ] ) . here
, we will try to give a possible explanation on the decrease in @xmath52 with decreasing thickness .
if there is no interaction between dipole moments , the relaxation strength @xmath114 is described as follows : @xmath115 where @xmath116 is the dipole number density , @xmath117 is the dipole moment of the relaxing unit , @xmath118 is the boltzmann constant , and @xmath83 is the temperature
. now it is assumed that _
@xmath119 neighboring dipole moments are relaxed in a cooperative manner_. then @xmath116 and @xmath117 should be replaced by @xmath120 and @xmath121 in eq.([delta_e ] ) , respectively . as a result
, we obtain the relaxation strength @xmath122 for such a cooperative motion in the following way : @xmath123 here , it can be expected that the number @xmath119 is proportional to the intrinsic length of the motion in question . because cooperative motions are restricted near surfaces or interfaces , the intrinsic length of the motions may be reduced and @xmath119 is decreased .
therefore , @xmath52 of the region near surfaces and interfaces can be expected to be smaller than that of the bulk system . by averaging along the direction normal to the film surface ,
the relaxation strength @xmath52 is expressed by @xmath124,\end{aligned}\ ] ] where @xmath125 is the effective thickness of the surface region , @xmath9 is the overall thickness , and @xmath99 and @xmath126 are the relaxation strength of the bulk region and the surface regions , respectively . here ,
the relation @xmath127 is satisfied . on the basis of this simple picture , the observed thickness dependence of @xmath52 in eq.([de1 ] ) can be reproduced .
the value of @xmath100 is given by @xmath128 .
therefore , the observed result on @xmath52 in the present measurements is consistent with the idea of cooperative motions .
we made dielectric relaxation measurements for thin films of poly(vinyl acetate ) and poly(methyl methacrylate ) supported on al deposited glass substrate .
the results obtained in the present measurements are summarized as follows : 1 .
the temperature @xmath1 corresponding to the peak in the dielectric loss due to the @xmath0-process at 100hz is measured as a function of film thickness for thin films of pvac with three different molecular weights @xmath16=1.2@xmath510@xmath6 , 1.8@xmath510@xmath6 and 2.4@xmath510@xmath6 . it is found that @xmath1 decreases gradually with decreasing film thickness and the decreasing rate of @xmath1 increases more and more in thinner films .
furthermore , @xmath1 has almost no molecular weight dependence .
the width of the distribution of the relaxation times of the @xmath0-process increases with decreasing film thickness , while the relaxation strength of the @xmath0-process decreases with decreasing film thickness .
3 . the thickness dependence of the temperatures @xmath1 and @xmath3 corresponding to the peaks in the dielectric loss due to the @xmath0- and @xmath2-processes at 40hz , respectively , are measured for thin films of pmma with @xmath16=4.9@xmath510@xmath6 . the temperature @xmath1 remains almost constant as the thickness is decreased down to a critical thickness @xmath85 , at which point it begins to decrease with thickness .
the value of @xmath84 is about 20 nm . on the other hand ,
the temperature @xmath3 decreases slightly with decreasing film thickness from the bulk to a critical thickness @xmath110 .
below @xmath110 @xmath3 decreases more rapidly with decreasing film thickness .
the value of @xmath110 is almost equal to @xmath85 .
the widths @xmath70 and @xmath71 of the peaks in dielectric loss in the temperature domain are measured for both the @xmath0- and @xmath2-processes in thin films of pmma with @xmath16=4.9@xmath510@xmath6 .
as the thickness is decreased , the width @xmath70 increases much faster than @xmath71 .
this indicates that the broadening of the distribution of relaxation times of the @xmath0-process is much more sensitive to the change in thickness than that of the @xmath2-process . in this paper , several length scales characteristic of the thickness change in @xmath1 , @xmath70 , @xmath3 , @xmath113 , @xmath52 , and @xmath64
have been introduced .
the physical quantities such as @xmath70 , @xmath71 , @xmath64 and @xmath52 have been found to be described by the linear equation of the reciprocal of @xmath9 .
this result suggests that a layer model be applicable to explain the thickness dependence , as shown in iv.d . in such cases ,
the characteristic lengths @xmath68 and @xmath100 are correlated with the thickness of the surface layer @xmath125 , which can be expected to be comparable to the size of the crr . on the other hand , the crossover thicknesses @xmath85 and @xmath110 are almost equal to the radius of gyration of the polymer chain and are expected to be molecular weight dependent .
these crossover thicknesses may be strongly associated with dynamics of polymer chains . in order to understand the present experimental results ,
a model will be required , where the nature of polymer dynamics is taken into account as well as the intrinsic nature of the glass transition .
the work was partly supported by a grant - in - aid from the ministry of education , science , sports and culture of japan .
`` proceedings of the 3rd international discussion meeting on relaxations in complex systems '' , j. non - cryst . solids .
* 235 - 237 * ( 1998 ) .
`` proceedings of yukawa international seminar on dynamics of glass transition and related topics '' , prog .
* 126 * ( 1997 ) .
angell , k.l .
ngai , g.b .
mckenna , p.f .
mcmillan , and s.w .
martin , j. appl . phys . * 88 * , 3113 ( 2000 ) . m.d .
ediger , c.a .
angell , and s.r .
nagel , j. phys .
chem . * 100 * , 13201 ( 1996 ) .
h. sillescu , j. non - cryst .
solids * 243 * , 81 ( 1999 ) .
g. adam and j.h .
gibbs , j. chem .
phys . * 43 * , 139 ( 1965 ) .
.the weight averaged molecular weight @xmath16 , the ratio of @xmath16 to the number averaged molecular weight @xmath129 , and radius of gyration of the polymers @xmath130 used in this studies [ cols="<,^,^,^",options="header " , ]
|
dynamics of thin films of poly(vinyl acetate ) ( pvac ) and poly(methyl methacrylate ) ( pmma ) have been investigated by dielectric relaxation spectroscopy in the frequency range from 0.1hz to 1mhz at temperatures from 263k to 423k .
the @xmath0-process , the key process of glass transition , is observed for thin films of pvac and pmma as a dielectric loss peak at a temperature @xmath1 in temperature domain with a fixed frequency . for pmma
the @xmath2-process is also observed at a temperature @xmath3 . for pvac , @xmath1 decreases gradually with decreasing thickness , and the thickness dependence of @xmath1 is almost independent of the molecular weight ( @xmath42.4@xmath510@xmath6 ) . for pmma
, @xmath1 remains almost constant as thickness decreases down to a critical thickness @xmath7 , at which point it begins to decrease with decreasing thickness .
contrastingly , @xmath3 decreases gradually as thickness decreases to @xmath7 , and below @xmath7 it decreases drastically . for both pvac and pmma ,
the broadening of the distribution of the relaxation times in thinner films is observed and this broadening is more pronounced for the @xmath0-process than for the @xmath2-process .
it is also observed that the relaxation strength is depressed as the thickness decreases for both the polymers . 2
|
high - redshift ( @xmath3 ) quasars are of major importance since they can provide information on the growth of the supermassive black holes and the evolution of active galactic nuclei ( agn ) in the early universe .
the mere existence of black holes with a few million solar masses ( or more ) at @xmath4 constrains the black hole growth and the accretion process ( e.g. * ? ? ?
* and references therein ) . among the high - redshift quasars ,
the radio - loud ones constitute a very attractive subsample , since their radio jets can be studied with the highest angular resolution , via very long baseline interferometry ( vlbi ) technique .
typically , radio - loud agn contain a flat - spectrum core and a steep - spectrum jet .
the higher redshift is , the fainter the agn radio jet will appear if observed at a fixed wavelength .
thus , if core - jet agn constitute the same population of objects throughout the redshift space , the apparent `` prominence '' of jets at higher redshifts must decrease @xcite : well pronounced jets at high redshifts must appear less frequent than at low redshifts .
additionaly , recent vlbi studies suggest that the naive expectation that relativistically beamed sources ( blazars ) should dominate the high - redshift radio - loud quasar population might not be true .
there seems to exist a population of steep - spectrum high - redshift ( @xmath3 ) radio - loud quasars @xcite , which may be very young radio quasars , similar to the gigahertz - peaked spectrum ( gps ) sources observed in the local universe @xcite . on the other hand , a few high - redshift , beamed blazars are known as well .
three are identified at @xmath5 : sdss j114657.79 + 403708.6 @xcite , sdss j102623.61 + 254259.5 @xcite , and q0906 + 6930 @xcite .
recently @xcite reported the discovery of a high - redshift , radio - bright quasar , sdss [email protected] ( j0131@xmath00321 , hereafter ) .
the source was first selected as a candidate high - redshift quasar @xcite using the optical infrared selection criteria based on the sloan digital sky survey ( sdss , * ? ? ?
* ) and _ wide - field infrared survey explorer _ ( _ wise _ , * ? ? ?
* ) photometry .
its optical spectrum was first measured with the yunnan fainter object spectrograph and camera @xcite in 2013 november
. later higher resolution optical and near - infrared spectra were obtained by the magellan echelette and folded port infrared echelette ( fire ) spectrographs in 2014 january . according to these measurements , the source has a redshift of @xmath6 @xcite .
using the optical spectra and the empirical scaling relation between the mg ii line width and the black hole mass @xcite , @xcite estimated the total bolometric luminosity and the black hole mass of j0131@xmath00321 to be @xmath7w and @xmath8 , respectively . using the latest relation of @xcite
, the black hole mass can be even higher , @xmath9 . according to the faint images of the radio sky at twenty centimeters ( first , * ? ? ?
* ) survey , j0131@xmath00321 is radio - bright with a flux density of 33.7mjy at 1.4ghz . here
we report on the results of our high resolution radio interferometric observation of the source with the european vlbi network ( evn ) . throughout this paper ,
we use flat @xmath10 cold dark matter cosmological model with @xmath11kms@xmath12mpc@xmath12 , @xmath13 , @xmath14 ( the same as used in the discovery paper , * ? ? ?
the exploratory evn observation of j0131@xmath00321 took place on 2014 dec 2 at 1.7ghz ( project code : rsg06 ) .
six antennas participated in this e - vlbi experiment : effelsberg ( germany ) , hartebeesthoek ( south africa ) , jodrell bank mk ii ( united kingdom ) , onsala ( sweden ) , toru ( poland ) , sheshan ( china ) . in an e - vlbi experiment @xcite , the signals received at the remote radio telescopes are transmitted over optical fiber networks directly to the central data processor for real - time correlation .
this was performed at the evn software correlator ( sfxc , * ? ? ?
* ) in the joint institute for vlbi in europe ( jive ) , dwingeloo , the netherlands , with 2 s integration time .
the observation was carried out in phase - reference mode @xcite , thus precise relative positional information could be obtained .
the phase - reference calibrator , j0123@xmath00348 is separated by @xmath15 from the target in the sky .
its coordinates are right ascension @xmath16 and declination @xmath17 .
the positional uncertainties are @xmath18 and @xmath19 milliarcseconds ( mas ) in right ascension and declination , respectively .
the observation lasted for 2 h , the total time spent on j0131@xmath00321 was nearly @xmath20min .
eight intermediate frequency channels ( ifs ) were used in both right and left circular polarizations .
each if was divided into 64 spectral channels .
the total bandwidth was @xmath21mhz . the nrao astronomical image processing system ( aips , * ? ? ? * ) was used for data calibration @xcite following standard procedures .
the interferometric visibility amplitudes were calibrated using the system temperatures and antenna gain curves measured at the telescopes .
after fringe - fitting performed for the phase calibrator , its visibility data were exported to difmap @xcite for imaging .
the standard hybrid mapping procedure was used , with several cycles of cleaning @xcite and phase self - calibration , then finally amplitude self - calibration .
antenna gain correction factors were determined in difmap and applied to the data in aips .
the correction factors on average were below @xmath22 per cent .
fringe - fitting was repeated for the calibrator source but now taking into account its clean component model ( representative of its brightness distribution ) to reduce the small phase variations due to its structure .
the obtained solutions were interpolated and applied to the target source , j0131@xmath00321 . in order to correct for the bandpass ,
four - four channels were discarded at the edges of each if
. then the calibrated and phase - referenced data of the target source were imaged in difmap .
no amplitude self - calibration was attempted for the target source . since j0131@xmath00321 turned out to be adequately bright ( see sect .
[ result ] ) , we were able to fringe - fit the source itself in aips .
the images obtained with and without using the phase - calibrator source agree within the uncertainties . in fig .
[ fig : evn ] , we show the image made from the fringe - fitted visibility data .
the weights of the data points were set inversely proportional to the amplitude errors ( natural weighting ) .
our evn observation revealed a single , compact radio component ( fig .
[ fig : evn ] ) . using the phase - referenced data , we derived the position of the brightness peak with the maxfit verb in aips : right ascension @xmath23 and declination @xmath24 .
we estimate that each coordinate is accurate within 1mas . the dominant contribution to the error
is the positional uncertainty of the phase - reference calibrator .
additional errors come from the thermal noise of the interferometer phases , and from the phase - referencing over @xmath25 in the sky .
the obtained position is @xmath26mas away from the coordinates given in the first catalogue . to quantify the source brightness distribution , we fitted the self - calibrated visibilities with a single circular gaussian model component in difmap .
the integrated flux density is @xmath27mjy , the full width at half maximum ( fwhm ) size is @xmath28mas .
the statistical errors were calculated following the formulae of @xcite .
the angular size corresponds to a projected linear size of @xmath29pc in the rest frame of the source .
the fitted size of the gaussian component is close to , but slightly larger than the minimum resolvable size ( see * ? ? ?
the calibrated visibility amplitudes as the function of the projected baseline length are shown in fig .
[ fig : radplot ] together with the fitted gaussian brightness distribution model ( solid line ) and a point source model with the same flux density as the gaussian model ( dashed line ) . in the figure
, we bin the data in @xmath30 radius , by putting 4110 visibility points in one bin and display weighted averages of the visibility amplitudes for illustration purposes only ; when fitting the model , all visibility points were used . because of the large uncertainty of the visibility measurements on the longest baselines , we use the size estimate as an upper limit and thus we can derive the minimum rest frame brightness temperature of j0131@xmath00321 as @xmath31 where @xmath32 is the flux density measured in jy , @xmath33 is fwhm size of the component in mas
, @xmath34 is the redshift , and @xmath35 is the observing frequency in ghz .
the resulting lower limit to the brightness temperature is @xmath36k .
this brightness temperature is an order of magnitude higher than those of the other high - redshift ( @xmath3 ) , non - blazar radio - loud agn @xcite , and it also exceeds that of sdss j114657.79 + 403708.6 @xcite , a blazar @xcite at @xmath37 by two orders of magnitude .
the 1.7 ghz monochromatic radio luminosity of the source is @xmath38whz@xmath12 .
this is comparable to those of the five @xmath39 radio - loud quasars studied by vlbi by @xcite and the three @xmath5 blazars @xcite .
assuming an intrinsic brightness temperature for the source ( @xmath40 ) and knowing a lower limit to its apparent brightness temperature , a lower limit to the doppler factor ( @xmath41 ) can be estimated .
often the equipartition brightness temperature ( @xmath42k , * ? ? ?
* ) is used as the intrinsic brightness temperature of the source .
this assumed value indicates moderate doppler boosting factor as a lower limit , @xmath43 .
the flux density of the source measured in our 1.7-ghz vlbi experiment is significantly larger than the value given in the 1.4-ghz first catalogue , @xmath44mjy and the value given in the nvss catalogue @xcite , @xmath45mjy .
this can be most straightforwardly explained with variability : j0131@xmath00321 was almost twice as bright at 2014.9 as it was at 2009.2 ( the mean epoch of the first observation ) and at 1993.9 ( the mean epoch of the nvss observation ) .
alternatively , an extremely inverted spectrum would be needed to explain the different flux density values .
the observing frequency of the first survey , @xmath46ghz , and of our evn observation , @xmath47ghz correspond to @xmath48ghz and @xmath49ghz in the rest frame of the source , respectively .
the spectral index ( @xmath50 defined as @xmath51 ) between these frequencies would be @xmath52 , exceeding the canonical value of @xmath53 for a single , optically thick , synchrotron self - absorbed source @xcite .
( spectral index values lower than @xmath53 straightforwardly explained as the result of inhomogeneity in the sources . ) additionally , there is no sign of such inverted spectrum across the eight ifs of our evn observation .
the flux density values measured at the different ifs agree within the errors .
thus , the more likely scenario is that j0131@xmath00321 brightened considerably with respect to its first flux density .
such variability is expected from a beamed , blazar - type source .
the flux density of the phase - reference caibrator in our evn observation , @xmath54mjy , is lower than the flux density values given for the source in the nvss and first catalogues , @xmath55mjy and @xmath56mjy , respectively .
thus , it is unlikely that systematic ( e.g. instrumental ) amplitude calibration errors have led to the high flux density value of the target source , j0131@xmath00321 .
@xcite conducted _ swift _ satellite observation of j0131@xmath00321 between 2014 oct 23 and 2014 dec 9 .
they described the optical uv x - ray spectral energy distribution ( sed ) of the source with a model containing a jet , an accretion disc , and a torus @xcite .
they concluded that the viewing angle of the jet is very close to the line of sight , @xmath57 .
our 1.7-ghz , exploratory e - evn observation revealed that the source is compact with relatively high brightness temperature and moderate doppler boosting factor .
assuming there is no significant misalignment between the scales probed via sed fitting and the radio - emitting jet observed with vlbi at 1.7ghz , using the viewing angle and the lower limit to the doppler factor , we can estimate the lorentz factor of the jet ( @xmath58 ) : @xmath59 where @xmath60 is the bulk speed of the jet material given in the units of the speed of light , @xmath61 .
the resulting lorentz factor is @xmath62 ( and @xmath63 ) . in the model of @xcite , the lorentz factor
is tied to the viewing angle ( @xmath64 ) , and thus is more than four times larger than ours .
the size of the radio emitting region ( @xmath65pc ) is less than eight times the dissipation radius given by @xcite ( @xmath66 m , corresponding to a linear size of @xmath67pc ) .
thus , to explain the existence of the two different lorentz factors , the jet would have to decelerate significantly within this relatively small distance . alternatively ,
since our size estimate is an upper limit , and thus the doppler factor estimate is a lower limit , using the lorentz factor ( @xmath68 ) and the viewing angle ( @xmath69 ) from the model of @xcite , the doppler factor can be calculated to be in the range between @xmath70 and @xmath71 .
assuming the equipartition brightness temperature for the intrinsic brightness temperature of the source , this would imply a measured brightness temperature in the range of @xmath72k .
for the same flux density that we measured , this could be achieved if the radio emitting source size would be in the range between @xmath73mas@xmath74mas corresponding to @xmath75pc@xmath76pc in the frame of the source .
( or for the same source size , the flux density should be @xmath77 times larger , which is however in disagreement with our measurement . ) finally , we note that the x - ray and radio observations were performed relatively close to each other but they were not strictly simultaneous , thus flux density variability of the source might hinder the comparison of the models deduced from the observations . in radio regime , the source variability is indeed detected on a longer timescale of a few years between our evn observation and the first data .
apart from the sed fitting , the blazar classification of the source is also supported by its flux density variability .
the three previously known @xmath5 blazars have black hole masses of @xmath78 . according to @xcite ,
the mass of the black hole in j0131@xmath00321 is @xmath79 , but can be as large as @xmath80 .
however , @xcite assuming a @xcite disk model , determined an even larger black hole mass of @xmath81 .
this larger mass value no longer requires super - eddington accretion to produce the bolometric luminosity of @xmath7w , the eddington ratio would be @xmath82 .
the existence of supermassive black holes with few billion solar masses at early cosmological epochs ( e.g. at redshift @xmath83 , * ? ? ?
* ) poses a challenge to eddington - limited black hole growth formation model .
@xcite showed that a highly spinning black hole with low accretion disk efficiency is able to explain the existence of a black hole with @xmath84 at @xmath1 . on the other hand
, it could also be that the seeds of such supermassive black holes had initially been massive ( @xmath85 ) and/or grew via super - eddington accretion @xcite .
we observed the newly discovered @xmath1 radio - bright quasar , j0131@xmath00321 with evn at @xmath47ghz .
it was detected as a single compact radio source .
we calculated a lower limit to the brightness temperature , @xmath86k .
assuming the equipartition brightness temperature as the intrinsic one , the observed value indicates moderate doppler boosting , @xmath43 .
the flux density of the source in our evn observation is almost twice as high as the the value given in the first catalogue , measured @xmath87 years earlier .
such flux density variability and the relatively high brightness temperature are indicative of the blazar nature of the source .
this is in agreement with the results of @xcite , who analysed the optical , uv , and x - ray measurements of the source .
vlbi observations at higher radio frequencies ( e.g. @xmath88ghz ) can provide a more robust limit on the source compactness and its brightness temperature .
additional constraint on the brightness temperature , thus on the doppler factor and the blazar nature of the source , can be obtained from variability brightness temperature @xcite provided by radio flux density monitoring of the source .
we are grateful to the chair of the evn program committee , tom muxlow , for granting us short exploratory e - vlbi observing time in 2014 dec .
the european vlbi network is a joint facility of european , chinese , south african and other radio astronomy institutes funded by their national research councils .
the e - vlbi research infrastructure in europe was supported by the european community s seventh framework programme ( fp7/2007 - 2013 ) under grant agreement ri-261525 nexpres .
the research leading to these results has received funding from the european commission seventh framework programme ( fp/2007 - 2013 ) under grant agreement no .
283393 ( radionet3 ) .
this research was supported by the hungarian scientific research fund ( otka nn110333 ) , and the china
hungary collaboration and exchange programme by the international cooperation bureau of the chinese academy of sciences ( cas ) .
ta thanks for the grant support from the china ministry of science and technology under grant no . 2013cb837900 .
|
few high - redshift , radio - loud quasars are known to date .
the extremely luminous , radio - bright quasar , sdss [email protected] was recently discovered at a redshift of @xmath1 .
we observed the source with high resolution very long baseline interferometry ( vlbi ) at 1.7ghz with the european vlbi network ( evn ) and found a single compact radio component .
we estimated a lower limit to the brightness temperature of the detected radio component , @xmath2k .
additionaly , when compared to archival radio data , the source showed significant flux density variation .
these two findings are indicative of the blazar nature of the source .
[ firstpage ] techniques : interferometric galaxies : active quasars : individual : sdss [email protected] galaxies : high - redshift
|
in an analytic approach @xcite , we have recently described the spectrum of light scattered by a three - level lambda - type system under two conditions .
one condition was that the incident - light frequency was much lower than the separation frequencies between the states situated in the ends of the arms of lambda configuration ( figure [ fig : lambda ] ) .
thus , the transitions in the arms had multiphoton character .
the other condition was that of weak depopulation of the initial state placed at the bottom of one arm ( the black dot in figure [ fig : lambda ] ) .
as long as the latter condition was fulfilled , our analytic approach succeeded in explaining all details of the spectra found by numerical integration of the schrdinger equation ( number of peaks , their internal structures , positions , relative heights and sensitivity to the light - pulse shape and strength ) .
taking the wavefunction of the system as @xmath0 , the basis of our approach in @xcite was a set of two coupled nonlinear equations for the ratios @xmath1 and @xmath2 of the schrdinger population amplitudes @xmath3 .
this set , found as [ eq : ric ] @xmath4 with the parameters in it clarified in figure [ fig : lambda ] , was analytically solved for the case of weak depletion of the initial state , i.e. , under the condition @xmath5 . under this condition , we first dropped the term @xmath6 in equation ( [ eq : ric1 ] ) , making this equation a quadratically nonlinear riccati - type equation known from the two - level problem @xcite . next
, the riccati equation for @xmath7 was solved in the zero - order approximation by neglecting the term quadratic in @xmath7 . a better solution to the riccati equation for @xmath7 , describing approximately the effect of nonlinearity in @xmath7 , was then found in a perturbative way and this solution was substituted to equation ( [ eq : ric2 ] ) . at a given @xmath7 , equation ( [ eq : ric2 ] ) is a nonhomogeneous linear first - order differential equation with known formal solution @xcite .
now we extend our approach @xcite to cover the case of significant depopulation of the initial state , particularly the spectacular case when the population is periodically exchanged between the two lower states lying in the bottoms of lambda , leaving the highest state practically unpopulated .
such a population exchange between states 1 and 3 was , in fact , obtained in the numerical calculations reported by both us ( figure 2(c ) in @xcite and the corresponding text ) and caldara and fiordilino ( figure 2(b ) in @xcite ) .
specifically , an effective exchange between the mentioned states occurred when these states were separated by nearly the energy of two photons .
this case of the two - photon separation of the lower states is the very one we focus our attention on throughout this paper . in terms of @xmath7 and @xmath8 ,
the case to be considered means that @xmath7 is still small ( @xmath9 ) but @xmath8 can be large . and @xmath10 much higher than the laser frequency @xmath11 . @xmath12 and @xmath13 are the instantaneous rabi frequencies for the electric - dipole 1 - 2 and 2 - 3 couplings , respectively.,scaledwidth=40.0% ]
in figure [ fig : lambda ] , we sketch the system under study .
the initial state of the system has its eigenfrequency @xmath14 .
the parity @xmath15 of the 1st state differs from that of the 2nd state and is the same as that of the 3rd state so the only non - zero dipole matrix elements in the system are @xmath16 and @xmath17 . instead of applying the variables @xmath7 and @xmath8 as in @xcite , we now prefer to work with the equations for the population amplitudes @xmath3 @xcite : [ eq : set1 ] @xmath18 where @xmath12 and @xmath13 are the instantaneous rabi frequencies for the electric - dipole @xmath19 and @xmath20 couplings , respectively , and @xmath21 , @xmath22 are the eigenfrequencies of the other states .
precisely , @xmath23 and @xmath24 , where @xmath25 is a slowly varying pulse shape function and @xmath26 refers to the laser - field oscillations as fast as the optical frequency @xmath27 .
moreover , @xmath28 and @xmath29 are the familiar time - independent one - photon rabi frequencies with @xmath30 being the peak electric - field strength . under the conditions
@xmath31 and @xmath32 , we shall first reduce equations ( [ eq : set1 ] ) to an effective two - state set . since the numerical calculations @xcite pointed to negligible population of level 2 when one meets nearly two - photon resonance between states 1 and 3 , we approximately eliminate @xmath33 from the set ( equations ( [ eq : set1 ] ) ) .
first , we formally integrate equation ( [ eq : b2 ] ) in the range @xmath34 . since this equation alone is a nonhomogeneous linear in @xmath33 first - order differential equation ,
its formal solution is @xcite @xmath35 we know that @xmath36 and that ( since @xmath31 ) these instantaneous rabi frequencies oscillate slowly in time with respect to the function @xmath37 .
however , we a priori assume that the amplitudes @xmath38 and @xmath39 also change slowly in time as compared to @xmath37 .
thus , the solutions for @xmath38 and @xmath39 , if found under this adiabatic assumption , will have a limited range of applicability .
this range will be defined later on . with this adiabatic assumption ,
the leading term in equation ( [ eq : solb2 ] ) is @xmath40 this @xmath41 , when substituted to equations ( [ eq : b1 ] ) and ( [ eq : b3 ] ) , gives a pair of equations for @xmath38 and @xmath39 : [ eq : set2 ] @xmath42 where @xmath43 and @xmath44 . above ,
@xmath45 and @xmath46 represent the instantaneous stark shifts of states 1 and 3 caused by direct coupling of these states to state 2 , while @xmath47 looks like an effective time - dependent coupling between states 1 and 3 through the intermediate state 2 . substituting @xmath48 and @xmath49
, we transform set ( [ eq : set2 ] ) into [ eq : set3 ] @xmath50 where @xmath51 with the phase @xmath52 depending on the optical field strength through the combined static stark shift @xmath53 . here , @xmath54 . when writing the final ( approximate ) form for the phase @xmath55 we have applied the condition that the pulse shape function @xmath25 changes slowly with respect to @xmath56 . introducing a new variable @xmath57 , set ( [ eq : set3 ] )
is then reduced to a single equation for this variable , namely the riccati - type equation @xmath58 this equation has structurally the same form ( but different @xmath59 ) as the one met in the problem of a 2-level system @xcite .
the corresponding two - level system equation reads @xmath60 , where @xmath61 and @xmath62 . in terms of the present @xmath63 and @xmath59 one obtains from equations ( [ eq : set3 ] ) the required amplitudes : [ eq : bric ] @xmath64\,,\\ b_3(t ) & = & x\,\exp\left[i\left(\int_0^t s(t^ { ' } ) x(t^ { ' } ) { \mathrm{d}}t^{'}-\phi(t)\right)\right]\,.\end{aligned}\ ] ] finally , using the fourier - bessel expansion to the term @xmath65 , i.e. , @xmath66 and also the relation @xmath67 satisfied by the bessel functions @xmath68 we write the present @xmath59 defined by equation ( [ eq : fulls ] ) as @xmath69}\nonumber\\ & = & \sum_{n=-\infty}^{\infty}s_n(t)e^{-i[(\omega_{31}-2n\omega_0)t+\delta_t]}\,.\nonumber\end{aligned}\ ] ] this @xmath59 was derived taking fast oscillations of the electric field in the optical pulse in the form of @xmath26 . for
the @xmath70 oscillations one should replace @xmath71 by @xmath72 in equation ( [ eq : sapp1 ] ) .
equations ( [ eq : fulls ] ) and ( [ eq : sapp1 ] ) for @xmath59 show a great role played by the term @xmath73 in the phase @xmath55 . first of all , it is a term allowing higher - order even - photon @xmath74 resonances in the system , i.e. , higher than two - photon resonance .
moreover , this term introduces a complex field dependence to a fixed resonance because , due to this term , the field appears in the argument of the bessel functions . in section iii ,
more details will be given on the role of this term in the phase @xmath55 .
the key riccati equation ( [ eq : riccati ] ) is seen to play a crucial role in the above presentation .
obviously , it also results from equations ( [ eq : ric ] ) for the variables @xmath7 and @xmath8 . to show this
, we first neglect the term quadratic in @xmath7 in equation ( [ eq : ric1 ] ) when considering the case of weak population of state 2 .
then , the simplified equation ( [ eq : ric1 ] ) is solved in the adiabatic approximation with the result @xmath75 analogous to equation ( [ eq : apb2 ] ) . substituting this @xmath7 to equation ( [ eq : ric2 ] ) and taking into account that @xmath76 gives the riccati equation ( [ eq : riccati ] ) .
in @xcite , an analytic solution to the riccati equation like equation ( [ eq : riccati ] ) was given but for the case of weak depopulation of the initial state 1 .
that solution is not valid for the present case because now the initial - state population can completely be moved to the third state under the condition of two - photon @xmath74 resonance .
the strong pumping of state 3 means intermediate and large values of @xmath77 and , thus , the dominant role of the first term on the right - hand side of equation ( [ eq : riccati ] ) . on the other hand , the second term on the right - hand side of equation ( [ eq : riccati ] )
is expected to be non - negligible in the opposite limit of weak pumping of state 3 , i.e. , when @xmath77 is small .
mathematically , if @xmath78 is the solution of the reduced equation ( [ eq : riccati ] ) , @xmath79 , and @xmath80 solves the other reduced equation ( [ eq : riccati ] ) , @xmath81 , then @xmath82 is an approximate solution of the complete riccati equation ( [ eq : riccati ] ) under the condition @xmath83 , where @xmath84 . substituting the above @xmath85 to equations ( [ eq : bric ] ) results in [ eq : bric2 ] @xmath86 where @xmath87 $ ] . in the following
, we shall consider a special form of these amplitudes .
let us assume that the pulse envelope function @xmath25 rises from 0 to 1 in a short few - cycle time @xmath88 and then keeps constant value @xmath89 for @xmath90 .
thus , for the rising time much shorter than the pulse length , we roughly approximate @xmath91 . as a consequence , when the dynamic even - photon @xmath74 resonance takes place ( @xmath92 , where @xmath93 if @xmath94 ; @xmath95 is negative if @xmath96 ) , @xmath59 of equation ( [ eq : sapp1 ] ) splits into its slowly and rapidly varying in time parts .
the slow resonance part is that with @xmath97 only , @xmath98 while the rest in @xmath59 is the other part .
this slow part will give dominant contribution to the integral @xmath99 rooted as the argument in the trigonometric functions in equations ( [ eq : bric2 ] ) .
thus , in analogy to the standard one - photon rotating - wave approximation ( rwa ) @xcite , we now make generalized rotating - wave approximation ( grwa @xcite ) in the case of a given even - photon @xmath74 resonance , i.e , neglect the rapidly oscillating terms in @xmath59 . in grwa , @xmath100 and is thus real entailing @xmath101 , @xmath102 and @xmath103 .
the last @xmath63 is seen to be the exact solution of the riccati equation ( [ eq : riccati ] ) in its grwa limit .
the grwa population amplitudes , obtained from equations ( [ eq : bric2 ] ) , are thus [ eq : solb ] @xmath104 after the rising time @xmath88 has passed , the grwa populations of levels 1 and 3 ( @xmath105 ) are seen to oscillate at a constant @xmath95-photon rabi frequency resulting from equation ( [ eq : slow ] ) by putting @xmath89 . the applicability range of the above special solution ( equations ( [ eq : solb ] ) ) results from the requirement that @xmath38 and @xmath39 change slowly in time with respect to @xmath106 , where @xmath31 .
the amplitudes will change slowly if four inequalities are fulfilled : @xmath107 , @xmath108 , @xmath109 and @xmath110 .
the first inequality is realized in the lambda - configuration of the levels only ( figure [ fig : lambda ] ) , while all other inequalities impose the upper limit on the applied electric field . since @xmath31 ,
the strongest field is such that both @xmath111 and @xmath112 are much smaller than @xmath113 allowing @xmath111 and @xmath112 to be as high as of the order of 1 . for this strongest field ,
one finds @xmath114 justifying the approximation for the bessel function @xmath115 . with this approximation ,
equation ( [ eq : slow ] ) for @xmath116 reduces to : [ eq : rabifreq ] @xmath117 where @xmath118 .
nearly the same solution as our equations ( [ eq : solb ] ) , but with @xmath116 given by equations ( [ eq : rabifreq ] ) rather than general equation ( [ eq : slow ] ) , can be obtained from the floquet - like analysis .
however , this analysis seems to be less comfortable and consumes much more time than our derivation .
the basis for this analysis is the series expansion for the atomic amplitudes @xmath119 and @xmath120 : @xmath121 when substituted to equations ( [ eq : set2 ] ) , this expansion results in the evolutions for the floquet - state amplitudes : [ eq : floqueteq ] @xmath122 under the condition of @xmath123-photon resonance ( @xmath124 ) , equations ( [ eq : floqueteq ] ) can be solved approximately in a truncated basis of states .
the states of this basis are those forming the shortest coupling chain between the initial state of amplitude @xmath125 and the final state of amplitude @xmath126 .
the number of intermediate states in this shortest chain depends on the multiphoton order @xmath95 . for @xmath127 ,
i.e. , the case of degeneracy ( @xmath128 ) there is no intermediate state between the initial and final states characterized by @xmath125 and @xmath129 , respectively .
there is also no intermediate state for @xmath130 but now the initial and final states are characterized by @xmath125 and @xmath131 , respectively .
however , for @xmath132 there are @xmath133 intermediate states characterized by @xmath134 . by adiabatic elimination of all intermediate states
one obtains a pair of equations for the amplitudes of the initial and final floquet states coupled by @xmath95-photon resonance : @xmath135 where @xmath116 is given by equations ( [ eq : rabifreq ] ) .
the last set looks structurally as equations ( [ eq : set2 ] ) and , under the conditions that led us to equations ( [ eq : solb ] ) , we have [ eq : floquet ] @xmath136\,,\\ b_3(t ) & \simeq & \beta_{p/2}e^{-i p \omega_0 t}\nonumber\\ & = & i\,\sin\left(\int_0^t \omega^{(p)}(t^{'}){\mathrm{d}}t^{'}\right)\nonumber\\ & \cdot & \exp\!\!\left[i\!\!\left(\!\!\delta t - p\omega_0 t+\frac{m_r^2}{2\omega_{21}}\!\!\int_0^t\!\!\
! f^2(t^{'}){\mathrm{d}}t^{'}\!\!\right)\!\!\right].\end{aligned}\ ] ] the advantage of our equations ( [ eq : solb ] ) is that they include the rabi frequencies in their general form given by equation ( [ eq : slow ] ) , while the solution from the floquet - like analysis ( equations ( [ eq : floquet ] ) ) includes the approximate rabi frequencies given by equations ( [ eq : rabifreq ] ) .
( we do not show the evolution of @xmath39 due to fast oscillations with frequency @xmath137 under slowly changing envelope ) .
( b ) two - photon rabi oscillations of the populations in states 1 and 3 . ( c )
coherent spectrum of scattered light .
the conditions : @xmath138 , @xmath139 , @xmath140 , @xmath141 , the pulse shape function @xmath25 increases to its maximum value of 1 by four optical cycles and then keeps this value .
solid lines - the analytic results from equations ( [ eq : solb ] ) , dotted lines - the results of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] ( we do not show the evolution of @xmath39 due to fast oscillations with frequency @xmath137 under slowly changing envelope ) .
( b ) two - photon rabi oscillations of the populations in states 1 and 3 . ( c )
coherent spectrum of scattered light .
the conditions : @xmath138 , @xmath139 , @xmath140 , @xmath141 , the pulse shape function @xmath25 increases to its maximum value of 1 by four optical cycles and then keeps this value .
solid lines - the analytic results from equations ( [ eq : solb ] ) , dotted lines - the results of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] ( we do not show the evolution of @xmath39 due to fast oscillations with frequency @xmath137 under slowly changing envelope ) .
( b ) two - photon rabi oscillations of the populations in states 1 and 3 . ( c )
coherent spectrum of scattered light .
the conditions : @xmath138 , @xmath139 , @xmath140 , @xmath141 , the pulse shape function @xmath25 increases to its maximum value of 1 by four optical cycles and then keeps this value .
solid lines - the analytic results from equations ( [ eq : solb ] ) , dotted lines - the results of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] , @xmath142 , @xmath143 , @xmath144 , @xmath145 , @xmath146 $ ] with @xmath147 and @xmath148 .
( c ) as ( a ) but now for @xmath149 , @xmath150 and the same all other parameters .
solid lines - the results after the adiabatic and generalized rotating - wave approximations , dotted lines - the result of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] , @xmath142 , @xmath143 , @xmath144 , @xmath145 , @xmath146 $ ] with @xmath147 and @xmath148 .
( c ) as ( a ) but now for @xmath149 , @xmath150 and the same all other parameters .
solid lines - the results after the adiabatic and generalized rotating - wave approximations , dotted lines - the result of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] , @xmath142 , @xmath143 , @xmath144 , @xmath145 , @xmath146 $ ] with @xmath147 and @xmath148 .
( c ) as ( a ) but now for @xmath149 , @xmath150 and the same all other parameters .
solid lines - the results after the adiabatic and generalized rotating - wave approximations , dotted lines - the result of numerical integration of equations ( [ eq : set1]).,title="fig:",scaledwidth=40.0% ] in figure [ fig1 ] , we show the quality of the analytic grwa population amplitudes ( [ eq : solb ] ) . we compare the results obtained from equations ( [ eq : solb ] ) ( solid lines ) with those of the numerical integration ( dotted lines ) of the starting equations ( [ eq : set1 ] ) . to ensure roughly the applicability of the analytic solution and the @xmath74 even - photon resonance
we have taken the same parameters as in @xcite , i.e. , @xmath138 , @xmath151 , @xmath152 , @xmath141 and , additionally , 4-cycle rise time of the pulse ( @xmath153 for @xmath154 and @xmath89 for @xmath155 ) .
the assumed @xmath156 , @xmath157 , and @xmath158 give @xmath159 so one meets the dynamic 2-photon @xmath74 resonance ( @xmath160 ) .
for the above frequency - strength parameters the analytic amplitudes ( [ eq : solb ] ) are reduced to the forms @xmath161 and @xmath162 $ ] , where @xmath163 results from the approximate equation ( [ eq : rabifreq2 ] ) for @xmath164 .
we have used these forms of @xmath38 and @xmath39 when drawing all analytic curves ( solid lines ) in figure [ fig1 ] .
figure [ fig1](a ) shows the evolutions of the real and imaginary parts of @xmath38 , figure [ fig1](b ) the 2-photon rabi oscillations of populations ( @xmath165,@xmath166 ) between levels 1 and 3 ( the population of level 2 does not exceed @xmath167 for the parameters chosen ) , and figure [ fig1](c ) the coherent part of the spectrum of light scattered by the system ( @xmath168 means squared modulus of the finite - time fourier transform of the induced dipole in the system : @xmath169 , where @xmath33 is to be calculated from equation ( [ eq : apb2 ] ) by applying the above @xmath38 and @xmath39 ) .
as seen , the analytic solution explains well the trends observed in the results of numerical integration of equations ( [ eq : set1 ] ) .
moreover , much better coincidence is now found between the analytic and numerical spectra in the low - frequency regime than in our previous paper ( figure 2(c ) in @xcite ) .
this low - frequency part of spectrum includes three components .
the strongest component simply corresponds to the elastic rayleigh scattering .
the two remaining components , @xmath170 and @xmath171 , are the result of nonlinearity in the interaction between the system and light . as the field intensity increases ( large @xmath172 and @xmath158 ) , the @xmath171 component becomes more pronounced , as seen in figure [ fig:2](b )
. one would expect the appearance of higher - order odd harmonics of @xmath11 in the spectrum as the intensity rises .
however , we would like to mention that , in the high - frequency regime of the numerical spectrum in figure [ fig1](c ) ( @xmath173 ) , one observes an additional low structure .
the height of this structure is about five orders of magnitude lower than that of the leading peak in figure [ fig1](c ) .
our analytic solution was not able to reproduce this additional low structure observed in the numerical spectrum .
this is , probably , the cost we pay for the approximate procedures , particularly the procedure of adiabatic elimination of @xmath33 from equations ( [ eq : set1 ] ) .
the adiabatic elimination has turned out to be the main reason of quantitative differences between the solid and dotted curves in figure [ fig1 ] .
it was found that this procedure works better for higher ratios @xmath113 . by increasing this ratio well above the assumed value 13
, the differences in the long - time population evolutions gradually decrease .
obviously , longer pulses are needed to observe the 2-photon rabi oscillations when increasing the @xmath113 ratio . in figure [ fig:2 ]
, we present a different comparison for the conditions assumed by caldara and fiordilino @xcite in their numerical integration of the schrdinger equation . these conditions were : @xmath174 , @xmath143 , @xmath175 , @xmath145 and the smooth pulse shape function @xmath146 $ ] , where @xmath147 and @xmath148 .
for these conditions , equations ( [ eq : slow ] ) and ( [ eq : rabifreq ] ) for @xmath116 need to be multiplied by the factor @xmath176 . after including this factor we were not able to solve the riccati equation ( [ eq : riccati ] ) in an exact analytic way
so , we solved numerically the grwa version of equations ( [ eq : set3 ] ) for @xmath177 and @xmath178 ( @xmath179 ) and then obtained the solid - line curves shown in figure [ fig:2 ] .
these curves , resulting from both the adiabatic and generalized rotating - wave approximations , are compared with the dot - line curves coming from the numerical integration of equations ( [ eq : set1 ] ) .
the comparison shows that , under the assumed conditions , the two approximations lead to results reflecting the main aspects of the exact ( numerical ) solution of equations ( [ eq : set1 ] ) .
again , by increasing the ratio @xmath113 ( from 19 to 50 ) much better agreement was found between the analytic and numerical curves ( see figure [ fig:2](c ) ) . in the same time scale as in figure [ fig:2](a ) , almost all the population is seen in figure [ fig:2](c ) to move from the initial state 1 to state 3 but without the return 2-photon transition from 3 to 1 .
the residual differences , still seen between the solid and dotted curves in figure [ fig:2](c ) , may suggest that the approximate grwa multiphoton rabi frequencies given by equations ( [ eq : rabifreq ] ) are not sufficiently accurate .
however , taking @xmath116 in the form of equation ( [ eq : slow ] ) has not removed the residual differences . as to the photon - emission spectrum shown in figure [ fig:2](b )
, we do not observe now any low structure in the high - frequency regime ( @xmath173 ) of the numerical spectrum ( dotted line ) .
this makes a difference when compared to the numerical spectrum in figure [ fig1](c ) .
with this paper we have extended the scope of validity of our recent analytic approach @xcite to the problems of multiphoton population transfer in a three - level lambda - type system and scattering of low - frequency light by this system.the previous approach covered the case of weak depopulation of the initial state only .
now , the present version of the approach allows strong depopulation of the initial state as well .
the amended approach was shown to explain the main numerical results , particularly the two - photon rabi oscillations of population between the states forming the bottoms of @xmath180 and the dominant ( if not all ) peaks in the spectrum of scattered light . though we considered the 2-photon resonance between the lower states in @xmath180 , the approach holds for any even - photon resonance between the mentioned states as well as for the case of degeneration of these states .
|
we generalize our analytic approach ( berent m and parzyski r 2009 _ phys .
rev . _ a * 80 * 033834 ) to include the effect of strong depopulation of the initial state when considering the multiphoton population transfer and scattering of low - frequency light by a 3-level system in the lambda - type configuration .
we discuss the quality of the approximations made , i.e. , the adiabatic and generalized rotating - wave approximations .
|
high - order models have recently attracted considerable interests in the lattice boltzmann ( lb ) community . for these models ,
high - order terms in the expanded distribution function are retained , thus multi - speed lattices have to be used . by doing so , we have a few benefits , e.g. consistent description of thermal flows , the galilean invariance of the transport coefficients , improved model capability for compressible and rarefied flows@xcite .
meanwhile , high - order models can still preserve the simplicity of the standard lb model .
high - order lb models are tererfore often applied to thermal flows , compressible flows , and rarefied flows @xcite . due to their kinetic origin ,
high - order lb models have shown to be able to approach the boltzmann model equation such as the bhatnagar - gross - krook ( bgk ) equation by increasing the expansion and quadrature order@xcite .
numerically , it is found that high - order models are capable of describing rarefaction effects up to the early transitional regime with slightly increased discrete velocities .
for example , the velocity - slip can be captured by a d2q16 model for the knudsen number up to @xmath0 where d@xmath1q@xmath2 refers to the standard @xmath1 dimenional and @xmath2 discrete velocities .
a recent study on thermal flows shows that thermal rarefaction effects can also be captured by using a moderate number of discrete velocities with an affordable computational cost@xcite .
in addition to capturing the non - equilibrium effects , high - order models are also useful for other applications .
for instance , by using a d2q25 model , daniel et at@xcite have shown that high weber number can be achievable for droplet collision simulations .
implementation of boundary conditions is crucial to application of high - order lb models .
the kinetic diffuse - reflection boundary condition has shown to be able to predict velocity - slip and temperature - jump at the solid boundary . moreover ,
the positivity of distribution function can always be maintained ( provided that the distribution function from the bulk is positive ) , which is key to numerical stability .
however , due to the multi - speed lattice of high - order models , the formulation of the kinetic diffuse - reflection boundary condition with the characteristic `` streaming and collision '' mechanism is yet to be developed .
the successful implementations so far for high - order models are based on various finite difference scheme @xcite , where the highly desirable `` streaming and collision '' mechanism disappears .
because of this `` streaming and collision '' mechanism , lb method is often regarded a particle method .
the purpose of this work is to formulate the kinetic diffuse reflection boundary condition to retain the `` streaming and collision '' feature .
the boltzmann - bgk equation can be dicretized using a systematic procedure ( see @xcite for the details ) to derive the lb governing equation , which can be written as @xmath3 where @xmath4 denotes the single - particle distribution function evaluated at a discrete velocity @xmath5 , @xmath6 is the truncated maxwellian distribution , while @xmath7 is the mean relaxation time . for convenience
, a non - dimensional system @xmath8 @xmath9 can be introduced in eq.([lbgk ] ) , where the symbols with hat represent the dimensional quantities .
the common notations are used to represent physical quantities , i.e. @xmath10 denotes density ; @xmath11 , dynamic viscosity ; @xmath12 , velocity ; @xmath13 , pressure ; @xmath14 , temperature ; @xmath15 , stress and @xmath16 , heat flux .
@xmath17 is the characteristic length of the system .
the symbols with subscript @xmath18 are the corresponding reference quantities . with this non - dimensional system ,
the equation of state becomes @xmath19 .
the mean relaxation time can be written explicitly as @xmath20 , which is related to the viscosity and pressure . meanwhile
, the macroscopic quantities can be obtained as @xmath21=\sum_{\alpha=1}^{d}f_{\alpha}\left[\begin{array}{c } 1\\ c_{\mbox{\ensuremath{\alpha},}i}\\ c_{<\mbox{\ensuremath{\alpha},}i}c_{\mbox{\ensuremath{\alpha},}j>}\\ \frac{1}{2}c_{\mbox{\ensuremath{\alpha},}i}c_{\mbox{\ensuremath{\alpha},}i}c_{\mbox{\ensuremath{\alpha},}j}\\ c_{\mbox{\ensuremath{\alpha},}i}c_{\mbox{\ensuremath{\alpha},}i } \end{array}\right],\label{eq : summac}\ ] ] where @xmath22 is the space dimension number and @xmath23 is the total discrete velocity number and the angle brackets @xmath24 indicates the trace - free part of the tensor .
the discrete velocity @xmath5 and its weights @xmath25 may be determined through several ways , e.g. , @xcite list explicitly various orders of discrete velocity sets . for convenience
, we use the notation @xmath26 to represent the discrete velocity set . with an appropriate @xmath27 ,
the explicit form of the truncated maxwellian distribution can be given as @xmath28\right.\label{eq : feq}\\ & + & \frac{c_{i}u_{i}}{6}[(c_{i}u_{i})^{2}-3u_{i}u_{i}+3(t-1)(c_{i}c_{i}-d-2)]\nonumber \\ & + & \frac{1}{24}[(c_{i}u_{i})^{4}-6(u_{i}c_{i})^{2}u_{j}u_{j}+3(u_{j}u_{j})^{2}]\nonumber \\ & + & \frac{t-1}{4}[(c_{i}c_{i}-d-2)((u_{i}c_{i})^{2}-u_{i}u_{i})-2(u_{i}c_{i})^{2}]\nonumber \\ & + & \frac{(t-1)^{2}}{8}\left.\left[(c_{i}c_{i})^{2}-2(d+2)c_{i}c_{i}+d(d+2)\right]\right\ } , \nonumber \end{aligned}\ ] ] in which the forth order expansion is used .
if an isothermal flow is concerned , the temperature @xmath14 should be set to @xmath29 .
moreover , to simulate incompressible flows , it is common to use only the second order terms of @xmath30 . with eq.([lbgk ] ) and ( [ eq : feq ] ) , the final issue is to choose an appropriate numerical scheme . following the spirit of `` streaming and collision '' mechanism @xcite , an implicit scheme @xmath31\\ & + & \frac{dt}{2\tau(\bm{x}+\bm{c}_{\alpha}dt , t+dt)}\left[f_{\alpha}^{eq}(\bm{x}+\bm{c}_{\alpha}dt , t+dt)-f_{\alpha}(\bm{x}+\bm{c}_{\alpha}dt , t+dt)\right],\end{aligned}\ ] ] can be constructed .
if we introduce a new variable @xmath32 to eliminate the implicitness , we will get the evolution equation for@xmath33 as @xmath34.\label{eq : scheme}\ ] ] the advantage of eq.([eq : scheme ] ) is that , if the discrete velocities are tied to discretization of the space and time by choosing @xmath27 with integer value , the evolution of @xmath35 can be accomplished in a way similar to a `` particle '' , which makes the lb method simple but still flexible . with the variable @xmath35 , conservative quantities like density can still be obtained by using eq.([eq : summac ] ) without changing form but some conversions are needed for shear stress and heat flux ( cf .
in addition , the mean relaxation time @xmath7 may be related to the local gas temperature for thermal problems .
the essential idea of the kinetic diffuse - reflection boundary condition is that an outgoing particle completely forgets its history and its velocity is re - normalized by the maxwellian distribution . to implement this diffuse - reflection principle for the high - order lb model , we will follow the procedure described in @xcite .
moreover , the discussion is based on the assumption that the effective particle - wall interaction time is small compared to any characteristic time of interest and no permanent adsorption occurs@xcite . for the high - order models , as `` particles '' from more than one layer of computational grids can hit the wall , we have to properly identify them in order to implement the boundary condition . for this purpose ,
@xmath36 layers of ghost grids are introduced ( see the example of the d2q17 lattice and its grid arrangement shown in fig.[fig : d2q17-lattice ] and [ fig : illustration ] ) , where @xmath36 can be determined via the corresponding maximum value of the discrete velocity heading towards the wall ( e.g. , @xmath37 for the d2q17 lattice ) . as a common practice
, the physical wall is located at the half grid space between the ghost and fluid grids . to further distinguish incoming and outgoing particles , we use @xmath38 and @xmath39 to represent their velocities respectively , where @xmath40 denotes the layer number of the ghost grid ranging from @xmath18 to @xmath41 .
similarly , the distributions of incoming and outgoing particles at layer @xmath40 are written as @xmath42@xmath43 and @xmath44 , where the superscripts @xmath45 and @xmath46 stand for ` incoming ' and ` outgoing ' .
the corresponding discrete velocities must satisfy the condition @xmath47 and @xmath48 , @xmath49 denotes the unit vector normal to the wall surface @xmath50 at @xmath51 and directed from the wall into the gas .
note in the present lattice system @xmath52 so the conditions are equivalent to @xmath53 and @xmath54 .
indeed , @xmath38 and @xmath39 are a symmetric pair . on the other hand ,
the known information of the wall , i.e. , the position , velocity and temperature , are represented by @xmath55,@xmath56 and @xmath57 .
obviously , the distribution @xmath42@xmath43 can be obtained by naturally streaming the distribution function at fluid grids into the corresponding ghost ones .
we need to determine the unknown distribution @xmath44 according to the principle of diffusion reflection .
similar to the derivation of the continuum version of diffusion - reflection condition @xcite , we first write down the mass of outgoing and incoming particles as , @xmath58 @xmath59 where @xmath60 stands for mass and @xmath61 denotes the volume of the grid cell .
it is worth noting again here that , due to the exact advection of the lb method ( cf .
eq.([eq : scheme ] ) ) , the flux term in the continuum version ( cf .
( 1.11.1 ) in @xcite ) , can be replaced by the distribution function itself .
hence , according to the mass conservation , we have , @xmath62 where @xmath63 is the so - called scattering probability .
immediately , we arrive at @xmath64 moreover , the scattering probability @xmath65 must satisfy the property of non - negativeness , normalization and reciprocity condition@xcite .
particularly , the normalization condition , corresponding to mass conservation under the assumption of no permanent adsorption , can be written as , @xmath66 so far , the discussion is still generic as we have not introduced any specific assumption for the diffuse - reflection principle .
therefore , the above formulation may also be used to derive other type of boundary condition .
if the assumption of the diffuse - reflection boundary condition is applied , the scattering probability can be easily calculated as @xmath67 hence , the distributions of outgoing particles can be written as @xmath68
to validate the proposed implementation of kinetic boundary condition , we consider steady couette flow confined in two parallel planar plates located at @xmath69 and @xmath70 and moving oppositely with the same speed .
all the quantities are presented in their non - dimensional form , and both isothermal and thermal conditions are considered .
therefore , three lattice systems , namely d2q17@xcite , d2q16@xcite and d3q121@xcite are tested , where the d2q17 and d2q16 models are appropriate for the isothermal cases and the d3q121 model for the thermal ones .
the d2q17 lattice is illustrated in fig.[fig : d2q17-lattice ] and the corresponding grid arrangement is shown in fig.[fig : illustration ] .
the details of three lattices are omitted here for simplicity , which can be found in @xcite , @xcite and @xcite . ,
scaledwidth=30.0% ] ) .
the point denotes the bulk grid while the dashed circle represents the ghost grid .
[ fig : illustration],scaledwidth=70.0% ] we restrict to the couette flow within the slip - flow regime , so we may be able to use solutions of the navier - stokes - fourier ( ns ) equations as reference .
for the ns solutions , it is necessary to apply the velocity - slip and temperature - jump boundary conditions so that the velocity and temperature profiles can be written as @xmath71 and @xmath72 where the knudsen number is defined as @xmath73 @xmath74 denotes magnitude of the component of interest of the wall velocity @xmath56 . for some relatively larger knudsen numbers we may also compare to the solution of the linearized boltzmann - bgk ( l - bgk ) equation .
we first evaluate the d2q17 and d2q16 models for isothermal flows which are presented in fig.[fig : isothermal - case ] and [ fig : isothermal - case-1 ] , where @xmath74 is set to be @xmath75 .
both models are simulated with @xmath76 computational grids in the direction of interest and the comparisons are made against the ns solutions for @xmath77 and the l - bgk solutions for @xmath78 respectively .
the results show that the boundary condition eq.([eq : dkboundary ] ) works correctly for the isothermal flows . for @xmath79 ,
the velocity profiles are captured well by the d2q17 model while some deviations from the l - bgk results are observed for larger knudsen numbers , particularly at @xmath80 ( see fig.[fig : isothermal - case ] ) .
however , this is of no surprise as it is known that these deviations are due to the lattice structure@xcite . a further comparison to the finite difference ( fd ) implementation of eq.([lbgk ] ) ( see the description in @xcite , where the numerical simulation is validated by the analytical solution in @xcite ) , confirms the appropriateness of the boundary implementation .
interestingly , the d2q16 model can given much better predictions for the velocity profile .
even at @xmath81 , it still gives satisfactory results , see fig .
[ fig : isothermal - case-1 ] .
the reason was already discussed in @xcite , title="fig:",scaledwidth=47.0%],title="fig:",scaledwidth=47.0% ] , title="fig:",scaledwidth=47.0%],title="fig:",scaledwidth=47.0% ] , title="fig:",scaledwidth=47.0%],title="fig:",scaledwidth=47.0% ] for further validation , we also simulate the thermal couette flows using the d3q121 model . for these flows
the relevant parameters are @xmath82 and @xmath83 while the wall temperatures are set to be @xmath29 and their speed @xmath74 is set to be @xmath84 .
as relatively small knudsen numbers are considered here , the results are compared to the ns solutions , see fig.[fig : thermal ] .
the subtle temperature jumps are well captured at the wall boundary .
these agreements again confirm the appropriateness of the boundary treatment .
the velocity profiles show similar behavior to the isothermal cases , so they are not presented in fig .
[ fig : thermal ] .
.[fig : thermal],title="fig:",scaledwidth=47.0%].[fig : thermal],title="fig:",scaledwidth=47.0% ] to evaluate the numerical accuracy , a convergence study is conducted for the thermal case of @xmath85 .
the simulations are run for six different grid resolutions @xmath86 in the direction of interest .
the results of @xmath87 is then chosen as reference and the global relative errors of the velocity and temperature are defined as @xmath88^{2}}{\sum_{j=1}^{n_{g}}u_{\delta}^{2}(y_{j})}},\ ] ] and @xmath89^{2}}{\sum_{j=1}^{n_{g}}(t_{\delta}(y_{j})-t_{w})^{2}}},\ ] ] where @xmath90 and @xmath91 represent the results of @xmath87 .
fig.[fig : logana ] shows that the second order accuracy is achieved globally . and @xmath92 on the grid number @xmath93 .
[ fig : logana],title="fig:",scaledwidth=47.0% ] and @xmath92 on the grid number @xmath93 .
[ fig : logana],title="fig:",scaledwidth=47.0% ]
to conclude , we have formulated the kinetic diffuse reflection boundary condition for high - order lb models with emphasis on retaining the `` streaming - collision '' mechanism .
the numerical tests for both isothermal and thermal couette flows show that the present boundary condition can capture velocity - slip and temperature - jump very well within the capacity of the corresponding lattices . in term of numerical accuracy , we show that the second order accuracy can be achieved globally .
the authors would like to thank dr .
xiaowen shan for many informative discussions .
the research leading to these results has received funding from the engineering and physical sciences research council u.k . under grants no .
ep / f028865/1 and ep/ i036117/1 .
|
the implementation of the kinetic diffuse boundary condition with the characteristic streaming - collision mechanism is studied for the high - order lattice boltzmann ( lb ) models .
the obtained formulation is also tested and validated numerically for three high - order lb models for both isothermal and thermal couette flows .
the streaming - collision mechanism ensures that high - order lb models can retain particle feature while go beyond the navier - stokes hydrodynamics .
|
massive o - type stars , including the progenitors of wolf - rayet stars , inject enormous quantities of energy into the interstellar medium ( ism ) in the form of dissociating and ionizing radiation and winds .
the radiation fields create circumstellar / interstellar and regions into which winds blow , creating large - scale swept - up expanding shells of gas known as stellar wind bubbles @xcite .
there are numerous analytical and numerical studies of the formation , structure , and evolution of stellar wind bubbles ( swbs ) , but there are few convincing observations of bubbles around wolf - rayet stars and massive o - stars .
the viith catalogue of galactic wolf - rayet stars indicates that only @xmath0 are associated with bubbles @xcite .
surveys are incomplete , but many of these proposed bubbles are not securely established . the interstellar environment of massive stars is observed to be extremely complex and is likely to be the combined result of inhomogeneous initial conditions and the influence of the stars .
our understanding of the physics of astrophysical flows in inhomogeneous media is incomplete @xcite . in particular the effects of conduction , diffusion ,
turbulence , magnetism , and of mixing and mass - loading processes are all poorly understood .
therefore , in the observational study of swbs and the environment of massive stars it is important to remain open to the possibility of phenomena not predicted by standard swb models . in this paper
we present one possible example of such a phenomenon .
we investigate the nature of a large - scale system of neutral atomic hydrogen filaments which apparently emanate from compact molecular clouds located in or near the perseus arm of our galaxy .
we investigate the possibility that these `` tails '' are the result of molecular cloud mass - loading within the wind flow of massive stars .
the observations and data processing are briefly outlined in the next section .
section [ sect : results ] describes the observational results . in section [ sec : driving_srcs ] we consider possible driving sources , and in section [ sec : environ ] we take a closer look at the environment of the two main candidates .
the discussion follows ( [ sec : discuss ] ) , in which we first consider the kinematics and timescales involved ( [ sec : kin_time ] ) , and then present an analysis within the framework of a mass - loaded flow model ( [ sec : mass - loaded ] ) .
our conclusions follow in section [ sec : conclude ] .
the data discussed in this paper were obtained as part of the canadian galactic plane survey , a long - term multi - wavelength imaging survey of the ism of the northern galactic plane @xcite .
table [ tbl : cgps_param ] summarizes the main parameters of the relevant cgps data .
the principal observational components of the cgps are arcminute resolution mosaics of 21 cm ( in all stokes parameters ) and 74 cm ( right hand circular polarization ) radio continuum , and 21 cm line emission , all observed with the synthesis telescope @xcite at the dominion radio astrophysical observatory of the herzberg institute of astrophysics .
complementary data produced as part of the cgps include mosaics of dust emission created using hires - processed iras data @xcite and of @xmath1co(@xmath2 ) emission from the fcrao outer galaxy survey @xcite .
the cgps co mosaics are the result of a reprocessing of the original ogs data , wherein correlated noise between pixels caused by the observing method was suppressed , and faint contaminating emission found in the reference positions of the survey was removed @xcite .
the reprocessed data were resampled and projected into the cgps mosaic system .
a mean brightness temperature image , averaged between lsr radial velocities @xmath3 and @xmath4 , is shown in figure [ fig : mean_tb_hi ] .
prominent in this image is a linear structure @xmath5 long , running at an angle @xmath6 north of west relative to the galactic plane . in the extreme southeast
this structure is a relatively narrow ( @xmath7 wide ) filament beginning near @xmath8 . near @xmath9 ,
the narrow filament is joined by a broader ( @xmath10 ) structure which runs parallel and to the south of it . this structure continues to broaden towards the northwest , reaching a maximum width of @xmath11 .
its northwest extremity lies near @xmath12 , where it blends into larger scale structures in the galactic plane .
the approximate centre of the linear structure is at @xmath13 ; we designate it hi [email protected]@xmath1460 , and for brevity refer to it as hi g137 in this paper .
the radial velocity width of hi g137 is @xmath15 , and there is a distinct velocity gradient along its length : velocities become more negative towards the northwest .
this velocity gradient clearly separates hi g137 from the galactic emission to the northwest ( figure [ fig : vel_gradient ] ) .
individual channel maps ( figure [ fig : channel_maps ] ) hint at the existence of linear substructures , resulting in the overall impression that hi g137 is composed of several parallel filaments .
we can detect no clear velocity structure in the direction transverse to the length of the filaments : this may be due to a combination of sensitivity and angular resolution limitations for individual filaments and confusion between them .
we are unable to make a direct measurement of the distance to hi g137 . there are no known regions in the direction of the filaments , and there are large discrepancies between kinematic and photometric distances in the galactic plane in this region in the sense that kinematic distances can be greatly over - estimated compared to the latter @xcite .
the cgps co data were searched for molecular clouds having positional and/or velocity coincidences with hi g137 .
we found no co emission near hi g137 within the velocity range of the structure . widening
the velocity range of our search revealed an irregular chain of co clouds in the southeast of hi g137 at @xmath16 extending @xmath17 in a direction perpendicular to the length of the structure .
of these , the cloud with the brightest peak co line brightness ( @xmath18 k ) , which we dub co [email protected]@xmath1448 ( co g138 for brevity ) , lies at the convergence point of the broad - structured component of hi g137 ( figure [ fig : co_clouds ] ) . the velocity difference between co g138 and hi g137 ranges between 5 and 20 .
the northern narrow filament of hi g137 appears to emerge from between a gap in the chain of co clouds . within this gap
there is a faint ( 1.3 k ) compact cloudlet at @xmath19 embedded within very faint ( @xmath20 k ) diffuse co emission .
the positional association between molecular clouds and the convergent point of the filaments suggests that hi g137 might be the result of a directed flow of dissociated gas from the molecular clouds .
if hi g137 is associated with the co clouds at @xmath21 , its distance is @xmath22 kpc , the approximate distance to the perseus arm at this longitude .
the source(s ) of the radiation and/or winds driving hi g137 will be at the same distance .
the former would have to be located on the sky such that the filaments point towards them and the co clumps lie between them and the filaments in a more or less collinear fashion .
the simbad database was used to search for o , b , and wolf - rayet type stars within @xmath23 of @xmath24 ( table [ tbl : stars ] ) . of the resulting ten stars ,
the wolf - rayet star wr 5 and the o - type supergiant hd 17603 currently have substantial stellar winds and ionizing radiation .
these two therefore merit closest attention .
wr 5 ( hd 17638 ) is a wolf - rayet star of spectral type wc6 @xcite .
it has a reddening of @xmath25 ( @xmath26 ) and a photometrically derived distance of 1.91 kpc @xcite .
hd 17603 is an o7.5ib(f ) star @xcite located @xmath27 northwest of wr 5 .
stars of type o(f ) are thought to be precursors of wolf - rayet stars .
the available ubv photometry for this star @xcite is consistent with a reddening @xmath28 ( @xmath29 ) .
the absolute magnitudes of of stars are not well determined : however assuming that this star has @xmath30 @xcite , it lies at a distance @xmath31 kpc .
both wr 5 and hd 17603 satisfy the positional criteria to be considered candidate driving sources .
in addition , both are hot , luminous objects with strong winds . for both stars ,
the apparent alignment of @xmath14co@xmath14star is very good , and these stars are likely to be the most luminous and energetic of those we have identified .
finally , their approximate distances are consistent with that proposed for hi g137 .
stellar wind parameters for both stars are listed in table [ tbl : wr5_hd17603 ] .
a final consistency check on the stellar and co cloud distances can be made by integrating the line profiles towards the stars from zero velocity to the velocity where an column corresponding to @xmath32 is found . for the wr 5 and hd 17603 sight lines ,
reddening implies an column of @xmath33 .
our data shows that this column density is reached when the line profiles are integrated from zero velocity out to @xmath34 , consistent with the radial velocity of co g138 .
in addition to the wolf - rayet and oib(f ) star discussed above , there are eight b stars within @xmath10 of the tip of hi g137 . among them
is eo per , a rapidly varying b0 supergiant .
a b0 supergiant has a mass of roughly 30 and so was approximately type o7.5 while on the main sequence .
according to @xcite , such a star has a wind terminal velocity of 2300 , slightly lower than the lower bound quoted for both wr 5 and hd 17603 , and a mass loss rate of @xmath35 , an order of magnitude smaller than the lower bound for the other two stars . at present
this star is likely to have a higher mass loss rate but much lower wind terminal velocity than wr 5 and hd 17603 , as well as a lower ionizing flux .
the remaining seven candidates have very limited observational data upon which to base determination of their type and distance . if we assume they are located at @xmath36 kpc and lie behind extinction of @xmath29 , then all these stars have photometry roughly consistent with early b type stars on the main sequence , although we can not rule out the possibility that some are actually o type stars .
based on low resolution ( @xmath37 ) observations , @xcite proposed that wr 5 is associated with an ovoidal deficiency , interpreted as a cavity , at velocities around @xmath38 .
we note that integrating out to @xmath38 in the line yields a column density only @xmath39 , corresponding to @xmath40 , which is too small to be consistent with the observed reddening to the two stars .
wr 5 and hd 17603 are well beyond the distance corresponding to that gas velocity and thus can not be associated with the ovoidal structure . our high resolution ( @xmath41 )
observations show that the distribution is highly structured in both morphology and velocity , and the `` cavity '' is by no means clearly distinguishable .
in addition , regions of depressed brightness within this `` cavity '' correspond in position and velocity with much of the co emission seen around @xmath38 , which suggests that the relative lack of emission is due at least in part to the hydrogen being in molecular rather than atomic form . for the velocity range within which we associate wr 5 and hd 17603 ( @xmath42 ) , our data do not show any convincing evidence for an shell surrounding a cavity .
however , it is possible that a gas shell does exist but is not clearly visible in the observations because of its low contrast or because parts of it are molecular .
the clear anti - correlation between the co clouds and the brightness distribution in the @xmath43 to @xmath44 velocity range and the relative lack of co emission within @xmath45 of the stars lend some support to the latter possibility .
@xcite cited evidence from iras skyflux images for a @xmath46-diameter dust shell around wr 5 .
such a shell is not obvious in the hires processed 12 , 25 , 60 , and @xmath47 m mosaics . to investigate
further , we used the 60 and @xmath47 m data to produce dust temperature and optical depth images .
the scattered dust concentrations interpreted by marston as a shell correlate much better with the velocity - integrated co data than with the velocity - integrated data , some with co emission in the velocity range around @xmath38 and others with the co around @xmath48 .
this suggests that what was interpreted as a shell is in fact the result of projection effects .
we inspected data for this region from the virginia tech spectral - line survey @xcite and the wisconsin h - alpha mapper @xcite northern sky survey .
the vtss has an angular resolution of @xmath49 but does not provide velocity information .
the wham data has a velocity resolution of @xmath50 but a lower angular resolution @xmath51 .
the vtss data suggest that wr 5 and hd 17603 both lie within a region relatively devoid of emission .
the morphology of this `` void '' and thus its possible association with the stars is uncertain because a comparison of the vtss image with velocity - integrated co emission suggests that absorption by the dust in molecular clouds strongly influences the observed brightness distribution .
we conclude that there is at present no conclusive evidence for an or dust shell / cavity structure around wr 5 or hd 17603 , although the stars appear to lie in regions relatively free of molecular , neutral atomic , and ionized gas .
any gas shell surrounding this region would have to be of low contrast or be rather inhomogeneous in its composition and structure : the latter implies a `` leaky '' shell within which a hot wind bubble might not be maintained .
the morphological relationships between the tails , the co clouds , and the stars raises the question of whether all these are physically related .
the possibility of a simple coincidence can not be ruled out , but an explanation for the morphology and velocity structure of the tails would still be needed . in the remainder of this paper
we investigate the possibility that winds from massive stars could be responsible for driving long narrow tails from compact molecular clouds in their vicinity . a viable physical model of hi
g137 must be consistent with all of the observational data we have presented above .
obviously , we need a stellar mechanism for converting molecular to neutral atomic hydrogen and accelerating the away from the molecular clumps in long narrow tails at a velocity of @xmath52 .
wr 5 and/or hd 17603 must be able to supply momentum and energy at the rates required over a time comparable to the dynamical timescale .
the molecular clumps must also be able to survive within the environment of the star(s ) for this time .
one question that immediately arises is the association of tails with only some of the co clouds in the @xmath16 velocity range .
standard galactic rotation curves predict a slope of the distance@xmath14velocity relationship at this location in the galaxy of @xmath53 .
the above velocity range thus potentially covers a line of sight range of @xmath54 and the velocity resolution of our co data ( @xmath55 ) corresponds to @xmath53 .
the empirical rotation curve of @xcite suggests the slope may be even steeper .
given the potentially very wide range of distances for the co clouds , only a small subset of the observed clouds may be within the range of influence of the tail - driving source(s ) .
additionally , flows from other co clouds may be subject to confusion if they have relatively low radial velocities .
determining the mass of extended structures is always subject to wide uncertainties , not least because of confusion with unrelated emission .
this is particularly acute for structures having large velocity dispersions .
we determined the mass of hi g137 by first removing an estimate of the background emission on a channel - by - channel basis .
this background was calculated by a linear interpolation anchored to velocity channels bracketing the velocity range of hi g137 which were judged to be free of emission from the latter .
the background subtracted cube was then integrated over a polygonal region enclosing hi g137 .
the solid angle of this region was 1.1 square degrees , and the mean column density derived was @xmath56 .
the total mass is @xmath57 and the total mass , assuming a hydrogen mass fraction of 0.77 , is @xmath58 .
the uncertainty in the mass estimate is at least @xmath59 .
the physical length of hi g137 projected onto the plane of the sky is @xmath60 pc . assuming @xmath61 kpc for hi g137 , this yields a total mass of the structure of @xmath62 g. assuming a systemic radial velocity of the mass reservoir of @xmath19 , the linear momentum contained in hi g137 is @xmath63 , where we have assumed that the space velocity of the flow can be deprojected from the radial velocity by using an inclination angle to the line of sight @xmath64 .
the mass - weighted mean outflow velocity is then @xmath65 , and the kinetic energy is @xmath66 erg .
the driving source(s ) presumably have injected momentum and energy isotropically into the ism .
the observed opening angle of hi g137 is @xmath67 : assuming that the three dimensional structure is conical and viewed at an inclination @xmath64 , the true opening angle is @xmath68 , and hi g137 subtends a fraction @xmath69 of a sphere .
therefore , @xmath70 ( @xmath71 ster ) , the maximum value occurring when @xmath72 ( when the outflow axis lies in the plane of the sky ) . over the lifetime of hi g137 ( discussed below ) , the driving source(s ) have supplied a total ( isotropic ) momentum and kinetic energy of , respectively , @xmath73 and @xmath74 , where the @xmath75 factors represent the efficiency of the transfer of momentum and energy to the .
the minimum value of @xmath76 and @xmath77 is obtained by minimizing the following two expressions with respect to @xmath64 : @xmath78 @xmath79 for @xmath80 , the minimum values are @xmath81 ( when @xmath82 ) , and @xmath83 erg ( when @xmath84 ) . a simple conical outflow model yields for the flow timescale @xmath85 yr , where we have used the mass - weighted mean outflow radial velocity @xmath86 and the mass - weighted mean angular extent @xmath87 .
we have noted above the presence of a strong uniform radial velocity gradient along the tails . if this were a manifestation of velocity - sorting rather than continuous acceleration , @xmath88 might be more accurately estimated by using the maximum outflow radial velocity observed ( @xmath89 ) rather than the mean velocity , resulting in a factor of 2 decrease in @xmath88 .
we will argue below that the velocity gradient is more likely to be the result of a steady acceleration of the tails , in which case the use of the mean velocity is approximately correct . using the dynamical timescale we can estimate the rate at which the driving source(s ) deliver ( isotropically ) momentum and kinetic energy , respectively @xmath90 and @xmath91 .
the minimum values of @xmath90 and @xmath91 are obtained by minimizing the two following expressions with respect to @xmath64 : @xmath92 @xmath93 we find @xmath94 ( when @xmath95 ) , and @xmath96 ( when @xmath97 )
. figure [ fig : dep_on_i ] shows plots of the dependence of @xmath76 , @xmath90 , @xmath77 , and @xmath91 on @xmath64 for the simple conical flow model .
the uncertainty in the values of the kinematic and energetic parameters is dominated by our lack of knowledge of the inclination angle , affecting in particular the value of the kinematic timescale . in our discussions below
we will assume that @xmath64 most likely lies within the range @xmath98 .
our estimates of the physical parameters of the tails are summarized in table [ tbl : tail_param ] .
the lowest velocity shift discernable between the co and , limited by background confusion , is @xmath99 , which is roughly comparable to the sound speed in neutral atomic hydrogen .
the systematic radial velocity gradient of hi g137 could be interpreted in terms of curvature along the line of sight or of expansion . however
, if hi g137 is the result of a flow from the co clouds , it is reasonable to assume that the filaments are radially directed away from the driving source and are thus approximately linear .
the velocity gradient would then reflect a continuous acceleration of gas along the filaments , leading to the picture of a flow embedded within an extended wind region .
this picture of long flowing tails from dense clouds driven by an external wind is similar to the situation described analytically by @xcite and numerically by @xcite of the quasi - steady hydrodynamic mixing and entrainment of dense clump gas by a diffuse subsonic or transsonic wind .
we propose that the neutral atomic tails are created in the dissociation and entrainment of molecular clouds enveloped within a strong stellar wind or winds .
the wind ablates and entrains cloud gas and accelerates it away from the cloud in long narrow tails as a mass - loaded flow . given the positional associations noted above , the source of the wind(s ) is likely to be in the vicinity of wr 5 and hd 17603 .
we note that it is very unlikely that these filaments are cometary tails caused by shadowing of radiation in a low - density region .
the flaring of the tails away from the axis of symmetry with increasing distance from the apex is inconsistent with observations @xcite and models @xcite of cometary clouds .
the dynamical timescale of hi g137 , a few million years , is far longer than the timescale for complete photoionization of the tails @xcite . this leaves the wind - driven model , but in order to establish the plausibility of this model there are several questions which must be addressed .
can the molecular clouds survive a few million years inside a stellar wind region ? what must be the conditions inside this region in order to account for the observed cloud mass loss rate and entrained gas acceleration , and how realistic are they ?
are the energetics plausible ?
what are the mechanisms by which the entrained gas may be dissociated ?
we address these questions in the following sections . for the proposed model to be feasible , the molecular clouds must survive their initial impact with the stellar wind , before settling down to a quasi - steady entrainment phase within the wind .
@xcite have performed an analytic and numerical study of the interaction between a fast , steady , planar shock and a dense isothermal , nonmagnetic , spherical , cold cloud .
more recently , @xcite have extended numerical studies to the case of shocks encountering multiple clouds in which cloud - cloud interactions can oocur . as summarized in both these papers ( we follow the formulation of @xcite in this paper ) ,
a key timescale is the `` cloud - crushing time '' @xmath100 , which is the time required for the internally transmitted shock to cross the cloud , @xmath101 . here
@xmath102 is the initial cloud radius and @xmath103 , the forward velocity of the shock through the cloud , is @xmath104 where @xmath105 is the ratio of the pressure just behind the cloud shock to that at the stagnation point and @xmath106 is the ratio of the pressure at the stagnation point to that in the far upstream wind .
both these quantities have values of roughly unity @xcite .
the velocity of the external shock is @xmath107 and the density contrast @xmath108 is the ratio of the initial cloud density to that of the external medium . in the numerical simulations of @xcite and @xcite , the cloud destruction time @xmath109 ( defined as the time at which the mass of the surviving cloud core is reduced to @xmath20 of its initial value )
was typically found to be @xmath110 .
the numerical simulations of @xcite suggest values of @xmath111 with the inclusion of dynamically significant magnetic fields .
the molecular clouds at the apex of the tails have a total mass @xmath112 .
their current radii are @xmath113 and they have h@xmath114 number densities @xmath115 . at their present stage of evolution , the clouds have gravitational virial parameters @xmath116 80150 , indicating that they are not bound by their self - gravity ( @xmath117 is the observed cloud velocity dispersion , @xmath118 , and @xmath119 is the cloud mass ) . if we assume that the total mass of the tails was originally in the molecular clouds , we find that the clouds have been reduced from their initial mass by a factor of about 5 : according to the formal definition of @xmath109 above , these clouds would be classified as largely destroyed .
if we set @xmath120 , the cloud destruction time is @xmath121 for shock velocities @xmath122 and an initial density contrast @xmath123 ( with @xmath124 a few parsec and @xmath125 ) , the clouds will survive for times of order a few @xmath126 , long enough for them to become enveloped by the wind .
we conclude that the molecular clouds can survive the passage of the initial shock , after which they are gradually destroyed while mass - loading the surrounding flow .
the mass - loading of flows by embedded clouds has been discussed in a series of papers by dyson and collaborators @xcite . for an initially pressure - confined cloud which becomes immersed in a wind flow ,
the pressure gradients across the cloud surface cause a lateral expansion of the cloud which mixes cloud gas into the wind .
@xcite show that the mass loss rate from the cloud is @xmath127 where @xmath128 is the isothermal sound speed in the cloud , @xmath129 and @xmath130 are the wind density and velocity , and @xmath131 is the mach number in the wind . in this expression
, @xmath132 is a constant of order unity for which @xcite provisionally assign a value @xmath133 .
assuming that the wind and cloud parameters ( except @xmath119 ) are roughly constant over the cloud lifetime , the cloud mass loss rate averaged over the entire cloud lifetime is @xmath134 , where @xmath135 is the maximum mass loss rate corresponding to the initial mass @xmath136 of the cloud .
therefore , @xmath137 where @xmath138 is the number density of the cloud ( we have assumed a mean mass per particle of @xmath139 ) . for simplicity , we assume that the initial cloud mass ( the current molecular cloud mass plus the mass in the tails ) was contained within a single cloud , thus @xmath140 .
the relative fraction of the initial mass in the tails and the clouds indicates that @xmath141 , where @xmath142 is the cloud lifetime predicted by integrating the cloud mass loss rate .
we thus appear to be observing the clouds near the half - way point in time of their destruction process .
the mass loss rate averaged over this time period ( the dynamical timescale @xmath88 ) is @xmath143 , roughly twice @xmath144 . using our value of @xmath136
, we obtain @xmath145 where we have assumed @xmath146 ( i.e. @xmath147 ) .
we have observed a value @xmath1481.3 3.9@xmath149 ( for @xmath150 ) , which implies that we can match the theoretically predicted mass loss rate if @xmath151 in the theory of @xcite , the isothermal sound speed @xmath128 is better identified as the characteristic speed at which the dense cloud can supply mass to the wind - cloud mixing region .
thus , we will assume that the effective value of @xmath128 is @xmath152 .
then , the above condition becomes @xmath153 and has a value of @xmath154 when @xmath155 .
the wind velocity experienced by the dense clouds can not be less than the maximum velocity observed in the tails ( @xmath15640 , depending on @xmath64 ) .
the maximum possible wind speed is however much larger : the winds of massive main sequence stars can have terminal velocities @xmath157 ( cf . table [ tbl : wr5_hd17603 ] ) . if @xmath158 , the mass - loading model predicts @xmath159 .
such large values of @xmath138 rule out the possibility that the tails are produced in the mass - loading of clouds immersed in a freely - expanding bare stellar wind .
direct momentum - driving of the tails by a wind is also ruled out : the total isotropic linear momentum supplied by the wind of a 4085 star during its main sequence lifetime of a few @xmath160 is approximately @xmath161
@xmath162 @xcite , which is at least two orders of magnitude smaller than @xmath163 .
the isotropic momentum flux predicted by the mass - loading model is @xmath164 where substituting our adopted range of values of @xmath165 leaves only one factor of @xmath130 ( the separation between the driving source(s ) and the co clouds is @xmath166 , where we have assumed @xmath167 ) .
equating @xmath168 with our observed value of @xmath169 , we see that wind speeds in the mass - loading region greater than @xmath170 ( @xmath171 ) would be sufficient to account for the momentum flux .
given that velocities in hi g137 may be as high as @xmath172 , @xmath130 must be significantly higher than the limit of @xmath171 we have obtained .
we note that the predicted value of @xmath165 in the mass - loading theory is very sensitive to the value of the `` mixing efficiency '' @xmath132 ( @xmath173 )
. the mixing efficiency may well be smaller @xcite .
lowering the value of the mixing efficiency by a factor of 2 or 3 would increase the required wind speeds in the mass - loading region by a factor of @xmath171 , to @xmath174 . using this higher value ,
the mass - loading theory predicts a wind density in the mass - loading region of @xmath17540 .
we conclude that the mass - loading scenario is workable in the context of the main sequence stellar winds from massive o stars if the fast low - density stellar wind is significantly loaded with mass and greatly slowed before interacting with the dense cloud .
the low inferred wind velocity is consistent with the relative narrowness of the tails as predicted theoretically @xcite and found numerically @xcite for the case of mass loading by a mildly supersonic ( transsonic ) wind .
a related point concerns the apparent complete mixing of mass injected into the wind before encountering the co clouds : in contrast the filaments clearly do not completely mix over a large distance . mixing may be greatly enhanced in a highly inhomogeneous medium having numerous clumps which are close enough togther for wind - driven interactions to occur @xcite .
in contrast , the co clouds associated with the filaments are likely to be the dominant mass concentrations in their vicinity and so are not subject to mixing driven by cloud@xmath14cloud interactions .
our model still requires that the wind transfer very large amounts of kinetic energy to the tails : several times @xmath176 , which is the same order of magnitude as the mechanical luminosity of the main sequence wind of a massive o star .
this is a severe constraint , since in the basic theory of adiabatic stellar wind bubbles , only a tiny fraction , @xmath177 , of the energy is in kinetic form within the bubble , the bulk being in the thermal energy of hot shocked gas and in the kinetic energy of the swept - up shell .
therefore a very small fraction of the energy is available to drive mass - loading of clouds trapped inside .
much higher efficiencies of energy conversion into kinetic form are possible if the stellar wind is itself significantly mass - loaded . @xcite
have shown that mass - loading of an initially highly supersonic wind by mass sources distributed smoothly throughout the region around the star can reduce the mach number of the wind without having the wind going through a global shock near the star .
if the mass loading is sufficiently heavy ( @xmath178 , where @xmath179 is the mass loaded into the wind and @xmath180 is the mass of the wind injected by the star ) , a large fraction of the stellar wind energy ( @xmath181 ) will remain in the form of kinetic energy within the wind region @xcite .
therefore , stellar winds which are heavily `` pre - loaded '' with mass may be able to provide the kinetic energy required to drive flows from larger - scale mass concentrations .
it also may potentially explain the low wind velocities and high wind densities that we infer .
in this regard , a scaled - up version of the scenario modelled numerically by @xcite , the mass - loading of the wind from the trapezium cluster by propylds , may be relevant .
@xcite showed that mass - loading could have very significant dynamical effects on the wind , including increased wind density and low mach numbers at large radii .
however , even with a high efficiency of momentum transfer to the filaments , the momentum input of the winds of wr 5 and hd 17603 integrated is too low by at least an order of magnitude .
it would seem then that our scenario requires the wind input from a group of many more o stars in addition to these two .
we have not been able to identify such a grouping of stars . since massive o stars form in ob clusters , short - lived stars such as wr 5 and hd 17603
do not have time to travel far from their birthplaces ( unless they are runaway stars ejected from a binary during the supernova explosion of the other component ) .
although it is not satisfactory to have to posit the existence of an unseen group of ob stars , the apparent close proximity of wr 5 and hd 17603 to each other makes this idea more plausible .
given the effects of visual extinction ( and possible anomalous extinction ) in a direction towards a major spiral arm , and the difficulties of finding , classifying , and obtaining distance estimates , it is perhaps not unlikely that such a grouping of stars could have escaped detection
. given the dynamical timescale of the filaments , it is quite possible that the parent molecular cloud in which the group formed has been largely destroyed by now , and some of its members no longer exist after having gone supernova ( we could speculate that supernova explosions might have contributed to the flow driving ) .
such a remnant grouping of stars might be very difficult to identify .
there are two plausible mechanisms for dissociating the molecular gas which forms the tails : photodissociation and shock dissociation .
stars like wr 5 and hd 17603 produce very large fluxes of ionizing photons ( @xmath182 ) , and thus the ionized skins of the molecular clouds may be observable in .
the brightness can be estimated using , e.g. , equation @xmath183 of @xcite , where we assume a cloud radius 2 , a ratio of ionized skin thickness to cloud radius of @xmath184 , and a source of ionizing photons @xmath185 located @xmath186 from the cloud .
the latter was estimated using equation @xmath187 in @xcite . the resulting surface brightness observed from the earth , assuming no extinction , corresponds to a photon specific flux of @xmath188 .
this is comparable to the surface brightnesses of the fainter photoevaporating cometary knots observed in the helix nebula by @xcite .
if the clouds suffer a similar extinction as wr 5 , the emission will be further attenuated by a factor of @xmath189 .
the large ( @xmath190 ) and patchy visual obscuration may make detection of emission from the clouds difficult ( see discussion in section [ sec : environ ] ) .
there is no evidence for radio continuum emission from ionized hydrogen from the cgps 1420 continuum data .
any region must thus be of very low radio surface brightness : a 100 path length through a 1 region would produce only a few tenths of a kelvin brightness temperature , which could not easily be distinguished from the general galactic background of several kelvin .
hot massive stars produce photons capable of dissociating molecular hydrogen at rates comparable to the ionizing photon rates @xcite . a rough evaluation of the possibility that the tails are photodissociated can be made by comparing the photodissociating rate into the solid angle of the tails , @xmath191 , to the dissociation rate implied by our observations .
@xmath192 is a typical stellar photodissociating photon rate for a very massive o star , @xmath193 is the probability that an absorption will lead to dissociation @xcite , and @xmath194 is the fractional solid angle subtended by the tails . we find that the two rates are roughly equal at @xmath195 dissociations when integrated over a period of a few @xmath126 .
photodissociation thus appears to be a plausible dissociation mechanism for the tails .
the other possibility is that shocks driven into the molecular clouds by the stellar wind(s ) may be capable of dissociating the h@xmath114 .
j - type shocks , in which the shock velocity @xmath103 is sufficiently high to preclude the existence of a magnetic precursor , are the most effective at dissociating molecular gas . in a strong j - type shock with @xmath196 moving into molecular gas with a pre - shock h@xmath114 number density @xmath197 , about @xmath198 of the h@xmath114 will be dissociated @xcite . for @xmath199 and a wind velocity @xmath130 of the order of a few hundred
, shock velocities will be of the order of @xmath200@xmath201 , which is sufficient to result in very significant dissociation .
the energy required to shock dissociate the molecular gas is roughly an order of magnitude smaller than that to accelerate it to the velocity of the tail .
however , shock velocities of this magnitude may reduce the cloud destruction time to values @xmath202 , which is a problem for the scenario we are proposing .
we conclude that photodissociation is the most workable mechanism for dissociating the molecular gas swept into the filaments .
we have discovered an unusual system of `` tails '' comprised of a number of linear or quasi - linear thin filaments which , if projected along their length , appear to have a common origin .
the feature , which we dub hi [email protected]@xmath1460 , has dimensions of @xmath203 ( length @xmath204 width @xmath204 velocity ) , although its width varies along its length .
located at one end of the filaments is a chain of molecular clouds that lie at a distance of approximately 2 . despite the molecular material being displaced from the atomic by @xmath205 , we propose that the molecular clouds and feature are located at the same distance and are dynamically related .
the lack of a traditional head - tail morphology , and the lack of any evidence of ionization along the leading edge of the molecular clouds , lead us to reject the possibility that this system is simply a cometary cloud ; instead , we propose that the feature is made up of material dissociated from the molecular clouds and entrained in the outflowing wind as a mass - loaded flow .
given the velocity difference between the atomic and molecular material , we estimate the dynamical age of the system to be on the order of a few @xmath126 years . the velocity gradient observed along the feature is , in our model , explained by continuous acceleration of the entrained material by the stellar wind .
a search for driving sources for the filaments identified the wolf - rayet star wr 5 and the of star hd 17603 as candidates .
both these stars are located near the apparent origin of the filaments , are approximately at the same distance as the molecular clouds , and have main sequence lifetimes comparable to the dynamical age of the system .
however , even with the high efficiency of momentum transfer predicted for a heavily mass - loaded wind , wr 5 and hd 17603 are incapable on their own to drive the filaments : we are forced to speculate that other , unidentified stars and/or old supernova explosions must also have been involved .
we can not of course discount the possibility that the positional associations between , co , and stars are simply accidental .
but , assuming that stellar winds ( or explosions ) have shaped the gaseous structures we have observed , we suggest that anisotropic structures similar to hi g137 should be searched for in the environments around massive stars ( and supernova remnants ) .
l l l 21 cm line & spectral resolution & 1.32 + & channel width & 0.82 + & angular resolution & 64 + & & + @xmath1co(@xmath2 ) line & spectral resolution & 0.98 + & channel width & 0.82 + & angular resolution & 100 + l l l l l wr 5 & wc6 @xmath206}$ ] & 138.87 & @xmath207 & + hd 17603 & o7.5ib(f ) @xmath208}$ ] & 138.77 & @xmath209 & + eo per & b0iab : e & 138.50 & @xmath210 & @xmath211 listed ra / dec differ slightly + cds 314 & b & 138.50 & @xmath210 & @xmath212 for these two stars + bd+56722 & b & 138.57 & @xmath213 & + als 7516 & b & 138.80 & @xmath209 & + als 7523 & b & 138.87 & @xmath214 & + zz per & b & 138.84 & @xmath215 & + als 7453 & b & 138.22 & @xmath216 & + als 7470 & b & 138.16 & @xmath217 &
+ l l l spectral type & wc6 @xmath218}$ ] & o7.5ib(f ) @xmath219}$ ] + v@xmath220 & 2365 @xmath206}$ ] & 2500 @xmath221}$ ] + & 10@xmath222 @xmath208}$ ] & 10@xmath223 @xmath224}$ ] + @xmath225 & @xmath226 & - + + ms spectral type @xmath227}$ ] & o5 & o5 + ms v@xmath220 & 2800 & 2800 + ms & 10@xmath228 & 10@xmath228 + ms @xmath225 & @xmath229 erg s@xmath230 & @xmath229 erg s@xmath230 + ms lifetime & 3.7 myr & 3.7 myr + ms wind energy & @xmath231 erg & @xmath231 erg + l l l average column density in hi g137 & @xmath232 & @xmath233 @xmath234 + total mass of hi g137 & @xmath235 & @xmath236 + flow time scale for hi g137 & @xmath88 & @xmath237 yr + linear momentum in hi g137 & @xmath238 & @xmath239 + kinetic energy in hi g137 & @xmath240 & @xmath241 erg + minimum total kinetic energy from driving sources & @xmath242 & @xmath243 erg + minimum total momentum from driving sources & @xmath163 & @xmath244 g cm s@xmath230 + minimum rate of kinetic energy delivery & @xmath245 & @xmath246 erg s@xmath230 + minimum rate of momentum delivery & @xmath247 & @xmath248 g cm @xmath249 +
|
we report the discovery of a system of unusual filaments which appear to be associated with molecular clouds in the perseus spiral arm of our galaxy .
we investigate the hypothesis that this system is the result of a directed flow of dissociated gas from clouds trapped within an extended wind flow from massive stars .
the wolf - rayet star wr 5 and the oib(f ) star hd 17603 are identified as candidate driving sources .
however , an examination of this hypothesis within the context of the theory of mass - loaded winds shows that these two stars alone can not account for the energetics and kinematics of the required spherically symmetric wind flow . unless the apparent association between , molecular gas , and stars is an accidental one
, we suggest that other as - yet unidentified stars must have contributed to driving the filaments .
|
main - sequence stars with masses greater than about @xmath4 follow two evolutionary paths ; either they form a degenerate core of o / ne / mg , or a degenerate fe core , which , after undergoing a type ii supernova core collapse , forms a proto - neutron star @xcite .
neutron stars ( nss ) are also expected to form through the accretion - induced collapse of a white dwarf @xcite . at birth
, nss are rapidly and differentially rotating , which makes them subject to various types of instabilities . among these ,
the dynamical bar - mode instability and the shear - instability are particularly interesting because of their potential role as sources of gravitational waves ( gws ) . indeed
, a newly born ns may develop a dynamical bar - mode instability when the rotation parameter @xmath5 , with @xmath6 the rotational kinetic energy and @xmath7 the gravitational binding energy , exceeds a critical value @xmath8 ( see , for instance , @xcite for some reviews ) . under these conditions ,
the rapidly rotating ns can become severely deformed , leading to a strong emission of gws in the khz range .
analytic investigations of the conditions under which these dynamical instabilities develop in self - gravitating rotating stars can be found in @xcite , but these are inevitably restricted to newtonian gravity or to simple shell models . to improve our understanding of these instabilities also in the nonlinear regimes , and to be able to extract useful physical information from the potential gw emission ,
it is clear that a general - relativistic numerical modeling is necessary .
this has been the focus of a number of recent works , e.g. , @xcite , which have provided important clues about the threshold for the instability and its survival under realistic conditions . as an example , for a polytropic relativistic star with polytropic index @xmath9 , the calculations reported in @xcite revealed that the critical value is @xmath10 , and that a simple dependence on the stellar compactness can be used to track this threshold from the newtonian limit over to the fully relativistic one @xcite .
furthermore , numerical simulations have also revealed that the instability is in general short - lived and that the bar - deformation is suppressed over a timescale of a few revolutions ( this was first pointed out in ref .
@xcite and later confirmed in ref .
@xcite , where it was interpreted as due to a faraday resonance ) .
one aspect of the bar - mode instability that so far has not received sufficient attention is about the occurrence of the instability in magnetized stars .
this is not an academic question since nss at birth are expected to be quite generically magnetized , with magnetic fields that have strengths up to @xmath11 g in ordinary nss and reaching strengths in excess of @xmath12 g in magnetars , if instabilities or dynamos have taken place in the proto - neutron star phase @xcite .
magnetic fields of this strength can affect both the structure and the evolution of nss @xcite , and it is natural to expect that they will influence also the development of the instability when compared to the purely hydrodynamical case . a first dynamical study in this direction has been carried out recently in ref .
@xcite , where the development of the dynamical bar - mode instability has been studied for differentially rotating magnetized stars in newtonian gravity and in the ideal - magnetohydrodynamics ( mhd ) limit ( i.e. , with a plasma having infinite conductivity ) .
not surprisingly , this study found that magnetic fields do have an effect on the development of the instability , but that this is the case only for very strong magnetic fields .
we here consider the same problem , but extend the analysis to a fully general - relativistic framework , assessing the impact that the results have on high - energy astrophysics and gw astronomy .
our investigation of the dynamics of highly - magnetized and rapidly rotating nss is also part of a wider study of this type of objects to explain the phenomenology associated with short gamma - ray bursts .
these catastrophic phenomena , in fact , are normally thought to be related to the merger of a binary system of nss @xcite , which could then lead to the formation of a long - lived hypermassive ns ( hmns ) @xcite . if highly magnetized , the hmns could then also lead to an intense electromagnetic emission @xcite .
this scenario has recently been considered in refs .
@xcite , where numerical simulations of an axisymmetric differentially rotating hmns were carried out .
the hmns had initially a purely poloidal magnetic field , which eventually led to a magnetically driven outflow along the rotation axis .
a similar setup has also been considered in a number of works , either in two - dimensional ( 2d ) @xcite or in three - dimensional ( 3d ) simulations @xcite , with the goal of determining whether or not the conditions typical of a hmns can lead to the development of the magnetorotational ( mri ) instability @xcite .
although this type of simulations in 3d still stretches the computational resources presently available , the very high resolutions employed in ref .
@xcite , and the careful analysis of the results , provided the first convincing evidence that the mri can develop from 3d configurations .
this has of course important consequences on much of the phenomenology associated with hmnss , as it shows that very strong magnetic fields , up to equipartition , will be produced in the hmns if this survives long enough for the mri to develop . in the simulations reported here we necessarily adopt much coarser resolutions and
hence we will not be able to concentrate our attention on the development of the mri .
rather , we will here extend our previous work on the dynamical bar - mode instability @xcite also to the case of magnetized stars , determining when and how magnetic fields can limit the development of the dynamical bar - mode instability .
our initial models correspond to stationary equilibrium configurations of axisymmetric and rapidly rotating relativistic stars .
more precisely , our initial models are described by a polytropic eos with adiabatic index @xmath13 and are members of a sequence with a constant rest - mass of @xmath14 and a constant amount of differential rotation .
interpreting the results of our simulations can be rather straightforward . because we work in the ideal - mhd limit , the magnetic field lines are `` frozen '' in the fluid and follow its dynamics ( see @xcite for a recent extension of the code to resistive regime ) . as a consequence , differential rotation drives the initial purely poloidal magnetic field into rotation , winding it up and generating a toroidal component . at early times
, the toroidal magnetic field grows linearly with time , tapping the ns s rotational energy . at later times , the growth starts deviating from the linear behavior and the magnetic tension produced by the very large magnetic - field winding , alters the angular velocity profile of the star . depending on the models adopted and the initial magnetic field strength , the magnetic winding could become the most efficient mechanisms for redistributing angular momentum , with the mri being the dominant one when the alfvn timescale becomes comparable to the magnetic winding timescale .
overall , we find that if the initial magnetic fields are @xmath15 g , then they have a negligible effect on the occurrence of the dynamical bar - mode instability , which develops in close analogy with the purely hydrodynamical case . on the other hand ,
if the initial magnetic fields are @xmath3 g , they can suppress the instability completely .
note that the precise threshold marking the stability region depends not only on the strength of the magnetic field , but also on the amount of rotation .
we trace this threshold by performing a number of simulations of a number of sequences having the same parameter @xmath16 but different magnetizations
. an important consequence of our results is that because the instability is suppressed in strongly magnetized nss , these can no longer be considered as potential sources of gws , at least via the dynamical bar - mode instability .
the organization of the paper is as follows . in sect .
[ sec : evolution ] we describe the numerical methods and the setup employed in our simulations , as well as the full set of equations we solve . in sect .
[ sec : initial ] we mention briefly the main properties of the stellar models adopted as initial data , together with the simplifications and assumptions we make . in sect .
[ sec : results ] we examine in great detail the effects of the presence of an initial poloidal magnetic field on differentially rotating stars covering a wide range in the parameter space .
we further discuss the qualitative and quantitative features of the evolution of models with the same total rest mass but different rest - mass density and angular momentum profiles that are known to be unstable to the purely hydrodynamic bar - mode instability .
finally , we investigate whether magnetic fields affect the stellar evolution even when the bar - mode instability does not develop .
our conclusions are drawn in sect .
[ sec : conclusions ] and two appendices discuss the influence of symmetries on the development of the instability and the convergence of our results . unless stated differently , we adopt geometrized units in which @xmath17 , @xmath18 , @xmath19 .
the simulations have been carried out using the general - relativistic ideal - mhd ( grmhd ) code ` whiskymhd ` @xcite .
the code provides a 3d numerical solution of the full set of the grmhd equations in flux - conservative form on a dynamical background in cartesian coordinates .
it is based on the same high - resolution shock - capturing ( hrsc ) techniques on domains with adaptive mesh refinements ( amr ) @xcite ) as discussed in @xcite .
the reconstruction method adopted is the one discussed in the piecewise parabolic ( ppm ) @xcite , while the harten - lax - van leer - einfeldt ( hlle ) approximate riemann solver @xcite has been employed in order to compute the fluxes .
the divergence of the magnetic field is enforced to stay within machine precision by employing the flux - cd approach as implemented in @xcite , but with the difference that we adopt as evolution variable the vector potential instead of the magnetic field .
this method ensures the divergence - free character of the magnetic field since the magnetic field is computed as the curl of the evolved vector potential using the same finite - differencing operators as the ones for computing the divergence of the magnetic field . because of the gauge invariance of maxwell equations
, a choice needs to be made and we have opted for the simplest one , namely , the algebraic maxwell gauge .
this choice can introduce some spurious oscillations close to the amr boundaries in highly dynamical simulations , but this has not been the case for the simulations reported here . on the other hand ,
it has allowed us to keep the divergence of the magnetic field essentially nearly at machine precision .
a more advanced prescription has been also introduced recently in ref .
@xcite ; this approach requires a certain amount of tuning for optimal performance and will be considered in future works .
additional information on the code can also be found in refs .
@xcite .
furthermore , to remove spurious post - shock oscillations in the magnetic field we add a fifth - order kreiss - oliger type of dissipation @xcite to the vector potential evolution equation with a dissipation parameter of @xmath20 .
finally , the evolution of the gravitational fields is obtained through the ` ccatie ` code , which provides the solution of the conformal traceless formulation of the einstein equations @xcite .
the time integration of the evolution equations is achieved through a third - order accurate runge - kutta scheme .
essentially all of the simulations presented in this paper use a 3d cartesian grid with four refinement levels and with outer boundaries located at a distance @xmath21 km from the center of the grid .
the finest resolution is @xmath22 km ( between @xmath23 and @xmath24 points across the stellar radius , depending on the model ) and the coarsest extends up to about @xmath21 km , namely more than five times the stellar radius .
unless stated differently , all of the simulations discussed hereafter have been performed imposing a bitant symmetry , i.e. , a reflection symmetry across the @xmath25 plane . for convenience
we report here the full set of the evolution equations we solve numerically which consists in the coupled systems of einstein and mhd equations , i.e. , @xmath26 where @xmath27 and @xmath28 are the ricci tensor , the metric tensor and the ricci scalar , respectively . on the electromagnetic side
, @xmath29 is the maxwell tensor , dual of the faraday tensor @xmath30 , @xmath31 is the current four - vector , and on the matter side @xmath32 is the rest - mass density , @xmath33 is the 4-velocity of the fluid satisfying the normalization condition @xmath34 .
the total energy - momentum tensor @xmath35 is the linear combination of the contributions coming from a perfect fluid , i.e. , @xmath36 , and from the electromagnetic fields , i.e. , @xmath37 @xmath38 where @xmath39 in the expressions above we recall that @xmath40 is the specific enthalpy , @xmath41 the specific internal energy .
hence , the energy density in the rest - frame of the fluid is just @xmath42 . at the same time
, the four - vector @xmath43 represents the magnetic field as measured in the comoving frame , so that the maxwell and faraday tensors are expressed as ( see @xcite for details ) @xmath44 where the second equalities introduce the electric and magnetic fields measured by an observer moving along a normal direction @xmath45 .
we further note that the @xmath46 terms appearing in eqs . , are absorbed in the definition of the magnetic field . in the interest of compactness
, we will not discuss here the detailed formulation of the eqs .
we use in the numerical solution and refer the interested reader to the following works where these aspects are discussed in detail : ref .
@xcite for the formulation of the einstein equations and the gauge conditions used , refs .
@xcite for the formulation of the mhd equations and the strategy for enforcing a zero divergence of the magnetic field , refs .
@xcite for the computational infrastructure and the numerical methods used .
what is however important to remark here is that we employ an `` ideal - fluid '' ( or gamma - law ) equation of state ( eos ) @xcite @xmath47 where @xmath48 is the adiabatic exponent , which we set to be @xmath9 .
more realistic eos could have been used , as done for instance in ref .
@xcite , and this will indeed be the focus of future work . at this stage , however , and because this is the first study of this type , the simpler analytic eos is sufficient to collect the first qualitative aspects of the development of the instability .
determining the convergence properties of our simulations is of course an essential validation of the results presented and a considerable effort has been put into performing these measures within the numerical setup used here . lacking an analytic solution that describes the fully nonlinear development of the bar , we can only perform self - convergence tests at this stage .
the results will be discussed below . however , there is a regime in our calculations in which we can exploit the knowledge of an analytic solution and this refers to the initial shearing of the poloidal magnetic field by the differentially rotating star .
it is in fact not difficult to show that within an ideal - mhd framework the induction equation predicts a growth of the toroidal magnetic field which is linear in time ( see , for instance , @xcite for a pedagogic presentation of the perturbed induction equation ) . to explore this regime
we have performed a large number of simulations of model ` u11 - 1.0e15 ` with varying resolution and monitored the growth of the square root of the toroidal magnetic energy @xmath49 [ cf .
, eq . ( [ eq : emag_tor ] ] ; we recall that the poloidal magnetic field is not expected to grow during this stage ( cf . , sect .
[ subsec : untablemodels ] ) . figure [ fig : grow_toroidal_res ] reports the results of these simulations relatively to the first @xmath50 ms , with different curves referring to different resolutions .
it is then evident that the curves are getting closer and closer to straight lines as the resolution increases . to measure whether a linear - in - time - growth is actually reached
we have actually computed the growth rate `` @xmath51 '' by fitting the square root of the magnetic energy with a trial function which is a power - law in time with undetermined growth rate , i.e. , with @xmath52 where the time interval has been selected to be between @xmath53 to @xmath54 ms . for three different resolutions . _
top panel : _ order of the self - convergence test , @xmath55 , shown as a function of time .
note that a convergence order around @xmath56 is measured before the bar - mode instability develops and shocks are produced ( gray - shaded area).,scaledwidth=48.0% ] also reported in the inset of fig .
[ fig : grow_toroidal_res ] are the values of @xmath51 ( colored symbols ) as a function of the resolution @xmath57 , as well as a fit for @xmath58 ( dashed line ) when assuming a second - order convergence with resolution , i.e. , assuming @xmath59 ( the point for @xmath60 km has been excluded from the fit ) . having made this assumption , we do find that the growth rate is in very good agreement with the one expected in this linear regime , with @xmath61 .
of course this result does not prove directly that we have second - order convergence over this period of time .
however , what it does prove is that if a second - order convergence is assumed , then our solution matches the expected perturbative one .
next we consider a more general calculation of the convergence order by performing again simulations of model ` u11 - 1.0e15 ` for a range of resolutions .
this time our results for the convergence are obtained by taking into account the data corresponding to the whole timescale of the simulations , i.e. , @xmath62 ms . also in this case we monitor the growth of the toroidal magnetic energy @xmath49 and report in the top panel of fig . [ fig : grow_toroidal_res_2 ] its evolution for three runs at resolutions : @xmath63 , and @xmath64 km , respectively .
the bottom panel of the same figure reports instead the convergence order @xmath55 , computed via a self - convergence test @xcite , when shown as a function of time . in this case
it is then possible to recognize that the code does indeed converge at around second order during the linear growth stage ( i.e. , for @xmath65 ms ) , in agreement with the results found in purely hydrodynamical simulations @xcite , or with the new resistive code @xcite .
however , as the bar - mode instability develops , the second - order convergence is lost and the convergence order reduces to one .
this is not surprising as the development of the bar also leads to the formation of shocks , which necessarily degrade our solution to a first - order convergence .
we also note that the large variations in the convergence order shown in the gray - shaded area of fig .
[ fig : grow_toroidal_res_2 ] ( i.e. , for @xmath66 ms ) are simply the consequence of the fact that the instability starts growing at different times for different resolutions and this inevitably leads to large excursions in @xmath55 .
because all the major considerations made about the onset and development of the bar deformation , as well as the estimates for the growth rates and frequencies , are obtained after looking at the first 20 ms of the evolution , we conclude that all of our results have been achieved with solutions converging at the expected rates .
|
we present three - dimensional simulations of the dynamical bar - mode instability in magnetized and differentially rotating stars in full general relativity .
our focus is on the effects that magnetic fields have on the dynamics and the onset of the instability . in particular , we perform ideal - magnetohydrodynamics simulations of neutron stars that are known to be either stable or unstable against the purely hydrodynamical instability , but to which a poloidal magnetic field in the range of @xmath0@xmath1 g is superimposed initially . as expected , the differential rotation is responsible for the shearing of the poloidal field and the consequent linear growth in time of the toroidal magnetic field .
the latter rapidly exceeds in strength the original poloidal one , leading to a magnetic - field amplification in the the stars .
weak initial magnetic fields , i.e. , @xmath2 g , have negligible effects on the development of the dynamical bar - mode instability , simply braking the stellar configuration via magnetic - field shearing , and over a timescale for which we derived a simple algebraic expression . on the other hand , strong magnetic fields , i.e. , @xmath3 g , can suppress the instability completely , with the precise threshold being dependent also on the amount of rotation . as a result ,
it is unlikely that very highly magnetized neutron stars can be considered as sources of gravitational waves via the dynamical bar - mode instability .
|
type ia supernovae are generally associated with the thermonuclear explosion of earth - sized white dwarf ( wd ) stars the final stages of the evolution of small and intermediate - mass stars .
these dense objects consist of carbon and oxygen and their pressure is dominated by degenerate electrons . in a binary system , the wd can gain mass from its companion by accretion .
however , there exists a fundamental limit , the chandrasekhar mass , beyond which the star is unstable against gravitational collapse . before reaching this stage , the density at the center of the wd reaches values at which nuclear reactions from carbon to heavier elements
this establishes a stage of convective carbon burning which is finally terminated when a thermonuclear runaway forms a flame .
the thermonuclear flame propagates outward and may give rise to an explosion of the star @xcite .
the process resembles premixed chemical combustion where two solutions supersonic detonation and subsonic deflagration are admissible . from observational constraints
it is clear that the flame starts out as a subsonic deflagration .
the ashes are buoyant and shear instabilities generate strong turbulence .
the interaction of the flame with turbulence is a key feature in the model and accelerates the flame .
consequently , one of the outstanding problems is to determine , whether this turbulent flame leads to explosions energetic enough to be consistent with observations , or whether a later transition to a supersonic detonation is both possible and necessary . in the freely expanding medium of sn ia explosions , such transitions are difficult to achieve .
the exact way of flame ignition is not well determined yet , but previous numerical studies @xcite suggest that the ignition process may proceed on only one side of the star s center in a few patches . while it has been shown that ignitions in multiple spots distributed around the center of the wd lead to strong explosions @xcite , one - sided ignitions are expected to release less energy .
the effects of the latter scenario on the explosion process are investigated in the present study .
it has been speculated that collisions between gravitationally bound laterally expanding regions of fuel and ash could , in principle , give rise to a `` gravitationally confined detonation '' @xcite .
this possibility is studied in detail .
the code used for simulating the thermonuclear supernova explosions on scales of the wd star has been described in detail previously @xcite . since it is not possible to resolve all relevant spatial scales of the problem , an approach similar to large eddy simulations
was chosen .
the modeling of the hydrodynamics is based on the prometheus @xcite implementation of the piecewise parabolic method @xcite .
our flame model consists of three components .
flame propagation is followed via a level - set approach .
it associates the flame with the zero level set of a signed distance function @xmath0 which is evolved in a suitable way @xcite to account for the propagation .
this flame prescription , however , does not resolve the scales that determine the propagation velocity . in the flamelet regime of turbulent combustion , which applies to the situation in thermonuclear supernovae ( possibly with the exception of the very late stages , e.g. @xcite )
the flame velocity is proportional to the turbulent velocity fluctuations .
these are determined from a turbulent subgrid - scale model @xcite featuring localized closures to take into account the non - isotropic nature of the problem .
the energy release due to nuclear burning is handled in a simplified way .
only five species are considered : carbon and oxygen of the initial wd material , @xmath1 mg representing intermediate mass elements , alpha particles , and @xmath2ni as a representative of iron group nuclei . at
fuel densities higher than @xmath3 , the material crossed by the flame is burned to nuclear statistical equilibrium , modeled as a mixture of alpha particles and @xmath2ni .
below , this density burning is incomplete and produces intermediate mass elements .
the appropriate amounts of energy are released in the ashes . at densities below @xmath4
the reactions cease .
the code is efficiently parallelized on the basis of domain decomposition in an mpi implementation . while two - dimensional simulations can be run easily on a multi - processor desktop computer , three - dimensional simulations and require a supercomputer .
the code has been used on several machines and proved good scaling behavior .
the three - dimensional simulations described here were carried out on a @xmath5 ^ 3 $ ] cells computational grid required about 20,000 cpu hours each on the nccs jaguar cray xt3 supercomputer .
simulations of this size were performed on 256 processors .
evolution of an explosion simulation with the flame ignition in a single bubble of radius @xmath6 displaced @xmath7 from the center of the wd .
top left panel : initial evolution of the flame front ( blue to green isosurfaces correspond to @xmath8 \ , \mathrm{s}$ ] ) .
other panels : later evolution with the logarithm of the density volume rendered and @xmath9 as blue isosurface indicating the flame front or , later , the approximate boundary between burnt and unburnt material.,scaledwidth=80.0% ] in our model , the flame is artificially ignited .
therefore the initial flame configuration is an undetermined parameter .
since it is not constrained tightly by complementary studies , several numerical experiments with asymmetric ignitions have been performed .
we considered initial flame configurations in form of a single spherical bubble placed at different radii off - center of the wd , more complex bubble - substructures and perturbations from perfect sphericity , and ( on average ) teardrop - shaped initial flames extending down to the center of the star or even overshooting it . in fig .
[ fig : evo ] , the deflagration flame was ignited in a single spherical bubble 200 km off - center .
this bubble is filled with hot ashes of lower density than the surrounding fuel and therefore is buoyant .
as it burns and floats , the flame shape evolves from a sphere to a torus in the first tenths of a second ( upper left panel of fig . [
fig : evo ] ) .
continuing towards the surface , the torus becomes distorted by growing features that eventually connect .
the energy released during the flame s transit from the center to the surface is too small to explode the wd , which remains gravitationally bound .
once the bubble of ash breaks out , it starts to sweep around the core , colliding on the diametrically opposite side of the star . in the collision ,
the temperature and the density of unburnt material are increased .
the question is whether this increase is sufficient to initiate a detonation .
a detonation wave would propagate inwards burning the core of the star and leading to an explosion . clearly , the strength of the collision depends upon the initial parameters mentioned above , all of which alter the energy release in the nuclear burning and hence the expansion of the wd prior the the surface material collision .
these parameters were tested in about 20 two - dimensional simulations . since such simulations impose an artificial cylindrical symmetry and , moreover , turbulence behaves differently in 2d and 3d simulations , neither the nuclear energy release nor the collision strength are expected to be realistic .
therefore , the 2d simulations merely provide a way of exploring the dependencies .
five three - dimensional simulations were carried out in addition to assess what might realistically happen in a supernova .
maximum fuel temperature reached in the region of collision of the material sweeping around the wd as a function of the nuclear energy release prior to reaching that temperature .
the 3d data point is marked , all other points correspond to 2d simulations.,scaledwidth=70.0% ] initiation of a detonation wave in the collision region depends on the temperature reached , the density , and the spatial extent of the collision region . for our 2d simulations
ignited in a spherical bubble , an inverse correlation exists between nuclear energy release prior to collision and the maximum temperature reached ( cf.fig .
[ fig : et1 ] ) .
this correlation arises naturally from the expansion of the star due to the energy release .
the more material is burnt in the flame on its way to the surface , the more the star expands and the weaker is the collision .
the energy release depends on the displacement of the initial flame from the center . initiating the flame at larger
radii leads to less burning and thus to stronger collisions of the surface material .
the collision strength is very sensitive to perturbations of the initial flame from perfect sphericity , either directly by imposing more complex initial configurations ( teardrop - like shapes , multiple initial flame kernels ) or by different numerical resolution and resulting discretization errors .
maximum temperature reached in the collision region for fuel above the given density thresholds as a function of the nuclear energy release prior to reaching the maximum collision temperature .
the shaded and dashed regions correspond to conditions where an initiation of a detonation is possible for temperatures reached in fuel of densities @xmath10 and @xmath11 , respectively.,scaledwidth=70.0% ] to trigger a detonation , a collision temperature over @xmath12 is necessary .
however , this threshold temperature needs to be reached in fuel with density higher than @xmath13 .
therefore , the maximum temperature in the collision region at a density exceeding these values was measured .
the results are plotted in fig .
[ fig : et2 ] . from this measurement
, we find that a detonation initiation is only possible in those 2d models that start with a spherical initial flame placed at unrealistically large radii ( @xmath14 ) off - center of the wd star ( cf .
[ fig : et2 ] ) .
3d simulations are cooler in the collimation region for the same nuclear energy release . however , except for one model ( included in figs .
[ fig : et1 ] and [ fig : et2 ] ) the 3d simulations release more energy in the nuclear burning than the 2d simulations . in the model shown in fig .
[ fig : evo ] , the energy release amounted to @xmath15@xmath16 and the temperature in the collision region barely exceeded @xmath17 .
for the one - sided teardrop - shaped initial flame the energy release was even greater and one singe - bubble ignition and the two - sided teardrop - ignition unbound the star .
all 3d simulations clearly failed to trigger a detonation .
we conclude , that although a detonation due to the colliding surface material is in principle possible for specific ( and artificial ) ignition configurations , it is unlikely to happen in nature . for models that remain gravitationally bound , failures to initiate a detonation will lead to pulsations of the wd star .
this may be a second chance for triggering a detonation ( `` pulsational delayed detonation scenario '' , @xcite ) and will be addressed in a follow - up study .
|
we explore the evolution of thermonuclear supernova explosions when the progenitor white dwarf star ignites asymmetrically off - center .
several numerical simulations are carried out in two and three dimensions to test the consequences of different initial flame configurations such as spherical bubbles displaced from the center , more complex deformed configurations , and teardrop - shaped ignitions .
the burning bubbles float towards the surface while releasing energy due to the nuclear reactions .
if the energy release is too small to gravitationally unbind the star , the ash sweeps around it , once the burning bubble approaches the surface .
collisions in the fuel on the opposite side increase its temperature and density and may in some cases initiate a detonation wave which will then propagate inward burning the core of the star and leading to a strong explosion .
however , for initial setups in two dimensions that seem realistic from pre - ignition evolution , as well as for all three - dimensional simulations the collimation of the surface material is found to be too weak to trigger a detonation .
|
the survival of ferromagnetic ordering under the disruption of frozen random fields @xcite and the onset of spin - glass ( sg ) order in systems characterized by random competing interactions @xcite are two central problems in the statistical mechanics of systems with quenched randomness . when properly rephrased , these two problems turn out to relate to core problems in other , quite different disciplines .
insight into these two closely related problems in any given system is obtained from the corresponding phase diagram as a function of the physical parameters , such as temperature and disorder strength .
finite - connectivity mean - field sg models @xcite and their _
p_-spin counterpart @xcite are important for two distinct reasons .
firstly , despite being mean - field in nature , they are believed to share common properties with finite - dimensional spin - glass systems @xcite .
secondly , the statistical mechanics of sg model systems with fixed finite connectivity relates to the study of various hard computational problems . in the last two decades
the analysis of finite - connectivity sg systems has also offered new tools for understanding the very nature of several optimization problems _ and _ improve algorithmic performances @xcite . in this paper
we first obtain the phase diagram of finite - connectivity mean - field sg models .
we then investigate the relation between features of the phase diagram and the performance of low density parity check ( ldpc ) codes , analyzing the effects induced by the replica symmetry breaking ( rsb ) nature of solutions obtained for these finite - connectivity sg models @xcite . in order to answer these questions
, we exploit the relation between finite - connectivity sg systems and a class of exactly solvable models ; we employ an efficient computational method that has been shown to be quite effective for understanding several features of finite - dimensional sg @xcite .
we will consider a generalized model unifying the frameworks for a few distinct problems .
particular attention will be given to two special cases : a ) the bethe sg which has been extensively studied under both replica symmetric ( rs ) and rsb anstze @xcite and recently considered in its ferromagnetically biased version @xcite .
b ) gallager ( ldpc ) error - correcting codes @xcite . while the former is well known within the statistical physics community , many in this community are less familiar with the latter and its links to the physics of disordered systems . reliable transmission of information in noisy conditions is one of the basic problems of modern communication technology . correspondingly , one of the central problems of information theory is channel coding .
methods to achieve reliable information transmission rely on the introduction of structured redundancy to the original message in order to compensate for corruption due to noise .
shannon derived rigorous bounds on the level of redundancy ( the code rate ) required to enable error - free communication for a given channel noise @xcite .
however , shannon s theorems are non - constructive and do not offer a practical coding scheme .
different methods , based on various redundancy construction recipes , have been proposed over the years @xcite . despite the fact that error - correcting codes are now widely used in a variety of applications , current performances of most methods are significantly below shannon s bound .
one family of error - correcting codes that have been shown to provide close - to - optimal performances is that of gallager s ldpc codes @xcite . in these codes
the structured redundancy is introduced through parity checks of boolean sums of randomly sampled message bits .
the relation between the parity - check error - correcting codes and sg models , and consequently between quite different disciplines as statistical physics and information theory , has been pointed to in the seminal work of sourlas @xcite .
decoding in ldpc codes corresponds to a class of sg models defined by an underlying lattice geometry that reflects the low density character of the code construction .
the importance of this very last point , that made the above correspondence useful and of practical relevance was not understood until recently @xcite .
methods developed in the statistical mechanics of diluted sg under the rs anstz have been successfully employed to compute macroscopic properties of such systems for different parameter values ; reasonable agreement with observed decoding bounds have been reported @xcite . to our knowledge
however , the only phase diagram that has been evaluated for ldpc codes is in the limit of large coordination / large multi - spin interaction numbers @xcite , which also relates to the phase diagram of the random energy model ( rem ) @xcite . in the following section we introduce the class of ising spin systems to be analyzed and its links to the bethe sg and ldpc decoding problems
this will be followed by section [ sec : rs ] where we review the rs equations and the corresponding solutions are obtained using a computational method adopted from the study of @xmath5-dimensional hierarchical models . as part of this section
we will present the method and briefly discuss the relation between finite - connectivity sg and @xmath5-dimensional hierarchical models . in section [ sec : rsb ]
we introduce the equations behind the rsb theory of finite - connectivity sg studied here and present results obtained for the bethe sg problem and ldpc error - correcting codes under two different transmission channels : the binary erasure channel ( bec ) and the binary symmetric channel ( bsc ) . in section [ sec : disc ]
we discuss the results obtained and point to future research directions .
the two different models examined in this work stem from a unified physical system of @xmath6-spin interaction ising model on a lattice of fixed connectivity @xmath7 .
the hamiltonian describing the system takes the form @xmath8 where @xmath9 represents @xmath10 binary variables , @xmath11 a set of indices related to interaction @xmath12 and @xmath13 the corresponding connectivity tensor with the following properties : @xmath14 the parameter @xmath15 is related to the strength of the random exchange interaction .
the various models considered in this work differ in the connectivity parameters and type of bias used , which corresponds to an additional component of the hamiltonian ( [ h_0 ] ) .
the bethe sg has been extensively studied in the past @xcite and represents a special case of the hamiltonian ( [ h_0 ] ) where one is only considering pairwise interactions , @xmath16 .
bethe sg with ferromagnetically biased exchange interaction @xmath15 , which we will focus on here , has been recently studied in @xcite , where the distribution of the quenched random variable @xmath15 takes the form @xmath17 and where the parameter @xmath18 $ ] is @xmath19 for a ferromagnet and @xmath20 for the unbiased bethe sg .
we consider a simple communication model whereby @xmath21 bit messages are encoded using ldpc codes and then communicated through two channel types : @xmath22 the binary symmetric channel ( bsc ) where bits are flipped with probability @xmath23 during transmission and @xmath24 the binary erasure channel ( bec ) where bits are received with probability @xmath25 and fail to arrive with probability @xmath23 .
a gallager ldpc code is defined by a binary matrix @xmath26 $ ] , concatenating two very sparse matrices known to both sender and receiver , with @xmath27 ( of dimensionality @xmath28 ) being invertible ; the matrix @xmath29 is of dimensionality @xmath30 .
ldpc codes can be employed over various finite fields @xcite ; we will restrict the treatment in this paper to binary symbols .
encoding refers to mapping the original @xmath21 dimensional binary message vector @xmath31 to an @xmath10 dimensional binary code word @xmath32 ( @xmath33 ) by taking the product @xmath34 , where all operations are performed in the field @xmath35 and are indicated by @xmath36 .
the generator matrix is @xmath37 \mbox { ( mod 2)}$ ] , where @xmath38 is the @xmath39 identity matrix , implying that @xmath40 and that the first @xmath21 bits of @xmath41 are set to the message @xmath42 . in
_ regular _ gallager codes the number of non - zero elements in each row of @xmath43 is chosen to be exactly @xmath6 .
the number of elements per column is then @xmath44 , where the code rate is @xmath45 ( for unbiased messages ) .
we will mostly focus here on regular code constructions .
the corruption process that is most intuitively linked to physical systems is the bsc , where the encoded vector @xmath41 is corrupted by flip noise represented by the vector @xmath46 with components independently drawn from @xmath47 .
the received vector takes the form @xmath48 .
decoding is carried out by multiplying the received message by the matrix @xmath43 to produce the _ syndrome _ vector @xmath49 from which an estimate @xmath50 for the noise vector can be inferred .
an estimate for the original message is then obtained as the first @xmath21 bits of @xmath51 .
the bayes optimal estimator ( also known as _ marginal posterior maximiser _ ( mpm ) ) for noise vector bits is defined as @xmath52 .
computing the mpm estimate directly is computationally hard , and is typically approximated in practice by employing an iterative inference algorithm such as belief propagation @xcite .
the connection to statistical physics becomes clear when the field @xmath35 is replaced by ising spins @xmath53 and mod @xmath54 sums by products @xcite .
the syndrome vector acquires the form of a multi - spin coupling @xmath55 where @xmath56 and @xmath57 .
the @xmath6 indices of nonzero elements in the row @xmath12 of @xmath58 are given by @xmath59 and where @xmath60 is the connectivity tensor is given in equation ( [ tensor ] ) .
after gauging @xmath61 the multi - spin coupling coefficient vanishes @xmath62 and the parity check component of the hamiltonian takes the form ( [ h_0 ] ) with @xmath63 forcing all candidate vectors to obey the parity checks .
this notation is used to distinguish the temperature from the parity check constraints and to facilitate the finite temperature study of error - correcting codes . prior knowledge of the biased noise vector is added through a second component to the hamiltonian ( [ h_0 ] ) . after gauging , the new term takes the form @xmath64 where @xmath65 is the nishimori condition @xcite ( where @xmath66 ) , which corresponds to the correct prior for the variables @xmath67 .
decoding at the nishimori condition corresponds to the mpm estimation @xcite .
this gives rise to the hamiltonian @xmath68 for the bsc .
another important estimator one should mention is the _ maximum a posteriori _ ( map )
estimator defined as @xmath69 .
this corresponds to zero temperature decoding @xcite .
codeword bits transmitted over a bec have a probability @xmath25 of being received ( in tact ) .
encoding is carried out similarly to the bsc , but decoding is based on inferring the codeword itself , rather than the noise vector . in the case of the bec ,
solution vectors are , by definition , codewords and should therefore obey @xmath70 , where @xmath71 are the received vectors ( with a certain number of unknown bits ) .
an estimate of the original message is then obtained by inferring the missing bits using mpm .
similar practical methods to those described in the bsc may be employed also in this case .
the statistical mechanics formulation of the problem is similar to that of the bsc , except for the second component added to the hamiltonian ( [ h_0 ] ) , that takes the form @xmath72 \ .\ ] ] code performance in the bec will be considered only at the zero temperature limit as in @xcite . a review of ldpc codes and their link to statistical mechanics has been presented in @xcite .
the analysis carried out here is a natural extension of similar calculations done in the past for the bethe sg @xcite model and @xmath23-spin ising models @xcite .
we therefore briefly review the rs theory of the model described by hamiltonian ( [ rfrb ] ) as both the bethe sg and ldpc decoding can be considered as special cases of the same model . to compute the free energy of the problem , consider the replica equality , @xmath73 where @xmath74 is the analytical continuation in the interval @xmath75 $ ] of the replicated partition function @xcite . following the standard derivation , e.g. as in @xcite , and employing the rs anstz , one obtains the following expression for the replica symmetric free - energy as a function of the local field distribution @xmath76 ( details can be found in appendix a ) @xmath77 there are both site and bond contributions , @xmath78 where @xmath79 denotes an average over the two sources of randomness in ( [ rfrb ] ) , and @xmath80 we denote @xmath81 where @xmath82 \nonumber \\ \beta a ( \ { x_j \},j)&= & \frac{1}{2 } \log [ r(+)r(- ) ] \ .\end{aligned}\ ] ] the polynomials @xmath83 are defined via , @xmath84 , \label{poly}\ ] ] where @xmath85 .
the above expression , in a similar form , has already been obtained in @xcite , and the equivalence with their expression is shown in appendix a. we prefer the current formulation because of its manifest simplicity ( the two contributions to the free - energy coming from sites and bonds are here in evidence ) and because it depends on the function @xmath86 , which will play a central role in the rsb case , examined in section [ sec : rsb ] .
the above writing also introduces the polynomials @xmath83 , already well known in the context of hierarchical lattices @xcite .
evaluating the saddle point equations of the above free - energy expression is the basic premise of the rs theory , so that the crux of the quenched randomness problem ( [ rfrb ] ) lies in the convolution @xcite , @xmath87 where @xmath88 , @xmath89 ; and where the recursion relations are defined via @xmath90 we can also rewrite for convenience the above convolution with the aid of the auxiliary distribution @xmath91 , as a set of two coupled equations ( cfr .
appendix a ) , @xmath92 to briefly explain the link between ldpc decoding and the replica formulation we will restrict this discussion to the bsc .
the replica analysis as well as decoding in other noisy channels follow a similar path @xcite . following the derivation of @xcite we write saddle point equations for the ldpc decoding problem , considering a special case of the hamiltonian ( [ rfrb ] ) and the saddle point equations
( [ crux2 ] ) , where @xmath93 is deterministic @xmath94 most of the numerical results obtained so far are obtained along the nishimori line @xmath95 . here
we investigate the phase diagram at @xmath96 prior temperature @xmath97 , for different choices of the parameters @xmath7 and @xmath6 . a typical case we will discuss in detail
is the one corresponding to @xmath98 , @xmath99 .
we will check the consistency between the results obtained and true decoding experiments , using belief propagation ( bp ) @xcite , at and away from the nishimori condition .
we will not review bp in the paper as it is a well known message passing algorithm @xcite with strong links to the bethe approximation and its variants @xcite .
the iterative algorithm is a microscopic equivalent of equations ( [ crux3 ] ) .
the set of ( rs ) equations ( [ crux2 ] ) have been considered and solved in a variety of different ways already .
this section introduces to a different method .
we propose an alternative approach to existing methods ( e.g. population dynamics or monte carlo techniques ) which solves equations of the form ( [ crux ] ) efficiently and accurately .
the main reason why it has not been used to study finite - connectivity sg models so far is because of the different context in which it was firstly introduced @xcite .
the method we propose allows to determine the phase diagram of diluted systems of the type studied here and analyze the nature and exact location of the phase boundaries .
it borrows tools and concepts from a different branch of research , namely the area of hierarchical models @xcite , to obtain accurate phase diagrams .
the iterative method we now describe has a corresponding computational effort that scales as @xmath100 , n being the number of bins introduced and @xmath6 the multi - spin interaction parameter .
equation ( [ crux ] ) complicates , after a few iterations , even a simple initial local field distribution .
the distribution @xmath76 and its conjugate distribution @xmath101 are represented via histograms .
each histogram is characterized by two quantities , @xmath102 and @xmath76 , where the latter is the associated probability .
equation ( [ crux3 ] ) is the convolution of @xmath103 distributions , taking into account the distribution of the random field @xmath104 , contained in the average @xmath105 .
this is numerically constructed from pairwise convolutions of distributions . for a @xmath1 rate gallager code , e.g. , three pairwise convolutions are required . a pairwise convolution ( or binary operation )
is achieved as follows : + ( 1 ) the histograms are placed on a grid properly drawn at each iteration step of the algorithm .
all histograms that fall within the same grid cell are combined in such a way as to preserve the averages across the domain of the local field ( and its conjugated ) distribution(s ) .
the histograms that falls outside the grid , representing a pre - determined small fraction of the overall probability , are similarly combined into a single histogram . in this procedure ,
a histogram that falls within a narrow band of a grid boundary is proportionally shared between neighboring bins .
the gridding is done separately for @xmath106 and @xmath107 .
+ ( 2 ) two gridded distributions are convoluted according to equation ( [ crux2 ] ) regenerating the original number of histograms .
+ the convolution of @xmath108 distributions is achieved , from pairwise convolutions , as follows : ( 1 ) the pairwise convolution contained in the first equation of ( [ crux2 ] ) is cycled once for , e.g. , a code rate @xmath1 .
( 2 ) the resulting distribution is pairwise - convoluted with the noise channel distribution @xmath109 . (
3 ) the resulting distribution is further self - convoluted according to the second equation in ( [ crux2 ] ) .
this completes the entire process , that is then iterated until convergence is achieved .
multiple cycling operations are also an interesting option being related to , e.g. , the source compression problem .
the flows of the probability distribution @xmath76 determine the phase diagram of the system .
most of our results have been obtained considering @xmath110 independent bins , corresponding to the flows of @xmath111 quantities ( the corresponding number of elementary operations considered is the number of bins to the power of @xmath112 ) .
each of the two grids at @xmath106 and @xmath107 have e.g. @xmath113 bins on each side of the average value .
+ similar set of equations to ( [ crux ] ) have been discussed in a variety of different contexts .
a very important one is that of real - space renormalization - group theory @xcite @xcite @xcite .
eventhough the algebraic form of the convolution corresponding to these two problems is the same , the two problems are indeed very different . in the case of hierarchical models the distribution of the exchange interaction is studied , while in the problem we are discussing in this work one considers the distribution of the local field .
this reflects the difference in the two underlying geometries .
the @xmath5-dimensional hierarchical lattice include loops that are one of the essential characteristics that makes hierarchical lattices better candidates , within exactly solvable models , to study the problem of finite - dimensional sg @xcite .
the similarity between equations ( [ crux3 ] ) and those related to hierarchical models is indicative of a deeper relation between the two ( both geometries are hierarchical in the sense of ref .
@xcite ) and will be the subject of future work .
the calculated phase diagram for the gallager ldpc code of rate @xmath1 and a bsc of flip rate @xmath23 as a function of the prior temperature @xmath97 are shown if fig .
[ ldpc_phase ] .
we report both the location of the dynamical noise transition @xmath114 , defined by the noise level at which suboptimal solutions emerge , and critical noise transition @xmath115 , where the ferromagnetic solution ceases to be dominant , for all ranges of temperature @xmath97 .
the phase diagram was calculated considering the flows of @xmath111 quantities dictated by the convolution ( [ crux ] ) . as mentioned before the above number should be taken to the power @xmath112 to obtain the number of elementary operations within , e.g. , the second equation of ( [ crux2 ] ) .
the location of the dynamical and critical transition values can be determined by analyzing the free - energy and the complexity as a function of the channel noise level for different values of the temperature @xmath116 ( fig .
[ bscrsfree ] ) .
we will refer noise level that corresponds to the emergence of sub - optimal solutions as the dynamical transition . at any value @xmath117 , the computed free - energy ( [ free - energy ] )
reduces to the ferromagnetic free - energy @xmath118 .
consider first temperature values in the proximity of the nishimori line .
the trajectories exhibit a crossover to an unstable distribution first , and then flow to the ferromagnetic sink below @xmath119 ( corresponding to a decoding success ) or to a stable non - ferromagnetic distribution ( corresponding to a decoding failure ) above it .
a similar behavior is observed for any temperature above the nishimori line , the stable distribution being different for each values of the temperature @xmath120 .
rather different behavior is observed below the nishimori line , where the unstable distribution is a strong coupling distribution , meaning that the average local field strength increases but the rescaled distribution remains constant .
finally , the trajectory flows again , either to the ferromagnetic sink or to a fixed local field distribution , which is stable under iteration of the recursion ( [ crux3 ] ) . the computed free - energy , according to the convolution ( [ crux3 ] ) , is higher than the ferromagnetic free - energy for all noise levels @xmath121 . above the critical noise value @xmath115 , the computed free - energy becomes lower than the ferromagnetic free - energy .
the location of the dynamical transition has been determined , at any temperature , considering the flows , under iteration , of the convolution ( [ crux ] ) , while the location of the critical noise threshold has been determined from the free - energy ( [ free - energy ] ) as the point where the computed and the ferromagnetic free - energies coincides . in the inset of fig .
[ ldpc_phase ] we also report in square boxes the results of real decoding experiments obtained as follows .
we iterated the probabilistic decoding algorithm @xcite , at different values of the noise channel @xmath23 , for an ensemble of @xmath111 networks of @xmath122 and @xmath123 nodes respectively .
the inset shows the calculated mean values , the corresponding error being smaller than the corresponding symbol size .
a precise estimate of the dynamical transition point has been obtained considering the percentage of decoding success within the network ensemble as a function of the noise level , for different system sizes , @xcite .
the decoding experiments were performed for different choices of the temperature . along the nishimori line ( see fig . [ ldpc_phase ] ) we find the value of @xmath124 in excellent agreement with the reported value for regular @xmath1 gallager codes @xcite . notice that the reentrant nature of both the dynamical and critical boundaries , as we found solving the convolution ( [ crux2 ] ) , is confirmed by the decoding experiments . in the range of noise levels @xmath125 , decoding is possible only in a range of finite temperatures , as indicated in the inset of fig .
[ ldpc_phase ] .
the boundary between decodeable and undecodeable regions reaches its maximal value at the nishimori temperature ; zero temperature decoding fails at the same noise level .
the reentrant nature of the critical phase boundary increases below the nishimori line , but eventually weakens as the temperature is decreases further , a phenomenon that has already been discussed in a variety of different contexts @xcite .
the fact that mpm and zero temperature bp / rs decoding do not coincide is not in contradiction with the results obtained in @xcite which indicate that map and mpm decoding provide the same critical transition point .
clearly , zero temperature bp / rs decoding does not provide map results due to rsb effects .
the latter and their effect on the phase diagram and on the reentrant nature of the phase boundaries will be further discussed in section [ sec : rsb ] .
an independent calculation has been performed at zero temperature . in this case
the recursion relations ( [ recursion ] ) simplify and one finds limiting expressions for the two functions @xmath126 , @xmath127 . implementing numerically such asymptotic expressions would require an independent calculation where one explicitly changes the recursion relations ( [ recursion ] ) to their asymptotic @xmath128 form .
the importance of the polynomials @xmath129 becomes clear at this point . at any temperature
considered , we distinguish the finite and zero temperature contributions , corresponding to factorize the largest exponent in the sum ( [ poly ] ) , so that the limit of zero temperature is particularly simple to handle in our approach and no explicit changes are required in the recursion relations ( [ recursion ] ) .
we observe that at zero temperature the local field distribution is not peaked on integers as one might expect .
we find that the relative weight of the background behind the integer - peaked distribution weakens as one consider the rsb solution , but restricting the distribution of the local fields to integer values , which results in a modification of the recursion relations ( [ recursion ] ) leads to different results . for ldpc ( rs )
decoding in the bsc this leads to an erroneous dynamical transition value of @xmath130 , marked in fig .
[ ldpc_phase ] by a black filled dot , well below the reported zero temperature value of @xmath131 .
similar behavior occurs in the binary erasure channel problem if one simply neglects the background distribution , as we will discuss in section [ sec : rsb ] .
the initial condition @xmath132 , at @xmath133 , has been used as initial condition for the bsc analysis . a dependence of the solution of the convolution ( [ crux3 ] ) on the initial condition for the local field has been observed .
we have studied the above convolution ( [ crux3 ] ) for different choices of the initial condition in the form @xmath134 the choice of @xmath135 , where @xmath23 is the flip rate of the channel results in the phase diagram of fig .
[ initial_cond ] . _
any _ choice of initial condition with @xmath136 ( including the case @xmath137 considered above ) leads to the dynamical transition line ( dashed line in fig .
[ ldpc_phase ] ) , except for the choice @xmath135 which leads to a different dynamical transition .
the location of the critical transition remains unchanged .
for instance , using the initial condition @xmath135 , the dynamical transition along the nishimori line is increased to @xmath138 , well above the known dynamical transition point .
the physical interpretation of this behavior remains to be explained .
since it results from a simple change in the initial conditions of ( [ crux ] ) , one might argue that a properly modified decoding algorithm would enable us to decode above current @xmath114 values ; this is a possibility that we can not rule out and further research is needed in order to elucidate this point .
we where able to perform a second set of decoding experiments that coincide with the dynamical transition line in fig .
[ initial_cond ] but only for noise level values below the value @xmath139 .
the reentrant nature of the dynamical transition line ( the dynamical transition at zero temperature being @xmath140 ) is reduced in this case . for a constant @xmath6 value one can select non - integer connectivity values of average @xmath141 and consider a modified form of the convolution ( [ crux3 ] ) that involves a linear combination of terms corresponding to different connectivity values .
non - integer values of the multi - spin interaction parameter @xmath6 can also be easily accommodated in ( [ crux3 ] ) .
it has been reported that irregular low - density parity - check codes outperform regular constructions @xcite .
we considered a generalized form of the convolution ( [ crux3 ] ) , ( an explicit form is not provided for brevity ) where both connectivity @xmath7 and multi - spin interaction @xmath6 take non - integer values such that the rate @xmath142 is preserved . in fig .
[ irregular ] we report the dynamical transition line for the regular construction @xmath143 ( @xmath1 ) already considered in the previous section together with the dynamical transition lines for irregular constructions of two values @xmath99 and @xmath144 and @xmath145 and @xmath98 , such that @xmath146 ( left to right ) and @xmath147 .
the dynamical transition value is observed to increase for the irregular constructions .
this does not give a direct information on the specific pattern of irregularity that one should consider but provides quantitative information on the average connectivity / multi - spin interaction values one should investigate further . as for the regular case discussed earlier , a phase diagrammatic approach to
the problem of irregular ldpc codes is important to understand the reason behind the successful decoding properties of irregular ldpc codes .
a similar behavior to the one shown in fig .
[ irregular ] , but less pronounced , has also been observed for ldpc decoding with the initial conditions ( [ ic ] ) when @xmath135 .
we considered the finite - connectivity model ( [ rfrb ] ) for the values of the connectivity @xmath149 and @xmath99 .
despite the fact that it does not correspond to any gallager ldpc code construction it is interesting to consider these connectivity parameters for two reasons : 1 ) they can represent code constructions in other variants of the ldpc family such as sourlas or mn codes @xcite , and as such may also be linked to other ldpc - based applications .
2 ) they give rise to a sg behavior which is of interest . in fig .
[ c9k4 ] we report the corresponding phase diagram obtained considering the convolution ( [ crux3 ] ) .
an sg phase occurs in this case . at low values of the flip rate @xmath23 the convolution ( [ crux ] ) flows to the ferromagnetic
sink . in the vicinity of the dynamical transition line ( marked in fig [ c9k4 ] as the solid black line ) the trajectories under iteration of the above convolution
cross over to an unstable distribution and then either flow to the ferromagnetic sink , below the dynamical transition , or to a stable non - ferromagnetic distribution . for temperatures below the nishimori line , at high enough values of the flip rate ,
the convolution instead flows to a spin - glass distribution .
the role of rsb for the problem ( [ rfrb ] ) is the subject of this and the following sections .
the one - step rsb solution for the bethe sg , as a special limiting case of the hamiltonian ( [ rfrb ] ) , has been studied in detail long ago . in this case
it is well known that the rs solution is incorrect @xcite .
recently , the onset of rsb phenomena in some limiting cases of the problem ( [ rfrb ] ) has been considered , relying on the variational approximation @xcite ( for a recent formulation see @xcite ) .
consider the equation for the global order parameter @xmath150 , firstly introduced for the bethe sg by mottishaw , @xcite in the general case of a multi - spin interaction parameter @xmath6 , @xmath151 \big ] .
\nonumber \\ \label{gopk}\end{aligned}\ ] ] where @xmath152 labels different replicas of the system .
the above expression holds in general , whether replica symmetry is broken or not ( see appendix b for the derivation ) . in the rs approximation
one assumes that different replicas are equivalent and the functional order parameter @xmath150 depends only on the variable @xmath153 . in the limit of @xmath154 , can be shown to satisfy , @xmath155 .
\nonumber \\ \label{rs}\end{aligned}\ ] ] this equation corresponds to the equation that have been written for the bethe sg , as soon as @xmath16 and for vanishing message prior temperature @xmath156 @xcite .
a derivation of equation ( [ rs ] ) is found in appendix b , where we show the equivalence of the above expression with the set of recursion relations ( [ crux ] ) , for the local field and conjugate distributions . in the one - step rsb approximation
one assumes instead that the functional order parameter @xmath150 depends only on the variables @xmath157 , where @xmath158 is the replica index , @xmath159 and @xmath160 . in this case one
finds that the corresponding equation for the functional order parameter @xmath161 becomes @xmath162 \times \nonumber \\
\big [ \frac{m}{2 } \sum_m a ( \ { r_{mj } \},j ) \big ] , \nonumber \label{rsb}\end{aligned}\ ] ] where the normalization is given by @xmath163 the corresponding one - step rsb convolution follows , assuming that the order parameter factorize according to @xmath164 the assumptions behind the _ factorized _ anstz ( [ factorized ] ) , where the local field distribution is independent from the intra - group index , are discussed in @xcite . in this case
the one - step rsb convolution ( [ rsb ] ) become @xmath165 \times \nonumber \\ & & \delta \big ( \hat{x}-u ( \ { x_j\},j ) \big ) , \label{crux_rsb}\end{aligned}\ ] ] where the normalization @xmath166 is given by @xmath167,\ ] ] and where @xmath168 .
the above equations are valid , under the assumption ( [ factorized ] ) , at finite temperatures and arbitrary connectivity values @xmath7 and @xmath6 . after its very first formulation @xcite the convolution ( [ crux_rsb ] ) has been studied in a range of different contexts , from random @xmath6-sat to the coloring problem @xcite . using the identity , @xmath169 = \big [ \frac { 2 \cosh ( u ( \ { x_j\},j ) ) } { \prod_j \cosh ( \beta x_j ) } \big ] ^ { \mu } , \label{identity}\ ] ] and substituting it in equation ( [ crux_rsb ] ) one recovers the one - step rsb equations that have been also derived using the variational approach @xcite .
this shows that the so called frozen variational anstz , considered in @xcite and the factorized anstz of @xcite are equivalent .
even though there exists some degree of arbitrariness @xcite in the way one redistributes the re - weighting between the two equations ( [ crux_rsb ] ) , we prefer the notation that involves the exponential of the energy term @xmath170 , rather than the less obvious ( and numerically more difficult to implement ) parallel expression that follows considering the right hand side of ( [ identity ] ) @xcite .
the simplified expression of the free - energy presented in section iii was obtained in virtue of the same identity ( [ identity ] ) , and the same degree of freedom writing the free - energy was already discussed for the bethe sg @xcite ( the right hand side of ( [ identity ] ) being , at @xmath16 the function @xmath171 discussed in ref .
@xcite ) .
we apply the method described in section iii to study the one - step rsb convolution ( [ crux_rsb ] ) of the bethe sg , which has been extensively studied and recently considered also in its ferromagnetically biased version @xcite . in this case
the distribution of the quenched random variable @xmath15 is @xmath172 and where the parameter @xmath18 $ ] is @xmath19 for a ferromagnet and @xmath20 for the unbiased bethe sg .
the bethe sg corresponds indeed to the original problem ( [ h_0 ] ) at @xmath16 .
we will present results obtained at zero temperature .
in the zero temperature limit we find the following expression for the free energy @xmath173 , @xmath174 where explicit expressions for the site and link contributions to the ground state energy are @xmath175 corresponding , at @xmath176 and @xmath16 , to the rs equation ( [ free - energy ] ) in the zero temperature limit and neglecting the prior term in ( [ rfrb ] ) .
similar expression have been discussed in @xcite . in the limit of zero temperature one
finds @xmath177 and @xmath178 , being @xmath179 and @xmath180 . in what follows we present results obtained under the @xmath181 anstz as well as for the full distribution .
the relative importance of non - integer values in the distribution that solves the above convolution ( [ crux_rsb ] ) is rather limited in this case .
the background behind the integer peaked values weakens as one consider the one - step rsb case , and may disappear in a fully replica symmetry broken scenario @xcite . in fig .
[ bethe_gse ] we show the ground state energy as a function of the replica symmetry breaking parameter @xmath182 for different values of the ferromagnetic bias using the integer peaked anstz and in the case of the full local field distribution .
the ground state energy one extrapolates including the background calculation is higher than the one obtained within the three peak anstz @xcite . in fig .
[ bethe_compl ] we show the corresponding complexity or configurational entropy , that can be obtained as a parametric plot in terms of the rsb parameter @xmath183 , the free - energy @xmath184 being its legendre transform @xcite .
we observe a positive complexity for all reported values of @xmath185 . at higher values of the ferromagnetic bias
@xmath186 a positive complexity is not observed anymore .
one should also consider that the above values should be extrapolated in the limit of large number of bins , so that one expects a slightly higher value @xmath187 . in fig .
[ bethe_mag ] we show the magnetization for in the replica symmetric and one - step replica symmetry broken solution at different values of the ferromagnetic bias . at values @xmath188
one recovers the results known for the unbiased bethe sg .
the solid line indicates the magnetization computed in the replica symmetric approximation ( notice that this is a computed magnetization for a finite number of bins , so that it approaches the expected value of @xmath189 from below ) , while the square boxes represent the magnetization in the one step rsb approximation without considering the background . for each value of @xmath190 , we extrapolated the magnetization that corresponds to the value of @xmath191 where the ground state energy has a maximum . similarly , the dashed line points to the results obtained considering the true convolution ( [ crux ] ) , that includes non integer values of the local field . the putative size of the so called @xmath192 , previously discussed @xcite , if any , is rather small and do not agree with previous estimates ( we observe the magnetization within the rsb calculation to vanish at values of @xmath193 , if one considers a large enough number of bins within the calculation , implying that the size of the mixed phase , if any , is vanishingly small ) .
it should be said however that the calculation of @xcite considered a full one step rsb calculation , in the sense of @xcite , where the factorized approximation is relaxed . to investigate this point better , we performed the following : instead of implementing the convolution ( [ crux ] ) within the factorized approximation @xcite
, we choose a population of @xmath194 distributions ( including a site dependence ) . at each step of the iteration process
we consider pairwise - convolutions of local field distributions corresponding to different sites , selected at random ; after a transient regime , we obtain the equilibrium population . waiting a number of iteration steps of the order of the population size , according to our calculation , the local field distributions become site independent , meaning that the factorized approximation provides similar results to those of the full rsb anstz . if one compares fig .
[ bethe_compl ] with the corresponding plot in ref .
@xcite , where a similar calculation of the complexity is presented , we see that the location of the maximum of the complexity as a function of @xmath182 _ decreases _ for increasing @xmath190 values ( see fig .
[ bethe_compl ] ) , as one might intuitively expect , increasing the ferromagnetic bias . in the case of the bec discussed in section
ii , the form of the convolution ( [ crux3 ] ) simplifies and the distribution of the local field , and its conjugated distribution become bimodal . despite its simplicity
the binary erasure channel problem is of practical relevance @xcite . for the bec one writes the convolution ( [ crux3 ] ) as @xmath195 where @xmath196 is a delta function located at @xmath197 , @xmath198 is the distribution of the effective local field and @xmath199 its conjugate distribution .
the resulting convolution @xcite involving the two distributions @xmath198 and @xmath199 , in the one - step rsb case , is given by @xmath200 \times \nonumber \\ & & \delta \big ( \hat{x}-u(\ { x_i\},j ) \big ) , \nonumber \\ \label{crux_bec}\end{aligned}\ ] ] where @xmath201 are the binomial coefficients of order @xmath112 and @xmath202 the set of normalization coefficients defined by @xmath203.\ ] ] one can readily check that the above convolution reduces to the variational approximation @xcite using identity ( [ identity ] ) .
we note that the variational anstz , as already discussed for the case of the bsc , coincides with the factorized anstz ( [ factorized ] ) @xcite .
we consider an explicit expression for the one step rsb free - energy that involves the binomial coefficients of order @xmath6 . within the factorized hypothesis @xcite ,
the expression for the free - energy is easily computed as a function of the rsb parameter @xmath182 .
we present results for the bec at zero temperature for values of the code rate r=1/4 and @xmath204 ( note that we also consider in what follows the integer peak anstz for the local field distribution @xcite in order to compare with the general case of non - integer local fields ) . in the inset of fig .
[ bec_1 ] we show the energy as a function of the erasure probability @xmath23 for the rate value @xmath204 within the integer valued local field approximation .
notice that the integer value approximation corresponds to having only three bins in our approach of section iii . in fig .
[ bec_2 ] we compare the results within and without the integer approximation for a code rate @xmath1 .
the dynamical transition is defined as the point where sub - optimal solutions emerge the corresponding complexity being non zero , as already discussed for the bsc channel ( see fig . [ bec_2 ] ) .
the location of the dynamical transition is also the point where the recursion ( [ crux_rsb ] ) flows to a subdominant distribution , rather than the ferromagnetic sink .
this coincides with the point where algorithms devised for real decoding experiments fail to decode the original message . within the rs framework
, we find that the location of the dynamical transition changes significantly if non - integer values of the local field are taken into account .
within the three peak anstz the local field distribution is simply defined as @xcite @xmath205 the calculation we performed within the integer local field approximation ( for rates @xmath204 and @xmath206 in fig .
[ bec_1 ] and fig .
[ bec_2 ] respectively ) might suggest to interpret the point @xmath207 for the rate value @xmath204 as the dynamical transition point @xcite . if however one compares ( see fig .
[ bec_2 ] ) with the results of the local field distribution that now includes non - integer values ( the thick solid line corresponding to @xmath208 ) we see that the location of the dynamical transition is rather different .
moreover the behavior of the local field probability @xmath209 around the value @xmath210 , within the @xmath211 anstz is rather suspicious . indeed , including non - integer values of the local field turns out to be crucial for having a quantitative description of the dynamical transition and possibly explains the disagreement with numerical experiments that has been reported lately for the r=1/2 bec problem @xcite .
if one avoids the _ three peak _ anstz , the nature of the flows dictated by the recursion ( [ crux_bec ] ) is smooth as it always leads to a well - behaved behavior of the probability densities @xmath212 and @xmath213 . up to this point
all considerations are still within the rs approximation .
moreover , considering the full convolution ( [ crux_bec ] ) as we do in what follows , the location of the dynamical transition in the one - step rsb solution becomes higher than that obtained within rs , which arguably implies that one may be able to construct the algorithmic analog of the rsb recursion ( [ crux_bec ] ) to improve decoding performances .
similarly , within the rsb calculation , the location of the critical transition is increased for the rate @xmath1 considered above and found to be very close to the shannon bound value @xmath214 ( cfr .
[ bec_5 ] ) .
+ we iterated the convolution ( [ crux_bec ] ) within the one - step rsb approximation described in section vi and computed the free - energy as a function of the rsb parameter @xmath182 for several values of the erasure probability @xmath23 , above the dynamical transition point @xmath215 as reported in fig .
[ bec_3 ] . in this range of @xmath23 values ,
the free - energy grows as a function of @xmath182 , dropping to the ferromagnetic free - energy value ( being @xmath216 for the bec ) at some critical @xmath217 . at higher values of the erasure probability @xmath23 ,
the free - energy eventually drops to zero after having reached its maximum ( see fig . [ bec_3 ] ) . according to the standard recipe of maximizing the free - energy with respect to @xmath12 @xcite
, we extrapolate the value of the dynamical transition within the one - step rsb calculation , finding the value of @xmath218 , indicated in fig .
[ bec_2 ] by the dotted line , well above the rs value @xmath219 .
the corresponding complexity is shown in fig .
[ bec_4 ] , for different values of the erasure probability considered .
we observe two branches , a physical ( right ) and an unphysical one ( left ) as explained in ref .
@xcite . from the former branch we extrapolated the rsb energy value corresponding to the noise channel probability .
notice that above the rs dynamical transition we do observe , for values of @xmath220 that the second branch , that is the relevant one , is not present .
this can be explained because the corresponding curves for the free - energy ( fig .
[ bec_3 ] ) do not reach a maximum and drop instead to the ferromagnetic energy value ( @xmath221 ) .
the dynamical transition within the rsb calculation increases .
if one considers the free - energy as a function of the rsb parameter @xmath182 at values of the erasure probability slightly higher that the rs critical transition point , one finds that the maximum is located at positive energies , implying that the critical transition point also changes under the one step rsb calculation . in this way
[ bec_5 ] ) we extrapolated the value of the critical transition , within the rsb approximation , to @xmath222 .
note that shannon s bound for the specific code rate considered is @xmath223 . in comparison , the critical rs transition point @xmath224 .
the fact that the dynamical and critical transition points increase within the one - step rsb calculation might be surprising at first . in the next section
we will show that a similar situation is observed for the bsc at temperatures below the nishimori condition @xcite , where it is known that the rs solution is dominant @xcite .
it is also reasonable to believe that the above one step rsb calculation is exact for this problem @xcite @xcite .
the above findings concerning the location of the critical ( rsb ) transition seem to support this belief . in this section
we return to the bsc ( [ rfrb ] ) and employ the same considerations discussed for the bec in the previous section .
we consider the convolution ( [ crux_rsb ] ) along the nishimori line @xmath225 .
the location of the dynamical transition has been discussed in section [ sec : rs ] and we found the value of @xmath226 within the rs approximation , in agreement with probabilistic decoding estimates .
not surprisingly @xcite , if one looks at the free - energy as a function of the rsb parameter @xmath12 for values of the bsc noise probability @xmath227 , one finds that the free - energy decreases ( see the inset in fig . [ bsc_1 ] ) , indicating that the rs solution is dominant , so that the dynamical transition is the same in the one - step rsb case .
@xmath228 . however , at temperatures well below the nishimori line instead , i.e. @xmath229 , we observe the following facts : above the rs dynamical transition @xmath230 the free - energy increases as a function of the rsb parameter @xmath182 , dropping to the corresponding ferromagnetic free - energy at some value @xmath217 , before reaching a maximum .
this means that the second , physical branch of the complexity is not present , the analysis following the same lines as for the bec calculation . at higher values of the flip rate ,
i.e. @xmath231 the second branch of the complexity ( see fig .
[ bsc_2 ] ) appears , indicating a non - zero complexity and that we are above the rsb dynamical transition .
the corresponding complexity for the ldpc decoding at @xmath232 is reported in fig .
[ bsc_2 ] .
these findings imply that the location of the dynamical transition is improved with respect with the rs value .
the same behavior has been observed , progressively reducing in importance at higher values of the temperature @xmath233 and @xmath234 , while at @xmath235 , corresponding to the nishimori line , this phenomenon is no longer observed .
the inset of fig .
[ bsc_1 ] shows the computed free - energy as a function of the rsb parameter @xmath182 at the nishimori temperature channel noise @xmath236 .
we conclude observing that , below the nishimori line , the reentrant nature of the dynamical transition is not observed within the rsb calculation .
we introduced a new method to compute the phase diagram of finite - connectivity sg model systems and discussed the connection with ldpc codes .
we extended the one - step rsb theory , originally formulated for the bethe sg problem , to the gallager ldpc decoding for two different type of corruption processes , bec and bsc .
for the bec at zero temperature we found that the location of the dynamical transition increases with respect to the rs case .
our estimate for the rsb critical transition value is a mere one percent away from capacity , differently from the rs estimate .
the critical and dynamical properties of ldpc decoding have also been studied for the case of a bsc .
the reentrant nature of the ldpc phase diagram and the corresponding changes that rsb implies in its low temperature properties have also been studied .
we found that below the nishimori line the dynamical transition increases within the rsb theory .
possible implications to improve the decoding performances of the state - of - the - art error correcting codes are one of the perspectives we wish to address .
we would like to construct the algorithmic analog of the convolution ( [ crux_rsb ] ) in order to improve the decoding performances as predicted by the rsb theory .
the method we used to determine the phase diagrams along this work is very promising .
it allows to study , without loss of continuity , rather different problems , ranging from the @xmath5-dimensional hierarchical sg to finite - connectivity mean - field models . in this sense
it would be very interesting to apply this method to clarify a few open issues .
on the one side we would like to investigate , within the one - step rsb theory , the finite temperature phase diagram of the bethe sg in a magnetic field . on the other , it would be interesting to explore the presence , if any , of rsb phenomena in @xmath5-dimensional sg hierarchical model systems .
this work was supported in part by the european community s human potential programme under contract number hprn - ct-2002 - 00319 .
y. imry and s. ma , phys .
* 35 * , 1399 ( 1975 ) .
s.f . edwards and p.w .
anderson , j. phys .
f * 5 * , 965 ( 1975 ) . k. wong and d. sherrington , j. phys .
a * 20 * , l785 ( 1987 ) .
m. mzard and g. parisi , europhys . lett . * 3 * , 1067 ( 1987 ) .
p. mottishaw , europhys . lett . * 4 * , 333 ( 1987 ) . c. dedominicis and p. mottishaw , j. phys .
a * 20 * , l375 ( 1987 ) . c. dedominicis and y.y .
goldschmidt , j. phys .
a * 22 * , 399 ( 1989 ) ; c. dedominicis and y.y .
goldschmidt , phys .
b * 41 * , 2184 ( 1990 ) .
h. rieger and s. kirkpatrick , phys .
b * 45 * , 9772 ( 1992 ) .
r. monasson , j. phys .
a * 31 * , 513 ( 1998 ) .
y.y . goldschmidt and p.y .
lai , j. phys a * 23 * , l775 ( 1990 ) . m. mzard and g. parisi , eur .
j. b * 20 * , 217 ( 2001 ) ; m. mzard and g. parisi , j. stat . phys . * 111 * , 1 ( 2003 ) .
m. mzard , g. parisi , and r. zecchina , science * 297 * , 812 ( 2002 ) .
dewar and p. mottishaw , j. phys .
a * 21 * , l1135 ( 1988 ) .
g. migliorini and a.n .
berker , phys .
b * 57 * , 426 ( 1998 ) .
t. castellani , f. krzakala , and f. ricci - tersenghi , cond - mat 0403053 .
gallager low density parity check codes , ire trans .
theory * it-8 * , 21 ( 1962 ) ; r.g .
gallager , low density parity check codes , research monograph series * 21 * , cambridge , ma : mit press ( 1963 ) .
shannon , mathematical theory of communication .
j. * 27 * , ( pt i ) 379 , ( pt ii ) 623 ( 1948 ) .
t. cover and j. thomas , elements of information theory , new york : wiley ( 1991 ) .
n. sourlas , nature * 339 * , 693 ( 1989 ) .
y. kabashima and d. saad , europhys . lett .
* 45 * , 97 ( 1999 ) .
r. vicente , d. saad , and y. kabashima , phys .
e * 60 * , 5352 ( 1999 ) .
a. montanari , eur .
j. b , * 23 * , 121 ( 2001 )
. b. derrida , phys .
b * 24 * , 2613 ( 1981 ) .
k. nakamura , y. kabashima and d. saad , europhys .
56 * , 610 - 616 , ( 2001 ) .
mackay , ieee trans .
theory * 45 * , 399 ( 1999 ) .
h. nishimori , j. phys .
c * 13 * , 4071 ( 1980 ) .
h. nishimori , j. phys .
japan * 62 * , 2973 ( 1993 ) .
y. iba , j. phys .
a * 32 * , 3875 ( 1999 ) .
s. franz , m. leone , a. montanari , and f. ricci - tersenghi , phys .
e * 66 * 046120 ( 2002 ) .
y. kabashima and d. saad , j. phys .
a * * , r1 ( 2004 ) .
d. bowman and k. levin , phys .
b * 25 * , 3438 ( 1982 ) . y. fu and p.w .
anderson , j. phys .
a * 19 * , 1605 ( 1986 ) .
m. mzard , g. parisi and m. virasoro , spin glass theory and beyond , singapore : world scientific ( 1987 ) .
t.murayama , y. kabashima , d.saad and r. vicente , phys .
e * 62 * , 1577 ( 2000 ) .
t. tanaka and d. saad , j. phys .
a * 36 * , 11143 ( 2003 ) .
f. kschischang and b. frey , ieee j. selected areas commun . * 2 * , 153 ( 1998 )
. j. pearl , probabilistic reasoning in intelligent systems , san francisco , ca : morgan kaufmann publishers inc .
yedida , w.t .
freeman , and y. weiss , advances in neural information processing systems * 13 * , t.k .
leen , t.g .
dietterich and v. treps ( eds . ) , cambridge , ma : mit press ( 2001 )
. m. opper and d. saad ( editors ) , advanced mean field methods - theory and practice , cambridge , ma : mit press ( 2001 ) .
d. andelman and a.n .
berker , phys .
b * 29 * , 2630 ( 1984 ) .
m. hinczewski and a.n .
berker , cond - mat 0507293 .
berker and s.r .
mackay , phys .
b * 33 * , 4712 ( 1986 ) .
m. kaufman and b. griffiths , phys .
b * 24 * , 496 ( 1981 ) ; b. griffiths and m. kaufman , phys . rev .
b * 26 * , 5022 ( 1982 ) .
i. kanter and d. saad , j. phys .
a * 33 * , 1675 ( 2000 ) .
t. richardson and r. urbanke , ieee trans .
theory , * 47 * , 599 ( 2001 ) .
s.r mackay , a.n .
berker , and s. kirkpatrick , phys .
lett . * 48 * , 767 ( 1982 ) . c. amoruso and a.k .
hartmann , cond - mat 0401464 .
j. van mourik , d. saad and y. kabashima , phys .
e * 66 * , 026705 ( 2002 ) .
y. kabashima , cond - mat 0506311 .
s. franz , m. leone , f. ricci - tersenghi , and r. zecchina , phys .
* 87 * , 127 ( 2001 ) .
m. mzard and r. zecchina , phys .
e , * 66 * , 56126 ( 2002 ) .
luby , m. mitzenmacher , m.a .
shokrollahi , and d.a .
spielman , ieee trans .
theory * 47 * , 569 ( 2001 ) .
t. richardson and r. urbanke , codes , systems , and graphical models , b. marcus and j. rosenthal ( eds . ) , new york , ny : springer ( 2001 ) . h. nishimori and d. sherrington , disordered and complex systems , p. sollich , a. coolen , l. hughston and r.streater ( eds . ) , new york , ny : aip , 67 ( 2001 ) ; p. le doussal and a.b .
harris , phys .
lett * 61 * , 625 ( 1988 ) .
considering the replicated partition function , @xmath237 one can perform the average over the corresponding underlying geometry in ( [ partition_f ] ) introducing an integral representation for the finite - connectivity constrain @xmath238 in the spirit of @xcite , defining the set of overlaps @xmath239 and conjugated functional order parameter @xmath240 , @xcite .
this leads , under the rs anstz , @xmath241 to the following expression for the free energy , @xmath242 \nonumber \\ & \times & \ln \big [ 1 + \tanh ( \beta j ) \prod_{j=1}^k \tanh ( \beta x_j ) \big ] \nonumber \\ & + & c \int dx d \hat{x } \pi(x ) \hat { \pi } ( \hat{x } ) \log \big ( 1+\tanh ( \beta x ) \tanh ( \beta \hat{x } ) \big ) \nonumber \\ & + & c \int dy \hat { \pi } ( \hat{x } ) \log \cosh ( \beta \hat{x } ) \nonumber \\ & - & e _ { \zeta } \int \prod_{l=1}^c dy_l \hat { \pi}(x_l ) \big [ 2 \cosh \beta ( \sum_{l=1}^c \hat{x}_l + f \zeta ) \big ] . \label{free - energy2}\end{aligned}\ ] ] the corresponding saddle point equations ( [ crux2 ] ) are derived differentiating the above expression with respect to the local field @xmath243 and its conjugated distribution @xmath244 . in order to obtain the free - energy ( [ free - energy ] ) , the second and third term in ( [ free - energy2 ] ) can be combined via equation ( [ crux2 ] ) , to obtain @xmath245 the fourth term can be written as @xmath246 where we wrote @xmath247 and where the last term in the free - energy expression above is nothing but @xmath248 . it is easy to see that several terms cancels out , together with the first term in equation ( [ free - energy2 ] ) , to obtain equation ( [ free - energy ] ) .
similar simplifications and their interpretation were discussed already in the case of the bethe sg @xcite .
consider the set of coupled equations ( [ crux2 ] ) , @xmath249 for the case of @xmath16 , with no random bias , corresponding to the bethe sg problem , let us define the laplace transform of the conjugated local field distribution as @xmath250 substituting the first equation of ( [ crux2b ] ) into the second and then into ( [ gop2 ] ) , one finds @xmath251 .
\nonumber \\\end{aligned}\ ] ] which is the expression for the global order parameter found in @xcite and corresponds to expression ( [ rs ] ) for @xmath16 .
the expression for arbitrary @xmath6 values of the multi - spin interaction simply involves multiple convolutions of the mottishaw global order parameter .
the meaning of the global order parameter ( [ gop2 ] ) can be understood observing that , e.g. in the case of the bethe sg considered here ( in a uniform magnetic field ) , the partition function is given by @xcite @xmath252 [ g_m ( \sigma_o^ { \alpha } ) ] ^c,\ ] ] where @xmath253 \big ] .
\\ \label{rsb2}\end{aligned}\ ] ] the subscript @xmath254 indicates the dependence of the global order parameter on the sub - shell tree structure . if one is interested in the thermodynamical limit , it is expected that the fixed point of the above recursion determines the equilibrium properties of the system .
a delicate point is the choice of the boundary conditions which are well known @xcite to play an important role in this problem .
the corresponding extension of equation ( [ rsb2 ] ) to the case of arbitrary @xmath6 values has been reported in equation ( [ gopk ] ) .
|
we obtain phase diagrams of regular and irregular finite connectivity spin - glasses .
contact is firstly established between properties of the phase diagram and the performances of low density parity check codes ( ldpc ) within the replica symmetric ( rs ) anstz .
we then study the location of the dynamical and critical transition of these systems within the one step replica symmetry breaking theory ( rsb ) , extending similar calculations that have been performed in the past for the bethe spin - glass problem .
we observe that , away from the nishimori line , in the low temperature region , the location of the dynamical transition line @xmath0 change within the rsb theory , in comparison with the ( rs ) case . for ldpc decoding over the binary erasure channel we find , at zero temperature and rate @xmath1 an rs critical transition point located at @xmath2
while the critical rsb transition point is located at @xmath3 , to be compared with the corresponding shannon bound @xmath4 . for the binary symmetric channel ( bsc )
we show that the low temperature reentrant behavior of the dynamical transition line , observed within the rs anstz , changes within the rsb theory ; the location of the dynamical transition point occurring at higher values of the channel noise .
possible practical implications to improve the performances of the state - of - the - art error correcting codes are discussed .
+ pacs numbers : 89.90+n , 89.70+c,05.50+q +
|
the standard model to understand the large - scale features of the matter distribution in the universe after decoupling from radiation can be hardly simpler : a collection of many identical point particles interacting with each other via the newtonian gravitational force in an expanding spatial background @xcite .
structure arises as a consequence of the gravitational instability of initially tiny density perturbations .
this model neglects relativistic effects , which become important only at scales of the order of the horizon and beyond , or when dealing with relativistic velocities .
the model also excludes nongravitational interactions , which are assumed to be relevant only at small enough scales .
the general solution to the dynamical evolution of this model is unknown due to the mathematical difficulties .
n - body simulations , which numerically solve the dynamical equations ( see eqs .
( [ newton ] ) in the next section ) , have been very helpful in understanding this evolution . in this work
i look for an analytical derivation of some of the relevant features in the formation of large - scale structures .
one is not usually interested in following the detailed path of each particle , but rather in some general properties that typically depend on the behavior of a large number of particles and which in the end are the kind of data provided by observations , e.g. , the smoothed density and velocity fields .
this naturally leads to the use of a _ coarse - grained _ or smoothed description : the evolution of a few , `` macroscopic '' variables is isolated by making suitable approximations for the evolution of the neglected degrees of freedom and for their influence on the relevant ones .
the idea of a spatial coarse - graining has a long history in the field of statistical mechanics , where it has proven quite successful as a powerful tool for extracting information about the dynamical evolution of many - body systems . in this work
i explore its application in the context of cosmological structure formation .
although the procedure of smoothing a field is widely used in cosmology , this application has mainly had a descriptive purpose . unlike this , i study systematically the _ dynamical _ evolution of the smoothed fields
. as we shall see , this method has the merit of providing a common framework for different models ( dust and adhesion ) of structure formation : it offers a clear explanation of the approximations involved in each model and of their physical meaning , so that it also opens the way to systematically relax them and obtain improved models . starting from the microscopic equations of motion of the particles , i derive an exact set of hydrodynamic - like equations for the ( coarse - grained ) mass - density and velocity fields ( sec . [ seccoarse ] ) .
these equations do not form an autonomous set , but rather constitute the first ones of an infinite hierarchy that must be truncated to become a useful tool .
i discuss in sec .
[ secsmallk ] how this can be achieved by resort to a _ large - scale expansion _ , based on the assumption that the dynamical influence of the small scales over the large - scale evolution is weak .
the lowest - order term of the expansion yields the dust model ( sec .
[ secdust ] ) , and it corresponds to a complete neglect of the structure below the coarsening scale .
this leads to an eventual failure of the model ( in the form of pancake - like singularities ) , which can be prevented by considering the first - order correction in the expansion ( sec .
[ secadhesion ] ) : it accounts for the dynamical influence of the structure below the coarsening scale , and i show by means of boundary - layer techniques that it gives rise to a robust `` adhesive '' behavior of the same kind as the adhesion model .
i end up in sec .
[ secconclusion ] with a discussion of the results .
the basic model is a system of nonrelativistic , identical point particles which ( i ) are assumed to interact with each other via gravity only ; ( ii ) look homogeneously distributed on sufficiently large scales , so that the evolution corresponds to an expanding friedmann - lematre cosmological background ; and ( iii ) deviations to homogeneity are relevant only on scales small enough that a newtonian approximation is valid to follow their evolution .
let @xmath0 denote the expansion factor of the friedmann - lematre cosmological background , @xmath1 the associated hubble function , and @xmath2 the homogeneous ( background ) density on large scales .
@xmath3 is the comoving spatial coordinate of the @xmath4-th particle , @xmath5 is its peculiar velocity , and @xmath6 its mass . in terms of these variables
the evolution is described by the following set of equations @xcite ( @xmath7 denotes a partial derivative with respect to @xmath3 ) : [ newton ] @xmath8 @xmath9 @xmath10 , \ ] ] @xmath11 where @xmath12 is the peculiar gravitational acceleration acting on the @xmath4-th particle .
finally , eqs .
( [ newton ] ) must be subjected to periodic boundary conditions in order to yield a newtonian description consistent with the friedmann - lematre solution at large scales @xcite .
to implement the idea of a spatial coarse - graining one employs a smoothing window @xmath13 : this smoothing window should define a bounded region of space and whose inner structure is smoothed out . in app .
[ apwindow ] i discuss the general properties i will require from a smoothing window .
let @xmath14 denote the ( comoving ) coarse - graining length scale .
a microscopic mass density field and a coarse - grained mass density field can be defined respectively as follows : [ density ]
@xmath15 @xmath16 the physical interpretation of the field @xmath17 follows straightforwardly from the properties of the smoothing window : it is proportional to the number of particles contained within the coarsening cell of size @xmath18 centered at @xmath19 .
a microscopic peculiar - momentum density field and the corresponding coarse - grained field can be defined in the same way : [ momentum ] @xmath20 @xmath21 one can introduce peculiar velocity fields @xmath22 and @xmath23 by definition as @xmath24 and similarly for @xmath22 .
the physical meaning of @xmath25 is also simple : it is the center - of - mass peculiar velocity of the subsystem defined by the particles inside the coarsening cell of size @xmath18 centered at @xmath19 . notice that @xmath26 is
_ not _ obtained by coarse - graining @xmath22 : from a dynamical point of view , it is more natural to coarse - grain the momentum rather than the velocity , since the former is an additive quantity for a system of particles
. finally , one can define peculiar gravitational acceleration fields @xmath27 and @xmath28 through a coarse - graining of the force : [ acceleration ] @xmath29 @xmath30 the field @xmath31 has the physical meaning of the center - of - mass peculiar gravitational acceleration of the subsystem defined by the coarsening cell at @xmath19 . from these definitions and eqs .
( [ newton]a ) and ( [ newton]b ) , it is straightforward to derive the evolution equations obeyed by the coarse - grained fields @xmath32 and @xmath26 ( from now on , @xmath33 is taken at constant @xmath19 , and @xmath34 means partial derivative with respect to @xmath19 ) : [ hydro ] @xmath35 @xmath36 where a new second - rank tensor field has been defined : @xmath37 [ { \bf u}_{mic}({\bf y } , t ) - { \bf u } ( { \bf x } , t ; l ) ] .\ ] ] the field @xmath38 is due to the velocity dispersion , i.e. , to the fact that the particles in the coarsening cell have in general a velocity different from that of the center of mass .
the trace of @xmath39 is proportional to the internal kinetic energy of the coarsening cell , that is , the total kinetic energy of the particles in the reference frame of the center of mass .
the physical meaning of eqs .
( [ hydro]a ) and ( [ hydro]b ) is simple : they are just balance equations , stating mass conservation and momentum conservation , respectively . the term @xmath40 represents a _
kinetic _ pressure due to the exchange of particles between neighboring coarsening cells ( just like in the ideal gas ) and it has the same physical origin as the convective term @xmath41 , i.e. , a nonlinear mode - mode coupling of the velocity field .
the difference is that the convective term couples only modes on scales @xmath42 , while the velocity dispersion term codifies the dynamical effect of the coupling of the modes on scales @xmath42 with the modes on scales @xmath43 .
the term @xmath44 codifies the gravitational interaction between the coarsening cells and it is shown later that it can be split in a similar manner into a contribution due to the large scales and another due to the coupling of the large scales with the small ones . although eqs .
( [ hydro ] ) look similar to the ordinary hydrodynamic equations , there is the important difference that these equations are _ exact _ : as one changes the smoothing length , the fields @xmath32 , @xmath26 , @xmath45 , @xmath39 change
but in such a way that the equations remain the same ( for example , upon increase of the smoothing length , part of the dynamical effect described by @xmath46 is shifted towards @xmath40 ) .
this property is reflected in that the equations are not an autonomous system for @xmath32 and @xmath26 .
in fact , they are just the first ones of an infinite hierarchy , as can be checked by computing the evolution equations for the fields @xmath28 and @xmath39 ( see eq .
( [ pieq ] ) , for example ) . to obtain a useful set of equations ,
it is necesary to truncate this hierarchy by looking for a functional dependence of @xmath28 and @xmath39 on @xmath32 and @xmath26
. this will be the task of the next section .
in this section i discuss the closure of the hydrodynamic hierarchy at the level of eqs .
( [ hydro ] ) . when the particle interaction is dominated by a fast - decaying ( `` collisional '' ) force , as in normal fluids , this truncation is achieved by the assumption of local equilibrium ( see , e.g. , refs .
@xcite ) . in this case
, the interaction determines a privileged smoothing scale @xmath14 and it drives the coarsening cells of size @xmath14 towards an approximate internal thermal equilibrium , as if isolated from each other .
the evolution of the large scales ( @xmath47 ) is then ruled by the interaction between neighboring coarsening cells . in this way , for example , eq .
( [ hydro]b ) becomes navier - stokes equation .
one can not , however , apply this approach to the problem in hand : a system with a long - ranged , unshielded interaction such as gravity does not obey the usual thermodynamics ( see , e.g. , the brief review in ref .
@xcite and the more technical remarks in ref .
moreover , this long range implies that the interaction is self - similar and does not pick up itself a favored coarsening length .
one must therefore make use of a different approach to close the hydrodynamic hierarchy . for this purpose ,
i introduce the _ large - scale expansion _ , which relies on the assumption that , in the context of cosmological structure formation , the evolution of the large scales is weakly influenced by what is going on in the small scales .
this assumption is further discussed in sec .
[ secconclusion ] . here
i show how to formulate this idea in order to write the fields @xmath28 and @xmath39 in eqs .
( [ hydro ] ) in terms of @xmath32 and @xmath26 .
let a tilde denote the fourier transform of any field : [ fourier ] @xmath48 @xmath49 where @xmath50 denotes the volume of periodicity for eqs . ( [ newton ] ) .
then the definition ( [ density]b ) can be written as @xmath51 . making use of the taylor - expansion ( [ taylorwin ] ) and formally inverting it ,
one gets : [ micdens ] @xmath52 \ , \tilde{\varrho}({\bf k } ; l ) , \ ] ] @xmath53 \ , \varrho({\bf x } ; l ) .\ ] ] consider first the field @xmath28 .
fourier - transforming its definition ( [ acceleration ] ) and using eqs .
( [ newton]c ) and ( [ newton]d ) , one finds : @xmath54 this expression collects the mode - mode coupling via gravity of the scales @xmath55 with every other scale .
take now a large scale @xmath56 : the assumption of weak coupling to the small scales then means that in the summation in eq .
( [ wfourier ] ) , the main contribution arises also from the large scales , @xmath57 .
one is then justified to insert the expansions ( [ micdens]a ) and ( [ taylorwin ] ) into eq .
( [ wfourier ] ) . transforming back into real space
yields the result : [ wsmallk ] @xmath58
@xmath59 @xmath60 thus , the unknown field @xmath28 has been written as a functional of the coarse - grained density field .
the combination @xmath61 ( @xmath62 ) can be _ formally _ viewed as a parameter which measures the influence of the small scales over the large scales and which has been assumed to be small : hence the name large - scale expansion .
the field @xmath63 , which can be called the _
macroscopic _ gravitational field , represents the gravitational field created by the monopole moment of the matter distribution in the coarsening cells , i.e. , as if they had no spatial extension ( @xmath64 ) .
it obeys the equations that one would have naively guessed and we see that this is not the whole story : there exist a correction due to the higher multipole moments that can then be properly called _ tidal _ correction and it represents the coupling of the large to the small scales induced by gravity .
hence , this decomposition into a macroscopic field and a tidal correction is analogous to the decomposition in eq .
( [ hydro]b ) into convection , @xmath65 , and velocity dispersion , @xmath39 .
the same procedure can be now applied to the field @xmath39 .
now , @xmath66 obeys an expansion like ( [ micdens]b ) , but @xmath22 does not , because it is not defined by a straightforward coarse - graining .
what one has in this case is slightly more involved : @xmath67 \ , { \bf j}}{[1 - { 1 \over 2 } b \ , ( l \ , \nabla)^2 + o(l \ , \nabla)^4 ] \ , \varrho } = \left [ 1 - { 1 \over 2 } b \ , ( l \ , \nabla)^2 - b l^2 { \nabla \varrho \over \varrho } \cdot \nabla + o(l \ , \nabla)^4 \right ] \ , { \bf u } .\ ] ] because of this minor complication , it is simpler in this case to work directly with the representation in real space , eq .
( [ veldis ] ) , into which the expansion ( [ micvel ] ) and the one corresponding to @xmath68 are now inserted .
because the smoothing window effectively restricts @xmath69 , the fields within the integral evaluated at point @xmath70 can be consistently taylor - expanded around point @xmath19 to order @xmath71 .
the final result reads : @xmath72 where a summation over the index @xmath4 is implied . taking this result and ( [ wsmallk ] ) into the hydrodynamic hierarchy ( [ hydro ] ) , and dropping the assumed small corrections @xmath73
, i finally obtain [ hydrosmallk ] @xmath35 @xmath74 \right\ } , \ ] ] @xmath59 @xmath60 this is an autonomous system of equations for the two coarse - grained fields @xmath32 and @xmath26 .
compared to the hydrodynamic equations of a normal fluid , we see that , because of the long range of the interaction , its contribution @xmath44 to the equation for momentum conservation can not be written as the divergence of a tensor , i.e. , as a nonideal correction to the kinetic contribution represented by @xmath39 .
also the expression to lowest - order for @xmath39 is another evident difference . in the next sections ,
i explore the dynamical evolution described by eqs .
( [ hydrosmallk ] ) .
in this section i consider the lowest order in the large - scale expansion .
this corresponds to formally setting @xmath64 in eq .
( [ hydrosmallk]b ) , thus yielding : [ dust ] @xmath75 @xmath76 @xmath59 @xmath60
this is the popular and throughly studied _ dust model _ ( see , e.g. , refs .
the large - scale expansion provides a clear picture of the approximations leading from eqs .
( [ newton ] ) to this model : by setting @xmath64 one assumes that the coarsening cells can be thought of as `` big particles '' lacking completely an internal structure ( fig . [ figdust ] ) .
this implies neglecting ( i ) the velocity dispersion @xmath39 compared to the convection term @xmath77 ; and ( ii ) the spatial extension of the cells and thus the tidal correction compared to the macroscopic gravitational field @xmath63
. it will be useful to briefly review some results for the dust model . in the linear regime of small fluctuations about the homogeneous cosmological background , eqs .
( [ dust ] ) can be linearized and the resulting set of linear differential equations solved . in particular , in the long - time limit ( but still within the linear approximation ) , the peculiar velocity and the gravitational acceleration are related by the condition of parallelism : @xmath78 with the function @xmath79 where @xmath80 is the growing mode of the ( small ) density contrast , i.e. , the growing solution of the equation @xmath81 .
the linear evolution eventually breaks down due to the growth of inhomogeneities by gravitational instability .
the solution to the fully nonlinear equations ( [ dust ] ) is not known , but a successful approximation in this regime is the _
zeldovich approximation _
( za hereafter ) @xcite , which turns out to be an exact solution of eqs .
( [ dust ] ) for some highly symmetric configurations . in the most general case , it can be understood as the extrapolation of the parallelism condition ( [ parallelism ] ) into eq .
( [ dust]b ) @xcite .
more properly , considering the way eqs .
( [ dust ] ) were derived , the actual approximation is the _ truncated _ za @xcite , since the density and velocity fields have been smoothed on a scale @xmath14 .
comparison with n - body simulations @xcite shows that the truncated za performs substantially better than the original za : this is no wonder within the present coarse - graining formalism , since the particles that must obey the `` dust '' evolution ( and thus be moved according to the za ) are not the n - body particles of the simulations , which follow eqs .
( [ newton ] ) , but rather the `` big particles '' which represent the coarsening cells .
if one introduces a rescaled velocity field @xmath82 , and uses @xmath80 as the new temporal variable , then eq . ( [ dust]b ) becomes in the za : [ zeldovich ] @xmath83 @xmath84 together with the irrotationality constraint following from eqs .
( [ parallelism ] ) and ( [ dust]d ) .
hence , the problem reduces to the curl - free evolution of a fluid under no forcing at all .
the solution to this equation is then inserted into the continuity equation ( [ dust]a ) to yield the evolved density field . as is well - known
, however , the za has the problem of giving rise to singularities in the fields , mainly sheet - like singularities or `` pancakes '' @xcite at which @xmath85 and @xmath86 . beyond this moment
, the za ceases to be valid .
the reason lies in the nonlinear , convective term in eq .
( [ zeldovich]a ) , which deforms the initial velocity field and generically leads to a multivalued velocity field ( _ shell crossing _ in the cosmological literature ) .
this feature is likely not exclusive to the za but a generic property of the fully nonlinear dust model , eqs .
( [ dust ] ) , as checked , e.g. , by the application of lagrangian perturbation techniques ( see @xcite and refs .
therein ) .
the generation of singularities by the convective term is not prevented by the gravitational acceleration @xmath63 , which , on the contrary , favors this process .
the emergence of singularities signals the unsuitability of the dust model beyond that time : in the next section i investigate how the correction to dust following from the large - scale expansion regularizes the singularities .
the singularities of the dust model arise because the approximation of the coarsening cells as `` big particles '' is bad when the density of these `` particles '' is large ( formally infinite ) : the interaction between them is no longer the simple macroscopic gravitational force @xmath87 , but it becomes more and more dependent on the internal structure of the coarsening cells . therefore , a first step to improve the dust model is to take into account the first - order correction following from the large - scale expansion and study eqs .
( [ hydrosmallk ] ) . obviously
, the nonlinearities represented by this correction make it even more difficult to solve these equations than in the case of the dust model .
hence , in order to understand the effects of the correction on the dust evolution , i introduce some further simplifying assumptions .
the idea is to consider the limit @xmath88 in eqs .
( [ hydrosmallk ] ) , so that the correction is irrelevant everywhere _ except at those places where the dust evolution would predict a singularity_. therefore , the evolution will follow the za almost everywhere and the parallelism approximation ( [ parallelism ] ) will be good : this implies in particular , via eqs .
( [ hydrosmallk]c ) and ( [ hydrosmallk]d ) , that @xmath26 is also curl - free and that @xmath89 . since the correction to dust in eq .
( [ hydrosmallk]b ) will be effective only near the singularities and these correspond in the za predominantly to pancakes , the correction can be computed under the approximation of a local plane - parallel collapse : @xmath32 and @xmath26 change only along the direction @xmath90 of the local plane - parallel collapse and @xmath91 . making use of these approximations ,
the correction term can be simplified as follows : [ correction ] @xmath92 @xmath93 \approx { 1 \over a \varrho } \nabla [ \varrho ( \nabla \cdot { \bf u})^2 ] \approx { 3 \over a } ( \nabla \cdot { \bf u } ) \nabla^2 { \bf u } \quad ( \propto - \varrho \nabla \varrho ) , \ ] ] where it has been taken into account that the correction will be effective only at those places where @xmath94 . in this limit
, the contribution ( [ correction]b ) due to the velocity dispersion will be dominant over that of the tidal correction ( [ correction]a ) . inserting this result in eq .
( [ hydrosmallk]b ) and in terms of the variables @xmath95 and @xmath96 previously introduced , i finally arrive at the following model : [ adhesion ] @xmath97 @xmath84 this is a closed set of equations for the velocity field : once solved , this field is used to compute the evolved density field from the continuity equation ( [ hydrosmallk]a ) , much in the same way as with the za .
the coefficient @xmath98 was called the _ gravitational multistream _ viscosity in ref . @xcite . because it multiplies the highest - order derivative in eq .
( [ adhesion]a ) , one can not simply drop this term in the limit @xmath99 , but one must rather apply the techniques of the boundary - layer theory to study this limit . the mathematical handling is collected in the next subsection ; here i simply quote the conclusion that , as expected , the correction regularizes the pancake - like singularities predicted by the dust model , which become robust structures , where more and more mass gets `` adhered '' .
another conclusion is that this `` adhesive '' behavior is rather insensitive to the detailed functional dependence on @xmath100 of the factor in front of @xmath101 : hence , the same behavior arises in particular if one simply drops the @xmath102 to obtain a linear correction .
but then , one recovers the original adhesion model @xcite , which was introduced as a phenomelogical ( and , when compared against n - body simulations @xcite , very successful ) correction to the za to go beyond the epoch of formation of singularities . in the framework of the large - scale expansion ,
the physical explanation of the adhesive behavior is clear ( fig .
[ figadh ] ) .
the corrections to dust ( [ correction ] ) behave as a force in the opposite direction to the density gradient , so that in the neighborhood of a pancake there is a competition between the inflow of matter driven by the gravitational attraction and this `` repulsion '' , which prevents the formation of a singularity .
the dominant contribution as @xmath103 is the correction ( [ correction]b ) due to the velocity dispersion and represents the conversion of `` coherent '' streaming kinetic energy into `` disordered '' internal kinetic energy within the coarsening cells @xcite ( the mechanism has the same origin as the pressure and viscous - like forces in a gas ) . the subdominant contribution ( [ correction]a ) from the tidal correction also exhibits this `` repulsive '' behavior , and means that the macroscopic field @xmath63 overestimates the gravitational attraction between coarsening cells . in this subsection
i apply the theory of boundary layers @xcite to the model ( [ adhesion ] ) .
the physical picture is that one can study the evolution almost everywhere setting @xmath104 .
however , there arise regions ( `` shocks '' ) spatially well separated from each other and of vanishing volume ( as @xmath99 ) where @xmath105 has a discontinuity , @xmath106 diverges and the effect of the term multiplied by @xmath98 must be taken into account .
the fact that these shocks correspond mainly to plane - parallel structures greatly simplifies the technical handling , since it reduces the problem to the solution of an ordinary differential equation .
the purpose is to study eqs .
( [ adhesion ] ) in the vicinity of a pancake , so that one takes a point at the pancake and introduces a new coordinate system moving rigidly with it and with one of the axis normal to the pancake . in this new noninertial reference frame eq .
( [ adhesion]a ) must be appended with a term @xmath107 which collects the acceleration due to the inertial forces , and the velocity @xmath105 is now understood as relative to this new reference frame .
new rescaled coordinates are defined as @xmath108 and the idea is to take @xmath109 in such a way that the fields and their derivatives ( and consequently also the thickness of the pancake ) remain finite in terms of the new coordinates @xmath110 even in the limit @xmath99 .
( the prime of the rescaled coordinates will be dropped hereafter to simplify the notation ) .
let @xmath111 , @xmath112 denote the coordinates normal and parallel to the plane of the pancake , respectively , and @xmath113 , @xmath114 the corresponding components of the velocity @xmath105 .
let @xmath115 denote the gradient with respect to @xmath112 .
in terms of these new coordinates , eqs .
( [ adhesion ] ) now read : [ epsilonadhs ] @xmath116 @xmath117 where i have generalized the model ( [ adhesion ] ) by allowing the prefactor of @xmath101 to be an arbitrary function of @xmath100 whose asymptotic behavior as @xmath118 is characterized by the power @xmath119 ( the case @xmath120 recovers the original adhesion model ) .
it has also been considered the case that @xmath98 ( and hence , the rescaling factor @xmath121 ) may depend on time .
the motivation for this generalization is twofold : ( i ) this kind of behavior corresponds to the models discussed in refs .
@xcite , where the hydrodynamic hierarchy ( [ hydro ] ) is closed by simply neglecting tidal corrections altogether and assuming @xmath122 , @xmath123 .
( ii ) the computation of the trace of @xmath39 from n - body simulations also yields this kind of behavior @xcite .
the factor @xmath121 must be chosen so as to absorb the explicit dependence on @xmath98 in eq .
( [ epsilonadhs]a ) and thus to render the coefficient of the highest - order derivative of order unity , namely @xmath124 and requiring @xmath125 , so that @xmath109 .
this latter constraint states that the correction to dust must grow fast enough with @xmath106 so as to succesfully oppose the gravitational attraction and overcome the formation of a singularity .
( [ epsilonadhs ] ) are now simplified by keeping only the dominant terms in the limit @xmath109 : first , one can set @xmath126 , because in terms of the rescaled coordinates the pancake looks like an infinite plane and the spatial variations along it take place over an infinitely large lenght scale : only the derivative normal to the pancake is relevant .
also , the acceleration terms @xmath127 and @xmath107 are finite in the neighborhood of pancakes .
finally , @xmath128 is of order @xmath121 at any finite time .
therefore , eqs .
( [ epsilonadhs ] ) reduce to the following system of ordinary differential equations : [ blayer ] @xmath129 @xmath130 the physical interpretation of these equations is straightforward .
the tangential velocity , @xmath114 , is smooth at the pancakes and so constant in the rescaled coordinates . the normal velocity , which looks discontinuous in the physical coordinates , is determined in rescaled coordinates by the balance between the convective transport towards the pancake ( governed in turn by the gravitational attraction of the pancake ) and the outwards `` pressure '' due to the velocity dispersion .
the boundary conditions to be imposed to these equations are : ( i ) @xmath131 , meaning that the rescaled coordinate system is centered at the pancake and moves with it ; ( ii ) @xmath132 for any @xmath111 , so that there is an inflow of matter towards the pancake ( and @xmath133 , see eq .
( [ densitypancake ] ) ) ; ( iii ) @xmath134 as @xmath135 , so that the derivative in physical coordinates is not divergent outside of the pancake . integrating twice eq .
( [ blayer]a ) with these boundary conditions , one finds an implicit solution @xmath136 : @xmath137 where @xmath138 is an integration constant .
knowning @xmath139 , one can get an approximation to the pancake density profile by applying the parallelism condition ( [ parallelism ] ) together with poisson equation ( [ hydrosmallk]c ) in the limit @xmath140 : @xmath141 whose solution reads @xmath142 ^{1 \over \gamma-1 } .\ ] ] a detailed study of the solution ( [ generalsol ] ) leads to the general picture shown in fig . [ figpanc ] : the fields remain univalued at all times and no singularity arises .
the parameter @xmath143 measures the pancake thickness in the physical coordinates .
this study also shows that for the cases @xmath144 the solution approaches so fast its asymptotic values at large distances that in fact @xmath145 at two finite values of the normal coordinate , @xmath146 and @xmath147 . in such cases ,
the solution ( [ generalsol ] ) must be replaced beyond these points by : @xmath148 , @xmath149 ( @xmath150 ) . from the general solution ( [ generalsol ] )
, one can compute the behavior of the solution @xmath151 in the limit @xmath109 for a fixed value @xmath152 of the physical coordinate @xmath153 : @xmath154 independently of the value of @xmath119 and the temporal dependence of @xmath98 .
the conclusion is therefore that the adhesive - like behavior is a robust property of the whole family of models ( [ epsilonadhs ] ) parametrized by @xmath119 : whenever the dust evolution ( @xmath104 ) predicts a singularity , this must be replaced by the `` adhesive prescription '' ( [ limit ] ) , which leads , via the continuity equation ( [ hydrosmallk]a ) , to a steady accretion of mass .
i have derived and applied a hydrodynamic - like formulation for the process of cosmological structure formation : this is a nontrivial statement , because one could believe that the collisionless nature of the basic model ( [ newton ] ) renders a `` fluid '' description impossible after shell - crossing , and that then one should employ a different approach , e.g. the bbgky hierarchy .
i have shown , however , that a `` fluid '' description is feasible after the breakdown of the dust model : what makes a difference with `` down - to - earth '' fluids is how the closure of the hydrodynamic hierarchy ( [ hydro ] ) is achieved .
for this pupose , i have introduced the large - scale expansion , which builds on the assumption that the large - scale evolution is insensitive to the small scales .
this assumption lies behind many reasonings in the cosmological literature : for example , behind the idea that on large scales the evolution follows a friedmann - lematre model , regardless of the small - scale inhomogeneities , and also behind the confidence on cosmological n - body simulations , where each n - body particle is so massive that it must correspond in the real world to a full structure in its own , composed of many smaller particles .
the plausibility of the assumption can be argued on the basis of the long range of gravity : the evolution can be expected to be dominated by the large scales provided there is enough large - scale power initially .
( one can then also expect that the validity of the large - scale expansion should depend on the initial and boundary conditions ) .
the good agreement with n - body simulations of the truncated za and the adhesion model @xcite , in which small - scale structure is completely disregarded , can be viewed as a support of this large - scale dominance . from the derived hydrodynamic equations ,
i have shown how the dust model arises as the lowest - order term in the large - scale expansion , while the first - order correction gives rise to a model which can be reduced to the adhesion model by further approximations .
this derivation provides a clear physical interpretation of the two models : the `` particles '' which are assumed to follow the dust model are not such but have an internal structure , and the `` interaction '' between these `` particles '' due to its internal structure explains the adhesive behavior . by applying boundary - layer techniques , i found in particular that this behavior is a robust property of the model , in the sense that it arises quite independently of the detailed dependence of the velocity dispersion and the tidal correction on the density and velocity fields .
it is interesting to compare this work with previous , related works dealing also with a hydrodynamic - like formulation of large - scale structure formation . in refs .
@xcite , the case is studied in which the hydrodynamic hierarchy is truncated by a _
phenomenological _ ansatz which writes the corrections to the adhesion model as a stochastic term ( a noise ) . the adhesion model , in turn , was justified in refs .
@xcite ( see ref .
@xcite for a relativistic generalization ) by disregarding the tidal correction altogether and by letting the velocity dispersion be given as @xmath155 .
the robustness of the adhesive property already mentioned explains why this behavior also arises in the models that follow from this truncation .
it is particularly instructive to compare with the work in ref .
@xcite , where a truncation relying on something else than phenomenology was studied .
i have been able to close the hydrodynamic hierarchy with less restrictive assumptions and thus shown that the adhesive behavior can still be recovered when assumptions ( a2 ) ( isotropic @xmath39 ) and ( a4 ) and ( a5 ) ( further restrictions on the form of @xmath39 ) in sec . 4 of ref .
@xcite are dropped .
assumption ( a1 ) seems in practice equivalent to the large - scale expansion , while i also employed assumption ( a3 ) ( parallelism ) in order to derive adhesion - like models .
but there is also a major improvement compared to that work : i do not assume the mean - field approximation from the outset ( and thus the vlasov equation , which was the starting point in ref .
in fact , i could derive an expression for the tidal correction and show that it also behaves `` adhesively '' . as explained in more detail in app .
[ appi ] , this correction is also the origin for the discrepancy between expression ( [ pismallk ] ) and the relationship @xmath156 derived in ref .
@xcite .
finally , i have shown that the derivation of adhesion - like models requires in principle further assumptions than the simple large - scale expansion .
thus , one could improve on these models by relaxing those assumptions and rather studying eqs .
( [ hydrosmallk ] ) .
an important difference between the adhesion - like models and eqs .
( [ hydrosmallk ] ) is that the corrections to dust in the latter generate vorticity , even if it is initially absent : this may be a relevant feature when modelling galaxy formation .
another difference is that now there is no need in principle to retain the condition @xmath157 : pancakes and other singularities are no longer of vanishing volume , but have an inner structure whose evolution could be studied with eqs .
( [ hydrosmallk ] ) . for such purpose
, the role of the length @xmath14 must be better understood .
i would like to thank c. beisbart , t. buchert and h. wagner for their useful comments on the manuscript .
a _ smoothing window _
@xmath13@xcite should behave as a _ window _ that defines a bounded region of space and as a _ smoothing _ filter that erases the structural details inside this region .
these conditions are implemented by requiring @xmath13 and its fourier transform @xmath158 to decay faster than any power as @xmath159 or @xmath160 , respectively .
the smoothing window acts as an integral kernel : if @xmath161 is a given field , then the associated coarse - grained field @xmath162 over the scale @xmath14 and its fourier transform @xmath163 are given by [ coarsegraining ] @xmath164
@xmath165 therefore , the coarse - grained field at point @xmath19 is just the ( weigthed ) addition of the original field over a region of size @xmath18 around that point .
two more conditions are also required : ( i ) @xmath166 , so that the contribution from the neighborhood of the center of the window is unweighted ; ( ii ) @xmath167 , so that the coarse - grained field has the same large - scale structure as the original field .
this latter condition implies the following normalization : @xmath168 so that @xmath169 , and @xmath170 .
the condition on the decay of @xmath158 for large @xmath171 prevents the window @xmath13 from having sharp borders , which thus becomes a `` fuzzy '' window . otherwise , the coarse - grained field @xmath162 could change in a discontinuous manner as the center of the window , @xmath19 , sweeps the system . in the same way , the fast decay of @xmath13 for large @xmath172 ( which implies that @xmath158 lacks sharp borders in fourier space ) implies the property of spatial locality for the coarse - grained field .
this also guarantees the existence of the taylor - expansion of @xmath158 at @xmath173 in the form : @xmath174 where the constant @xmath175 is related to the quadrupole moment of the smoothing window .
the required constraints on the smoothing window exclude a step function ( either in real or in fourier space ) .
a useful function which satisfies the constraints is a gaussian : @xmath176 .
in this appendix i study the evolution equation for the velocity dispersion and show that ( [ pismallk ] ) is a solution of the equation .
the purpose is to compare with the result in ref .
@xcite for @xmath39 , also obtained from the evolution equation for this tensor .
starting from the definition ( [ veldis ] ) and eqs .
( [ newton]a ) and ( [ newton]b ) , one can obtain the following equation for the temporal evolution of the velocity dispersion : @xmath177 where a summation over repeated indices is implied , and a second - rank and a third - rank tensor fields have been defined : [ pandl ] @xmath178 [ { \bf w}_{mic}({\bf y } , t ) - { \bf w } ( { \bf x } , t ; l ) ] + \nonumber \\ \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \qquad\qquad + [ { \bf w}_{mic}({\bf y } , t ) - { \bf w}({\bf x } , t ; l ) ] [ { \bf u}_{mic}({\bf y } , t ) - { \bf u } ( { \bf x } , t ; l ) ] \big\ } , \end{aligned}\ ] ] @xmath179 [ { \bf u}_{mic}({\bf y } , t ) - { \bf u } ( { \bf x } , t ; l ) ] [ { \bf u}_{mic}({\bf y } , t ) - { \bf u}({\bf x } , t ; l ) ] .\ ] ] the term @xmath180 accounts for the change of velocity dispersion due to the exchange of particles between the coarsening cells , while the term @xmath181 represents the change due to the gravitational interaction . to get a physical picture ,
just consider the equation for the trace of @xmath39 ( the internal kinetic energy ) : then @xmath180 reduces to the equivalent of the kinetic contribution to the heat flux in the usual hydrodynamics , while @xmath181 becomes the power performed by the gravitational interaction .
compared to eq .
( 4c ) of @xcite , eq .
( [ pieq ] ) contains the extra term @xmath181 , and the reason is that this term drops if tidal corrections are neglected . indeed , if one performs a large - scale expansion of @xmath181 and @xmath180 in the same way as explained in sec .
[ secsmallk ] , one obtains in ref .
@xcite , eq .
( [ pieq ] ) ( with @xmath184 , as explained ) was solved for the trace of @xmath39 by imposing the constraint of shear - free flow : it is then found that @xmath185 , where @xmath186 is determined by the initial velocity dispersion . this must be viewed as a solution valid only for early times ;
when the tidal corrections grow , the term @xmath181 becomes relevant : @xmath39 eventually forgets its initial condition , due to the term @xmath187 in eq .
( [ pieq ] ) , and becomes `` slaved '' to the density and velocity fields , as given by eq .
( [ pismallk ] ) .
this is a mechanism similar to the `` slaving '' represented by the parallelism condition ( [ parallelism ] ) in the linearized dust evolution .
s. bildhauer and t. buchert , prog .
phys . * 86 * , 653 ( 1991 ) ; t. buchert , mon . not .
. soc . * 254 * , 729 ( 1992 ) ; l. a. kofman , in _ iuap proceedings on nucleosynthesis in the universe _ , edited by k. sato ( kluwer : dordrecht , 1991 ) .
s. n. gurbatov , a. i. saichev , and s. f. shandarin , mon . not .
. soc . * 236 * , 385 ( 1989 ) ; d. h. weinberg , and j. e. gunn , mon . not .
r. astron .
soc . * 247 * , 260 ( 1990 ) ; l. kofman , d. pogosyan , s. f. shandarin , and a. l. melott , astrophys . j. * 393 * , 437 ( 1992 ) ; a. l. melott , s. f. shandarin , and d. h. weinberg , astrophys . j. * 428 * , 28 ( 1994 ) ; b. s. sathyaprakash _
et al . _ , mon . not .
. soc . * 275 * , 463 ( 1995 ) .
|
a hydrodynamic formulation of the evolution of large - scale structure in the universe is presented .
it relies on the spatially coarse - grained description of the dynamical evolution of a many - body gravitating system .
because of the assumed irrelevance of short - range ( `` collisional '' ) interactions , the way to tackle the hydrodynamic equations is essentially different from the usual case .
the main assumption is that the influence of the small scales over the large - scale evolution is weak : this idea is implemented in the form of a _ large - scale expansion _ for the coarse - grained equations .
this expansion builds a framework in which to _ derive _ in a controlled manner the popular `` dust '' model ( as the lowest - order term ) and the `` adhesion '' model ( as the first - order correction ) .
it provides a clear physical interpretation of the assumptions involved in these models and also the possibility to improve over them .
phys .
rev .
d ; submitted : may 18 , 2000 ; accepted : july 18 , 2000
|
a characteristic signature of young rotation - powered pulsars is the phenomena of `` glitches '' , sudden discontinuous changes in their spin periods ( e.g. , see lyne & graham - smith 1998 ) .
the physical causes of these glitches are not understood .
suggestions include sudden changes in the neutron star ( ns ) crust configuration ( `` starquakes '' ) , abrupt reconfiguration of the magnetic field , or perhaps to the sudden unpinning of vortices in the superfluid neutrons in the inner part of the ns crust . for the latter , the amplitude of the glitch provides an estimate of the fractional part of the moment of inertia carried by superfluid neutrons ( lyne et al .
1996 ) .
the largest glitches have relative amplitudes ( @xmath4 ) of several parts per million , but the range of amplitudes covers many orders of magnitude . often there is a partial recovery back toward the pre - glitch rotation rate on a time scale of @xmath5 days , however , the spin - down rate may be permanently altered .
lyne ( 1995 ) noted that the amount of recovery in the rotation rate tends to be inversely proportional to the characteristic age of the pulsar .
for some pulsars ( e.g. crab ) the glitches are accompanied by a persistent increase in the spin - down rate with a relative amplitude of a few @xmath6 .
this increase may be caused , for example , by changes in the alignment of the magnetic field because of starquakes ( allen & horvath 1998 ; link et al .
1998 ) . in this paper
we present preliminary results on the first detection of glitches from the two lmc crab - like pulsars .
all data presented herein were obtained with the pca instrument on - board the rossi x - ray timing explorer ( _ rxte _ ) for the purpose of monitoring the 16 ms pulsar psr j0537 - 6910 ( marshall et al .
however , the 50 ms psr b0540 - 69 , located @xmath7 away , lies well within the rxte @xmath8 ( fwhm ) field of view and was thus simultaneously observed .
the duration and spacing between observations were optimized to phase - link the pulses from psr j0537 - 6910 which is more then sufficient to allow a similar analysis for psr b0540 - 69 , which is both slower and brighter than psr j0537 - 6910 .
the pca is sensitive to x - rays in the @xmath9 kev band , with a spectral resolution @xmath10 at 6 kev and a time resolution of @xmath11 . to optimize the pulsar detection , we selected @xmath12 kev photons from the top layer of the pca only .
photon arrival times were corrected to the solar system barycenter .
our timing analysis procedure is summarized below ( see zhang et al .
2001 for details ) . for each observation we folded the arrival times into 20 phase bins over a range of frequencies centered on an initial value found from an fft of the data set ( without accounting for any @xmath13 ) .
from the folded profile we determined a pulse time - of - arrival ( toa ) for each observation in an iterative manner , initially using a linear fit to the toa s then adding higher order derivative terms as needed . using the above prescription we fitted the toa s for the psr 0540 - 69 data sets .
a linear fit was used as a quadratic term was not needed to improve the fit .
however , we found an apparent timing glitch had occurred , too small to show up directly in a frequency residuals but large enough to be evident in the analysis of the toas .
an improved fit was obtained by breaking up the data set up into two groups and fitting each separately .
figure 1 presents the pulse arrival times fitted with the ephemeries of zhang et al .
we determine that the glitch occurred at mjd @xmath14 .
the glitch at this epoch changed in frequency and frequency derivative @xmath15 and @xmath16 , respectively ; the magnitude of @xmath17 for this glitch is @xmath18 times larger than the one deeter , nagase , & boynton ( 1999 ) suggested .
our timing result for psr b0540 - 69 is most consistent with the extrapolated ephemeris of deeter et al .
( 1987 to 1991 ) and agrees to within the uncertainties , while that of boyd ( spanning 1979 to 1993 ) over - predicts our measured frequencies by tens of micro - hertz .
our braking index of @xmath19 is significantly lower than than the values of @xmath20 ( boyd et al .
1995 ) and @xmath21 ( deeter et al . 1999 ) .
eikenberry et al . (
1998 ) found @xmath22 whose errors are dominated by timing noise .
however , all previous measurements explicitly assumed that the pulsar had not suffered any glitches during the observation intervals .
given our new result , and the apparent inconsistent in the reported braking index , it is likely that there may have been many other glitches in the timing history of psr b0540 - 69 . in particular , when we combine the data points in table 1 of boyd et al .
( 1995 ) with the frequency measurements of the work presented herein , we obtain a braking index of @xmath23 , significantly lower than @xmath20 . based on the glitch rates from other crab - like pulsars , such as psr j537 - 6910 ( see below ) and on vela ( nine glitches in 25 yr ; lyne et al .
1996 ) , it is certainly possible that psr b0540 - 69 could have an average glitch rate of one per year . since our measurements come from phase - linked data sets , they likely provide the most accurate measure of the secular braking index . following a similar analysis
as described above , for psr j0537 - 6910 we find six large glitches occurred during the first 900 days of our monitoring campaign . the intervals between glitches ranged between @xmath24 to @xmath25 days .
the relative increases of the pulse frequency ranged from @xmath26 to @xmath27 , and relative amplitude ranged from @xmath28 in frequency .
glitches with amplitudes as small as @xmath29 could have easily been detected .
their absence suggests that small glitches are relatively rare in psr j0537 - 6910 . with the exception of the jan 2000 observations , we are able to maintain the cycle - count of the pulse period for all data between the glitches .
the long term spindown of the pulsar continues to average @xmath3 hz / s , i.e. , the same value previously reported . with continued monitoring
we should be able to determine the braking index , @xmath30 for psr j0537 - 6910 ; this may allow us to distinguish between possible physical mechanisms contributing to the spindown . although psr j0537 - 6910 is the oldest and most rapidly rotating crab - like pulsar , the observed rate of glitching appears to be the highest of all known crab - like systems .
the frequency and size of the glitches are comparable to the extreme values for any pulsar .
the pulsars exhibiting the largest known rates of glitches are psr j1740 - 3015 , for which nine glitches have been observed in 8 years ( shemar & lyne 1996 ) , and psr j1341 - 6220 , for which 12 glitches have been seen in 8.2 years ( wang et al . 2000 ) .
most of these glitches were relatively small , however , with typical relative amplitudes of @xmath31 .
however , the vela pulsar has had nine major glitches in 25 years of comparable size as those found for psr j0537 - 6910 ( lyne et .
al 1996 ) .
allen , m.p . & horvath , j.e .
1997 , , 287 , 615 .
boyd , p. t. , et al .
1995 , apj , 448 , 365 caraveo , p. a. , bignami , g. f. , mereghetti , s. , & mombelli , m. 1992 , apj , 395 , l103 deeter , j. e. , nagase , f. , & boynton , p. e. 1999 , apj , 512 , 300 eikenberry , s. s. , fazio , g. g. , & ransom , s. m. 1998 , apj , 492 , 754 link , b. et al .
1998 , , 508 , 838 .
lyne , a.g .
1995 , in lives of neutron stars , p. 167 .
lyne , a.g .
1996 , , 281 , l14 .
lyne , a. g. , pritchard , r. s. , graham - smith , f. , & camilo , f. 1996 , nature , 381 , 497 lyne , a. g. , pritchard , r. s. , & smith , f. g. 1988 , mnras , 233 , 667 marshall , f. e. , gotthelf , e. v. , zhang , w. , middleditch , j. , & wang , q. d. 1998 , apj , 499 , l179 shemar , s.l . & lyne , a.g .
1996 , , 282 , 677 .
wang , n. , manchester , r.n . , pace , r.t . ,
bailes , m. , kaspi , v.m . , stappers , b.w . , &
lyne , a.g .
, mnras , 317 , 843 zhang , w. , marshall , f. e. , gotthelf , e. v. , middleditch , j. , & wang , q. d. 2001 apj , 54 , l177
|
we are undertaking an extensive x - ray monitoring campaign of the two crab - like pulsars in the large magellanic clouds , psr b0540 - 69 and psr j0537 - 6910 .
we present our current phase - connected timing analysis derived from a set of 50 pointed x - ray observations spanning several years . from our initial 1.2 yr monitoring program of the young 50 ms pulsar psr b0540 - 69 , we find the first compelling evidence for a glitch in its rotation .
this glitch is characterized by @xmath0 and @xmath1 .
taking into account the glitch event , we derive a braking index of @xmath2 , significantly lower than previous reported .
for the 16 ms pulsar , psr j0537 - 6910 , we recorded 6 large glitch events during a period of nearly 3 years , the highest rate of all known crab - like systems . despite the extreme timing activity
, the long term spin - down of this pulsar continues to average @xmath3 hz / s
. 73 # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
many problems in science and engineering can be expressed as a partial differential equation ( pde ) .
finding accurate numerical solutions is often important for these applications , but there are many challenges associated with computing these solutions .
nonlinear pdes can cause numerical methods to destabilize@xcite . finding accurate solutions can require high resolution , large amounts of memory , small time - steps , and a lot of computation time .
numerical methods that compute solutions for pdes must approximate derivatives .
one way to achieve this is with a finite difference ; one of the simplest such methods is the forward difference given by @xmath0 ( * ? ? ?
1 ) shows that the finite difference approximation will converge to the true value at a rate polynomial in @xmath1 ; that is , as @xmath1 gets smaller , the approximation gets better at a polynomial rate .
getting a very accurate approximation of the derivative may require @xmath1 to be extremely small , which may exhaust memory requirement or take a lot of time , especially in the 3d case .
4 ) goes on to show that spectral methods converge faster than any polynomial , so we can get very accurate approximations with a relatively larger @xmath1 .
this can save a lot of time and memory .
if periodic boundary conditions are used , we use the fourier transform to compute approximate spatial derivatives .
this paper considers the sine - gordon equations in 1 , 2 , and 3 dimensions .
the 1d problems considered here are all suited to being run on one processor .
however , as the dimensionality is increased , several issues make using multiple processors necessary .
computing multi - dimensional fourier transforms amounts to computing multiple one - dimensional fourier transforms .
these one - dimensional transforms are independent and can be computed more quickly in parallel .
memory requirements also increase dramatically as the dimensionality increases .
problem sizes of interest , especially in 3d , may require more memory than one machine can reasonably have , so using multiple computers may be necessary . using multiple computers
also gives more total cache , which generally improves performance .
the sine - gordon equation is a nonlinear partial differential equation given by @xmath2 this equation arises in many different applications , including propagation of magnetic flux on josephson junctions , sound propagation in a crystal lattice , and several others discussed in @xcite .
we consider the 1 , 2 , and 3 dimensional cases given by @xmath3 the one dimensional sine - gordon equation is exactly integrable .
this means a large class of exact solutions can be found using the inverse scattering transform , including a special class of solutions called solitons@xcite .
solitons are nonlinear effects that lead to travelling waves that are localized and permanent .
solitons are interesting physically and as a way to check numerical codes . for the sine - gordon equation
, some of the explicit exact solutions can not be easily evaluated@xcite . for these solutions
, it may be may be easier to compute the solution numerically and can still give great insight .
the sine - gordon equation is known to have soliton solutions in the 1 and 2 dimensional cases@xcite .
the sine - gordon equation has a hamiltonian @xmath4 given by @xmath5 we give each component of the hamiltonian a name and interpretation@xcite .
@xmath6 is the kinetic energy , @xmath7 is the strain energy , and @xmath8 is the potential energy .
@xmath4 is the total energy and is constant .
these quantities are useful in evaluating the accuracy of numerical schemes and the nature of a solution . the discrete fourier transform ( dft )
can be used to approximate derivatives .
the dft and inverse dft are given by @xmath9&=\frac{1}{2\pi}\sum_{k =- n/2 + 1}^{n/2 } \hat{v}_ke^{ikx_j},\ ; j=1,\ldots , n\end{aligned}\ ] ] conceptually , we can think of the dft as decomposing @xmath10 $ ] into into a set of orthogonal complex exponentials .
each @xmath11 represents the magnitude and phase of its corresponding complex exponential . for smooth functions ,
the fourier coefficients converge faster than any polynomial@xcite . using the fast fourier
transform ( fft ) algorithm , the dft can be computed in @xmath12 time@xcite .
computing the derivative in fourier space is a simple multiplication by the wave number @xmath13 and @xmath14 .
@xmath15}{\partial x } & = \frac{1}{2\pi}\sum_{k =- n/2 + 1}^{n/2 } \frac{\partial}{\partial x}\hat{v}_ke^{ikx_j},\ ; j=1,\ldots , n \\ & = \frac{1}{2\pi}\sum_{k =- n/2 + 1}^{n/2 } ik\hat{v}_ke^{ikx_j},\ ; j=1,\ldots , n\end{aligned}\ ] ] following @xcite and @xcite , we use a leapfrog method to approximate the second derivative in time with a central difference given by : @xmath16 where @xmath17 is the time step size and @xmath18 is the approximation to the function @xmath19 at the @xmath20th time step .
we approximate the spatial derivative with a spectral method .
differentiation in real space becomes multiplication by the wave number in fourier space , giving @xmath21 where @xmath22 and @xmath23 are the wave numbers for the @xmath24 and @xmath25 dimensions , respectively .
in the 1d ( and 2d ) case , the @xmath23 ( and @xmath26 ) wave numbers are not present . we note that because the sine function is nonlinear , @xmath27 and we must recompute @xmath28 separately from @xmath29 for every time step . putting everything together , we have @xmath30 moving terms in around to compute @xmath31 gives @xmath32 \end{split}\ ] ]
the 1d , 2d , and 3d codes are all implemented differently to illustrate many programming techniques .
the 1d code is relatively straight - forward fortran implementation .
the 2d cases of interest are small enough to fit on one gpu , so the 2d code is implemented in c using nvidia s cuda .
the ffts and time - stepping scheme are run entirely on the gpu .
we have found that gpu codes can run approximately 50 times faster for this type of code@xcite .
the 3d problems of interest use grid sizes that are too large to easily run on one computer , so we provide a parallel mpi implementation based on @xcite . the 2decomp library is used to provide an efficient abstraction for performing 3d ffts@xcite . the 1d and 2d plots were generated using python s matplotlib@xcite .
the 3d plots were generated using visit@xcite .
visit is useful for these because it can run in parallel .
the 3d codes can generate a large amount of data , so this feature is very important .
one type of solution for the sine - gordon equation is called a breather .
a breather is a nonlinear mode that is localized in space and oscillates with time . in @xcite
, enz shows that these breathers can be interpreted as de broglie waves ; that is , a particle that is also a wave . in the 1d case , a breather is given by@xcite@xcite @xmath33 where @xmath34 is a parameter that determines the size and frequency of the pulse and @xmath35 is the speed of the pulse .
shows a numerical solution with a stationary and moving breather colliding . after colliding ,
both breathers continue on their original trajectories .
@xcite give initial conditions for stationary and moving breathers in 2d .
@xmath36\ ] ] for @xmath37 and @xmath38 constant this gives a stationary breather , otherwise the breather is moving .
shows the simulation results for initial conditions for a stationary and moving breather . shows the results of running the simulation for the 2d ring solitons .
a 3d pseudo - breather was constructed based on @xcite .
it is called a pseudo - breather because , like a breather , it oscillates in place .
the initial conditions are @xmath39
so far , the 3d codes have only been run on relatively small machines with small grid sizes .
further work includes moving to a larger machine and larger grid sizes , as well as trying more complex interactions between a larger number of solitons .
this research was conducted as an internship with the blue waters undergraduate petascale education program ( bw - upep ) .
many thanks to shodor and the national computational science institute ( ncsi ) .
h. childs , e. brugger , k. bonnell , j. meredith , m. miller , b. whitlock , and n. max .
a contract based system for large data visualization . in _ visualization , 2005 .
ieee _ , pages 191198 .
ieee , 2005 .
|
the sine - gordon equation is a nonlinear partial differential equation .
it is known that the sine - gordon has soliton solutions in the 1d and 2d cases , but such solutions are not known to exist in the 3d case .
several numerical solutions to the 1d , 2d , and 3d sine - gordon equation are presented and comments are given on the nature of the solutions .
|
cold atomic systems have provided an ideal playground for the experimental studies of the unitary evolution of thermally isolated quantum systems .
@xcite apart from the experimental interest , studies of quantum evolution have shed light on some fundamental questions such as thermalization .
@xcite a widely - studied scenario is the unitary evolution of a quantum state when the hamiltonian is suddenly changed a `` quantum quench '' , @xcite which has been the subject of many recent studies ( see refs .
[ ] for a few examples ) . a challenging question , which has been addressed theoretically in one dimensional systems , is the evolution of the entanglement entropy following the quench .
@xcite more generally , such studies are concerned with the fate of an out - of - equilibrium quantum state with a hamiltonian that changes with some time - dependent protocol .
the case where the coupling constants are changed across a quantum phase transition is of special interest . in recent experiments with spinor condensates , for example
, the formation of ferromagnetic domains was observed in a quench to a hamiltonian with a symmetry - breaking ferromagnetic ground state .
@xcite even though the interactions after the quench favor a certain type of order , it is not at all obvious that the order should emerge out of equilibrium . since symmetry - breaking order has a local order parameter , the non - equilibrium state can support the formation of ordered domains .
but what about quenches across a topological phase transition ?
would any of the topological characteristics of the equilibrium phase show up in the non - equilibrium wave - function ?
another interesting aspect of the non - equilibrium dynamics is the issue of thermalization and whether a large system can act as an effective heat bath for its subsystems .
if thermalization occurs , due to the presence of some excess energy , the non - equilibrium state at long times should be in some sense _ close _ to a thermal state at a finite temperature , which implies volume scaling for the entanglement entropy .
the qualitative discussions above motivate a detailed analysis of a quantum quench across a topological phase transition into the topological phase .
we study the specific case of the kitaev toric code model @xcite with spin - polarized initial states . a related quantum quench problem with topologically ordered initial states
was recently studied numerically in ref .
we focus on the time evolution of the entanglement entropy because it is intimately related to topological order through the topological entanglement entropy .
@xcite entanglement entropy is also related to thermalization since , as mentioned above , volume scaling of the entropy as opposed to the area law is generically expected of thermal states .
we compute exactly the bipartite entanglement entropy as a function of time .
this is an example where exact calculation can be done in 2 + 1d ( thus far , analytical results for the time - evolution of entanglement had focused on 1 + 1d systems ) .
we show that the entanglement entropy respects the area law at all times and if one adds disorder to the coupling constants , the recurrences or revivals are suppressed at long times and the entanglement entropy per unit `` area '' ( boundary length ) @xmath0 equals @xmath1 in the thermodynamic limit when the initial polarization is in the @xmath2 or @xmath3 direction .
we find the dependence of this saturation value on the initial state for an arbitrary direction of the initial magnetization . a new ingredient that comes in when considering these initial states is order - one corrections to the entanglement entropy from the convex and concave corners in the subsystem boundary .
we find that the general form of the generated entanglement entropy is given by @xmath4 where @xmath0 is the perimeter of the subsystem and @xmath5 ( @xmath6 ) the number of convex ( concave ) corners in its boundary .
the functions of time @xmath7 , @xmath8 and @xmath9 are independent of the subsystem geometry .
from the form of the entanglement entropy above , we find that the topological entropy vanishes at all times .
we study the effects of breaking the integrability in a perturbative approach .
we identify the mechanism by which an integrability - breaking magnetic field can change the area law to the volume law as expected of thermalized states .
the physical picture is that the area law survives at low orders in an expansion in the integrability - breaking field .
going to higher orders however increases the thickness of a ring - like region around the subsystem boundary that contributes to the entanglement entropy . eventually at high enough orders , the ring will collapse onto itself making volume scaling possible . in the perturbative treatment , we explicitly find that the topological entanglement entropy remains zero at low orders .
we expect that the excess energy and the finite density of defects should prevent the formation of any topological order in a sudden quench even if one breaks the integrability .
the paper is organized as follows . in section [ sec : quenchesxz ] we discuss quantum quenches for spin - polarized initial states in the @xmath2 or @xmath3 direction .
we calculate the entanglement entropy analytically and discuss the effects of disorder in the coupling constants . in section [ sec : corners ] , we study the role of the sharp corners in the subsystem boundary in the generation of entanglement by exactly calculating the second order renyi entropy for an arbitrary direction of the initial magnetization .
we then discuss the effects of breaking the integrability , which we treat perturbatively , in section [ sec : non - integrable ] .
we conclude the paper in section [ sec : conclusion ] .
consider the toric code hamiltonian in a field @xcite @xmath10 the hilbert space of the above hamiltonian is given by spins living on the bonds of a square lattice on a torus and the sums are over all plaquettes , stars and bonds respectively .
the plaquette and star operators are defined as follows @xmath11 where @xmath12s are the pauli matrices . for the usual kitaev model the coupling constants @xmath13 and @xmath14
are independent of the plaquettes and the stars but here we keep the @xmath15 and @xmath16 indeces to allow for disorder in the couplings .
note that for the model in zero magnetic field ( @xmath17 ) , the ground state is independent of the actual values of the coupling constants @xmath13 and @xmath14 as long as they are all positive
. one can study the dynamics of this model by changing the coefficients @xmath13 , @xmath14 and @xmath18 in time with some protocol . in this section ,
we consider a quantum quench where the system is initially in a fully magnetized state and then evolves with the integrable toric code hamiltonian , namely , we are initially in the ground state for @xmath19 and @xmath20 for all stars and plaquettes .
we then turn off the field @xmath18 at @xmath21 ( which does not change the quantum state ) and turn on the toric code part of the hamiltonian as follows : @xmath22 and @xmath23 ( @xmath24 ) .
consider for now , the case where @xmath25 ( @xmath26 ) .
the ground state of the usual model ( without disorder in the couplings ) exhibits a topological phase transition : for @xmath27 the system is a paramagnet and for @xmath28 , it is topologically ordered .
this topological phase is robust against weak disorder in the couplings .
let us call the initial state @xmath29 .
in the case where the initial magnetization is in the @xmath2 direction @xmath30 in the @xmath31 basis . the initial state is a tensor product of single spin states and thus has no entanglement . in the paramagnetic initial state
we have @xmath32 for all spin operators . before addressing the question of the entanglement
entropy evolution , let us consider the simple problem of what happens to the @xmath33 expectation value . as a function of time , we have @xmath34 where @xmath35 is the evolution operator given by @xmath36 since all star and plaquettes commute with one another , we have dropped the time ordering operator .
if we consider a sudden quench , we have @xmath37 and @xmath38 .
let us focus on sudden quenches for a more compact notation .
at the end of the calculation we can replace @xmath39 and @xmath40 with the corresponding integrals over time - dependent couplings for a more general protocol . the expectation value of @xmath41 , a spin between two nearest - neighbor plaquettes @xmath15 and @xmath42 is then given by @xmath43 we have @xmath44 .
therefore we get @xmath45 in the completely clean case ( a theoretical abstraction ) , the average of @xmath46 over time is equal to @xmath47 but any amount of disorder will eventually lead to a state which unlike the initial paramagnet has a vanishing time - averaged expectation value for @xmath33 .
we next consider the main problem of this section , namely the evolution of the entanglement entropy .
the density matrix of the system can be written as @xmath48 where @xmath35 is the time evolution operator and @xmath49 is a generic initial state . by inserting the resolution of identity twice , we can write , @xmath50 and @xmath51 . any star or plaquette operator acts on four spins which may all belong to @xmath52 , all belong to @xmath51 or some belong to @xmath52 and some to @xmath51 .
in this last case , the star or plaquette operator must lie at the boundary of the two subsystems .
we can then write @xmath53 where @xmath54 and @xmath55 only act on spins in @xmath52 and @xmath51 respectively and @xmath56 only depends on star and plaquette operators at the boundary of the two subsystems .
the three operators @xmath54 , @xmath55 and @xmath56 commute with one another . each basis state can be written as the tensor product of states for spins in @xmath52 and @xmath51 , @xmath57 .
we now integrate out the spins in @xmath51 to find the reduced density matrix for subsystem @xmath52 , @xmath58 and obtain , @xmath59 by @xmath60 in the above expression .
this amounts to calculating the trace in a rotated basis .
we then substitute eq .
( [ eq : u ] ) into eq .
( [ eq : rho_a ] ) above and obtain latexmath:[\[\rho_a(t)=\sum_{\alpha_a , \sigma_a , \beta_b } \langle \alpha_a \beta_b | u_a u_\partial|\omega\rangle \langle \omega | u_a^\dag u_\partial^\dag | \sigma_a \beta_b \rangle that only acts on @xmath51 spins has no effect on the time evolution of the reduced density matrix for subsystem @xmath52 .
in order to calculate the entanglement entropy between the two subsystems we need to find @xmath62 . we give a replica index to @xmath63 , @xmath64 and @xmath12 and drop the subsystem index for brevity .
states @xmath65 are in subsystem @xmath52 and @xmath66 in @xmath51 . using @xmath67 we obtain , @xmath68 with @xmath69
once again because traces are basis - independent we can replace @xmath65 by @xmath70 in the above equation and we obtain an expression for @xmath62 that only depends on @xmath56 and not the whole @xmath35 .
let us define the following commuting operators @xmath71 where @xmath72 ( @xmath73 ) is the set of all boundary plaquettes ( stars ) which act on both @xmath52 and @xmath51 spins .
we then obtain @xmath74 with @xmath69 .
we now focus on the simple case where initially all spins are aligned in the @xmath2 direction . in this case
, @xmath75 is an eigenstate of @xmath76 and we can therefore substitute @xmath77 in eq .
( [ eq : tr_rho_a_2 ] ) above .
also @xmath78 and @xmath79 are chosen in the @xmath33 basis so the action of @xmath80 on a basis state is to flip the four spins around the plaquette @xmath15 . for concreteness , we assume that the partition @xmath51 contains all spins which lie inside a simply connected @xmath81 rectangular region .
we then have @xmath82 boundary plaquettes .
let us consider the action of @xmath83 on a state @xmath84 . since @xmath85 , we have @xmath86 notice that @xmath87 only if the state the state @xmath88 is obtained by acting with a certain number of @xmath72 plaquette flips on @xmath89 .
let us consider then ising spin variables living at the center of @xmath72 plaquettes .
a plaquette is flipped for @xmath90 and not flipped for @xmath91 .
notice that , on a torus , flipping a certain set of plaquettes @xmath92 and flipping all other plaquettes that do not belong to @xmath92 have the same effect on the spins , i.e. they describe the same element in the group of plaquette flips .
however , since we have eliminated @xmath54 and @xmath55 from the formulation of the problem by a basis rotation , we can uniquely specify a state with nonvanishing @xmath93 by a set of flipped _ boundary _ plaquettes .
let the state @xmath94 be labeled by @xmath95 for @xmath96 .
if the partitions are not too small , it is not possible for a set of boundary plaquette flips acting on @xmath97 to create a state with the same @xmath51 spins and different @xmath52 spins or vice versa .
therefore @xmath98 and we obtain using eq .
( [ eq : tr_rho_a_2 ] ) @xmath99 using eq .
( [ eq : upm ] ) we find that @xmath100 is a product of terms that are equal to @xmath101 for each unflipped plaquette with @xmath102 and @xmath103 for each flipped plaquette with @xmath104 .
we can then write @xmath105 switching the order of the sum and product after plugging eq .
( [ eq : matrix - element ] ) into eq .
( [ eq : trace - rho - n - x ] ) , we obtain @xmath106 we have used the fact that for @xmath107 , @xmath108 .
summing over @xmath109 for each boundary plaquette simply gives @xmath110.\ ] ] by analytic continuation of @xmath111 , we write the entanglement entropy as @xmath112 and we finally obtain , @xmath113.\ ] ] for the clean kitaev model , the entanglement entropy is an oscillatory function of time whose maximum is proportional to the number of boundary plaquettes or the perimeter of the subsystem .
for this initial condition , the star term in the toric code hamiltonian has no effect on the evolution of the entanglement entropy because the initial state is an eigenstate of that term .
similarly if initially all spins are in the @xmath3 direction , the plaquette term will have no effect .
the completely clean system will never reach a steady state .
if there is some disorder in the coupling constants however , the entanglement entropy per unit boundary length will saturate to a steady state value in the thermodynamic limit .
suppose the number of boundary plaquettes @xmath0 is very large .
as @xmath114 , due to the presence of some disorder in couplings @xmath115 , the phases @xmath116 are equally likely to take any value between @xmath117 and @xmath118 modulo @xmath118 .
therefore , taking the thermodynamic limit first we can write @xmath119 with @xmath120 where @xmath121 is a random variable between @xmath117 and @xmath122 drawn from a uniform distribution . the average entanglement entropy is then given by @xmath123 we have @xmath124 which implies that for any realization of disorder @xmath125 indicating that in the thermodynamic limit , the entanglement per unit area length saturates to a steady - state value @xmath126 .
notice that these results , namely the survival of area law at all times and the oscillatory behavior in the entanglement entropy in the absence of disorder are generic features of integrable quenches where the integrals of motion are operators with local support .
the entanglement entropy per unit boundary length relaxes in the thermodynamic limit to a steady state value in the presence of disorder in the coupling constants .
because the system can reach a steady state , it is natural to ask if the steady state of a subsystem has the properties of a thermalized state described by a gibbs or a generalized gibbs ensemble .
@xcite for such thermalized states however , we expect the entanglement entropy of a small subsystem to scale as its volume in the limit of long times .
the survival of the area law at all times which follows trivially from having local integrals of motion indicates the absence of thermalization even to a generalized gibbs ensemble .
let us now consider a more generic case where the initial density matrix is not diagonal in either @xmath33 or @xmath127 basis and both terms in the kitaev hamiltonian affect the entanglement dynamics .
we find that even though in the thermodynamic limit the area law still applies , the physics of the out of equilibrium entanglement is much richer .
we find that here there are order - one corrections to the entanglement entropy from the concave and convex corners of the subsystem .
subleading corrections to the entanglement entropy arising form the sharp corners in the boundary were found before for _ static _ entanglement entropy in the logarithmic corrections to the area law in a class of conformal quantum critical points .
@xcite here we find that corners can also make subleading contributions to the _ dynamical _ entanglement .
we find that when subsystem @xmath51 consists of spins inside a region with @xmath5 convex and @xmath6 concave corners as seen in fig .
[ fig : corners ] , the entanglement entropy after a quantum quench behaves as @xmath128 where @xmath0 is the perimeter of the subsystem and @xmath7 , @xmath8 and @xmath9 are functions of time that do not depend on the subsystem geometry . for a subsystem with a mostly smooth boundary and only a few sharp corners ,
the leading term is the @xmath129 but for a subsystem with a corrugated boundary , the entanglement entropy can be different in the thermodynamic limit .
notice that the number of concave and convex corners are not independent .
suppose subsystem @xmath51 consists of spins inside of an area with @xmath130 connected regions , each with a certain number of handles such that the total number handles is @xmath131 .
as an example in fig .
[ fig : corners ] , we have @xmath132 and @xmath133 .
we argue that @xmath134 this is because for a rectangular simply connected region , we have @xmath135 and @xmath136 and transforming the shape by adding or subtracting plaquettes on the square lattice does not change @xmath137 .
therefore each simply connected region makes a contribution of @xmath138 to @xmath139 and by a similar argument , each hole contributes @xmath140 . to see how the contribution from sharp corners comes in ,
consider an initial state with all the spins in the positive @xmath141 direction .
let us work in the the @xmath142 basis with @xmath143 , @xmath144 and @xmath145 .
the action of a star or plaquette operator in this basis is then to flip the corresponding four @xmath142 spins modulo a phase factor of @xmath146 from @xmath80 .
we will see later that this sign has no effect on the entanglement entropy .
we assume for simplicity that there is a minimum distance of @xmath147 times the lattice spacing between all corners .
consider a convex corner as in the right hand side of fig .
[ fig : ycorner ] .
the @xmath64 spins live in the dashed blue ( color online ) bonds and the @xmath63 spins on the solid black bonds .
flipping the corner plaquette represented by a red square has the same effect on @xmath64 spins as flipping the two boundary stars represented by red crosses .
these operations of course have a different effect on @xmath63 spins .
because we assumed a minimum distance between the corners , these corner star and plaquette operators uniquely correspond to one corner .
similarly as in the left hand side of fig .
[ fig : ycorner ] , flipping the corner star at a concave corner has the same effect on the @xmath63 spins as flipping the two corresponding plaquettes .
we see from this discussion that the effects of stars and plaquette flips mix at the corners , which leads to a different entanglement generation than along a smooth boundary .
the details of the derivation will be given in appendix [ app_a ] .
let us write for now the result for the entanglement entropy for a subsystem @xmath51 consisting of the bonds inside an @xmath148 rectangular region . a boundary star or plaquette that is not located at a corner is refereed to as a regular star or plaquette .
as seen in fig .
[ fig : ycorner ] , at each convex corner , one plaquette and two stars are special . for our @xmath149 subsystem , we have four convex corners , @xmath150 regular boundary plaquettes and @xmath151 regular boundary stars and the entanglement entropy for a clean system after a sudden quench is given by @xmath152 where the contributions from regular plaquettes ( @xmath153 ) , regular stars ( @xmath154 ) and convex corners ( @xmath155 ) are given by @xmath156 \end{aligned}\ ] ] with @xmath157 the above solution conforms to the generic form eq .
( [ eq : entropy - corners ] ) for @xmath158 , @xmath135 , @xmath136 , @xmath159 and @xmath160 . in appendix
[ app_a ] , we give a derivation of the above result in the more general setting of an arbitrary subsystem geometry .
if instead of a sudden quench we turn on the coupling constants with some protocol @xmath161 and @xmath162 , we just need to replace @xmath39 and @xmath40 by their integrals over time . using the partitions of ref . , shown in fig .
[ fig : partitions ] , the topological entanglement entropy is given by @xmath163 from the form of eq .
( [ eq : entropy - corners ] ) , we can then see that although there are order - one corrections to the area law , the topological entanglement entropy remains exactly zero at all times independently of how the toric code hamiltonian is turned on .
it seems plausible that for _ any _ initial state , as long as the integrability is not broken ( with a field @xmath18 for example ) , evolution with the kitaev hamiltonian does not change the topological entanglement entropy for any time dependence of the coupling constants .
this is because this integrable evolution does not allow propagation of information .
the von neumann entropy is a special case ( limit of @xmath164 ) of the order @xmath111 renyi entropy defined as @xmath165 hereafter we refer to the second order ( @xmath166 ) renyi entropy as the renyi entropy .
the renyi entropy @xmath167 is simpler to calculate than the von neumann entropy and is an equally valid measure of entanglement .
it has been shown that the topological entanglement entropy calculated by any order renyi entropy contains the same information in the abelean topological phases @xcite such as the toric code .
for an integrable quench from a paramagnetic state with magnetization in an arbitrary direction , we exactly calculate the renyi entropy @xmath168 .
we consider the clean case with position - independent coupling constants .
with disorder in the couplings , the entropy saturates to the time - average of the clean case similarly to the simpler initial conditions considered before . in the @xmath169 basis
, an arbitrary paramagnetic state can be written as @xmath170 the calculation is rather involved and is done by a transfer matrix method explained in appendix [ app_b ] .
we find the same general features for an arbitrary initial magnetization as for the initial magnetization in the @xmath141 direction .
let us summarize these features below .
the renyi entropy can be written as a sum of contributions from the perimeter of the subsystem as well as the convex and concave corners .
these contributions are oscillatory functions of time in the absence of disorder in the couplings which at long times , saturate to the values given by the time - average of the clean case if we have some disorder in the couplings .
the saturation values are independent of the final coupling constants and the protocol used for turning them on .
they do however depend on the initial state . for a large subsystem with only a few sharp corners ,
the leading term for the renyi entropy is the first term in eq .
( [ eq : entropy - corners ] ) , namely @xmath171 .
the saturation value of the renyi entropy per unit boundary length is plotted in fig .
[ fig : plot ] as a function of the @xmath172 and @xmath173 angles which determine the initial state through @xmath174 and @xmath175 .
notice that this saturation value is a unique function of @xmath176 and @xmath177 .
to study the effects of breaking the integrability , we focus on one simple case .
namely , we start with a state that has all spins aligned in the @xmath2 direction .
as opposed to the integrable case where the magnetic field is turned off before turning on the integrable toric code hamiltonian , the the toric code hamiltonian is turned on in the presence of a small magnetic field @xmath18 .
we can not perform exact analytical calculations in this case .
we can however discuss the effects of the magnetic field by calculating a formal expansion for the ( renyi ) entanglement and topological entropies in powers of @xmath18 .
the expansion can only be considered as a perturbative treatment if the toric code coupling constants are quickly pushed above @xmath18 and the time scales considered are not too large .
two main insights are derived from this analysis .
first we find indications as to how breaking the integrability can eventually lead to volume scaling in the entanglement entropy .
the physical picture is that in the expansion in powers of @xmath18 , the contribution to the entanglement entropy comes from ring - shaped areas around the subsystem boundary whose thickness grows at higher and higher orders in @xmath18 . eventually ,
at a high enough order , these correlated regions meet and the terms in the expansion no longer respect the area law . even for a small @xmath18 ,
these high order terms become important at long times making volume scaling of the entanglement entropy possible as expected of thermalized states . secondly
, we find that in this expansion in powers of @xmath18 , the topological entanglement entropy remains zero at low orders until we reach an order proportional to the system size , suggesting that a small integrability - breaking term is not enough for generating topological entanglement .
a useful notion that we introduce in this section is the entanglement entropy in the interaction picture .
we are used to the shcrdinger , heisenberg and interaction pictures for calculating the expectation values of observables , which are of course independent of the picture .
if one defines the entanglement entropy as a property of the wave function alone , then since the wave function does not change in the heisenberg picture , calculating the entanglement entropy in that picture leads to the nonsensical result that time evolution does not change the entanglement of a quantum state . an interesting alternative is to try and define the entanglement entropy entirely in terms of the correlation functions of appropriate operators .
we do not pursue this path here .
we argue instead that as explained below , the entanglement entropy calculated only from the wave function in the interaction picture can be a useful concept .
let us first consider a _
non - interacting _
hamiltonian @xmath178 such as a paramagnet .
consider a partitioning of the system into subsystems @xmath52 and @xmath51 .
the evolution operator for @xmath178 can be written as @xmath179 where @xmath180 ( @xmath181 ) only acts on the degrees of freedom in @xmath52 ( @xmath51 ) .
notice that acting with @xmath182 on any wave function does not change its entanglement properties .
now consider a hamiltonian @xmath183 with the corresponding evolution operator @xmath184 .
the wave functions at time @xmath185 in the schrdinger and interaction pictures are respectively given by @xmath186 now since acting with @xmath187 or @xmath188 on any wave function does not change its entanglement , the interaction picture wave function has the same entanglement entropy as the schrdinger picture .
note however the calculation may be much simpler in the interaction picture . in the discussion above , we made use of the interaction picture entanglement entropy because we had a hamiltonian @xmath178 whose evolution operator did not change the entanglement entropy . at the next level of complexity , where using the interaction picture wave function may be useful for calculating the entanglement entropy , the evolution operator of the _ unperturbed _ hamiltonian ( @xmath187 ) may have another known property .
for example it may only create entanglement entropy scaling with the subsystem area or it could be incapable of changing the topological entanglement .
this is the case for the integrable toric code hamiltonian .
now if we are interested in finding whether a perturbation can generate entanglement entropy with volume scaling or whether it can generate topological entanglement entropy , we can perform the calculation for the wave function @xmath189 in the interaction picture .
let us go back to our quench problem and write the wave function in the interaction picture with @xmath190 and @xmath191 starting with the initial state @xmath192 with all the spins in the positive @xmath2 direction , the wave - function at time @xmath185 in the interaction picture is given by @xmath193 where @xmath194 is given by @xmath195 for @xmath196 for the simplicity of notation , we focus on the sudden quench and write @xmath197 as @xmath39 , bearing in mind however that for a more general protocol , we can substitute the integral at the end of the calculation
. we can explicitly calculate @xmath198 and the result is @xmath199 with the operators @xmath200 , @xmath0 and @xmath201 defined as follows @xmath202{p1.eps}}\end{picture}+\begin{picture}(22,0)\put(2,-6){\includegraphics[width=0.035\textwidth]{p2.eps}}\end{picture}+ \begin{picture}(22,0)\put(2,-6){\includegraphics[width=0.035\textwidth]{p3.eps}}\end{picture}+\begin{picture}(22,0)\put(2,-6){\includegraphics[width=0.035\textwidth]{p4.eps}}\end{picture}\;\right ) \nonumber \\ d & = & \sum _ { d}\;\begin{picture}(22,0)\put(2,-6){\includegraphics[width=0.07\textwidth]{d.eps}}\end{picture}\qquad.\end{aligned}\ ] ] the sums are over all bonds , plaquettes and dominos and the graphical notation represents a product of pauli matrices .
for example if the bonds in a plaquette are numbered clockwise from @xmath203 to @xmath204 with the top bond being number @xmath203 we have @xmath205{p1.eps}}\end{picture}=\sigma_y^1\sigma_z^2\sigma_z^3\sigma_z^4 , \qquad \begin{picture}(22,0)\put(2,-6){\includegraphics[width=0.035\textwidth]{p2.eps}}\end{picture}=\sigma_z^1\sigma_y^2\sigma_z^3\sigma_z^4 , \qquad \cdots\ ] ] the closed form in eq .
( [ eq : vhat ] ) is simply derived from the repeated substitution of the following commutation relations @xmath206= 2 i p , \qquad \left[\sum_p b_p,\;p \right]= -4 i ( c+d)\\ & & \left[\sum_p b_p,\;c+d \right]= 4 i p\end{aligned}\ ] ] in the baker - hausdorff expansion of @xmath207 @xmath208\right.\\ & & \left .
+ \frac{(-i \lambda t)^2}{2\:!}\left[\sum_p b_p,\left[\sum_p b_p,\sum_i \sigma^x_i \right]\right]+\cdots\right ) \end{aligned}\ ] ] and resumming the series .
the renyi entropy for the interaction picture wave function @xmath189 is given by @xmath209 where summation over repeated indices is implied and as before , the @xmath63 and @xmath64 spins belong to the @xmath52 and @xmath51 subsystems respectively .
first let us consider the first order term in the @xmath210 ( the expression for the renyi entropy without the log ) .
we use the shorthand notation @xmath211 and similarly for the hermitian conjugate . the first order term vanishes as seen below @xmath212 the second order contribution to @xmath210 is given by @xmath213.\end{aligned}\ ] ] let us assume that the whole system is on a torus and the total number of bonds ( spins ) is @xmath214 .
the number of plaquettes @xmath215 and the number of dominos in the system is @xmath216 .
let us introduce another shorthand notation for integrals involving the functions @xmath217 , @xmath15 and @xmath218 , @xmath219 so we have for example @xmath220 and @xmath221 .
we denote the number of plaquettes ( dominos ) with all bonds in subsystem @xmath52 by @xmath222 ( @xmath223 ) and similarly by @xmath224 ( @xmath225 ) for subsystem @xmath51 .
we then have @xmath226 the factors of @xmath227 come from the fact that each term in @xmath0 is a sum of four operators that flip the spins around a plaquette . noting that because @xmath228 is obviously symmetric under the exchange @xmath229 , we have @xmath230 and similarly for @xmath231 and @xmath232 .
we then obtain @xmath233 which using @xmath234 gives @xmath235 this expression demonstrates the physical picture for building volume entanglement which we alluded to before .
we observe from the above expression that at this order the contribution to the entropy is from a ring - shaped area of thickness two lattice spacings around the subsystem boundary . at higher orders , the thickness of this correlated region , which contributes to the entanglement entropy , grows until the ring closes in on itself , making volume scaling possible .
notice that in contrast to exact area law which is possible in a finite system , since the renyi entropy is symmetric under exchanging the subsystems , volume scaling can only hold exactly for a finite subsystem in an _ infinite _ system and as a good approximation in the limit when a subsystem is much smaller than the whole system .
it is a simple exercise in geometry to count @xmath236 and @xmath237 of the two subsystems for partitions shown in fig .
[ fig : partitions ] and verify directly that no topological entanglement is generated at order @xmath238 .
the details of this calculation will be given in appendix [ app_c ] .
we will also show in that appendix that at the next order with a non - vanishing contribution to the renyi entropy ( @xmath239 ) , the topological term will still vanish due to an exact cancellation . similarly to the volume scaling ,
topological entanglement is non - perturbative and does not show up in an expansion in powers of @xmath18 in the thermodynamic limit . for a finite system
, we expect non - vanishing contributions to the topological entropy to appear at orders that scale with the system size .
let us comment here that on physical grounds , in the case of a sudden quench , we do not expect any topological entanglement to form even at infinitely long times .
a trend toward forming volume entanglement observed at low orders suggests that in the limit of large times , the generic thermalization behavior is expected .
although we can not give a mathematical proof , the fragility of topological order to thermal fluctuations @xcite and the extensive excess energy in the system suggests that the non - equilibrium state , after a sudden quantum quench , will likely bear no signature of the equilibrium topological order of the phase .
we emphasize that the expansion used here can not be treated as a perturbative treatment in the adiabatic limit where the toric code coupling constants are turned on slowly because in that case @xmath240 for short times .
we know that in the case of adiabatic evolution , we approximately remain in equilibrium , i.e. in the ground state manifold , and topological order emerges after a time of order system size .
we studied the quench dynamics of the entanglement entropy in the kitaev toric code model . in the case of the integrable
quench we found the entropy exactly as a function of time .
the leading order contribution was found to scale as the subsystem perimeter with an oscillatory behavior that gets suppressed in the presence of some disorder in the coupling constants leading to the saturation of the entanglement entropy to disorder - independent values respecting the area law .
we obtained results for the generated entanglement entropy as a function of an initial paramagnetic state with arbitrary magnetization .
we studied the effect of the sharp corners in the subsystem boundary and found that in analogy with the static entanglement discussed in the literature , they make subleading contributions to the dynamically generated entanglement entropy as well .
we showed that the topological entanglement entropy remains zero in this quantum quench .
the results of the integrable quench as well as the analysis of the integrability - breaking perturbations strongly suggest that the non - equilibrium state after a quench into the topological phase does not develop any of the topological structure present in the equilibrium phase .
we used the concept of the entanglement entropy in the interaction picture which simplifies perturbative approaches to the calculation of the entanglement entropy .
this allows us to tease apart in this particular problem the effects of the time evolution with an unperturbed hamiltonian that can not generate either entanglement with volume scaling or topological entanglement and focus only the part of the evolution that can possibly lead to volume scaling or topological entanglement . by studying the effects of an integrability - breaking perturbation
, we found a mechanism where volume scaling of the generated entanglement entropy can emerge at high orders in perturbation theory as expected of thermalized states . the notion of thermalization after a quench naturally ties in with topological order through the fragility of topological order to thermal fluctuations .
we are grateful to c. castelnovo , a. hamma , m. haque , a. polkovnikov and m. rigol for helpful discussions .
this work was supported by in part by the doe grant de - fg02 - 06er46316 .
we explicitly calculate in this appendix the contributions to the entanglement entropy from regular boundary stars and plaquettes as well as the convex and concave corners when the initial magnetization is in the @xmath141 direction .
first let us find the number of regular boundary stars ( @xmath241 ) and plaquettes ( @xmath242 ) in terms of the subsystem perimeter and the number of convex and concave corners .
we define the perimeter @xmath0 as the total number of boundary plaquettes or boundary stars ( they are equal ) . we then have @xmath243 once again we use eq .
( [ eq : tr_rho_a_2 ] ) which generally holds for any initial state @xmath49 . the only intermediate states ( in the @xmath142 basis ) which contribute to the @xmath62 are the ones connected to @xmath244 by some star or plaquette flips .
we label these states by variable @xmath245 and @xmath246 living on the boundary plaquettes and stars respectively . a negative @xmath247 ( @xmath248 )
indicates that the corresponding plaquettes ( star ) is flipped with respect to @xmath249 .
@xmath250 where @xmath251 , @xmath252 , @xmath253 and @xmath254 label regular plaquettes , regular stars , convex corner plaquettes and concave corner stars respectively . for @xmath255 , @xmath256 ( @xmath257 )
are variables living on the two stars ( plaquettes ) corresponding to the convex ( concave ) corners .
notice unlike regular star variables the two stars at a convex corner are labeled by the corner plaquette and similarly the two plaquettes at a concave corner are labeled by a corner star since a convex ( concave ) corner uniquely corresponds to a corner plaquette ( star ) .
now suppose two such states have the same @xmath64 spins .
all regular and concave corner plaquette and star flips must be the same for these two states . as seen in fig .
[ fig : ycorner ] however , at a convex corner the @xmath247 or @xmath248 variables for these two states could be different if both the convex corner plaquette and the corresponding two corner stars simultaneously have opposite flips .
then for two states @xmath88 and @xmath258 ( which can be labeled as eq .
( [ eq : labeling ] ) because they contribute @xmath62 ) all their @xmath247 and @xmath248 variables must be the same except for the ones corresponding to convex corners , namely @xmath259 and @xmath260 which follow the relation @xmath261 with auxiliary variable @xmath262 .
similarly we can introduce variables @xmath263 for concave corner and for two states @xmath88 and @xmath264 we must have @xmath265 introducing a replica index @xmath266 on @xmath267 and @xmath268 , we can write @xmath269 let us for the moment focus on the renyi entropy @xmath270 for simplicity
. the calculation for @xmath271 will be very similar .
first we can see from the form of the the states above that the sign factors from @xmath80 acting on states in the @xmath142 basis have no effect .
a @xmath80 plaquette flip can only give a different sign for the matrix elements involving @xmath94 and @xmath272 at a corner .
but then the plaquette flip at that corner would also give a different sign for the matrix elements involving @xmath273 and @xmath274 and the overall sign for the product of the four matrix elements appearing in @xmath275 will be positive .
the calculation is then similar to the simpler cases with the initial magnetization in the @xmath2 and @xmath3 directions except for each convex or concave corner there is an additional auxiliary variable to sum over .
again , each matrix element can be written as a product similar to eq .
( [ eq : matrix - element ] ) . by simply bringing together terms
we can write @xmath275 in the following form : @xmath276 where the summation is over all @xmath247 , @xmath248 , @xmath277 and @xmath278 variables for both replicas .
we can take the sum eq .
[ eq : pre - factorization ] inside the products by switching their order and perform the sum for individual terms .
this gives the factorized form below @xmath279 where we will compute the @xmath280s below .
assuming no disorder and taking the log of both sides , we immediately obtain our main result for the entropy @xmath281 which conforms to the general expression eq .
( [ eq : entropy - corners ] ) . for regular stars and plaquttes , the terms appearing in the above eq .
( [ eq : pre - factorization ] ) are identical to what we obtained when the initial spins were in the @xmath2 or @xmath3 direction .
for example @xmath282 let us now define the following function for a more compact notation @xmath283 we can then write @xmath284 ^ 2,\qquad f_{s_r}=\left[g(\psi_{s_r},\mu_{s_r}t)\right]^2.\ ] ] at the corners , we have a more complicated structure and for example , @xmath285 , in addition to @xmath286 , also depends on @xmath287 and @xmath288 for @xmath255 . notice that for the renyi entropy @xmath289 and therefore @xmath290 .
we can explicitly write @xmath291 it can be easily shown that @xmath292 using the above identity we can write @xmath293 a similar expression with @xmath294 , @xmath295 and @xmath296 gives @xmath297
. we can now perform the summations and we obtain @xmath298 with @xmath166 for the renyi entropy and similarly for @xmath299 with @xmath295 . to calculate @xmath300 we first perform the sum over @xmath247s and @xmath248s and then over @xmath277s and we obtain after some algebra @xmath301 with @xmath166 for the renyi entropy and @xmath302 , @xmath303 , @xmath304 and @xmath305 .
we can similarly find @xmath306 for concave corners by switching @xmath295 .
the generalization to @xmath307 is similar and leads to the appearance of a different @xmath111 in eq .
( [ eq : convex_f ] ) and eq .
( [ eq : regular_f ] ) .
in this appendix , we discuss the transfer matrix method used in calculating the renyi entropy for an arbitrary initial magnetization in the positive @xmath308 direction . the initial state can be written as @xmath309 where @xmath310 and @xmath311 are the eigenstates of @xmath33 .
let us begin by writing an explicit equation for @xmath312 .
using @xmath166 in eq .
( [ eq : tr_rho_a_2 ] ) we can write @xmath313 consider the first of the four matrix elements appeasing in the equation above , namely @xmath314 . we represent the @xmath94 spin configuration in a basis where boundary spins , i.e. spins belonging to a boundary star or plaquettes , are in the @xmath169 basis and all other spins in the @xmath315 basis . in a schematic notation explained below ,
we can then write the matrix element as @xmath316\end{aligned}\ ] ] where in @xmath317 and @xmath318 we can more generally make the substitution @xmath319 .
the notation in the above equation needs some explanation . as before we have introduced ising variables @xmath320 on the @xmath0 boundary plaquettes . for a given configuration of these variables @xmath92 ,
@xmath321 and @xmath322 are the number of flipped and unflipped boundary plaquettes respectively and @xmath323 and @xmath324 are the number of up and down _ boundary _ spins in the state obtained from @xmath325 after flipping the plaquettes represented by @xmath92 .
we have introduced a new notation where @xmath63 denotes only the boundary spins in subsystem @xmath52 that belong to a boundary plaquette and the other boundary spins in a , which only belong to a boundary star , are denoted by @xmath326 .
similarly boundary spins in subsystem @xmath51 that belong to a boundary star are labeled @xmath64 and the rest @xmath327 as seen in fig .
[ fig : boundary_spins ] .
assuming that the total number of boundary spins is @xmath51 , we can relate @xmath328 and @xmath329 to the spins variables above as follows @xmath330 and @xmath331 in the sums above each @xmath63 and @xmath327 spin is multiplied by the ising variable @xmath247 of the boundary plaquette it belongs to .
each @xmath64 spin belongs to two boundary plaquettes and is therefore multiplied by the two corresponding @xmath247 variables . in the notation explained above , we now write the term @xmath332 as follows @xmath333 where @xmath334 like the one - dimensional classical ising model in a magnetic field , we can now integrate out the @xmath247 ising variables using the transfer matrix method . the terms containing @xmath247
are promoted to @xmath335 transfer matrices where each transfer matrix @xmath336 is between two adjacent boundary plaquettes and @xmath337 the other degrees of freedom can be systematically integrated out by repeatedly using the following trace identity @xmath338.\ ] ] repeated use of the identity eq .
( [ eq : identity ] ) leads to enlarged transfer matrices as discussed below .
tracing out all degrees of freedom will lead to @xmath339 transfer matrices the trace of whose product gives the renyi entropy .
notice that at concave and convex corners , we get a different @xmath335 transfer matrix .
the @xmath335 transfer matrices at the corners will be given at the end of this appendix for completeness .
we discuss below the procedure for obtaining the final @xmath339 transfer matrices from the @xmath335 ones along a smooth boundary . at the sharp corners ,
we can obtain the @xmath339 transfer matrices from the corresponding convex and concave corner @xmath335 matrices essentially in the same way .
let us now explicitly write @xmath336 in eq .
( [ eq : two - by - two ] ) along a smooth boundary . for a bond @xmath340 between two adjacent boundary plaquettes as show in the top left corner of fig .
[ fig : transfer_all ] , we have @xmath341 \times } \\ & & \left ( \begin{array}{ll } h^{\frac{1}{2}}a^{\frac{1}{2}({\beta^\prime}^i+\alpha^i)}&0 \\ 0&h^{-\frac{1}{2}}a^{-\frac{1}{2}({\beta^\prime}^i+\alpha^i ) } \end{array } \right ) \left(\begin{array}{ll } a^{\frac{1}{2}\beta^i}&a^{-\frac{1}{2}\beta^i } \\ a^{-\frac{1}{2}\beta^i}&a^{\frac{1}{2}\beta^i } \end{array } \right).\end{aligned}\ ] ] the next step is to consider @xmath342 we can now use the trace identity eq .
[ eq : identity ] to bring together terms with the same @xmath64 and @xmath327 and trace out these degrees of freedom to get @xmath343 where the @xmath344 transfer matrix above is given by @xmath345 t_1\ : t_2 \ : t_3 \ : t_4\ ] ] with the following @xmath346 matrices @xmath347 using the trace identity above to bring together the degrees of freedom via the tensor product of matrices is the main strategy for calculating the renyi entropy .
these calculations look tedious but are greatly simplified using mathematica . the next step is to bring together the same degrees of freedom in @xmath348 using the trace identity which results in @xmath349 matrices that only depend on @xmath63 and @xmath326 degrees of freedom for both replicas .
notice that so far we have integrated out all @xmath64 and @xmath327 degrees of freedom ( for both replicas ) .
now since the @xmath326 degrees of freedom do not interact we can do the sum directly .
the @xmath63 degrees of freedom have an ising type interaction so the transfer matrix method can be applied again promoting functions of @xmath63 ( which appear in @xmath349 matrices ) to @xmath350 matrices ( because there are @xmath351 replicas ) .
this operation can be done very simply in mathematica . at the end of the day
the renyi entropy will be given by the trace of the product of @xmath0 ( the number of boundary plaquettes ) @xmath339 transfer matrices .
if the boundary of the subsystem is largely smooth , i.e. only has a few sharp corners , the different transfer matrices at the corners will give an order - one contribution to the renyi entropy and the leading order area law term is @xmath352 where @xmath353 is the largest eigenvalue of the smooth boundary @xmath339 transfer matrix constructed above .
an exact calculation of the entanglement entropy is needed for evaluating the topological entropy .
therefore corrections from concave and convex corners must be taken into account by using different transfer matrices at the corners in the boundary .
there is a small complication due to the the fact that a convex ( concave ) corner plaquette has an additional @xmath63 ( @xmath64 ) spin .
let us review our method : we first consider two matrix elements each given by the trace of a product of @xmath335 matrices , we then bring together the same degrees of freedom and explicitly sum over @xmath64s to get the @xmath350 transfer matrices . now having two @xmath64s to sum over does not change anything except we have to do a few more summations to get @xmath354 . with our method of first integrating out the @xmath51 spins and then the @xmath52 spins ,
the additional @xmath63 spin at a convex corner gives a jump in the indices of the transfer matrices at the end of the day but this can be easily accounted for by inserting a matrix @xmath355 between the transfer matrices and calculating the trace of their product as usual .
let us now write the @xmath335 matrices for the corners .
first consider the transfer matrix @xmath356 for entering a convex corner as seen in the top right corner of fig .
[ fig : transfer_all ] . each transfer matrix is written for a @xmath64 bond and the @xmath63 spins before ( @xmath357 ) and after @xmath358 that bond . as mentioned above the discontinuity in the @xmath63 index ( @xmath358 for a transfer matrix
is not equal to @xmath357 for the next transfer matrix ) can be taken care of by inserting the matrix @xmath359 ( @xmath350 because we have two replicas ) at the end of the day between the @xmath360 transfer matrices . for exiting a convex corner plaquette , with
the notation of the bottom left corner of fig . [
fig : transfer_all ] , we can write a @xmath361 transfer matrix @xmath362 and similarly a transfer matrix @xmath363 for a concave corner ( bottom right corner of fig .
[ fig : transfer_all ] ) . in terms of the spin variables shown on the corresponding corner of fig .
[ fig : transfer_all ] , these matrices are given by @xmath364\times \left ( \begin{array}{ll } h^{\frac{1}{2}}a^{\frac{1}{2}({\beta^\prime}^i+\alpha^i_{\rm b})}&0 \\ 0&h^{-\frac{1}{2}}a^{-\frac{1}{2}({\beta^\prime}^i+\alpha^i_{\rm
b } ) } \end{array } \right ) \left(\begin{array}{ll } a^{\frac{1}{2}\beta^i}&a^{-\frac{1}{2}\beta^i } \\
a^{-\frac{1}{2}\beta^i}&a^{\frac{1}{2}\beta^i } \end{array } \right ) \left(\begin{array}{ll } a^{\frac{1}{2}\alpha^i_{\rm a}}&0 \\
0&a^{-\frac{1}{2}\alpha^i_{\rm a } } \end{array } \right ) \nonumber\\ t^{r>}_{2\times 2}&= & a^ { \frac{1}{2 } { \alpha^\prime}^i } \exp \left [ i \mu t\ : { \alpha^\prime}^i \beta^i \alpha^i_{\rm b } \alpha^i_{\rm a } \right ] \times \left ( \begin{array}{ll } h^{\frac{1}{2}}a^{\frac{1}{2}\alpha^i_{\rm b}}&0 \\
0&h^{-\frac{1}{2}}a^{-\frac{1}{2}\alpha^i_{\rm b } } \end{array } \right ) \left(\begin{array}{ll } a^{\frac{1}{2}\beta^i}&a^{-\frac{1}{2}\beta^i } \\
a^{-\frac{1}{2}\beta^i}&a^{\frac{1}{2}\beta^i } \end{array } \right)\nonumber\\ t^<_{2\times2}&= & \exp \left [ { i \mu t\ : \alpha^i_{\rm b } \alpha^i_{\rm a } \beta^i \beta^{i+1 } } \right ] \times \left ( \begin{array}{ll } h^{\frac{1}{2}}a^{\frac{1}{2}(\beta_i+{\beta^\prime}^i+\alpha^i_{\rm b})}&0 \\
0&h^{-\frac{1}{2}}a^{-\frac{1}{2}(\beta_i+{\beta^\prime}^i+\alpha^i_{\rm b } ) } \end{array } \right ) \left(\begin{array}{ll } a^{\frac{1}{2}\beta^{i+1}}&a^{-\frac{1}{2}\beta^{i+1 } } \\ a^{\frac{1}{2}\beta^{i+1}}&a^{-\frac{1}{2}\beta^{i+1 } } \end{array } \right).\end{aligned}\ ] ] notice that for a concave corner , the transfer matrix is written between two neighboring boundary plaquettes which in this case do not have a common edge because the plaquette in the middle lies entirely in subsystem @xmath51 and is therefore not a boundary plaquette .
the procedure for going from the initial @xmath361 transfer matrix to the final @xmath360 is analogous to the smooth boundary case discussed above and a detailed discussion would not be illuminating .
we have explicitly verified using the transfer matrix method described in this appendix that the order - one correction to the area law is entirely from the convex and concave corners and the topological entanglement entropy remains identically zero .
we stated in section [ sec : non - integrable ] that eq .
( [ eq : second - order ] ) gives zero topological entanglement at order @xmath238 .
consider the partitions of fig .
[ fig : partitions ] with the spins inside ( but not on the boundary of ) the blue regions belonging to subsystem @xmath51 and all other spins to @xmath52 . for partition
i for example , the outer and inner boundaries of the subsystem @xmath51 are squares of side @xmath366 and @xmath367 respectively .
the whole system is on an @xmath368 torus . in terms of these dimensions
, we can explicitly find the quantities @xmath369 , @xmath370 , @xmath371 and @xmath372 for the four different partitions and we observe that the value of each quantity for partitions ii and iii , which are obviously equal , is the average of those for partitions i and iv .
this immediately implies that the topological entanglement entropy identically vanishes at second order in @xmath18 . for completeness
we include the results of this simple geometry exercise for partitions i and iv below @xmath373 where we have assumed @xmath374 .
next , we discuss the fourth order term .
the combinatorics get more complicated at higher orders but we can calculate the terms in the expansion by direct enumeration .
this gives semi - analytical results with the time - dependence entering analytically through the multiple integrals depending on functions @xmath217 , @xmath15 and @xmath218 and the geometry dependence computed for a system of given dimensions by computer enumeration .
we have performed these calculation for several systems . here
we present the details for an @xmath375 system shown in fig .
[ fig : partition8by8 ] .
if @xmath376 , we have @xmath377 let us assume we have coefficients @xmath378 and @xmath379 for partition @xmath380 .
we can then write the fourth order term in the topological renyi entropy as @xmath381 .
\end{aligned}\ ] ] to find the fourth order term in @xmath382 , in addition to the @xmath7 coefficients that we found before , we need to calculate the coefficients @xmath383 , @xmath384 and @xmath385 for the different partitions .
looking at the structure of the eq .
( [ eq : renyi - explicit ] ) , we observe that at fourth order we need to insert four @xmath386s into the four bins @xmath387 which give a total of @xmath388 terms of five different categories .
there is one term of type @xmath389 ( with one @xmath386 in each @xmath390 ) and @xmath391 @xmath392 , @xmath393 @xmath394 , @xmath391 @xmath395 and @xmath204 @xmath396 terms .
first consider the @xmath389 term .
we define a matrix @xmath397 such that @xmath398 and obtain @xmath399.\ ] ] now since one @xmath386 term acting on @xmath400 can only change the state by flipping either one plaquette or one domino , the @xmath84 states with a non - vanishing @xmath401 can be enumerated and the matrix @xmath397 can be constructed .
the elements of the matrix @xmath397 contain integrals @xmath402 , @xmath403 and @xmath404 depending on which term in @xmath386 contributes to the matrix element .
we then obtain for example the following expression for partition 1 of the @xmath375 system @xmath405}.\end{aligned}\ ] ] we can similarly calculate @xmath406 for other partitions .
the next two categories are @xmath407 and @xmath394 .
the expression for @xmath408 and @xmath409 are given below .
@xmath410 and @xmath411 computing these expressions in terms of the functions @xmath217 , @xmath218 and @xmath7 with direct enumeration yields the following expression for the topological renyi entropy .
@xmath412\\ & & + ( d)^4 - 4(d , d)^2 -16\left[(p , d)+(d , p)\right]^2\\ & & -48(p)^4 + 256(p)^2(p , p)-320(p , p)^2 \big{\rbrace } .
\end{aligned}\ ] ] notice that the last category @xmath413 is independent of the partitions and does not contribute to the topological entanglement entropy at fourth order in @xmath18 .
we can show that the expression in eq .
( [ eq : renyi - topo ] ) above identically vanishes using @xmath414 , @xmath415 and the following integral identity @xmath416 m. greiner _
et al_. , nature ( london ) * 419 * , 51 ( 2002 ) ; a. k. tuchman _
et al_. , phys .
a * 74 * , 051601 ( 2006 ) ; t. kinoshita , t. wenger , and d. s. weiss , nature ( london ) * 440 * , 900 ( 2006 ) ; s. hofferberth _ et al_. , _ ibid_. * 449 * , 324 ( 2007 ) .
a. polkovnikov and v. gritsev , nat
. phys . * 4 * , 477 ( 2008 ) ; r. barankov and a. polkovnikov , phys .
* 101 * , 076801 ( 2008 ) ; c. de grandi , r. a. barankov , and a. polkovnikov , phys . rev . lett .
* 101 * , 230402 ( 2008 ) .
m. levin and x .- g .
wen , phys .
lett . * 96 * , 110405 ( 2006 ) .
s. trebst , p. werner , m , troyer , k shtengel and c. nayak , phys .
* 98 * , 070602 ( 2007 ) .
j. vidal , s. dusuel and k. schmidt , phys .
b * 79 * , 033109 ( 2009 ) ; j. vidal , r. thomale , k. schmidt and s. dusuel , phys .
b * 80 * , 081104 ( 2009 ) . i. s. tupitsyn , a. kitaev , n. v. prokofev , and p. c. e. stamp , 82 , 085114 ( 2010 ) .
m. rigol , v. dunjko , v. yurovsky and m. olshanii , phys .
lett . * 98 * 050405 ( 2007 ) .
|
we study quantum quenches in the two - dimensional kitaev toric code model and compute exactly the time - dependent entanglement entropy of the non - equilibrium wave - function evolving from a paramagnetic initial state with the toric code hamiltonian .
we find that the area law survives at all times . adding disorder to the toric code couplings
makes the entanglement entropy per unit boundary length saturate to disorder - independent values at long times and in the thermodynamic limit .
there are order - one corrections to the area law from the corners in the subsystem boundary but the topological entropy remains zero at all times .
we argue that breaking the integrability with a small magnetic field could change the area law to a volume scaling as expected of thermalized states but is not sufficient for forming topological entanglement due to the presence of an excess energy and a finite density of defects .
|
coherence is a central theme in quantum physics . in feynman s
words on quantum theory @xcite , it is _ the only mystery_. a good definition of coherence depends on two things @xcite : the state of the system , @xmath0 , and the alternatives under consideration , usually attached to different eigenvalues of an observable @xmath1 . whenever @xmath0 is written in a basis ,
the non - diagonal elements of @xmath0 are called _ coherences _ ( with respect to that basis ) since they can originate the oscillating patterns which characterize interference .
curiously , measurements of @xmath1 can not show the effects of such coherences .
the mean value of @xmath1 is simply the wheighted sum of its proper values .
however , observables which do not commute with @xmath1 can reveal them .
naturally , when @xmath0 is diagonalized , there is no relevant coherence with respect to that basis .
a state which is diagonal in whatever basis can show no coherence at all . in finite dimensions
this is the case of the totally mixed state , @xmath2 , @xmath3 denoting the identity operator and @xmath4 the dimension of the state space . in this sense , @xmath5
can also be called _ the _ incoherent state .
quantum systems are not isolated .
they naturally interact with their environment .
when such interaction is taken into account , a _ decoherence _ process usually appears @xcite . in this context
, decoherence can be understood as the vanishing of the non - diagonal terms of the density matrix @xmath0 , which usually occurs as an exponential decay with a timescale @xmath6 much smaller then the usual damping time @xmath7 .
decoherence is even seen as the bridge from quantum mechanics to the classical world , in washing out the interference terms . in the last decades
, a special consequence of `` the only mystery '' has gained central status in quantum mechanics : _
entanglement_. a good definition of entanglement also depends on two things @xcite : the state of the system and a division of such system in meaningfull parties @xcite .
the most paradigmatic context is the bipartite scenario , when the parts are attached to remote experimentalists , alice and bob .
the state space then takes the form of a tensor product @xmath8 .
observables @xmath9 and @xmath10 are _
local_. states of the form @xmath11 show no correlation at all , in the sense that local observables are statistically independent .
states of the form @xmath12 where @xmath13 is a probability distribution , are called _ classically correlated _ or _ separable _ @xcite .
the results of measurements of local observables on separable states can be strongly correlated . a good example , in spin notation
, is the state @xmath14 .
the measurements of the @xmath15-component of spin in both parts are perfectly correlated . but
this correlation is classical .
a bipartite state is entangled if it can not be written as eq . .
recently , interesting effects were recognized in the open system treatment of entangled - state dynamics . in particular ,
if two qubits are subjected to very natural reservoirs , their entanglement can vanish in finite time , even when coherence only vanishes asymptotically @xcite .
this effect was latter called _ sudden death of entanglement _
it is conjectured to be a quite generic result .
a proposal for its experimental detection was already made @xcite and implemented @xcite .
however , it is yet considered as a surprising effect , waiting for a simple geometric explanation @xcite . in this article
we give such a geometric interpretation to the sudden death of entanglement .
this interpretation allows for the explanation of the necessary and sufficient conditions for it to happen .
for simplicity , we concentrate in the two qubit scenario , despite the fact that large dimensions and large number of parties can bring new interesting developements @xcite .
entanglement can be viewed as a consequence of non - local coherence .
in fact , a state like @xmath16 exhibits coherence with respect to the nonlocal obervable @xmath17 .
this example is very illustrative , since its entanglement can vanish in a finite time or only asymptotically , depending on the reservoir it is subjected to . for a sudden death situation , one can use independent spontaneous decay with positive temperature @xcite , while an asymptotic regime is obtained for a dephasing noise @xcite or for independent spontaneous decay at null temperature @xcite . in this work ,
the separation between alice and bob is fixed ( in the sense that we only use the canonical tensor product structure @xcite ) , and coherence will be studied with respect to the usual computational basis . for convenience
, we will adopt a notation similar to ref .
@xcite , by using as a basis @xmath18 the global density matrix takes the form @xmath19},\ ] ] and the local ( reduced ) density matrices are @xmath20},\\ \rho _ b & = & { \left [ \begin{array}{cc } \rho _ { 11 } + \rho _ { 33 } & \rho _ { 12 } + \rho _ { 34}\\ \rho _ { 21 } + \rho _ { 43 } & \rho _ { 22 } + \rho _ { 44 } \end{array } \right]}. \ ] ] the first important geometric aspect appears here .
our problem starts on a manifold of ( real ) dimension @xmath21 given by the positive operators in @xmath22 with unit trace .
if we want system @xmath1 not to show any coherence with respect to the computational basis , we need to have @xmath23 while if we want to see no coherence for the computational basis of system @xmath24 we must obey @xmath25 as each of these equations are complex , they represent a system of two real equations each . accordingly , each set of states incoherent with respect to local @xmath26 is a subset of dimension @xmath27 .
if we want both local coherences to vanish then we get an even smaller subset , of dimension @xmath28 . from now on
, we will adopt the strategy of focus on the so called @xmath29 states ( due to the appearance of their density matrices , see eq . )
@xcite , defined by @xmath30 this set has real dimension @xmath31 . for any natural volume measure ,
all those sets of local incoherent states have zero measure ( like a line in @xmath32 ) . on the opposite direction ,
the set of unentangled states is known to have positive measure ( in particular , the same dimensionality as the ambient ) @xcite . in a sense ( _ e.g. _ by random choices ) it is much more probable to pick an unentangled state than a locally incoherent state . explicitly reparameterizing the set , we consider density matrices of the form @xmath33},\ ] ] which shows no local coherence with respect to the computational basis . despite being `` small '' in the mathematical sense described above
, this set contains some important subsets like the states which can be written as a mixture of bell states ( under the conditions @xmath34 , @xmath35 , and @xmath36 ) , and werner states @xcite ( @xmath34 , @xmath35 , @xmath37 , and @xmath38 ) .
let us discuss a little bit more the notions of non - local coherence and separability inside this set .
the block structure of density matrices of the form ( more evident if the basis is re - ordered ) strongly simplifies the use of peres - horodecki criterion @xcite to test for the existence of entanglement .
it implies that @xmath0 is entangled iff @xmath39 or @xmath40 .
the positivity of @xmath0 implies @xmath41 and @xmath42 , hence only one of the above conditions can be fulfilled . by the same conditions , if one of the populations is null , say @xmath43 , then @xmath37 and for any non - null value of @xmath15 the state is entangled
this is a possible strategy to obtain persistent entanglement @xcite .
it is interesting to note that the elements @xmath15 and @xmath44 are non - local coherences , in the sense that they appear in the mean values of obervables like @xmath45 or @xmath46 .
sudden death of entanglement @xcite can then be characterized as the finite time disentanglement of a state , even when non - local coherences are still present , and usually decay exponentially .
now let us treat the two qubits as an open system .
there is a large collection of natural reservoirs that can be considered ( and an even larger one that can be engineered ) , but let us just divide them into classes .
the first question is if they have a single asymptotic state ( in the sense that all initial states tend asymptotically to the same point ) or not .
a good example of the first case is spontaneous decay , while a dephasing noise is an example of the second .
if the system is subjected to a reservoir with only one asymptotic state , three very diferent situations can happen ( see fig .
[ oneasymp ] ) : i ) the asymptotic state can be an internal point of the separable states set ( _ e.g. _ for random noise or finite temperature thermal reservoirs for independent qubits ) ; ii ) the asymptotic state can be separable but at the boundary of such set ( _ e.g. _ zero temperature spontaneous decay of two independent qubits ) ; or iii ) the asymptotic state can be entangled ( _ e.g. _ low temperature reservoirs for some interacting two - qubit hamiltonians or collective decay of two otherwise independent qubits ) . [
cols="^,^,^ " , ] the generalization of situation iii ) of the previous case , corresponding to the new situation iii ) when @xmath47 , is also geometrically possible , but the non - convexity of @xmath48 makes it harder to find dynamically . if one takes two different entangled states , there is no guarantee that their convex combinations remain entangled .
the most famous example is for a pair of bell states : any equal mixture of two bell states is separable .
another important restriction is that @xmath49 can not contain a whole set of projectors @xmath50 for any two - dimensional subspace of @xmath51 .
, a quadric complex surface , while each two - dimensional subspace of @xmath22 corresponds to a projective line @xmath52 .
this two sets must intersect , by the bezout theorem @xcite . ] with respect to the dynamics of entanglement , this situation behaves in a way analogous to the single entangled asymptotic point : after a finite amount of time , only entangled states can be found . to obtain krauss operators for a reservoir like
this is yet an open problem .
the remaining possibility happens when @xmath49 contains both separable and entangled states .
for this situation , all the behaviours previously discussed are , in principle , possible .
let us just discuss one general and interesting example of this class : two non - interacting identical qubits subjected to collective phase noise .
in this situation , the asymptotic states are of the form with @xmath37 .
this is a five dimensional manifold ( remember @xmath15 is complex and @xmath53 ) containing both separable and entangled states . in this example , initial states like @xmath54}\ ] ] exhibit asymptotic decay of entanglement , while @xmath55}\ ] ] with @xmath56 show sudden death ; on the other hand , initially separable states remain separable for all times , while entangled states like @xmath57}\ ] ] remain fixed as a consequence of the existence of a decoherence free subspace @xcite .
to conclude , we affirm that the curious effect of sudden death of entanglement is an example of a large class of interesting behaviours which open quantum system can show .
it also brings together a rich geometrical flavour .
this geometrical reasoning makes simpler the decision of when entanglement should be destroyed in finite time , when it can even be created by damping , and when qualitatively different time evolutions can coexist in the same system .
a natural question which appears from this work is how to give examples , in theory and in laboratory , for each different possible geometrical situation .
another is about the existence of a maximum number of possible `` entanglement resurrections '' under some natural restrictions on the reservoirs .
generalizations are also in order : large dimensional bipartite systems can exhibit new features with respect to sudden death of entanglement and its geometry .
for example , has bound entanglement any special behaviour with respect to sudden death ?
another natural generalization deals with multipartite entanglement .
it is clear from the geometrical picture that the effect can take place in such context , but as now different kinds of entanglement appear , the classification scheme becomes much trickier .
this work was originated in discussions with dr .
marcelo frana santos and daniel cavalcanti .
the author thanks their suggestions , as well as comments from dr .
karol yczkowski and raphael drumond .
financial support from cnpq , fapemig , and prpq - ufmg are kindly acknowledged .
this work is part of the millenium project for quantum information , cnpq .
feynman , r.b .
leighton , and m. sands , the feynman lectures on physics , vol 3 ( 1965 ) .
bartlett , t. rudolph , and r.w .
spekkens , int . j. quant
. info . * 4 * , 17 ( 2006 ) .
d. giulini _ et al .
_ , decoherence and the appearance of classical world in quantum theory ( 1996 ) .
terra cunha , j.a .
dunningham , and v. vedral , quant - ph/0606149 , to appear in _ proc .
a _ ( doi:10.1098/rspa.2007.1887 ) .
p. zanardi , d.a .
lidar , and s. lloyd , , 060402 ( 2004 ) .
werner , , 4277 ( 1989 ) .
k. yczkowski _
et al . _ ,
a * 65 * , 012101 ( 2001 ) .
l. disi , lect .
notes phys .
* 622 * , 157 ( 2003 ) .
dodd and j.j .
halliwell , phys .
a * 69 * , 052105 ( 2004 ) .
t. yu and j.h .
eberly , , 193306 ( 2002 ) .
t. yu and j.h .
eberly , , 165322 ( 2003 ) .
t. yu and j.h .
eberly , , 140404 ( 2004 ) .
t. yu and j.h .
eberly , , 140403 ( 2006 ) .
t. yu and j.h .
eberly , opt .
commun . * 264 * , 393 ( 2006 ) .
m. frana santos _
et al . _ , , 040305 ( 2006 ) .
et al . _ ,
science * 316 * , 579 ( 2007 ) .
t. yu and j.h .
eberly , quant - ph/0703083
. f. lastra _
et al . _ , , 062324 ( 2007 ) .
t. yu and j.h .
eberly , arxiv : quant - ph/0503089v3 . k. yczkowski _
et al . _ , , 883 ( 1998 ) .
a. peres , , 1413 ( 1996 ) ; m. horodecki , p. horodecki , and r. horodecki , phys .
a * 223 * , 1 ( 1996 ) .
z. ficek and r. tana , , 024304 ( 2006 ) . o. osenda and g.a .
raggio , phys . rev . a * 72
* , 064102 ( 2005 ) .
d.w . berry and m.r .
dowling , , 062301 ( 2006 ) .
j. harris , algebraic geometry - a first course , springer - verlag ( 1992 ) .
lidar , i.l .
chuang , and k.b .
whaley , , 2594 ( 1998 ) .
|
in open quantum systems , entanglement can vanish faster than coherence .
this phenomenon is usually called sudden death of entanglement . in this paper sudden death of entanglement
is discussed from a geometrical point of view , in the context of two qubits .
a classification of possible scenarios is presented , with important known examples classified .
sugestions of theoretical and experimental examples are given as well as large dimensional and multipartite versions of the effect are briefly discussed .
|
the power spectrum of primordial fluctuations is well measured on cosmological scales using the cosmic microwave background ( cmb ) and large scale structure @xcite .
these observations provide compelling evidence that the fluctuations originated in an era of inflation .
however , there is a great deal of new information waiting to be accessed in small - scale fluctuations and in higher - order statistics ( non - gaussianity ) . in this paper
we explore both new regimes by examining how object number counts constrain the probability distribution of primordial fluctuations . by looking at very small objects ( primordial black holes and ultracompact minihalos ) ,
we constrain the power in fluctuations on smaller scales than the cmb and large scale structure currently probe , @xmath0 mpc@xmath1 .
since these objects are also very rare , we constrain the level of non - gaussianity by limiting the abundance of extreme primordial overdensities . counts of rare objects are a useful probe of the primordial inhomogeneities in the gravitational field and their evolution : different objects probe different scales , different cosmological eras , and different particle physics and astrophysics . primordial black holes ( pbhs ) have been used as a probe of small - scale power for many years @xcite .
their abundance limits the allowed fluctuation power on very small scales , but the constraint is weak compared to the numbers expected from extrapolating a scale - invariant spectrum down from cmb scales @xcite .
more recently , a much stronger constraint on small - scale power was obtained by considering ultracompact minihalos ( ucmhs ) of dark matter @xcite .
the only drawback of this approach is that the strongest limits @xcite require dark matter to annihilate into standard model ( sm ) particles , which can then be sought using standard indirect dark matter detection techniques @xcite .
furthermore , ucmh constraints on the small - scale power obtainable by microlensing @xcite are still stronger than pbh constraints , and apply even for non - annihilating dark matter .
the abundance of pbhs was also considered early on as a probe of non - gaussianity @xcite . in a non - gaussian distribution
, the number of rare objects is different from the gaussian expectation , and rarer objects are typically sensitive to higher moments of the distribution .
number counts pick up _ any _ deviation from gaussianity in a model - independent way and provide complementary constraints to those on the individual correlation functions , such as the shape of the 3-point function in momentum space ( the bispectrum ) . however , there are important limitations on what can be learned from number counts on both the analytic and observational sides .
first , in general one only knows the non - gaussian probability density function ( pdf ) approximately , usually in terms of a few of the lowest - order moments .
this limits how far out onto its tail the distribution is known , and so limits the utility of looking at very rare objects to constrain particular models .
furthermore , the best controlled approximations to weakly non - gaussian pdfs are asymptotic expansions , so for a given level of non - gaussianity one trusts the expansion only so far out onto the tail . the greater the level of non - gaussianity , the more limited the range of utility of the asymptotic expansion . on the observational side
, one must know the amplitude of fluctuations ( the variance of the distribution ) very precisely in order to find sharp constraints on the level of non - gaussianity .
the relationship between mass and the observable signature of the objects must be also be known accurately and precisely .
finally , extremely rare objects are not necessarily expected to be present in any finite sample ( e.g. in any single survey or even our universe ) , which makes drawing conclusions from their absence , or from very small number statistics , difficult .
nevertheless , constraints on the abundance of rare objects currently provide the only probe of the primordial fluctuations on very small scales , @xmath2 mpc@xmath1 . on larger scales , @xmath3 mpc@xmath4mpc@xmath1 , @xmath5-type spectral distortions of the cmb
also provide constraints @xcite .
we will return to the complementarity of these probes in the conclusions . here
we treat the analytic difficulties mentioned above carefully and develop a general prescription to use number counts to constrain weakly non - gaussian primordial fluctuations .
we apply this technique to derive bounds on non - gaussianities with ucmhs and pbhs .
previous authors have used pbhs to constrain particular cases of the non - gaussian pdf or particular inflation scenarios .
bullock and primack @xcite , followed by green and liddle @xcite and ivanov @xcite , considered potentials with a localized feature that can both boost power dramatically on some scales and generate strong non - gaussianity because of mode - coupling near the sharp feature .
pina avelino @xcite worked out constraints on a @xmath6 distribution for the fluctuations .
hidalgo @xcite considered a non - gaussian pdf from only the first term in the edgeworth expansion ( @xmath7 from eq .
( [ edgeworthhier ] ) ) .
klimai and bugaev @xcite constrained two - field inflation models .
most recently byrnes et al @xcite considered constraints on weak or strong non - gaussianity of the ` local ' type @xcite , where the primordial curvature is @xmath8 , where @xmath9 is a gaussian random field .
young & byrnes @xcite and kohri et al .
@xcite went on to also consider non - gaussianity in the curvaton scenario .
our results generalize all previous work on weakly non - gaussian models , demonstrating how constraints can be obtained from any object abundance and for any weakly non - gaussian distribution .
we use an extended press - schechter @xcite approach where the abundance of an object is determined by the overdensity required for its formation ( @xmath10 ) , the mass variance smoothed over the region that forms the object ( @xmath11 ) , and similarly smoothed higher - order moments of the density fluctuations .
these variables can be straightforwardly computed for _ any _ object : one need only compute the collapse threshold and smoothed statistics ( by propagating the primordial fluctuations forward to the appropriate time of formation using transfer functions and a growth factor ) .
this method is a straightforward extension of previous work on an analytic , weakly non - gaussian mass function for galaxy clusters @xcite .
however , here we provide more precise criteria for where the analytic expansions are reliable in terms of the rareness of the object ( the threshold @xmath12 ) and the level of non - gaussianity .
in addition , we consider a wider range of non - gaussian scenarios by parameterizing deviations from gaussianity with both the value of the dimensionless skewness ( @xmath13 ) and a rule for how the higher moments scale with respect to the skewness .
most previous work assumes one particular choice , but physical mechanisms generating the primordial fluctuations do not all follow that pattern , and the distinction is important for interpreting constraints . by demonstrating the computational limits of the edgeworth expansion , we also clarify why strongly non - gaussian models must be evaluated on a case - by - case basis .
although we focus on small - scale constraints from pbhs and ucmhs in this work , the techniques we develop are equally applicable to limits on the abundances of high - redshift clusters , which provide strong constraints on non - gaussianity on larger scales @xcite .
section [ sec : expansiontechnique ] develops our formalism for describing weakly non - gaussian distributions and determining the abundance of objects when the primordial fluctuations are described by a non - gaussian distribution .
sections [ sec : pbh ] and [ sec : ucmh ] apply this technique to pbhs and ucmhs respectively , culminating with our final constraint plots in figure [ fig : ucmh+pbh ] ( which a casual reader can skip to directly ) .
we conclude by comparing current and future constraints on small scale power and non - gaussianity on a range of scales from a variety of existing and proposed techniques .
assuming that objects form from regions of the density field above some threshold , we need a probability distribution @xmath14 of perturbation amplitudes @xmath15 and a threshold @xmath16 above which to count the objects . here
@xmath17 gives the minimum density contrast required for a perturbation to later collapse into an object of type @xmath18 , ucmhs@xmath19 ; @xmath20 @xcite , while @xmath21 is a scale - dependent quantity @xcite , although typically @xmath22 @xcite .
assuming a flat universe , the fractional density of matter in objects of type @xmath23 is [ eq : betax ] ( ) = 2a_^p(_r)d_r , where @xmath14 is a function of quantities smoothed on the scale @xmath24 associated with objects of mass @xmath25 .
for a generic pdf , the factor @xmath26 can be determined by ensuring that counting all objects recovers the total mass density . here
we will just use the press - schechter factor of 2 ( @xmath27 ) as a good enough approximation since we are only concerned with a subset of objects .
in addition , we have approximated the integral as extending to infinity , even though for some rare objects ( e.g. ucmhs ) that are not quite as rare as others ( e.g. pbhs ) there will be a finite upper limit associated with @xmath16 of the ` next rarer ' object . extending the integral
to infinity is a very good approximation as long as @xmath16 is dissimilar for different rare objects , as @xmath14 is always a steeply - falling function of @xmath15 .
if the probability distribution is gaussian , we have _ x , gaussian()=erfc , [ eq : betaxgauss ] where @xmath28 and @xmath11 is calculated in the cosmological era relevant for formation of objects of type @xmath23 . in general we will suppress the subscript @xmath23 on @xmath29 , @xmath16 and @xmath30 , except where talking about a specific type of object .
our convention follows @xcite : @xmath24 refers to the comoving radius of the region that forms the object , @xmath31 is the corresponding wavenumber , and both @xmath17 and @xmath11 are computed when the region enters the hubble horizon .
this is an analytic technique , which should be adjusted to agree with numerical simulations before it is trusted to give precise predictions of object abundance .
we can describe weakly non - gaussian distributions as an expansion in moments about a gaussian . a generic expression , generalizing the edgeworth expansion ,
was written down by petrov @xcite [ eq : petrov ] p()d&=&e^-^2/2\{1+_s=1^_r^s_\{k_m}h_s+2r ( ) .
+ & & ._m=1^s()^k_m}. in this expression , the second summation is over all sets of integers @xmath32 that satisfy [ eq : diophantine ] k_1 + 2k_2+
+sk_s = s ; @xmath33 are hermite polynomials : .
] h_n()=(-1)^ne^^2/2e^-^2/2 ; and @xmath34 .
reduced cumulants " are defined as s_n , r .
to decide how to organize the terms in this series and truncate it somewhere , we need additional input about the relative importance of higher - order moments for the physical case under consideration .
the choices are more naturally phrased in terms of the connected part of the dimensionless moments ( the cumulants ) of the real space density contrast , smoothed on scale @xmath24 : [ eq : mnrs ] _
n , r=. the moments @xmath35 can equally well be thought of as functions @xmath36 of the wavenumber @xmath37 corresponding to the smoothing scale @xmath24 .
we will use these two notations interchangeably , and sometimes we do not write out the scale - dependence explicitly at all .
two choices motivated by particle physics @xcite are hierarchical scaling _
n ( ^_)^n-2 , and feeder scaling _
n ^n , n3 , where @xmath38 is the amplitude of fluctuations in the primordial curvature , @xmath39 , and @xmath40 is a parameter indicating the strength of the interaction that sourced the non - gaussianity . in specific models one can work out the coefficients relating @xmath41 , where @xmath42 is the parameter for which cmb experiments quote constraints on non - gaussianity . using simple local models with each scaling to determine representative combinatorics$ ] for the hierarchical scaling ( in that case @xmath43 ) .
we use a two field extension @xmath44 $ ] , with @xmath45 for the feeder scaling ( then @xmath46 ) . ] , we can express the typical size of all higher order moments ( @xmath47 ) in terms of the third moment @xmath48 by @xmath49 these are of course just two examples of how the moments could scale .
for currently developed inflation models , they are representative of the full range of well motivated possibilities ( which we discuss in more detail in the next section ) .
some models ( quasi - single field inflation @xcite ) are a hybrid between the two scalings , or have additional numerical coefficients ( e.g. two - field local models ) .
any skewness in @xmath50 must be positive ( assuming that higher moments are sub - dominant ) for non - gaussianities to increase the production of rare objects like pbhs , as the positive tail of the distribution must be enhanced relative to the core .
once the scaling of the moments has been established , eq .
( [ eq : petrov ] ) can be written in terms of the dimensionless smoothed moments @xmath35 , and the series can be organized in a form appropriate for any scaling . making use of eq .
( [ eq : diophantine ] ) , we find @xmath51 for the two scalings above , we have [ edgeworthhier ] p_1^(h)(,r)&=&h_3 ( ) + p_2^(h)(,r)&=&h_4()+()^2h_6 ( ) + p_3^(h)(,r)&=&h_5()+h_7 ( ) + & & + ( ) ^3h_9 ( ) for the hierarchical scaling ( this is the usual edgeworth expansion ) and [ edgeworthaxion ] p_1^(f)(,r)&=&h_3 ( ) + p_2^(f)(,r)&=&h_4 ( ) + p_3^(f)(,r)&=&h_5 ( ) + p_4^(f)(,r)&=&(+()^2)h_6 ( ) + p_5^(f)(,r)&=&(+)h_7 ( ) for the feeder scaling . with this organization , the series in eq .
( [ eq : petrovm ] ) is an asymptotic expansion for pdfs with moments of either scaling .
organizing the expansion in this way also exposes the key difference between the hierarchical and feeder scalings : @xmath52 , while @xmath53 , which implies that higher - order terms in the expansion make a larger contribution to feeder distributions than they do to hierarchical distributions . then arranging the sums according to each scaling and performing the integral in eq .
( [ eq : betax ] ) , we find for objects of type @xmath23 , [ eq : finalbeta ] ^(h)(_,x)&=&erfc ( ) + & & + 2_s=1^_\{k_m}_h h_s+2r-1(_,x ) + & & _ m=1^s()^k_m ; + ^(f)(_,x)&=&erfc ( ) + & & + 2_s=1^_\{k_m}_fh_s+1(_,x ) + & & _ m=1^s()^k_m , where the sets @xmath54 are again non - negative integer solutions to @xmath55 and @xmath34 , but the sets @xmath56 are solutions to . now for either scaling , truncating the series at some finite @xmath57 in the sums above keeps all terms up to the same order in @xmath13 : @xmath58 for hierarchical scalings and @xmath59 for feeder scalings . we will work out constraints for the single parameter @xmath13 characterizing the two classes of distribution ( hierarchical and feeder ) we have motivated using their particular petrov expansions shown above
first , we illustrate how the petrov - type expansions work in figure [ fig : chisquare ] , which shows a class of examples of the edgeworth expansion compared to the corresponding complete distribution .
the panels show three @xmath6 distributions with varying levels of non - gaussianity and three truncations of the edgeworth expansion for each case . as the figures show , the edgeworth series is accurate out to larger values of @xmath60 when the distribution is closer to gaussian .
but , for a given level of non - gaussianity , there is a value of @xmath60 beyond which the series is not an accurate fit and adding more terms does not improve the fit .
a study comparing a variety of expansions of non - gaussian distributions , with additional examples , can be found in @xcite .
@xmath61{chik2pdfalt.pdf } & \includegraphics[width=0.5\textwidth , angle=0]{chik4pdf.pdf } \\
\includegraphics[width=0.5\textwidth , angle=0]{chik20pdf.pdf } & \includegraphics[width=0.5\textwidth , angle=0]{chik20pdftails.pdf } \\ \end{array}$ ] the parametrization of non - gaussianity in terms of the dimensionless moments and their relative importance is the relevant organization for any observables sensitive to the pdf .
there is a straightforward mapping between the dimensionless skewness and the power spectrum and bispectrum of any particular model ( and choice of cosmological parameters ) .
for example , for local and equilateral bispectral templates , the dimensionless skewness in the dark matter density perturbations at horizon crossing for modes that enter the horizon during radiation domination is @xmath62 we have integrated over the bispectra , using a tophat window function and the transfer function described in the appendix of @xcite .
we have assumed that both the power spectrum and @xmath63 parameters are constant over the range where the integrals peak , @xmath64 , which is not too restrictive since only about a decade in @xmath37 contributes significantly to the integrals .
the dimensionless moments are straightforward to calculate for any inflation model , and the hierarchical and feeder scalings we use are representative of the full range of primordial non - gaussianity that has its origin in the particle physics of inflation . since the moments are determined from a perturbation - theory expansion of some lagrangian , they do indeed scale in a simple way depending on a single parameter that determines the overall level of interaction together with order - one coefficients .
any scenario where the non - gaussian interactions can be modeled by a single source , including inflaton self - interactions and curvaton models , gives rise to the hierarchical scaling .
scenarios where non - gaussian fields couple to the primary source of curvature perturbation to generate the non - gaussianity have the feeder scaling or a hybrid of the two types .
since a pure feeder scaling is the most non - gaussian scaling ( for a given skewness ) known to be generated from inflation , the purely hierarchical and purely feeder scalings nicely bracket the behavior of all known inflation models . finally ,
although we are dealing with much smaller scales than what is seen in the cmb , the behavior of the power spectrum and higher - order correlations on all scales can be computed in any given inflation model .
so , the parameters constrained by probes of very small scales provide complementary constraints on the models . in addition , models where the non - gaussianity changes on small scales typically have correlated changes in the amplitude of the power spectrum .
the two effects are not arbitrary , as both come from the dynamics of the fields sourcing inflation .
a full study of the relationship between the evolution of the power spectrum and amplitude of non - gaussianity has not yet been done for a very wide variety of models , but see figures 1 and 2 in @xcite for an example .
our joint constraints on the power spectrum and non - gaussianity are particularly valuable for these models .
we will return to this point when we present our results . to calculate the abundance of rare objects
, we must truncate the @xmath29 expansions given by eq .
( [ eq : finalbeta ] ) , and we must determine the accuracy of the truncated series .
for a given level of non - gaussianity , there will be some region in the high-@xmath60 tail of the pdf for which the truncated expansion of the distribution is not reliable .
furthermore , for a given value of @xmath60 , there is also a level of non - gaussianity beyond which the expansion is not reliable .
this maximum accessible value of @xmath13 can be increased by including higher - order terms , but only up to a point .
eventually the inclusion of more terms will stop improving the accuracy of the truncated sum , and adding additional terms makes the expansion less reliable ( e.g. , see the blue long - dashed curves in figure [ fig : chisquare ] ) . in this section ,
we determine the optimal number of terms to include in the expressions for the pdf and @xmath29 .
we also describe how we evaluate the accuracy of the resulting expressions and find the maximum value of @xmath13 that can be reliably constrained by a given upper bound on @xmath29 . some previous work on the validity of truncating this series for the hierarchical case only , and applying it
to cluster number counts can be found in @xcite .
a recent study of observational constraints on the feeder and hierarchical scalings from x - ray detected clusters demonstrates that current data analysis is sensitive to the choice of scaling @xcite . to calculate the number of objects of a given type
, we must be able to trust the pdf for values of @xmath65 that contribute to the @xmath29 integral given by eq .
( [ eq : betax ] ) .
fortunately , only a limited range of @xmath60 values make a significant contribution to @xmath29 because the pdf decreases rapidly as @xmath60 increases ( @xmath66 $ ] ) .
therefore , we only need to trust the pdf for @xmath67 , where @xmath68 is chosen so that truncating the @xmath29 integral at @xmath68 has a minimal impact on @xmath29 .
more explicitly , we demand that < 0.02 [ eq : vmaxcond ] so that the pdf for values of @xmath69 contributes less than two percent of the @xmath29 integral . if the observational upper bound on @xmath29 is greater than @xmath70 , then eq .
( [ eq : vmaxcond ] ) is satisfied for all relevant @xmath71 values and all @xmath13 values for which we can trust the pdf for @xmath67 if we set , where @xmath72 for feeder models and @xmath73 for hierarchical models .
since all the bounds on ucmhs and most of the bounds on pbhs are less restrictive than @xmath74 , we adopt these expressions for @xmath68 in our analysis .
using stricter bounds on @xmath29 to constrain non - gaussianity would require a larger value for @xmath68 , which could prevent these improved constraints on @xmath29 from providing improved constraints on @xmath13 using asymptotic expansions of the pdf .
@xmath61{fig2a } & \includegraphics[width=0.5\textwidth , angle=0]{fig2b } \end{array}$ ] next , we evaluate the error in the pdf over the range @xmath67 as a function of @xmath13 and @xmath71 . since the series expansion of the pdf is an asymptotic expansion , the error introduced by truncating the series at @xmath75 terms is the same order as the @xmath76 term in the expansion.th term in the expansion according to the power of @xmath13 in that term . for hierarchical models ,
the @xmath75th term is proportional to @xmath77 and corresponds to the @xmath78 term in eq .
( [ eq : petrovm ] ) . for feeder models ,
the @xmath75th term is proportional to @xmath79 , and receives contributions from all terms with @xmath80 in eq .
( [ eq : petrovm ] ) . ] given @xmath13 and @xmath71 , we compute the ratio of the @xmath76 term in the pdf to the sum of the first @xmath75 terms and find its maximum value in the range @xmath67 : this ratio estimates the maximum fractional error in the truncated pdf over the range of @xmath60 values that contribute significantly to @xmath81 . in our analysis , we restrict ourselves to values of @xmath13 for which this maximum fractional error is less than twenty percent .
we use this criterion to define a computationally accessible " region in the @xmath82 parameter space . given that the pdf is accurate to within @xmath83 for all @xmath60 values that significantly contribute to @xmath81 , we know that the fractional error in @xmath81 is at most @xmath83 for these values of @xmath13 and @xmath71 , and we consider the truncated expansion to be trustworthy in this region .
we expect that the error in @xmath29 is actually lower than 20% because the errors in the pdf are less than 20% for most values of @xmath60 between @xmath71 and @xmath68
. furthermore , these errors may be positive or negative and therefore may partially cancel when the pdf is integrated to obtain @xmath29 .
this expectation was confirmed when we applied our procedure to the @xmath84 probability distributions and compared the values of @xmath29 obtained by integrating the truncated pdf expansion over all @xmath85 to the values obtained by integrating the true pdf ( given in the caption to figure [ fig : chisquare ] ) .
we tested series that were truncated at @xmath86 and @xmath87 and we considered the series trustworthy only if the estimated pdf errors were less than 20% within the range @xmath67 .
we estimated the errors in the pdf using the first term not included in the expansion ; we did not compare the truncated pdf to the true pdf .
for all truncations , the @xmath29 values obtained from the truncated series with @xmath88 and @xmath37 values in that truncation s trustworthy region differed from the true @xmath29 values by less than 6% .
figure [ procedure ] shows the region of the @xmath71-@xmath13 plane that is computationally accessible for @xmath89 and @xmath90 for both hierarchical and feeder models , and @xmath91 for feeder models ; the trustworthy regions lie below the thick curves . for hierarchical models ,
we see that adding terms with @xmath92 does not significantly change the upper bound on computationally accessible @xmath13 values for @xmath93 . for larger values of @xmath71 , adding more terms to the pdf brings larger values of @xmath13 into the trustworthy region . for feeder models , adding more terms decreases the range of accessible @xmath13 values for small values of @xmath71 , but improves the range for larger values of @xmath71 . however , for @xmath94 , adding more terms does not significantly change the maximum trustworthy value of @xmath13 .
@xmath61{pdfsfull } \includegraphics[width=0.5\textwidth , angle=0]{pdfstail } \end{array}$ ] to determine the optimal number of terms to consider when computing the pdf , we must know what @xmath71 values we are interested in .
the shaded regions in figure [ procedure ] show the regions that are excluded by different observational upper bounds on @xmath29 . to maximize the power of a given upper bound on @xmath29 , we want to choose @xmath75 to maximize the portion of the corresponding excluded region that lies within the computationally accessible region ( i.e. beneath the thick curves in figure [ procedure ] ) . for example ,
the darkest two shaded areas show the values excluded by the bound @xmath95 . for hierarchical models
, we should choose @xmath96 when evaluating this constraint .
for feeder models @xmath91 is the best choice because it maximizes the excluded range of @xmath71 values , but we could reach higher @xmath13 by also considering pdfs with fewer terms . in practice , however , changing the number of terms in the expansion for @xmath29 only changes the excluded regions for values of @xmath71 near the excluded region s boundary and for untrustworthy @xmath13 values that are far above all the thick curves in figure [ procedure ] .
therefore , we should choose @xmath75 such that the computationally accessible excluded region reaches the highest possible values of @xmath71 .
we can also trust this excluded region for all @xmath13 that are computationally accessible for any smaller value of @xmath75 .
our default value for @xmath75 is 16 for hierarchical models and 17 for feeder models .
we chose this value of @xmath75 due to computational limitations , but from figure [ procedure ] we see that including more terms offers no benefits for hierarchical models with upper bounds on @xmath29 greater than @xmath97 . for feeder models , including more terms could marginally extend the computationally accessible region excluded by @xmath95 , but it would not have any impact for lower or higher bounds on @xmath29 .
figure [ fig : pdftails ] compares several gaussian and non - gaussian pdfs ( left panel ) and their positive tails ( right panel ) .
the distributions have parameters found using the procedure above and the bound on the abundance of ucmhs at @xmath98 , for which @xmath99 and @xmath100 . the thick solid black curve is the gaussian distribution that saturates the bound on @xmath29 , with @xmath101 .
the blue solid curve is the non - gaussian distribution with hierarchical scaling and the maximum trustworthy value of @xmath13 that gives @xmath102 ( @xmath103 for @xmath104 ) .
this case also defines a minimum variance @xmath105 for a hierarchical distribution that saturates the bound on @xmath29 ; the dotted blue curve is the gaussian distribution with that same value of @xmath106 . in the right hand panel
, this gaussian distribution has very little area under the positive tail and lies nearly on top of the @xmath107-axis .
the long - dashed red curve is the non - gaussian distribution with feeder scaling and the maximum trustworthy value of @xmath13 that gives @xmath102 ( @xmath108 for @xmath109 ) .
this curve has @xmath110 , and the dotted red curve in the right panel shows the tail of a gaussian distribution with the same value of @xmath106 . in the right panel , this curve would be nearly indistinguishable from the red dashed curve .
the dot - dashed curve in the right panel is one way of visualizing the difference between the feeder and hierarchical scalings : it shows a hierarchical non - gaussian pdf with values of @xmath106 and @xmath13 that saturate @xmath29 when used in a _ feeder _ pdf ( @xmath108 and @xmath110 , as shown by the red dashed curve ) .
clearly , for a particular value of @xmath13 , the feeder scaling gives a pdf that is overall substantially more non - gaussian than the hierarchical scaling .
@xmath61{deltanu } \includegraphics[width=0.5\textwidth , angle=0]{deltabetavsnu } \end{array}$ ]
now we can get some measure of the maximum shift in the pdf that can come from adding weak non - gaussianity in a controlled way .
figure [ fig : deltanu ] illustrates this in two ways .
the left hand panel illustrates how a constraint on the variance @xmath106 derived assuming a gaussian distribution can be shifted to a constraint on a non - gaussian distribution with a new ( smaller ) variance .
the vertical axis shows the relative shift in @xmath88 between the gaussian distribution that gives abundance fraction @xmath29 and the maximally non - gaussian distribution , under control according to our criteria above , that gives the same @xmath29 . to use this for any particular object
, @xmath88 should be written in terms of the appropriate @xmath111 and @xmath11 for the object of interest .
@xmath112 can be likewise be converted to a non - gaussian parameter in the primordial distribution .
this plot gives a sense of how accurately the variance of fluctuations must be known in order for number counts to provide an independent constraint on non - gaussianity using petrov expansions .
for instance , if @xmath113 in the left panel of figure [ fig : deltanu ] , then adding weak non - gaussianity in a controlled way has the same effect on the abundance of rare objects as increasing @xmath11 by @xmath114 .
if the uncertainty in @xmath11 exceeds @xmath115 , then number counts with the corresponding value @xmath29 can not independently constrain non - gaussianity without knowing the full pdf of the density fluctuations .
the right hand panel of figure [ fig : deltanu ] shows the maximum fractional change in the abundance @xmath29 that results from adding controlled non - gaussianity for a given object and fixed variance ( so fixed @xmath88 ) : @xmath116/\beta(0,\nu_\mathrm{c})$ ] .
notice that the primary limit on how much @xmath29 can change is our ability to accurately determine the pdf .
since the feeder scaling produces a distribution that is overall much more non - gaussian than the hierarchical for a fixed value of @xmath13 , we can not accurately calculate at as large of @xmath88 ( rarer objects ) or for as large of shifts in @xmath29 as for the hierarchical case .
again , the effects of more non - gaussian distributions can be computed if one knows the full distribution rather than just a finite number of moments .
for a gaussian spectrum of fluctuations , the present - day fractional density of black holes has typically been used to constrain the spectral index @xmath117 under the assumption of a scale - free spectrum @xcite .
the resulting bound , @xmath118 ( based on e.g. veritas searches at @xmath119mpc@xmath1 @xcite ) , is easily satisfied by the current experimentally measured value @xmath120 @xcite , with little or no running . in terms of generalized , scale - dependent gaussian power spectra , pbhs
have been used to place bounds on the amplitude of curvature perturbations at the level of @xmath121 @xcite , on scales corresponding to @xmath122mpc@xmath1 .
stronger constraints than this can be found if the analysis is done using a top - hat window function rather than a gaussian , which is arguably a more correct method @xcite .
present observational bounds on pbhs @xcite limit @xmath123 to values ranging from @xmath124 for solar mass pbhs , to @xmath125 for a small range of pbh masses around @xmath126 g , to @xmath127 for even smaller masses .
a pbh mass @xmath128 corresponds to horizon mass @xmath129 @xcite , and therefore wavenumber @xmath130 here @xmath131mpc@xmath132mpc@xmath1 @xcite and @xmath133 @xcite are the wavenumber and horizon mass respectively at equality , and @xmath134 , @xmath135 , @xmath136 @xcite are the respective degrees of freedom at equality and the time of horizon entry .
we use the current limits on @xmath123 from @xcite , along with eq .
( [ eq : finalbeta ] ) and the requirement that it represents a well - behaved expansion , to determine the range of values of @xmath137 that are excluded by non - observation of pbhs . for this calculation
we draw upon the expressions for @xmath11 in a generalized gaussian power spectrum , given in appendix b of @xcite ; in particular , we employ the top - hat window function advocated in that paper .
we give our limits in figure [ fig : ucmh+pbh ] as a function of the power spectrum of curvature perturbations at any given @xmath37 : @xmath138 we show the excluded areas as dark shaded regions , for both hierarchical and feeder scaling , and at three different example scales .
the corresponding limits on @xmath29 at these scales are @xmath139 ( @xmath140mpc@xmath1 ) , @xmath141 ( @xmath142mpc@xmath1 ) and @xmath143 ( @xmath144mpc@xmath1 ) .
more positive non - gaussianity ( larger @xmath48 ) results in stronger exclusions , up until the point at which our expansion eq .
( [ eq : finalbeta ] ) begins to break down . above these very large values of @xmath48 ,
the excluded regions shrink with additional non - gaussianity , as more of the available parameter space is made inaccessible by the increasingly poor behavior of the expansion .
ucmhs were initially discussed in early papers by berezinsky and collaborators @xcite , and have been recently studied in the context of lensing signatures @xcite , gamma - ray and neutrino signals from dark matter annihilation or decay @xcite , and impacts on the cmb @xcite . to date the strongest limits come from gamma - ray observations @xcite , indicating that @xmath145 for a scale - free spectrum , and @xmath146@xmath147 for generalized spectra , on scales corresponding to @xmath148mpc@xmath1
. the corresponding direct limits on @xmath29 at the two example scales of figure [ fig : ucmh+pbh ] accessible with ucmhs are @xmath149 ( @xmath140mpc@xmath1 ) and @xmath150 ( @xmath142mpc@xmath1 ) . to explore the impact of ucmhs on the allowed parameter space for non - gaussianity , we have implemented eq .
( [ eq : finalbeta ] ) in the analysis framework of @xcite , using combined 1-year _
fermi_-lat limits on the abundance of ucmhs from galactic and extragalactic source searches for details of our slight improvement of the treatment of sources over @xcite . ] , as well as the overall diffuse gamma - ray flux in the direction of the galactic poles .
whilst 2-year source searches ( e.g. @xcite ) and corresponding diffuse data are available , we conservatively adopt 1-year sensitivities to be certain that no dark matter sources were found in the observing period we consider .
the reader is referred to @xcite for extensive discussion of this issue .
our gamma - ray limits assume the canonical thermal cross - section for dark matter annihilation ( @xmath151@xmath152s@xmath1 ) into @xmath153 quarks and anti - quarks , and a relatively conservative dark matter mass of 1tev ( smaller masses would result in stronger limits ) . in figure [
fig : ucmh+pbh ] , we show the ability of ucmhs and pbhs to exclude different values of @xmath48 for the two scaling scenarios , as a function of the departure of the underlying gaussian power from the current gaussian limit .
dark shaded regions are excluded at 95% confidence level ( cl ) . by expressing the underlying gaussian power _
relative to present limits on gaussian power _ from each class of compact object , it is possible to compare the limits on @xmath154 from ucmhs and pbhs directly . although ucmhs provide a much stronger limit than pbhs on the gaussian power ( as known from @xcite ) , pbhs provide a stronger lever arm for constraining non - gaussianity below their gaussian limit than ucmhs . on the other hand ,
our expansion is well - behaved to larger @xmath48 for ucmhs , allowing them to exclude stronger non - gaussianities than pbhs .
the reference power @xmath155 that we have used for normalising the curves shown in figure [ fig : ucmh+pbh ] is the observational limit on the gaussian power at each scale .
for pbhs at @xmath156mpc@xmath1 , this corresponds to @xmath157 . for ucmhs ,
the limit on the gaussian power is much stronger .
precisely how much stronger depends on the latest redshift at which ucmhs can be assumed to undergo gravitational collapse . assuming that only objects collapsing before a redshift of @xmath158 form ucmhs , the reference power is @xmath159 at @xmath160mpc@xmath1 , and @xmath161 at @xmath162mpc@xmath1 .
setting @xmath158 follows earlier work @xcite , and is a very conservative choice .
the defining characteristic of a ucmh is that it forms in isolation from a spherical collapse ; the processes of radial collapse and secondary infall give ucmhs extremely steep dark matter density profiles , thereby distinguishing them from halos formed at late times .
exactly when the approximation of radial infall breaks down is not yet known @xcite , as this depends sensitively on the velocities of dark matter particles and interactions between growing overdensities .
more detailed simulation work is sorely needed to answer this question . in principle ucmhs
may form at lower redshifts , as long as the presence of different - sized overdensities does not spoil the assumption of spherical collapse .
the smaller the redshift , the more unlikely it is that this requirement will be fulfilled .
later collapse redshifts allow smaller amplitude perturbations to form ucmhs , reducing @xmath21 and increasing @xmath163 for a given combination of @xmath48 and @xmath164 .
if ucmhs continue to be formed as late as @xmath165 , their non - observation begins to constrain non - gaussianities on small scales , even if the power spectrum is near the level observed on cmb scales : @xmath166 at @xmath160mpc@xmath1 , and @xmath167 at @xmath162mpc@xmath1 .
we stress again however , that there exists _ no observational evidence whatsoever _ that the power on small scales is actually this low ( see figure [ fig : planckps ] ) ; even with a very conservative collapse redshift of @xmath158 ucmhs therefore place significant new constraints on non - gaussianity at small scales .
we have provided a general technique for using the abundance of any object to constrain weakly non - gaussian statistics of the primordial fluctuations .
we suggest precise criteria for the amplitude of non - gaussianity that can be accurately constrained using petrov - type expansions and emphasize that there is not a one - to - one mapping between the overall level of non - gaussianity and the size of the skewness ( @xmath13 ) .
this is important when using the abundance of rare objects to constrain non - gaussianity : number counts are sensitive to _ any _ deviation from gaussianity , but a measured value of @xmath29 does not constrain a parameter like @xmath42 ( typically interpreted as the amount of skewness ) without additional information about the higher moments of the pdf . in general pbhs
provide stringent constraints on non - gaussianities only in the presence of fluctuations with relatively large variance , from @xmath168 at intermediate ( @xmath160mpc@xmath169 ) and small ( @xmath170mpc@xmath169 ) scales , to @xmath171 at very small scales .
notably though , at very small scales pbhs provide the only available limit on the allowed amount of non - gaussianity .
ucmhs provide stronger constraints , and so may be more likely to eventually also constrain non - gaussianity , but do not reach as small scales as pbhs . for both ucmhs and pbhs , it is straightforward to map our constraints on @xmath13 to constraints on a non - gaussian parameter in the primordial fluctuations .
for example , for the local ansatz @xmath172 $ ] the smoothed , integrated , dimensionless skewness @xmath173 @xmath174 is very nearly equal to @xmath175 .
for this case , if the variance is large enough , the abundance of ucmhs constrains @xmath176 and pbhs constrain @xmath177 on smaller scales .
however , it is important to remember that this does not mean ucmhs are less powerful quite the opposite . for the constraint from either object
, the maximum value of @xmath42 corresponds to @xmath178 , so those values of @xmath42 correspond to equally non - gaussian distributions .
the difference in constraints phrased as a value of @xmath42 comes from the difference in the variance of the fluctuations .
for reference , recall that on cmb scales , where the amplitude of fluctuations is currently measured , @xmath179 corresponds to @xmath180 . to constrain non - gaussianity ,
both pbhs and ucmhs require significantly more power than extrapolating the cmb spectrum from the standard six parameter fit would suggest , but pbhs require even more power than ucmhs .
so , given a prediction for the power on small scales , ucmhs are more likely to constrain non - gaussianity .
in addition , current observational constraints on cosmological models that allow more than six parameters show that additional power on small scales is hardly ruled out .
figure [ fig : planckps ] shows the power spectra allowed by fits to the _ planck _ satellite data @xcite .
cdm " fit models the power spectrum amplitude ( @xmath181 ) with a constant tilt ( @xmath182 ) . if additional scale dependence of @xmath183 and/or a non - zero amplitude of tensor modes ( @xmath184 ) are allowed , the best fit power spectrum can be different . from top to bottom at large @xmath37 , the allowed parameters and data sets shown ( where wp = _ wmap _ polarization data ) are : ( 1 ) @xmath185cdm + @xmath186 + @xmath187 , _ planck_+wp ( solid blue ) ; ( 2 ) @xmath185cdm + @xmath184 + @xmath186 , _ planck_+wp ( dashed blue ) ; ( 3 ) @xmath185cdm + @xmath184 , _ planck_+wp ( solid red ) ; ( 4 ) @xmath185cdm , _ planck_+wp ( dashed black ) ; ( 5 ) @xmath185cdm + @xmath186 , _ planck_+wp ( solid green ) ; ( 6 ) @xmath185cdm + @xmath184 + @xmath186 , _ planck_+wp+bao ( dashed red ) .
_ planck _ does not report the new best fit value of @xmath181 for the models with more than 6 parameters , but the 2@xmath106 uncertainty in the amplitude for the minimal model is within the thickness of the lines .
the @xmath185cdm and @xmath185cdm + @xmath184 models ( lines 3 and 4 ) are reported at a pivot of @xmath188mpc@xmath1 , while the others have a pivot of @xmath189mpc@xmath1 .
[ fig : planckps],height=245 ] the constraints we quote here assume that the dark matter particle is a standard thermal relic with detectable annihilation products @xcite .
unfortunately , near - future astrometric microlensing searches for ucmhs can only constrain @xmath190 @xcite . for these large values of @xmath29 ,
ucmhs are not sufficiently rare to be a useful probe of non - gaussianity because @xmath191 , which is smaller than the crossing point between gaussian and skewed pdfs . consequently
, changing @xmath13 does not significantly affect the abundance of relatively common objects @xmath192 . aside from the abundances of rare objects ,
the only other probe of the small - scale inhomogeneities is the spectral distortion of the cmb , specifically the @xmath5-distortion , which is sensitive to the fluctuations on scales on the order of @xmath193 mpc@xmath1@xcite . as with pbhs and ucmhs ,
cmb @xmath5-distortion can also constrain non - gaussianity , but the nature of the constraints on both characteristics is qualitatively different . first , cmb @xmath5-distortion is sensitive to the total power in fluctuations over a wider range of scales , while the objects are sensitive to the power only very near the scale of the object , @xmath194 .
second , @xmath5-distortion is sensitive to only the squeezed limit of the bispectrum rather than the total skewness .
that is , this measurement can constrain the amplitude in the correlation of three momentum modes with @xmath195 @xcite . if non - gaussianity follows the local ansatz ( which has a strong signal in the squeezed bispectrum ) , a futuristic cmb probe like pixie @xcite could constrain @xmath196 for a scale - invariant power spectrum and @xmath179 for an enhanced small - scale power spectrum that saturates current bounds ( @xmath197 in both cases ) @xcite . as can be seen from the top two panels in figure [ fig : ucmh+pbh ] ,
the @xmath5-distortion bound would be about an order of magnitude better than the bounds from object counts at that scale if the non - gaussianity is local and hierarchical .
the bounds may be similar for non - gaussianity with a stronger , feeder type scaling and a local bispectrum .
the @xmath5-distortion probe is then quite complementary to the abundance of pbhs and ucmhs in terms of the range of scales probed and the relationship between the observables and the primordial fluctuations .
j. yana galarza and s. shandera were supported in the early stages of this work by the perimeter institute for theoretical physics .
research at perimeter institute is supported by the government of canada through industry canada and by the province of ontario through the ministry of research and innovation .
s. shandera has also been supported by the eberly research funds of the pennsylvania state university .
the institute for gravitation and the cosmos is supported by the eberly college of science and the office of the senior vice president for research at the pennsylvania state university .
a. erickcek supported in part by the canadian institute for advanced research .
p. scott is supported by the banting program , administered by the natural science and engineering research council of canada .
our analysis here actually includes a slightly improved form of the point source analysis of @xcite . following from eq . ( 18 ) in @xcite , if there is a single ucmh present somewhere in the milky way , we identify the probability @xmath198 of actually seeing that ucmh as @xmath199 here @xmath200 is the true distance of the ucmh from earth , @xmath201 is the maximum observable distance required for detection at cl @xmath107 , @xmath202 is the probability of the ucmh residing within distance @xmath201 of earth , @xmath203 is the dark mass of the milky way and @xmath204 is the dark mass within @xmath201 .
this leads to an improved version of eq .
( 24 ) in @xcite , which gives the maximum allowed fraction of matter in ucmhs today @xmath205 here @xmath206 is the fraction of matter in the universe that is dark , @xmath207 is the present - day ucmh mass , and @xmath208 is the desired cl of the limit @xmath209 .
the relationship between @xmath29 and @xmath209 can be found in eq .
( 21 ) of @xcite .
( [ fmax_better ] ) is valid for all @xmath210 . for @xmath211
, it agrees with eq .
( 24 ) in @xcite ; whilst both forms are valid for the values of @xmath107 and @xmath208 we use here ( @xmath212 , @xmath213 , as in @xcite ) , the latter breaks down as @xmath214 , so eq . ( [ fmax_better ] ) is more general .
e. komatsu _ et .
_ , _ seven - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation _ , _ astrophys.j.suppl .
_ * 192 * ( 2011 ) 18 , [ http://xxx.lanl.gov/abs/1001.4538[arxiv:1001.4538 ] ] .
j. dunkley , r. hlozek , j. sievers , v. acquaviva , p. ade , _ et .
_ , _ the atacama cosmology telescope : cosmological parameters from the 2008 power spectra _ , _
_ * 739 * ( 2011 ) 52 , [ http://xxx.lanl.gov/abs/1009.0866[arxiv:1009.0866 ] ] .
r. keisler , c. reichardt , k. aird , b. benson , l. bleem , _ et .
_ , _ a measurement of the damping tail of the cosmic microwave background power spectrum with the south pole telescope _ ,
_ astrophys.j .
_ * 743 * ( 2011 ) 28 , [ http://xxx.lanl.gov/abs/1105.3182[arxiv:1105.3182 ] ] .
b. a. reid , w. j. percival , d. j. eisenstein , l. verde , d. n. spergel , _ et .
_ , _ cosmological constraints from the clustering of the sloan digital sky survey dr7 luminous red galaxies _ , _ mon.not.roy.astron.soc .
_ * 404 * ( 2010 ) 6085 , [ http://xxx.lanl.gov/abs/0907.1659[arxiv:0907.1659 ] ] .
n. sehgal , h. trac , v. acquaviva , p. a. ade , p. aguirre , _ et .
_ , _ the atacama cosmology telescope : cosmology from galaxy clusters detected via the sunyaev - zeldovich effect _ , _ astrophys.j .
_ * 732 * ( 2011 ) 44 , [ http://xxx.lanl.gov/abs/1010.1025[arxiv:1010.1025 ] ] .
, p. mcdonald _ et .
_ , _ the lyman - alpha forest power spectrum from the sloan digital sky survey _ ,
_ astrophys.j.suppl .
_ * 163 * ( 2006 ) 80109 , [ http://xxx.lanl.gov/abs/astro-ph/0405013[astro-ph/0405013 ] ] .
a. s. josan , a. m. green , and k. a. malik , _ generalised constraints on the curvature perturbation from primordial black holes _ , _ phys.rev .
_ * d79 * ( 2009 ) 103520 , [ http://xxx.lanl.gov/abs/0903.3184 [ arxiv:0903.3184 ] ] .
a. s. josan and a. m. green , _ gamma - rays from ultracompact minihalos : potential constraints on the primordial curvature perturbation _ , _ phys.rev . _ * d82 * ( 2010 ) 083527 , [ http://xxx.lanl.gov/abs/1006.4970[arxiv:1006.4970 ] ] .
t. bringmann , p. scott , and y. akrami , _ improved constraints on the primordial power spectrum at small scales from ultracompact minihalos _ ,
_ phys.rev .
_ * d85 * ( 2012 ) 125027 , [ http://xxx.lanl.gov/abs/1110.2484[arxiv:1110.2484 ] ] .
f. li , a. l. erickcek , and n. m. law , _ a new probe of the small - scale primordial power spectrum : astrometric microlensing by ultracompact minihalos _ , _ phys.rev .
_ * d86 * ( 2012 ) 043519 , [ http://xxx.lanl.gov/abs/1202.1284[arxiv:1202.1284 ] ] .
j. s. bullock and j. r. primack , _ non - gaussian fluctuations and primordial black holes from inflation _ , _ phys .
* d55 * ( 1997 ) 74237439 , [ http://xxx.lanl.gov/abs/astro-ph/9611106[astro-ph/9611106 ] ] .
w. hu , d. scott , and j. silk , _ power spectrum constraints from spectral distortions in the cosmic microwave background _ , _
_ * 430 * ( 1994 ) l5l8 , [ http://xxx.lanl.gov/abs/astro-ph/9402045 [ astro - ph/9402045 ] ] .
j. chluba , a. l. erickcek , and i. ben - dayan , _ probing the inflaton : small - scale power spectrum constraints from measurements of the cmb energy spectrum _ , _ astrophys.j . _
* 758 * ( 2012 ) 76 , [ http://xxx.lanl.gov/abs/1203.2681[arxiv:1203.2681 ] ] .
a. m. green and a. r. liddle , _ constraints on the density perturbation spectrum from primordial black holes _ , _ phys .
_ * d56 * ( 1997 ) 61666174 , [ http://xxx.lanl.gov/abs/astro-ph/9704251 [ astro - ph/9704251 ] ] .
c. t. byrnes , e. j. copeland , and a. m. green , _ primordial black holes as a tool for constraining non - gaussianity _ , _ phys.rev . _
* d86 * ( 2012 ) 043512 , [ http://xxx.lanl.gov/abs/1206.4188[arxiv:1206.4188 ] ] .
m. loverde , a. miller , s. shandera , and l. verde , _ effects of scale - dependent non - gaussianity on cosmological structures _ , _ jcap _ * 0804 * ( 2008 ) 014 , [ http://xxx.lanl.gov/abs/0711.4126 [ arxiv:0711.4126 ] ] .
v. v. petrov , _ limit theorems of probability theory : sequences of independent random variables _ , vol . 4 of _ oxford studies in probability_. clarendon press , oxford / oxford university press , new york , 1995 .
s. chongchitnan and j. silk , _ a study of high - order non - gaussianity with applications to massive clusters and large voids _ , _
_ * 724 * ( 2010 ) 285295 , [ http://xxx.lanl.gov/abs/1007.1230 [ arxiv:1007.1230 ] ] .
a. m. green , a. r. liddle , k. a. malik , and m. sasaki , _ a new calculation of the mass fraction of primordial black holes _ , _ phys
* d70 * ( 2004 ) 041502 , [ http://xxx.lanl.gov/abs/astro-ph/0403181 [ astro - ph/0403181 ] ] .
s. chongchitnan and g. efstathiou , _ accuracy of slow - roll formulae for inflationary perturbations : implications for primordial black hole formation _ , _ jcap _ * 0701 * ( 2007 ) 011 , [ http://xxx.lanl.gov/abs/astro-ph/0611818[astro-ph/0611818 ] ] .
t. bringmann , c. kiefer , and d. polarski , _ primordial black holes from inflationary models with and without broken scale invariance _ , _
phys.rev . _
* d65 * ( 2002 ) 024008 , [ http://xxx.lanl.gov/abs/astro-ph/0109404[astro-ph/0109404 ] ] .
v. berezinsky , v. dokuchaev , and y. eroshenko , _ small - scale clumps in the galactic halo and dark matter annihilation _ , _
phys.rev . _
* d68 * ( 2003 ) 103003 , [ http://xxx.lanl.gov/abs/astro-ph/0301551 [ astro - ph/0301551 ] ] .
v. berezinsky , v. dokuchaev , and y. eroshenko , _ destruction of small - scale dark matter clumps in the hierarchical structures and galaxies _ , _ phys.rev .
_ * d73 * ( 2006 ) 063504 , [ http://xxx.lanl.gov/abs/astro-ph/0511494[astro-ph/0511494 ] ] .
v. berezinsky , v. dokuchaev , and y. eroshenko , _ anisotropy of dark matter annihilation with respect to the galactic plane _ , _ jcap _ * 0707 * ( 2007 ) 011 , [ http://xxx.lanl.gov/abs/astro-ph/0612733 [ astro - ph/0612733 ] ] .
e. zackrisson , s. asadi , k. wiik , j. jonsson , p. scott , k. k. datta , m. m. friedrich , h. jensen , j. johansson , c .- e .
rydberg , and a. sandberg , _
hunting for dark halo substructure using submilliarcsecond - scale observations of macrolensed radio jets _ , _
mon.not.roy.astron.soc . _
* 431 * ( 2013 ) 21722183 , [ http://xxx.lanl.gov/abs/1208.5482[arxiv:1208.5482 ] ] .
v. berezinsky , v. dokuchaev , y. eroshenko , m. kachelriess , and m. a. solberg , _ superdense cosmological dark matter clumps _ , _ phys.rev . _ * d81 * ( 2010 ) 103529 , [ http://xxx.lanl.gov/abs/1002.3444 [ arxiv:1002.3444 ] ] .
v. berezinsky , v. dokuchaev , y. eroshenko , m. kachelriess , and m. a. solberg , _ annihilations of superheavy dark matter in superdense clumps _ , _ phys.rev .
_ * d81 * ( 2010 ) 103530 , [ http://xxx.lanl.gov/abs/1002.3445[arxiv:1002.3445 ] ] .
yang , l. feng , x .- y .
huang , x. chen , t. lu , _ et .
_ , _ constraints on ultracompact minihalos from extragalactic @xmath215-ray background _ , _ jcap _ * 1112 * ( 2011 ) 020 , [ http://xxx.lanl.gov/abs/1112.6229[arxiv:1112.6229 ] ] .
y. yang , g. yang and h. zong , _ neutrino signals from ultracompact minihalos and constraints on the primordial curvature perturbation _ , _
phys.rev . _
* d87 * ( 2013 ) 103525 , [ http://xxx.lanl.gov/abs/1305.4213[arxiv:1305.4213 ] ] .
d. zhang , _ impact of primordial ultracompact minihaloes on the intergalactic medium and first structure formation _ , _
mon.not.roy.astron.soc . _
* 418 * ( 2011 ) 18501872 , [ http://xxx.lanl.gov/abs/1011.1935 [ arxiv:1011.1935 ] ] .
y. yang , x. huang , x. chen , and h. zong , _ new constraints on primordial minihalo abundance using cosmic microwave background observations _ , _ phys.rev . _ * d84 * ( 2011 ) 043506 , [ http://xxx.lanl.gov/abs/1109.0156[arxiv:1109.0156 ] ] .
j. ganc and e. komatsu , _ scale - dependent bias of galaxies and mu - type distortion of the cosmic microwave background spectrum from single - field inflation with a modified initial state _ , _ phys.rev . _ * d86 * ( 2012 ) 023518 , [ http://xxx.lanl.gov/abs/1204.4241[arxiv:1204.4241 ] ] . a. kogut , d. fixsen , d. chuss , j. dotson , e. dwek , _ et .
_ , _ the primordial inflation explorer ( pixie ) : a nulling polarimeter for cosmic microwave background observations _ , _ jcap _ * 1107 * ( 2011 ) 025 , [ http://xxx.lanl.gov/abs/1105.2044[arxiv:1105.2044 ] ] .
|
we describe a general procedure for using number counts of any object to constrain the probability distribution of the primordial fluctuations , allowing for generic weak non - gaussianity .
we apply this procedure to use limits on the abundance of primordial black holes and dark matter ultracompact minihalos ( ucmhs ) to characterize the allowed statistics of primordial fluctuations on very small scales .
we present constraints on the power spectrum and the amplitude of the skewness for two different families of non - gaussian distributions , distinguished by the relative importance of higher moments .
although primordial black holes probe the smallest scales , ultracompact minihalos provide significantly stronger constraints on the power spectrum and so are more likely to eventually provide small - scale constraints on non - gaussianity .
|
recent advances in producing synthetic spin - orbit coupling fields in cold atoms , both bosonic @xcite and fermionic , @xcite opened a new field of research in cold matter physics reviewed in refs .
spin - orbit coupling of the rashba and dresselhaus forms , as well as effective magnetic fields leading to the zeeman - like splitting for the corresponding pseudospin , can be designed there by optical means . in these systems
the pseudospin is formed by coupling hyperfine atomic levels with highly coherent resonant laser radiation . since this
coupling strongly depends on the detuning of the laser frequency from the resonance , the movement of an atom also modifies its interaction with light via the doppler shift linear in the atom velocity .
this effect is seen as the effective spin - orbit coupling .
there are at least two advantages of the synthetic spin - orbit couplings over those observed in solids @xcite .
first , the effect of the coupling on cold atoms can be considerably stronger than that on electrons in semiconductors .
second , it may be designed for a particular purpose , and can be switched on and off as required .
the relative strength of the coupling can make possible observation of new macroscopically ordered phases ( see e.g. @xcite ) , topological states ( e.g. @xcite ) , and nontrivial spin dynamics @xcite . flexibility in designing the fields suggests new applications of cold atoms in dynamical systems , for example , developing of new quantum measurement techniques .
below we use this coupling in order to simulate a measurement on spin precessing in an external magnetic field , a fundamental problem in many branches of physics .
conversely , spin - orbit coupling can be seen as a means of controlling the motion of atoms . in particular
, we will explore the measurement s ability to manipulate coherent motion of atoms in a bose - einstein condensate with spin - orbit coupling , where the effect of the coupling is enhanced by a large number of participating bosons in the same quantum state .
the rest of the paper is organized as follows . in sections ii and iii
we introduce coupled spin - orbit dynamics and show that it can serve as a quantum meter for the ( pseudo)spin degrees of freedom , with the center of mass of the atom playing the role of a von neumann pointer . in sec .
iv we study decoherence produced on the spin by the measurement . in sec .
v we use simple results from the measurement theory in order to describe the spacial distribution of the atomic cloud and identify zeno , ergodic , and intermediate regimes of the measurement . in section vi we will suggest possible experimental implementations .
brief conclusions are given in sec .
we begin with the rashba hamiltonian for an atom of mass @xmath0 moving in one dimension @xcite in the presence of an external magnetic field @xmath1 ( @xmath2 ) : @xmath3 where @xmath4 is the momentum , @xmath5 is the larmor frequency , @xmath6 is the spin - orbit velocity , and @xmath7 are the pauli matrices . the spin - orbit coupling term in eq .
results in breaking the galilean invariance @xcite , so that the system can have a non - zero velocity even for a zero total momentum .
the system has a characteristic length , the distance @xmath8 which an atom travels at @xmath6 in @xmath9 of the larmor period . with the help of @xmath10
we define the dimensionless variables @xmath11 , @xmath12 , @xmath13 , and @xmath14 , which we will use below unless stated otherwise .
the wavefunction of an atom with a pseudo - spin 1/2 is a two - component spinor @xmath15 which , in the new variables , satisfies the schrdinger equation : @xmath16 { \bm \psi}(x , t).\ ] ] we assume that the spin - orbit coupling is activated at @xmath17 , when the center of mass of the atom is described by a wave packet @xmath18 and its spin is in an initial pure state @xmath19}$ ] . with no loss of generality we consider an atom initially at rest , so that @xmath20 is peaked around @xmath21 .
( an additional constant field due to the atom s nonzero momentum can be absorbed in @xmath22 . ) finally , we choose @xmath22 to lie in the @xmath23-plane making an angle @xmath24 with the @xmath25-axis , @xmath26 .
the subsequent dynamics is not trivial : for each momentum @xmath27 , the spin orbit coupling adds an extra magnetic field , proportional to @xmath27 , along the @xmath25-axis , so that the resultant strength of the field in which the spin precesses , is uncertain .
the anomalous
spin - dependent contribution to the atom s velocity , @xmath28 $ ] , is proportional to the @xmath25-component of the spin , so the atom s translational motion is also complicated .
we can write the wave function as an integral over the atom s momenta , @xmath29}dp,\q\end{aligned}\ ] ] where @xmath30t\}$ ] is the evolution operator for a spin precessing in the composite field @xmath31 with components @xmath32 .
its matrix elements are @xmath33 where @xmath34
equivalently , @xmath15 can be written as a convolution in the coordinate space , @xmath35 where the wave packet @xmath36 describes the motion of the atom without spin - orbit coupling , and the sub - states @xmath37 are given by @xmath38 } dp.\end{aligned}\ ] ] both representations , ( [ 5 ] ) and ( [ 7 ] ) have advantages , which we will explore below .
we begin by revisiting classical measurements . consider a classical system with coordinate @xmath39 and momentum @xmath40 , coupled to a pointer , whose coordinate and momentum are @xmath41 and @xmath27 , respectively .
for the full hamiltonian we write @xmath42 where @xmath43 is the system s hamiltonian , and @xmath44 is some dynamical variable of interest @xcite . setting the pointer to zero , @xmath45 , and giving it an initial momentum , @xmath46
, we easily find @xmath47 where @xmath48 is the trajectory of a system governed by the modified hamiltonian @xmath49 .
thus , by reading the final pointer position , @xmath50 , we also measure the time average @xmath51 along a trajectory which is perturbed by the meter , unless the pointer is at rest , @xmath52 .
equation ( [ 4 ] ) is a quantum version of eq.([2 ] ) , with the @xmath25-component of the spin , @xmath53 playing the role of the measured variable , the center of mass of the atom with the position @xmath41 playing the role of the pointer , and the coupling to the magnetic field - the role of @xmath54 .
we have , therefore , a quantum measurement which we will analyze in detail .
the connection with the classical measurement above becomes even more evident if we slice the time interval into @xmath55 subintervals of length @xmath56 , send @xmath55 to infinity , and use the trotter formula to factorize @xmath57\epsilon\}=\exp\{-i(\mathbf{n}\cdot \bm{\sigma})\epsilon\}\exp(-ip\sigma_z\epsilon)$ ] . in this way , we obtain for the spin a set of virtual feynman paths taking the values @xmath58 , @xmath59 at each discrete time , and define for each path an evolution operator @xmath60 \times \\ & & \hspace{-1cm}|s_{k+1}\ra\la s_{k+1}|\exp[-i(\mathbf{n}\cdot\bm{\sigma})\epsilon]|s_{k}\ra\times \ldots \nonumber \\ & & \hspace{0.5cm}\la s_2|\exp[-i(\mathbf{n}\cdot \bm{\sigma})\epsilon]|s_{1}\ra\la s_1| \nonumber\\ & & = \exp\left[-ip\sum_{k=1}^k s_k\epsilon\right ] \hat{\mathcal{u}}(t,\rm path ) .
\nonumber \end{aligned}\ ] ] the sum in the first exponent is proportional to the time average of @xmath61 , defined along a spin s feynman path in a similar way the time average of @xmath44 , @xmath62 was earlier defined in eq.([3 ] ) for a classical trajectory .
the resulting time average of @xmath61 along a given path @xmath63 is : @xmath64\ra_t = t^{-1}\lim_{k\to \infty } \sum_{k=1}^k s_k\epsilon \equiv t^{-1}\int_0^t s(t')dt'.\end{aligned}\ ] ] interchanging integration over @xmath27 with summation over the paths , we can rewrite eq.([8 ] ) as a sum over only such evolutions for which the time average of @xmath61 equals @xmath65 in the form : @xmath66},\end{aligned}\ ] ] where @xmath67 , acting on the spin s degrees of freedom , is defined in eq.([10a ] ) .
this is a quantum analog of eq.([3 ] ) - the atom arrives at a location @xmath41 provided the time average of @xmath61 is equal to @xmath65 .
there are , however , important differences .
first , with many virtual paths involved , @xmath68 is a distributed quantity , whose value may lie between @xmath69 ( the spin spends all the time in the @xmath70 state ) and @xmath71 ( the spin spends all the time in the @xmath72 state ) .
second , we can only determine @xmath68 to a finite accuracy .
if @xmath73 in eq.([7 ] ) is peaked around @xmath74 with the width @xmath75 , by finding the particle in point @xmath41 one can ascertain the value of @xmath68 with an error @xmath76 .
finally , even though the mean momentum of the atom remains zero at all times , a narrow @xmath77 requires a spread of @xmath27 s around @xmath21 , and additional fields produced by these non - zero momenta perturb the spin s motion .
the perturbation is greater for more accurate measurements .
it is likely to result in decoherence of the initially pure spin state @xmath19}$ ] , which will discuss next .
taking the trace of the pure state @xmath78 over the atom s position @xmath41 , we find spin s reduced density matrix @xmath79 to be given by an incoherent sum , @xmath80 plus a spin - orbit field induced by @xmath27 ( cf
@xcite ) . from ( [ 6 ] )
we see that the matrix elements of @xmath81 contain terms which oscillate as functions of @xmath27 ever more rapidly as the time increases . for a smooth wave packet , such as a gaussian one , @xmath82
the integrals involving these oscillatory terms will vanish as @xmath83 , leaving the spin in a steady state @xmath84 which , in general , depends on @xmath75 and @xmath19}$ ] . the same applies to the averages of spin components @xmath85 $ ] , which will reach steady values at long times .
for example , for a spin initially polarized along the @xmath25-axis , @xmath86}=(1,0)^t,\end{aligned}\ ] ] we have @xmath87 the time dependence of @xmath88 , @xmath89 and @xmath90 is shown in fig.1 .
the type of decoherence in the spin subspace depends on the direction of the external magnetic field , and varies from gaussian , @xmath91 , for large @xmath75 and @xmath92 , to slow @xcite , @xmath93 for narrow wave packets , or @xmath24 close to @xmath94 .
the reason for this behavior , partially illustrated in fig.[fig : fig1 ] , can be seen as follows .
for example , the expression for the out - of - plane component @xmath95 contains oscillatory integrals of the type @xmath96dp,\ ] ] with @xmath97 defined in eq.([6a ] ) . expanding the phase @xmath97 to the second order in @xmath98 , @xmath99 , and evaluating the gaussian integrals , yields @xmath100^{1/4 } } \times \\ & & \quad\exp\left[-\frac{2 t^2(\delta x)^2\cos^2\theta}{(\delta x)^4\sin^2 \theta+4t^2}\right ] .
\nonumber\end{aligned}\ ] ] equation ( [ int ] ) smoothly interpolates between the inverse power law @xmath101 for @xmath102 , and the gaussian behavior @xmath103 $ ] for @xmath75 sufficiently large , at @xmath104 or @xmath105 . in the trivial case
@xmath106 all the fields are directed along the @xmath25-axis , there is no spin precession , and @xmath61 is conserved .
( components are shown near the lines ) for a relatively weak spin - orbit coupling @xmath107 , @xmath108 and @xmath19}=(1,0)^t$ ] .
the decoherence seen in @xmath109 is approximately gaussian , as indicated by the dashed line . as expected from eq.([12 ] ) for a large @xmath110 , we obtain @xmath111 here we use physical units for length and time . ]
to study spatial distribution of the atomic cloud we begin by noting from eq .
( [ 4 ] ) that the mean anomalous velocity of the atom is given by @xmath112}=\overline{\sigma}_{z}(t)$ ] , so that its mean position at a time @xmath113 is @xmath114 this is another quantum analog of the classical equation ( [ 3 ] ) .
it shows that once the @xmath25-component of the spin settles into its stationary value ( see fig.[fig : fig1 ] ) , the center of mass of the atomic cloud @xmath115 will move with a constant velocity .
equation ( [ 13 ] ) does not , however , tell us what the cloud looks like . to learn more about the cloud s shape we return to the convolution formula ( [ 7 ] ) .
it involves two known quantities : the wave packet @xmath73 describing the atom _ as it would be without the so interaction _ , and , via eq.([10 ] ) , the amplitude distribution of @xmath68 for a spin precessing in the magnetic field @xmath116 , also _ without the so interaction_. general properties of such distributions are known ( see , e.g. , @xcite ) , which simplifies our analysis . in the following
we will assume that the mass of the atom is sufficiently large to neglect the spreading of the wave packet ( a justification will be given in sect .
vi below ) .
thus , we have @xmath117^{1/4}\exp(-x^2/(\delta x)^2).\q\ ] ] the advantage of applying eq.([7 ] ) is that we effectively look at the details of the green s function ( [ 8 ] ) through a sort of a microscope with a resolution @xmath75 .
the smaller @xmath75 is , the more of the fine structure of @xmath37 is revealed .
the larger is @xmath75 is , the less information about the distribution of @xmath68 we obtain .
it is convenient to introduce the probability amplitudes to have @xmath118 provided the spin starts in the state @xmath119 and ends in the state @xmath120 , @xmath121\xi_{ij}.\end{aligned}\ ] ] deforming the integration paths in eq.([8 ] ) into contours in the complex @xmath27-plane , for @xmath122 we find ( @xmath123 ) @xmath124 and @xmath125 for @xmath126 and @xmath105 otherwise .
general properties of such amplitudes , as are mentioned above , are known in quantum measurement theory @xcite .
the diagonal terms , @xmath127 , have @xmath128-singularities responsible for the zeno effect .
the effect arises if , for a given @xmath113 , the error with which one measures a time average , tends to zero .
it locks the measured system in the eigenstates of the measured quantity , in our case , @xmath61 . for a time that is long compared to the period of the larmor precession , that is for @xmath129 , all @xmath130 are highly oscillatory , with stationary regions near @xmath131 , where @xmath132^t , are the spin states polarized along and against the effective magnetic field @xmath22 , respectively .
the stationary points account for the ergodic property of quantum motion @xcite . if the accuracy @xmath75 is kept finite , while the measurement time increases , the unperturbed hamiltonian @xmath133 overcomes the restrictions imposed by the meter .
the spin s density matrix becomes diagonal in the eigenbasis of @xmath134 , and the system explores its hilbert space in such a way that the time average of a variable tends to its average in the instantaneous von neumann measurement made on this diagonal mixed state @xcite . to prove the above , for @xmath129 we evaluate the integrals in ( [ 15 ] ) by the saddle point method , which yields @xmath135 , \nonumber \\ & & \xi_{12}(x , t)=i\sqrt{1/2\pi\beta}(t^2-x^2)^{-1/4 } \nonumber \\ & & \qquad\qquad\times \sin[\beta(t^2-x^2)^{1/2}+\pi/4].\end{aligned}\ ] ] it is seen that @xmath130 in eq.([14 ] ) have stationary points at @xmath136 in our case , the accuracy of measuring @xmath68 , @xmath137 changes with time , but so do the amplitude distributions @xmath138 , and one can not say _ apriori _ whether the improvement in the accuracy is rapid enough to lock the system in the zeno effect , or whether it would lead to the ergodic regime involving stationary points of the propagator described above .
next we will show that both situations are possible , depending on the ratio between the width of the wave packet @xmath77 , and the characteristic distance @xmath10 .
thus , for a narrow initial wavepacket , @xmath139 , and @xmath129 from ( [ 12 ] ) we have @xmath140 this is the zeno effect - at long times the spin is locked in its initial state @xmath141 @xcite .
accordingly , the time average @xmath68 also tends to unity , with only the @xmath128-term in the first of eqs.([15 ] ) contributing to @xmath15 in eq.([7 ] ) . as a result , the initial wave packet @xmath77 travels as a whole to the right at a constant speed @xmath6 , as can be seen in fig.[fig : fig2](a ) . for @xmath142 and @xmath129 eq.([12 ] )
yields @xmath143 .
the position of the atom is determined by the stationary points @xmath144 of the regular part of @xmath145 .
expanding the phases of the cosine and sine in eq.([16 ] ) around @xmath146 to the second order in @xmath147 and evaluating resulting gaussian integrals yields @xmath148|\mathbf{n}\ra + \\
\nonumber c^-\exp[-\gamma^-(x - x^{-})^2]|-\mathbf{n}\ra\end{aligned}\ ] ] where @xmath149\sqrt{\delta x(1\pm \cos\theta ) } } { 2^{3/4}\pi^{1/4}\sqrt{t^2\sin^2\theta \pm i(\delta x)^2 t/2 } } , \nonumber \\ & & \gamma^\pm=-\frac{(\delta x)^2\pm 2it\sin^2 \theta}{4t^2\sin^4\theta+(\delta x)^4}.\end{aligned}\ ] ] this is an example of the ergodic behavior - we have two gaussians , each corresponding to the spin polarized along and against the magnetic field .
the gaussians move in the opposite directions with velocities proportional to the long time average @xmath68 for the spin in the @xmath150 or @xmath151 state ( see fig . [
fig : fig2](b ) ) . by ergodicity @xcite ,
these time averages equal the ensemble averages @xmath152 and @xmath153 . with @xmath154 ,
the gaussians do not overlap , and the spin s mixed stated becomes @xmath155 so that @xmath156 equals @xmath157 , as is mentioned at the beginning of this paragraph .
the atomic density in the intermediate regime @xmath158 is shown in fig .
[ fig : fig2](c ) . here
a measurement resolves the oscillations of @xmath122 in ( [ 16 ] ) with a period larger than @xmath75 as can be obtained directly from eqs .
( [ 8]),([14 ] ) and ( [ 16 ] ) , while finer details such as more rapid oscillations are lost in the convolution ( [ 7 ] ) . ,
@xmath159 for @xmath19}=(1,0)^t$ ] : a ) in the zeno regime the atom moves to the right at the speed @xmath6 ; b ) in the ergodic regime the wave packet is split into two parts moving with the velocities @xmath160 .
the solid line is the exact numerical calculation , and the dashed line shows approximation ( [ 19 ] ) ; c ) interference fringes in the intermediate regime @xmath161 . here
we use physical units for length and time . ]
an experimental realization of the spin - orbit system described here may consist in collecting a large number of spin - polarized non- or weakly interacting bosonic atoms in the ground state of a quasi one - dimensional trap with a harmonic confinement along the @xmath41-axis . at @xmath17 the potential
is collapsed , and the spin - orbit coupling is switched on . then , depending on the initial confinement and the direction of the external magnetic field one may be able to drive the condensate as a whole ( see fig . [
fig : fig2](a ) ) , split it in two spin - polarized parts moving in the opposite directions ( see fig .
[ fig : fig2](b ) ) , or observe interference fringes similar to those shown in fig .
[ fig : fig2](c )
. now we return to physical units . to check the viability of the experiment , one may use as the energy scale the recoil energy @xmath162 , where @xmath163 is the wavelength of the laser light used to induce the spin - orbit coupling . for @xmath164 , one can achieve in experiment @xmath165 cm / s @xcite . with the characteristic frequency of the trap , @xmath166 , of the order of @xmath167 @xcite ,
the width of the initial gaussian state becomes @xmath168 .
the spreading rate of the initial gaussian , @xmath169 , is then about @xmath170 . as a result ,
the displacement of the wave packet due to spin - orbit coupling is much larger than the packet spreading , and the spreading does not influence the classification of asymptotic measurement regimes in fig.[fig : fig2 ] .
these can be obtained for a constant wave packet width since at given @xmath24 and @xmath171 they are determined solely by the @xmath110 ratio .
although the fringe pattern in fig.[fig : fig2](c ) can be influenced by the packet broadening at very long times , fringes will remain as a fingerprint of the intermediate regime .
taking into account that @xmath172 we obtain @xmath173 . by changing @xmath174 e.g. , by tuning the phase of the laser fields , in the range from zero to @xmath175
, one can go from the zeno to the ergodic regime .
finally , the repulsion between atoms can be neglected if the mean interaction energy per atom is small compared to @xmath176 - a condition easily achieved for typical bose gases .
in summary , the mechanism behind manipulation of a cold atom cloud by switching on and off the spin - orbit interaction is that of a quantum measurement .
the measurement , performed on the spin component coupled to the atom s momentum , can probe both the zeno and the ergodic regimes . depending on the external zeeman - like field
, it can lead to different decoherence regimes for the monitored spin , ranging from a gaussian decay to an algebraic law .
the effects of the measurement on the atoms density profiles include translation , spin - dependent fragmentation , and formation of interference fringes .
we acknowledge support of the mineco of spain ( grant fis 2009 - 12773-c02 - 01 ) , the government of the basque country ( grant `` grupos consolidados upv / ehu del gobierno vasco '' it-472 - 10 ) , and the upv / ehu ( program ufi 11/55 ) .
this type of locking can result in the inefficient spin - flip transitions caused by an external field in the presence of a strong spin - orbit coupling : r. li , j. q. you , c. p. sun and f. nori , phys .
lett . * 111 * , 086805 ( 2013 ) .
|
we show that by switching on a spin - orbit interaction in a cold - atom system , experiencing a zeeman - like coupling to an external field , e.g. , in a bose - einstein condensate , one can simulate a quantum measurement on a precessing spin . depending on the realization ,
the measurement can access both the ergodic and the zeno regimes , while the time dependence of the spin s decoherence may vary from a gaussian to an inverse fractional power law .
back action of the measurement forms time- and coordinate - dependent profiles of the atoms density , resulting in its translation , spin - dependent fragmentation , and appearance of interference patterns . u
|
general relativity predicts that gravitational collapse of massive objects leads to space - time singularities in rather general circumstances in our universe@xcite and such singularities might be accompanied with the blow up of physical quantities ( energy density , pressure and curvature of space - time , etc ) . the known physical laws including general relativity itself will break down in the neighborhood of the space - time singularity and hence the quantum theory of gravity is believed to be necessary to describe the physical phenomena in such a region .
one of the important issues is whether the space - time singularities formed in our universe are visible or not for the observer ( us ) far from the region where the gravitational collapse occurs .
the cosmic censorship conjecture proposed by penrose gave a strong motivation to investigate this problem@xcite .
roughly speaking , this conjecture states that the singularity is not visible for any observer if it is resulted from physically reasonable initial conditions .
however , this conjecture has not yet been proven .
rather candidates of the counterexample of this conjecture have been found .
the simplest example is the gravitational collapse of spherically symmetric dust fluid .
there is an exact solution of einstein equations for this case ; the so - called lematre - tolman - bondi ( ltb ) solution .
after eardley and smarr have pointed out the occurrence of a central shell focusing naked singularity of marginally bound dust collapse@xcite , several theoretical efforts revealed the genericity of the shell focusing naked singularity formation in the ltb solution@xcite .
there are several researches for more general spherically symmetric system and those revealed that non - vanishing pressure does not necessarily prevent the formation of the central shell focusing naked singularity@xcite . on the other hand , there are not so much researches for non - spherical systems as the spherically symmetric cases since general analytic approach is impossible .
nakamura et al .
and nakamura and sato performed numerical simulations for axisymmetric perfect fluid system@xcite .
their results suggest that naked singularities might be formed if the initial internal energy of the fluid elements is very small and initial configuration is sufficiently elongated .
this result is consistent with the hoop conjecture proposed by thorne , which states that black holes with horizons form when and only when a mass @xmath3 gets compacted into a region whose circumference in every direction is @xmath4@xcite .
shapiro and teukolsky also performed numerical simulations for the axisymmetric collision - less - particle system and showed a possibility of naked singularity formation which is also consistent with the hoop conjecture@xcite .
the critical behavior of the axisymmetric gravitational waves is also a candidate of the naked singularity formation@xcite .
an example which can be investigated analytically is the shell focusing naked singularity formation in the szekeres solution@xcite . anyway , even in the non - spherical cases , there are several examples of naked singularity formation .
there might be readers who state that numerical examples can not be regarded as candidates of naked singularity formation since the numerical simulation can not reveal the global structure of the space - time and further can not deal with infinite quantities ( the numerical technique recently proposed by h@xmath5bner might be able to avoid this difficulty@xcite ) .
however , the present authors would like to stress that if the region of sufficiently high energy , large pressure and large space - time curvature is visible from an observer far from the region , it should be regarded as a naked singularity .
the region with extremely large physical quantities will be described by the quantum theory of gravity which is little known and hence such region is equivalent to a space - time singularity in practical sense .
an important point of view for the naked singularity formation process has been proposed by nakamura et al.@xcite .
they pointed out that the visible strong curvature region may be a candidate of strong gravitational wave sources .
gravitational waves generated in such a region will propagate out from there since there is no event horizon . from this point of view , chiba investigated cylindrical gravitational collapse but his result is not consistent to the nakamura et al s conjecture@xcite .
the numerical simulations by shapiro and teukolsky essentially agrees with chiba s result with respect to the gravitational radiation .
however , we should be concerned about the numerically covered domain since the gravitational waves might be emitted form just the neighborhood of the naked singularity .
recently , present authors investigated the behavior of aspherical perturbations in the ltb space - time and showed that the weyl curvature of even mode perturbations corresponding to outgoing gravitational waves diverges@xcite .
although this result was obtained by numerical integration of linearized einstein equations , the numerical stability is guaranteed much better than that of the numerical simulation of full einstein equations .
further , linearized einstein equations were solved as the characteristic initial value problem and hence the numerically covered domain of the space - time by this analysis is much wider than the previous two numerical simulations .
the results of the linear perturbation analysis of the ltb space - time suggest that gravitational waves will be emitted in the formation process of naked singularities .
the purpose of the present paper is to re - analyze the dynamics of perturbations of the ltb space - time in the framework of the newtonian approximation . in order that the singularity of the spherically symmetric space - time is naked , `` the gravitational potential '' @xmath6 is smaller than unity in the neighborhood of the singularity , where @xmath3 is the misner - sharp mass function and @xmath7 is the areal radius .
the central shell focusing naked singularity of the ltb space - time satisfies this condition and further the gravitational potential vanishes even at this singularity .
the speed of the dust fluid is also much smaller than the speed of light before and at the central shell focusing naked singularity formation .
therefore the newtonian approximation seems to be available , even though the space - time curvature is infinite at the singularity .
the advantage of the newtonian approximation scheme is that the dynamics of perturbations of the dust fluid and gravitational waves generated by the motion of the dust fluid are separately estimated ; the evolution of the perturbations of the dust fluid are obtained by the newtonian dynamics and the gravitational radiation is obtained by the quadrupole formula .
hence the semi - analytic estimate of the gravitational radiation due to the matter perturbation of the ltb space - time is possible if we adopt the newtonian approximation .
the results of the newtonian perturbation analysis well agrees with the relativistic perturbation analysis by the present authors .
this suggests that the newtonian analysis will be a powerful tool in the analysis of some category of naked singularity .
however , we should stress that the neighborhood of the naked singularity is not newtonian in ordinary sense , because there is indefinitely strong tidal force .
further , studies of quantum effects have revealed that the violent quantum particle creation occurs in this space - time@xcite .
the particle creation is just a highly relativistic phenomenon .
the newtonian approximation scheme is available to describe the dynamics of the neighborhood of the naked singularity but the situation is not newtonian .
this paper is organized as follows . in sec.2
, we briefly review the ltb solution . in sec.3 , we consider the newtonian approximation of a spherically symmetric dust collapse and show the relation between the eulerian , lagrangian and synchronous - comoving ( sc ) coordinate systems in this approximation scheme .
further in this section , we show the validity of the newtonian approximation scheme even in the neighborhood of the central shell focusing naked singularity as long as we adopt the eulerian coordinate system .
the basic equations for the even mode of perturbations are presented in sec.4 . in order to estimate the gravitational radiation generated by aspherical perturbations of dust fluid
, we need the knowledge about mass - quadrupole moment .
we show an explicit expression of the quadrupole moment in sec.5 .
then we show the numerical calculation in sec.6 and the asymptotic behavior of the mass - quadrupole moment is presented in sec.7 .
finally , sec.8 is devoted for summary and discussion . in this article
, we adopt @xmath8 unit and basically follow the convention and notation in ref.@xcite .
the greek indices mean components of tensors , while the latin indices except for @xmath9 and @xmath10 represent a type of a tensor .
we use the sub- or super - scripts @xmath9 and @xmath10 to denote spatial components of tensors .
in general , the gravitational collapse of a spherical dust ball produces a shell focusing naked singularity at the symmetric center .
this singularity can be locally or globally naked in accordance with the initial rest - mass density configuration .
the solution of the einstein equations describing such a situation is known as the ltb space - time .
the line element is given by @xmath11 where @xmath12 is the line element of an unit 2-sphere .
the stress - energy tensor of the dust fluid is given by @xmath13 where @xmath14 is the rest - mass density and @xmath15 is the 4-velocity of the dust fluid element . in the coordinate system ( [ eq : sc - line ] ) , the components of the 4-velocity @xmath16 is given by @xmath17 einstein equations and the equation of motion for the dust fluid lead to the equations for the areal radius @xmath7 and rest - mass density @xmath14 as @xmath18 where @xmath19 and @xmath20 are arbitrary functions .
eq.([eq : energy - eq ] ) might be regarded as an energy equation of the dust fluid element at @xmath21 . from this point of view
, the function @xmath19 corresponds to the specific energy of the dust fluid element and the function @xmath20 can be regarded as the gravitational mass function .
the solution is completely fixed by choosing the specific energy @xmath19 and initial rest - mass density profile @xmath22 , or equivalently the mass function @xmath20 .
the shell focusing naked singularity is formed only at @xmath23 . for simplicity
, hereafter we will focus on the marginally bound case @xmath24 .
since our interest is on the behavior of the space - time near the central singularity , this restriction might not lose generality of the conclusion .
the solution of eq.([eq : energy - eq ] ) is given by @xmath25^{2\over3 } , \label{eq : radius}\ ] ] where @xmath26 is an arbitrary function which corresponds to the moment of the singularity formation .
we choose the time of singularity formation as @xmath27 so that @xmath7 agrees with @xmath21 at @xmath28 . in the spherically symmetric space - time
, we can naturally introduce a `` gravitational potential '' @xmath29 defined by @xmath30 as hayward discussed , the gravitational potential @xmath29 is deeply related to the formation of trapped region and hence gives a measure of the strength of the gravitational field@xcite . to see the behavior of the gravitational potential @xmath29 and the velocity @xmath31 of the dust fluid element
, we adopt the following initial density configuration for @xmath32 @xmath33 where @xmath34 and @xmath35 are positive constants .
since a sufficient condition for the nakedness of the central shell focusing singularity is @xmath36@xcite , the above initial density distribution necessarily leads to the naked singularity at the symmetric center @xmath23 . here
note that the moment of the central shell focusing singularity is @xmath37 by the above initial density profile . the core radius of the above configuration is given by @xmath38
if we set appropriate @xmath34 and @xmath35 , the space - time is globally naked singular . in fig.1
, we depict the gravitational potential @xmath29 with @xmath39 and @xmath40 .
we see that the gravitational potential @xmath29 is much smaller than unity even at the moment of the central shell focusing naked singularity formation .
note that form eq.([eq : energy - eq ] ) , @xmath41 is also much smaller than unity .
hence in this example , the newtonian approximation scheme seems to be available and in reality it is true as will be shown in the next section .
in this section , we consider gravitational collapse of a spherically symmetric dust fluid in the framework of the newtonian approximation and show that the newtonian approximation is valid even at the moment of the central shell focusing naked singularity formation if the initial condition is appropriately set up as in the case of the example in the previous section . for the purpose to follow the motion of a dust sphere , the lagrangian coordinate is more suitable than the eulerian one .
the transformation matrix between the eulerian and lagrangian coordinate systems is given by @xmath47 where regarding @xmath48 and @xmath49 as the independent variables , a dot means a partial derivative with respect to @xmath48 while a prime denotes a partial derivative with respect to @xmath49 .
then the line element in the lagrangian coordinate system is obtained as @xmath50 equations for the dust fluid and newtonian gravitational potential are given by @xmath51 where @xmath52 and @xmath53 are regarded as arbitrary functions .
since the equation for the areal radius @xmath7 is the same as that of the ltb space - time , its solution for the marginally bound collapse @xmath54 is given by the same functional form as eq.([eq : radius ] ) , @xmath55^{2\over3 } , \label{eq : n - radius}\ ] ] where @xmath56 is an arbitrary function which determines the moment of singularity formation .
here we consider the newtonian approximation of the example given in the previous section .
hence we choose the moment of the singularity formation as @xmath57 so that @xmath7 agrees with @xmath49 at @xmath58 . as for the initial density configuration
, we adopt the same functional form as eq.([eq : initial - density ] ) , @xmath59 where @xmath60 and @xmath61 are positive constants .
the above choice guarantees the regularity of all the variables before the singularity formation and that the central shell focusing singularity is formed at @xmath62 . imposing a boundary condition @xmath63 for @xmath64 , the solution of eq.([eq : l - poisson ] )
is formally expressed as @xmath65 where @xmath66 the behavior of @xmath67 is the same as that of @xmath29 as shown in fig.1 .
on the other hand , the newtonian gravitational potential @xmath43 of @xmath68 and @xmath69 is depicted in fig.2(a ) .
form these figures , we find that the condition ( [ eq : n - condition ] ) is satisfied even at the moment of the central shell focusing naked singularity formation . here
it is worthwhile to note that the right hand side of eq.([eq : l - poisson ] ) at @xmath70 diverges at the moment of the central shell focusing naked singularity formation , @xmath71 where @xmath72 however since the power index of @xmath49 is larger than @xmath73 , @xmath43 itself is finite at @xmath70 even at the moment of the central shell focusing singularity formation @xmath62 . in order
that the newtonian approximation is successful , temporal derivatives of all the quantities should always smaller than the radial derivatives of those .
here we shall focus on the neighborhood of the central shell focusing naked singularity only . for this purpose
, we introduce a new variable @xmath74 defined by @xmath75 where @xmath76 .
then we consider a limit @xmath77 with fixed @xmath74 .
it should be noted that @xmath49 also goes to zero by this limiting procedure .
the mass function @xmath78 , rest - mass density @xmath14 and areal radius @xmath7 behave as @xmath79 all these variables are proportional to the power of @xmath80 and the coefficients of those are functions of @xmath74 .
it is easy to see that their derivatives with respect to @xmath48 or @xmath49 also take the same functional structure with respect to @xmath80 and @xmath74 .
thus the @xmath49-dependent part @xmath81 of the newtonian gravitational potential will also behave in the manner @xmath82 where @xmath83 is a constant and @xmath84 is a function of @xmath74 . substituting the above equation into eq.([eq : l - poisson ] ) and using the asymptotic behavior ( [ eq : f - ap ] ) and ( [ eq : r - ap ] ) ,
we obtain @xmath85 in order that the dependence of both sides in the above equation on @xmath80 agrees with each other , @xmath83 should be equal to @xmath86 .
integration of eq.([eq : phi - ap - equation ] ) leads to @xmath87 in order to see the asymptotic dependence of @xmath88 on @xmath48 , we differentiate eq.([eq : phi2-def ] ) to obtain @xmath89 the integrand in the right hand side of the above equation behaves near the origin at the moment of the central shell focusing singularity formation as @xmath90 therefore , the integral in eq.([eq : dt - phi ] ) does not have finite value at @xmath62 .
since as shown in the above , this divergence comes from the irregularity of the integrand at the origin @xmath70 , we shall estimate the contribution near the origin to the integral in eq.([eq : dt - phi ] ) .
we again consider the limit of @xmath91 with fixed @xmath74 and obtain @xmath92 substituting the above equation into eq.([eq : dt - phi ] ) and integrating it with respect to @xmath48 , we obtain @xmath93 as a result , in the limit of @xmath91 with fixed @xmath74 , @xmath43 is expressed in the form @xmath94 we depict numerically obtained @xmath43 together with the above asymptotic form in fig.2(b ) .
the asymptotic estimate agrees with the numerical result quite well .
now we have known that in the limit of @xmath95 with fixed @xmath74 , all the variables behave as @xmath96 where @xmath97 is some function of @xmath74 and @xmath98 is a constant .
the derivatives of @xmath99 with respect to @xmath100 and @xmath7 are expressed by using its derivatives with respect to @xmath48 and @xmath49 as @xmath101 from eqs.([eq : l - matrix1 ] ) and ( [ eq : l - matrix2 ] ) , we find @xmath102 then the following relation is derived , @xmath103 inserting eq.([eq : z - ap ] ) into the above equation , we obtain @xmath104 the above equation means that in the limit of @xmath91 with fixed @xmath74 , the following inequality holds @xmath105 therefore , the order counting of the newtonian approximation is guaranteed even in the neighborhood of the central naked singularity .
here it is worthy to notice that in the limit of @xmath91 with fixed @xmath74 , @xmath106 is much larger than @xmath107 because of @xmath108 hence the vicinity of the central shell focusing naked singularity is not newtonian situation in ordinary sense . in reality
, the present authors have shown that the violent quantum particle creation occurs in such a situation@xcite .
this means that the formation of the central shell focusing singularity is a highly relativistic phenomenon even though it is well described by the newtonian approximation in the eulerian coordinate system . in the ltb solution ,
the sc coordinate system is adopted .
we should note that the sc coordinate system is different from the lagrangian one .
we consider the following coordinate transformation @xmath109 assuming eq.([eq : n - condition ] ) , derivatives of the areal radius @xmath7 with respect to the lagrangian time coordinate @xmath48 and the radial coordinate @xmath49 are written as @xmath110 hence the line element in the new coordinate system @xmath111 up to the lowest order in the sense of eq.([eq : n - condition ] ) is written as @xmath112 from the above equation , we find that the coordinate transformation ( [ eq : sc - matrix1 ] ) and ( [ eq : sc - matrix2 ] ) leads to the sc coordinate system .
especially , the above line element is completely the same as the relativistic one in the marginally bound case .
further we obtain the equations of the lowest order in the sc coordinate system are given in completely the same form as that in the relativistic one : @xmath113 it is worthwhile to note that in the sc coordinate system , the newtonian gravitational potential @xmath43 does not appear .
we consider non - spherical linear perturbations in the system of the spherically symmetric dust ball described in the previous section .
first , we consider perturbations in the eulerian coordinate system .
the line element is written as @xmath114 where @xmath115 is a perturbation of the newtonian gravitational potential . using the transformation matrix ( [ eq : l - matrix1 ] ) and
( [ eq : l - matrix2 ] ) , we obtain the perturbed line element in the background lagrangian coordinate system as @xmath116 hereafter we discuss the behavior of perturbations in this coordinate system .
the density @xmath117 and 4-velocity @xmath118 are written in the form @xmath119 by definition of the lagrangian coordinate system , the components of the background 4-velocity is given by @xmath120 from the normalization of the 4-velocity , we find @xmath121 the order - counting with respect to the expansion parameter @xmath122 of the newtonian approximation is given by @xmath123 then the equations for the perturbations are given by @xmath124 where @xmath125 and @xmath126 is a contravariant component of the background 3-metric . here
we focus on the axisymmetric even mode of perturbations .
hence the perturbations are expressed in the form @xmath127 from eqs.([eq : rho - eq1 ] ) , ( [ eq : v - eq1 ] ) and ( [ eq : p - eq1 ] ) , we obtain @xmath128 comparing the basic equations for the relativistic perturbations of @xmath129 in ref.@xcite to the above equations , we find the correspondence between the newtonian and relativistic variables ; @xmath130 , @xmath131 and @xmath132 .
hereafter we focus on the quadrupole mode @xmath129 and therefore omit the subscript @xmath133 to specify the multi - pole component of the perturbation variable .
the mass - quadrupole moment @xmath134 is given by @xmath135,\ ] ] where @xmath136 for a function @xmath137 with sufficiently rapid fall off for @xmath138 , we find that @xmath139_{0}^{\infty } = \int_{0}^{\infty}\partial_{\tau}(r'g)dx.\end{aligned}\ ] ] using the above formula , we obtain @xmath140 the power @xmath141 carried by the gravitational radiation at the future null infinity @xmath142 is given by @xmath143 where @xmath144 is the retarded time .
the weyl scalar @xmath145 carried by outgoing gravitational waves at the future null infinity is estimated as @xmath146 where @xmath147 is the weyl tensor , and @xmath148 and @xmath149 are two of the null tetrad basis whose components in the spherical polar coordinate are given by @xmath150 it should be noted that the power @xmath141 is proportional to the square of the 3rd - order derivative of @xmath151 while the weyl scalar @xmath145 is proportional to the 4th - order derivatives of @xmath151 .
in this section , we explain the procedure to numerically integrate the system of partial differential equations for the quadrupole perturbation variables and show the time evolution of the mass - quadrupole moment .
we assume that all the perturbation variables are regular before the central shell focusing singularity formation and hence are written in the form of the taylor series as @xmath152 in the numerical calculation , we focus on the above asterisked variables rather than the original variables . to obtain the solution of eq.([eq : p - eq ] ) , we introduce a variable @xmath153 defined by @xmath154 substituting the above form of @xmath155 into eq.([eq : p - eq ] ) , we obtain the equation for @xmath153 as @xmath156 integrating the above equation , we obtain @xmath157 where we have chosen the integration constant so that @xmath158 is finite for @xmath64 .
further integration leads to @xmath159 where @xmath160 and we have chosen the integration constant so that @xmath153 vanishes at the origin , @xmath70 . hence @xmath161 is written as @xmath162 for the numerical calculation , we rewrite the basic equations ( [ eq : rho - eq])@xmath163([eq : ut - eq ] ) for the perturbation variables . differentiating the conservation law ( [ eq : rho - eq ] ) with respect to time @xmath48 and then using eqs.([eq : rho - eq])@xmath163([eq : ut - eq ] ) , we obtain a second order differential equation with respect to @xmath48 for @xmath164 as @xmath165 where @xmath166 the equations for @xmath167 and @xmath168 are written as @xmath169 eqs.([eq : dphi - eq2])@xmath163([eq : ut - eq2 ] )
constitute a closed system of equations , which is solved numerically .
the background solution is given by ( [ eq : n - radius ] ) , ( [ eq : n - s - time ] ) and ( [ eq : n - initial - density ] ) .
the parameters @xmath60 and @xmath61 in @xmath170 are chosen to be @xmath68 and @xmath69 , respectively .
then the core radius @xmath171 of the background rest - mass density distribution becomes @xmath172 we start the numerical integration at @xmath58 .
the initial data of the density perturbation @xmath164 is set up in the form @xmath173 where @xmath174 is a positive constant smaller than unity .
we set @xmath175 . from eqs.([eq :
ux - eq ] ) and ( [ eq : ut - eq ] ) , we find @xmath176 hence @xmath177 is written in the form @xmath178 where @xmath179 is an arbitrary function .
as for the initial data , we set @xmath180 and @xmath181 where @xmath182 is a positive constant smaller than unity and @xmath183 is also a constant . from eq.([eq : rho - eq ] ) , the initial data for @xmath184 should be @xmath185 the temporal integration of eqs.([eq : vrho - eq2])@xmath163([eq : ut - eq2 ] ) is performed by the second order runge - kutta method . in order to obtain @xmath161 and @xmath186
, we numerically integrate eqs.([eq : cal - a ] ) and ( [ eq : cal - b ] ) also by the second order runge - kutta method and then substitute the results into eqs.([eq : dphi - eq2 ] ) and ( [ eq : dphi - def ] ) .
the number of spatial grid points is @xmath187 and the outermost grid point corresponds to @xmath188 .
since we numerically integrate the second order differential equation with respect to @xmath48 for the density perturbation @xmath164 , the conservation law ( [ eq : rho - eq ] ) is not trivially satisfied by the numerical error .
hence we use eq.([eq : rho - eq ] ) as a check of accuracy of the numerical integration .
we define the error function @xmath189 as @xmath190^{-1}.\ ] ] we perform the numerical integration until @xmath191 .
the function @xmath192 in @xmath193 is less than @xmath194 .
hence the numerical integration is almost consistently performed near the origin and hence we can observe the asymptotic behavior of the contribution near the central shell focusing singularity to the mass - quadrupole moment in the limit of @xmath195 . in fig3 , we show @xmath196 , @xmath197 and @xmath198 .
the numerical integration was performed from @xmath199 to @xmath191 . from this figure
, we find that those variables do not show power - law behavior with respect to @xmath200 but rather their behavior is oscillatory and growing . at this stage , the mass - quadrupole moment
is determined mainly by the motion of dust fluid in the outside region and the contribution in the neighborhood of the central shell focusing singularity is still negligible .
however , the contribution near the origin will be dominant in the limit @xmath195 as will be shown in the next section .
in order to get information of the asymptotic behavior of the mass - quadrupole moment , we should carefully examine the asymptotic behavior of the perturbation variables near the origin .
for this purpose , we introduce @xmath74 defined in eq.([eq : w - definition ] ) and then consider the limit @xmath77 with fixed @xmath74 .
since all the background variables appearing in the equations of the perturbations are proportional to the power of @xmath80 and the coefficients of those are the functions of @xmath74 as eqs.([eq : f - ap])@xmath163([eq : r - ap ] ) and ( [ eq : n - phi - ap ] ) , we expect that the perturbation variables also behave in the same manner as the background variables and hence we assume @xmath201 where we have assumed that the contribution of @xmath179 to @xmath167 is negligible in the limit of @xmath195 . in fig.4 , we depict the numerical results for the asterisked variables at the origin as functions of @xmath200 in the case of @xmath202 . from this figure
, we can easily find that the hatted variables show the power - low behavior and grow monotonically .
the power indices are estimated as @xmath203 due to this growing behavior , in reality , the contribution of @xmath179 to @xmath167 becomes negligible in the limit @xmath195 .
although we do not show the case of @xmath204 , we have confirmed that the behavior of the asterisked variables in the case of @xmath204 is almost identical to the case of @xmath202 in the limit @xmath195 .
next we depict the following normalized perturbation variables with respect to @xmath74 for @xmath205 at various time steps for the time interval @xmath206 in figs.5 , 6 and 7 ; @xmath207 note that @xmath208 at @xmath191 becomes 381 times larger than that at @xmath209 . as expected , the spatial configuration of these variables with respect to @xmath74 are almost time independent .
hence the assumptions ( [ eq : rho - asp])@xmath163([eq : p - asp ] ) are justified by the numerical calculation .
now , by virtue of our knowledge about the asymptotic forms ( [ eq : rho - asp])@xmath163([eq : p - asp ] ) , more rigorous discussion about the evolution of the mass - quadrupole moment is possible .
substituting eqs.([eq : rho - asp])@xmath163([eq : p - asp ] ) into eqs.([eq : rho - eq])@xmath163([eq : p - eq ] ) , and using the asymptotic behavior of the background variables ( [ eq : f - ap])@xmath163([eq : r - ap ] ) , we obtain @xmath210 \delta\tau^{-q-7/3}=0,\label{eq : conti - ap}\ ] ] @xmath211 and @xmath212 \delta\tau^{-r+5/3 } -{2\over9}w^{4}{\delta}^{*}_{\rho}\delta\tau^{-p+2}=0.\label{eq : p - ap}\ ] ] since the power of @xmath80 should be balanced in each equation , we obtain @xmath213 eqs.([eq : conti - ap])@xmath163([eq : p - ap ] ) constitute a closed system of ordinary differential equations . in order to obtain a solution of the above equations , we also need numerical integration .
however , here we are going to search for regular and gentle solutions of the ordinary differential equations .
this is much easier than the previous numerical integration of the partial differential equations of which solutions show singular behavior .
further high numerical accuracy is guaranteed and therefore the following analysis might be called `` semi - analytic '' . by an appropriate manipulation@xcite , we obtain a single decoupled equation for @xmath214 as @xmath215 where @xmath216 and @xmath217 giving an appropriate boundary condition , we can numerically solve eqs.([eq : conti - ap])@xmath163([eq : p - ap ] ) as a kind of the eigen value problem to obtain @xmath218 and the solution for @xmath214 .
the boundary condition at @xmath219 is given by the following procedure .
first , we assume that @xmath220 , @xmath214 and @xmath221 are @xmath222 functions . by this assumption ,
the perturbation variables are written in the form of the taylor series as , @xmath223 then substituting the above expressions into eqs.([eq : rho - eq])@xmath163([eq : p - eq ] ) , we obtain @xmath224 where @xmath10 is non - negative integer . from the above equations , we obtain @xmath225 since eq.([eq : u - eq ] ) is linear , the value of @xmath226 has no special meaning and hence we set @xmath227 , for simplicity .
then the boundary condition at @xmath219 for eq.([eq : u - eq ] ) is uniquely determined as @xmath228 we numerically integrate eq.([eq : u - eq ] ) outward from @xmath219 by 4th - order runge - kutta method .
the behavior of @xmath214 depends on the value of @xmath218 . in order to set up the boundary condition at the outer numerical boundary
, we consider the behavior of eq.([eq : u - eq ] ) in the limit of @xmath229 , @xmath230 therefore , @xmath231 behaves as @xmath232 inserting the above equation into eq.([eq : u - eq - ap ] ) , we obtain a 4th - order algebraic equation for @xmath233 .
the solutions of this equation are given by @xmath234 in order to pick up a solution from the above four independent asymptotic solutions , we consider @xmath235 in the limit of @xmath236 , @xmath237 here we impose a condition that @xmath235 is non - zero and finite for @xmath238 at @xmath239 , where @xmath240 is a positive infinitesimal number .
this condition leads to @xmath241 hence by varying @xmath218 , we search for the solution which behaves @xmath242 the numerical calculation reveals that the above behavior is realized when @xmath243 from eq.([eq : power ] ) , we obtain @xmath244 the above values agree with the numerical results ( [ eq : n - results ] ) quite well .
we depict the data for @xmath245 , @xmath214 and @xmath246 in figs.5 , 6 and 7 together with the variables @xmath247 , @xmath248 and @xmath249 obtained by the numerical calculation of eqs.([eq : vrho - eq2])@xmath163([eq : ut - eq2 ] ) , where @xmath250 and @xmath251 are defined by eqs.([eq : p - rho - def ] ) and ( [ eq : p - phi - def ] ) , respectively .
we see the quite excellent agreement .
now we examine the mass - quadrupole moment @xmath252 and its time - derivatives @xmath253 . in order to see the contribution of the central singularity to @xmath253
, we consider the integrand in the right hand side of eq.([eq : delt - q ] ) .
using eqs.([eq : f - ap ] ) , ( [ eq : r - ap ] ) and ( [ eq : rho - asp ] ) , we obtain @xmath254 from the above equation , we obtain @xmath255 we consider the integral of @xmath256 from @xmath70 to @xmath257 to see the contribution of the central shell focusing naked singularity to the time derivatives of the mass - quadrupole moment . here we take a limit @xmath91 with fixed @xmath258 and then consider the limit @xmath259 . as a result ,
we obtain @xmath260 the above equation and eq.([eq : ap - results ] ) show that the contribution of the central singularity to @xmath253 diverges for @xmath195 if and only if @xmath10 is larger than or equal to 4 .
this result and the quadrupole formula imply that the metric perturbation corresponding to the gravitational radiation and its first order temporal derivative is finite but the second order temporal derivative diverges .
hence the power @xmath141 of the gravitational radiation is finite but the curvature @xmath145 carried by the gravitational waves from the central naked singularity diverges ; see eqs.([eq : g - power ] ) and ( [ eq : curvature ] ) .
this agrees with the relativistic perturbation analysis .
further , we find that in the limit of @xmath195 , @xmath261 this result is also consistent with the relativistic perturbation analysis@xcite .
we analyzed the even mode perturbations of @xmath129 in the spherically symmetric dust collapse in the framework of the newtonian approximation and estimated the gravitational radiation generated by these perturbations by the quadrupole formula .
since we treat separately the dynamics of the matter perturbations and the gravitational waves in the wave zone , we can estimate the asymptotic behavior semi - analytically , where `` semi - analytically '' means that we know it by solving the gentle ordinary differential equations .
this is the great advantage of the newtonian approximation . as a result
, we found that the power carried by the gravitational waves from the neighborhood of the naked singularity at the symmetric center is finite .
however , the space - time curvature associated to the gravitational waves becomes infinite in accordance with the power law .
this result is consistent with recently performed relativistic perturbation analysis by iguchi et al.@xcite .
furthermore , the power index obtained by the newtonian analysis also agrees with the relativistic perturbation analysis quite well .
the agreement between the results of the newtonian and relativistic analyses implies that the perturbations themselves are always confined within the range to which the newtonian approximation is applicable . here
we will focus on the metric perturbation , @xmath161 . since the asymptotic solution of @xmath161 has the same form as eq.([eq : z - ap ] )
, we immediately find that in the limit of @xmath91 with fixed @xmath74 , @xmath262 and hence the assumption @xmath263 of the newtonian approximation is valid in the eulerian coordinate system .
we can also see 2nd - order derivatives . in the same limit , we find @xmath264 from the above equations , we obtain @xmath265 the above equation means that in the limit of @xmath91 with fixed @xmath74 , the following inequality is also satisfied , latexmath:[\[|\partial_{t}^{2}{\delta}^{*}_{\phi}|\ll however as mentioned in sec.2 , the gravitational collapse producing the shell focusing globally naked singularity is not newtonian in ordinary sense . the same is true for the perturbation variables because @xmath267 in the limit of @xmath91 with fixed @xmath74 . even though the newtonian approximation is valid for the eulerian coordinate system , the newtonian order counting breaks down if we adopt the lagrangian coordinate system as the spatial coordinates .
we are grateful to h. sato for his continuous encouragement .
we are also grateful to t. nakamura , h. kodama , t.p .
singh , a. ishibashi and s.s .
deshingkar for helpful discussions .
this work was supported by the grant - in - aid for scientific research ( no .
05540 ) and for creative basic research ( no .
09np0801 ) from the japanese ministry of education , science , sports and culture .
99 r. penrose , phys .
* 14 * , 57 ( 1965 ) .
s. w. hawking , proc .
london * a300 * , 187 ( 1967 ) .
s. w. hawking and r. penrose , proc .
london * a314 * , 529 ( 1970 ) .
r. penrose , riv .
nuovo cim . * 1 * , 252 ( 1969 ) .
eardley and l. smarr , phys .
d * 19 * , 2239 ( 1979 ) .
d. christodoulou , commun .
phys . * 93 * , 171 ( 1984 ) .
r.p.a.c .
newman , class .
quantum grav .
* 3 * , 527 ( 1986 ) .
joshi and i.h .
dwivedi , phys , rev , d * 45 * , 5357 ( 1993 ) .
t.p . singh and p.s .
joshi , class . quantum grav . * 13 * , 559 ( 1996 ) .
s. jhingan , p.s .
joshi and t.p .
singh , class .
quantum grav .
* 13 * , 3057 ( 1996 ) .
a. ori and t. piran , gen .
. gravit . *
20 * , 7 ( 1988 ) .
a. ori and t. piran , phys .
d * 42 * , 1068 ( 1990 ) .
t. harada , phys .
d * 58 * , 104015 ( 1998 ) .
t. harada , h. iguchi and k. nakao , phys .
d * 58 * , 104015 ( 1998 ) .
t. nakamura , k. maeda , s. miyama and m. sasaki , _ proceedings of the 2nd marcel grossmann meeting on general relativity _ edited by r. ruffini ( amsterdam : north - holland ) ,
p675 ( 1982 ) . t. nakamura and h. sato , prog .
. phys . * 67 * , 1396 ( 1982 ) .
thorne , in _ magic without magic ; john archibald wheeler _ , edited by j.klauder ( frieman , san francisco , 1972 ) , p231 . *
58 * , 104015 ( 1998 ) .
s. l. shapiro and s. a. teukolsky , phys .
lett . * 66 * , 994 ( 1991 ) .
a. m. abrahams and c. r. evans , phys .
lett . * 70 * , 2980 ( 1993 ) .
p. s. joshi and a. krolak , class . and quantum grav .
* 13 * , 3969 ( 1996 ) .
p. h@xmath5bner , phys . pev . *
d53 * , 701 ( 1996 ) .
t. nakamura , m. shibata and k. nakao , prog .
89 * , 821 ( 1993 ) .
t. chiba , prog .
. phys . * 95 * , 321 ( 1996 ) .
h. iguchi , t. harada and k. nakao , prog .
. phys . * 103 * , 53 ( 2000 ) s. barve , t.p .
singh , c. vaz and l. witten , nucl .
b * 532 * , 361 ( 1998 ) .
s. barve , t.p .
singh , c. vaz and l. witten , phys .
d * 58 * , 104018 ( 1998 ) . c. vaz and l. witten , phys .
b * 442 * , 90 ( 1998 ) .
t. harada , h. iguchi and k. nakao , phys.rev.*d61 * , 101502(r ) ( 2000 ) .
t. harada , h. iguchi and k. nakao , phys.rev.*d * in press .
r. m. wald , _ general relativity _
( the university of chicago press , 1984 ) .
hayward , phys . rev . *
d53 * , 1938 ( 1996 ) in order to avoid miss calculations , we have used the _ mathematica_.
|
spherical dust collapse generally forms a shell focusing naked singularity at the symmetric center .
this naked singularity is massless . further the newtonian gravitational potential and speed of the dust fluid elements are everywhere much smaller than unity until the central shell focusing naked singularity formation if an appropriate initial condition is set up .
although such a situation is highly relativistic , the analysis by the newtonian approximation scheme is available even in the vicinity of the space - time singularity .
this remarkable feature makes the analysis of such singularity formation very easy .
we investigate non - spherical even - parity matter perturbations in this scheme by complementary using numerical and semi - analytical approaches , and estimate linear gravitational waves generated in the neighborhood of the naked singularity by the quadrupole formula .
the result shows good agreement with the relativistic perturbation analysis recently performed by iguchi et al .
the energy flux of the gravitational waves is finite but the space - time curvature carried by them diverges . 0.5 cm pacs number(s ) : 04.25.nx,04.30.db,04.20.dw @ @.2326ex # 1^to-1.4@-@@ ken - ichi nakao@xmath0 , hideo iguchi@xmath1 and tomohiro harada@xmath2 + _ @xmath0department of physics , osaka city university _
+ _ osaka 558 - 8585 , japan _
+ _ @xmath1department of earth and space science , graduate school of science , osaka university
_ + _ toyonaka , osaka 560 - 0043 , japan _ + _ @xmath2department of physics , waseda university _
+ _ oh - kubo , shinjuku - ku , tokyo 169 - 8555 , japan _ +
|
upon a rotation of angle @xmath12 of the spin quantization axis in the @xmath66 plane , the rotated of up and down density operators take the following form in the basis of the z - axis eigenstates .
@xmath67 @xmath68 thus , these are the local mean field operators in a collineal calculation only allowing the development of magnetism along this particular quantization axis .
this choice of mean field operators restrics the avaible hilbert space to the collinear solutions , whereas the inclusion of the exchange term will lead to the full non - collinear solutions .
even though the anisosotrpy calculations were performed on a zigzag edge , localized edge states also appear on chiral nanoribbons @xcite .
thus , this anisotropy effect will appear as well in chiral ( n , m ) nanoribbons due to the local lattice imbalance . as a particular example
, we show the selfconsistent band structure for in - plane and off - plane edge ferromagnetic configurations for a ( 8,1 ) chiral ribbon , as reported in @xcite .
it is observed that , in the same fashion as in the zigzag ribbon , the off - plane magnetic solution host gapless edge states whereas the in - plane solution opens up a gap in the topological edge states , being this last one the lowest energy configuration of the system .
|
the independent predictions of edge ferromagnetism and the quantum spin hall phase in graphene have inspired the quest of other two dimensional honeycomb systems , such as silicene , germanene , stanene , iridiates , and organometallic lattices , as well as artificial superlattices , all of them with electronic properties analogous to those of graphene , but much larger spin - orbit coupling .
here we study the interplay of ferromagnetic order and spin - orbit interactions at the zigzag edges of these graphene - like systems .
we find an in - plane magnetic anisotropy that opens a gap in the otherwise conducting edge channels , that should result in large changes of electronic properties upon rotation of the magnetization .
magnetic anisotropy , a technologically crucial property , is driven by spin - orbit interaction , which is normally the underdog in the competition with the other two terms that control ferromagnetism , namely , kinetic and coulomb energy@xcite . as a result ,
magnetic anisotropy energy in conventional ferromagnets is at least 2 orders of magnitude smaller than the curie temperature and the fermi energies ( or the band - gap , in the case of insulators ) . for the same reason ,
transport properties in ferromagnetic metals are only weakly dependent on the magnetic orientation , and typical values for the anisotropic magnetoresistance ( amr ) are below 3 percent@xcite . here
we study magnetic anisotropy in a class of systems for which the balance between these three energy scales is very different from the usual , which leads to two dramatic consequences , very different from conventional ferromagnetism .
first , the conducting properties change from metal to insulator , depending on the magnetization orientation , an effect that , to the best of our knowledge , has never been reported .
second , the magnetic moment magnitude depends strongly on the magnetic orientation , and it can change even vanish in some directions , a phenomenon dubbed colossal magnetic anisotropy@xcite .
the class of systems in question are the zigzag edges of two dimensional honeycomb crystals@xcite whose electronic properties can be described with a tight - binding model with a single orbital per site and kane - mele spin - orbit interactions@xcite .
this includes several materials , such as group iv two dimensional crystals ( graphene@xcite , as silicene@xcite , germanene@xcite , and stanene@xcite ) , the double layer perovskyte iridates @xcite , and metal organic frameworks ( mof ) @xcite .
in addition , given that the existence of non - dispersive edge states occurs at the zigzag edge of any system described with the dirac equation@xcite , the results discussed here should also be valid for the so - called designer dirac fermions formed in `` artificial graphene '' formed by decoration of two dimensional electron gases with honeycomb arrangements@xcite . ignoring spin - orbit and coulomb interactions altogether , these 2d crystals are zero band - gap semiconductors with dirac - like dispersion close to the fermi energy .
zigzag edges in these systems are known to host localized edge states that , when both coulomb and spin - orbit coupling are neglected , are non - dispersive , sub lattice polarized , and lie precisely at the fermi energy , at half - filling@xcite . the ensuing large density of states results in a stoner instability that leads to ferromagnetic order at the edge@xcite .
on the other hand , kane - mele spin - orbit interaction , a second - neighbor spin dependent hopping that conserves the spin component @xmath0 perpendicular to the two dimensional crystal@xcite , has dramatic consequences in these honeycomb crystals .
it opens a topologically non - trivial gap in bulk and the emergence of in - gap spin - filtered dispersive edge states : for a given spin projection @xmath0 , electrons propagate along one direction only , preventing back - scattering even in the presence of time - reversal symmetric disorder .
importantly , the slope of the edge bands is proportional to the kane - mele spin - orbit coupling , which controls thereby the density of states at the fermi energy .
the interplay of spin - orbit and coulomb repulsion on the otherwise non - dispersive edge states leads to the strong magnetic anisotropy effects anticipated above . and
@xmath1 , scaledwidth=50.0% ] to model this kind of systems , we use the so called kane - mele - hubbard hamiltonian@xcite , which provides a minimal model to study the effect of the coulomb interactions on the topologically protected edge states : @xmath2 where @xmath3 are the spin projections of the spin along the axis perpendicular to the to the two dimensional crystal , @xmath4 stands for first neighbor and @xmath5 for second , @xmath6 for clockwise or anti - clockwise second neighbor hopping @xcite . for simplicity
, we neglect the rashba coupling,@xcite . in the case of planar honeycomb systems , such as graphene ,
the rashba term is null .
for buckled group iv crystals , such as silicene , germanene and stanene , the magnitude of the rashba is one order of magnitude lower than the pure spin - orbit.@xcite the hubbard term reads : @xmath7 where @xmath8 denotes the occupation operator of site @xmath9 with spin @xmath10 along an arbitrary quantization axis .
we treat the hubbard interaction in the collinear mean field approximation , enforcing the magnetization to lie along the axis @xmath11 , that we take as the quantization axis ( see fig.[f2]a ) .
this approach permits to study solutions with different @xmath12 and compare their properties .
rotations in the @xmath13 plane leave the results invariant , due to the symmetry of the kane - mele spin - orbit coupling . in general , the coulomb interaction term evaluated in the mean field approximation leads to two self - consistent potentials terms , direct and exchange . in the case of the hubbard model in the collinear approximation ,
only the direct term survives : @xmath14\ ] ] where the notation explicitly shows the spin quantization axis is taken along @xmath15 and @xmath16 stand for the average of the occupation operator calculated within the ground state of the mean field hamiltonian : @xmath17 as usual , this defines a self - consistent problem that we solve by iteration . because of the spin - orbit kane - mele term in @xmath18 , mean field solutions with different @xmath12 are not equivalent .
notice as well that the @xmath19 term is non - diagonal when represented in the basis of eigenstates of @xmath20 and @xmath21@xcite .
we pay special attention to the atomic magnetization , along the @xmath15 in site @xmath9 : @xmath22}{2}\ ] ] and we take @xmath23 . in order to study the zigzag edges it is convenient to study ribbons , that define a one dimensional crystal ( see fig .
1 ) with two edges . a given unit cell of the one dimensional crystal is formed by @xmath24 units of 4 atoms . in the following
we characterize the width of the ribbons by @xmath24 . for finite @xmath25 , and as long as @xmath26 is not too large , we find solutions with ferromagnetic order at the edges .
the magnetic moment calculated self - consistently is non - negligible only at the edge atoms . attending to their mutual magnetization orientation ,
ribbons yield two types of solutions with ferromagnetic edges : parallel ( fm ) and antiparallel ( af ) . for sufficiently wide ribbons
the inter - edge coupling is negligible and both solutions have identical properties .
the first important result of the paper is shown in figure 1 .
whereas off - plane magnetization @xmath27 leads to a conducting solution , found in previous works@xcite , the in - plane magnetization opens a gap .
therefore , transport properties of zigzag edges will change dramatically upon rotation of the magnetization direction , in contrast with conventional metallic ferromagnets .
this metal - insulator transistion will is developed as well in chiral edge ribbons @xcite , which have been widely reported @xcite .
the second important result of the manuscript is shown in fig.[f2]b .
the ground state energy @xmath28 is minimal for @xmath29 , i.e. , for in - plane magnetization , which means that spontaneous magnetic order in this system leads to insulating behavior . .
( a ) scheme of the edge magnetization for two different angles @xmath30 and @xmath12 .
ground state total energy ( per unit cell , with 2 magnetic atoms per cell ) ( b ) , gap(c ) and magnetization ( d ) as a function of @xmath12 .
( e ) and ( f ) evolution of the band structure for different values of @xmath12 for the fm ( e ) and af(f ) configurations .
@xmath31 , u = t , @xmath1.,scaledwidth=50.0% ] the results of figure 1 can be understood as follows . in the absence of magnetic order ,
two spin - filtered in - gap edge states with opposite velocities exist at each edge@xcite , resulting in a two - fold degeneracy ( not shown ) .
ferromagnetic order with @xmath32 breaks time reversal symmetry but does not mix spins .
thus , magnetic order merely yields a spin - dependent shift that breaks the two - fold degeneracy , as seen in fig.[f1]a , for the fm case . for the af configuration
, there is an extra symmetry that restores the double degeneracy : the combined action of spatial inversion , that results in a exchange of the atoms of the two interpenetrating triangular sub lattices @xmath33 and @xmath34 that form the honeycomb , and time reversal , that exchange @xmath10 and @xmath35 , leave the system invariant .
thus the spin @xmath10 band localized at the at the @xmath33 type edge is degenerate with the spin @xmath35 band localized in the opposite edge .
the situation is radically different when the magnetization lies in plane .
representing the self - content potential in the basis of the @xmath36 spin filtered edge states , with spin quantized along the @xmath37 axis , the effect of the in - plane magnetization is to mix bands with opposite spins . as a result ,
a band gap opens at the @xmath38 point where the non - interacting edge bands cross .
the evolution of the bands as the magnetization is rotated from almost off - plane ( left ) to almost in - plane ( right ) is shown if fig .
[ f2])(e , f ) .
it is apparent that the band gap ( fig .
[ f2])(c ) ) is maximal for in - plane magnetization ( @xmath29 ) and null for off - plane @xmath39 .
the preference for in - plane magnetization can also be connected with the variation of the magnitude of the edge magnetic moment with @xmath12 ( fig.[f2]d ) .
these two results naturally explain the fact that the ground state energy is minimal for in - plane magnetization . at half - filling ,
all the valence bands are occupied and the conduction bands are empty .
therefore , increasing the band - gap decreases the total energy .
the gap opening as long as magnetization is not off - plane will certainly have dramatic consequences on the transport properties along the edges .
a result similar to this has been obtained recently@xcite , using a kane - mele model where magnetic order is externally driven , and modeled by a magnetic exchange potential that arises from proximity rather than spontaneously , as discussed here .
the results of figures 1 and 2 are for a specific choice of @xmath40 and @xmath41 , and for a ribbon with @xmath31 sites , wide enough to decouple the two edges .
we now discuss how the results depend , quantitatively , on the specific values of the spin - orbit coupling , ribbon width and @xmath25 .
the evolution of several energy differences between af / fm and in - plane off - plane configurations , as a function of the ribbon width @xmath24 is shown in fig.[f3]a . for large @xmath24
it is apparent that fm and af have the same ground state and anisotropy energy .
in addition , the edge gap ( [ f3]b ) also becomes independent on the magnetic configuration at large width .
.energy scales for different graphene - like honeycomb materials . [
cols="<,^,^,^",options="header " , ] [ table ] in figure ( fig.[f3](c)(e ) ) ) we plot the dependence of the magnetic anisotropy and magnetic moment ( both in and off - plane ) on the magnitude of the spin - orbit coupling @xmath42 , for two different values of @xmath25 .
attending to the difference between the magnitude of the magnetic moment in the off - plane and in - plane cases ( fig.[f3](e ) ) , three different regions are found .
for very small @xmath42 the magnetic moment is the same for out of plane and in - plane magnetization and the magnetic anisotropy energy depends quadratically on @xmath42 . from this standpoint ,
the behavior of the zigzag edge is similar to conventional magnets , although a small gap , open for in - plane magnetization . for wide enough ribbons , the value of this gap is given by @xmath43 where @xmath44 is the bulk gap opened by spin - orbit interaction and @xmath45 is the exchange splitting gap , which is a decreasing function of @xmath46 , giving rise to the curve seen in figure [ f3](d ) .
for intermediate values of @xmath42 it is apparent that the magnetic moment magnitude is different for in - plane and off - plane orientations , but in both cases finite . in this region
the anisotropy energy scales approximately linear with @xmath42 , and the band - gap of in - plane magnetization is still a linear function of @xmath42 . finally , above a
given critical value @xmath47 ( @xmath48 ) for @xmath49 ( @xmath50 ) , the system enters in the so - called colossal@xcite magnetic anisotropy regime , for which magnetic order is only possible in - plane , and the magnetic solutions off - plane do not exist .
increasing spin - orbit coupling beyond this point starts to reduce the band - gap and the magnetic order altogether , which leads to a reduction of the magnetic anisotropy energy ( fig.[f3](e ) ) we thereby expect that graphene , silicene and germanene are in the small @xmath42 region , mof is in the intermediate region and the stanene zigzag edge could show the colossal magnetic anisotropy effect .
notice that in the intermediate region the magnetic anisotropy energy per magnetic atom can be extremely large .
for instance , for stanene , taking @xmath50 , intermediate @xmath51mev and @xmath52 , we obtain of @xmath53mev , significantly larger than record materials such as @xmath54@xcite .
na@xmath55iro@xmath56 and related systems@xcite offer also a fascinating possibility of real tuning of the effective @xmath46 by strain,@xcite , which would make it possible to build devices with strain - tunable anisotropy .
given the spread of estimates of the actual values of @xmath25 for a given material , as well as the fact that it different substrates can result in different values of @xmath25 , we address the question of how the results above depend on the strength of the on - site coulomb repulsion @xmath25 . at finite @xmath46
there is a critical @xmath57 below which the edges are non - magnetic@xcite , and a second @xmath58 above which the entire honeycomb lattice becomes antiferromagnetic@xcite . for @xmath59
only the edge is magnetic , and its magnetic anisotropy energy is a non - monotonic function of @xmath25 .
it increases first , reflecting the increase of the magnetic moment , and then it decreases slightly , reflecting the reduction of the ratio @xmath26 . as @xmath25 approaches @xmath60 the magnetic anisotropy overshoots because the bulk becomes magnetic as well . .
( b ) gap , for in - plane magnetization solution , for fm and af solutions , as a function of @xmath24 .
( c , d , e ) evolution with the strength of the spin - orbit coupling @xmath46 : anisotropy energy ( c ) , gap ( d ) and edge magnetization ( e ) .
( f ) evolution of the anisotropy energy with the on - site hubbard interaction . in ( a - b ) @xmath50 and @xmath61 . in ( c - d ) @xmath31 .
, scaledwidth=50.0% ] we now discuss the physical effects not covered within our two main approximations , namely , treating the interactions at the mean field level and ignoring the rashba spin - orbit term . in one dimension ,
collective spin fluctuations are expected to destroy the infinitely long - range order described by mean field theory .
still , for ribbons shorter than the spin correlation length @xmath62 the mean field theory provides a fair description , very much like density functional theory describes properly the magnetization of clusters and nanomagnets .
the spin correlation length @xmath62 in graphene edges , calculated within the spin wave approximation and ignoring spin - orbit coupling@xcite , is @xmath63 for @xmath64 .
the magnetic anisotropy barrier to rotate the spins out of plane for the approximately 15 tin atoms of a zigzag stanene edge that long would be @xmath6560 mev .
inclusion of the rashba coupling would have two consequences .
first , lack of inversion symmetry would split the bands in the case of af configurations .
second , it would break the in - plane @xmath13 magnetic symmetry at the edges . in conclusion
, we have studied the magnetic anisotropy of the ferromagnetic phase of the zigzag edges of graphene and graphene - like systems , that can be described with a single orbital hubbard model model on a honeycomb lattice with spin - orbit coupling described with the kane - mele hamiltonian .
this includes a large class of two dimensional crystals , such as silicene@xcite , germanene@xcite , stanene@xcite , iridates@xcite and metal organic frameworks@xcite .
since the electronic dispersion of the non - interacting edge states is fully determined by the spin - orbit coupling , the resulting magnetic anisotropy effects , computed within a mean field approximation , turn out to be very strong : the system undergoes a metal to insulator transition when the magnetization is rotated out of the normal and for large values of @xmath46 the magnetic solutions are only stable for in - plane magnetization . for all values of the spin - orbit interaction
we find that the ground state energy occurs for in - plane magnetization and the edge states are gapped .
jfr acknowledges financial supported by mec - spain ( fis2010 - 21883-c02 - 01 ) and generalitat valenciana ( acomp/2010/070 ) , prometeo .
this work has been financially supported in part by feder funds .
we acknowledge financial support by marie - curie - itn 607904-spinograph .
30 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , _ _ ( , , ) .
, , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , ,
* * , ( ) . , , , * * , ( ) .
, , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) , issn , . , , ,
, * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
see supplemental material at [ url ] which includes ref.@xcite for details of the collineal mean field antsaz and the anisotropy effect in chiral edges .
d. gonsalbez - martinez , d. soriano , j. j. palacios , j. fernandez - rossier , solid state .
communications 151 , 10751083 , ( 2011 ) , , , , , , , , , , , * * , ( ) .
, , , , , , , , , , , * * , ( ) . , * * , ( ) . , , , , ( ) , . , , , * * , ( ) , . , , ,
* * , ( ) . ,
( ) , . , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . * supplemental material * in this supplemental material we present the mean field operators needed to perform an angle dependent collineal mean field hubbard calculation , as well as the angle dependent ground states of chiral hubbard kane - mele honeycomb ribbons
|
the kondo effect @xcite in samples with reduced dimensions ( thin films , narrow wires ) is one of the most challenging problems in the field .
most of the experiments @xcite have shown that the kondo contribution to the resistivity is suppressed when the sample size is reduced or the disorder in the sample is increased .
in addition the different thermopowers of samples with different thickness gave further evidences for the existence of size dependence @xcite .
the previously examined samples were au(fe ) , cu(fe ) , and cu(cr ) alloys , i.e. , alloys with integer spin impurities .
surprisingly , however , very weak size dependence has been found recently in cu(mn ) alloys @xcite . the first possible explanation related to the size of the kondo screening cloud @xcite
was ruled out both theoretically @xcite and experimentally @xcite . in the limit of strong disorder the experiments
might be well explained with the theory of phillips and co - workers based on weak localization @xcite . in the dilute limit
the theory of the spin - orbit - induced magnetic anisotropy proposed by the authors @xcite was able to explain every experiment in samples with reduced dimensions , small disorder and integer spin impurities for thin layers @xcite .
recently an elegant method was developed by fomin and co - workers @xcite which can be applied for a general geometry . according to this theory
@xcite the spin - orbit interaction of the conduction electrons on the non - magnetic host atoms can result in a magnetic anisotropy for the magnetic impurity .
this anisotropy can be described by the hamiltonian @xmath3 where @xmath4 is the normal direction of the experienced surface element , @xmath5 is the spin of the impurity , and @xmath6 is the anisotropy constant which is always positive and inversely proportional to the distance of the impurity from the surface . due to this
anisotropy the spin multiplet splits according to the value of @xmath7 . in the case of integer spin ( e.g. , @xmath1 for fe ) ,
the lowest level is a singlet with @xmath8 thus at a given temperature the impurities close enough to the surface , where the splitting is greater than @xmath9 , can not contribute to the kondo resistivity .
when the sample size becomes smaller , more and more impurity spins freeze out , reducing the amplitude of the kondo resistivity @xcite .
the theory predicts , however , different behavior for samples with half - integer impurity spin ( e.g. , @xmath0 for mn ) which is recently in the center of interest . in the one - half case when the anisotropy looses its meaning @xcite , in fact
no anomalous size dependence has been found for ce impurities by roth and co - workers @xcite . in half - integer @xmath10 case
the lowest level is a doublet ( @xmath11 ) thus even an impurity close to the surface ( large anisotropy ) has a contribution to the kondo resistivity . even accepting that the surface anisotropy reduces the free spin of the manganese to a doublet , it is far from trivial that in this case the size dependence is drastically suppressed , therefore an elaborate theory is required . in this paper
we calculate the kondo resistivity in thin films of magnetic alloys with @xmath0 impurity with helps of the simple model and mrg calculation of ref .
fitting to the kondo resistivity the @xmath12 function , we calculate the coefficient @xmath13 in terms of the film thickness which is quite different from the @xmath1 case .
these results will be compared to the recent experiment by jacobs and giordano @xcite presented in thin cu(mn ) films and we compare also the case of alloys with different kondo temperatures .
in the presence of the spin - orbit - induced surface anisotropy the kondo hamiltonian is @xmath14 where @xmath15 ( @xmath16 ) creates ( annihilates ) a conduction electron with momentum @xmath17 , spin @xmath18 and energy @xmath19 measured from the fermi level . the conduction electron band is taken with constant energy density @xmath20 for one spin direction , with a sharp and symmetric bandwidth cutoff @xmath21 .
@xmath22 stands for the pauli matrices , @xmath23 s are the effective kondo couplings and @xmath24 is the anisotropy hamiltonian when the quantization axis is parallel to @xmath4 . applying the callan - symantzik multiplicative renormalization group ( mrg ) method to the problem , the next to leading logarithmic scaling equations for the dimensionless couplings @xmath25 were calculated for any impurity spin in ref .
there was a multiple step scaling performed , corresponding to the freezing out of different intermediate states due to the surface anisotropy .
after exploiting the @xmath26 symmetries of the problem , the scaling equations ( see eq . ( 20 ) and ( 21 ) in ref .
@xcite ) were solved numerically in terms of the scaling parameter @xmath27 .
the results for @xmath0 , @xmath28 , and @xmath29k , i.e. , @xmath30k for cu(mn ) , can be seen in fig .
we can see from the plot that , when @xmath31 is large enough , at @xmath32 every coupling can be neglected , except the @xmath33 and @xmath34 . this corresponds to the freezing out of the given higher @xmath7 states , but it can be seen well that the two lowest energy states are still significant even for large anisotropy . the kondo resistivity calculated from the running couplings at @xmath32 by solving the boltzmann equations ( see ref .
@xcite ) is @xmath35 where @xmath36 is the impurity concentration , @xmath37 is the fermi energy , @xmath38 is the electron distribution function in the absence of an electric field , and @xmath39 is a function of the running couplings at @xmath32 , spin factors , the strength of the anisotropy @xmath40 , and the temperature , defined in ref . @xcite . for thin films the kondo resistivity calculated in the frame of a simple model where the two surfaces contribute to the anisotropy constant in an additive way as
@xmath41 is @xmath42 where @xmath43 is the proportionality factor of the spin - orbit - induced surface anisotropy , @xmath44 is the thickness of the film and we have used the fact that the kondo contribution to the resistivity is smaller by a factor of @xmath45 than the residual normal impurity contribution ( see for the details ref .
@xcite ) . in fig .
[ fig2 ] the resistivity as a function of @xmath46 can be seen for different @xmath47 for @xmath0 , @xmath28 , and @xmath29k , i.e. , @xmath30k as for cu(mn ) . thus reducing the thickness of the film for given @xmath43 , the kondo contribution to the resistivity
is reduced comparing to the bulk value , but for given thickness there is a temperature below that the reduction turns into an increase . fitting the logarithmic function @xmath48 to the kondo resistivity
, the plots of the coefficient @xmath49 as a function of the thickness can be seen in fig .
here we can better see this very different behavior from the @xmath1 case ( cf .
first of all , the kondo amplitude is reduced comparing to the bulk value , but the dependence on the thickness is much weaker than for @xmath1 .
secondly , for small @xmath47 s the coefficient does not go to zero as for @xmath1 , but has a minima at @xmath50 and then changes sign , begins to increase .
this corresponds to that for an @xmath0 impurity in the presence of the anisotropy , the lowest energy states are the @xmath51 doublet which give a contribution to the resistivity even for large anisotropy ( small distance from the surface ) and which can also be larger than the bulk value as a consequence of the spin factors in the scaling equations . because of the large domain of the microscopic estimation of the anisotropy constant ( @xmath52 ) , we can not give a precise microscopic prediction for the place of the minima . according to the the fits on the experimental results on au(fe ) @xcite and cu(fe ) @xcite , @xmath53
from which we can obtain a rough estimation for the place of the minima as @xmath54 .
however , the theoretical calculation is not reliable in that region where @xmath33 is already in the strong coupling limit .
thus the minima may be a sign of the breakdown of the weak coupling calculation .
we have fitted in the @xmath55 k temperature regime , thus below @xmath56k where the electron - phonon interaction can still be neglected @xcite and well above the kondo temperature where the weak coupling approximation is justified , thus our perturbative calculation and the logarithmic function for the kondo resistivity is valid .
these results are in good agreement with the recent experiments of jacobs and giordano @xcite on thin films of cu(mn ) . in fig .
[ fig4 ] we have examined what like is the function @xmath57 for different kondo temperatures corresponding to different @xmath58-spin alloys .
it can be seen from the figure that a minima in the @xmath59 function for small @xmath47 s is present at pretty much the same place ( i.e. , @xmath60 ) .
the character of the @xmath13 coefficient comparing to @xmath61 would roughly the same , but the absolute measure of the thickness dependence and the reduction from the bulk value become larger for larger @xmath2 when we fit in the same temperature regime corresponding to the larger bulk value ( see the figure caption of fig . [ fig4 ] ) .
in this paper we have reexamined our previous calculation @xcite about the kondo resistivity in thin films of alloys with @xmath0 impurities in the presence of spin - orbit - induced surface anisotropy .
first we presented our results on kondo resistivity for @xmath0 and @xmath30k , i.e. , for cu(mn ) , and fitted the @xmath48 function on it .
the @xmath13 function in terms of the film thickness is plotted in fig .
[ fig3 ] from which we can see that there is a reduction in resistivity comparing to the bulk value , but the @xmath13 coefficient depends much weaker on the thickness than in case of @xmath1 , and for small @xmath47 s @xmath13 does not go to zero , but it has a minima at ca .
@xmath62 below that it increases which may be only a sign of the breakdown of the weak coupling calculation .
these results are in good agreement with the recent experiment on cu(mn ) films @xcite .
we do not get , however , their factor of 30 difference from the bulk value .
then we have examined what like is the @xmath57 coefficient for different kondo temperatures , i.e. , for different @xmath58-spin alloys .
we have find that the character of the @xmath57 function is the same , but the reduction from the bulk value and the thickness dependence is somewhat stronger for larger @xmath2 .
summarizing , the level structure of the impurity is very different for integer and half - integer spin and at the surface the spin has a degenerate ground state in the latter case , but we obtained an essentially suppressed size dependence which is far from being trivial .
the actually dependence is an interplay between different effects as the strength of the anisotropy and the different amplitudes and temperature dependence of coupling strength of the electron induced transitions between levels .
the drastically different behavior for the integer and half - integer cases found experimentally and determined theoretically provides a further support of the surface anisotropy as the origin of the size dependence .
we would like to thank n. giordano for helpful discussions .
this work was supported by grant otka @xmath63 and @xmath64 .
one of us ( a.z . ) benefited from the hospitality of the meissner institute and the lmu in munich and supported by the humboldt foundation .
99 j. kondo , prog .
* 32 * , 37 ( 1964 ) .
see for a review m. a. blachly and n. giordano , phys .
* b * * 51 * , 12537 ( 1995 ) ; j. f. ditusa , k. lin , m. park , m. s. isaacson , and j. m. parpia , phys . rev . lett .
* 68 * , 678 ( 1992 ) ; g. neuttiens , j. eom , c. strunk , v. chandrasekhar , c. van haesendonck , and y. bruynserade , europhys .
* 34 * , 623 ( 1996 ) . c. strunk , m. henny , and c. schnenberger , g. neuttiens and c. van haesendonck , phys .
* 81 * , 2982 ( 1998 ) . t. m. jacobs and n. giordano , europhys .
lett . * 44 * , 74 ( 1998 ) .
g. bergmann , phys .
* 67 * , 2545 ( 1991 ) .
v. barzykin and i. affleck , phys .
lett . * 76 * , 4959 ( 1996 ) .
m. a. blachly and n. giordano , phys . rev .
* b * * 49 * , 6788 ( 1994 ) .
i. martin , yi wan , and philip phillips , phys .
lett . * 78 * , 114 ( 1997 ) .
o. jsghy , a. zawadowski , and b. l. gyorffy , phys .
lett . * 76 * , 2378 ( 1996 ) ; o. jsghy and a. zawadowski , phys . rev . * b * * 57 * , 11598 ( 1998 ) . o. jsghy and a. zawadowski , phys . rev .
* b * * 57 * , 11609 ( 1998 ) .
l. borda and a. zawadowski , cond - mat/9903054 , submitted to phys . rev . * b*. v. m. fomin , v. n. gladilin , j. t. devreese , c. van haesendonck , and g. neuttiens , solid state commun . * 106 * , 293 ( 1998 ) .
c. roth , c. srgersand , and h. v. lhneysen , phys . rev .
* b * * 54 * , 534 ( 1996 ) .
o. jsghy and a. zawadowski , unpublished .
|
recently , for spin @xmath0 impurities quite different size dependence of the kondo contribution to the resistivity was found experimentally than for @xmath1 . therefore previous calculation about the effect of the spin - orbit - induced magnetic anisotropy on the kondo amplitude of the resistivity
is extended to the case of @xmath0 impurity spin which differs from the integer spin case as the ground state is degenerated . in this case
the kondo contribution remains finite when the sample size goes to zero and the thickness dependence in the kondo resistivity is much weaker for cu(mn ) .
the behavior of the kondo coefficient as a function of the thickness depends on the kondo temperature , that is somewhat stronger for larger @xmath2 . comparing our results with a recent experiment in thin cu(mn ) films , we find a good agreement .
|
we present a rigorous mathematical examination of the problem of josephson vortices and of the meissner effect in thin - layer josephson - junction stacks and layered superconductors , with a static external magnetic field @xmath13 applied parallel to the layers ( along the @xmath14 axis , see fig . 1 . )
we consider periodic systems composed of an arbitrary number @xmath0 of identical superconductor / insulator ( s / i ) junctions ( @xmath15 , where @xmath2 is the number of s - layers , with @xmath16 being the layering axis ) .
our starting point is the microscopic gibbs free - energy functional derived in ref . .
mathematical structure of this functional is analogous to that of the phenomenological lawrence - doniach model .
@xcite thus , the treatment of our paper fully applies to the latter model as well .
mathematically , both josephson vortices in moderate fields @xmath17 and the inhomogeneous meissner state are described by solutions of a system of nonlinear second - order differential equations for phase differences , @xmath18 , with square - integrable first - order derivatives.@xcite ( for brevity , we call here such solutions localized ) .
our approach is substantially based on the observation of a nontrivial property of the differential equations for @xmath18 : we show that the problem of finding localized solutions for @xmath18 can be reduced to solving a standard initial value problem .
using this key mathematical result , we establish an exact criterion of the existence of localized solutions .
the existence criterion , in turn , allows us to obtain a complete classification of physical localized solutions .
we have found three types of such solutions : meissner solutions , topological vortex - plane solutions,@xcite and topological incoherent vortex solutions .
meissner solutions are localized near the side boundaries @xmath19 and @xmath20 .
in contrast to the well - known single - junction case,@xcite the meissner solutions in stacks with @xmath21 turn out to be characterized by several josephson lengths @xmath3 ( @xmath4 lengths for even @xmath2 , and @xmath22 lengths for odd @xmath2 ) .
the meissner solutions persist up to a certain superheating field of the meissner phase , @xmath6 .
we derive an exact expression for @xmath6 as an explicit function of @xmath2 .
we show that the field @xmath6 simultaneously determines the penetration field for vortex planes .
( see below . )
thermodynamically stable vortex - plane
solutions represent a chain of @xmath23 josephson vortices ( one vortex per each i - layer ) , positioned in the symmetry plane @xmath24 .
these solutions are uniquely determined by the vortex - penetration conditions at @xmath25 .
they are characterized by the same set of @xmath3 as the meissner solutions .
such solutions were previously obtained for infinite ( @xmath26 ) layered superconductors.@xcite under the name of the coherent mode or the in - phase mode
they are well - known in double - junction stacks.@xcite for @xmath27 , the existence of the coherent mode was predicted in ref . .
( the authors of ref . specially emphasized the importance of this mode for practical applications . ) besides giving a proof of the existence and stability of the vortex planes in the general case @xmath15 , we derive exact analytical expressions for their self - energy @xmath28 and for the lower critical field @xmath10 .
incoherent vortex solutions comprise single - vortex solutions , vortex solutions with @xmath29 vortices in the plane @xmath30 as well as different vortex - antivortex configurations .
all such solutions satisfy the existence criterion . however , in contrast to the vortex - plane solutions , they prove to be thermodynamically unstable and do not meet the vortex - penetration conditions at any @xmath31 .
it should be noted that the single - vortex solutions obtained in this paper have no resemblance to hypothetical abrikosov - type vortices , introduced without proper mathematical justification in some previous publications.@xcite besides being thermodynamically unstable , the actual single - vortex solutions are not uniquely determined by asymptotic boundary conditions .
they are accompanied by singular phase - difference distribution in all @xmath0 junctions , and their spatial dependence is characterized , in general , by @xmath0 length scales .
section ii of the paper is devoted to exact mathematical formulation of the problem . in section iii , we derive all major physical and mathematical results sketched above .
the general consideration of this section is illustrated by several concrete examples in section iv .
in particular , we analyze 1 - 4-junction stacks and investigate a transition to the layered - superconductor limit ( @xmath12 ) .
the obtained results are discussed in section v. appendices a - c contain some additional mathematics , relevant to the subject of our study .
we begin by writing down the microscopic gibbs free - energy functional@xcite of a periodic structure consisting of alternating @xmath2 superconducting ( s ) and @xmath0 insulating ( i ) layers ( @xmath15 ) : @xmath32 = % \frac{h_{c}^{2}(t)}{4\pi } aw_{z}\left [ \sum_{n=0}^{n-1}\int\limits_{-l}^{l}dy% \left [ -f_{n}^{2}(y)+\frac{1}{2}f_{n}^{4}(y)\right . \right.\ ] ] @xmath33 ^{2}f_{n}^{2}(y)% \right]\ ] ] @xmath34 \right ] \right.\ ] ] @xmath35 ^{2}\right ] \right ] , \label{1.1}\ ] ] @xmath36 @xmath37 @xmath38 here @xmath39 ; @xmath40 is the s - layer thickness ; @xmath41 is the period , and @xmath42 is the length of the structure in the @xmath14 direction ( @xmath43 ) ; the length of the structure in the @xmath44 direction is @xmath45 ; @xmath46 [ @xmath47 and @xmath48 are , respectively , the reduced modulus and the phase of the pair potential @xmath49 in the @xmath50th superconducting layer : @xmath51 with @xmath52 being the microscopic gap at temperature @xmath53 ; @xmath54 is the bcs coherence length ; @xmath55 and @xmath56 are , respectively , the ginzburg - landau ( gl ) coherence length and the penetration depth ; @xmath57 is the incidence - angle - dependent tunneling probability of the i - layer between two successive s - layers ; @xmath58 is the thermodynamic critical field ; @xmath59 is the vector potential . the local magnetic field @xmath60 $ ] obeys the maxwell equation @xmath61 with boundary conditions @xmath62 , \ ] ] @xmath63 , \ ] ] where @xmath64 is a static external magnetic field applied along the @xmath14 axis .
( see fig . 1 . )
the sum of the first three phase- and field - independent terms on the right - hand side of ( [ 1.1 ] ) represents the condensation energy .
the fourth term is the kinetic energy of the intralayer currents .
the last two terms are the josephson energy and the field energy , respectively .
expression ( [ 1.1 ] ) is valid under the conditions @xmath65 @xmath66 @xmath67 @xmath68 conditions ( [ 1.9 ] ) ( @xmath69 is the critical temperature of an isolated s - layer ) and ( [ 1.10 ] ) ensure the applicability of the gl - type expansion within each s - layer .
condition ( [ 1.11 ] ) corresponds to the thin s - layer limit , whereas condition ( [ 1.12 ] ) is employed here for the sake of mathematical simplicity only .
being a first - order expansion in @xmath70 , equation ( [ 1.1 ] ) applies in fields @xmath71 , where @xmath72 is the upper critical field .
mathematical treatment of the functionals of the type ( [ 1.1 ] ) is described in full detail in ref . , section iii : one minimizes ( [ 1.1 ] ) with respect to @xmath73 and @xmath74 , @xmath75 , imposes the gauge @xmath76 and eliminates @xmath75 by integration .
the result is a closed , complete set of coupled nonlinear mean - field equations for the reduced modulus of the pair potential @xmath73 and the phase differences @xmath18 , together with relations for all physical quantities of interest . to simplify mathematical analysis of the mean - field equations , we introduce dimensionless units by @xmath77 @xmath78 @xmath79 where the quantities on the left - hand side are dimensional , with @xmath80 being the josephson penetration depth
( @xmath81 is the density of the josephson current in a single junction with thick electrodes ) and @xmath82 being the superheating ( penetration ) field of the infinite layered superconductor.@xcite in our dimensionless units , for example , the flux quantum is @xmath83 , and the lower critical field of the infinite layered superconductor@xcite is @xmath84 . in the dimensionless form ,
the mean - field equations for @xmath73 and @xmath18 read @xmath85 ^2}{\epsilon ^2f_0 ^ 3(y)}+\frac 12\left [ f_0(y)-f_1(y)\cos \phi _ 1(y)\right ] \right ] , \ ] ] @xmath86 ^2}{\epsilon ^2f_n^3(y)}\right.\ ] ] @xmath87 \right ] , \quad 1\leq n\leq n-2,\ ] ] @xmath88 ^2}{\epsilon ^2f_{n-1}^3(y)}\right.\ ] ] @xmath89 % \right ] ; \ ] ] @xmath90 @xmath91 -\frac 1{f_{n-1}^2(y)}\left [ h_n(y)-h_{n-1}(y)\right ] -\epsilon ^2h_n(y)\ ] ] @xmath92 @xmath93 @xmath94 where @xmath95 and the local magnetic field in the @xmath50th insulating layer ( @xmath96 ) is given by @xmath97 @xmath98 note that although in most physical situations @xmath99 , for the sake of mathematical generality , in this paper , @xmath100 is supposed to satisfy only the weak inequality ( [ 1.21 ] ) . because of the obvious property @xmath101 , equation ( [ 1.22 ] ) implies that the phase differences @xmath18 meet the condition @xmath102 the dimensionless gibbs free energy @xmath103 , normalized via the relation @xmath104 in terms of the mean - field quantities @xmath73 , @xmath18 and @xmath105 has the form @xmath106
@xmath107 ^2 + 2r(t)\left [ h_n(y)-h\right ] ^2\right]\ ] ] @xmath108 .\ ] ] here , the two terms in the second line on the right - hand side are the kinetic energy of the intralayer currents and the field energy , respectively .
the intralayer current in the @xmath50th s - layer @xmath109 ( normalized to @xmath110 ) and the density of the josephson current between the @xmath50th and the @xmath111th s - layers @xmath112 ( normalized to @xmath113 ) are given by @xmath114 , \quad 0\leq n\leq n-1,\ ] ] and @xmath115 respectively .
relations ( [ 1.16])-([1.26 ] ) provide a complete , self - consistent description of the thin - layer periodic s / i structure in the temperature range ( [ 1.9 ] ) and in fields @xmath71 . in this paper
, we will be interested in physical solutions for @xmath18 with a square - integrable first - order derivative , localized within a spatial range of order unity .
( for brevity , we call such solutions localized . ) therefore we assume the condition @xmath116 moreover , we assume that the temperature range satisfies the condition of the weak - coupling limit @xmath117 one can obtain a perturbative solution for @xmath73 and @xmath18 up to any desired order in @xmath118 , starting from the zero - order solution to ( [ 1.16 ] ) , ( [ 1.17 ] ) , @xmath119 and the zero - order equations for @xmath18 , @xmath120 where @xmath105 are given by ( [ 1.22 ] ) with @xmath121 and satisfy the boundary conditions ( [ 1.19 ] ) . for most applications , it is sufficient to consider expressions for physical quantities only in leading order in @xmath118
. thus , for example , substituting ( [ 1.29 ] ) and the solution of ( [ 1.30 ] ) into ( [ 1.24 ] ) immediately yields a first - order expansion for the gibbs free energy , because first - order corrections to the condensation - energy term cancel out .
a detailed mathematical analysis of eqs . ( [ 1.30 ] ) is the subject of section iii .
here we point out that these equations can be transformed into a very useful for application form by solving for @xmath105 ( see appendix a for mathematical details ) : @xmath122 @xmath123 @xmath124 where @xmath125 are given by ( [ b.9 ] ) , and @xmath126 is given by ( [ b.5 ] ) . by ( [ b.8 ] ) , and ( [ 1.20 ] ) , ( [ b.13 ] ) , expression ( [ 1.31 ] ) explicitly satisfies boundary conditions ( [ 1.19 ] ) and @xmath127 .
moreover , the @xmath44-independent quantities @xmath128 in ( [ 1.31 ] ) have clear physical meaning : being solutions of ( [ 1.30 ] ) with @xmath129 , they describe distribution of the local magnetic field within i - layers in the homogeneous meissner state ( see section iii of ref . ) . also note that @xmath130 , which is a reflection of the symmetry of the problem .
[ by comparison , in an infinite layered superconductor @xmath131 , and @xmath132 where @xmath133 are defined by ( [ b.15 ] ) . ]
in addition , we point out that equations of the popular phenomenological lawrence - doniach model@xcite also can be reduced to the dimensionless form ( [ 1.16])-([1.20 ] ) , ( [ 1.22])-([1.26 ] ) , with @xmath118 being a phenomenological parameter , @xmath134 , and @xmath135 normalized to @xmath136 .
( see ref . for more details . )
thus , all the consideration of this paper fully applies to the lawrence - doniach model as well .
by differentiation with respect to @xmath44 , integrodifferential equations ( [ 1.30 ] ) reduce to a system of @xmath0 ordinary nonlinear second - order differential equations @xmath137 , \ ] ] @xmath138 , \quad 2\leq n\leq n-2,\ ] ] @xmath139\ ] ] with boundary conditions ( [ 1.20 ] ) . consider eqs .
( [ 1.34 ] ) on the whole axis @xmath140 .
two simple properties of ( [ 1.34 ] ) are quite obvious : if @xmath141 ( @xmath142 ) is a solution , the functions @xmath143 given by @xmath144 and @xmath145 are also solutions . [
the latter is a result of the fact that @xmath44 does not enter explicitly the right - hand side of ( [ 1.34 ] ) . ]
our conclusions about the existence of localized solutions to ( [ 1.34 ] ) will be substantially based on another key property , which we formulate as a lemma : * lemma*. consider an arbitrary interval @xmath146 $ ] and @xmath147 . the initial value problem for eqs .
( [ 1.34 ] ) with arbitrary initial conditions @xmath148 , @xmath149 has a unique solution in the _ whole _ interval @xmath150 .
this solution has continuous derivatives with respect to @xmath44 of arbitrary order and continuously depends on the initial data .
( for the proof of the * lemma * , see appendix b. ) it is worth noting that the existence and uniqueness of a smooth solution to the initial value problem in the _ whole _
interval @xmath150 is rather nontrivial for _ nonlinear _ differential equations : for such equations , theorems of existence and uniqueness are usually valid only locally , in the neighborhood of initial data.@xcite in our case , global character of the solution and its infinite differentiability are ensured by the fact that @xmath151 enter the right - hand side of eqs .
( [ 1.34 ] ) only as arguments of the sine .
note that because of the arbitrariness of the interval @xmath150 , the solution can be uniquely continued onto the whole axis @xmath152@xmath153.@xcite now we will show that the problem of finding localized solutions to ( [ 1.34 ] ) can be reduced to the standard initial value problem . differentiating ( [ 1.31 ] ) with respect to @xmath44 yields
@xmath154 multiplying ( [ 1.37 ] ) by @xmath155 , summing over the layer index @xmath50 with the use of ( [ b.11 ] ) and performing integration , we arrive at the first integral of eqs .
( [ 1.34 ] ) : @xmath156 where @xmath157 is the constant of integration .
now let us choose an arbitrary point @xmath158 $ ] , where @xmath159 is sufficiently large [ see the condition ( [ 1.27 ] ) ] .
we are looking for localized solutions of eqs .
( [ 1.34 ] ) that in the region @xmath160 where @xmath161 is determined by the maximum positive eigenvalue of the symmetric matrix @xmath162 ( see appendix a ) , satisfy the asymptotic conditions @xmath163 @xmath164 for any @xmath165 . by inserting ( [ 1.40 ] ) and ( [ 1.41 ] ) into ( [ 1.38 ] ) , we establish the value of the constant of integration : @xmath166 .
( [ 1.38 ] ) becomes @xmath167 substituting initial values @xmath148 , @xmath168 into ( [ 1.42 ] ) , we obtain the general _ criterion of the existence _ of localized solutions to ( [ 1.34 ] ) : @xmath169 indeed , the * lemma * guarantees the existence , uniqueness and differentiability of a solution for arbitrary @xmath170 and @xmath171 in the whole interval @xmath172 $ ] . owing to our special choice of the constant of integration in ( [ 1.42 ] ) , the solution determined by @xmath173 and @xmath171 obeying ( [ 1.43 ] ) will necessarily satisfy asymptotic conditions ( [ 1.40 ] ) , ( [ 1.41 ] ) in the region ( [ 1.39 ] ) .
moreover , this solution will automatically satisfy the conditions @xmath174 in the region ( [ 1.39 ] ) for any @xmath165 and @xmath175 , which can be verified by repeated differentiation of ( [ 1.37 ] ) and application of ( [ 1.40 ] ) , ( [ 1.41 ] ) .
below , we apply the criterion ( [ 1.43 ] ) to concrete physical situations .
let @xmath176 for definiteness .
consider an intermediate length scale @xmath177 such that @xmath178 ( see fig . 1 . )
the meissner boundary value problem is specified by the boundary conditions ( [ 1.20 ] ) and @xmath179 for any @xmath165 . for @xmath180 , we write using ( [ 1.22 ] ) : @xmath181 according to ( [ 1.31 ] ) , we make the identification @xmath182 @xmath183 where @xmath184 stands for the field penetrating through the @xmath19 interface . using the fact that @xmath185 , and @xmath130 , we establish the following important properties : @xmath186 @xmath187 relations ( [ 1.51 ] ) are a result of the symmetry of the problem .
they imply that the number of independent equations describing the meissner solution is @xmath4 for even @xmath2 and @xmath22 for odd @xmath2 .
the solution for @xmath141 in the region @xmath188 can be obtained from the solution in the region @xmath180 using the property @xmath189 resulting from the general relation ( [ 1.23 ] ) . in the region
@xmath190 , @xmath191 , and we have @xmath192 now we apply relation ( [ 1.43 ] ) for @xmath193 . in view of the boundary conditions @xmath194 [ see ( [ 1.20 ] ) ] , we get @xmath195 where @xmath196 ^{-\frac 12}.\ ] ] physical interpretation of the quantity @xmath6 is straightforward .
the maximum value of the left - hand side of ( [ 1.53 ] ) is unity , which corresponds to @xmath197 on the right - hand side .
hence @xmath6 is the maximum external field for which a meissner solution is still possible , i.e. the _ superheating _ field of the meissner state.@xcite moreover , the maximum of the left - hand side is achieved when all @xmath198 , which , by ( [ 1.51.1 ] ) , gives @xmath199 for _ all _ @xmath50 .
the total flux of the penetrating field at @xmath200@xmath201 @xmath202 , \ ] ] is exactly equal to the total flux carried by a
vortex plane ,
i.e. a chain of @xmath0 vortices positioned in the plane @xmath24 .
( [ 1.60 ] ) below . ]
these conditions correspond@xcite to _ simultaneous _ and _ coherent _ penetration of josephson vortices into all the junctions .
therefore @xmath6 can be also regarded as the _ penetration _
field for a vortex plane .
note that @xmath6 for @xmath203 is always higher than the penetration field of an infinite layered superconductor @xmath204 .
@xcite thus , the meissner boundary value problem , eqs .
( [ 1.20 ] ) and ( [ 1.46 ] ) , for @xmath197 reduces to the initial value problem with @xmath205 and @xmath206 ( @xmath142 ) .
let @xmath207 be the corresponding unique solution . by continuous dependence of a solution of the initial value problem on initial data
( see the * lemma * ) , we can argue that the existence of the unique meissner solution @xmath207 at @xmath197 guarantees the existence of a unique meissner solution at any @xmath208 with a certain set of @xmath170 ( @xmath209 ) satisfying ( [ 1.53 ] ) .
in other words , we have proved that the meissner boundary value problem , eqs .
( [ 1.20 ] ) and ( [ 1.46 ] ) , has a unique solution for any @xmath210 .
finally , we want to emphasize one more important property of meissner solutions in ( @xmath0)-junction periodic s / i structures that was not realized in previous publications.@xcite owing to the fact that these solutions satisfy a system of coupled second - order differential equations ( @xmath4 equations for even @xmath211 and @xmath22 equations for odd @xmath2 ) , they are necessarily characterized by _
josephson lengths @xmath3 ( @xmath4 lengths for even @xmath211 and @xmath22 lengths for odd @xmath2 ) , in contrast to a single junction with only one @xmath212 . all @xmath3 are determined by positive eigenvalues of the symmetric matrix @xmath213 .
( see appendix a. ) by a vortex - plane solution we understand a chain of @xmath0 josephson vortices ( one vortex per each i - layer ) positioned at @xmath24 .
analogously , an antivortex - plane solution is a chain of @xmath0 josephson antivortices ( one antivortex per each i - layer ) .
such solutions are characterized by the symmetry
@xmath214 [ see ( [ 1.23 ] ) ] and asymptotic boundary conditions@xcite @xmath215 @xmath216 for _ all _ @xmath165 and _ any _ @xmath217 [ the plus sign in ( [ 1.55 ] ) and ( [ 1.56 ] ) corresponds to a vortex plane in fields @xmath218 , whereas the minus sign corresponds to an antivortex plane in fields @xmath219 . ] the existence and uniqueness of these solutions follows immediately from the * lemma * , the criterion ( [ 1.43 ] ) and the results of the previous subsection .
indeed , by ( [ 1.55 ] ) , a vortex ( antivortex ) plane satisfies the conditions @xmath220@xmath221 for all @xmath50 . under these conditions , the left - hand side of ( [ 1.43 ] )
reaches its maximum .
as shown in the previous subsection , the maximum condition corresponds to the unique choice @xmath222 on the right - hand side of ( [ 1.43 ] ) , consistent with the vortex - penetration conditions ( see the next paragraph ) .
hence the initial values @xmath223 determine a unique localized solution in the region @xmath190 that automatically meets the asymptotic boundary conditions ( [ 1.56 ] ) , ( [ 1.57 ] ) .
thermodynamic stability of vortex - plane ( antivortex - plane ) solutions is ensured by the fact that they satisfy the vortex - penetration conditions for @xmath224 .
let @xmath176 for definiteness .
( in what follows , we consider only vortex planes .
the discussion of antivortex planes is quite analogous . )
a vortex - plane solution can be constructed from the meissner solution @xmath225 in the region @xmath180 , discussed in the previous subsection . indeed , using the properties ( [ 1.35 ] ) , ( [ 1.36 ] ) , we obtain a solution @xmath226 in the region @xmath227 that satisfies the initial conditions @xmath228 , @xmath229 .
this solution can be continued@xcite into the region @xmath230 . by the uniqueness of a solution to the initial value problem ,
the obtained solution coincides with the vortex - plane solution in the interval @xmath190 .
the local magnetic field in the presence of a vortex plane is given by general relations ( [ 1.31 ] ) and ( [ 1.47])-([1.49 ] ) , where @xmath184 [ with @xmath231 should be interpreted as the magnetic field induced by the vortex plane itself . owing to the vortex - plane initial conditions
@xmath232@xmath233 , @xmath234 .\ ] ] ( see fig .
the symmetry relations ( [ 1.51 ] ) apply to vortex - plane solutions too .
thus , the number of independent equations describing a vortex plane is @xmath4 for even @xmath211 and @xmath22 for odd @xmath2 , as in the case of meissner solutions .
consequently , spatial dependence of vortex - plane solutions is characterized by the same set of @xmath3 as that of the meissner solutions .
the flux @xmath235 through the @xmath50th i - layer can be found using ( [ 1.32 ] ) , ( [ 1.56 ] ) and ( [ b.13 ] ) : @xmath236 .\ ] ] the total flux carried by a vortex plane is @xmath237 .\ ] ] note that in contrast to josephson junctions with thick electrodes@xcite and infinite layered superconductors,@xcite the flux carried by a josephson vortex in a finite thin - layer s / i structure _ is not quantized _ and is always smaller than the flux quantum @xmath83 .
( this fact has been already pointed out in ref . . ) to determine the thermodynamic _ lower critical _
field @xmath10 at which the vortex - plane solutions become energetically favorable , we must calculate the difference between the gibbs free energy in the presence of a single vortex plane , @xmath238 , and the gibbs free energy of the homogeneous meissner state , @xmath239 [ the sum of phase - independent terms in ( [ 1.24 ] ) ] . substituting ( [ 1.31])-([1.33 ] ) into ( [ 1.24 ] ) and using ( [ b.7 ] ) , ( [ 1.42 ] ) , in first order in @xmath240 , we obtain : @xmath241 @xmath242 , \ ] ] where the total flux @xmath243 is given by ( [ 1.60 ] ) . the first term on the right - hand side of ( [ 1.62 ] )
should be interpreted as the self - energy of the vortex plane : @xmath244 @xmath245 @xmath246 .\ ] ] note that @xmath28 is exactly twice the energy of the josephson currents in ( [ 1.24 ] ) .
besides , formulas ( [ 1.62 ] ) and ( [ 1.63 ] ) , with corresponding reinterpretation of @xmath238 , @xmath243 and @xmath28 , also hold for incoherent vortex solutions considered in the next subsection .
they also apply to an infinite layered superconductor , taking account of the substitution @xmath247 , where @xmath133 are defined by ( [ b.15 ] ) .
in this latter case , the self - energy should be additionally minimized with respect to the _ phases _ @xmath248 , which immediately yields the exact solution@xcite with @xmath249 and @xmath250 .
[ in the case @xmath203 , the minimization with respect to @xmath251 is not allowed by the boundary conditions ( [ 1.19]).@xcite ] from ( [ 1.62 ] ) , we get : @xmath252 for thermodynamically stable solutions , we must necessarily have @xmath253 .
it is straightforward to verify that this condition is met by the vortex - plane solutions . using ( [ 1.37 ] ) , ( [ 1.55 ] ) ( with the plus sign ) , the initial values @xmath254 and integrating by parts , we convert @xmath28 into the form @xmath255 .\ ] ] the first term on the right - hand side of ( [ 1.65 ] ) is positive , because in the region @xmath227 all @xmath18 satisfy the relation @xmath256 . by the use of ( [ 1.37 ] ) and ( [ b.14 ] ) , we obtain the following strict inequalities : @xmath257 @xmath258 hence , @xmath259 as anticipated .
note that in all special cases admitting exact analytical solutions ( @xmath26,@xcite and @xmath260 , see section iv ) , @xmath261 .
finally , we want to point out that in contrast to vortex - plane solutions in infinite layered superconductors,@xcite where @xmath262 and @xmath263 for all @xmath50 , in finite structures the intralayer currents @xmath264 , in general , are not equal to zero , as can be easily seen from ( [ 1.25 ] ) . only for even number of junctions ( @xmath2 is odd ) , in the central s - layer @xmath265 , by the symmetry ( [ 1.51 ] ) . a single josephson vortex positioned in the @xmath266th i - layer at @xmath24 obeys symmetry relations @xmath267 [ see ( [ 1.23 ] ) ] and asymptotic boundary conditions@xcite @xmath268 @xmath269 @xmath270 moreover , @xmath271 in the region @xmath272 , and @xmath273 in the region @xmath274 .
hence , @xmath18 must satisfy the relations @xmath275 [ see ( [ 1.48 ] ) for @xmath231 and the initial conditions @xmath276 @xmath277 a necessary condition of the existence of such solutions is provided by the general criterion ( [ 1.43 ] ) and has the form @xmath278 relation ( [ 1.75 ] ) imposes only one constraint on @xmath0 quantities @xmath279 and can be satisfied by different sets of @xmath171 .
therefore , in contrast to the vortex - plane problem , a solution to the single - vortex problem for any given vortex position @xmath266 is _ not unique_. the determination of the optimum set of @xmath171 may require an additional examination of the vortex self - energy ( see section iii.b ) . besides failing to meet the requirement of uniqueness , single - vortex solutions , in general , break the symmetry ( [ 1.51 ] ) ,
inherent to the original integrodifferential equations ( [ 1.18 ] ) , ( [ 1.30 ] ) minimizing the gibbs free - energy functional ( [ 1.1 ] ) .
( the only exclusion is a stack with an odd number of junctions @xmath0 and @xmath280 . ) even more important is the fact that single - vortex configurations do not satisfy the vortex - penetration conditions for any @xmath176 . indeed , starting from a single - vortex solution in the region @xmath190 , we can construct a solution in the region @xmath180 satisfying the meissner boundary conditions ( [ 1.46 ] ) .
( compare with the reverse procedure of the construction of a vortex - plane solution from the meissner solution @xmath281 described in the previous subsection .
) however , the solution thus obtained will not represent any physics , because it can not meet the physical boundary conditions @xmath282 for any @xmath176 : according to ( [ 1.74 ] ) , @xmath283 and @xmath171 with @xmath284 have _ different _ signs .
physically , this means that isolated vortices can not penetrate the periodic s / i structure at any static @xmath176 , and the penetration field for _ isolated vortices _ can not be defined . on the basis of these observations ,
we conclude that , in contrast to vortex - plane solutions , single - vortex solutions are _ thermodynamically unstable _ and do not represent any solutions to ( [ 1.18 ] ) , ( [ 1.30 ] ) for @xmath176 .
this situation has a simple mathematical explanation . in contrast to differential equations ( [ 1.34 ] ) , integrodifferential equations ( [ 1.18 ] ) , ( [ 1.30 ] ) explicitly contain the external magnetic field @xmath64 .
[ see the explicit expressions for @xmath105 , eqs .
( [ 1.22 ] ) . ] in the case of eqs . ( [ 1.34 ] ) , the external field @xmath64 enters only via the boundary conditions ( [ 1.20 ] ) at @xmath285 , whereas single - vortex configurations are required to satisfy asymptotic boundary conditions ( [ 1.69])-([1.71 ] ) at @xmath286 .
it is therefore not surprising that eqs .
( [ 1.34 ] ) , restricted to the interval @xmath287 $ ] , may possess redundant solutions that do not satisfy eqs .
( [ 1.18 ] ) , ( [ 1.30 ] ) .
[ note that the exact equations ( [ 1.18 ] ) , valid for arbitrary @xmath118 , do not , in general , reduce to any differential equations , if @xmath288const . ] the flux through the @xmath50th i - layer due to the vortex in the @xmath266th i - layer can be found using ( [ 1.32 ] ) and ( [ 1.69 ] ) , ( [ 1.70 ] ) : @xmath289 .\ ] ] the total flux carried by the vortex in the @xmath266th layer is @xmath290 .\ ] ] for @xmath203 , the total flux @xmath243 is not quantized and is less than the flux quantum @xmath83 .
( see the previous subsection . )
note that the total flux carried by a single vortex in the @xmath266th i - layer is exactly equal to the flux through the @xmath266th i - layer in the case of a vortex plane , given by eq .
( [ 1.59 ] ) with @xmath291 .
the consideration of other localized incoherent vortex solutions to ( [ 1.34 ] ) ( i.e. , solutions with @xmath29 vortices in the plane @xmath24 , and vortex - antivortex configurations ) can be done along the same lines .
all these solutions are thermodynamically unstable too .
we want to underline that our general conclusions about the form of static single - vortex configurations completely agree with the results of numerical calculations of ref . .
( see , in particular , fig .
5 therein . )
on the other hand , these single - vortex configurations have no resemblance to hypothetical abrikosov - type vortices introduced without appropriate mathematical justification in refs . .
let alone thermodynamic instability , the actual single - vortex solutions to ( [ 1.34 ] ) are accompanied by singular phase - difference distribution in _ all _ @xmath0 junctions , satisfying ( [ 1.72])-([1.74 ] ) and the existence condition ( [ 1.75 ] ) .
their spatial dependence is characterized , in general , by @xmath0 length scales , which is inherent to localized solutions of a system of coupled second - order differential equations .
these intrinsic features of topological solutions in discrete periodic s / i structures can not be reproduced by any imitation of abrikosov vortices , typical of continuum type - ii superconductors .
it is also worth reminding that abrikosov vortices in the london approximation are described by _ linear _ partial differential equations,@xcite whereas the ordinary differential equations ( [ 1.34 ] ) are essentially _ nonlinear_. the principle of superposition of solutions is not valid for nonlinear equations .
unfortunately , this basic point is sometimes disregarded in literature.@xcite
in this simplest case , the only nonzero element of the matrix @xmath293 , given by ( [ b.9 ] ) , is @xmath294 by ( [ 1.37 ] ) , a single phase difference @xmath295 satisfies the usual static sine - gordon equation @xmath296 with the josephson length@xcite @xmath297 note that @xmath212 , given by ( [ 4.3 ] ) , for @xmath99 is much smaller than the josephson length of a single junction with thick electrodes , which in our dimensionless units is @xmath298.@xcite from ( [ 1.33 ] ) and ( [ 1.54 ] ) , we get the local field in the homogeneous meissner state @xmath299 and the superheating ( penetration ) field @xmath300 respectively . for @xmath99 , the superheating ( penetration ) field @xmath301 , given by ( [ 4.5 ] ) , is much higher than the corresponding field@xcite of a single junction with thick electrodes @xmath302 . for the fields @xmath303 , the meissner solution in the region @xmath180 up to first order in @xmath304
is given by @xmath305 } { h_s+\sqrt{h_s^2-h^2}},\ ] ] @xmath306 @xmath307 \exp % \left [ -\frac{\left ( l+y\right ) } { \lambda _ j}\right ] } { \left [ h_s+\sqrt{% h_s^2-h^2}\right ] ^2+h^2\exp \left [ -\frac{2\left ( l+y\right ) } { \lambda _ j}% \right ] } , \ ] ] @xmath308\ ] ] @xmath309 ^2-h^2\exp \left [ -% \frac{2\left ( l+y\right ) } { \lambda _ j}\right ] \right ] \exp \left [ - \frac{% \left ( l+y\right ) } { \lambda _ j}\right ] } { \left [ \left [ h_s+\sqrt{h_s^2-h^2}% \right ] ^2+h^2\exp \left [ -\frac{2\left ( l+y\right ) } { \lambda _ j}\right ] % \right ] ^2},\ ] ] @xmath310 , \ ] ] @xmath311 ^2\right ] .\ ] ] the meissner solution in the region @xmath188 can be obtained from ( [ 4.6])-([4.11 ] ) by means of the substitution @xmath312 , @xmath313 . in the region
@xmath190 , the solution is @xmath314 @xmath315 @xmath316 , \ ] ] @xmath317 ^2.\ ] ] in the region @xmath318 the vortex ( antivortex ) solution satisfies the initial conditions @xmath319 , @xmath320 [ eq . ( [ 1.57.1 ] ) ] and has the form @xmath321 , \ ] ] @xmath322 , \ ] ] @xmath323 \sinh { } \left [ \frac y{\lambda _ j}\right ] .\ ] ] the quantities @xmath324 , @xmath325 and @xmath326 are given by ( [ 4.7 ] ) , ( [ 4.10 ] ) and ( [ 4.11 ] ) , respectively , with @xmath327 taken from ( [ 4.17 ] ) .
note that the field induced by a vortex at @xmath24 , in agreement with ( [ 1.61 ] ) , is @xmath328 and not @xmath329 , as in the case of a single junction with thick electrodes.@xcite by inserting ( [ 4.1 ] ) and ( [ 4.16 ] ) into ( [ 1.63 ] ) , we obtain the vortex self - energy : @xmath330 the vortex flux , according to ( [ 1.60 ] ) , is @xmath331 and the lower critical field , by ( [ 1.64 ] ) , is @xmath332 thus , for @xmath99 , the vortex flux @xmath333 , and the lower critical field ( [ 4.19 ] ) is much larger than the corresponding field of a single junction with thick electrodes @xmath334 , in agreement with ref . .
in the double - junction case , the nonzero matrix elements ( [ b.9 ] ) of @xmath336 are @xmath337 the corresponding @xmath338 matrix @xmath162 ( see appendix a ) has two positive eigenvalues : @xmath339 and @xmath340 , with the lengths @xmath341 according to ( [ 1.54 ] ) , the superheating ( penetration ) field is @xmath342 which is smaller than the corresponding single - junction value ( [ 4.5 ] ) , in agreement with ref . .
the application of ( [ 1.33 ] ) yields the value of the local field in the homogeneous meissner state : @xmath343 the meissner solution in the fields @xmath344 obeys the symmetry relations ( [ 1.51 ] ) .
the substitution of @xmath345 into ( [ 1.37 ] ) , using ( [ 4.20 ] ) , yields @xmath346 thus , all the results of the single - junction case , eqs .
( [ 4.6])-([4.15 ] ) , apply if we substitute @xmath347 ( note that @xmath348 ) , make the identification @xmath349 @xmath350 and take the values of @xmath6 and @xmath351 from ( [ 4.22 ] ) and ( [ 4.23 ] ) , respectively .
moreover , @xmath352 and @xmath353 in the region @xmath318 the vortex - plane ( antivortex - plane ) solution describes two vortices ( antivortices ) [ one vortex ( antivortex ) per i - layer ] and satisfies the initial conditions @xmath354 , @xmath355 , @xmath356 , @xmath357 [ eq . ( [ 1.57.1 ] ) ] . by ( [ 1.51 ] ) , it obeys the symmetry ( [ 4.23.1 ] ) and eq .
( [ 4.24 ] ) , with @xmath358 .\ ] ] thus , explicit expressions for @xmath359 and @xmath360 can be obtained from single - junction eqs .
( [ 4.17 ] ) , ( [ 4.18 ] ) , taking account of the substitution @xmath361 .
the quantities @xmath362 , @xmath363 and @xmath364 are given by ( [ 4.7 ] ) , ( [ 4.10 ] ) and ( [ 4.11 ] ) , respectively , with @xmath351 taken from ( [ 4.23 ] ) . for @xmath365 and @xmath366 , we have ( [ 4.26 ] ) and ( [ 4.27 ] ) , respectively .
the vortex - plane self - energy is @xmath367 and the flux is @xmath368 which immediately leads to the lower critical field : @xmath369 as can be seen by comparing ( [ 4.28 ] ) with the single - junction expression ( [ 4.18.1 ] ) , the energy per vortex in the double - junction stack is higher . finally ,
the field induced by the vortex plane at @xmath24 , according to ( [ 1.61 ] ) , is @xmath370 as can be easily seen , equations ( [ 1.34 ] ) with @xmath335 admit in the region @xmath190 another exact topological solution , namely an incoherent vortex - antivortex solution @xmath371 where @xmath372 is given by @xmath373 .\ ] ] with @xmath374 and @xmath375 , the vortex - antivortex solution explicitly satisfies the existence criterion ( [ 1.43 ] ) . by ( [ 1.63 ] ) ,
the self - energy of the vortex - antivortex solution is @xmath376 which is lower than the self - energy of the vortex - plane solution ( [ 4.28 ] ) .
however , the vortex - antivortex solution does not satisfy integrodifferential equations ( [ 1.30 ] ) with @xmath335 : these equations do not possess the symmetry ( [ 4.31 ] ) for any @xmath377 .
the vortex - penetration conditions also can not be met , because @xmath378 . moreover ,
the gibbs free energy ( [ 1.24 ] ) of the vortex - antivortex pair is always positive with respect to the gibbs free energy of the meissner state : the flux @xmath243 carried by this pair is exactly equal to zero .
thus , the static vortex - antivortex solution is thermodynamically unstable , in agreement with the general consideration in section iii . as shown in ref . , the vortex - antivortex solution can be realized in the dynamic regime , in the presence of an external current applied to the central s - layer . consider a configuration with a single vortex in one i - layer ( say , with @xmath379 ) and no vortices in the other .
the solution representing this configuration does not possess any symmetry .
as shown in appendix c , the self - energy of a single vortex , @xmath380 , satisfies the exact inequality @xmath381 , where @xmath28 and @xmath382 are given by ( [ 4.28 ] ) and ( [ 4.33 ] ) , respectively . on the other hand , the total flux carried by a single vortex , according to ( [ 1.77 ] ) with @xmath335 and @xmath383 ,
is @xmath384 which is exactly half the total flux of the vortex plane ( [ 4.29 ] ) .
hence , the gibbs free energy of a single vortex is positive with respect to that of the meissner state for @xmath385 .
this example clearly illustrates thermodynamic instability of single - vortex solutions discussed in section iii .
the two second - order differential equations describing the single - vortex configuration can be reduced@xcite to one fourth - order equation @xmath386 @xmath387 \left [ \left ( 2+\epsilon ^2\right ) \cos \phi _ 2\frac{d\phi _ 2 } { dy}-\epsilon ^2\frac{d^3\phi _ 2}{dy^3}\right ] ^2}{1-\left [ \left ( 2+\epsilon ^2\right ) \sin \phi _
2-\epsilon ^2\frac{d^2\phi _ 2}{dy^2}\right ] ^2}\ ] ] @xmath388 ^2}\left [ \frac{\left ( 2+\epsilon ^2\right ) ^2 - 1}{% \epsilon ^4}\sin \phi _ 2-\frac{2+\epsilon ^2}{\epsilon ^2}\frac{d^2\phi _ 2}{% dy^2}\right ] = 0.\ ] ] the phase difference @xmath389 can be found without any additional integration from the relations @xmath390 , \quad -\frac \pi 2\leq \phi _
1\leq \frac \pi % 2,\ ] ] @xmath391 + \pi , \quad \frac \pi 2<\phi _ 1<\pi , \ ] ] @xmath392 -\pi , \quad -\pi < \phi _ 1<-\frac \pi 2.\ ] ] the initial conditions for the vortex solution are given by the relations @xmath393 , @xmath394 , @xmath395 , @xmath396 and must satisfy the existence criterion ( [ 1.75 ] ) : @xmath397 } { \epsilon ^2}.\ ] ] in view of the condition @xmath398 , the appropriate solution of ( [ 4.36 ] ) for @xmath399 is @xmath400 \left [ 4\left ( 2+\epsilon ^2\right ) -\epsilon ^2\beta _ 1 ^ 2\right ] } \right ] , \ ] ] @xmath401 by the symmetry ( [ 1.68 ] ) , @xmath402 .
the initial condition on @xmath403 can be obtained from the relation @xmath404\ ] ] that follows directly from ( [ 1.34 ] ) : @xmath405 \left [ 4\left ( 2+\epsilon ^2\right ) -\epsilon ^2\beta _ 1 ^ 2\right ] } .\ ] ] thus , we have obtained a complete formulation of the single - vortex problem in terms of the standard initial value problem . in agreement with general consideration of section
iii , the problem admits a family of single - vortex solutions parameterized by @xmath406 .
unfortunately , there are no general methods of analytical integration of nonlinear differential equations of order higher than two .
however , numerical integration of ( [ 4.34 ] ) with the above - derived initial conditions should pose no problem .
moreover , it is not difficult to obtain asymptotics of the single - vortex solution in the region @xmath407 . for @xmath408 , equations ( [ 4.34 ] ) , ( [ 4.35 ] ) can be linearized : @xmath409 @xmath410 the solution of ( [ 4.39 ] ) , ( [ 4.40 ] ) obeying the vortex asymptotic conditions is straightforward : @xmath411 -c_2(\beta _
1)\exp \left [ \frac y{\lambda _ 2}\right ] , \ ] ] @xmath412 + c_2(\beta _ 1)\exp \left [ \frac y{\lambda _ 2}\right ] , \quad y\ll -\lambda _ { j1},\ ] ] where @xmath413,@xmath414 are constants with respect to @xmath44 , parameterized by the initial value @xmath406 .
the asymptotics for @xmath415 can be obtained from ( [ 4.41 ] ) by the use of the symmetry relations ( [ 1.68 ] ) : @xmath416 + c_2(\beta _
1)\exp \left [ -\frac y{\lambda _ 2}\right ] , \ ] ] @xmath417 -c_2(\beta _ 1)\exp \left [ -\frac y{\lambda _ 2}\right ] , \quad y\gg \lambda _ { j1},\ ] ] expressions ( [ 4.41 ] ) , ( [ 4.42 ] ) illustrate a very important property of single - vortex solutions in multilayer structures : their spatial dependence is characterized , in general , by @xmath0 different length scales , which agrees with the conclusions of ref . .
the consideration of all solutions to ( [ 1.34 ] ) in the case @xmath21 , including thermodynamically unstable incoherent vortex configurations , requires the use of @xmath419 second - order nonlinear differential equations , or , equivalently , of one nonlinear equation of order @xmath420 .
therefore , for the sake of mathematical simplicity , from now on we concentrate only on meissner solutions and topological vortex - plane solutions , both obeying the symmetry ( [ 1.51 ] ) .
( the analysis of incoherent vortex configurations for @xmath21 can be done using the algorithm worked out in the previous subsection . ) for @xmath418 , the meissner solutions and the vortex - plane solutions satisfy the relation @xmath421 . the relevant two equations for @xmath389 and @xmath422 can be given the form @xmath386
@xmath423 \left [ \left ( 2+\epsilon ^2\right ) \cos \phi _ 2 \frac{d\phi
_ 2}{dy}-\epsilon ^2\frac{d^3\phi _ 2}{dy^3}\right ] ^2}{1-\frac 14% \left [ \left ( 2+\epsilon ^2\right ) \sin \phi _
2-\epsilon ^2\frac{d^2\phi _ 2 } { dy^2}\right ] ^2}\ ] ] @xmath424 ^2}\left [ \frac{\left ( 2+\epsilon ^2\right ) ^2 - 2}{\epsilon ^4}\sin \phi
_ 2-\frac{2+\epsilon ^2}{\epsilon ^2}\frac{% d^2\phi _ 2}{dy^2}\right ] = 0,\ ] ] @xmath425 , \quad -\frac \pi 2\leq \phi _
1\leq \frac \pi 2,\ ] ] @xmath426 + \pi , \quad
\frac \pi 2<\phi _ 1<\pi , \ ] ] @xmath427 -\pi , \quad -\pi < \phi _ 1<-% \frac \pi 2.\ ] ] note that eqs . ( [ 4.43 ] ) , ( [ 4.44 ] ) have the same mathematical structure as equations for single vortices in the case of a double - junction stack ( [ 4.34 ] ) , ( [ 4.35 ] ) . according to ( [ 1.54 ] ) , the superheating ( penetration ) field for @xmath418 is @xmath428 } } { \epsilon \sqrt{10 + 3\epsilon ^2}}.\ ] ] the local field in the homogeneous meissner state , according to ( [ 1.33 ] ) , is given by the relations @xmath429 in fields @xmath430 , the general criterion of the existence of meissner solutions ( [ 1.53 ] ) takes the form @xmath431 = \frac{h^2 } { h_s^2},\ ] ] where @xmath432 , @xmath433 .
thus , equations ( [ 4.43 ] ) , ( [ 4.44 ] ) can be linearized : @xmath434 @xmath435 .\ ] ] the meissner solution of ( [ 4.48 ] ) , ( [ 4.49 ] ) in the region @xmath436 , obeying the boundary conditions @xmath437 is @xmath438 + \frac{\left ( \sqrt{2}-1\right ) } { \sqrt{2+\sqrt{2}% + \epsilon ^2}}\exp \left [ -\frac{\left ( y+l\right ) } { \lambda _ { j2}}\right ] % \right ] , \ ] ] @xmath439 -% \frac{\left ( \sqrt{2}-1\right ) } { \sqrt{2+\sqrt{2}+\epsilon ^2}}\exp \left [ -% \frac{\left ( y+l\right ) } { \lambda _ { j2}}\right ] \right ] , \ ] ] where @xmath440 we observe that spatial dependence of the solution ( [ 4.51 ] ) is characterized by two different josephson lengths , @xmath441 and @xmath442 , in agreement with the general consideration of section iii .
moreover , @xmath443 , and @xmath444 .
the values @xmath445 and @xmath446 meet the existence criterion ( [ 4.47 ] ) , as they should .
using ( [ 4.51 ] ) , we can derive explicit expressions for all physical quantities of interest from general formulas of section ii .
the meissner solution in the region @xmath188 can be obtained from ( [ 4.51 ] ) by means of the substitution @xmath447 . in the region @xmath190 , we have @xmath448 , and @xmath449 , as usual . the vortex - plane solution to ( [ 4.43 ] ) , ( [ 4.44 ] ) in the region @xmath450 obeys the initial conditions @xmath354 , @xmath355 and @xmath451 , @xmath452 [ eq . (
[ 1.57.1 ] ) ] . by the symmetry ( [ 1.55 ] )
, we also have @xmath402 . the initial condition on @xmath403
is derived from the relation @xmath453\ ] ] that follows from ( [ 1.34 ] ) : @xmath454 in this way , we arrive at a complete formulation of the initial value problem for the vortex - plane configuration .
in contrast to the single - vortex problem considered in the previous subsection , the above - derived initial conditions do not contain any arbitrariness , in full agreement with our general conclusion in section iii about the uniqueness of the vortex - plane solutions .
although the vortex - plane problem admits only numerical integration , the asymptotics in the regions @xmath455 and @xmath456 can be readily obtained .
thus , on the basis of linearized eqs .
( [ 4.48 ] ) , ( [ 4.49 ] ) , we have : @xmath457 -c_2\exp \left [ \frac y{\lambda _ { j2}}\right ] \right ] , \ ] ] @xmath458 + c_2\exp \left [ \frac y{\lambda _ { j2}}\right ] , \quad y\ll -\lambda _ { j1};\ ] ] @xmath459 -c_2\exp
\left [ -\frac y{\lambda _ { j2}}\right ] \right ] , \ ] ] @xmath460 -c_2\exp % \left [ -\frac y{\lambda _ { j2}}\right ] , \quad y\gg \lambda _ { j1},\ ] ] where @xmath461 and @xmath462 are positive constants .
the solution in the region @xmath463 is represented by taylor series expansions : @xmath464 @xmath465 finally , the total flux carried by the vortex plane , by ( [ 1.60 ] ) , is @xmath466 and the exact value of the induced field at @xmath24 , by ( [ 1.61 ] ) , is @xmath467 ( see fig .
2 . ) for @xmath468 , the meissner solutions and the vortex - plane solutions satisfy the relation @xmath469 and @xmath470 . the equations for @xmath389 and @xmath422 reduce to the form @xmath471
@xmath472 \left [ \left ( 1+\epsilon ^2\right ) \cos \phi _
2\frac{d\phi _ 2 } { dy}-\epsilon ^2\frac{d^3\phi _ 2}{dy^3}\right ] ^2}{1-\left [ \left ( 1+\epsilon ^2\right ) \sin \phi _ 2-\epsilon ^2\frac{d^2\phi _ 2}{dy^2}\right ] ^2}\ ] ] @xmath473 ^2}\left [ \frac{\left ( 2+\epsilon ^2\right ) \left ( 1+\epsilon ^2\right ) -1}{\epsilon
^4}\sin \phi _
2-\frac{2+\epsilon ^2}{% \epsilon ^2}\frac{d^2\phi _ 2}{dy^2}\right ] = 0,\ ] ] @xmath474 , \quad -\frac \pi 2\leq \phi _
1\leq \frac \pi % 2,\ ] ] @xmath475 + \pi , \quad \frac \pi 2<\phi _ 1<\pi , \ ] ] @xmath476 -\pi , \quad -\pi < \phi _ 1<-\frac \pi 2.\ ] ] mathematical structure of eqs .
( [ 4.56 ] ) , ( [ 4.57 ] ) is analogous to that of eqs .
( [ 4.43 ] ) , ( [ 4.44 ] ) of the 3-junction stack , which allows us to skip some details in what follows .
according to ( [ 1.54 ] ) , the superheating ( penetration ) field for @xmath468 is @xmath477 } } { \epsilon \sqrt{5 + 2\epsilon ^2}}.\ ] ] the local field in the homogeneous meissner state , according to ( [ 1.33 ] ) , is @xmath478 for the fields @xmath430 , the meissner solution in the region @xmath436 has the form @xmath479 + \frac{\left ( \sqrt{5}+1\right ) \left ( 3-\sqrt{5}\right ) } { \sqrt{\frac{3+\sqrt{5}}2% + \epsilon ^2}}\exp \left [ -\frac{\left ( y+l\right ) } { \lambda _ { j2}}\right ] % \right ] , \ ] ] @xmath480 -\frac{\left ( 3-\sqrt{5}\right ) } { \sqrt{\frac{3 + \sqrt{5}}2+\epsilon ^2}}\exp \left [ -\frac{\left ( y+l\right ) } { \lambda _ { j2}}% \right ] \right ] , \ ] ] where @xmath481 by comparison with @xmath441 , @xmath482 of the 3-junction case , eq .
( [ 4.52 ] ) , the josephson lengths given by ( [ 4.61 ] ) are larger , which illustrates a general tendency : josephson lengths increase with increasing @xmath2 . as in the 3-junction case , @xmath443 , and @xmath483 .
the values @xmath484 and @xmath485 satisfy the existence criterion ( [ 1.53 ] ) , as expected .
the vortex - plane solution to ( [ 4.56 ] ) , ( [ 4.57 ] ) in the region @xmath450 is characterized by the initial conditions @xmath486 , @xmath394 and @xmath487 , @xmath452 [ eq . ( [ 1.57.1 ] ) ] . in addition , we have @xmath402 and @xmath488 , as in the 3-junction case .
the asymptotics of the vortex - plane solution in the region @xmath489 are @xmath490 -\left ( \sqrt{5}+1\right ) c_2\exp \left [ \frac y{% \lambda _ { j2}}\right ] \right ] , \ ] ] @xmath458 + c_2\exp \left [ \frac y{\lambda _ { j2}}\right ] , \quad y\ll -\lambda _ { j1};\ ] ] @xmath491 -\left ( \sqrt{5}+1\right ) c_2\exp \left [ -%
\frac y{\lambda _ { j2}}\right ] \right ] , \ ] ] @xmath460 -c_2\exp % \left [ -\frac y{\lambda _ { j2}}\right ] , \quad y\gg \lambda _ { j1},\ ] ] where @xmath492 . in the region
@xmath493 , we have : @xmath464 @xmath465 the total flux carried by the vortex plane is @xmath494 and the exact value of the induced field at @xmath24 is @xmath495 ( see fig .
the limit @xmath497 $ ] ( @xmath498 $ ] stands for the integer part of @xmath499 ) corresponds to the situation when a periodic thin - layer s / i structure can be regarded as a layered superconductor rather than merely a ( @xmath0)-josephson - junction stack .
indeed , in this limit the superheating ( penetration ) field , eq .
( [ 1.54 ] ) , becomes @xmath500 and for @xmath501 tends to the limiting value of an infinite layered superconductor @xmath204.@xcite ( we remind that the lower critical field of an infinite layered superconductor , according to refs .
, is @xmath502 . )
the total flux carried by a vortex plane , by ( [ 1.60 ] ) , in the considered limit is @xmath503 , \ ] ] which for @xmath501 tends to the limiting value of one flux quantum @xmath83 per vortex , as expected for an infinite layered superconductor.@xcite moreover , according to ( [ 1.33 ] ) , for i - layers whose index @xmath50 satisfies the condition @xmath504
\ll n\ll n-1-\left [ \frac 1\epsilon \right ] , $ ] we have @xmath505 in other words , for @xmath506 , the inside junctions exhibit the complete meissner effect in the region @xmath190 .
thus , the integer @xmath507 $ ] determines the number of junctions near the boundaries @xmath508 and @xmath509 that feel the influence of the superconductor / vacuum interfaces .
far from the boundaries , i.e. in the region @xmath507 \ll x\ll n-1-\left [ \frac 1% \epsilon \right ] $ ] , one can apply all the results of the theory of infinite layered superconductors.@xcite
within the framework of standard methods of the theory of ordinary differential equations , we have obtained a complete mathematical description of josephson vortices and of the meissner effect in periodic thin - layer s / i structures .
a rigorous examination of the properties of eqs . ( [ 1.34 ] ) allowed us to establish the general existence criterion ( [ 1.43 ] ) , which formed a solid basis for our subsequent physical and mathematical conclusions .
one of the most striking physical consequences is that the derivation of the exact expression for the vortex - penetration field @xmath6 , eq .
( [ 1.54 ] ) , did not require any explicit solution of ( [ 1.34 ] ) . although explicit analytical solutions to eqs .
( [ 1.34 ] ) proved to be possible only in a limited number of cases discussed in section iv , numerical integration of these equations should pose no problem owing to the algorithm worked out in the paper .
all the three types of localized solutions obtained in the paper possess a number of interesting physical and mathematical properties .
for example , the meissner solutions are characterized by several different josephson lengths @xmath510 ( @xmath4 lengths for even @xmath2 , and @xmath22 lengths for odd @xmath2 ) .
unfortunately , this important fact was not noticed in previous publications.@xcite we think that our result may prove to be useful in view of the current experimental efforts@xcite to verify the interlayer tunneling model of high-@xmath511 superconductivity@xcite by measuring the @xmath512-axis penetration depth . [ the penetration of the parallel magnetic field with a distribution of length scales has been recently observed@xcite in the organic layered superconductor @xmath513-(bedt - ttf)@xmath514cu(ncs)@xmath514 . ]
mathematically , josephson vortices , represented by both the vortex - plane solutions and incoherent vortex solutions , are static sine - gordon - type solitons .
they satisfy the standard@xcite asymptotic boundary conditions , eqs .
( [ 1.56 ] ) , ( [ 1.57 ] ) , and eqs . ( [ 1.69])-([1.71 ] ) .
the expression for their self - energy ( [ 1.63 ] ) is a direct generalization of the well - known expression for a single junction.@xcite the thermodynamically stable vortex - plane solutions demonstrate a profound difference between josephson - vortex formation in weakly - coupled multilayer structures and abrikosov - vortex formation in continuum type - ii superconductors ( isotropic or not ) .
the proof of the existence of such solutions in the general case of ( @xmath0)-junction stacks establishes relationship to the well - know coherent mode
( alias the in - phase mode ) in a double - junction stack@xcite and the recently obtained@xcite vortex - plane solutions in infinite ( @xmath515 ) layered superconductors .
however , in contrast to the latter two cases , the vortex - plane solutions for @xmath27 are characterized by several @xmath3 , as the meissner solutions . moreover , in contrast to the case @xmath26 , the intralayer currents , @xmath264 , in the presence of a vortex plane are not equal to zero , with the exception of @xmath516 in the case of odd @xmath2 .
the single - vortex solutions are not uniquely determined by the asymptotic boundary conditions ( [ 1.69])-([1.71 ] ) , as follows from the existence criterion ( [ 1.75 ] ) .
their spatial dependence is characterized , in general , by @xmath0 length scales .
in contrast to the vortex planes , isolated josephson vortices can not penetrate the periodic s / i structure at any @xmath517 and do not satisfy the original integrodifferential equations ( [ 1.18 ] ) , ( [ 1.30 ] ) minimizing the gibbs free - energy functional ( [ 1.1 ] ) .
although the self - energy of incoherent vortex solutions may be lower than the self - energy of a vortex plane , their gibbs free energy is positive with respect to the gibbs free energy of the meissner state for @xmath518 ( @xmath10 is the lower critical field for the vortex - plane solution ) , as is illustrated by our analysis of a double - junction stack in section iv .
however , incoherent vortex solutions can be realized in the dynamic regime.@xcite in layered superconductors , isolated vortices may also emerge as metastable topological entities owing to pinning by extended defects .
we emphasize that our general conclusions about the form of single - vortex configurations stand in full agreement with numerical results of ref . .
finally , our exact results clearly show that the phase - difference equations ( [ 1.34 ] ) do not admit any solutions in the form of abrikosov - type vortices , suggested in refs . .
as we have already pointed out , @xcite the hypothesis of abrikosov - type solutions in infinite layered superconductors leads to incorrect estimates of the value of the lower critical field @xmath519 .
unfortunately , it seems that some other theoretical predictions , based on the assumption of the existence of abrikosov - type solutions , should also be revised .
equations ( [ 1.30 ] ) can be regarded as a nonhomogeneous finite difference equation for @xmath105 with respect to the layer index @xmath50 , subject to boundary conditions ( [ 1.19 ] ) . according to general theory of such equations,@xcite its solution for @xmath520
can be represented in the form @xmath521 where @xmath522 is the particular solution of ( [ 1.30 ] ) satisfying the boundary conditions @xmath523 and @xmath524 @xmath525 is the solution of the homogeneous form of ( [ 1.30 ] ) ( with the zero right - hand side ) meeting the boundary conditions @xmath526 the quantities @xmath125 in ( [ b.2 ] ) are matrix elements of @xmath527 matrix green s function @xmath528 .
they obey the nonhomogeneous finite difference equation @xmath529 ( @xmath530 is the kronecker index ) with the boundary conditions @xmath531 the explicit form of @xmath125 is @xmath532 .\ ] ] the following properties of @xmath125 can be easily verified using ( [ b.7 ] ) and ( [ b.9 ] ) : @xmath533 @xmath534 @xmath535 @xmath536\ ] ] @xmath537 , \quad 1\leq n\leq n-1,\ ] ] @xmath538 , \ ] ] note that matrix green s function for an infinite layered superconductor , @xmath539 , is determined by the matrix elements @xmath540 that satisfy the summation rule @xmath541 consider now a @xmath542 matrix @xmath543 , whose matrix elements are given by the right - hand side of ( [ b.9 ] ) with @xmath544 .
of special physical importance are positive eigenvalues of @xmath162 : they determine characteristic length scales of localized solutions to ( [ 1.34 ] ) .
indeed , in the asymptotic region where @xmath545 and @xmath546 , equations ( [ 1.37 ] ) can be linearized with the result @xmath547 the substitution @xmath548 $ ] yields @xmath549 which is exactly an eigenvalue equation for @xmath162 , with @xmath550 being a positive eigenvalue .
by introducing new functions @xmath551 @xmath552 we convert ( [ 1.34 ] ) into an equivalent normal system of @xmath553 first - order equations @xmath554 @xmath555 @xmath556 , \ ] ] @xmath557 , \quad n+1\leq i\leq n-3,\ ] ] @xmath558 , \ ] ] subject to initial conditions @xmath559 @xmath560 to prove the statement of the * lemma * , it is sufficient to observe that all @xmath561 are continuous functions of their arguments for @xmath562 and @xmath563 ( @xmath564 ) .
moreover , their partial derivatives with respect to @xmath565 satisfy the relation @xmath566 for @xmath562 and @xmath567 ( @xmath568 ) .
thus , the lipschitz conditions with respect to @xmath565 are met for @xmath562 and @xmath567 ( @xmath569 ) , which immediately guarantees@xcite the existence and uniqueness of a solution to ( [ a.2 ] ) , satisfying arbitrary initial conditions ( [ a.3 ] ) , in an arbitrary interval @xmath146 $ ] such that @xmath570 .
continuous dependence of the solution on initial data is a result of continuous dependence of @xmath571 on their arguments and of the condition ( [ a.4 ] ) .
infinite differentiability of the solution automatically follows from infinite differentiability of @xmath572 with respect to their arguments .
to determine the upper and the lower bounds of the self - energy of single - vortex solutions in a double - junction stack , we must find the extrema of the right - hand side of ( [ 1.63 ] ) with @xmath335 under the normalization condition @xmath573 ^2= \text{const}<\infty .\ ] ] this leads to the variational principle @xmath574 ^2\right ] = 0,\ ] ] where @xmath575 is a lagrange multiplier . performing the variation with the use of the boundary conditions @xmath576 , we arrive at the eigenvalue problem for the @xmath338 matrix @xmath543 , determined by the matrix elements ( [ 4.20 ] ) : @xmath577 the only two eigenvalues of @xmath162 are @xmath578 and @xmath579 , with @xmath441 and @xmath580 given by ( [ 4.21 ] ) .
the larger eigenvalue , @xmath581 , corresponds to the vortex - plane solution with @xmath582 and determines the upper bound ( [ 4.28 ] ) for the self - energy .
the smaller eigenvalue , @xmath579 , corresponds to the vortex - antivortex solution with @xmath583 and determines the lower bound ( [ 4.33 ] ) for the self - energy .
thus , for the self - energy of the single - vortex solutions , @xmath380 , we necessarily have @xmath584 1 . fig .
1 . the geometry of the problem ( schematically ) . here
@xmath585 ; @xmath586 , @xmath587 ; @xmath176 .
2 . fig .
2 . the distribution of the self - induced field of a vortex plane @xmath588 at @xmath24 , for @xmath589 : a ) a 3-junction stack , @xmath590 ; b ) a 4-junction stack , @xmath591 .
|
we present an exact mathematical description of josephson vortices and of the meissner effect in periodic thin - layer superconductor / insulator structures with an arbitrary number of identical junctions @xmath0
( @xmath1 , where @xmath2 is the number of superconducting layers ) in terms of localized solutions to a system of differential equations for phase differences . we establish a general criterion of the existence of localized solutions .
we show that meissner solutions are characterized by several josephson lengths @xmath3 ( @xmath4 lengths for even @xmath2 , and @xmath5 lengths for odd @xmath2 ) .
we derive an exact expression for the superheating field of the meissner state , @xmath6 , as an explicit function of @xmath7 . for josephson vortices , we find two basically different types of topological solutions : vortex - plane solutions and incoherent vortex solutions .
thermodynamically stable vortex - plane
solutions represent a chain of @xmath0 vortices ( one vortex per each insulating layer ) .
they are characterized by the same set of @xmath8@xmath9 as the meissner solutions .
we obtain exact analytical expressions for their self - energy and for the lower critical field @xmath10 .
incoherent vortex solutions comprise solutions with @xmath11 vortices and different vortex - antivortex configurations .
in contrast to the vortex - plane solutions , they prove to be thermodynamically unstable , and their spatial dependence is characterized , in general , by @xmath0 length scales . as an illustration , we analyze 1 - 4-josephson - junction stacks and investigate a transition to the layered superconductor limit ( @xmath12 ) .
|
given a partition @xmath6 of @xmath7 , let @xmath8 be the associated irreducible complex representation of the symmetric group @xmath9 .
the _ kronecker coefficients _
@xmath10 are the tensor product multiplicities : @xmath11 to determine these coefficients and understand their properties has been one of the major problems in combinatorics and representation theory for nearly a century .
people are particularly interested in finding combinatorial interpretation for these coefficients .
they hope that kronecker coefficients count some combinatorial objects , eg .
, lattice points in polytopes .
more recently , the interest in computing kronecker coefficients has intensified in connection with _ geometric complexity theory _
@xcite , pioneered as an approach to the p vs. np problem . despite of a large body of work ,
those coefficients remain very mysterious .
the complete solution of analogous problems for the general linear groups @xmath12 certainly bring new hope to this old problem .
the tensor multiplicities of irreducible representations of @xmath12 , known as _ littlewood - richardson coefficients _ , can be computed by the littlewood - richardson rule . according to knutson - tao @xcite
, they are counted by the lattice points in the so - called hive polytopes .
similar results were also obtained by derksen - weyman in @xcite via an invariant - theoretic approach .
the link between these two approachs was established in @xcite through fomin - zelevinsky s cluster algebras @xcite and their quiver with potential models @xcite . in this paper
, we use the similar ideas in @xcite to study the kronecker coefficients
. it may be quite frustrating to study the kronecker coefficients via a purely finite - group theoretic approach .
however , by the schur - weyl duality they can be interpreted as multiplicities below @xmath13 then calculation of kronecker coefficients becomes a problem in invariant theory .
analogous to @xcite , our first main result is make this link completely explicit via quiver representations .
let @xmath14 be the _ flagged kronecker quiver _ @xmath15 and @xmath16 be the dimension vector defined by @xmath17 .
let @xmath18 be a @xmath19-dimensional vector space and @xmath20 be the @xmath0-dimensional vector space spanned by the arrows from @xmath21 to @xmath5 .
consider the product of special linear group @xmath22 acting naturally on the quiver representation space @xmath23 the _ semi - invariant ring _
@xmath24^{{\operatorname{sl}}_{{\beta_l}}}$ ] is graded by a weight @xmath25 and a weight @xmath6 of @xmath26 : @xmath27 for any pair of partitions @xmath28 and @xmath29 of _ length _ no greater than @xmath5 , we can associate a weight vector @xmath30 ( see lemma [ l : si(kml ) ] ) .
[ t : fkm ] let @xmath31 be a triple of partitions of length no greater than @xmath32 and @xmath0 respectively , then @xmath33 where @xmath34 is the group of upper triangular matrices in @xmath26 , and @xmath35 is the weight defined by @xmath36 .
it is well - known that the kronecker coefficients are invariant under the permutation of @xmath37 and @xmath29 . in the literature ,
it is usually written as @xmath38 .
unfortunately , this symmetry is blurred from our construction , so we keep the notation @xmath10 throughout . to study the kronecker coefficients , we need more structure in these semi - invariant rings .
it turns out that when @xmath1 all these semi - invariant rings are _ upper cluster algebras _ introduced by fomin - zelevinsky and berenstein @xcite .
here is the crucial result .
[ t : crucial ] the semi - invariant ring @xmath39 is the upper cluster algebra @xmath40 . moreover , the _ generic cluster character _
maps the lattice points in a rational polyhedral cone @xmath3 bijectively onto a basis of @xmath41 .
we have a concrete description of the cone @xmath3 . here
, @xmath42 is the ice _ diamond quiver _ to be introduced in section [ s : diamond ] . when @xmath43 , it is displayed in figure [ f : diamond4 ] . the initial _
cluster variables _ together with the _ coefficients _ @xcite in @xmath44 can be explicitly described in terms of _
schofield s semi - invariants _ @xcite .
we want to point out here that all these upper cluster algebras strictly contain the corresponding cluster algebras @xmath45 .
this might suggest that upper cluster algebras are more important than cluster algebras .
as mentioned above , we dream that kronecker coefficients count lattice points in some polytopes . with the help of theorem [ t : crucial ] , we are getting closer to this dream .
we realize the relevant kronecker coefficients as the difference of the numbers of lattice points in two polytopes .
all these polytopes are obtained from the cone @xmath3 by hyperplane sections defined by the weights @xmath46 and @xmath6 .
we denote such a polytope by @xmath47 ( or @xmath48 if no restriction on @xmath6 ) .
[ t : second ] let @xmath28 and @xmath29 be two partitions of length no greater than @xmath5 , and @xmath49 be the associated weight
. then 1 .
@xmath50_2 = \sum_{{{\sf g}}\in { { \sf g}}_{\diamond_l}(\sigma ) \cap { \mathbb{z}}^{\diamond_l^0 } } { \bold{z}}^{\lambda({{\sf g}})}.\ ] ] 2 .
let @xmath6 be a partition of length @xmath51 , and @xmath52 , then @xmath53 here , @xmath54_2 $ ] is the @xmath4-truncated kronecker product of symmetric functions ( see section [ s : kc ] ) , and @xmath55 is the @xmath6-weight of @xmath2 ( see section [ s : ccs ] ) . in this paper , we treat the case of @xmath1 only .
there are several reason for this .
first , @xmath1 is the first nontrivial case .
we will see that to prove these results is already quite difficult .
to author s best knowledge , there is no general result like this in the literature .
second , we do find cluster structures for @xmath56 , but there are many different cluster structures for fixed @xmath5 and @xmath0 .
however , we believe that for @xmath1 the cluster structure is unique , though we do not have a proof for this . finally , we will see in the subsequential paper that the cluster structure established here is a part of the building blocks for bigger @xmath0 .
so we think that this case deserves our special care .
here is the outline of this paper . in section [ s
: kc ] , we recall some basic facts on the kronecker coefficients . in section [ s : si ] , we recall the work of schofield , derksen - weyman , etc .
, on the semi - invariant rings of quiver representations . in section [ s : fkm ]
, we introduce the flagged kronecker quivers and prove theorem [ t : fkm ] .
we also introduce a special class of semi - invariants , which will be our choice of initial cluster . in section [ s :
ca ] , we recall the definition of graded cluster algebras and their upper bounds . in section [ s :
cc ] , we recall the ice quivers with potentials and their representations following @xcite .
we recall the generic cluster character in the current setting following @xcite . in section [
s : diamond ] , we introduce the ice diamond quiver with potential @xmath57 . we give in theorem [ t : lp_gl ] a concrete description of the domain of the generic cluster character as the lattice points in the polyhedral cone @xmath3 . in section
[ s : cs ] , we partially prove our first main result - theorem [ t : crucial ] .
the full proof is quite complicated so we put off it to the last section . in section [ s :
ccs ] , we draw several consequences from the cluster structure .
we prove our second main result - theorem [ t : second ] . as a minor result , in proposition [ p : guinv ] we estimate all possible @xmath2-vectors in the @xmath34-invariants . in section [
s : inv ] , we introduce the _ extremal unimodular fan _ as a tool to present upper cluster algebras . as an application , we compute several relevant invariant rings . in section [
s : proof ] , we fully prove theorem [ t : crucial ] .
our vectors are exclusively row vectors .
all modules are right modules . for a quiver
@xmath58 , we denote by @xmath59 the set of vertices and by @xmath60 the set of arrows . for an arrow @xmath61 ,
we denote by @xmath62 and @xmath63 its tail and head .
arrows are composed from left to right , i.e. , @xmath64 is the path @xmath65 . throughout the paper , the base field @xmath66
is algebraically closed of characteristic zero .
unadorned @xmath67 and @xmath68 are all over the base field @xmath66 , and the superscript @xmath69 is the trivial dual for vector spaces . for any representation @xmath70 , @xmath71 is the dimension vector of @xmath70 . for direct sum of @xmath7 copies of @xmath70
, we write @xmath72 instead of the traditional @xmath73 .
a _ partition _ @xmath6 is a weakly decreasing sequence @xmath74 eventually being zeros .
we often do not write the trailing zeros , but the zero partition @xmath75 is also valid .
the _ length _ of @xmath6 denoted by @xmath76 is the number of indices @xmath77 for which @xmath78 is non - zero .
the _ size _ of @xmath6 denoted by @xmath79 is the sum of @xmath78 s .
if @xmath6 has size @xmath7 , we also say that @xmath6 is a partition of @xmath7 , and write @xmath80 .
when @xmath6 contains some repetition , we may use superscripts to simplify the notation .
for example , @xmath81 is the partition @xmath82 .
the _ transpose _ or _
conjugate _ of @xmath6 is the partition @xmath83 defined by @xmath84 we fix an algebraically closed field of characteristic @xmath85 as our base field @xmath66 . for @xmath86
let @xmath9 be the symmetric group of @xmath7 letters .
the irreducible representations of @xmath9 over @xmath66 are parameterized by partitions of @xmath7 . for a partition @xmath6 ,
we denote by @xmath8 the irreducible representation corresponding to @xmath6 .
the category of all representations of @xmath9 is semisimple .
the _ kronecker coefficients _
@xmath10 are the tensor product multiplicities : @xmath11 the _ trivial representation _ and the _ sign representation _ are the two one - dimensional representations corresponding to the partition @xmath87 and @xmath88 .
they are characterized by the property that @xmath89 in particular , we have that @xmath90
. since all representations of @xmath9 are self - dual , @xmath10 is invariant under the permutations of @xmath91 and @xmath29 .
the _ schur functor _
@xmath92 is by definition @xmath93 where @xmath18 is a @xmath66-vector space .
it follows from the definition that [ l : sw ] we have the following decomposition @xmath94 by the _ schur - weyl duality _ , @xmath95 is an irreducible _ polynomial _ representation of the general linear group @xmath96 if @xmath97 .
it is well - known that the category of all polynomial representations of @xmath96 is also semisimple , and the irreducible representations are parameterized by partitions of length no greater than @xmath98 .
the tensor product multiplicities @xmath99 are called _ littlewood - richardson coefficients _ ( lr coefficient for short ) : @xmath100 the lr coefficients can be computed using the littlewood - richardson rule @xcite .
we fix the dimension of the vector space @xmath18 , say @xmath101 .
let @xmath102 be the space of homogeneous symmetric polynomials of degree @xmath7 in @xmath0 variables @xmath103 .
a basis for this space is given by the _ schur polynomials _ @xmath104 .
let @xmath105 be the algebra of symmetric polynomials .
the _ character map _
@xmath106 maps a representation @xmath70 of @xmath96 to @xmath107 , where @xmath108 is subspace of @xmath70 of weight @xmath6 .
it is well - known @xcite that @xmath109 .
in particular , the structure constants in @xmath110 is also given by the lr coefficients : @xmath111 the _ @xmath0-truncated kronecker product _ in @xmath110 is by definition @xmath112_m : = \sum_{\lambda\mid \ell(\lambda)\leq m}g_{\mu,\nu}^\lambda s_\lambda.\ ] ] as is usual in the theory of _ symmetric functions _ , we can use inverse limit ( see @xcite ) to obtain a _ schur function _ @xmath113 for each partition , depending on countably many variables .
we recall that the _ kronecker product _ on the algebra of symmetric functions is defined by @xmath114 the kronecker product and the ordinary product satisfy the littlewood s identity : @xmath115 using this identity and jacobi - trudi formula , we can express kronecker coefficients in terms of multiple lr coefficients . for a set @xmath116 of @xmath0 partitions @xmath117 , the _
multiple _ lr coefficients @xmath118 are the multiplicities of @xmath95 in the multiple tensor product : @xmath119 for @xmath120 , we write @xmath35 for the vector in @xmath121 such that @xmath122 .
note that @xmath35 is not necessarily a partition .
[ l : k2c ] assume that @xmath123 , then @xmath124 we first recall that the jacobi - trudi formula ( * ? ?
* a.5 ) expresses @xmath125 as the determinant of the matrix @xmath126 .
so @xmath127 therefore @xmath128 [ c : k2c ] let @xmath129 be a triple of partition of @xmath7 with @xmath130 .
we set @xmath131 , then @xmath132 in practice , if we apply the above formula to compute kronecker coefficients , we may find that most of the lr coefficients are zeros .
so to really make this formula effective , we need to understand the non - vanishing conditions for lr coefficients .
such conditions are conjectured by a. horn , and solved by klyachko and knutson - tao @xcite , and many other people , remarkably @xcite . to state these conditions , we define for any positive decreasing sequence @xmath133 a partition
@xmath134 [ l : horn ] let @xmath129 be a triple of partitions satisfying @xmath135 and @xmath136 .
then @xmath137 for any @xmath138 satisfying @xmath139 and @xmath140 .
since @xmath141 , as a rather trivial consequence of lemma [ l : horn ] , we have that @xmath142 are all necessary .
these inequalities are essentially due to h. weyl .
it follows that @xmath143 .
another easy consequence on the length is that @xmath144 . as an illustration ,
let us compute one family of kronecker coefficients .
[ l : g=2 ] we have that @xmath145 for @xmath146 , and @xmath147 with @xmath148 and @xmath149 .
we assume that @xmath150 .
the case of @xmath151 can be treated similarly . by the restriction on the length , to make @xmath152 contribute to @xmath153 in corollary [ c : k2c ] , we must have that @xmath154 .
applying weyl s inequalities for @xmath155 , we have that @xmath156 for @xmath157 .
in particular , @xmath158 .
but @xmath159 in @xmath160 for @xmath161 .
so the only choice for @xmath162 is @xmath163 and @xmath164 . similarly applying weyl s inequalities for @xmath165
, we get @xmath166 .
we claim that the only choice for @xmath167 is @xmath168 .
since @xmath169 , we do not have many choices left .
other choices can be easily ruled out by other horn s inequalities .
we note that there are six more inequalities coming from @xmath170 and their symmetry . finally , we compute by the littlewood - richardson rule that @xmath171 and @xmath172 .
hence , @xmath145 .
let us briefly recall the semi - invariant rings of quiver representations @xcite . let @xmath58 be a finite quiver without oriented cycles . for a dimension vector @xmath173 of @xmath58 , let @xmath18 be a @xmath173-dimensional vector space @xmath174 .
we write @xmath175 for the @xmath77-th component of @xmath18 .
the space of all @xmath173-dimensional representations is @xmath176 the product of general linear group @xmath177 acts on @xmath178 by the natural base change .
define @xmath179 by @xmath180 we are interested in the rings of semi - invariants @xmath181^{{\operatorname{sl}}_\beta}.\ ] ] the ring @xmath182 has a weight space decomposition @xmath183 where @xmath46 runs through the multiplicative _ characters _ of @xmath184 .
we refer to such a decomposition the @xmath46-grading of @xmath182 .
recall that any character @xmath185 can be identified with a weight vector @xmath186 @xmath187 since @xmath58 has no oriented cycles , the degree zero component is the field @xmath66 @xcite .
let us understand these multihomogeneous components @xmath188\mid g(f)=\sigma(g)f , \forall g\in{\operatorname{gl}}_\beta \}.\ ] ] for any projective presentation @xmath189 , we view it as an element in the homotopy category @xmath190 of bounded complexes of projective representations of @xmath58 .
the _ weight vector _
@xmath191 of @xmath192 is the corresponding element in the grothendieck group of @xmath190 . concretely , suppose that @xmath193 and @xmath194 for @xmath195 , then @xmath196 . here
, we use the notation @xmath197 for @xmath198 , where @xmath199 is the indecomposable projective representation corresponding to the vertex @xmath77 . from now on , we will view a weight @xmath46 as an element in the dual @xmath200 via the usual dot product .
we assume that @xmath201 and @xmath202 .
we apply the functor @xmath203 to @xmath192 for @xmath204 @xmath205 since @xmath202 , @xmath206 is a square matrix . following schofield @xcite , we define @xmath207 we give a more concrete description for the map @xmath206 .
+ * concrete description of @xmath208 * recall that a morphism @xmath209 can be represented by a matrix whose entries are linear combination of paths . applying @xmath203 to this morphism is equivalent to that we first transpose the matrix of @xmath192 , then substitute paths in the matrix by corresponding matrix representations in @xmath210 .
we set @xmath211 as a function on @xmath178 .
it is proved in @xcite that @xmath212 .
in fact , [ t : inv_span ] @xmath213 s span @xmath214 over the base field @xmath66
. it is easy to see that if @xmath215 , then @xmath192 resolves some representation @xmath70 , say of dimension @xmath216 @xmath217 from the long exact sequence @xmath218 we see that @xmath216 and @xmath46 are related by @xmath219 , where @xmath220 is the _ euler form _ of @xmath58 . in this case , we call @xmath46 the weight vector corresponding to @xmath216 , and denote it by @xmath221 ; and conversely we call @xmath216 the dimension vector corresponding to @xmath46 , and denote it by @xmath222 .
it also follows from that @xmath223 if and only if @xmath224 or @xmath225 .
we want to point out that the function @xmath213 is determined , up to a scalar multiple , by the homotopy equivalent class of @xmath192 , and thus by the isomorphism class of @xmath70 .
if one hopes to define the function @xmath213 uniquely in terms of representations , one can take the _ canonical resolution _
@xmath226 of @xmath70 .
this is schofield s original definition in @xcite .
we define the subgroup @xmath227 to be the kernel of the character map .
the invariant ring of its action is @xmath228^{{\operatorname{gl}}_\beta^\sigma}=\bigoplus_{n\geqslant 0 } { \operatorname{si}}_\beta(q)_{n\sigma}.\ ] ] a representation @xmath229 is called _ @xmath46-semi - stable _ if there is some non - constant @xmath230 such that @xmath231 .
it is called _ stable _ if in addition the orbit @xmath232 is closed of dimension equal to @xmath233 .
based on hilbert - mumford criterion , king provided a simple criterion for the stability of a representation .
* proposition 3.1 ) [ l : king ] a representation @xmath70 is @xmath46-semi - stable ( resp .
@xmath46-stable ) if and only if @xmath234 and @xmath235 ( resp .
@xmath236 ) for any non - trivial subrepresentation @xmath237 of @xmath70 . let @xmath238 be the set of all weights @xmath46 such that @xmath239 is non - empty .
the next theorem is an easy consequence of king s stability criterion and theorem [ t : inv_span ] . by a general @xmath173-dimensional representation , we mean in a sufficiently small zariski open subset ( sufficient " here depends on the context ) .
following @xcite , we use the notation @xmath240 to mean that a general @xmath173-dimensional representation has a @xmath241-dimensional subrepresentation .
* theorem 3 ) [ t : cone ] we have @xmath242 in particular , @xmath238 is a saturated semigroup .
recall that a dimension vector @xmath216 is called a _
root if @xmath243 .
it is called _ schur _ if @xmath244 for @xmath70 general in @xmath245 . for a real schur root @xmath216
, @xmath246 has a dense orbit for any @xmath247 .
[ l : gsreal ] if @xmath216 is a real schur root , then @xmath248 for any @xmath247 .
a weight is called _ extremal _ in @xmath238 if it lies on an extremal ray of @xmath249 .
[ l : irreducible ] if @xmath46 is an indivisible extremal weight , then any semi - invariant function @xmath250 of weight @xmath46 is irreducible .
let @xmath251 be the vector space spanned by arrows from @xmath77 to @xmath252 .
let @xmath253 be the group @xmath254 acting naturally on @xmath255 .
this induces an action on the space of paths @xmath256 , and thus induces an action on the space of presentations @xmath257 .
the group @xmath253 also acts on the representation spaces @xmath178 by twisting the action of arrows : @xmath258 we will write this action exponentially , that is , @xmath259 and @xmath260 for @xmath261 and @xmath262 .
the action on the representations induces an action on the semi - invariant rings : @xmath263 , because the action of @xmath253 commutes with the action of @xmath184 @xmath264 on the other hand , the action on the presentations also induces an action on the semi - invariant rings : @xmath265 .
it turns out that two induced action are the same .
indeed , we recall that @xmath266 .
it is clear from our description of @xmath206 that @xmath267 , so @xmath268 .
[ d : ginv ] let @xmath269 be a subgroup of @xmath253 .
a presentation @xmath262 is called _
coherently @xmath269-invariant _ if there is an algebraic group homomorphism @xmath270 such that @xmath271 .
[ l : ginv ] if @xmath192 is coherently @xmath269-invariant , then @xmath213 is @xmath269-semi - invariant .
it is proved in ( * ? ? ?
* proposition 5.13 ) that @xmath272 is semi - invariant as a function on the presentation space @xmath257 under the action of @xmath273 .
suppose that @xmath271 , then @xmath274 for some character @xmath275 .
hence @xmath276 , which means that @xmath213 is @xmath269-semi - invariant .
consider the following quiver @xmath277 @xmath278 we see that the full subquiver of two vertices @xmath279 and @xmath280 is an @xmath0-arrow kronecker quiver , so we call it a _ flagged kronecker quiver_. the full subquiver of negative ( resp
. positive ) vertices are called its negative ( positive ) arm .
we fix a schur root @xmath173 of @xmath277 .
it is easy to see that @xmath173 is nondecreasing ( resp .
nonincreasing ) along the negative ( positive ) arm .
let @xmath18 be a @xmath173-dimensional vector space , and @xmath281 be the @xmath0-dimensional vector space spanned by arrows from @xmath279 to @xmath280 .
let us decompose the coordinate ring @xmath282 \\ = & k\big[\bigoplus_{i={1}}^{l_1 - 1 } { \operatorname{hom}}(v_{-i},v_{-(i+1 ) } ) \oplus \bigoplus_{i={1}}^{l_2 - 1 } { \operatorname{hom}}(v_{i+1},v_{i } ) \oplus { \operatorname{hom}}(v_{-l_1},v_{l_2})\otimes w \big]\\ = & \bigotimes_{i={1}}^{l_1 - 1 } { \operatorname{sym}}(v_{-i}\otimes v_{-(i+1)}^ * ) \otimes \bigotimes_{i={1}}^{l_2 - 1 } { \operatorname{sym}}(v_{i+1}\otimes v_{i}^ * ) \otimes { \operatorname{sym}}(v_{-l_1}\otimes v_{l_2}^ * \otimes w)\\ \intertext{by cauchy 's formula \cite{fh } , this decomposes into a direct sum over all tuples of partitions $ { \boldsymbol{\lambda}}=(\lambda_{-l_1 } , \dots , \lambda_{-1 } , \lambda_0,\lambda_1,\dots,\lambda_{l_2})$ of the summands } & \bigotimes_{i={1}}^{l_1 - 1 } \left ( { { \bf s}}_{\lambda_{-i}}v_{-i}\otimes { { \bf s}}_{\lambda_{-i}}v_{-(i+1)}^ * \right)\otimes \bigotimes_{i={1}}^{l_2 - 1 } \left ( { { \bf s}}_{\lambda_{i}}v_{i+1}\otimes { { \bf s}}_{\lambda_{i } } v_{i}^*\right ) \otimes { { \bf s}}_{\lambda_0}(v_{-l_1}\otimes v_{l_2}^ * ) \otimes { { \bf s}}_{\lambda_0}w \\ \intertext{which is by lemma \ref{l : sw } } & \bigoplus_{|\lambda|=|\mu|=|\nu|=|\lambda_0| } g_{\mu,\nu}^\lambda \bigg ( \bigotimes_{i={1}}^{l_1 - 1 } \left({{\bf s}}_{\lambda_{-i}}v_{-i}\otimes { { \bf s}}_{\lambda_{-i}}v_{-(i+1)}^ * \right ) \otimes \bigotimes_{i={1}}^{l_2 - 1 } \left ( { { \bf s}}_{\lambda_{i}}v_{i+1}\otimes { { \bf s}}_{\lambda_{i } } v_{i}^ * \right ) \\ & \hspace{3.05 in } \otimes \big ( { { \bf s}}_{\mu}v_{-{l_1}}\otimes { { \bf s}}_{\nu}v_{l_2}^ * \otimes { { \bf s}}_{\lambda}w\big ) \big).\end{aligned}\ ] ] so the invariant ring @xmath283^{{\operatorname{sl}}_\beta}$ ] @xmath284 [ l : si(kml ) ] let @xmath46 be a weight on @xmath277 , and @xmath285 be two partitions of @xmath7 equal to @xmath286 , and @xmath287 , then @xmath288 the proof is similar to that of ( * ? ? ?
* proposition 1 ) . for @xmath289 to contribute to @xmath290
, the partition @xmath291 must be the square partition @xmath292 . in this case ,
@xmath293 . for the next one @xmath294 to contribute
, the partition @xmath295 must be equal to @xmath291 with some extra rows of size @xmath296 . in this case
, @xmath294 is also @xmath66 .
we continue this procedure , and conclude that @xmath297 must be equal to @xmath298 .
similarly , we have that @xmath299 .
we thus get the required equality .
let @xmath34 and @xmath300 be the subgroups of upper - triangular and diagonal matrices in @xmath26 .
the semi - invariant ring @xmath301 is also graded by the @xmath300-weights @xmath6 .
we write @xmath302 for the weight @xmath6 component of @xmath290 .
we refer to this grading as the extended @xmath46-grading or @xmath303-grading .
[ c : dimuinv ] the dimension of @xmath304 is equal to @xmath305 .
we can also decompose the coordinate ring of @xmath290 in another way .
@xmath282 \\ = & \bigotimes_{i={1}}^{l_1 - 1 } { \operatorname{sym}}(v_{-i}\otimes v_{-(i+1)}^ * ) \otimes \bigotimes_{i={1}}^{l_2 - 1 } { \operatorname{sym}}(v_{i+1}\otimes v_{i}^ * ) \otimes { \operatorname{sym}}(v_{-l_1}\otimes v_{l_2}^*)^{\otimes m}.\\ \intertext{by cauchy 's formula \cite{fh } , this decomposes into a direct sum over all tuples of partitions $ { \boldsymbol{\lambda}}=(\lambda_{-l_1 } , \dots , \lambda_{-1},\lambda_1,\dots,\lambda_{l_2})$ of the summands } & \bigotimes_{i={1}}^{l_1 - 1 } \left ( { { \bf s}}_{\lambda_{-i}}v_{-i}\otimes { { \bf s}}_{\lambda_{-i}}v_{-(i+1)}^ * \right)\otimes \bigotimes_{i={1}}^{l_2 - 1 } \left ( { { \bf s}}_{\lambda_{i}}v_{i+1}\otimes { { \bf s}}_{\lambda_{i } } v_{i}^*\right ) \otimes \left ( \bigoplus_{\lambda_0 } { { \bf s}}_{\lambda_0}v_{-l_1}\otimes { { \bf s}}_{\lambda_0 } v_{l_2}^ * \right)^{\otimes m } \\ = & \bigotimes_{i={1}}^{l_1 - 1 } \left ( { { \bf s}}_{\lambda_{-i}}v_{-i}\otimes { { \bf s}}_{\lambda_{-i}}v_{-(i+1)}^ * \right)\otimes \bigotimes_{i={1}}^{l_2 - 1 } \left ( { { \bf s}}_{\lambda_{i}}v_{i+1}\otimes { { \bf s}}_{\lambda_{i } } v_{i}^*\right ) \\ & \hspace{1.83 in } \otimes \left(\bigoplus_{\mu,\nu , { \boldsymbol{\eta}}=\{\eta_1,\eta_2,\dots,\eta_m\ } } c_{{\boldsymbol{\eta}}}^\mu c_{{\boldsymbol{\eta}}}^\nu s^{\mu}v_{-l_1}\otimes s^{\nu } v_{l_2}^*\right).\end{aligned}\ ] ] so the invariant ring @xmath283^{{\operatorname{sl}}_\beta}$ ] @xmath306 by a similar argument as in lemma [ l : si(kml ) ] , we obtain the following lemma .
[ l : si(kml)2 ] in the same setting as lemma [ l : si(kml ) ] , we have that @xmath307 moreover , each @xmath308 counts the dimension of @xmath302 .
it follows from lemma [ l : si(kml)2 ] and [ l : k2c ] that [ c : si(kml)2 ] we have that @xmath309 for the rest of the paper , we will mainly concern about the case when @xmath310 and @xmath173 is the _ standard dimension vector _ @xmath311 .
we note that for any pair of partitions @xmath312 with @xmath313 , there is a weight @xmath46 such that @xmath314 in lemma [ l : si(kml ) ] .
let @xmath315 be the unit vector supported on the vertex @xmath77 .
we use the convention that @xmath316 is the zero vector .
it follows from theorem [ t : cone ] that if @xmath317 , then @xmath318 we consider the following set of weight vectors labeled by @xmath319 and @xmath320 satisfying @xmath149 : @xmath321 for each weight vector @xmath322 , we associate a projective presentation @xmath323 given by @xmath324 to simplify our exposition , we denote @xmath325 and write @xmath326 for both @xmath323 and @xmath327 .
we introduce an involution @xmath328 on @xmath256 sending the path @xmath329 to @xmath330 .
this induces a map @xmath331 in this language ,
we can write @xmath332 as @xmath333 . recall that @xmath213 is the schofield s semi - invariant function attached to a presentation @xmath192 .
we write @xmath334 for @xmath335 .
it is useful to study the @xmath26-action on these special semi - invariants . recall that the @xmath336 acts on @xmath337 by @xmath338 where @xmath339 are the matrices of @xmath340 and @xmath341 are matrices on the arms
let @xmath342 be the transposition in the weyl group @xmath343 .
the following lemma is obvious .
[ l : wt - var ] the element @xmath342 permutes @xmath344 and @xmath345 .
the @xmath300-action is given by @xmath346 [ l : level1 ] each @xmath347 is extremal in @xmath348 .
moreover , for @xmath349 , @xmath350 is spanned by @xmath351 . by lemma
[ l : si(kml ) ] , @xmath191 corresponds to the partitions @xmath352 . since @xmath353
, @xmath191 lies in @xmath348 and @xmath354 .
@xmath191 must be extremal because of .
we observe that @xmath355 is a coherently @xmath34-invariant presentation with @xmath356 or @xmath357 depending on the sign of @xmath355 . here
@xmath358 is the trivial path corresponding to the vertex @xmath77 .
hence @xmath335 is @xmath34-invariant and spans @xmath350 .
let @xmath359 be the subgroups of lower - triangular matrices in @xmath360 .
it follows that for @xmath361 , @xmath351 is @xmath359-invariant , and @xmath362 is thus @xmath360-semi - invariant .
the latter statement is also clear from the fact that @xmath363 corresponds to an exceptional representation @xcite .
we will see in section [ s : cs ] that @xmath364 is our choice of an ( initial ) cluster .
consider the quiver @xmath365 @xmath366 we list below the concrete form for the semi - invariant functions @xmath367 . @xmath368
we follow mostly section 3 of @xcite .
the combinatorial data defining a cluster algebra is encoded in an _ ice _ quiver @xmath369 with no loops or oriented 2-cycles .
the first @xmath370 vertices of @xmath369 are designated as _
mutable _ ; the remaining @xmath371 vertices are called _
frozen_. if we require no arrows between frozen vertices , then such a quiver is uniquely determined by its _ @xmath372-matrix _
@xmath373 . it is a @xmath374 matrix given by @xmath375 [ d : qmu]let @xmath376 be a mutable vertex of @xmath369 .
the _ quiver mutation _
@xmath377 transforms @xmath369 into the new quiver @xmath378 via a sequence of three steps . 1 . for each pair of arrows @xmath379 ,
introduce a new arrow @xmath380 ( unless both @xmath381 and @xmath382 are frozen , in which case do nothing ) ; 2 . reverse the direction of all arrows incident to @xmath376 ; 3 . remove all oriented 2-cycles .
[ d : seeds ] let @xmath383 be a field containing @xmath66 .
a _ seed _ in @xmath383 is a pair @xmath384 consisting of an ice quiver @xmath369 as above together with a collection @xmath385 , called an _ extended cluster _ , consisting of algebraically independent ( over @xmath66 ) elements of @xmath383 , one for each vertex of @xmath369 . the elements of @xmath386 associated with the mutable vertices are called _ cluster variables _ ; they form a _ cluster_.
the elements associated with the frozen vertices are called _ frozen variables _ , or _ coefficient variables_. a _ seed mutation _
@xmath377 at a ( mutable ) vertex @xmath376 transforms @xmath384 into the seed @xmath387 defined as follows .
the new quiver is @xmath378 .
the new extended cluster is @xmath388 where the new cluster variable @xmath389 replacing @xmath390 is determined by the _ exchange relation _ @xmath391
we note that the mutated seed @xmath392 contains the same coefficient variables as the original seed @xmath384 .
it is easy to check that one can recover @xmath384 from @xmath392 by performing a seed mutation again at @xmath376 .
two seeds @xmath384 and @xmath392 that can be obtained from each other by a sequence of mutations are called _ mutation - equivalent _ , denoted by @xmath393 .
[ d : cluster - algebra ] the _ cluster algebra @xmath394 _ associated to a seed @xmath384 is defined as the subring of @xmath383 generated by all elements of all extended clusters of the seeds mutation - equivalent to @xmath384 .
note that the above construction of @xmath394 depends only , up to a natural isomorphism , on the mutation equivalence class of the initial quiver @xmath369 .
so we may drop @xmath386 and simply write @xmath395 .
an amazing property of cluster algebras is [ t : laurent ] any element of a cluster algebra @xmath394 can be expressed in terms of the extended cluster @xmath386 as a laurent polynomial , which is polynomial in coefficient variables .
since @xmath394 is generated by cluster variables from the seeds mutation equivalent to @xmath384 , theorem [ t : laurent ] can be rephrased as @xmath396 where @xmath397 $ ] .
note that our definition of @xmath398 is slightly different from the original one in @xcite , where the coefficient variables are inverted in @xmath399 $ ] .
the upper cluster algebra with seed @xmath384 is @xmath400 any ( upper ) cluster algebra , being a subring of a field , is an integral domain ( and under our conventions , a @xmath66-algebra ) .
conversely , given such a domain @xmath401 , one may be interested in identifying @xmath401 as an ( upper ) cluster algebra . as an ambient field @xmath383 , we can always use the quotient field @xmath402 .
the challenge is to find a seed @xmath384 in @xmath402 such that @xmath403 .
the following lemma is very helpful .
* corollary 3.7 ) [ l : rca ] let @xmath401 be a finitely generated unique factorization domain over @xmath66 .
let @xmath384 be a seed in the quotient field of @xmath401 such that all elements of @xmath386 and all elements of clusters adjacent to @xmath386 are irreducible elements of @xmath401 .
then @xmath404 . the original conclusion in ( * ?
* corollary 3.7 ) is that @xmath405 .
however , the proof implies this stronger result ( see the comment before the proof of ( * ? ? ? * proposition 3.6 ) ) .
[ d : wtconfig ] a _ weight configuration _
@xmath406 on an ice quiver @xmath369 is an assignment for each vertex @xmath381 of @xmath369 a ( weight ) vector @xmath407 such that for each mutable @xmath376 , we have that @xmath408 a _ mutation _
@xmath377 at a mutable vertex @xmath376 transforms @xmath406 into a weight configuration @xmath409 of the mutated quiver @xmath410 defined as @xmath411 by slight abuse of notation , we can view @xmath406 as a matrix whose @xmath381-th row is the weight vector @xmath407 . in matrix notation , the condition is equivalent to @xmath412 is a zero matrix .
it is not hard to see that for any weight configuration of @xmath369 , the mutation can be iterated .
[ d : full ] a weight configuration @xmath406 is called _ full _ if the null rank of @xmath413 is equal to the rank of @xmath406 .
the null space of @xmath413 is called the _ grading space _ of the cluster algebra @xmath395 . given a weight configuration @xmath406 of @xmath369 , we can assign a multidegree ( or weight ) to the ( upper ) cluster algebra @xmath394 by setting @xmath414 for @xmath415 .
then mutation preserves multihomogeneousity .
we say that this ( upper ) cluster algebra is @xmath406-graded , and denoted by @xmath416 .
we refer to @xmath417 as a graded seed .
in @xcite and @xcite , the mutation of quivers with potentials is invented to model the cluster algebras .
we emphasis here that ice quivers considered here can have arrows between frozen vertices .
following @xcite , we define a potential @xmath20 on an ice quiver @xmath369 as a ( possibly infinite ) linear combination of oriented cycles in @xmath369 .
more precisely , a _
potential _ is an element of the _ trace space _
@xmath418 $ ] , where @xmath419 is the completion of the path algebra @xmath420 and @xmath421 $ ] is the closure of the commutator subspace of @xmath419 .
the pair @xmath422 is an _ ice quiver with potential _ , or iqp for short . for each arrow @xmath423 , the _ cyclic derivative _
@xmath424 on @xmath425 is defined to be the linear extension of @xmath426 for each potential @xmath20 , its _ jacobian ideal _
@xmath427 is the ( closed two - sided ) ideal in @xmath419 generated by all @xmath428 .
jacobian algebra _
@xmath429 is the quotient algebra @xmath430 .
if @xmath20 is polynomial and @xmath429 is finite - dimensional , then the completion is unnecessary to define @xmath429 .
this is the case throughout this paper .
the key notion introduced in @xcite is the _ mutation _ of quivers with potentials and their decorated representations . for an ice quiver with _ nondegenerate
( see @xcite ) , the mutation in certain sense lifts " the mutation in definition [ d : qmu ] .
since we do not need mutation in an explicit way , we refer readers to the original text .
a _ decorated representation _ of a jacobian algebra @xmath431 is a pair @xmath432 , where @xmath433 , and @xmath434 is a finite - dimensional @xmath435-module .
let @xmath436 be the set of decorated representations of @xmath429 up to isomorphism .
there is a bijection between two additive categories @xmath436 and @xmath437 mapping any representation @xmath70 to its minimal presentation in @xmath438 , and the simple representation @xmath439 of @xmath435 to @xmath440 .
suppose that @xmath441 corresponds to a projective presentation @xmath442 .
the _ @xmath2-vector _
@xmath443 of a decorated representation @xmath441 is the weight vector of its image in @xmath444 , that is , @xmath445 .
a potential @xmath20 is called _ rigid _ on a quiver @xmath369 if every potential on @xmath369 is cyclically equivalent to an element in the jacobian ideal @xmath427 .
such a qp @xmath422 is also called _
rigid_. it is known ( * ? ? ?
* proposition 8.1 , corollary 6.11 ) that every rigid qp is @xmath4-acyclic , and the rigidity is preserved under mutations . in particular ,
any rigid qp is nondegenerate . from now on
, we assume that @xmath422 is a nondegenerate iqp .
let @xmath385 be an ( extended ) cluster . for a vector @xmath446 , we write @xmath447 for the monomial @xmath448 . for @xmath449 , we set @xmath450 where @xmath451 is the @xmath376-th row of the matrix @xmath373 , and let @xmath452 .
suppose that an element @xmath453 can be written as @xmath454 where @xmath455 is a rational polynomial not divisible by any @xmath456 , and @xmath457 .
if we assume that the matrix @xmath373 has full rank , then the elements @xmath458 are algebraically independent so that the vector @xmath459 is uniquely determined @xcite .
we call the vector @xmath459 the ( extended ) _ @xmath2-vector _ of @xmath460 .
definition implies at once that for two such elements @xmath461 we have that @xmath462 .
so the set @xmath463 of all @xmath2-vectors in @xmath464 forms a sub - semigroup of @xmath465 .
[ l : independent ] assume that the matrix @xmath373 has full rank .
let @xmath466 be a subset of @xmath464 with well - defined @xmath2-vectors .
if @xmath467 s are all distinct , then @xmath468 is linearly independent over @xmath66 . to any @xmath469 we associate the _ reduced _ presentation space @xmath470_+),p([-{{\sf g}}]_+)),\ ] ] where @xmath471_+$ ] is the vector satisfying @xmath471_+(u ) = \max({{\sf g}}(u),0)$ ] .
we denote by @xmath472 the cokernel of a general presentation in @xmath473 .
reader should be aware that @xmath472 is just a notation rather than a specific representation .
if we write @xmath474 , this simply means that we take a general presentation in @xmath473 , then let @xmath70 to be its cokernel . [
d : mu_supported ] a representation is called _ @xmath28-supported _ if its supporting vertices are all mutable .
a weight vector @xmath475 is called _ @xmath28-supported _ if @xmath472 is @xmath28-supported .
let @xmath476 be the set of all @xmath28-supported vectors in @xmath477 .
it turns out that for a large class of iqp the set @xmath476 are given by lattice points in some rational polyhedral cone .
such a class includes the ice hive quiver @xmath478 introduced in @xcite and the quiver @xmath42 to be introduced in the next section .
we define the _ generic character _
@xmath479 by @xmath480 where @xmath481 is the variety parameterizing @xmath482-dimensional quotient representations of @xmath70 , and @xmath483 denotes the topological euler - characteristic .
we note that if @xmath384 is @xmath406-graded , then @xmath484 is multihomogeneous of degree @xmath485 .
[ c : gcc ] suppose that iqp @xmath422 is non - degenerate and @xmath373 has full rank .
the generic character @xmath486 maps @xmath476 ( bijectively ) to a set of linearly independent elements in @xmath464 containing all cluster monomials .
[ d : model ] we say that an iqp @xmath422 _ models _ an algebra @xmath487 if the generic cluster character maps @xmath476 ( bijectively ) onto a basis of @xmath487 . if @xmath487 is the upper cluster algebra @xmath488 , then we simply say that @xmath422 is a _ cluster model_.
we define an ice quiver @xmath42 whose vertices are labeled by @xmath319 and @xmath320 where @xmath489 and @xmath149 .
we also use the convention that @xmath490 and @xmath491 , and notation that @xmath492 and @xmath493 . instead of @xmath494 ,
we write @xmath495 for the set of vertices of @xmath42 .
we define an involution on the set of vertices @xmath496 we will use the notation @xmath497 for both @xmath319 and @xmath498 . note that there is another involution @xmath499 .
we freeze the vertices with @xmath500 , i.e. , of label @xmath501 .
a vertex labeled by @xmath502 or @xmath503 is called a _
horizontal vertex _ ; a vertex labeled by @xmath504 is called a _ vertical vertex_. a vertex
is called _ special _ if @xmath505 .
any rest vertex is called _
general_. we draw arrows according to the following recipe .
@xmath506 however , we draw two ( type @xmath507 ) arrows instead of one from @xmath508 to @xmath509 . when actually drawing the quiver , as a convention we put horizontal ( resp .
vertical ) vertices horizontally as on the @xmath510-axis ( vertically as on the @xmath511-axis ) ; put positive ( resp .
negative ) vertices over ( below ) the @xmath510-axis .
the sign of the @xmath510-coordinate is determined by the sign of @xmath512 .
unfortunately , we have to bend the type @xmath507 arrows between special vertices .
an arrow is called _ nonnegative _ ( resp .
_ nonpositive _ ) if it lies on or above ( resp . on or below ) the @xmath510-axis . as an illustration , @xmath513 is display in figure [ f : diamond4 ] .
the type @xmath514 ( resp . @xmath372 and
@xmath507 ) arrows are in red ( resp .
blue and yellow ) .
@xmath515 the next two lemmas are straightforward .
the @xmath372-matrix of @xmath42 is of full rank .
recall the weight vectors @xmath516 defined in section [ s : fkm ] .
[ l : wc ] the assignment @xmath517 on the vertices of @xmath42 defines a weight configuration @xmath518 on @xmath42 .
moreover , @xmath518 can be extended to a full weight configuration @xmath519 by assigning @xmath520 to @xmath497 .
let @xmath61 ( resp . @xmath521 and @xmath522 ) denote the sum of all type @xmath514 ( @xmath372 and @xmath507 ) arrows .
let @xmath523 and @xmath524 denote the sum of all nonnegative and nonpositive type @xmath514 arrows , and similar for @xmath525 and @xmath526 .
we put the potential @xmath527 on the quiver @xmath42 .
then the jacobian ideal is generated by the elements @xmath528 we say a path @xmath370 in @xmath42 _ nonnegative _ ( resp . _ nonpositive _ ) if all its composite arrows are nonnegative ( nonpositive )
. such a path can be ( uniquely ) written as @xmath529 by . moreover , if a path is equivalent to another path in the jacobian algebra @xmath530 , then they must have the same length unless they are equivalent to the zero path .
the iqp @xmath57 is rigid .
we need to show that every cycle up to cyclic equivalence is zero in the jacobian algebra . by cyclic equivalence and relations , cycles of form @xmath531 can be written as @xmath532 for some frozen @xmath381 , which is zero in the @xmath533 by .
now let @xmath250 be any cycle in @xmath42 .
we must have that the numbers of three types of arrows are the same in @xmath250 .
we can write @xmath250 as @xmath534 , where each @xmath535 is either nonnegative or nonpositive .
so each @xmath535 can be written as @xmath536 .
we first assume that none of @xmath537 is zero .
we observe that @xmath538 can change order except for @xmath539 and @xmath540 with @xmath382 on the @xmath510-axis .
but these two cases can not happen simultaneously in the procedure of moving @xmath521 right and moving @xmath61 left in @xmath536 .
so we can get at least one cycle of form @xmath531 .
we have treated the cases when @xmath541 and when some @xmath535 have all types of arrows .
suppose that @xmath542 . since @xmath250 is a cycle , there is some @xmath535 such that @xmath543 with both @xmath544 and @xmath545 on the @xmath510-axis .
we can also assume that @xmath535 has only two types of arrows .
but @xmath535 can not be of form @xmath546 or @xmath547 , because otherwise the arrow preceding or succeeding @xmath535 must be of the third type on the @xmath510-axis .
so let us assume that @xmath548 .
then by , we can write @xmath535 as @xmath549 , which is zero in @xmath533 by . let @xmath370 be a path @xmath550 in a quiver @xmath369 , where @xmath551 s may not be all distinct .
we can attach a _
representation @xmath552 of @xmath369 with its basis elements @xmath553 labeled by @xmath551 .
the action of arrows on @xmath552 is given by @xmath554 and @xmath555 otherwise . a _ tri - broken path _ from @xmath556 to @xmath497 denoted by @xmath557 is the path @xmath558 we also define @xmath559 and @xmath560 as the paths @xmath561 note that a tri - broken path may not be literally broken thrice in the figure ( see figure [ f : diamond4 ] ) . the module @xmath562 has dimension vector @xmath563 note that the @xmath564 because the path passes @xmath565 twice .
for a subpath @xmath566 of @xmath370 of form @xmath567 , we use the notation @xmath568 for the _ quotient _ path @xmath569 . note that there is one - to - one correspondence between subpaths of this form and subrepresentations of @xmath552 .
conversely for @xmath570 , we write @xmath571 for @xmath566 .
let @xmath572 be the category of ( not necessarily finite - dimensional ) modules of @xmath533 .
the algebra @xmath533 is in fact finite - dimensional ( corollary [ c : fdjl ] ) , but this is not clear at this stage .
[ l : proj_fp ] we have the following projective presentations in @xmath572 . @xmath573 we only prove the first one .
the rest is quite obvious . without loss of generality , we only deal with the positive cases . for any path @xmath370 in @xmath42 , we denote by @xmath574 $ ] its equivalence class in @xmath533 .
let @xmath370 be a path such that @xmath574\in p_{i;j , k}^\pm$ ] is not in the image of @xmath64 .
it is easy to see that the first arrow of @xmath370 must be of type @xmath514 unless @xmath370 is trivial .
if the second arrow of @xmath370 is of type @xmath372 , then @xmath574 $ ] lies in the image of @xmath64 .
if the second arrow of @xmath370 is of type @xmath507 , then by @xmath370 is equivalent to a zero path .
inductively using and , we can show that the maximal nonnegative subpath of @xmath370 is of form @xmath575 .
if @xmath576 is nontrivial , then @xmath577 .
it is easy to see that the first arrow of @xmath566 must be of type @xmath507 unless @xmath566 is trivial . if the second arrow of @xmath566 is of type @xmath514 or @xmath372 , then by or we get a contradiction on the statement of the last paragraph .
inductively using and ( or ) we can show that the maximal nonpositive subpath of @xmath566 is of form @xmath578 .
if @xmath579 is nontrivial , then @xmath580 .
it is easy to see that the first arrow of @xmath581 must be of type @xmath372 unless @xmath581 is trivial . if the second arrow of @xmath581 is of type @xmath514 or @xmath507 , then by or we get a contradiction on the statement of the last paragraph . inductively using and ( or )
we can show that @xmath581 is of form @xmath582 .
the conclusion is that @xmath370 must be a subpath of the tri - broken path @xmath583 .
this is exactly what we desire . by a similar argument of lemma [ l : proj_fp ]
, we have that [ l : injres_fp ] we have the following injective presentations @xmath572 . @xmath584 [ c : fdjl ] the jacobian algebra @xmath533 is finite - dimensional .
first we show by induction on @xmath252 that all @xmath585 and @xmath586 are finite - dimensional for @xmath587 .
@xmath588 and @xmath589 are the 1-dimensional simple module .
then we use , and , to finish the induction .
next we claim that each @xmath590 and @xmath591 is finite - dimensional for @xmath592 .
since @xmath585 and @xmath586 is finite - dimensional for @xmath587 , there are only finitely many paths in @xmath590 ( dual paths in @xmath591 ) passing @xmath593 .
so our claim is equivalent to that @xmath590 and @xmath591 is finite - dimensional if we restrict to the subquiver where all vertices @xmath497 with @xmath594 . but
this can be showed using a similar argument for @xmath500 . by peeling " off the quiver ,
inductively we can show that each @xmath585 is finite - dimensional so that the jacobian algebra is finite - dimensional .
a vertex @xmath381 is called _ maximal _ in a representation @xmath70 if all its subrepresentations are not supported on @xmath381 .
* lemma 6.5 ) [ l : hom=0 ] suppose that a representation @xmath300 contains a maximal vertex .
let @xmath474 , then @xmath595 if and only if @xmath596 for all subrepresentations @xmath597 of @xmath300 . to simplify the notation , for a frozen vertex @xmath587 ( resp .
@xmath598 and @xmath599 ) , we set @xmath600 ( resp . @xmath601 and @xmath602 ) .
we note that every @xmath603 contains a maximal vertex .
[ l : tiequi ] let @xmath474 , then @xmath604 for each frozen @xmath381 if and only if @xmath605 for each frozen @xmath381 . since each subrepresentation of @xmath603 is also a subrepresentation of @xmath586 ,
one direction is clear .
conversely , let us assume that @xmath604 for each frozen @xmath381 .
we prove that @xmath605 by induction on @xmath66 . for @xmath606
, we have that @xmath607 . now
suppose that it is true for @xmath608 , that is , @xmath609 . by lemma [ l : injres_fp ] , @xmath610 is equivalent to @xmath611 .
let the notation @xmath612 stand for @xmath613 if @xmath370 is the path @xmath550 .
we define a cone @xmath614 by @xmath615 for the following paths @xmath370 . 1 . all strict subpaths of @xmath616 , 2 .
all subpaths of @xmath617 , 3 . and the trivial path @xmath618 .
[ t : lp_gl ] the set of lattice points @xmath619 is exactly @xmath620 . moreover , all the defining conditions of @xmath3 are essential .
due to lemma [ l : hom=0 ] and [ l : tiequi ] , it suffices to show that @xmath3 is defined by @xmath596 for all subrepresentations @xmath597 of @xmath603 and all @xmath603 .
we notice that these defining conditions are the union of the defining conditions of but the latter conditions are redundant because @xmath621 and @xmath622 is the dimension vector of a subrepresentation of @xmath623 . for the last statement , we notice that each path @xmath370 in the defining conditions contains a unique frozen vertex .
so if one defining condition is a positive combination of others , their paths must share a common frozen vertex .
but this is clearly impossible . given a weight configuration @xmath406 of a quiver @xmath369 and a convex polyhedral cone @xmath624 , we define the ( not necessarily bounded ) convex polytope @xmath625 as @xmath626 cut out by the hyperplane sections @xmath627 .
let @xmath367 be the schofield s semi - invariant function @xmath351 .
we consider @xmath629 as our choice of the initial cluster . [ l:2arm ] a general representation
@xmath70 in @xmath630 has a following representative .
its matrices on the negative arm are @xmath631 ; on the positive arm are @xmath632 ; on the arrow @xmath633 is @xmath634 , where @xmath635 is a @xmath636 identity matrix for @xmath637 and @xmath85 is a zero column vector .
moreover , the evaluations of @xmath70 on @xmath344 and @xmath638 are the @xmath639\times[1,i - j]$]-minor and @xmath640\times [ j+1,i]$]-minor of @xmath641 . here the interval notation @xmath642 $ ] means the @xmath77-th up to @xmath252-th rows ( columns ) .
if we forget the arrow @xmath643 , then we get a quiver of type @xmath644 with the dimension vector shown below @xmath645 & 2\ar[r ] & \cdots \ar[r ] & l\ar[r ] \ar[r ] & l \ar[r ] & \cdots \ar[r ] & 2 \ar[r ] & 1 } } \ ] ] this dimension vector decomposes canonically as @xmath646 , where @xmath647 this means that a general representation of that dimension vector is a direct sum of general representations of dimension @xmath648 , which is exactly what we desired
. the second statement can be easily verified by elementary matrix calculation .
we define a restriction map @xmath649 sending @xmath70 to @xmath650 defined by @xmath651 in the language of @xcite , the restriction map is the composition of two _ vertex removals _ at @xmath21 and @xmath5 . in particular , it follows from @xcite that [ l : embedding ] the restriction map @xmath649 induces an embedding @xmath652 . [ l : ag - independent ] the semi - invariants @xmath367 in @xmath44 are algebraically independent .
we consider a representation @xmath653 whose matrices on the negative arm are @xmath631 ; on the postive arm are @xmath632 ; on the arrows @xmath340 are the following block matrices @xmath654 where @xmath514 is an @xmath655 matrix in general position , @xmath656 is the row vector in generic variables @xmath657 , @xmath658 is the column vector in generic variables @xmath659 , @xmath660 are zero vectors of the same size as @xmath661 , and @xmath662 is another generic variable .
then according to lemma [ l:2arm ] , @xmath663 is easily seen to be a linear function in @xmath664 and @xmath662 .
similarly , @xmath665 is a linear function in @xmath666 and @xmath662 . in particular
, we see that the set @xmath667 is algebraically independent .
we will finish the proof by induction on @xmath5 . if @xmath668 , the statement is clear .
now we assume that the statement is true for @xmath669 . in view of lemma
[ l : embedding ] , we can view @xmath670 as a subset in @xmath671 .
now suppose that @xmath672 are algebraically dependent , that is , there is a non - zero polynomial @xmath673 $ ] such that @xmath674 its total degree on @xmath675 must be strictly positive because elements in @xmath670 are algebraically independent .
we evaluate @xmath370 at the representation @xmath70 .
since @xmath676 are algebraically independent , we conclude that there is some function @xmath677 vanishes at @xmath70 , where @xmath678 is the subring generated by @xmath670 . by lemma [ l:2arm ] and [ l : embedding ] , the restriction of @xmath70 on the subquiver @xmath679 is a general representation of dimension @xmath680 .
but a general representation in @xmath681 is stable , so there is no such @xmath682 .
we get a contradiction .
recall the weight configuration @xmath518 for @xmath42 . for any mutable vertex @xmath683 , we set @xmath684 .
we have that @xmath685 as before , we use the convention that @xmath686 and @xmath687 is the zero vector space . if we mutate @xmath688 at @xmath376 , then we get a corresponding mutated weight vector @xmath689 : @xmath690 for @xmath689 with @xmath376 positive , we associate a corresponding projective presentation @xmath691 : @xmath692 if @xmath693 is negative , we associate a projective presentation @xmath694 , where the first @xmath328 is the map defined in , and @xmath695 .
finally we can associate a semi - invariant function @xmath696 for each mutable vertex @xmath376 @xmath697 [ l : uinv ] for @xmath698 with @xmath699 , @xmath691 is coherently @xmath34-invariant so @xmath696 is @xmath34-invariant . by the symmetry we only need to prove for @xmath376 positive .
it is easy to see that @xmath700 .
then the assignment @xmath701 defines the algebraic group homomorphism @xmath702 as in definition [ d : ginv ] .
then our claim follows from lemma [ l : ginv ] .
[ l : su_irreducible ] any @xmath696 is an irreducible polynomial . according to lemma [
l : irreducible ] , it suffices to show that @xmath689 is indivisible and extremal in @xmath703 .
the first case is obvious because we observed that any @xmath704 must satisfy @xmath705 . for the second case
, we observe that any factor of @xmath696 is also semi - invariant because of ( * ? ? ?
* theorem 3.1 ) .
we restrict @xmath696 on the subvariety @xmath706 by lemma [ l:2arm ] , @xmath707 is dense in @xmath708 . we conclude that @xmath696 is irreducible if and only if its restriction on @xmath709 is irreducible .
but @xmath710 , which is clearly irreducible .
[ l : s ] we have the following relations @xmath711 due to symmetries , we only need to prove the first one when @xmath712 takes @xmath713 and @xmath349 .
we first assume that @xmath714 , i.e. , @xmath376 is a general vertex .
let @xmath715 we notice that according to lemma [ l : level1 ] and [ l : uinv ] , each @xmath716 is @xmath34-invariant with @xmath717-degree equal to @xmath718 . by corollary
[ c : dimuinv ] and lemma [ l : g=2 ] , we have that @xmath719 hence , we must have that @xmath720 , for some @xmath721 . to determine @xmath722
, we consider a special representation @xmath70 , whose matrices on the negative arm are @xmath631 ; on the positive arm are @xmath723 ; and on the arrow @xmath633 is @xmath634 as in lemma [ l:2arm ] .
the entries of matrix @xmath641 are all zeros except for the diagonal of the @xmath724\times [ 1,i - j+1 ] , [ j , i+1]\times[1,i - j+1]$ ] , and @xmath725\times [ 1,i - j]$]-submatrix .
the entries on the first and third diagonals are all @xmath726 s ; the entries on the second diagonal are all @xmath85 s except that @xmath727 and @xmath728 generic
. then @xmath729 and @xmath730 ( resp . @xmath731 and @xmath732 )
are given by the determinants of lower ( upper ) triangular matrices with diagonal all @xmath726 s .
it can be easily verified by the elementary matrix calculation that @xmath733 where @xmath734 and @xmath735 .
we get @xmath736 by solving a linear system . to show the first relation for @xmath737 , we observe that these two cases are in fact degenerated from the cases when @xmath714 .
to make this more precise , we recall that it suffices to show the identity evaluating at a general representation .
we take a general representation @xmath70 with a form as in lemma [ l:2arm ] .
we can see that the desired identity for @xmath738 can be specialized from the proven identity for @xmath739 by setting the first row of @xmath641 to be zeros .
the second relation can be easily verified by evaluating at a general representation @xmath70 as in lemma [ l:2arm ] .
@xmath740 from the second relation and the symmetry , we also get that @xmath741 moreover by setting @xmath742 and our convention , we get that @xmath743 [ p : contain ] the upper cluster algebra @xmath744 is a @xmath303-graded subalgebra of @xmath39 . due to lemma
[ l : ag - independent ] and [ l : wc ] , such an assignment does define a graded cluster algebra @xmath745 . by ( * ? ? ?
* theorem 3.17 ) , any @xmath39 is a ufd . due to lemma
[ l : su_irreducible ] and [ l : s].(2 ) , we can apply lemma [ l : rca ] and get the containment @xmath746 . by our construction and lemma
[ l : wt - var ] , the grading @xmath519 is consistent with the @xmath303-grading of @xmath39 .
we will see in remark [ r : strictupper ] that the upper cluster @xmath41 strictly contains the cluster algebra @xmath747 .
[ t : cs ] the semi - invariant ring @xmath39 is the graded upper cluster algebra @xmath744 .
moreover , the generic cluster character maps the lattice points in @xmath3 ( bijectively ) onto a basis of @xmath41 .
we put off the proof to the last section .
it is a rather complicated combinatorial proof using several results from @xcite and @xcite .
however , we still suggest readers carefully reading it because it tells the story of how we find the cluster structure .
our first result says that the weighted counting of @xmath2-vectors gives @xmath4-truncated kronecker products .
let @xmath748 be the extended part of the weight configuration @xmath749 , that is , @xmath750 .
[ p:2kp ] let @xmath46 be a weight of @xmath751 , then @xmath752_2.\ ] ] by theorem [ t : cs ] , the generic character maps the lattice points in @xmath47 bijectively to a basis in @xmath753 .
applying the character map @xmath106 to the equality in lemma [ l : si(kml ) ] , we get our desired result . note that by the symmetry we have that @xmath754 , though this is not clear from the defining condition of @xmath3 .
recall our definition that @xmath755 but @xmath756 .
[ t : kc ] let @xmath46 be a weight of @xmath751 and @xmath6 be a partition of length @xmath51 .
then @xmath757 this follows immediately from corollary [ l : si(kml)2 ] and theorem [ t : cs ] .
we knew that the semi - invariant ring @xmath758 is the upper cluster algebra associated to the quiver @xmath759 @xmath760 let us use proposition [ p:2kp ] to compute @xmath761_2 $ ] . in this case , we consider the weight @xmath762 .
we find that there are six lattice points in the @xmath2-vector cone @xmath763 .
we list them with their @xmath6-weights on the right .
of course , by the symmetry we only need half of them .
@xmath764 so @xmath761_2 = x^{(3,0)}+2x^{(2,1)}+2x^{(1,2)}+x^{(0,3 ) } = s_{(3)}+s_{(2,1)}$ ]
. now we use theorem [ t : kc ] .
we see that @xmath765 and @xmath766 .
it is clear that theorem [ t : kc ] is more efficient than proposition [ p:2kp ] .
[ ex : l=2 ] in this example , we study concretely the @xmath34-action on the initial cluster . to simplify the notation
, we relabeled the vertices of @xmath759 as follows .
@xmath767 the cluster variables @xmath768 are four known @xmath34-invariants .
let us compute the @xmath34-action on the other two .
we abuse @xmath376 for the unipotent element @xmath769 . according to the standard highest weight theory , if @xmath770 , then @xmath771 , where @xmath772 . by explicit calculation
, we find that @xmath773 where @xmath774 and @xmath775 .
using this we can easily check that @xmath776 and @xmath777 are also @xmath34-invariant .
the latter is the well - known _
hyperdeterminant_. moreover ,
they satisfy @xmath778 . note that @xmath779 and @xmath780 have well - defined @xmath2-vectors : @xmath781 their @xmath2-vectors are @xmath782 and @xmath783 .
since @xmath784 , they spanned their corresponding weight spaces .
note that @xmath785 , so the kronecker coefficients are not _
saturated_. does any @xmath303-graded @xmath34-invariant function in @xmath786 have a well - defined @xmath2-vector ? if a @xmath303-graded @xmath34-invariant function does have a well - defined @xmath2-vector , we will see that it must lie on the _ facet _
( i.e. , codimension-1 face ) of @xmath3 defined by @xmath787 .
this is really a facet due to the second statement of theorem [ t : lp_gl ] .
we define a ( full - dimensional ) subcone @xmath788 of the facet by adding one additional condition that @xmath789 .
this condition ensures that the @xmath6-weights of @xmath2-vectors in @xmath788 are dominant , so it is clearly necessary .
[ p : guinv ] all possible @xmath2-vectors of @xmath34-invariants span a full - dimensional subcone in @xmath788 .
let @xmath790 and @xmath791 , then we have that @xmath792 .
let @xmath793 be the set of all possible @xmath2-vectors in the @xmath34-invariants .
we explained in section [ s : cc ] that @xmath793 is a semigroup .
so by lemma [ l : independent ] the degree of the hilbert polynomial of @xmath794 is no less than the dimension of the cone @xmath795 .
we first claim that @xmath793 can not span a full - dimensional subcone of @xmath3 .
the quotient @xmath796 of the integral @xmath34-invariants is contained in the rational @xmath34-invariants @xmath797 . since the action of the @xmath726-dimensional group @xmath34 is non - trivial
, the transcendence degree of @xmath797 is one less than @xmath798 ( ( * ? ? ?
* corollary of theorem 2.3 ) ) .
our first claim follows .
now it suffices to find @xmath799 linearly independent @xmath2-vectors in @xmath793 satisfying @xmath787 .
we have seen in example [ ex : l=2 ] that @xmath800 lies in @xmath793 . besides this one , we consider the following list of @xmath2-vectors .
@xmath801 it is easy to verify by theorem [ t : lp_gl ] that these @xmath2-vectors are in the cone @xmath3 .
so they correspond to weights orthogonal to @xmath16 . by calculation , those weights are @xmath802 by checking with the euler form of @xmath751 , we can verify that they correspond to real schur roots , and thus correspond to @xmath34-invariants by lemma [ l : gsreal ] .
finally we observe that together with @xmath803 for @xmath349 , these @xmath2-vectors form a linearly independent set because each @xmath2-vector has a
leading term " @xmath804 with respect to the lexicographical order .
the index of each leading term corresponds to a vertex of @xmath42 except for @xmath508 .
unfortunately the kronecker coefficient @xmath305 is _ not _ counted by lattice points in @xmath805 .
for example , the @xmath2-vector @xmath806 corresponds to @xmath807 ( see example [ ex : l=2 ] ) but @xmath785 .
let @xmath808 be a _ pointed _ rational cone with _ generator _
@xmath809 , that is , points in @xmath810 are positive linear combination of @xmath811 and there exists a hyperplane @xmath812 for which @xmath813 . the cone @xmath810 is called _ simplicial _ if it has precisely @xmath814 linearly independent generators . a simplicial cone
is called _ unimodular _ if its generators form a @xmath815-basis of @xmath816 .
a collection @xmath817 of simplicial @xmath814-dimensional cones is a _ triangulation _ of the @xmath814-dimensional cone @xmath810 if it satisfies : 1 .
2 . for any @xmath819 is a common face of @xmath820 and @xmath821 .
we say that @xmath810 can be triangulated using no new generators if there exists a triangulation @xmath817 such that the generators of any @xmath822 are generators of @xmath810 .
it is well - known ( * ? ? ?
* section 2.6 ) that 1 .
any pointed rational cone can be triangulated into unimodular cones ; 2 .
any pointed cone can be triangulated into simplicial cones using no new generators
. however , in general pointed rational cones may not be triangulated into unimodular cones using no new generators .
let @xmath823 be the cone generated by all @xmath2-vectors of the upper cluster algebra @xmath464 .
it is clear that for any frozen vertex @xmath824 , @xmath825 is a generator of @xmath823 .
we denote by @xmath826 the set @xmath827 .
an _ extremal unimodular fan _ of @xmath464 is a triangulation of @xmath823 into unimodular cones with generators @xmath828 , where each @xmath829 is a generator of @xmath823 .
the set of generators of each unimodular cone is called a _
fake cluster_. we can guarantee neither the existence nor the uniqueness of the extremal unimodular fan of an upper cluster algebra .
but when it exists , we shall see that it is a crucial tool to present an upper cluster algebra with infinitely many clusters . here
at least we can easily see that the upper cluster algebra is generated by elements of extremal @xmath2-vectors if lattice points in @xmath823 index a basis of @xmath464 .
this idea will be elaborated somewhere else .
we believe that an extremal unimodular fan exists for a huge class of upper cluster algebras including @xmath41 .
each @xmath41 has an extremal unimodular fan . in some cases
, the extremal unimodular fan agrees with the cluster fan . as a nontrivial example , the cluster algebra @xmath830 @xcite is one such case . in general , even if the cluster algebra has only finitely many clusters , a fake cluster may _ not _ be a cluster . for any unimodular cone @xmath831 with generators @xmath832 ,
the generating function for the set of integer points in @xmath831 is a multiple geometric series @xmath833 using this observation , it is easy to compute the multigraded hilbert series of the upper cluster algebras .
let us come back to example [ ex : l=2 ] . recall that @xmath834 and @xmath835 .
we set @xmath836 .
reader should be aware that @xmath837 is _ not _ a cluster variable , because @xmath838 is not a _ real _ @xmath2-vector in the sense of ( * ? ? ?
* definition 4.6 ) . to simplify notation ,
we will write @xmath839 for the weight vector @xmath840 . 1 .
@xmath758 is minimally presented by @xmath841/(x_2 ^ 2x_5+x_1 ^ 2x_6+x_3x_4- x_1x_2x_2^\circ).\ ] ] its @xmath46-graded character is equal to @xmath842 2 .
@xmath843 is minimally presented by @xmath844/ ( w_1 ^ 2 + 4x_3x_4x_5 - w_2x_1 ^ 2).\ ] ] its @xmath303-graded hilbert series is equal to @xmath845 ( 1 ) .
the cone @xmath763 admits a triangulation by four unimodular cones .
they are spanned by the unions of @xmath846 with the following four pairs of vectors @xmath847 by definition , this is an extremal unimodular fan . in particular
, @xmath758 is generated by the initial cluster together with @xmath848 and @xmath837 .
the weights of @xmath849 are all extremal in @xmath850 , but @xmath851 .
so it is minimally generated by @xmath852 , and satisfies @xmath853 .
since @xmath854 is irreducible and the krull dimension of @xmath758 is @xmath855 , we conclude that it is the only relation . using the extremal unimodular fan and the inclusion - exclusion , we get the generating function for the lattice points in @xmath763 : @xmath856 by theorem [ t : cs ] , we remain to transfer the @xmath2-vector weights to the @xmath857-weights .
we do get the desired graded character because @xmath858 .
( 2 ) . using the omega package @xcite
, we can turn a rational expression for a formal series @xmath859 into a rational expression for @xmath860 .
applying to the graded character , we get the desired @xmath857-graded hilbert series .
let us consider the subalgebra @xmath861 generated by @xmath862 .
since the last five generators have linearly independent @xmath857-degrees , they are algebraically independent .
recall from example [ ex : l=2 ] , they satisfy the relation @xmath863 .
since @xmath682 is irreducible , we conclude that @xmath401 is presented by @xmath864/(r)$ ] .
now we rewrite the @xmath857-graded hilbert series as @xmath865 which is the same the hilbert series of @xmath401 .
hence , we must have that @xmath866 .
[ r : strictupper ] since the @xmath46-weight of @xmath837 is extremal , the non - cluster variable @xmath837 is contained in any homogeneous generating set of @xmath758 ( in fact @xmath867 ) .
this provides an example of an acyclic cluster algebra which is strictly contained in its upper cluster algebra .
this is different from the result in @xcite due to the difference in the definition of upper cluster algebras .
moreover , we believe that @xmath843 ( and @xmath868 in general ) is _ not _ an ( upper ) cluster algebra , though we do not have a proof for that . as our last example , we want to show readers that the cluster structure still exists even if the two flags have different length , i.e. , @xmath869 .
recall from @xcite that @xmath870 whenever @xmath871 .
so by the @xmath872-symmetry of the kronecker coefficients , the semi - invariant ring below controls all ( full ) kronecker products of @xmath873 if @xmath874 .
consider the quiver with dimension vector @xmath875 obtained from @xmath876 by removing the vertex @xmath877 and @xmath878 ( see below theorem [ t : tripleflag ] ) .
@xmath879 by proposition [ p : removal ] , we find that the semi - invariant ring @xmath880 is the upper cluster algebra of the following quiver .
the right one is a relabeling of the left one .
@xmath881{{$_{4;2 ^ 2}$ } } } } \qquad\qquad \ckrontwofourtwonum{{\framebox[1.2\width]{{$9$}}}}\ ] ] note that the ice quiver is not connected .
if we delete the frozen vertex @xmath882 , then the associated upper cluster algebra is the semi - invariant ring for the quiver with dimension vector @xmath883 : @xmath884 it suffices to study this semi - invariant ring .
we keep that @xmath885 , and let @xmath886 using @xcite we find that together with @xmath826 they constitute generators of the @xmath2-vector cone .
@xmath887 is minimally generated by @xmath888 except for @xmath889 these vectors all correspond to extremal weights of @xmath890 .
the @xmath2-vector cone admits a triangulation by @xmath891 unimodular cones .
the cones are spanned by the unions of @xmath826 with the following sets of vectors @xmath892 we have seen in the previous example that @xmath848 can be generated from others .
we do get the desired minimal generating set . using the extremal unimodular fan and the inclusion - exclusion
, we can get the generating function for the lattice points in @xmath893 . after transferring the weights as in the previous example
, we obtain the @xmath46-graded character of @xmath887 , which is a rather long expression . with the help of the omega package @xcite
, we can recover a old result in @xcite on the generating function @xmath894 , where the summation runs through all triple of partitions @xmath129 with @xmath895 and @xmath896 . with a little effort
, one can show that the ring @xmath897 is minimally generated by 13 invariants of @xmath898-degree equal to @xmath899
by corollary [ c : gcc ] , the image of the lattice points in @xmath48 under the generic character @xmath900 is linearly independent in @xmath901 for any weight @xmath46 . due to proposition [ p :
contain ] , it suffices to show that the dimension of @xmath901 is counted by the lattice points in @xmath48 .
we will prove this in three steps . in step one , we show that the semi - invariant ring of certain ( noncomplete ) triple flag is an upper cluster algebras , and describe its iqp model .
in step two , we use orthogonal projection to show that the semi - invariant ring of certain triple - flagged kronecker quiver representations is an upper cluster algebra , and describe its iqp model and @xmath2-vector cone . in step three , by embedding the quiver @xmath751 to the triple - flagged kronecker quiver , we show that there are required number of lattice points in @xmath48 .
we first recall the main result from @xcite .
the _ complete triple flag _ of length @xmath7 is the quiver with dimension vector @xmath902 as indicated below @xmath903 the _ ice hive quiver @xmath478 of size @xmath7 _ is an ice quiver in the plane with @xmath904 vertices arranged in a triangular grid .
we label the vertices as shown in figure [ fig : hive ] .
note that three vertices @xmath905 and @xmath906 are missing , and all boundary vertices are frozen .
there are three types of arrows : @xmath907 , and @xmath908 .
the quiver @xmath478 can be equipped with a rigid potential @xmath909 ( see @xcite for detail ) .
let @xmath911 be the unit vector in @xmath912 supported on the @xmath77-th vertex of the @xmath61-th arm of @xmath913 . by convention
@xmath914 is the zero vector .
we define a ( full ) weight configuration @xmath915 by the assignment @xmath916 on the vertices of @xmath478 .
next , we need to recall another result about _ vertex removal _ from @xcite .
let @xmath58 be a quiver , and @xmath173 a dimension vector of @xmath58 .
for a fixed vertex @xmath920 , let @xmath921 be the quiver such that @xmath922 , and @xmath923\mid h(a)=t(b)=r\},\ ] ] where @xmath924)=t(a)$ ] and @xmath925)=h(b)$ ] .
let @xmath926 be the restriction of @xmath173 on @xmath927 . for an iqp @xmath422 , _ deleting _ a set of vertices
@xmath928 is the operation that we delete all vertices in @xmath929 for @xmath369 , and set all incoming and outgoing arrows of @xmath930 to be zero in @xmath20 .
we denote the new iqp by @xmath931 . for an iqp @xmath422 , _ freezing _ a set of vertices @xmath932 is the operation that we set all vertices in @xmath932 to be frozen , but we do not delete arrows between frozen vertices .
we also keep the same potential @xmath20 .
the new quiver is denoted by @xmath933 .
a set of vertices @xmath932 is called _ attached _ to another set of vertices @xmath929 if @xmath934 ( resp .
@xmath935 ) whenever @xmath936 ( resp .
@xmath937 ) . for such a pair @xmath938
, we form a new iqp @xmath939 by deleting @xmath929 and freezing @xmath932 .
we denote by @xmath940 the restriction of @xmath406 on @xmath941 .
* theorem 3.7 and 5.17 ) [ t : removal ] suppose that @xmath182 is a naturally graded upper cluster algebra @xmath942 , and @xmath373 has full rank .
let @xmath682 be a vertex of @xmath58 satisfying @xmath943 and that @xmath944 for all @xmath824 except for a set of frozen vertices @xmath929 , where all @xmath945 ( or all @xmath946 ) .
then we have that @xmath947 is the naturally graded upper cluster algebra @xmath948 , where @xmath932 is attached to @xmath929 .
we start with the quiver with dimension vector @xmath949 .
we remove the last @xmath950 vertices ( i.e. , from the @xmath951-th to the @xmath952-th ) for each arm .
we denote the resulting quiver with dimension vector by @xmath953 .
we repeatedly apply theorem [ t : removal ] to those vertices in the order @xmath954 for each arm , and obtain an iqp denoted by @xmath955 .
let @xmath956 be the restriction of @xmath957 on @xmath958 .
roughly speaking , @xmath958 is a hexagonal subquiver of @xmath959 centered at the vertex @xmath960 with edge length equal to @xmath5 arrow length , and all boundary vertices frozen . when @xmath961 , the quiver is displayed in figure [ f : hex4 ] .
let @xmath58 be a finite quiver without oriented cycles . for two dimension vector @xmath966 , we write @xmath967 if @xmath968 for @xmath969 general in @xmath245 and @xmath178 .
we recall that an _ exceptional sequence of dimension vector _ @xmath970 is a sequence of real schur roots of @xmath58 such that @xmath971 for any @xmath972 .
it is called _ quiver _ if @xmath973 for any @xmath972 .
it is called _ complete _ if @xmath974 .
the _ quiver _
@xmath975 of a quiver exceptional sequence @xmath976 is by definition the quiver with vertices labeled by @xmath977 and @xmath978 arrows from @xmath979 to @xmath977 .
according to ( * ? ? ?
* theorem 4.1 ) , @xmath980 .
given an exceptional sequence of dimension vector @xmath981 with @xmath716 general in @xmath982 , we consider the ( right ) _ orthogonal subcategory _
@xmath983 defined by @xmath984 according to @xcite , this abelian subcategory is equivalent to the module category of another quiver denoted by @xmath985 .
we say that @xmath58 _ projects _ to @xmath985 through @xmath826 . in this case
there is a unique quiver exceptional sequence @xmath986 of @xmath58 such that the concatenation of two sequences @xmath987 is a ( complete ) exceptional sequence and @xmath988 .
we refer to @xcite for the construction of @xmath976 .
we define a linear isometry ( with respect to the euler forms ) @xmath989 by @xmath990 conversely we define a linear map @xmath991 left inverse to @xmath992 as the composition @xmath993 , where the first map @xmath994 is given by the recursion @xmath995 both maps induce a linear map on the weights , still denoted by @xmath992 and @xmath991 .
we write @xmath996 and @xmath997 for @xmath998 and @xmath999 .
to effectively compute @xmath997 later , we can use the formula @xmath1000 it turns out that the quiver @xmath1002 projects to the following triple - flagged kronecker quiver @xmath1003 through the exceptional sequence @xmath1004 , where @xmath1005 and @xmath1006 .
we prove this claim by explicitly putting the real schur roots in @xmath976 on the quiver of @xmath976 .
readers can easily check that @xmath1007 and @xmath976 is a quiver exceptional sequence with the prescribed quiver .
@xmath1008 where @xmath1009 and @xmath1010 .
we label the vertices and arms of this quiver as follows @xmath1011 let @xmath1012 be the dimension vector of @xmath1003 which is @xmath1013 along each arm
. we can easily check by that @xmath1014 .
suppose that @xmath182 is an upper cluster algebra @xmath1016 naturally graded by the weights of semi - invariants .
let @xmath1017 be a vertex of @xmath369 such that @xmath1018 corresponds to a real schur root @xmath1019 of @xmath58 .
we define the _ projection _ of the weight configuration @xmath1020 through @xmath1017 as the pair @xmath1021 , where @xmath1022 .
* theorem 3.7 and 5.6 ) [ t : project ] suppose that @xmath182 is a naturally graded upper cluster algebra @xmath1023 , and @xmath373 has full rank .
if @xmath1024 is real and extremal in @xmath1025 , then @xmath1019 is schur and @xmath1026 is the naturally graded upper cluster algebra @xmath1027 .
we note that if @xmath1029 is an exceptional sequence with each @xmath1030 extremal in @xmath238 , then each @xmath1031 is extremal in @xmath1032 .
so we can apply theorem [ t : project ] to the sequence of vertices @xmath1033 of @xmath958 corresponding to the exceptional sequence @xmath826 of @xmath1003 .
we obtain an iqp denoted by @xmath1034 .
it follows that the semi - invariant ring @xmath1035 is the graded upper cluster algebra @xmath1036 , where the projected weight configuration @xmath1037 is given by @xmath1038 we remind the readers of our convention that @xmath914 is the zero vector , and @xmath1039 .
the projected weight configuration @xmath1037 can be easily checked using . to prepare for our next step ,
we relabel the vertex of @xmath1040 according to the rule @xmath1041 ( see figure [ f : hathexagon ] for an example ) . moreover , instead of the quiver with dimension vector @xmath1042 , we consider its _ reflection - equivalent _ one . when a vertex @xmath250 is a sink or source for a quiver @xmath58 ( without potential ) , we have the _ bgp - reflection functor _ @xmath1043 ( see @xcite ) . at the level of @xmath1044-groups
, the functor @xmath1043 sends a dimension vector @xmath173 to @xmath1045 .
this induces an action on the weight vectors , sending a weight @xmath46 to @xmath1046 satisfying @xmath1047 where @xmath1048 is the number of arrows from @xmath250 to @xmath77 .
the semi - invariant ring @xmath182 is invariant under the bgp - reflection functors .
more precisely , for each positive arm , we apply a sequence of reflections ( from right to left ) @xmath1052 , where @xmath1053 .
we get the following quiver @xmath1054 with the same dimension vector ( denoted by @xmath1055 ) .
@xmath1056 it is easy to compute from that this sequence of reflection acts on weights by @xmath1057 after this sequence of reflection , the weight configuration @xmath1037 transforms into a new one denoted by @xmath1058 .
it can be verified by that @xmath1059 here , @xmath712 is the sign of @xmath1060 .
we keep the convention that @xmath1061 and @xmath1039 .
we summarize our result in step two .
the potential @xmath1065 is obtained from @xmath909 by deleting a few vertices , so it is given by the sum of all smallest " triangle cycles .
similar to the proof of ( * ? ? ?
* theorem 6.8 ) and theorem [ t : lp_gl ] , we can show that @xmath1066 is given by the lattice points inside a rational polyhedral cone @xmath1067 .
the cone @xmath1067 has a following description : @xmath1068 is in @xmath1067 if and only if @xmath1069 the quiver @xmath751 is a subquiver of @xmath1054 .
we make a convention that the arms of @xmath751 is identified with the arms of @xmath1054 labeled with @xmath1070 .
then a weight @xmath46 of @xmath751 can be extended by zeros to a weight of @xmath1054 still denoted by @xmath46 .
the semi - invariant ring @xmath786 can be naturally embedded into @xmath1071 : @xmath1072 where @xmath46 runs through all weights supported on @xmath751 .
to finish the proof , we need to show that there are the same number of lattice points in the polytopes @xmath48 and @xmath1073 . for this ,
let us recall that a matrix @xmath300 is called _ totally unimodular _ if all its minors equal to @xmath1074 , or @xmath1075 .
this implies that a totally unimodular matrix maps lattice points to lattice points .
if a nonsingular square matrix is totally unimodular , then by cramer s rule its inverse is also totally unimodular .
let @xmath300 be an @xmath1076 totally unimodular matrix with @xmath1077 . if @xmath300 is of full rank , then it has a totally unimodular right inverse as follows .
we take a full rank square submatrix @xmath597 of @xmath300 . without loss of generality , we can assume @xmath597 to be first @xmath0 columns of @xmath300 .
we extend the inverse of @xmath597 by zeros to an @xmath1078 matrix , which is our desired right inverse .
we recall ghouila - houri s characterization of totally unimodular matrix .
[ l : tu ] a matrix @xmath300 is totally unimodular if and only if each collection of columns of @xmath300 can be split into two parts so that the sum of the columns in one part minus the sum of the columns in the other part is a vector with entries only @xmath1074 , and @xmath1075 .
we define the linear map @xmath1081 as follows @xmath1082 we use the convention that @xmath1083 is the zero vector , so in particular @xmath1084 .
@xmath702 is clearly of full rank , so we remain to check that @xmath1085 ( 1 ) . to verify the total unimodularity , we observe that the matrix of @xmath702 can be partition into blocks indexed by @xmath252 and a sign @xmath712 . if @xmath1086 , the rows and columns of each block @xmath1087 are indexed by @xmath1088 or @xmath1089 . if @xmath1090 or @xmath5 , then the block has a single entry @xmath726 .
since no any two blocks have common rows or columns , it suffices to verify the total unimodularity on each block .
but this is rather trivial by lemma [ l : tu ] .
next , we write the definition of @xmath702 in coordinates : @xmath1094 . we find that @xmath1095 except for the following @xmath1096 we see that @xmath1097 ( resp .
@xmath1098 ) is a dimension vector of a subrepresentation of @xmath1099 ( resp .
@xmath1100 ) , and @xmath1101 is a sum of dimension vectors of a subrepresentation of @xmath1099 and a subrepresentation of @xmath1100 .
now we can easily verify the following @xmath1102 where @xmath1103 if @xmath1104 , @xmath1105 if @xmath1106 , @xmath1107 if @xmath1108 ; otherwise @xmath1109 for @xmath1110 .
we can rewrite the first three equations as follows : @xmath1111 where @xmath1112 is a dimension vector of some submodule of @xmath1113 , @xmath1114 and @xmath1115 are dimension vectors of some submodules of @xmath1099 .
note that the negative term of the third equation causes cancelation only .
we observe that all dimension vectors of submodules of @xmath1113 and @xmath1099 arise this way . by symmetry
, we have the similar equations for @xmath1116 , and @xmath1117 .
we conclude that @xmath1118 satisfies the defining condition of @xmath1067 if and only if @xmath2 satisfy the defining conditions of @xmath3 .
this implies that @xmath702 maps @xmath48 ( bijectively ) onto @xmath1073 .
therefore , the two polytopes have the same number of lattice points . 99 a. berenstein , s. fomin , a. zelevinsky , _ cluster algebras .
iii . upper bounds and double bruhat cells , _ duke math . j. 126 ( 2005 ) , no .
1 , 152 . h. derksen , j. fei , _ general presentations of algebras , _ adv .
( 2015 ) , 210237 .
h. derksen , j. weyman , _ semi - invariants of quivers and saturation for littlewood - richardson coefficients , _
( 2000 ) , no .
3 , 467479 . h. derksen , j. weyman , _ the combinatorics of quiver representation , _ ann .
fourier ( grenoble ) 61 ( 2011 ) , no .
3 , 10611131 .
h. derksen , j. weyman , a. zelevinsky , _ quivers with potentials and their representations i , _ selecta math .
14 ( 2008 ) , no .
1 , 59119 . h. derksen , j. weyman , a. zelevinsky , _ quivers with potentials and their representations ii , _
( 2010 ) , no .
3 , 749790
. m. domokos , a. zubkov , _ semi - invariants of quivers as determinants , _ transform .
groups 6 ( 2001 ) , no .
y. dvir , _ on the kronecker product of @xmath1119 characters , _ j. algebra 154 ( 1993 ) , 125140 .
j. fei , _ on the @xmath269-invariant modules , _ arxiv:1402.6021 .
cluster algebras and semi - invariant rings i. triple flags , _ arxiv:1411.4693 . j. fei , _ cluster algebras and semi - invariant rings ii .
projections , _ arxiv:1508.05563 .
s. fomin , p. pylyavskyy , _ tensor diagrams and cluster algebras , _ arxiv:1210.1888
. s. fomin , a. zelevinsky , _ cluster algebras .
i. foundations , _ j. amer .
( 2002 ) , no .
2 , 497529 . s. fomin , a. zelevinsky , _ cluster algebras .
coefficients , _ compos . math . 143
( 2007 ) , no .
1 , 112164 .
w. fulton , _ intro to toric variety , _ a first course .
graduate texts in mathematics , 129 .
readings in mathematics .
springer - verlag , new york , 1991 .
w. fulton , j. harris , _ representation theory , _ a first course .
graduate texts in mathematics , 129 .
readings in mathematics .
springer - verlag , new york , 1991 .
k. igusa , k. orr , g. todorov , j. weyman , _ cluster complexes via semi - invariants , _ compositio math . 145
( 2009 ) , 10011034 . v. g. kac , _ infinite root systems , representations of graphs and invariant theory i , _ invent .
56 ( 1980 ) , 5792 .
king , _ moduli of representations of finite - dimensional algebras , _ quart .
oxford ser .
( 2 ) 45 ( 1994 ) , no .
180 , 515530 .
a. knutson , t. tao , _ the honeycomb model of @xmath1120 tensor products .
i. proof of the saturation conjecture , _
soc . 12 ( 1999 ) , no .
4 , 10551090 .
i. g. macdonald , _ symmetric functions and hall polynomials , _ oxford math .
monographs , 2nd enlarged ed . 1995 .
g. muller , _
@xmath1121 for locally acyclic cluster algebras , _ sigma 10 ( 2014 ) , 094 , 8 pages .
mulmuley , _ geometric complexity theory vi : the flip via saturated and positive integer programming in representation theory and algebraic geometry .
_ m. herceg , m. kvasnica , c.n .
jones , m. morari , multi - parametric toolbox ( mpt ) , http://people.ee.ethz.ch/~mpt/. i. pak , g. panova , _ unimodality via kronecker products , _ j. algebraic combin . 40 ( 2014 ) , no .
4 , 11031120 .
p. plamondon , _ generic bases for cluster algebras from the cluster category , _ int math res notices ( 2013 ) 2013 ( 10 ) : 23682420
. j. patera , r. t. sharp , _ generating functions for plethysms of finite and continuous groups , _ j. phys . a 13 ( 1980 ) ,
2 , 397416 . v. l. popov , e. b. vinberg , _ invariant theory , in : algebraic geometry .
iv , encyclopaedia of mathematical sciences , _ vol .
55 , springer - verlag , berlin , 1994 , 123284 .
a riese ( with g.e .
andrews , p paule ) , _ omega package , _ risc linz , 2008 , http://www.risc.jku.at/research/combinat/software/ergosum/risc/omega.html .
a. schofield , _ semi - invariants of quivers , _ j. london math
( 2 ) 43 ( 1991 ) , no .
3 , 385395 .
a. schofield , _ general representations of quivers , _ proc .
london math .
( 3 ) 65 ( 1992 ) , no .
1 , 4664 .
a. schrijver , _ theory of linear and integer programming , _ wiley - interscience series in discrete mathematics .
john wiley & sons , ltd . ,
chichester , 1986 .
a. schofield , m. van den bergh , _ semi - invariants of quivers for arbitrary dimension vectors , _ indag .
12 ( 2001 ) , no .
1 , 125138 .
|
we relate the @xmath0-truncated kronecker products of symmetric functions to the semi - invariant rings of a family of quiver representations .
we find cluster algebra structures for these semi - invariant rings when @xmath1 .
each @xmath2-vector cone @xmath3 of these cluster algebras controls the @xmath4-truncated kronecker products for all symmetric functions of degree no greater than @xmath5 . as a consequence , each relevant kronecker coefficient is the difference of the number of the lattice points inside two rational polytopes .
we also give explicit description of all @xmath3 s . as an application
, we compute some invariant rings . [ multiblock footnote omitted ]
|
finding unconventional superconductors and understanding their superconducting mechanism is one of the main themes in condensed matter physics @xcite .
the term unconventional " firstly means the superconducting pairing mechanism is not phonon - mediated .
this usually manifests as a superconducting dome neighbouring a magnetic order in the phase diagram , and spin fluctuations are considered as the major pairing glue @xcite .
secondly , the term unconventional " means the wave function of cooper pairs is not @xmath4-wave .
symmetry imposed nodes ( gap zeros ) are often observed , such as in @xmath5-wave cuprate superconductors and heavy - fermion superconductor cecoin@xmath6 @xcite , and in @xmath7-wave superconductor sr@xmath8ruo@xmath0 @xcite .
note that the iron - based superconductors are exceptions , likely with the form of @xmath9-wave @xcite .
the superconducting gap symmetry and structure provide important clue to the underlying pairing mechanism .
unconventional superconductivity usually resides in quasi - two - dimensional ( q2d ) compounds , such as cuprate and iron - based superconductors , cecoin@xmath6 , sr@xmath8ruo@xmath0 , and organic superconductors @xmath10-(bedt - ttf)@xmath8x @xcite . when further reducing the dimensionality , namely in quasi - one - dimensional ( q1d ) superconductors represented by the organic compounds ( tmtsf)@xmath8x ( x = pf@xmath11 , clo@xmath0 ) @xcite , the pairing symmetry and mechanism are also likely unconventional @xcite .
in this sense , low dimensionality is important for the appearance of unconventional superconductivity @xcite .
recently , the ternary telluride ta@xmath0pd@xmath1te@xmath2 @xcite was found to be a new layered superconductor with q1d characteristics @xcite .
the @xmath12 is about 4.5 k at ambient pressure .
it has relatively flat ta@xmath13pd@xmath13te layers in the ( @xmath1403 ) plane , which contains pdte@xmath8 , tate@xmath1 , and ta@xmath8te@xmath0 chains along the crystallographic @xmath15 axis , as illustrated in fig .
[ crystal ]
. it will be very interesting to check whether there exists unconventional superconductivity in this low - dimensional compound . in this paper , we present the low - temperature thermal conductivity measurements of ta@xmath0pd@xmath1te@xmath2 single crystal down to 80 mk , which clearly demonstrates that there are nodes in the superconducting gap .
furthermore , a superconducting dome in the temperature - pressure phase diagram is revealed by resistivity measurement under pressures up to 21.9 kbar .
these results suggest unconventional superconductivity in ta@xmath0pd@xmath1te@xmath2 .
we discuss the possible origin of this novel superconducting state .
[ t ] crystal structure of ta@xmath0pd@xmath1te@xmath2 .
( a ) a view parallel to the @xmath16 plane .
the compound crystallizes in space group @xmath172/m with a monoclinic unit cell of @xmath18 = 17.687(4 ) , @xmath15 = 3.735(1 ) , @xmath19 = 19.510(4 ) , and @xmath20 = 110.42@xmath21 .
the crystal structure has relatively flat ta@xmath13pd@xmath13te layers in the ( @xmath1403 ) plane , which is the largest natural surface of as - grown single crystals . for convenience ,
we define @xmath22 direction as [ 301 ] , so that the ( @xmath1403 ) plane is the @xmath23 plane .
@xmath24 is the direction perpendicular to the @xmath23 plane .
the pd atoms are octahedrally coordinated , forming edge - sharing pdte@xmath8 chains along the @xmath15 axis .
( b ) a three - dimensional perspective view along the @xmath15 axis .
the pdte@xmath8 chains are separated by tate@xmath1 chains and ta@xmath8te@xmath0 double chains.,title="fig:",width=321 ]
single crystals of ta@xmath0pd@xmath1te@xmath2 were grown by a self - flux method @xcite .
the shiny crystals in flattened needle shape have the longest dimension along @xmath15 axis ( the chain direction ) . the largest natural surface with typical dimensions of 2.5 @xmath25 0.25 mm@xmath26 is in ( @xmath1403 ) plane , which is the @xmath23 plane in fig . [ crystal](a ) .
the thickness along the @xmath24 direction is about 0.1 mm .
the dc magnetization measurements were performed in a squid ( mpms , quantum design ) , with an applied field of @xmath27 oe parallel to the @xmath15 direction .
four contacts were made directly on the sample surfaces with silver paint , which were used for both resistivity and thermal conductivity measurements along the @xmath15 direction at ambient pressure .
the resistivity was measured in a @xmath28he cryostat from 300 k to 2 k , and in a @xmath29he cryostat down to 0.3 k. the thermal conductivity was measured in a dilution refrigerator , using a standard four - wire steady - state method with two ruo@xmath8 chip thermometers , calibrated _ in situ _ against a reference ruo@xmath8 thermometer . for resistivity measurements under pressure ,
the contacts were made with silver epoxy .
samples were pressurized in a piston - cylinder clamp cell made of be - cu alloy , with daphne oil as the pressure media .
the pressure inside the cell was determined from the @xmath12 of a tin wire .
[ t ] ( a ) the dc magnetization at @xmath27 oe for ta@xmath0pd@xmath1te@xmath2 single crystal , with both zero - field - cooling ( zfc ) and field - cooling ( fc ) processes .
( b ) the resistivity @xmath30 along the @xmath15 direction of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s1 ) in zero field
. the data between 7 and 25 k can be well fitted to @xmath30 = @xmath31 + @xmath32 , giving residual resistivity @xmath31 = 3.96 @xmath33 cm and @xmath34 = 2.26 , as shown in the inset .
, title="fig:",width=226 ]
figure [ resistivity](a ) shows the typical low - temperature dc magnetization of ta@xmath0pd@xmath1te@xmath2 single crystal . with zero - field - cooling process
, a sharp diamagnetic superconducting transition is observed at @xmath35 k. in fig .
[ resistivity](b ) , the resistivity of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s1 ) in zero field is plotted .
the resistivity decreases smoothly from room temperature to @xmath12 . fitting the data between 7 and 25 k to @xmath30 = @xmath31
+ @xmath32 gives residual resistivity @xmath36 @xmath33 cm and @xmath34 = 2.26 .
the @xmath12 = 4.3 k is defined by @xmath37 = 0 , which agrees well with the magnetization measurement .
[ t ] ( a ) low - temperature resistivity of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s1 ) in magnetic field @xmath38 up to 3 t. ( b ) the upper critical field @xmath39 of sample s1 , defined by @xmath40 .
the dashed line is a guide to the eye , which points to @xmath41 t. the inset shows the field dependence of @xmath31 .
( c ) the @xmath39 of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s6 ) for @xmath42 , @xmath22 , and @xmath24 . the extrapolated @xmath39 values at zero temperature are 7.9 , 4.8 , and 3.2 t , respectively .
, title="fig:",width=226 ] to determine the upper critical field @xmath39(0 ) of ta@xmath0pd@xmath1te@xmath2 , we measure the resistivity of sample s1 down to 0.3 k in various magnetic fields along @xmath24 direction .
figure [ fieldresistivity](a ) shows the low - temperature resistivity in fields up to 3 t. the temperature dependence of @xmath39 , defined by @xmath40 on the resistivity curves in fig .
[ fieldresistivity](a ) , is plotted in fig . [ fieldresistivity](b ) .
the dashed line is a guide to the eye , which points to @xmath43 t. the inset shows the field dependence of @xmath31 , which manifests positive magnetoresistance , with @xmath31(2 t ) = 4.54 @xmath33 cm . the anisotropy of @xmath39 along @xmath15 , @xmath22 , and @xmath24 directions for sample s6 is shown in fig .
[ fieldresistivity](c ) .
the resistivity data of sample s6 are not shown , and the @xmath39 is also defined by @xmath40 .
along three directions , @xmath44 7.9 , 4.8 , and 3.2 t are estimated from fig .
[ fieldresistivity](c ) .
the initial @xmath39 slopes are -1.61 , -0.85 , and -0.53
t / k , which corresponds to the @xmath39 ratio of @xmath45 near @xmath12 for @xmath46 . according to anisotropic ginzburg - landau ( gl ) theory @xmath47 @xcite , the resistivity ratio @xmath48 is roughly estimated .
this anisotropy is consistent with the quasi-1d structure of layered ta@xmath0pd@xmath1te@xmath2 .
note that this ratio is much smaller than those of quasi-1d superconductors limo@xmath11o@xmath49 and ( tmtsf)@xmath8pf@xmath11 @xcite .
low - temperature heat transport is an established bulk technique to probe the superconducting gap structure @xcite .
the thermal conductivity results of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s1 ) are presented in fig .
[ thermal conductivity ] .
figure [ thermal conductivity](a ) shows the temperature dependence of thermal conductivity in magnetic field @xmath38 up to 2 t , plotted as @xmath50 vs @xmath51 .
the thermal conductivity at very low temperature can be usually fitted to @xmath50 = @xmath18 + @xmath52 @xcite , in which the two terms @xmath53 and @xmath54 represent contributions from electrons and phonons , respectively .
the power @xmath55 is typically between 2 and 3 , due to specular reflections of phonons at the boundary @xcite . since all the curves in fig .
[ thermal conductivity](a ) are roughly linear , we fix @xmath55 to 2 .
the low values of @xmath55 have been previously observed in several superconductors , for example , cu@xmath56tise@xmath8 ( @xmath57 2.27 ) and kfe@xmath8as@xmath8 ( @xmath57 2 ) @xcite .
here , we only focus on the electronic term .
[ t ] ( a ) low - temperature thermal conductivity of ta@xmath0pd@xmath1te@xmath2 ( sample s1 ) in magnetic fields up to 2 t , applied along the @xmath24 direction .
all the curves are roughly linear .
the solid lines are fits to @xmath50 = @xmath58 .
the dashed line is the normal - state wiedemann - franz law expectation @xmath59(2 t ) , where @xmath60 is the lorenz number 2.45 @xmath25 10@xmath61 w @xmath62 k@xmath63 and @xmath31(2 t ) = 4.54 @xmath33 cm .
( b ) normalized @xmath64 of ta@xmath0pd@xmath1te@xmath2 as a function of @xmath65 .
similar data of the clean @xmath4-wave superconductor nb @xcite , the dirty @xmath4-wave superconducting alloy inbi @xcite , the multiband @xmath4-wave superconductor nbse@xmath8 @xcite , and an overdoped @xmath5-wave superconductor tl-2201 @xcite are also shown for comparison .
the normalized @xmath66 of ta@xmath0pd@xmath1te@xmath2 clearly mimics that of tl-2201 . , title="fig:",width=226 ] [ t ] ( a ) and ( b ) low - temperature resistance of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s8 ) under various pressures up to 21.9 kbar .
( c ) the pressure dependent @xmath12 , defined by @xmath67 .
there is a clear superconducting dome , with maximum @xmath12 = 6.7 k at optimal pressure @xmath68 = 3.1 kbar .
( d ) the pressure dependence of the exponent @xmath34 of resistance .
there is a clear minimum of @xmath34 near @xmath68.,title="fig:",width=321 ] in zero field , the fitting gives a residual linear term with coefficient @xmath69 mw k@xmath63 @xmath70 .
this value is more than 30@xmath71 of the normal - state wiedemann - franz law expectation @xmath72(0 t ) = 6.19 mw k@xmath63 @xmath70 , where @xmath73 2.45 @xmath25 10@xmath61 w
@xmath62 k@xmath63 is the lorenz number and @xmath31(0 t ) = 3.96 @xmath33 cm . in nodeless superconductors ,
all electrons become cooper pairs as @xmath74 and there are no fermionic quasiparticles to conduct heat .
therefore there is no residual linear term of @xmath10 , i.e. , @xmath75 .
however , for unconventional superconductors with nodes in the superconducting gap , the nodal quasiparticles will contribute a finite @xmath64 in zero field @xcite .
for example , @xmath64 = 1.41 mw k@xmath63
@xmath70 for the overdoped @xmath5-wave cuprate superconductor tl@xmath8ba@xmath8cuo@xmath76 ( tl-2201 ) , which is about 36@xmath71 of its @xmath77 @xcite , and @xmath64 = 17 mw k@xmath63 @xmath70 for the @xmath7-wave superconductor sr@xmath8ruo@xmath0 , which is about 9@xmath71 of its @xmath77 @xcite . the significant @xmath64 ( @xmath78 ) of ta@xmath0pd@xmath1te@xmath2 in zero field rules out the case that it results from a small nonsuperconducting metallic portion in the sample ,
thus it is a strong evidence for the presence of nodes in the superconducting gap @xcite . from fig .
[ thermal conductivity](a ) , a small field @xmath79 t has significantly increased the @xmath50 . above @xmath80 t
, @xmath50 tends to saturate . for @xmath81 and 2 t , @xmath82 and 5.16 @xmath83 0.04 mw k@xmath63
@xmath70 were obtained from the fittings , respectively .
the value of @xmath64 for @xmath84 t roughly meets the normal - state wiedemann - franz law expectation @xmath59(2 t ) = 5.40 mw k@xmath63 @xmath70 , which validates our method of extrapolating to @xmath85 .
we take @xmath86 t as the bulk @xmath39(0 ) of ta@xmath0pd@xmath1te@xmath2 . to choose a slightly different bulk @xmath39(0 ) does not affect our discussions of the field dependence of @xmath64 below . in fig .
[ thermal conductivity](b ) , the normalized @xmath64 of ta@xmath0pd@xmath1te@xmath2 is plotted as a function of @xmath65 , together with the clean @xmath4-wave superconductor nb @xcite , the dirty @xmath4-wave superconducting alloy inbi @xcite , the multiband @xmath4-wave superconductor nbse@xmath8 @xcite , and the overdoped @xmath5-wave cuprate superconductor tl-2201 @xcite . for ta@xmath0pd@xmath1te@xmath2
, the field dependence of @xmath64 clearly mimics the behaviour of tl-2201 .
this rapid increase of @xmath64 in magnetic field further rules out the case that the significant @xmath64 results from a small nonsuperconducting metallic portion in the sample , since it should not change so dramatically in magnetic field .
the rapid increase of @xmath64 in magnetic field should come from the volovik effect of nodal quasiparticles , thus provides further evidence for nodes in the superconducting gap @xcite . to our knowledge , so far all nodal superconductors have unconventional pairing mechanism @xcite . in this regard
, the nodal gap we demonstrate from thermal conductivity results suggests unconventional superconductivity in ta@xmath0pd@xmath1te@xmath2 . to get further clue to the pairing mechanism in ta@xmath0pd@xmath1te@xmath2
, we map out its temperature - pressure phase diagram by resistivity measurement under pressures .
figure [ pressureresistivity](a ) and [ pressureresistivity](b ) present the low - temperature resistivity of ta@xmath0pd@xmath1te@xmath2 single crystal ( sample s8 ) under various pressures up to 21.9 kbar . at ambient pressure
, the @xmath12 is 4.2 k , defined by @xmath87 . with increasing pressure ,
the @xmath12 first increases sharply to 6.7 k at 3.1 kbar , enhanced by 60@xmath71 .
then it decreases slowly to 1.8 k at 21.9 kbar .
the non - monotonic pressure dependence of @xmath12 is plotted in fig .
[ pressureresistivity](c ) , which shows a clear superconducting dome . a temperature - pressure ( @xmath12 vs @xmath7 ) or temperature - doping ( @xmath12 vs @xmath88 ) superconducting dome has been commonly observed in many unconventional superconductors , including heavy - fermion superconductors , cuprate superconductors , iron - based superconductors , and q2d organic superconductors @xcite . for example ,
the heavy - fermion superconductor cecoin@xmath6 manifests a @xmath12 vs @xmath7 superconducting dome , and the unconventional superconductivity may result from the antiferromagnetic spin fluctuations @xcite . theoretically , it has been shown that unconventional superconductivity can also appear on the boarder of a density transition , and the superconductivity is mediated by density fluctuations @xcite .
this may be the case of the pressure - induced superconductivity in 1@xmath51-tise@xmath8 , with the superconducting dome appearing around the critical pressure related to the charge - density wave ( cdw ) meltdown @xcite .
for ta@xmath0pd@xmath1te@xmath2 , recent electronic structure calculations showed that its electronic states are mostly derived from te @xmath7 states with small ta @xmath5 and pd @xmath5 contributions , which places the compound far from magnetic instabilities @xcite .
two scanning tunneling microscopy ( stm ) studies on ta@xmath0pd@xmath1te@xmath2 found commensurate modulations along atom chains , which may arise from cdw @xcite .
cdw usually appears in low - dimensional compounds , such as ttf - tcnq , nbse@xmath1 , and nbse@xmath8 @xcite , therefore it is not surprising that cdw exists in layered ta@xmath0pd@xmath1te@xmath2 with q1d structure .
the absence of resistivity anomaly in the resistivity curve suggests that the cdw in ta@xmath0pd@xmath1te@xmath2 is quite weak , since the robustness of cdw can be reflected from the resistivity anomaly , as seen in ttf - tcnq , nbse@xmath1 , and nbse@xmath8 @xcite .
to examine whether the superconducting dome relates to a cdw meltdown in ta@xmath0pd@xmath1te@xmath2 , we carefully fit the resistance data up to 25 k to @xmath30 = @xmath31 + @xmath32 for each pressure , and plot the pressure dependence of the exponent @xmath34 in fig .
there is a clear minimum of @xmath34 near the optimal pressure @xmath68 = 3.1 kbar .
similar pressure dependence of @xmath34 has been observed in 1@xmath51-tise@xmath8 @xcite . for 1@xmath51-tise@xmath8 ,
the @xmath34 = 3 at pressure above 40 kbar is attributed to the phonon - assisted @xmath89 interband scattering , and the suppression of @xmath34 in the 20 - 40 kbar pressure region signifies the presence of cdw fluctuations around a critical pressure of 30 kbar @xcite . in this context ,
@xmath68 = 3.1 kbar is likely the critical pressure where the cdw in ta@xmath0pd@xmath1te@xmath2 is completely suppressed .
if this is the case , the nodal superconductivity in ta@xmath0pd@xmath1te@xmath2 may originate from the cdw fluctuations .
we study the superconducting gap structure of new layered superconductor ta@xmath0pd@xmath1te@xmath2 by low - temperature thermal conductivity measurements .
the significant @xmath64 in zero magnetic field and its rapid field dependence suggest nodes in the superconducting gap .
further measurements of resistivity under pressure reveal a superconducting dome in the temperature - pressure phase diagram .
these results indicate unconventional superconductivity in ta@xmath0pd@xmath1te@xmath2 . with the recent stm evidence for the existence of cdw and our observation of the suppression of exponent @xmath34 near the optimal pressure @xmath68 , ta@xmath0pd@xmath1te@xmath2 may provide a rare platform to study the unconventional superconductivity near a cdw instability . clarifying the pairing symmetry and mechanism of this new layered superconductor will give us new understandings of unconventional superconductivity .
we thank x. h. chen , j. k. dong , and d. l. feng for helpful discussions .
this work is supported by the natural science foundation of china , the ministry of science and technology of china ( national basic research program no .
2012cb821402 and 2015cb921401 ) , program for professor of special appointment ( eastern scholar ) at shanghai institutions of higher learning , and stcsm of china ( no .
15xd1500200 ) .
m. r. norman , _ the challenge of unconventional superconductivity _ , science * 332 * , 196 ( 2011 ) . c. c. tsuei and j. r. kirtley , _ pairing symmetry in cuprate superconductors _ , rev .
phys . * 72 * , 969 ( 2000 ) .
k. an , t. sakakibara , r. settai , y. onuki , m. hiragi , m. ichioka , and k. machida , _ sign reversal of field - angle resolved heat capacity oscillations in a heavy fermion superconductor _ cecoin@xmath6 _ and _ @xmath90 _ pairing symmetry _ , phys .
lett . * 104 * , 037002 ( 2010 ) .
a. p. mackenzie and y. maeno , _ the superconductivity of _ sr@xmath8ruo@xmath0 _ and the physics of spin - triplet pairing _ , rev .
phys . * 75 * , 657 ( 2003 ) .
p. j. hirschfeld , m. m. korshunov , and i. i. mazin , _ gap symmetry and structure of _ fe-_based superconductors _ , rep .
* 74 * , 124508 ( 2011 ) .
d. jrome , a. mazaud , m. ribault , and k. beechgard , _ superconductivity in a synthetic organic conductor _
( tmtsf)@xmath8pf@xmath11 , j. phys . lett . * 41 * , l95 ( 1980 ) . k. bechgaard , k. carneiro , m. olsen , f. b. rasmussen , and c. s. jacobsen , _ zero - pressure organic superconductor : di-(tetramethyltetraselenafulvalenium)-perchlorate _ [ ( tmtsf)@xmath8clo@xmath0 ] , phys .
46 * , 852 ( 1981 ) .
w. zhang and c. a. r. s de melo , _ triplet versus singlet superconductivity in quasi - one - dimensional conductors _ , adv .
phys . * 56 * , 545 ( 2007 ) .
i. i. mazin , _ superconductivity gets an iron boost _ , nature ( london ) * 464 * , 183 ( 2010 ) . a. mar and j. a. ibers , _ synthesis , crystal structure and electrical conductivity of a new layered ternary telluride _ ta@xmath0pd@xmath1te@xmath2 , j. am .
, dalton trans . 639 ( 1991 ) .
w. h. jiao , z. t. tang , y. l. sun , y. liu , q. tao , c. m. feng , y. w. zeng , z. a. xu , and g. h. cao , _ superconductivity in a layered _ ta@xmath0pd@xmath1te@xmath2 _ with _ pdte@xmath8
_ chains _ , j. am . chem . soc . * 136 * , 1284 ( 2014 ) . j .- f .
mercure , a. f. bangura , x. f. xu , n. wakeham , a. carrington , p. walmsley , m. greenblatt , and n. e. hussey , _ upper critical magnetic field far above the paramagnetic pair - breaking limit of superconducting one - dimensional _
li@xmath92mo@xmath11o@xmath49 _ single crystals _ ,
rev . lett . * 108 * , 187003 ( 2012 ) . i. j. lee , m. j. naughton , g. m. danner , and p. m. chaikin , _ anisotropy of the upper critical field in _
( tmtsf)@xmath8pf@xmath11 , phys . rev . lett . * 78 * , 3555 ( 1997 ) .
g. mihly , i. kzsmrki , f. zmborszky , and l. forr , _ hall effect and conduction anisotropy in the organic conductor _
( tmtsf)@xmath8pf@xmath11 , phys . rev . lett . * 84 * , 2670 ( 2000 ) .
h. shakeripour , c. petrovic , and l. taillefer , _ heat transport as a probe of superconducting gap structure _ , new j. phys . * 11 * , 055065 ( 2009 ) .
m. sutherland , d. g. hawthorn , r. w. hill , f. ronning , s. wakimoto , h. zhang , c. proust , etienne boaknin , c. lupien , l. taillefer , r. x. liang , d. a. bonn , w. n. hardy , r. gagnon , n. e. hussey , t. kimura , m. nohara and h. takagi , _ thermal conductivity across the phase diagram of cuprates : low - energy quasiparticles and doping dependence of the superconducting gap _ , phys .
b * 67 * , 174520 ( 2003 ) .
s. y. li , j .- b .
bonnemaison , a. payeur , p. fournier , c. h. wang , x. h. chen , and l. taillefer , _ low - temperature phonon thermal conductivity of single - crystalline _ nd@xmath8cuo@xmath0 : _ effects of sample size and surface roughness _ ,
b * 77 * , 134501 ( 2008 )
. s. y. li , g. wu , x. h. chen , and l. taillefer , _ single - gap s - wave superconductivity near the charge - density - wave quantum critical point in _ cu@xmath93tise@xmath8 , phys .
lett . * 99 * , 107001 ( 2007 ) . j. k. dong , s. y. zhou , t. y. guan , h. zhang , y. f. dai , x. qiu , x. f. wang , y. he , x. h. chen , and s. y. li , _ quantum criticality and nodal superconductivity in the _ feas_-based superconductor _ kfe@xmath94as@xmath94 , phys . rev . lett . *
104 * , 087005 ( 2010 ) . c. proust , e. boaknin , r. w. hill , l. taillefer , and a. p. mackenzie , _ heat transport in a strongly overdoped cuprate : fermi liquid and a pure d - wave bcs superconductor _ , phys .
lett . * 89 * , 147003 ( 2002 ) .
m. suzuki , m. a. tanatar , n. kikugawa , z. q. mao , y. maeno , and t. ishiguro , _ universal heat transport in _
sr@xmath8ruo@xmath0 , phys .
lett . * 88 * , 227004 ( 2002 ) .
j. lowell and j. b. sousa , _ mixed - state thermal conductivity of type _ , j. low .
phys . * 3 * , 65 ( 1970 ) .
j. willis and d. ginsberg , _ thermal conductivity of superconducting alloy films in a perpendicular magnetic field _ , phys .
b * 14 * , 1916 ( 1976 ) .
e. boaknin , m. a. tanatar , j. paglione , d. hawthorn , f. ronning , r. w. hill , m. sutherland , l. taillefer , j. sonier , s. m. hayden , and j. w. brill , _ heat conduction in the vortex state of _ nbse@xmath8 : _ evidence for multiband superconductivity _ ,
lett . * 90 * , 117003 ( 2003 ) . v. a. sidorov , m. nicklas , p. g. pagliuso , j. l. sarrao , y. bang , a. v. balatsky , and j. d. thompson , _ superconductivity and quantum criticality in _
cecoin@xmath6 , phys .
lett . * 89 * , 157004 ( 2002 ) .
p. monthoux and g. g. lonzarich , _ density - fluctuation - mediated superconductivity _ , phys .
b * 69 * , 064517 ( 2004 ) .
p. monthoux , d. pines , and g. g. lonzarich , _ superconductivity without phonons _ , nature ( london ) * 450 * , 1177 ( 2007 ) .
a. f. kusmartseva , b. sipos , h. berger , l. forr , and e. tuti , _ pressure induced superconductivity in pristine _ 1@xmath51-tise@xmath8 , phys .
103 * , 236401 ( 2009 ) .
d. j. singh , _ multiband superconductivity of _ ta@xmath0pd@xmath1te@xmath2 _ from _ te _ p states _
, phys . rev .
b * 90 * , 144501 ( 2014 ) .
z. y. du , d. l. fang , z. y. wang , y. f. li , g. du , h. yang , x. y. zhu , and h .- h .
wen , _ anisotropic superconducting gap and elongated vortices with caroli - de gennes - matricon states in the new superconductor _ ta@xmath0pd@xmath1te@xmath2 , arxiv:1412.3993 .
q. fan , w. h. zhang , x. liu , y. j. yan , m. q. ren , m. xia , h. y. chen , d. f. xu , z. r. ye , w. h. jiao , g. h. cao , b. p. xie , t. zhang , and d. l. feng , _ scanning tunneling microscopy study of superconductivity , magnetic vortices , and possible charge - density wave in _
ta@xmath0pd@xmath1te@xmath2 , phys .
b * 91 * , 104506 ( 2015 ) .
a. andrieux , h. j. schulz , d. jerome , and k. bechgaard , _ conductivity of the one - dimensional conductor tetrathiafulvalene - tetracyanoquinodimethane _
( ttf - tcnq ) _ near commensurability _ , phys .
lett . * 43 * , 277 ( 1979 ) .
a. briggs , p. monceau , m. nunez - regueiro , j. peyrard , m. ribaultl , and j. richard , _ charge density wave formation , superconductivity and fermi surface determination in _
nbse@xmath1 : _ a pressure study _
, j. phys .
c : solid st . phys . , * 13 * , 2117 ( 1980 ) .
h. n. s. lee , h. mckinzie , d. s. tannhauser , and a. wold , _ the low - temperature transport properties of _ nbse@xmath8 , j. appl .
, * 40 * , 602 ( 1969 ) .
|
we measured the low - temperature thermal conductivity of a new layered superconductor with quasi - one - dimensional characteristics , the ternary telluride ta@xmath0pd@xmath1te@xmath2 with transition temperature @xmath3 4.3 k. the significant residual linear term of thermal conductivity in zero magnetic field and its rapid field dependence provide evidences for nodes in the superconducting gap . by measuring resistivity under pressures ,
we reveal a superconducting dome in the temperature - pressure phase diagram .
the existence of gap nodes and superconducting dome suggests unconventional superconductivity in ta@xmath0pd@xmath1te@xmath2 , which may relate to a charge - density wave instability in this low - dimensional compound .
|
it is now widely recognized that multiband superconductors are omnipresent @xcite .
this recognition may be triggered by mgb@xmath0 @xcite where there exist distinctive two bands ; the 3d @xmath8-band and 2d like @xmath9-band@xcite .
they play different role in forming superconductivity , in particular in magnetic properties under an applied field , such as symmetry of vortex lattices@xcite or the form factors probed by small angle neutron scattering ( sans ) experiments@xcite . to understand its detailed magnetic response , a two band model is indispensable .
in fact different dimensionality of the band structures between the @xmath8-band and @xmath9-band gives rise to rotation of the triangular vortex lattice under varying field@xcite .
the form factors of sans experiments clearly demonstrate a gradual change of the two components of the @xmath8-band and @xmath9-band as field varies@xcite .
it is also true for other materials among unconventional and conventional superconductors where multiband description is essential , such as heavy fermion superconductors @xcite and iron pnictides @xcite .
we have been seeing that the pauli paramagnetic effect ( ppe ) is important when combined with this multiband effect in certain superconductors , which give rise to a variety of unexpected phenomena .
typical examples are the oldest heavy fermion superconductors cecu@xmath0si@xmath0@xcite and ube@xmath10 @xcite , and also kfe@xmath0as@xmath0 @xcite , which necessitate the multiband description in fully understanding of their vortex properties . those include the hidden first order transition phenomenon @xcite and the disparity @xcite between the vortex lattice anisotropy and fermi velocity anisotropy as discussed later . here
we study the interplay between ppe and multiband effects in a uniaxial anisotropic superconductor in which each band has a different uniaxial anisotropy : for the case of two bands which we consider in this paper we can envisage two possible situations as schematically illustrated in fig .
let us consider the first case shown in figs .
[ fig1 ] ( a ) and ( b ) . in the absence of ppe the two orbital limited upper critical fields @xmath11 and @xmath12 cross at a in @xmath13 versus @xmath14 plane ( @xmath14 is the angle from the @xmath15 plane ) , each of which is characterized by the effective mass anisotropy @xmath16 for two bands @xmath17 and @xmath6 , assuming @xmath18 . as indicated in fig .
[ fig1 ] ( a ) the four divided regions are characterized by each @xmath16 .
in particular , along @xmath19 the characteristic anisotropy of the total system is switched at the intersecting point a from @xmath20 to @xmath21 as @xmath14 increases .
when traversing at a higher @xmath13 , the anisotropy @xmath20 for the @xmath4 band is sensed only .
now let us switch on ppe , then both orbital limited @xmath22 are suppressed towards lower fields , especially @xmath20 if we assume that the superconducting gaps for the two bands such that @xmath23 .
this is because the pauli limited fields @xmath24 .
the resulting phase diagram is shown in fig .
[ fig1 ] ( b ) . as seen from it the crossing point a is removed and three regions are now occupied by @xmath21 while the @xmath20 region is hidden deep inside at lower fields and finite @xmath14 s .
in particular , along @xmath19 the @xmath21 region persists all the way from @xmath25 to @xmath26 .
thus the higher field scan is sensing only the @xmath21 anisotropy while a lower field scan is sensing @xmath27 as @xmath14 increases .
this non - trivial anisotropy evolution is caused by the interplay between the effective mass anisotropy @xmath28 and ppe .
another possible phase diagrams are depicted in figs .
[ fig1](c ) and ( d ) where @xmath29 is assumed , keeping @xmath18 .
the @xmath4 ( @xmath6 ) band with @xmath30 ( @xmath31 ) is now major ( minor ) . in the absence of ppe shown in fig .
[ fig1](c ) the two orbital limited @xmath22 curves are not crossed , thus the higher ( lower ) field region is occupied by @xmath20(@xmath32 ) . in the presence of ppe
those curves are both suppressed downwards .
thus the two regions ( @xmath20 and @xmath32 ) are simply shifted downwards , keeping its phase diagram topologically unchanged .
note that @xmath33 .
therefore , a higher ( lower ) field scan is exclusively sensing the @xmath20 ( @xmath32 ) anisotropy of the total system as @xmath14 increases .
( color online ) schematic phase diagrams in the @xmath13 versus @xmath14 plane .
the @xmath6 band major scenario without ppe ( a ) and with ppe ( b ) . the @xmath4 band
major scenario without ppe ( c ) and with ppe ( d ) .
the effective mass anisotropy is assumed to be @xmath18 .
, width=340 ] the purpose of this paper is to investigate the physics of sr@xmath0ruo@xmath1 through the studies of the mixed state properties in this multiband superconductor where the interplay between the effective mass ( or fermi velocity ) anisotropy and ppe is important in understanding of the vortex lattice state in @xmath13 vs @xmath14 plane .
this superconductor is known to have the strong two dimensional uniaxial anisotropies for the three bands @xcite , the @xmath4 band ( @xmath34180 ) the @xmath5 band ( @xmath35120 ) and the @xmath6 band ( @xmath3660 ) . since the density of states ( dos )
@xmath37 at the fermi level ( @xmath38=0.53 , @xmath39=0.10 and @xmath40=0.37 of the total dos ) , the @xmath5 band is neglected in this paper for simplicity . in the following we consider a two band model : the @xmath4 band with @xmath38=0.53 and the @xmath6 band with @xmath40=0.47 . in sr@xmath0ruo@xmath1
there are several outstanding unsolved issues : \(1 ) which bands is the major band for superconductivity , either the @xmath4 band or the @xmath6 band ?
@xmath23 or @xmath29 ?
\(2 ) can the ppe resolve the observed disparity between the vortex lattice anisotropy @xcite 60 and the @xmath3 anisotropy @xcite 20 ?
the first issue ( 1 ) has been extensively debated , in particular in connection with the pairing mechanism of this material to stabilize the chiral p - wave state @xcite .
we approach this issue from the different view point by analyzing the mixed state vortex states .
the second issue ( 2 ) has been investigated by us @xcite based on the single band picture . in this paper we revisit it based on a more realistic two band model , which is able to allow us to study the issue ( 1 ) .
we base our computations on the quasiclassical theory @xcite .
the original single band theory is extended to various multiband cases , including mgb@xmath0 @xcite , iron pnictides @xcite .
the applicability of the quasiclassical theory @xcite is given in general by the condition @xmath41 with @xmath42 the fermi wave number and @xmath43 the coherence length under the assumption that the normal state properties are described by a fermi liquid theory .
for sr@xmath0ruo@xmath1 the three bands @xmath5 , @xmath6 and @xmath4 are @xmath42=0.304 , 0.622 and 0.753 ( @xmath44 ) , respectively .
the in - plane and @xmath45 axis coherence lengths @xmath43 are 660 and 33 ( @xmath46 ) , respectively @xcite . thus for any combinations @xmath41 is well satisfied .
we also notice that the lattice constants @xmath47=3.862@xmath46 and @xmath45=12.722@xmath46 are short enough compared with those coherence lengths , thus sr@xmath0ruo@xmath1 is a three dimensional normal metal .
moreover , most physical quantities , including various transport coefficients and thermodynamic properties can be consistently and coherently described by a fermi liquid theory as explained in details by mackenzie and maeno @xcite .
therefore we can quite safely apply the quasiclassical theory to sr@xmath0ruo@xmath1 . by self - consistently solving microscopic quasiclassical eilenberger equation with two bands ,
we calculate a variety of the physical quantities relevant to available experiments , such as the form factors probed by sans experiments @xcite , magnetic torques @xcite , and the vortex lattice anisotropy @xmath48 which differs generally from the effective mass anisotropy @xmath16 mentioned above .
the arrangement of this paper is as follows : after introducing eilenberger theory extended to multibands @xcite with ppe in sec .
2 which was done in our previous paper @xcite , we construct a model system for considering sr@xmath0ruo@xmath1 by fixing the several model parameters in sec .
we compare the two scenarios , one is the @xmath6 major and the other @xmath4 in equal footing , finding that the former is better than the latter relative to the existing experiments .
then we come to the main theme of the present paper in sec.4 ; computations and analyses of the form factors and vortex lattice anisotropies as a function of the angle @xmath14 compared with the data of sans experiments @xcite . in sec .
5 we examine the magnetic torques which are also measured recently by kittaka et al@xcite .
the final section 6 is devoted to discussions and conclusion .
this paper is an extension of our previous work based on a single band model@xcite and also closely related to our two band model calculations@xcite .
we start with the free energy @xmath49 in the quasiclassical theory @xcite extended to the two band case @xcite , which is given by @xmath50 @xmath51 with @xmath52 , @xmath53 and @xmath54 @xmath55 @xmath56 the flux quantum @xmath57 .
@xmath58 is the fermi velocity at @xmath59 of the band j. the fermi surface average @xmath60 is normalized within each band as @xmath61 .
here we introduced the interaction matrix @xmath62 with 2@xmath632 for two bands where @xmath64 is the pairing interaction on the @xmath65 band and @xmath66 for @xmath67 is the cooper pair transfer between the @xmath68 and @xmath65 bands .
@xmath69 , @xmath70 , and @xmath71 are the quasiclassical green s functions for the @xmath65 band .
@xmath72 is the pair potential and @xmath73 describes the gap symmetry of the @xmath65 band in reciprocal space , which allows us to choose the gap form depending upon each band as will be done in the following .
the vector potential @xmath74 and the internal field @xmath75 are related to @xmath76 with @xmath77 uniform field . by following the same procedure by eilenberger @xcite , the functional derivatives with respect to @xmath70 , and @xmath78 yield the so - called eilenberger equation extended to the two band case : @xmath79\right\}f_j=\delta_j(\bm{k}_j , \bm{r})g_j , \\ & \left\{\omega_n+i\mu b(\bm{r})-\bm{v}_j\cdot\left[\bm{\nabla } -i\bm{a}(\bm{r})\right]\right\}{f}^{\dagger}_j=\delta_j^*(\bm{k}_j , \bm{r})g_j .
\end{split}\label{eq : eilenberger}\ ] ] this form is understandable because the fourth term in the free energy eq . , which includes the green s functions , is separable in the band index , thus the resultant equation of the functional derivative should be separable for each band .
the stationary conditions of eq . with respect to the functionals
@xmath80 and the vector potential @xmath81 give rise to a complete set of the self - consistent equations extended to the two band case , which are given below ; eqs . and .
this complete set of the self - consistent equations coincides and is consistent with those obtained previously @xcite .
the electronic state is calculated by solving the eilenberger equation eq . in the clean limit @xcite , including the pauli paramagnetic effect ( ppe ) due to the zeeman term @xmath82 @xcite , where @xmath83 is a renormalized bohr magneton related to the so - called maki parameter @xmath84 .
the quasiclassical green s functions @xmath85 , @xmath86 , and @xmath87 with the band index @xmath65 depend on the direction of the fermi momentum @xmath88 for each band , the center - of - mass coordinate @xmath89 for the cooper pair , and matsubara frequency @xmath90 with @xmath91 .
they are calculated in a unit cell of the triangle vortex lattice .
the unit of fermi velocity @xmath92 is defined by @xmath93 , where the density of states ( dos ) in the normal state at each fermi surface is defined by @xmath94 . throughout this paper , temperatures , energies , lengths , and magnetic fields are , respectively , measured in units of the transition temperature @xmath95 , @xmath96 , @xmath97 , and @xmath98 .
we calculate the spatial structure of @xmath99 in a fully self - consistent way .
the pairing potential @xmath100 is calculated by the gap equation @xmath101 which is coupled via the interaction matrix @xmath102 .
we use the energy cutoff @xmath103 .
the vector potential is also self - consistently determined by @xmath104\right\rangle_{\bm{k}_j}\!,\label{eq : potential}\end{aligned}\ ] ] which includes the contribution of the paramagnetic moment @xmath105 with @xmath106\right\rangle_{\bm{k}_j}\!\right)\!.\label{eq : mpara}\end{aligned}\ ] ] here @xmath107 is the averaged flux density mentioned above , the normal state paramagnetic moment @xmath108 , and @xmath109 .
the ginzburg - landau ( gl ) parameter @xmath110 is the ratio of the penetration depth to coherence length for @xmath111 . using doria - gubernatis - rainer scaling @xcite
, we obtain the relation @xcite of @xmath107 and the external field @xmath13 .
the total magnetization @xmath112 including both the diamagnetic and the paramagnetic contributions is derived .
we solve eq .
with @xmath113 for the electronic state .
the local density of states ( ldos ) is given by @xmath114 where @xmath115\right\rangle_{\bm{k}_j}\!,\end{aligned}\ ] ] with @xmath116 @xmath117 for up ( down ) spin component .
we typically use the smearing factor @xmath118 .
the dos is obtained by the spatial average of the ldos as @xmath119 .
we consider a simplified model of a two - band system with a larger superconducting gap band ( band-1 ) and a smaller gap band ( band-2 ) . as a model of the fermi surfaces , we use two quasi - two dimensional fermi surfaces with rippled cylinder shapes .
the fermi velocity is assumed to be @xmath120 at @xmath121 on the fermi surfaces @xcite .
we consider a case @xmath122 , to produce large anisotropy ratio of the coherence lengths , @xmath123 with @xmath124 where @xmath60 indicates an average over the fermi surface on each band .
the magnetic field orientation is tilted by @xmath125 from the @xmath45 axis towards the @xmath15 plane .
since we set @xmath126 axis to the vortex line direction , the coordinate @xmath127 for the vortex structure is related to the crystal coordinate @xmath128 as @xmath129 .
we set unit vectors of the vortex lattice as @xmath130 with @xmath131 and the vortex lattice anisotropy is defined by @xmath132 . the anisotropic ratio @xmath133 , which comes from the fermi velocity anisotropy , @xmath134 with @xmath135 and 2 .
to discuss @xmath107-dependence of the internal field distribution @xmath136 , we consider flux line lattice ( fll ) form factor @xmath137 , which is obtained by fourier transformation of the internal field distribution as @xmath138 with the wave vector @xmath139 .
@xmath140 and @xmath141 are integers .
the unit vectors in reciprocal space are given by @xmath142 and @xmath143 .
the @xmath126-component @xmath144 from @xmath145 gives the intensity of conventional non - spinflip sans .
the transverse component , @xmath146 , is accessible by spin - flip sans experiments.@xcite
in order to determine the model parameters appropriate for sr@xmath0ruo@xmath1 , we start out to fix the gap magnitudes @xmath148 and @xmath149 and their nodal structures . we first analyze the electronic specific heat data @xmath150 at zero field @xcite by solving self - consistently the eilenberger equation eq . for the uniform system without using the phenomenological so - called @xmath5 model @xcite .
here we have assumed that the minor component induced only by the cooper pair transfer coupling @xmath151 .
the direct attractive coupling among the minor component is vanishing ; @xmath152 . as seen from fig.[fig2 ]
the @xmath150 data is equally well explained either by \(a ) the @xmath4 scenario with @xmath153 at @xmath154 where both bands have line nodes as coincided with other authors@xcite , \(b ) the @xmath6 scenario with @xmath155 where the @xmath6 band has a full gap and the @xmath4 band line nodes .
it is apparent that the linear @xmath156 behavior of @xmath150 at lower @xmath156 indicates that the nodal gap is necessary somewhere ; for the @xmath4 scenario in both bands and for the the @xmath6 scenario only in the @xmath4 band .
this nodal structure difference between them is decisive for the two scenarios as will be seen shortly .
( color online ) the specific heat data ( open symbols ) @xcite analysis by the @xmath6 major scenario ( bold line ) with @xmath155 where the line node ( full ) gap is on the @xmath4 ( @xmath6 ) band and by the @xmath4 major scenario ( dotted line ) with @xmath153 where the both bands contain the line nodes . for comparison , the standard bcs case with the full gap
is also shown ( thin line ) .
, width=188 ] ( color online ) the upper critical field ratio @xmath157 for the two directions ( @xmath158 and @xmath147 ) .
the two cases for the @xmath6 ( filled triangles ) and @xmath4 scenarios ( filled squares ) are compared with the experimental data ( open symbols ) @xcite .
this shows that the @xmath6 scenario is superior to the @xmath4 scenario .
, width=188 ] ( color online ) the comparison of the experimental data ( open symbols ) @xcite of the specific heat at @xmath156=60mk as a function of @xmath159 with the theoretical results ( filled symbols ) for the @xmath6 and @xmath4 scenarios ( @xmath160 ) .
this shows too much low energy excitations released at lower fields in the @xmath4 scenario because the nodal gap exists in the minor band .
, width=188 ] according to kittaka et al @xcite , the upper critical field ratio @xmath157 for the two directions ( @xmath158 and @xmath147 ) is @xmath156-dependent , implying that ppe becomes stronger as the field applied to the @xmath15 plane increases . here
we calculate @xmath157 for both scenarios and depict the results in fig .
[ fig3 ] . since near @xmath95 the intrinsic effective mass anisotropy governs its tending limit ( @xmath161 ) , namely , @xmath162 180 ( 60 ) for the @xmath4 ( @xmath6 ) scenario .
the experimental data @xcite shown support the @xmath6 scenario within the experimental accuracy where there is no indication of the ratio with tending to 180 .
this is one of the most clear signatures for the @xmath6 major scenario . by adjusting the @xmath163 parameters for both scenarios
the best fittings are accomplished in @xmath163=0.04 ( 0.02 ) for the @xmath6 ( @xmath4 ) scenario . from now on we use those values in the following calculations . in order to further distinguish between the two scenarios , we take up the experimental data @xcite of the field dependence of @xmath164 for @xmath147 at the available lowest temperature @xmath156=60 mk , which best mimics the zero - energy density of states @xmath165 at @xmath154 in the theoretical calculation . in fig .
[ fig4 ] we compare the experimental data with the theoretical values as a function of the field applied parallel to the @xmath45 axis . as seen from it , the @xmath4 scenario overestimates the data while the @xmath6 scenario underestimates it .
if taking into the finite @xmath156 effect in the experimental data , the theoretical curves should move up when considering thermal excitations , resulting in further departure of the @xmath4-major curve while the @xmath6-major curve comes closer to the data .
we also note from fig .
[ fig4 ] that too much low energy excitations are released at lower fields in the @xmath4 scenario because the nodal gap exists in the minor band , causing the overestimate .
this fact , which has been unnoticed so far is quite fatal for the @xmath4 scenario .
thus it is clear from those two criteria that the @xmath6 scenario is far better than the @xmath4 scenario .
the best fitting also yields the @xmath110 value for @xmath147 , that is , @xmath110=2.7 whose value is used in the following computations .
but we should keep in mind that this low @xmath110 value delicately depends on the particular sample used .
this quantity is known to be sample dependent .
the form factors of the longitudinal components @xmath166 and @xmath167 which are measured by sans experiments @xcite for @xmath147 are compared with the two scenarios whose magnitudes are multiplied by a factor of 1.7 for the @xmath6 scenario and 3.5 for the @xmath4 scenario .
this is partly because the actual @xmath110=2.7 determined by @xmath164 fitting previously might be different from the sans experiment ( in fact @xmath168=58mt and @xmath110=2.0 differ from @xmath168=75mt and @xmath110=2.3 in the best samples @xcite ) .
since the form factor magnitudes are sensitive to the @xmath110 value ( @xmath169 ) , it is permissible to adjust it to fit with data . as seen from fig .
[ fig5 ] the field dependences of @xmath166 and @xmath167 are far better explained by the @xmath6 scenario than by the @xmath4 scenario .
( color online ) the longitudinal form factors @xmath166 and @xmath167 as a function of @xmath170 .
the experimental data ( open symbols ) @xcite are compared with the two scenarios .
the theoretical results ( filled symbols ) are multiplied by a factor 1.7 ( 3.5 ) for the @xmath6 ( @xmath4 ) case ( @xmath160 ) .
this shows that the @xmath6 scenario is superior to the @xmath4 scenario .
, width=321 ] in fig .
[ fig6 ] we decompose the longitudinal form factors @xmath166 and @xmath167 into the two contributions of the @xmath6 band and @xmath4 band in the case of the @xmath6 scenario .
it is seen that the minor @xmath4 contribution amounts to @xmath171 of the total at this temperature @xmath160 for both @xmath166 and @xmath167 .
thus in order to understand the field dependence of the form factors , the multiband effect is essential , which further becomes clear later .
note in passing as mentioned in introduction on mgb@xmath0 cubitt et al @xcite discover the additional contribution of the minor @xmath8 band to the main @xmath9 band contribution at lower @xmath13 .
this general trend here supports their discovery ( see fig . 2 in ref . ) .
note that the relative weight of the main @xmath6 contribution and minor @xmath4 contribution depends on temperature , field and field orientation @xmath14 .
generally as @xmath156 and @xmath13 decrease , the minor @xmath4 contribution increases because the two order parameters are more competitive there .
( color online ) the longitudinal form factors @xmath166 and @xmath167 are decomposed into the the @xmath6 band contribution and @xmath4 band contribution in the case of the @xmath6 scenario ( @xmath160 ) .
, width=302 ] since we have determined the gap ratios for each scenario , it is possible to establish the phase diagrams on @xmath13 vs @xmath14 plane . for @xmath147
the two orbital limited @xmath172 ratio ( @xmath173 ) with the obvious notations is written as @xmath174 it is known@xcite that the fermi velocity ratio @xmath175 .
thus @xmath176 for the @xmath6 scenario while @xmath177 for the @xmath4 scenario , those corresponding to fig .
1(a ) and fig .
1(c ) at @xmath26 respectively . on the other hand , at @xmath25 @xmath178 this yields @xmath179 for the @xmath6 scenario while @xmath180 for the @xmath4 scenario , thus giving rise to the situations depicted in fig
. 1(a ) and fig .
1(c ) at @xmath25 respectively . note that the crossing point a of the two orbital limited @xmath3 s in fig .
1(a ) is located at around @xmath181 .
the form factors ff are a sensitive and useful probe measured by sans in order to detect the field distribution in the vortex state .
the sans experiments @xcite on sr@xmath0ruo@xmath1 are performed .
they find the transverse components of ff as a function of @xmath14 near the @xmath15 plane . here
we obtain ff by evaluating the field distribution via a self - consistent solution of eq . .
it is seen from figs .
[ fig7 ] ( a ) and ( b ) that the angle @xmath14 dependences of the transverse component of ff for two fields @xmath182 ( high field ) and @xmath183 ( low field ) exhibit a different characteristic ; the high field data show a simple monotonic decrease after taking a maximum towards @xmath184 at which the superconducting - normal state transition takes place .
the monotonic transverse ff curve is similar to that of the single band case shown in previous papers@xcite .
this high field scan ( @xmath182 ) corresponds to the horizontal scanning path in fig.[fig1](b ) where only the @xmath21 region is sensed .
the maximum position @xmath185 coincides with those of the single band result @xcite with @xmath186 .
the overall features of the experimental data for @xmath187 t and 0.7 t are well reproduced as seen from fig .
[ fig7 ] ( a ) except for a few data points at higher angles . on the other hand , for the low field result ( @xmath183 ) shown in fig.[fig7 ] ( b )
, it is seen that the theoretical ff curve coincides with the maximum angle of the experimental data at @xmath188 t and 0.25 t .
however , it deviates from those data after that .
the experimental data tend to vanish at round @xmath189 , which is by far from the known @xmath190 at @xmath188 ( 0.25)t , which should be ultimate vanishing angle for ff amplitude .
we will discuss this discrepancy shortly .
here we just point out that those low field data correspond to the low field scanning paths in fig.[fig1](b ) where the crossover of the regions @xmath191 is sensing with increasing @xmath14 .
( color online ) the transverse form factors @xmath192 of the sans data ( open symbols ) @xcite at @xmath13=0.7 t and @xmath13=0.5 t are compared with the theoretical results ( filled symbols ) for @xmath182 ( a ) and the data at @xmath13=0.25 t and 0.15 t with @xmath183 ( @xmath193 ) ( b ) . , width=302 ] ( color online ) the decomposition of the transverse components of the form factors @xmath194 ( a ) and @xmath192 ( b ) at high field @xmath182 ( @xmath193 ) . the decomposition of the longitudinal components of the form factors @xmath195 ( c ) and @xmath196 ( d ) at high field @xmath182 ( @xmath193 ) .
, title="fig:",width=302 ] ( color online ) the decomposition of the transverse components of the form factors @xmath194 ( a ) and @xmath192 ( b ) at high field @xmath182 ( @xmath193 ) .
the decomposition of the longitudinal components of the form factors @xmath195 ( c ) and @xmath196 ( d ) at high field @xmath182 ( @xmath193 ) .
, title="fig:",width=302 ] ( color online ) the decomposition of the transverse components of the form factors @xmath194 ( a ) and @xmath192 ( b ) at low field @xmath183 ( @xmath193 ) . the decomposition of the longitudinal components of the form factors @xmath195 ( c ) and @xmath196 ( d ) at low field @xmath183 ( @xmath193 ) . , title="fig:",width=302 ] ( color online ) the decomposition of the transverse components of the form factors @xmath194 ( a ) and @xmath192 ( b ) at low field @xmath183 ( @xmath193 ) . the decomposition of the longitudinal components of the form factors @xmath195 ( c ) and @xmath196 ( d ) at low field @xmath183 ( @xmath193 ) .
, title="fig:",width=302 ] we decompose the ff contributions from the major @xmath6 band and the minor @xmath4 band for the transverse components @xmath197 ( a ) and @xmath198 ( b ) and the longitudinal components @xmath199 ( c ) and @xmath199 ( d ) as shown in fig . [ fig8 ] for the high field and fig .
[ fig10 ] for the low field .
the relative weight of the minor component in this transverse ff is around @xmath200 of the total ff and their peak contribution coincides with the maximum position @xmath201 for both @xmath182 ( high field ) and @xmath183 ( low field ) data . as for the longitudinal components of @xmath199 and @xmath199 , it is reasonable to see that the maximum angle @xmath201 is shifted to a higher angle than @xmath198 .
this is because when @xmath202 is approached in which the `` effective '' field virtually increases because the decreasing @xmath203 means decreasing @xmath204 and hence enhances ppe , thus pushing up the longitudinal ff @xmath205 . in order to clearly see the different angle dependences @xmath206 for the two cases and to understand the discrepancy of the low field data mentioned above shown in fig .
[ fig7 ] ( b ) , we replot those theoretical results normalized by the own @xmath207 in fig . [ fig12 ] and compare those with the corresponding single band results @xcite where we have adjusted the vertical scale so that the slopes of these curves near @xmath207 coincide with each other .
it is now seen clearly that \(1 ) the high field result @xmath182 belongs to the single band universality curve .
this is because the high field scanning path is sensing only the @xmath21 region in fig .
[ fig1 ] ( b ) .
this is virtually same as in the single band case .
namely after taking the maximum the ff curve simply goes to vanish at @xmath207 .
\(2 ) the low field result @xmath183 behaves differently from those single band universality curves and exhibits a `` bimodal '' @xmath14 dependence where the ff peak is additionally enhanced .
thus the curve just after taking the maximum tends to vanish earlier than at @xmath207 which is the ultimate vanishing angle .
let us come back to understand the low field ff data shown in fig .
[ fig7 ] ( b ) .
the bimodal @xmath206 structure at the low field shown in fig .
[ fig10 ] ( b ) where the slope just after the maximum differs from the slope at higher angles as @xmath14 increases .
this crossover angle @xmath208 .
this low field scan corresponds to the low field scanning path in fig .
1(b ) where the crossover of the regions @xmath191 is sensing with increasing @xmath14 .
in particular , the central peak @xmath14 region ( @xmath209 ) in the bimodal structure corresponds to the middle @xmath20 region in this crossover .
this theoretical curve is compared with the sans data @xcite at @xmath210 t in fig.[fig7 ] ( b ) .
although the present experimental data do not exhibit this bimodal ff structure , the existing data are understood as coming only from the middle @xmath20 region among the bimodal structure because the extrapolated vanishing angle @xmath211 .
this critical angle is too small compared with @xmath212 which must be the ultimate vanishing angle of the ff intensity .
thus we interpret this so that the existing ff data are coming for the peak @xmath20 region of the bimodal ff distribution .
this interpretation is strengthened later when analyzing the vortex lattice anisotropy @xmath48 .
( color online ) the comparison of the transverse form factors @xmath192 with the single band results ( open symbols ) @xcite for @xmath213=1.5 , 3.0 , and 4.5 and the present multiband results ( filled symbols ) depicted in fig .
[ fig7 ] for @xmath213=2.0 and 0.5 .
the vertical scale is adjusted so that the slopes of these curves near @xmath207 coincide with each other .
it is clear that the low field theoretical result ( @xmath183 ) behaves differently , exhibiting a bimodal structure .
, width=226 ] figure [ fig13 ] shows the results of the vortex lattice anisotropy @xmath48 compared with the sans experiments @xcite where for a given @xmath213 and @xmath14 the self - consistent solutions of eq .
are optimized for varying @xmath48 as shown in fig .
[ fig14 ] to seek the free energy minimum where the free energy form is given by eq . .
in general the optimized @xmath48 is that given neither by the effective mass anisotropies @xmath214 with @xmath21=60 nor @xmath215 with @xmath20=180 alone given by eq . .
this is obvious because those two anisotropies are coupled and competed by the multiband effect . the higher field results in fig . [ fig13 ]
( a ) follow rather well those given by the above formula of eq . for @xmath214 for the higher angles @xmath216 .
we note that at @xmath25 where @xmath217=70 corresponds precisely to the anisotropy @xmath218 for @xmath193 . between the angles @xmath219 the theoretical results deviate upwards , that is , @xmath220 . the high field scan in fig.1(b )
is barely touching the @xmath20 region whose anisotropy can be certainly larger than @xmath21=60 because @xmath20=180 .
the low field theoretical results shown in fig .
[ fig13 ] ( b ) follow the @xmath214 curve for @xmath221 and remarkably exceed the @xmath222 curve in @xmath223 .
this window in @xmath14 corresponding to the @xmath20 region appears because there the @xmath20=180 anisotropy further modifies the @xmath48 value upwards , simultaneously enhancing the transverse ff .
we point out a theoretical fact that as @xmath224 increases , @xmath225 becomes larger .
we anticipate that the results at lower temperatures than the present one at @xmath193 would improve the quantitative fittings of @xmath48 and simultaneously @xmath225 .
note that @xmath226=52 corresponds to the single band anisotropy for the @xmath6 band at lower fields @xcite .
the @xmath210 t data nicely fit our theoretical result .
the important point here is that not only the experimental data exceed the @xmath222 line , but also there exists a wider window @xmath223 where the experimental data deviate from the single band @xmath214 curve .
( color online ) the comparison of the experimental data ( open symbols ) @xcite for @xmath224 for @xmath187 t with the high field results of @xmath182 ( filled symbols ) ( a ) and @xmath210 t with the low field results of @xmath183 ( filled symbols ) ( b ) ( @xmath193 ) .
the curves are drawn by the effective mass formula eq . with @xmath186 ( dotted line ) and @xmath227 ( dashed line ) . ,
width=340 ] ( color online ) the free energy curves for various angles @xmath14 as a function of @xmath48 for @xmath213=2.0 ( filled symbols ) and @xmath213=0.5 ( open symbols ) in order to emphasize that our computations are performed accurately enough to resolve the subtle differences of the @xmath48 values .
, width=264 ] we demonstrate in fig .
[ fig14 ] that the optimal @xmath48 is determined careful enough , which can be larger than the corresponding @xmath222 case in eq .. for example , for our @xmath213=0.5 case at @xmath228 our @xmath48=32 while @xmath48=29 for @xmath229 and @xmath48=28 for @xmath230 .
thus it can be said that the multiband effect helps enhancing @xmath48 beyond @xmath229 .
we illustrate the @xmath14 dependences of several physical quantities of interest which are the basis of the form factors and magnetic torque calculations as shown shortly . the free energy @xmath231 is shown in fig .
[ fig15 ] , from which we will evaluate the magnetic torques . since the transition at @xmath207 is of second order at those fields and temperature ( t=0.5t@xmath232 ) , @xmath231 becomes zero smoothly at the transition point @xmath207 . as seen from fig .
[ fig16 ] , the two order parameters ; the major @xmath233 and minor @xmath234 start decreasing from @xmath25 towards @xmath207 as a function of @xmath14 .
the two curves change smoothly in parallel because the minor component @xmath234 is induced by the major one @xmath233 through the cooper pair transfer @xmath151 in the absence of @xmath235 . in this case
we expect no hidden first order transition phenomena@xcite where there is an abrupt change of the two order parameters inside the superconducting state . in fig .
[ fig17 ] the total paramagnetic susceptibility @xmath236 and decomposed @xmath237 and @xmath238 are displayed as a function of @xmath14 . since we assume the density of states @xmath239 and @xmath240 of the total dos , the corresponding paramagnetic values @xmath241 as expected
. we also point out that the base paramagnetic moment values at @xmath25 is large because the calculations are done at rather high temperature @xmath193 .
needless to say , at t@xmath242 @xmath243 should vanish at @xmath25 .
as @xmath14 increases @xmath236 becomes larger because the system is approaching the transition point at @xmath207 where the normal value @xmath244 must be recovered , that is , @xmath245 .
( color online ) the angle dependences of the free energy for @xmath213=2.0 ( filled symbols ) and @xmath213=0.5 ( open symbols ) , showing a smooth change of the second order transition at @xmath207 where @xmath246 and @xmath247 respectively .
, width=188 ] ( color online ) the angle dependences of the order parameter amplitudes @xmath233 and @xmath234 at the center of the vortex unit cell for @xmath213=2.0 ( filled symbols ) and @xmath213=0.5 ( open symbols ) . , width=188 ] ( color online ) the angle dependences of the paramagnetic susceptibility @xmath236 for @xmath213=2.0 ( filled symbols ) and @xmath213=0.5(open symbols ) , which are decomposed into @xmath237 and @xmath238 . ,
magnetic torque @xmath248 defined by @xmath249 provides several important information on a uniaxial anisotropic superconductor .
we can know the intrinsic anisotropy of a system by the peak position of the torque curve @xmath248 .
this can be easily performed by using a phenomenological theory based on london theory@xcite for single band superconductors . in our previous papers@xcite
we examine the applicability of this approach and propose a modification to this . extending this single band eilenberger calculations , here we show the results of the torque curves @xmath248 for the present two band model and use those to analyze the data on sr@xmath0ruo@xmath1 where the torque curves are measured recently@xcite .
we first display our results for the torque curves in fig .
[ fig18 ] where our results for @xmath182 and @xmath183 are compared with the experimental data for @xmath13=0.5 t ( former ) and @xmath13=0.2 t ( latter ) .
it is seen from fig .
[ fig18 ] ( a ) at the high field @xmath182 that the fitting is done well , such as the peak position and the @xmath207 value . since the high field result is sensing only on the @xmath21 region , it is reasonable that the theoretical curve nicely explains the experimental data at @xmath13=0.5 t and the peak position @xmath250 coincides with the single band result @xcite .
this peak position @xmath251 also coincides with @xmath252 of the form factor ( @xmath182 ) .
thus the four values @xmath251 and @xmath252 both for theory and experiment are coinciding with each other , leading us to firmly conclude that in high fields the system is virtually in the single band - like @xmath21 region .
( color online ) the angle dependences of the torques ( filled symbols ) for @xmath213=2.0 together with the experimental data ( open symbols ) @xcite of @xmath13=0.5 t ( a ) and @xmath213=0.5 with the data of @xmath13=0.2 t ( b ) , width=302 ] ( color online ) the torque curves as a function of @xmath14 for @xmath213=2.0 and @xmath213=0.5 at @xmath193 . those are compared with the single band results @xcite for @xmath213=1.5 , 3.0 , and 4.5 .
the vertical scale is adjusted so that the slopes of these curves near @xmath207 coincide with each other .
, width=226 ] on the other hand , the low field result ( @xmath183 ) shown in fig .
[ fig18 ] ( b ) explains the experimental torque curves for @xmath13=0.2 t , but the peak position differs slightly from each others .
we also notice here that the theoretical @xmath251@xmath253@xmath252 and @xmath252 ( @xmath254 ) agrees with the experimental data as shown before .
it is interesting to compare our torque curves with those for the single band case which are displayed in fig .
[ fig19 ] .
it is seen that the two torque curves nicely correspond to the single band curves , which is in contrast with the ff case shown in fig .
[ fig12 ] where the low field curve markedly deviates from the single band case .
this means that the magnetic torque is a rather insensitive probe to see the subtle , but important multiband effect . in other words ,
the form factor measurement is sensitive enough to distinguish the multiband effect from the single band effect .
this is because the torques @xmath249 comes from the total free energy @xmath231 while the ff is probing the particular fourier component of the spatially modulated magnetic field in the mixed state selectively .
thus it is natural to expect that the ff measurement is more sensitive than the torque measurement in picking up the multiband effect .
( color online ) the comparison of the @xmath48 with the single band cases @xmath255 and @xmath256 and also with the @xmath3 anisotropy @xmath257 .
( a ) high field case @xmath258 where those three data @xmath48 , @xmath255 and @xmath259 coincide with each other , and ( b ) low field case @xmath260 where the lower ( high ) angle region corresponds to the @xmath6 ( @xmath4 ) band anisotropy .
the vertical dashed line at around @xmath261 denotes its boundary .
the dotted ( dashed ) curves indicates the angle dependence of the effective mass formula eq . for @xmath186 @xmath262 .
, width=340 ]
in order to describe the transverse components of ff and @xmath48 data @xcite near the @xmath15 plane which are the main themes of the paper , we have done the calculations based on both scenarios , first focusing on the @xmath147 physics to fix the model parameters . here
we critically examine those scenarios comparatively .
it is obvious that the multiband scenario superior than the single band scenario @xcite because the former includes the latter as a limiting case .
a question is how effectively the single band scenario can describe those data or how the multiband description is inevitable for the sr@xmath0ruo@xmath1 physics . as already demonstrated and also shown in fig .
[ fig20 ] , the overall features of those data can be reproduced by the single band model with @xmath21=60 in the higher fields above @xmath263 t in the @xmath13 vs @xmath14 plane where there is a little trace for the existence of the @xmath4 band with @xmath20=180 ( see fig .
[ fig20](a ) ) . in this sense
the single band picture is enough for this region .
in contrast , however , it is necessary to retain the @xmath4 band contribution in addition to the major @xmath6 band contribution in the low field region below 0.5 t ( see fig . [ fig20](b ) ) .
the former contribution is hidden and not explicit where the crossover occurs around @xmath261 as indicated in fig .
[ fig20](b ) .
it is necessary and indispensable to take it into account for explaining the ff and @xmath48 data @xcite .
this is a part of the reasons why our single band theory @xcite is successful to understand the mixed state properties of the present strongly uniaxial anisotropic superconductor .
moreover , depending upon the physical quantities of interest , the multiband effect is not so apparent even in the low field region .
namely , the torque curves are quite insensitive to the presence of the minor band and all the theoretical torque curves collapse into a universal curve by appropriate scaling as discussed before ( see fig . [ fig19 ] ) . the question on which band the gap is large
is much discussed in connection with the pairing mechanism to stabilize the chiral p - wave state by many authors @xcite . here
we take a different view to consider this question through the analysis of the ff , @xmath48 and torques .
if the @xmath4 band is major , this survives in the high field region of the @xmath13 vs @xmath14 plane over the minor @xmath6 band .
then the sans experiment should detect it , manifesting itself in @xmath48 , that is , @xmath264 as @xmath265 . however , both high fields and low filed data shown in fig . [ fig13 ] indicate @xmath266 or so , which is direct evidence that the @xmath6 band with @xmath230 is major .
this conclusion is also supported by the @xmath3 ratio @xmath267 when @xmath161 because near @xmath95 this ratio directly reflects the intrinsic fermi velocity anisotropy @xcite , or the coherent length anisotropy .
thus those experimental data obviously reveal that the @xmath6 band is major with certainty .
in general the two band contains the three pairing parameters @xmath268 ( @xmath235 ) is the attractive interaction for the major ( minor ) band , and @xmath269 is the cooper pair tunneling term or proximity induced term .
here we set @xmath235=0 in this paper because there is no or little indication for @xmath270 , which causes the so - called hidden first order phenomena @xcite .
namely at @xmath271 inside @xmath3 certain physical quantities exhibit a sudden change as a function of @xmath13 , such as the sommerfeld coefficient @xmath272 or the magnetization curve as observed in cecu@xmath0si@xmath0 @xcite , ube@xmath10 @xcite and kfe@xmath0as@xmath0 @xcite .
this means that @xmath273 in sr@xmath0ruo@xmath1 and the minor @xmath4 band pairing is exclusively induced by the major @xmath4 band through @xmath151 .
because of several technical reasons our main computations have been done at relatively high @xmath156 , namely @xmath193 , which are nevertheless successful in capturing the characteristic features in the ff , @xmath48 and torques in a qualitative level .
to understand those data in a quantitative level it is necessary to go into lower @xmath156 which is demanding computationally . here
we anticipate possible outcomes if performing it . as shown in fig .
[ fig7 ] we have mentioned that the low filed ff ( @xmath213=0.5 ) differs from that in @xmath213=2.0 , which belongs to the single band universality class .
the ff in low angle @xmath14 is enormously enhanced by the assistance from the minor band which amounts almost 10@xmath274 even at @xmath193 .
in low @xmath156 calculation this enhancement of the ff should increase and the ff angle dependence becomes more similar to the data shown in fig .
[ fig7 ] ( b ) .
this expectation is reasonable because at lower @xmath156 the induced paramagnetic moments are confined in the vortex core , making more contrast in the spatial field distribution and thus enhancing the ff amplitude .
see also fig .
[ fig17 ] where the paramagnetic susceptibility or paramagnetic moments of the minor @xmath4 band are almost exhausted at @xmath14=0 to its normal value , meaning that is not confined in the core .
thus , the paramagnetic moments are spreading out the whole space uniformly .
we can deepen our understandings of the present material by comparing similar multiband superconductors , cecu@xmath0si@xmath0
@xcite , ube@xmath10 @xcite and kfe@xmath0as@xmath0 @xcite .
the first two are heavy fermion materials known to have multiband with all full gaps belonging to the spin singlet category where @xmath3 is strongly suppressed and the hidden first order like anomalies at @xmath271 exist .
thus as mentioned @xmath235 is indispensable for those systems .
according to the recent sans experiment @xcite the typical multiband fe pnictide kfe@xmath0as@xmath0 is similar to sr@xmath0ruo@xmath1 in the point that the vortex anisotropy @xmath48 observed differs from the @xmath3 anisotropy .
this can be also explained by ppe .
thus the present theoretical framework , which is quite general might be able to explain those systems .
we have assigned the nodal structure that the dominant @xmath6 band is a full gap while the minor @xmath4 is vertical line nodes . in order to reproduce
the square vortex lattice oriented to the ( 110 ) direction observed for @xmath147 which dominates the whole space measured in the @xmath13 vs @xmath156 phase @xcite .
the gap ( or near ) nodes should be in this direction in reciprocal space @xcite , implying a @xmath7 like nodal structure , contrary to the claim by deguchi et al @xcite who measure the specific heat by rotating the field direction and see the four - fold oscillation patterns whose minima are located for the ( 100 ) direction .
they conclude that the @xmath4 band has a @xmath275 like nodal structure .
however , this assignment is difficult to explain the square vortex lattice orientated along the ( 110 ) direction for @xmath147 . even if we take into account the in - plane fermi velocity anisotropy in the @xmath6 and @xmath5 bands whose minima are oriented along the ( 100 ) direction @xcite , thus preferring the square lattice oriented along the ( 100 ) direction as discussed previously @xcite .
thus the best way to avoid this difficulty is to simply consider that the @xmath4 band nodal structure is @xmath7 like .
the specific heat oscillation experiment done above 120mk is still too high to see the sign changing of the oscillation pattern because the @xmath4 band is minor where the expected sign changing temperature theoretically @xcite and experimentally @xcite usually located at 0.1@xmath276mk must be lowered .
this kind of the band - dependent nodal structure differs from the idea based on the symmetry protected nodal structure where the all plural bands are governed by the same gap symmetry . from this point of view
, it is possible that the @xmath7 nodal structure may not be truly sign - changing symmetry , rather an extended s - wave type with the anisotropic gap whose minima are along the ( 110 ) direction .
some authors @xcite assert that the horizontal line nodes compatible to a chiral triplet state @xmath277 . from the present point of view
, the sommerfeld coefficient @xmath278 behavior for @xmath147 may not be consistent with this which gives a too larger @xmath278 at low field region than the experimental data @xcite shown in fig .
[ fig4 ] . in this connection
we should mention that no one succeeded in explaining the @xmath278 behavior @xcite for @xmath158 .
in particular , it is viewed that the initial rise of @xmath158 is often assigned to the @xmath279 dos because the plateau of @xmath278 for this mid field region @xmath280 t seems to correspond to @xmath279 dos ( 43@xmath274 of the total ) , which was taken as supporting evidence of the @xmath4 major scenario .
since we can not accept this view anymore , a full understudying of the @xmath278 behavior for @xmath158 remains mystery .
there remain certainly important experiments to put forth the research front of this interesting material : nmr experiments have been extensively done @xcite , showing the absence of any change of the knight shift @xcite below @xmath95 .
this is interpreted as freely rotatable d - vector in a spin triplet pairing , keeping always it perpendicular to an external field which is as small as @xmath2mt , a big mystery because according to the recent arpes @xcite the spin - orbit coupling which locks the d - vector to the crystalline lattice is an order of 200mv . since the recent magnetization experiment @xcite is directly able to detect the spin susceptibility change at the first order transition @xmath281 which amounts to @xmath171 drop compared with the normal value .
this is in sharp contrast with the knight shift experiment by nmr which calls for reexamination of nmr experiments .
here we propose the @xmath282 measurement to detect the fflo state @xcite expected to exist along @xmath281 below @xmath2830.8k where the first order transition is observed .
recently anomalous @xmath284 enhancement is observed @xcite when entering the fflo state due to the appearance of the zero energy state at the domain walls where the fflo order parameter is @xmath8-shifted .
there is a good chance to observe it if the second phase below @xmath281 really exists , which we believe so .
a necessary condition is that @xmath285 is short enough so that the spin - lattice relaxation @xmath282 process is dominated through those zero energy states as in @xmath286-(bedt - ttf)@xmath0cu(ncs)@xmath0 case @xcite . in this connection
we mention a recent @xmath163sr experiment @xcite that probes the peculiar vortex morphology at low fields of @xmath147 and related theory based on hidden criticality associated with multi - bandness @xcite .
based on the microscopic eilenberger theory extended to a multiband case , we have studied the mixed state properties of a uniaxial anisotropic type ii superconductor , focusing on the interplay between the pauli paramagnetic effect and multiband effect .
a two band model calculation is set up and applied to sr@xmath0ruo@xmath1 .
we have succeeded in reproducing the data both of the form factors of sans experiments @xcite and magnetic torque experiment @xcite .
that leads us to the conclusion that to understand the physics in sr@xmath0ruo@xmath1 it is indispensable to consider both the pauli paramagnetic effect and multiband effect simultaneously , which conspire to give rise to a variety of mysteries associated with the pairing symmetry determination in this material .
as agreed with the previous identification based on the single band analysis @xcite , the pairing symmetry in sr@xmath0ruo@xmath1 is either singlet which is most likely or triplet with the d - vector locked in the @xmath15 plane which is less likely . the @xmath6 ( @xmath4 )
band is major ( minor ) with the mass anisotropy 60 ( 180 ) .
namely , the @xmath6 ( @xmath4 ) band has a lager ( smaller ) gap .
the gap structure is a full gap in the @xmath6 band while in the @xmath4 band it is @xmath7 like .
this simple picture was difficult to reach because of the extreme two - dimensionality of this material which prevents conventional experimental access .
now the dedicated and refined experimental tools @xcite which are able to align the magnetic field direction accurately within 1@xmath287 enable us to uncover the physics of sr@xmath0ruo@xmath1 .
the authors thank m. r. eskildsen , s. kuhn , and c. rastovski for providing us unpublished sans data and also m. ichioka , m. ishihara , and y. amano for discussing on the theoretical side , and s. kittaka , a. kasahara , t. sakakibara , k. ishida , y. maeno , s. yonezawa and m. takigawa for discussing on the experiment side .
k. m. is supported by grant - in - aid for scientific research no .
26400360 and no .
25103716 from the japan society for the promotion of science .
r. cubitt , m. r. eskildsen , c. d. dewhurst , j. jun , s. m. kazakov , and j. karpinski , phys .
lett . * 91 * , 047002 ( 2003 ) .
s. kittaka , y. aoki , y. shimura , t. sakakibara , s. seiro , c. geibel , f. steglich , h. ikeda , and k. machida , phys .
lett . * 112 * , 067002 ( 2014 ) .
s. kittaka , y. aoki , n. kase , t. sakakibara , t. saito , h. fukazawa , y. kohori , k. kihou , chul - ho lee , a. iyo , h. eisaki , k. deguchi , n. k. sato , y. tsutsumi , and k. machida , j. phys .
. jpn . * 83 * , 013704 ( 2014 ) .
s. kittaka , a. kasahara , t. sakakibara , d. shibata , s. yonezawa , y. maeno , k. tenya , and k. machida , phys .
b * 89 * , 220402 ( r ) ( 2014 ) .
m. ishihara , y. amano , m. ichioka , and k. machida , phys .
b * 87 * , 224509 ( 2013 ) .
m. ichioka and k. machida , phys .
b * 76 * , 064502 ( 2007 ) ; m. ichioka , k. m. suzuki , y. tsutsumi , and k. machida , in _ superconductivity - theory and applications _ , edited by a. m. luiz ( intech , croatia , 2011 ) , chap.10 .
s. kittaka , t. nakamura , y. aono , s. yonezawa , k. ishida , and y. maeno , phys .
b * 80 * , 174514 ( 2009 ) .
s. kittaka , t. nakamura , y. aono , s. yonezawa , k. ishida , and y. maeno , j. phys .
. ser . * 150 * , 052112 ( 2009 ) . c. g. fatuzzo , m. dantz , s. fatale , p. olalde - velasco , n. e. shaik , b. dalla piazza , s. toth , j. pelliciari , r. fittipaldi , a. vecchione , n. kikugawa , j. s. brooks , h. m. rnnow , m. grioni , ch .
regg , t. schmitt , j. chang , arxiv:1503.01574 .
|
we investigate the mixed state properties in a type ii multiband superconductor with uniaxial anisotropy under the pauli paramagnetic effects .
eilenberger theory extended to a multiband superconductor is utilized to describe the detailed vortex lattice properties , such as the flux line form factors , the vortex lattice anisotropy and magnetic torques .
we apply this theory to sr@xmath0ruo@xmath1 to analyze those physical quantities obtained experimentally , focusing on the interplay between the strong two - dimensional anisotropy and the pauli paramagnetic effects .
this study allows us to understand the origin of the disparity between the vortex lattice anisotropy ( @xmath260 ) and the @xmath3 anisotropy ( @xmath220 ) . among the three bands
; @xmath4 with the effective mass anisotropy @xmath2180 , @xmath5 with @xmath2120 , and @xmath6 with @xmath260 , the last one is found to be the major band , responsible for various magnetic responses while the minor @xmath4 band plays an important role in the vortex formation .
namely , in a field orientation slightly tilted away from the two dimensional basal plane those two bands cooperatively form the optimal vortex anisotropy which exceeds that given by the effective mass formula with infinite anisotropy .
this is observed by small angle neutron scattering experiments on sr@xmath0ruo@xmath1 .
the pairing symmetry of sr@xmath0ruo@xmath1 realized is either spin singlet or spin triplet with the d - vector strongly locked in the basal plane .
the gap structure is that the major @xmath6 band has a full gap and the minor @xmath4 band has a @xmath7 like gap .
|
the focus of research in nonlinear dynamics has evolved from temporal instabilities over simple spatial patterns to complex spatio - temporal behavior and the control or synchronization of such dynamics .
complex spatio - temporal behavior in reaction - diffusion equations , which is in a wider sense the class of equations dealt with also in electrochemistry , might be found when instabilities breaking time and space symmetries interact .
a generic case is the interaction of turing @xcite and hopf bifurcation in a two component activator - inhibitor system in which the involved species diffuse .
complex spatio - temporal dynamics has been found near this codimension - two point theoretically @xcite as well as experimentally @xcite . in electrochemical systems that can be described by a two component model one variable typically is of electrical nature and the associated transport mechanism is migration rather than diffusion @xcite .
the decisive variable for the dynamics of the electric circuit is the double layer potential @xmath0 , measuring the voltage drop across the interface between the working electrode and the electrolyte solution @xcite .
local perturbations in the double layer potential are mediated through the electric field in the electrolyte .
thus , spatial inhomogeneities in the double layer potential are felt not only by its nearest neighbors , but by a whole range of neighboring sites which makes the coupling nonlocal @xcite .
the degree of non - locality depends on the geometry of the electrochemical cell , most importantly on the positions of the working ( we ) , counter ( ce ) and reference electrode ( re ) with respect to each other @xcite .
furthermore it can be shown that the galvanostatic operation mode ( constant current control ) introduces an additional global coupling into the system @xcite .
the role of the second variable in two component activator - inhibitor systems in electrochemistry is played , e. g. , by the chemical concentration of the reacting species in the double layer or by the density of adsorbed molecules on the we . over the last decade global coupling
has been an active area of research .
global coupling is present in systems that are subject to external control , e.g. via an electric circuit ( such as in electrochemical @xcite , semiconductor @xcite and gas discharge @xcite systems ) or via the electric control of the temperature in catalytic reactors @xcite .
but global coupling may also be due to transport processes that happen on time scales much faster than all other relevant time scales in the system , e. g. fast mixing in the gas phase @xcite .
a variety of other systems are described by dynamics that include global coupling , e.g. , ferromagnetic @xcite , biological @xcite , and chemical systems in which the global coupling can be light induced @xcite .
abstract theoretical models are discussed , e. g. , in @xcite .
results regarding electrochemical systems with global coupling have been reported for systems with an n - shaped current - voltage characteristic ( termed n - ndr systems : -shaped egative ifferential esistance ) for different types of global coupling . in these systems the double layer potential acts as
an activator and global coupling introduced by the galvanostatic control mode was shown to accelerate front motion @xcite thus having a synchronizing effect on the spatial dynamics .
desynchronizing global coupling of the activator was shown to stabilize potential fronts , leading to two stationary potential domains @xcite . also the formation of pulses and standing waves was observed @xcite .
in electrochemical systems with an s - shaped current - voltage characteristic ( s - ndr ) the roles of activator and inhibitor are reversed , leading to a global coupling of the inhibitor under galvanostatic conditions .
this leads to the opposite effects opposed to n - ndr systems , i.e. current domains that are stabilized by the constant current constraint @xcite .
similar results on accelerated and decelerated fronts in globally coupled semiconductors with s- or z - shaped current - voltage characteristics have also been obtained @xcite . in the present paper we focus on the latter case of global coupling of the inhibitor in an electrochemical system with an s - shaped current - voltage curve .
furthermore , we consider systems with high electrolyte conductivity .
in such systems the migration coupling is so efficient that any spatial variation in @xmath0 can be neglected , which results in an additional global coupling @xcite .
the set of equations to be investigated is thus of the general form : @xmath1 where @xmath2 stands for the activator variable , whose dynamics comprises an autocatalytic chemical step .
the angular brackets denote the spatial average over the spatial domain @xmath3 .
@xmath4 is autocatalytic in @xmath2 ; @xmath5 exhibits a monotonic characteristic as a function of @xmath0 and @xmath2 . @xmath6
denote the characteristic times for changes in @xmath0 and @xmath2 , respectively .
a formally very similar set of equations describes the dynamics in bistable semiconductor systems operated via an external load resistance @xcite .
the formation and dynamics of current density patterns in bistable semiconductors was extensively studied @xcite . in this respective class of semiconductor systems
the current - voltage characteristic also resembles the shape of an s , which points to the fact that the roles of the dynamic variables are very similar to the electrochemical model : the voltage drop @xmath7 across a semiconductor device acts effectively as inhibitor , and it is subject to a global constraint imposed by the external electric circuit .
the role of the activator variable might be played by different physical quantities , such as the electron temperature @xcite , the concentration of excess carriers @xcite , the charge density in resonant tunneling structures @xcite ( note that for bistable resonant tunneling structures the current - voltage characteristic is z - shaped resulting in an _ activatory _ , not inhibitory effect of the global constraint ) , the voltage drop across pn - junctions in thyristors @xcite or the interface charge density in a heterostructure hot electron diode ( hhed ) @xcite .
the dynamic equations are of the same structural form as eq .
( [ eq : algphi]),([eq : algtheta ] ) ; only the local nonlinear functions @xmath4 and @xmath5 differ from the electrochemical model . for the current density dynamics in a class of models originally derived for the hhed in one or two spatial dimensions under galvanostatic ( current - controlled ) conditions , interesting complex spatio - temporal patterns termed `` spiking '' and `` breathing '' current filaments were found @xcite .
recently , a sufficient condition for the onset of such complex spatio - temporal dynamics was given @xcite .
realizing the obvious similarities , we show in this paper that the methods ( e. g. for analyzing the dynamics ) developed for the semiconductor system can be applied to gain insight into the interaction of different instabilities in the electrochemical system .
results regarding the possibility of the occurrence of complex spatio - temporal behavior and the mechanisms that lead to such behavior are given .
it is emphasized whether the different dynamical regimes depend upon the general structural form of the equations , especially regarding the influence of global coupling , or if they are due to special properties of the underlying physical or chemical system , and thus the local dynamics .
hence a comparison of electrochemical and semiconductor systems gives considerable insight into generic complex dynamics of globally coupled bistable systems .
the paper is organized as follows . in the next section
we introduce the electrochemical model , discuss its important parameters and the mechanisms leading to global coupling in the model . in the following section we characterize the dynamics of the model by linear stability analysis along the lines developed for the semiconductor model and by numerical simulations . in the discussion
we compare the important features of the two models .
the mechanism leading to complex spatio - temporal behavior in both models is different and this difference is explored in this section in some depth .
we summarize our results in the last section and give a short outlook to applications in terms of experimental verifications and transfer of the electrochemical results to the semiconductor model .
the central variable in electrochemical pattern formation is the double layer potential @xmath0 , the potential drop across the interface between the we and electrolyte solution .
the dynamic evolution equation for @xmath0 can be deduced from the local charge balance at the electrode / electrolyte interface . to make things as transparent as possible and to facilitate later comparison with the semiconductor model
, we employ the idealized geometry shown in fig .
[ fig:0 ] . we and ce are equally sized rectangular plates positioned parallel to each other in a box - like cell with otherwise insulating walls .
this geometry imposes no - flux boundary conditions for @xmath0 and @xmath2 ; there will be no spatial inhomogeneities of the electric field at the interface imposed by this geometry .
for very high electrolyte conductivities @xmath8 , spatial inhomogeneities in the double layer potential are damped out very fast via the efficient coupling through migration currents .
it follows that spatial variations of @xmath0 can be neglected .
this effectively introduces a global coupling in the system , since local perturbations in @xmath0 are felt instantaneously in the whole double layer . in the following we additionally assume current controlled conditions .
galvanostatic control is known to introduce an additional global coupling into the system @xcite . assuming a specific double layer capacitance @xmath9 , the dynamic equation for the double layer potential reads @xmath10 where @xmath11 is the reaction current density and @xmath12 denotes the imposed current density .
the activator variable @xmath2 describes the evolution of the coverage of the we by an adsorbate or the concentration of a chemical species in the reaction plane .
its dynamics will be modeled by an equation of the form ( [ eq : algtheta ] ) , where we restrict our system to one spatial dimension ( 1d ) since the qualitative behavior should also be captured on 1d domains .
1d domains also resemble the situation of a very large aspect ratio of the rectangular domain , where one spatial dimension is too small to allow for spatial instabilities and can thus be eliminated .
we use the following model functions for the local dynamics of the activator and the reaction current density @xmath13 @xmath14 originally derived to describe pattern formation observed in a reaction , in which a reaction inhibiting adsorbate undergoes a first order phase transition due to lateral interactions of the adsorbate molecules @xcite .
the transformations leading to dimensionless units differ from the ones given in @xcite ; the derivation is given in appendix [ app1 ] .
note the non - polynomial nature of the function @xmath4 .
the dimensionless set of equations is thus @xmath15 subject to the boundary conditions @xmath16 we normalize space to the interval @xmath17 $ ] for computational convenience and thus @xmath18 this leads to the proportionality of the parameters ; @xmath19 and @xmath20 still can be changed independently since also other physical quantities enter these parameters ( cf .
appendix [ app1 ] ) . in fig .
( [ fig:1]a ) the nullclines of the system are depicted for a current density @xmath12 that is set in the range of the negative differential resistance in the current - voltage characteristic ( see fig .
[ fig:1]b ) .
the s - shaped current - voltage - characteristic is depicted together with the load line @xmath21 in fig .
[ fig:1]b .
this physically more intuitive -plane representation will be used in the following .
the parameters @xmath22 , @xmath23 and @xmath5 are fixed throughout this paper at the values @xmath24 , @xmath25 and @xmath26 ( cf .
appendix [ app1 ] ) .
the dynamics is determined by the model parameters @xmath19 , essentially proportional to @xmath27 , the relaxation time ratio of activator and inhibitor @xmath28 ( independent of @xmath29 ) , and the general excitation level controlled by the imposed current density @xmath12 .
the relaxation time ratio can be accessed easily via the concentrations of the reacting and adsorbing species ; @xmath12 is set by the galvanostatic control unit .
the numerical results discussed in section [ homlistatdo ] were obtained using pseudo spectral decomposition in space @xcite employing 15 spatial cosine - modes ( the results do not change when a larger number of modes is chosen ) . for the integration in time the routine lsode @xcite and for continuation of steady states and limit cycles
the package auto @xcite was used .
in this section we consider the spatially uniform fixed points of the system ( [ eq : nondimphi]),([eq : nondimtheta ] ) and their bifurcations .
the uniform steady state @xmath30 is given by , and corresponds to the homogeneous s - shaped current - voltage characteristic ( fig .
[ fig:1]b ) . perturbing the steady state with a perturbation @xmath31 ( consistent with the boundary conditions ) , the temporal evolution of the perturbation
is given by the eigenvalues of the jacobian matrix @xmath32 @xmath33 and stability @xmath34 implies that @xmath35 .
the jacobian reads @xmath36 subscripts denote partial derivatives with respect to the subscripted variable and evaluation at the steady state ( e. g. ) . for brevity
we denote .
the stability of the fixed point with respect to homogeneous fluctuations can be determined by inspecting @xmath37 and @xmath38 obviously in general since and , which follows from the fact that the branch of negative differential resistance ( ) is caused solely by the activator variable @xmath2 , equivalent to saying that in general . however ,
might change sign on the ndr - branch since and , which leads to an oscillatory instability ( hopf bifurcation ) of the homogeneous steady state ( denoted by a superscript `` h '' , cf .
table [ tab:2 ] ) at .abbreviations for bifurcation points [ cols= " < , < " , ]
the model equations ( [ eq : functheta]),([eq : funcir ] ) are retained via the transformations of the variables according to @xmath39 and with the introduction of the parameters @xmath40 with the values given in table [ tab:1 ] the parameters @xmath23 , @xmath22 and @xmath5 are fixed to was given as in previous papers @xcite , but the above value fits the physical situation better . ] , and .
they correspond to such physical values as free adsorption sites or interaction strength .
@xmath20 depends on the well accessible concentration of the reacting species , which can be varied over several decades ; ; @xmath12 can be set by the galvanostatic control unit and typical values will be of order .
in fig . [ fig:11 ] the bifurcations and phase portraits near the codimension - two point in which the domain bifurcation and hopf bifurcation of the homogeneous steady state meet ( dh ) is shown .
the additional branches not shown in fig . [ fig:8 ]
are a hopf bifurcation of the unstable stationary domain leading to an unstable inhomogeneous limit cycle and the pitchfork bifurcation of periodic orbits that stabilizes the homogeneous limit cycle born in the hopf bifurcation of the homogeneous steady state and which is the origin of another unstable inhomogeneous limit cycle ( cf .
[ fig:4]a ) .
both branches terminate in the dh .
the respective phase portraits ( insets ) depict the dynamics schematically in a projection on the plane spanned by the eigenvectors of the two complex conjugate eigenvalues describing the hopf bifurcation of the homogeneous fixed point resp . the stationary unstable domain .
the third direction describes the subcritical domain bifurcation ( spatial mode ) .
|
spatio - temporal bifurcations and complex dynamics in globally coupled intrinsically bistable electrochemical systems with an s - shaped current - voltage characteristic under galvanostatic control are studied theoretically on a one - dimensional domain .
the results are compared with the dynamics and the bifurcation scenarios occurring in a closely related model which describes pattern formation in semiconductors . under galvanostatic control
both systems are unstable with respect to the formation of stationary large amplitude current domains .
the current domains as well as the homogeneous steady state exhibit oscillatory instabilities for slow dynamics of the potential drop across the double layer , or across the semiconductor device , respectively .
the interplay of the different instabilities leads to complex spatio - temporal behavior .
we find breathing current domains and chaotic spatio - temporal dynamics in the electrochemical system . comparing these findings with the results obtained earlier for the semiconductor system , we outline bifurcation scenarios leading to complex dynamics in globally coupled bistable systems with subcritical spatial bifurcations .
|
a quantum optical system utilizing trapped neutral atoms or ions for qubits is one of promising candidates for implementing a quantum computer @xcite .
actually , there have been numerous theoretical @xcite and experimental @xcite achievements showing the positive prospects for it .
the number of qubits in such a system is , however , obviously limited by the size of the trapping structure , while one of the essential factors for a useful quantum computer is the scalability .
this difficulty could be overcome by connecting partially implemented quantum computation nodes to form a quantum network . for any unitary operation for the whole quantum network to be possible , controlled unitary operations between two nodes
should be performed as well as local unitary operations at each node @xcite .
there have been various ways for doing it by means of one or more ancilla qubits @xcite .
the underlying idea is to use ancilla qubits to transfer the quantum information between two nodes and perform local two - qubit operations at the nodes so that the overall process in effect results in the desired global two - qubit operation .
one method to accomplish the task is to shuttle ancilla qubits or a quantum node itself physically to a particular position where local interaction between an ancilla qubit and a quantum node is possible @xcite .
this method , however , can not be directly applicable to neutral atom quantum computers .
another method feasible for neutral atom quantum computers as well is to exploit a photon - mediated interaction between two nodes , such as in entanglement generation @xcite , quantum state transfer @xcite , and quantum teleportation @xcite . on the other hand ,
there have also been a scheme in which no ancilla qubit is involved @xcite .
the scheme , however , uses a quantum interferometer and the complex atomic structure of several hyperfine levels instead . in this work
, we introduce a simple scheme to do a controlled unitary operation between two distant atoms .
a common quantum communication setup @xcite , in which two atoms each trapped in an optical cavity directly communicate through a quantum transmission line such as an optical fiber connecting the two cavities , is considered .
in contrast to earlier methods , no ancilla atomic qubit is involved in our scheme and the gate operation is done by a simple coherent process .
and @xmath0.,scaledwidth=40.0% ] the schematic representation of our scheme is depicted in fig .
[ fig : scheme ] . atom a and b are trapped in cavity a and b , respectively , and two cavities are connected through an optical fiber of length @xmath1 . the decay rate of the cavity is @xmath2 and the spontaneous emission rate of the atom is @xmath3 . the outside mirror , i.e.
, the mirror on the side not connected to the fiber , of each cavity is assumed to be of 100% reflectivity .
each solid arrow represents a transition by a classical field of rabi frequency @xmath4 ( @xmath5 ) , and each dotted arrow represents a transition by a cavity mode of atom - cavity coupling rate @xmath6 ( @xmath7 ) .
we assume the lamb - dicke limit @xcite , thus @xmath6 is assumed to be a constant .
each transition is detuned by an amount of @xmath8 or @xmath9 as shown in fig .
[ fig : scheme ] .
a qubit is represented by two ground hyperfine levels @xmath10 and @xmath0 .
@xmath10 does not participate in the transition .
the desired two qubit operation is the controlled phase shift operation , which is accomplished in three steps as the following : @xmath11 in the first step , only the state @xmath12 is transferred to @xmath13 while other states remain unchanged .
it is done with high precision by the well - known technique of adiabatic passage @xcite . for this step ,
two classical fields of rabi frequency @xmath14 and @xmath15 are applied adiabatically to atom b in order . here
, detuning parameter @xmath16 has to be much larger than atom - cavity coupling rate @xmath17 ( @xmath18 ) for no cavity photon to be generated during the adiabatic passage process .
the third step is simply the inverse of the first step , and is also achieved by adiabatic passage .
the most important and nontrivial part of our scheme is the second step , in which only the state @xmath19 acquires a phase @xmath20 while other states remain unchanged . in the remainder of this paper , we concentrate on explaining the second step of operation ( [ eq : step ] ) . in the second step , a classical field of rabi frequency @xmath21
is applied to atom a adiabatically , whereas both the classical fields of rabi frequencies @xmath22 and @xmath23 for atom b are turned off .
let us assume that @xmath24 has a gaussian form : @xmath25 .
\label{eq : rabi}\ ] ] if the initial state of atom a is @xmath26 , this operation has no effect on the system . if atom a is initially in state @xmath27 , however , this operation transfers the population to @xmath28 , during which a single photon is generated in cavity a and emitted out of the cavity @xcite .
the time width @xmath29 of the emitted photon pulse is of order @xmath30 . with that in mind
, we investigate the system in two regimes .
first , we consider the short - distance regime in which the interaction time between two distant cavities is sufficiently short so that the whole system can be regarded to remain in a steady state at all times .
this regime is represented by the following condition : @xmath31 where @xmath32 is the speed of light . in this case , the whole system can be treated within the context of adiabatic theorem . if the system is initially in state @xmath33 , the hamiltonian in the rotating frame is written as @xmath34,\\ h_{b}&=0 , \\ h_{c}&=\sum_{n=-\infty}^{\infty } n\delta\omega\ , c_{n}^{\dagger}c_{n } \\ & \quad+i\kappa'\sum_{n=-\infty}^{\infty}[a_{a}^{\dagger}c_{n}+(-1)^{n}a_{b}^{\dagger}c_{n}-h.c .
] , \end{split } \label{eq : hamiltonian1}\ ] ] where @xmath35 , @xmath36 , and @xmath37 are the hamiltonians for atom a , atom b , and the fiber , respectively , @xmath38 ( @xmath39 ) is the field operator for cavity a ( cavity b ) , @xmath40 is the field operator for the @xmath41th fiber mode , @xmath42 is the frequency difference between two adjacent fiber modes , and @xmath43 is the effective cavity decay rate . the factors @xmath44 are introduced to model the phase difference @xmath45 between two ends of the fiber for every second modes .
hamiltonian @xmath35 has the dark state @xmath46 where @xmath47 is given by @xmath48 . here
, we represent a state of the atom - cavity system as @xmath49 where @xmath50 is the atomic state and @xmath41 is the cavity photon number .
the dark state of the total hamiltonian @xmath51 is also derived as @xmath52 where @xmath53 and @xmath54 denote the vacuum fiber and one photon in the @xmath55th fiber mode , respectively . from this expression , it is clear that after the classical pulse operation given by eq .
( [ eq : rabi ] ) the system just returns to its initial state since both the initial and the final values of @xmath56 are 0 , i.e. , @xmath57 . during this operation
, the dark state acquires no dynamical phase since the energy of dark state @xmath58 is 0 .
consequently , @xmath59 in the second step of operation ( [ eq : step ] ) is justified .
if the system is initially in state @xmath60 , the hamiltonian ( [ eq : hamiltonian1 ] ) is modified so that the atom - cavity interaction in cavity b is involved .
we assume a large detuning ( @xmath61 ) and take advantage of adiabatic elimination @xcite .
the effective hamiltonian for atom b now reads @xmath62 this hamiltonian can be regarded as a perturbation to hamiltonian ( [ eq : hamiltonian1 ] ) .
let @xmath63 be the perturbed eigenstate of dark state ( [ eq : dark ] ) ( with @xmath64 ) and @xmath65 be its eigenenergy .
the value of @xmath65 is no longer zero for nonzero @xmath24 . after the classical pulse of eq .
( [ eq : rabi ] ) , the system also returns to its initial state @xmath60 as in the previous case . during this operation , however , the state @xmath63 acquires a dynamical phase @xmath20 given by @xmath66 where @xmath67 is the operation time .
the value of @xmath20 depends linearly on the width @xmath68 of the classical pulse ( [ eq : rabi ] ) and increases as the height @xmath69 increases .
the dependency of @xmath20 on @xmath69 can be obtained by numerical simulation .
( solid curve ) and @xmath70 ( dotted curve ) . @xmath71 and @xmath72
denote the acquired phases during the operation for the initial states @xmath73 and @xmath19 , respectively . in case of @xmath74 ,
( b ) probabilities @xmath75 ( solid curve ) and @xmath76 ( dotted curve ) , and ( c ) fidelity @xmath77 for an initial state @xmath78 .
the selected numerical parameters are @xmath79 and @xmath80.,scaledwidth=45.0% ] we carry out numerical simulations , by directly solving the schrdinger equation without adiabatic approximations , for the second step of operation ( [ eq : step ] ) with a set of selected parameters : @xmath79 and @xmath80 . in the numerical simulation
, we also take into account the photon loss in the fiber by introducing photon loss rate @xmath81 of the fiber and adding terms @xmath82 in hamiltonian ( [ eq : hamiltonian1 ] ) . if a photon is lost due to the lossy fiber or the spontaneous decay of the atom , the system collapses into one of the ground states losing its phase information . in the case of the system being collapsed into @xmath27 or @xmath13 , it only takes the effect of lowering the fidelity of the whole operation .
if the system is collapsed into @xmath28 or @xmath12 , however , we are faced with another problem that the third step of opertation ( [ eq : step ] ) does not transform the state into one in the qubit subspace . in order that the resulting state be confined in the qubit subspace
, we thus perform optical pumping , after the second step , by applying two classical fields corresponding to transitions @xmath83 and @xmath84 , which induce population transfers from @xmath28 to @xmath27 and from @xmath12 to @xmath13 , respectively .
such an optical pumping process takes no effect when the state already exists in the desired subspace . for the initial state @xmath33 , let @xmath85 be the probability that no photon is lost during the operation , and let @xmath86 be the phase that is acquired when no photon is lost
let us also assign a probability @xmath87 and a phase @xmath88 in the same manner for the initial state @xmath60 .
the whole process is then summarized as the following operator - sum representation @xcite ( omitting indices a and b for brevity ) : @xmath89 where @xmath90 in fig .
[ fig : phi](a ) , we plot @xmath91 ( solid curve ) and @xmath70 ( dotted curve ) with respect to @xmath69 .
our numerical results indicate that the acquired phases are nearly independent of the fiber loss rate @xmath81 in our parametric regime .
as predicted above , the initial state @xmath73 acquires only a small phase @xmath71 , whereas the initial state @xmath19 acquires a phase @xmath72 which increases as @xmath92 increases .
the small phase change @xmath71 can be compensated by means of a single - qubit phase operation so that @xmath93 becomes the only relevant phase . in fig .
[ fig : phi](b ) , we plot @xmath75 ( solid curve ) and @xmath76 ( dotted curve ) with respect to @xmath81 . here , we have chosen @xmath74 , in which case @xmath93 is found to be @xmath45 .
given an initial state @xmath94 , the fidelity of the operation is easily derived as @xmath95^{1/2}. \end{split } \label{eq : fidelity}\ ] ] with the same parameters as above and an initial state @xmath96 , we plot in fig .
[ fig : phi](c ) the fidelity @xmath77 with respect to @xmath81 .
the fidelity decreases with the fiber loss , but it remains high as long as @xmath97 .
this observation , along with the fact that the photon loss is dominated by the spontaneous decay of the atom when @xmath97 , leads to a conclusion that the spontaneous decay of the atom does not have critical effect since it is suppressed due to dark state evolution and the large detuning .
now , we consider another regime , namely , the long - distance regime in which the time width @xmath98 of the single photon pulse leaking out from cavity a satisfies the following condition : @xmath99 in this case , the input / output process at each cavity can be treated separately .
we set the detuning parameter as @xmath100 for this regime .
first , we consider the output process at cavity a. as we have considered in the previous regime , hamiltonian @xmath35 has the dark state @xmath101 . if atom a is in state @xmath102 initially , the population is transferred to @xmath103 as @xmath24
is gradually increased from zero . during this adiabatic passage process , a single photon leaks out from the cavity . in the adiabatic limit ,
the pulse shape @xmath104 of the emitted photon can be calculated analytically as @xcite @xmath105 .
\label{eq : pulse}\ ] ] the output photon propagates through the fiber and reflects at cavity b. the input / output process at cavity b is described by the boundary condition @xcite : @xmath106 and the quantum langevin equation : @xmath107&-[b , a_{b}^{\dagger}]\left(\frac{\kappa}{2}a_{b}-\sqrt{\kappa}c_{in}(t)\right)\\ & + [ b , a_{b}]\left(\frac{\kappa}{2}a_{b}^{\dagger}-\sqrt{\kappa}c_{in}^{\dagger}(t)\right ) , \end{split } \label{eq : langevin}\ ] ] where @xmath108 and @xmath109 are the input and output field operator , respectively , and @xmath110 is any operator for atom - cavity system b. let us assume that the time derivative of any operator for system b vanishes , i.e. , @xmath111 .
this assumption is justified since the input photon pulse is generated by an adiabatic process .
we also assume the strong coupling limit @xmath112 . for the initial state of @xmath113 , the hamiltonian reads @xmath114 .
the time derivative of @xmath115 is thus derived from eq .
( [ eq : langevin ] ) as @xmath116 . from this equation , we get @xmath117 , and by substituting it into eq .
( [ eq : boundary ] ) the relationship between the cavity input and output is derived as @xmath118 . on the other hand ,
the hamiltonian for the initial state of @xmath13 reads @xmath119 in this case , we derive the time derivative of @xmath120 as @xmath121 .
since @xmath122 has a nonzero value , we come to a conclusion that @xmath123 . thus the input / output relationship reads @xmath124 .
consequently , the single photon reflected at cavity b acquires a different phase 0 or @xmath45 according to the state of atom b @xcite .
after the reflection at cavity b , the photon finally reaches cavity a. by applying an appropriate classical field of rabi frequency @xmath24 , the photon is completely absorbed in atom a and the atomic population is transferred to @xmath102 by adiabatic passage @xcite .
complete absorption of the photon is guaranteed if no photon is reflected during this operation , for which @xmath24 has to be adjusted to satisfy the following condition
@xcite : @xmath125 the different phase of 0 or @xmath45 acquired at atom b results in the different phase of the final atomic state .
the resulting atomic state thus acquires a conditional phase @xmath126 as given in the second step of operation ( [ eq : step ] ) .
we note that the value of @xmath20 is insensitive to the particular system parameters , which is a strong point of the scheme . of the classical pulse , ( b ) real parts of the amplitudes of states @xmath33 ( solid curve ) and @xmath60 ( dotted curve ) , and ( c ) fidelity of the operation for an initial state @xmath78 .
the selected numerical parameters are @xmath127 , @xmath128 , and @xmath129.,scaledwidth=45.0% ] in fig .
[ fig : regime2 ] , we numerically demonstrate a typical gate operation process in this long - distance regime .
the selected numerical parameters are @xmath127 and @xmath128 . in fig .
[ fig : regime2](a ) , we plot rabi frequency @xmath24 with respect to time .
we set rabi frequency @xmath24 differently from a gaussian shape given by eq .
( [ eq : rabi ] ) to take advantage of an analytic solution for the cavity input / output equations ( [ eq : pulse ] ) and ( [ eq : matching ] ) @xcite . at the beginning of the gate operation , a classical field of rabi frequency @xmath24 satisfying @xmath130\mathrm{sech}[2t'/\delta_{t}]}$ ] , where @xmath131 and @xmath132 , is applied to atom a. this classical field generates a cavity photon , which leaks out from the cavity with a pulse shape given by @xmath133 $ ] .
the generation of the single photon pulse and the reflection of this pulse at cavity b are simulated numerically . here
cavity b is assumed to be located at @xmath129 . in order to absorb the photon into atom - cavity
a without reflection , rabi frequency @xmath24 is finally adjusted to satisfy @xmath134\mathrm{sech}[2t''/\delta_{t}]}$ ] , where @xmath135 and @xmath136 , as shown in fig .
[ fig : regime2](a ) .
we have confirmed from the numerical results that nearly no photon is emitted from cavity a during this final process .
the real parts of the amplitudes of states @xmath33 and @xmath60 are plotted as a solid curve and a dotted curve , respectively , in fig .
[ fig : regime2](b ) .
it is clearly shown that the acquired phase corresponds to the second step of operation ( [ eq : step ] ) .
the above rather idealized analysis gives the basis for the following generalized one in which the photon loss of the fiber is taken into account .
we again introduce two probabilities @xmath75 and @xmath76 as in the previous analysis for the short - distance regime . @xmath75 and @xmath76 denote the probabilities that no photon is lost during the operation for the initial states @xmath73 and @xmath19 , respectively . the analysis laid out in case of the short - distance regime is applied for the current case in the same manner . by means of the same sort of optical pumping
, the state can be confined in the qubit subspace , and the fidelity is also given by eq .
( [ eq : fidelity ] ) .
the above numerical simulation gives @xmath75 and @xmath76 in case of no photon loss in the fiber .
the values are found to be @xmath137 and @xmath138 , which gives the fidelity @xmath139 when the initial state is chosen as @xmath78 .
if the fiber is not lossless and the photon loss rate per unit length of the fiber is denoted as @xmath140 , we get modified probabilities @xmath141 and @xmath142 for the two initial states , respectively . by substituting these probabilities into eq .
( [ eq : fidelity ] ) and choosing the initial state as @xmath78 , we plot in fig .
[ fig : regime2](c ) the fidelity @xmath77 of the operation with respect to @xmath140 . as in the short - distance regime ,
the numerical results show that the gate works faithfully when the photon loss in the fiber is small , namely @xmath143 , and the spontaneous decay of the atom does not have critical effect .
in summary , we have shown that a two - qubit controlled unitary operation between two distant atoms is allowed by simply connecting them through a quantum transmission line .
we have analyzed it in two regimes , namely , the short - distance regime and the long - distance regime . the scheme is based on the adiabatic passage and the cavity qed interaction . provided a single photon is reliably transmitted between two cavities , the gate works with high fidelity due to the inherent resistance of the adiabatic evolution against spontaneous decay . since our scheme is much simpler than other indirect methods , it is expected to improve the scalability of quantum computers based on trapped neutral atoms or ions .
j. i. cirac and p. zoller , phys .
lett . * 74 * , 4091 ( 1995 ) ; t. pellizzari , s. a. gardiner , j. i. cirac , and p. zoller , phys . rev . lett . * 75 * , 3788 ( 1995 ) ; j. pachos and h. walther , phys . rev . lett . * 89 * , 187903 ( 2002 ) ; x. x. yi , x. h. su , and l. you , phys .
* 90 * , 097902 ( 2003 ) .
j. hong and h .- w .
lee , phys .
. lett . * 89 * , 237901 ( 2002 ) ; x .-
feng , z .- m .
zhang , x .- d .
li , s .- q .
gong , and z - z .
xu , phys .
lett . * 90 * , 217902 ( 2003 ) ; l .- m .
duan and h. j. kimble , phys .
lett . * 90 * , 253601 ( 2003 ) .
|
we propose a scheme for implementing a controlled unitary gate between two distant atoms directly communicating through a quantum transmission line . to achieve our goal ,
only a series of several coherent pulses are applied to the atoms .
our scheme thus requires no ancilla atomic qubit .
the simplicity of our scheme may significantly improve the scalability of quantum computers based on trapped neutral atoms or ions .
|
it is well established that both the total and spectral solar irradiance ( si , henceforth ) are modulated by the emergence and evolution of the magnetic field in the solar atmosphere . in particular , si variations measured on time scales from days to several solar rotations can be explained by accounting for the radiative effect of the presence of magnetic flux in sunspots , faculae , and network regions ( e.g. * ? ? ?
. however , there is still uncertainty in measurements of si variations on time scales of the activity cycle and longer ( e.g. * ? ? ?
* ; * ? ? ?
* ) , as well as in the determination of individual contributions of different magnetic features to the measured variations ( e.g. * ? ? ?
* ; * ? ? ?
the role of the quiet sun , that is those areas on the solar surface away from magnetically active and network regions , the so called inter network regions , is particularly debated .
for instance , long - term si variations unexplained by some recent models have been attributed either to a 0.2 k global change in the sun s effective temperature @xcite or to quiet sun variations associated with small - scale magnetic field changes during the activity cycle @xcite .
two main difficulties hinder the verification of these hypotheses .
first , investigations of temperature variations would be achievable only by measurements of absolute radiometric contribution of the quiet sun , and such measurements are not available yet .
second , the small - size magnetic features ( 100 km ) that are seen permeating the whole solar surface in high - resolution observations ( * ? ? ?
* and references therein ) , are unresolved in full - disk observations usually employed for si reconstructions @xcite , so that their contribution to si variations is hidden .
note that the origin of this so called quiet - sun magnetic field is also controversial , as it is not clear yet whether and to what extent , it results from the decay of active regions , turbulent dynamo effects or both ( e.g. * ? ? ?
* ; * ? ? ?
variations of quiet sun magnetic flux over the cycle would indicate a global dynamo origin of the field ; nevertheless , most of such measurements seem to indicate no temporal variations of quiet sun field ( * ? ? ?
* and references therein ) thus suggesting a local dynamo origin .
whether this result is due to lack of sensitivity of instrumentation and the observational techniques employed , or to a true constancy of the field over the cycle is still an open issue .
the presence of magnetic field influences the emitted radiation in magnetically sensitive lines directly , by the zeeman effect , and indirectly , by changing the thermodynamic properties of the plasma ( e.g. * ? ? ?
variations of line properties have been therefore employed to study the properties of quiet sun .
for instance , @xcite derived , by the investigation of the subtle broadening of a large sample of lines , an upper limit for the magnetic field amplitude of quiet sun network of 1000 - 2000 gauss and an upper limit of 90 gauss for intranetwork regions .
moreover , line parameters of a set of fraunhofer lines were observed from 1974 to 2006 @xcite , with the aim of inferring long - term variations of the photospheric temperature .
analysis of these data lead , however , to inconclusive results @xcite . in this study
we investigate the effect of magnetic features on the shape of two magnetically sensitive photospheric lines , the 630.2 and 617.3 nm , by analysing high - resolution spectro - polarimetric data of noaa * 11172 * observed on march 17th , 2011 at the dunn solar telescope with the interferometric bidimensional spectrometer ( ibis * ? ? ?
* ; * ? ? ?
qualitative comparison with results from numerical synthesis has been performed to disentangle between the direct and indirect effects of magnetic field on the measured line shapes variations .
we then investigate the effects of unresolved magnetic features on the shapes of the two lines by analysing ibis data spatially degraded to mimic medium resolution observations .
the 630.2 and 617.3 nm lines were chosen because they have been frequently employed for photospheric magnetic field studies .
moreover , data at these spectral ranges are currently acquired with several instruments , like the hinode / sot @xcite , solis / vsm @xcite , and sdo / hmi @xcite .
the article is organized as follows : in sec .
2 we describe the observations , the data reduction and the numerical synthesis ; in sec . 3
we discuss the line formation properties of the two lines ; in sec .
4 we present the results derived from analyses of observational data ; in sec .
5 we describe results obtained from synthesis ; in sec .
6 we investigate the effects of unresolved magnetic features ; in sec . 7
we discuss the results and draw our conclusions .
we investigated the behavior of the two lines at 617.3 and 630.2 nm in data acquired at the nso dunn solar telescope with ibis . in order to interpret the results derived from observations in terms of properties of the observed solar plasma
, we also synthetized the two lines in semi - empirical atmospheric models representative of quiet sun and magnetic regions .
+ + the ibis data set employed for this study was taken on march 17th 2011 from 14:19 ut to 15:19 ut .
it consists of 60 spectro - polarimetric scans of the two lines , each sampled at 22 spectral positions ; the nearby @xmath0 630.28 nm telluric line was also sampled at 6 spectral positions and served for line position calibration purposes . a full scan of the two lines took @xmath1 60 seconds , so that each line was scanned in @xmath1 30 seconds .
this temporal cadence is short with respect to typical evolution time scales of solar photospheric phenomena , so that the integrity of each line during a scan can be assumed to be preserved .
the field of view ( fov ) was approximately 45@xmath290 arcsec and included a facular region and a small pore located at [ 16.5n,10.3e ] in between the leading and following parts of noaa 11172 .
white light ( 621.3 nm ) frames imaging the same fov were acquired contemporaneously to spectro - polarimetric data .
the spatial sampling of ibis and wl frames was 0.17pixel@xmath3 .
the high order adaptive optics system of the dst @xcite was tracking on a small pore in the fov during the acquisition , thus improving image quality and stability . the data set was reduced and compensated for instrumental blue - shift , and for instrument and telescope polarization by standard procedures ( i.e. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
residual aberrations introduced by the atmospheric seeing were estimated by applying multi frame blind deconvolution technique @xcite to wl images and were employed to destretch the spectro - polarimetric images .
the residual crosstalk between the stokes parameters and the rms noise level were estimated to be @xmath4 and of the order of @xmath5 the intensity in the nearby continuum , respectively .
an infrared blocker in the optical path reduced the spectral stray - light contamination , so that its residual , estimated as described in @xcite is below 8 @xmath6 and 4 @xmath6 for 617.3 nm and 630.2 nm , respectively . for the purpose of this study
, we analyzed a subsample of 10 scans having the highest rms - contrast in wl images .
figure [ dataexample ] shows an example of the analysed data .
we estimated the magnetic flux density along the line - of - sight in the fov by applying the center of gravity method ( cog ; * ? ? ?
* ) to the measured right- and left- circularly polarized signals .
@xcite showed that the cog provides good estimates of the magnetic flux at the formation height of typical photospheric line core for fields lower than 1.5 kgauss . to estimate uncertainties of magnetic flux density values derived with the cog we applied standard error propagation to the cog formula .
we found that the uncertainty is on average equal to 8 gauss for flux density values smaller than 150 gauss ; it increases monotonically for larger flux density values , although the relative uncertainty is approximately constant and equal to 3@xmath6 .
the scatter of values increases with the decrease of the estimated flux .
the average value of the uncertainties computed over the whole fov is 9 gauss , with a standard deviation of 6 gauss .
note , however , that the finite polarimetric sensitivity of ibis sets a lower limit of confidence on magnetic flux density estimates . to estimate this limit we computed the average absolute flux over those pixels in the fov whose total polarization signal , where @xmath7 is the continuum intensty of stokes i , and u , v and q are the corresponding stokes parameter . is lower than 3 @xmath8 and found a value of 20 gauss with a standard deviation of 10 gauss .
we therefore defined as quiet - sun all those pixels whose absolute magnetic flux density is smaller that 30 gauss ( that is , below one standard deviation from the mean ) . ] next , in order to characterize the shapes of the lines , we computed the core intensity ( ic , hereafter ) , the equivalent width ( eqw , hereafter ) , and the full width at half maximum ( fwhm , hereafter ) of the two lines at each image pixel of the fov .
in particular , the ic and fwhm were computed by fitting the line sampling with a gaussian function .
we computed the three line shape parameters only for those pixels in the fov with a magnetic flux below 1000 gauss , thus excluding the small pore and the area around it from the analysis , but including network and facular elements .
this allowed us to investigate the dependence of line shape parameters on magnetic flux density variations .
we performed synthesis of the two line profiles in various one - dimensional semi - empirical static models representative of various thermodynamic conditions of the plasma and for different values of magnetic field strengths .
in particular , we employed fal - c , fal - e , fal - h and fal - p @xcite models , representative of quiet sun , network , facular and bright facular regions , respectively .
the spectral synthesis was performed in local thermal equilibrium with the rh code @xcite .
the land factor is 2.5 for both lines ; the other atomic data adopted for the synthesis are summarized in table [ tabellaid ] .
the three line parameters ic , fwhm and eqw were then computed for emergent spectra calculated at vertical line of sight .
ccc line ( nm ) & @xmath9 & @xmath10 ( ev ) + 617.3 & -2.88 & 2.22 + 630.2 & -1.24 & 3.68 +
* in order to discuss the results obtained from the two lines we first investigated their formation heights .
to this aim we analysed results derived from spectral synthesis of the two lines through fal - c and fal - p models , the latter with uniform magnetic field of 1000 gauss*. we then computed the response function ( rf , hereafter ) of ic of both lines to perturbations of either the temperature ( t_rf ) or magnetic field ( b_rf ) on both atmosphere models .
figure [ response_t ] shows the t_rf of ic for both lines and models .
we find that the 617.3 nm line core in fal - c forms at a wide range of photospheric heights , from 0 to 600 km , with a peak at 200 - 300 km above optical depth unity at 500 nm ( in agreement with * ? ? ?
the core of the 630.2 nm line instead forms in a slightly narrower range of photopsheric heights , from 180 to 600 km , and peaks slightly higher in the atmosphere at approximately 300 km .
figure [ response_t ] also shows that the core of the two lines forms in the deepest layers of the photosphere in magnetic features . this is the result of the lower temperature of fal - p with respect to fal - c at heights below optical depth unity in the latter ( fontenla et al .
1999 ) , which is also found in computations of magnetic flux tube models ( e.g. * ? ? ?
* ; * ? ? ?
* ) and full 3-d magneto hydro dynamic simulations ( e.g. * ? ? ?
* ) . both computed @xmath11 peak at 100 km , with contribution from higher photospheric layers for the 630.2 nm line and from layers below optical depth unity for the 617.3 nm line .
the bumps of both t_rf at approximately 200 km reflect the change of the temperature gradient with height in fal - p model .
figure [ response_b ] shows the b_rf of ic for both lines in the fal - p model .
the two b_rf are quite similar , as both peak at approximately 190 km , with the b_rf of the 617.3 nm line having again a slightly larger contribution from the deepest layers than the one calculated for the 630.2 nm line .
the results derived from the synthesis indicate therefore that both lines form in similar regions of the solar atmosphere , with the 630.2 nm line sampling slightly higher atmospheric heights than the 617.3 nm line .
in this section we investigate the dependence of the three line shape parameters on magnetic flux density . note that we found that the flux density values estimated from the 630.2 nm data are on average lower than those obtained from the 617.3 nm set , and that the discrepancy increases with the magnetic flux density , with a maximum of approximately 40% at 1000 gauss . because , as shown in 3 , the two lines form in similar regions of the solar atmosphere , this discrepancy is unlikely to reflect a stratification of the magnetic field with height . to verify this hypothesis we applied the cog to synthetic spectra of the two lines computed in the fal - p model with a stratified magnetic field and obtained magnetic field estimates in agreement within 4% .
we conclude that the discrepancy that we found in our data is most likely due to the telluric blending affecting the 630.2 nm line . in the following ,
we therefore refer to flux density estimates derived from the 617.3 nm data only .
results from our measurements are presented in fig .
[ shapes ] .
dot symbols show the dependence of ic , eqw , and fwhm on the magnetic flux density for both lines , while asterisk symbols indicate the average value of the same line parameters computed over bins of magnetic flux .
note that the bins were increasingly larger in order to guarantee sufficient statistics .
we found that the ic and fwhm increase with the magnetic flux for both lines , whereas the eqw shows the opposite trend .
all three line shape parameters show saturation . for both ic and eqw
this saturation occurs at approximately 300 gauss for both lines , while the fwhm saturates at flux density values above 700 gauss and 500 gauss for the 617.3 nm and the 630.2 nm line , respectively .
note that the maximum relative variations with magnetic flux density ( i.e. relative difference between values obtained at the largest flux and those obtained for quiet regions ) are approximately 40% for ic for both lines , 55% and 36% for fwhm in the case of 617.3 nm and 630.2 nm respectively , and -15% and -8% for eqw , respectively .
the measurements present large scatter , especially at the lowest flux density values , even though the standard deviations of the values distributions in each bin ( error bars in the plots ) are rather small .
the observed scatter of values reflects the sensitivity of line shape parameters to temperature and velocity variations of the plasma .
the larger the magnetic flux , the more the temperature and velocity fluctuations of convective plasma motions are inhibited ( e.g. * ? ? ?
* ; * ? ? ?
* ) , so that the dispersion of the line shape parameters becomes smaller and smaller with the increase of the magnetic flux density . note that the dispersion for the 617.3 nm line is larger than that obtained for the 630.2 nm data , especially when comparing ic .
this is due to the lower excitation potential of the 617.3 nm line , which makes the core intensity of this line more sensitive to temperature and velocity fluctuations @xcite .
the ic in particular depends on both temperature and magnetic field in a manner that is difficult to disentangle .
results from numerical models and simulations of magnetic flux tubes show in fact that the photospheric temperature within a tube rises following an increase of magnetic field , due to the intensification of the lateral heating effects from the surrounding plasma ( e.g. * ? ? ?
as also qualitatively shown by results obtained from numerical synthesis reported in sec .
[ appenidx : numsin ] * , an increase in either of these quantities , the field strength , via the zeeman effect , or the temperature , via irradiation , leads to an increase of ic , so that it is difficult to discern between the two effects .
the trend with the magnetic flux density of fwhm found for both lines suggests instead that this line shape parameter is more sensitive to magnetic field variations and velocity effects than to temperature .
in fact , from the rise of temperature within flux tubes mentioned above , we would expect the fwhm to decrease with the magnetic flux .
this decrease is compensated on one hand by the fact that magnetic flux concentrations tend to be located in or nearby downflow regions , which are characterized by shallower and broader profiles , and on the other hand , to the increase of the splitting of the lines with the magnetic flux .
this latter effect becomes more and more important with the increase of the field ( and therefore of the flux ) , due to the inhibition of convective flows mentioned above .
this interpretation is confirmed by results obtained from numerical synthesis of line profiles discussed in sec .
[ appenidx : numsin ] .
we also note that at low magnetic flux density values , the fwhm of 617.3 nm line increases slower with the magnetic flux density than for the 630.2 nm .
this indicates that for the former line the effect of temperature counteracting zeeman broadening is larger than for the 630.2 nm .
note that the maximum difference between the fwhm at saturation and the fwhm of average quiet sun agree within the errors for the two lines ( i.e. is approximately 0.06 ) , as expected from zeeman splitting theory for lines with equal land factor . finally , the dependence of eqw on the magnetic flux reflects variations of both the width and depth of the studied lines .
the rapid saturation of eqw with the magnetic flux is due to the fact that the line core intensity increases more rapidly than the fwhm for a given increase of magnetic flux . * as formally shown in the appendix [ seca1 ] , and confirmed by results obtained from numerical synthesis shown in sec .
[ appenidx : numsin ] , * the eqw of the two iron lines is almost independent of the magnetic field and is mostly sensitive to temperature variations .
figure [ shapes ] therefore suggests that the magnetic features in our data are characterized by similar temperatures .
in previous section we have analysed the dependence of the three shape parameters of the two iron lines on the magnetic flux density . as already noticed , this dependence is not explicit , because of the indirect effects of the magnetic field on other physical parameters . in particular ,
the indirect effect on the temperature stratification highly affects the shape of a line . in order to separately investigate the effects of magnetic field and temperature on line shape parameters we have therefore analysed results obtained from numerical synthesis through various one dimensional models , as described in sec . [
data : synthesis ] .
results are presented in this section . to investigate the effects of temperature on the shapes of line profiles
we have compared results obtained from the various fal models , which are characterized by different temperature stratifications .
plots in fig .
[ eqwvstsim ] show the relative differences of the line parameters computed for fal - e , fal - h and fal - p models with respect to line parameter obtained with fal - c model .
since the three models are characterized by increasing temperature at the formation heights of the two lines , the plots show that the increase of temperature causes ic to increase and the fwhm and eqw to decrease .
while for ic and eqw we found changes up to @xmath12 and more , the fwhm changes do not exceed @xmath13 . to investigate the effects of the magnetic field , we considered line profiles obtained with fal - c model in the case of fields constant with geometric height and whose strength varied from 50 to 1000 gauss .
results reported in fig .
[ eqwvsbsim ] show the relative difference between line parameters obtained for the different field strength values with respect to those obtained in the case of no magnetic field .
we note that the increase of the magnetic field strength leads to a large increase of ic and fwhm ( more than @xmath14 in the case of ic ) , especially at the largest magnetic strength values , while eqw stays fairly constant ( variations are smaller than @xmath15 for both lines ) , thus confirming the analytical derivation presented in appendix [ seca1 ] .
both plots show that the two lines are differently affected by changes of plasma physical properties , mainly because of their different atomic parameters .
the plot in fig .
[ response_t ] shows for instance that the difference of the shapes of the t_rfs obtained for fal - c and fal - p models is larger for the 617.3 nm line than for the 630.2 nm line .
in particular , while the t_rfs of the 630.2 nm line obtained with the two models appear mostly shifted in depth respect to each other , the two t_rfs obtained for the 617.3 nm line also show different shapes .
the 617.3 nm line therefore experiences larger changes due to temperature effects than the 630.2 nm . a direct comparison of plots shown in fig .
[ eqwvstsim ] and in fig .
[ eqwvsbsim ] is not straightforward , as they show the variations of the line parameters with respect to two different physical quantities , the temperature and the magnetic field strength , respectively .
nevertheless , results from magnetic flux tube models show that within magnetic flux tubes in radiative equilibrium the temperature at the formation heigths of the two investigated lines is positively correlated with the field strength , so that , for instance , model fal - p can reasonably have associated a field strength of 1000 - 1500 gauss .
this allows a comparison of the trends of the line parameters with respect to the two physical quantities .
the plots show that ic increases with the increase of both temperature and field strenght , in a manner difficult to disentangle .
opposite trends are instead found in the case of the fwhm , as an increase of temperature makes this line parameter to decrease while an increase of magnetic field strength makes it to increase .
comparison of the value of fwhm variation obtained for fal - p with the value obtained for 1000 gauss show that magnetic field effects are more likely to dominate .
variations of the eqw found in the case of the increase of the temperature are more than one order of magnitude larger than those obtained when increasing the magnetic field strength , again confirming that the direct effect of zeeman splitting on this line parameter is negligible for the two lines .
finally , it is important to notice that variations found from the numerical synthesis presented here are meant to qualitatively interpret observational results presented in sec .
[ measure ] , as they are representative of the ideal case of unitary magnetic filling factor and observational effects like finite spatial and spectral resolution and scattered light have not been taken into account .
in this section we investigate whether and to what extent , line shape parameters are affected by small amounts of magnetic flux not resolved in the observations . to this purpose we analysed average line shape parameters
obtained in images that were synthetized by replacing line profiles of pixels whose magnetic flux density exceeds a certain threshold with the average quiet - sun profile . by varying the magnetic flux threshold , images of successively lower average magnetic flux density ( in particular from approximately 90 to * 10 * gauss ) were constructed . note that the quiet sun profile was computed averaging profiles of those pixels of the original images whose measured magnetic flux density is lower than 30 gauss ( see sec .
the line shape parameters were then determined from the average profiles of each of these synthetic maps , thus mimicing observations that do not resolve the remaining magnetic features .
figure [ shaveg ] shows the relative differences between line shape properties from the synthetic frames ( characterized by different values of the average magnetic flux ) and those of the synthetic frame with the lowest magnetic flux density ( i.e. 10 gauss ) . in the case of ic and
fwhm the error bars are derived from the uncertainties of the fitting of the line to a gaussian function ( see sec .
2 ) , while for the eqw the errors are estimated by the @xcite formula .
we found that the average line shape parameters vary by a few percents with the increase of the average flux density .
the maximum variation that we found are approximately @xmath16 in the case of ic for both lines , 4@xmath6 and 2.5@xmath6 in the case of fwhm for the 630.2 nm and the 617.3 nm , respectively , and -1.8@xmath6 and -3.5@xmath6 in the case of eqw .
variations are larger in 617.3 nm in the case of eqw , as a result of the already mentioned larger sensitivity of this line to temperature . for the same reason , larger variations of fwhm with the magnetic flux density
are instead found for the 630.2 nm . as explained above , in fact , for the 617.3 nm line the effect of temperature counteracting the zeeman line broadening
is larger , thus reducing the increase of the fwhm of this line with the flux .
the measured line shape parameters of the two lines vary with magnetic flux because of the direct effect of the zeeman splitting and the indirect effect of temperature variations induced by suppression of convective motions .
ic and fwhm increase with the magnetic flux , while the opposite trend was found for eqw ( fig .
[ shapes ] ) .
results from numerical synthesis show that in the case of ic , the trends reflect both the zeeman splitting and the temperature effects in a manner that is difficult to disentangle , whereas the fwhm is mostly sensitive to the zeeman effect , in particular for larger field strenght .
we have also shown that the dependence on magnetic field of the eqw is negligible at the values of magnetic flux investigated , so that the observed trend must be ascribed to temperature variations induced by the presence of magnetic field concentrations .
we also found that the three line shape parameters show saturation with the magnetic flux density ( fig .
[ shapes ] ) .
interpretation of these results in terms of magnetic flux tube models suggests that features characterized by small magnetic flux values are mostly spatially unresolved in our observations and that the pixels characterized by the largest magnetic flux values ( above the saturation threshold ) belong to spatially resolved features that may result from the clustering of magnetic elements .
according to flux tube models , in fact , an increase of the magnetic field leads to an increase of the temperature within the tube in line - forming layers . instead of a saturation we should therefore observe ic and fwhm to monotonically increase and eqw to monotonically
decrease with the flux .
in particular , the saturation of eqw suggests that the magnetic features that we observed are mostly characterized by very similar photospheric temperatures .
this is expected from models of clusters of magnetic flux tubes ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and is in agreement with results recently reported by @xcite .
it is worth to notice that also @xcite concluded that the profiles of the 525.0 and 525.3 nm lines suggest the `` quantization '' of the magnetic flux in the photosphere .
we also investigated the effects of unresolved magnetic field on the shapes of line profiles .
results reported in fig .
[ shaveg ] show that variations of a few percent in the average shapes of the investigated lines are produced for modest ( below 100 gauss ) values of average magnetic flux .
* note that the average flux values investigated are compatible with those employed in irradiance reconstruction techniques to identify quiet sun regions ( * ? ? ?
* ; * ? ? ?
* and references therein ) .
* in the case of fwhm , larger variations with the average magnetic flux are found for the 630.2 nm than for the 617.3 nm , whereas in the case of eqw larger variations are found for the 617.3 nm .
the 630.2 nm line is therefore more favorable to investigate magnetic field variation , while the 617.3 nm is more suitable to investigate effects of temperature variations .
it is interesting to notice that our estimates of eqw variations are in agreement with values reported by @xcite for the 608.27 nm and 624.07 nm lines .
these authors analyzed snapshots from magneto hydrodynamic simulations to investigate the variations of average eqw values of some iron lines with the increase of the average magnetic flux in the simulations .
comparison of our fig .
[ shaveg ] with their fig .
4 reveals that the agreement is better for the 617.3 nm line , as this has an excitation potential similar to that of 608.27 nm and 624.07 nm lines .
their plots also indicate that the variations of eqw mostly reflect variations of the average temperature in their simulations .
the small dependence in their plots on the average magnetic flux is due to the wavelenght shift of the opacity discussed in the appendix . for completeness
, we notice that those authors also reported a large dependence of eqw of the 1.5 @xmath17 line on the magnetic flux .
this is due on one hand to the fact that this line forms in regions where the variation of temperature induced by the magnetic field is negligible , on the other to its large land factor so that this line is not in the weak field regime . the amount of variations of line shape parameters due to the presence of unresolved magnetic flux that we found is measurable with instruments like hinode / sot , solis / vsm and sdo / hmi .
line shape variations of the investigated lines observed at disk center can be employed for studies of long term temporal variations of physical properties of quiet sun . in particular
, the results presented here show that it would be possible to disentangle thermal and magnetic effects by comparing variations of different properties of the two photospheric lines .
for instance , a decrease over the magnetic cycle of eqw together with an increase of fwhm would indicate variations of quiet sun temperature induced by the presence of unresolved magnetic features . on the other hand , a decrease of the eqw together with a decrease of the fwhm
would indicate an increase of the effective temperature of the plasma .
finally , it is worth to notice that @xcite showed that the same profiles of fe i 630.1 and 630.2 nm lines can be produced by different thermodynamic stratifications , magnetic field and microturbulence values , so that results obtained from spectro - polarimetric inversions should be treated with caution .
however , the method that we propose is less affected by this degeneracy . in fact
, we have shown that the eqw parameter of the two iron lines is almost insensitive to magnetic field effects , while the fwhm is mostly sensitive to magnetic field effects .
the eqw variations can therefore be utilized to derive information on temperature variations , while the fwhm can be utilized to derive information on magnetic flux variations .
note , however , that a proper modeling of the observed data ( e.g. * ? ? ?
* ) is required to quantitatively estimate these variations . in particular , modifications of the velocity distributions , which can result from either variations of the magnetic flux or of the effective temperature , can also affect the shapes of lines .
nevertheless , since velocity distributions are known to narrow in the case of increase of magnetic flux ( e.g. * ? ?
* ; * ? ? ?
* ) , while they broaden and skew toward upflow velocities in the case of an increase of the effective temperature ( e.g. * ? ? ?
* ; * ? ? ?
* ) , we expect velocity effects not to hamper the possbility of disentangling magnetic flux and effective temperature variations from observations of properties of the two lines .
we plan to investigate these effects in the future by analysing results from numerical simulations of the solar photosphere .
this study was supported by the istituto nazionale di astrofisica ( prin - inaf-2010 ) and the agenzia spaziale italiana ( asi / ess / i/915/01510710 ) .
the nso is operated by the association of universities for research in astronomy , inc .
( aura ) , for the national science foundation .
ibis was built by inaf / osservatorio astrofisico di arcetri with contributions from the universities of firenze and roma tor vergata , the national solar observatory , and the italian ministries of research ( mur ) and foreign affairs ( mae ) .
we are grateful to a. tritschler for providing the ibis data reduction code .
here we investigate the dependence of the eqw of a magnetically sensitive spectral line on the magnetic field .
it is common practice to express the emergent line profile @xmath18 in the presence of zeeman splitting as the sum of three gaussians ( e.g. * ? ? ?
* ) : @xmath19 the two gaussian functions shifted by the amount @xmath20 represent the left- and right - hand circularly polarized @xmath21 profiles , while the gaussian function centered at @xmath22 represents the profile of the linearly polarized @xmath23 component . @xmath24 and @xmath25 are the core intensity and doppler width of the line , in the absence of magnetic field ; @xmath26 is the inclination angle with respect to the line of sight ; @xmath27 , where c is a constant , @xmath28 is the effective land factor , @xmath22 is the central wavelength of the line , and b is the magnetic field strength . in the case of magnetic field
oriented along the line of sight ( @xmath29 ) the previous formula simplifies to : @xmath30 the eqw of the line in that case is , by definition : @xmath31 \label{eq3a}\ ] ] the two terms in the sum are equal as they differ only for by a shift within the infinite domain . since the two terms represent the eqws of the left ( @xmath32 ) and right ( @xmath33 ) circular polarized signal , then @xmath34 ( note that it is easy to verify that these equalities hold for any shape of the profiles , as far as the two circular polarized signals are symmetric ) .
we want now to study the variation of the eqw with the magnetic field .
to this aim we differentiate eq.[eq3a ] with respect to @xmath35 : @xmath36 \label{eq5a}\ ] ] making the variable change @xmath37 , and changing the order of integration and differentiation , we obtain : @xmath38 \label{eq6a}\ ] ] where we have assumed that the rest wavelength is much larger then the zeeman splitting .
it is easy to see that the right - hand side of eq .
[ eq6a ] is null , as the terms in the integrand that are proportional to @xmath39 are uneven , and the terms that are proportional to @xmath40 cancel .
thus , under the assumption that @xmath24 and @xmath41 do not depend on the field , the eqw of a zeeman split line is independent of the field strength .
howevere , this assumption does not hold in a stratified atmosphere , because the zeeman splitting changes the opacity at a given wavelength , and therefore alters the height and physical parameters the line samples . indeed , as shown by previous studies ( e.g. * ? ? ?
* ) , the eqw of a magnetic sensitive line is sensitive to the strength of the magnetic field .
these variations are small in the weak field case ( i.e. @xmath42 ) , and are non - negligible only for large fields .
moreover , the sensitivity of the eqw to the field depends on the atomic parameters of the line , as these determine its opacity .
|
the contribution of the quiet sun to solar irradiance variability , due either to changes of the solar effective temperature or to the presence of unresolved magnetic field , is still poorly understood . in this study
we investigate spectral line diagnostics that are sensitive to both temperature variations and the presence of small scale unresolved magnetic features in these areas of the solar atmosphere .
specifically we study the dependence on the magnetic flux density of three parameters describing the shape of two magnetically sensitive lines , at 630.2 nm and 617.3 nm , namely the line core intensity ( ic ) , full width at half maximum ( fwhm ) , and the equivalent width ( eqw ) . to this aim
we analyze observations of active region noaa 11172 , acquired with ibis at the dunn solar telescope , as well as results from numerical synthesis .
our results show that ic is sensitive to both temperature and magnetic flux density variations , fwhm is mostly affected by magnetic field changes , and eqw is mostly sensitive to temperature .
variations of a few percent of the measured line parameters are found in observational data that was spatially degraded to represent quiet - sun , disk - centre , medium resolution observations .
it is therefore possible to disentangle magnetic from pure thermodynamic effects by comparison of temporal variations of the eqw and the fwhm of either the two lines .
|
the expansion of the universe inevitably leads to a drop of the temperature of the primordial fireball .
this cooling provides natural conditions that are believed to precipitate a series of phase transitions in which `` false '' symmetric , high temperature phases are transformed into low temperature broken symmetry `` true '' vacuum . as the universe undergoes such phase transitions , the selection of the low - temperature , broken symmetry phase can only occur locally , within the causally correlated regions .
kibble @xcite first noted that this symmetry breaking process may leave relics of the high energy phase which will be trapped by the topologically stable configurations of the broken symmetry phase . in principle , such topological defects could be massive enough to leave an observational imprint in the cosmic microwave background and perhaps also produce the seeds needed for matter structure formation @xcite .
there are three principal types of topological defects @xcite .
they differ by their dimensionality .
monopoles are point - like objects that would be disastrous from the cosmological point of view .
they have a tendency to dominate the matter content of the universe .
inflation was invented in part to dilute them and thus to prevent observational inconsistencies .
membranes or domain walls are two - dimensional , and they almost certainly could also lead to the same disaster as monopoles . specifically , they would induce unacceptably large distortions of the cosmic microwave background .
by contrast , one - dimensional topological defects , cosmic strings , could provide the source of density perturbations needed to seed structure formation @xcite .
symmetry breaking phase transitions which occur in condensed matter physics are described by theories which are formally analogous to those involved in the cosmological context ( for a review see @xcite ) .
condensed matter phase transitions have , however , one crucial advantage : they can be studied in the laboratory . with this in mind , more than a decade ago one of us suggested @xcite that the cosmological mechanism for defect formation can be studied experimentally in the condensed matter context .
a number of experiments have been carried out following that proposal .
for instance , liquid crystal experiments demonstrated that copious production of topological defects does indeed happen in symmetry breaking phase transitions @xcite .
more recently , superfluid he@xmath0 and he@xmath1 experiments have also tested the topological defect formation picture @xcite .
liquid crystal phase transitions , however , are of the first order type . in first order phase transitions ,
the process which is responsible for the appearance of the new phase is nucleation .
namely , small regions of the medium undergo thermal activation which takes them over the potential barrier separating `` false '' and `` true '' vacua . as a result , bubbles of a certain ( critical ) size appear and form seeds of the new phase .
eventually , through growth and coalescence of these bubbles , the new phase replaces the old one .
a much more interesting and less understood dynamics takes place in second order ( landau - ginzburg like ) phase transitions . for this reason , he@xmath0 and he@xmath1 systems ,
for which the transition to a superfluid phase proceeds without nucleation , have recently attracted significant interest as examples of cosmological phase transition analogues . the defect formation scenario is based on two assumptions .
the first assumption is of the qualitative nature : it asserts that regions of the broken symmetry phase , which are causally disconnected , must select the new low temperature phase independently . as a result , when a symmetry - breaking phase transition with a non - trivial vacuum manifold occurs simultaneously in a sufficiently large volume , topological defects will be formed .
the second assumption is of the quantitative nature : it involves specifying the process responsible for the causal propagation of `` signals '' that allow the choice of the new vacuum to occur in a coordinated ( rather than independent ) fashion .
this is the assumption that yields a prediction of the density of topological defects .
both assumptions are of course necessary @xcite , but while the first one is straightforward , the second one requires much more specific physical input .
the original discussion of the scenario for defects formation was based on an idea similar to that of nucleation in first order phase transitions @xcite .
it was thought that the initial density of the topological defects is set at the so - called ginzburg temperature @xmath2 , namely the temperature below the critical temperature @xmath3 at which thermally activated transitions between the correlation - sized volumes of the new broken symmetry phase cease to occur . below @xmath2 , the ( free ) energy barrier , separating `` false '' and `` true '' vacua , becomes prohibitively large for correlation - length sized thermal fluctuations . under this approach
then , the density of defects is set by the correlation length at @xmath2 .
one of the key predictions of the original papers @xcite on `` cosmological '' phase transitions in superfluids was that this thermally activated process decide the density of defects .
instead it was proposed that the characteristic correlations length is set by the dynamics of the order parameter in the vicinity of the critical temperature @xmath3 .
this `` non - equilibrium '' scenario @xcite represents a significant change of the point of view , and yields a prediction for the density of defects which is rather different from the `` equilibrium '' estimate based on the ginzburg temperature .
in the vicinity of the critical temperature @xmath3 of a second - order phase transition , the contribution from the potential @xmath4 to the free energy of the system is given by @xmath5 with @xmath6 and @xmath7 positive parameters and @xmath8 the dimensionless temperature parameter .
the last term in ( [ eq:1 ] ) has been added , so @xmath9 at the minimum of the potential for @xmath10 . here
@xmath11 represents the , an abstract measure of the degree to which the symmetry in question has been broken . above the critical temperature
, the potential ( [ eq:1 ] ) has a single minimum at @xmath12 .
as the temperature drops below @xmath3 , the shape of the potential changes ; @xmath13 becomes non - negative . instead of a single minimum at @xmath12 ,
the potential has a minimum value @xmath9 at @xmath14 . at the same time , the potential @xmath4 develops a barrier centered at @xmath12 with magnitude @xmath15 separating " the minimum values of the potential .
depending on the symmetry that is broken , the manifold of equivalent vacuum states for the order parameter will have different topologies , and thus describe a different set of topological defects @xcite .
for instance , in the case of domain walls , the order parameter is a real scalar field with discrete @xmath16 symmetry .
the vacuum manifold in this case consists of the two values @xmath17 .
simple models of cosmic strings are , on the other hand , represented by a a complex scalar field with @xmath18 symmetry of phase transformations , @xmath19 .
the degenerate vacuum states are given by @xmath20 , with @xmath21 an arbitrary phase .
thus , the vacuum manifold for cosmic strings is the circle @xmath14 in the complex @xmath11-plane . as mentioned before ,
since all the points in the vacuum manifold are completely equivalent , the choice of the expectation value of the order parameter is determined from the initial thermal ( random ) fluctuations through its dynamics .
a measure of the extent , in space and time , to which perturbations of the order parameter return to their equilibrium value is given by the correlation length @xmath22 and dynamical relaxation time @xmath23 .
specifically , the correlation function of perturbations away from equilibrium behaves as @xmath24 and @xmath25 furthermore , in the vicinity of the critical temperature @xmath3 , the equilibrium values of @xmath22 and @xmath23 are given by @xmath26 and @xmath27 respectively , with @xmath28 and @xmath29 .
that is , at the critical temperature ( @xmath30 ) , the order parameter has a universal behavior characterized by the simultaneous divergence of both @xmath22 and @xmath23 .
before presenting the results of the density of topological defects from our numerical study , let us review the original estimate of the density of defects put forward in @xcite .
this estimate was based on a linear quench , @xmath31 with @xmath32 the time before ( @xmath33 ) and after ( @xmath34 ) the transition .
( the idea behind the argument does not rely on this particular time dependence of the temperature parameter @xmath13 , but it seems to be a likely approximation when narrow range of temperatures near critical @xmath3 is involved , as it will be the case here . ) in ( [ eq : quench ] ) , @xmath35 represents the quench timescale .
the other relevant timescale which we shall call symmetry breaking timescale is defined by @xmath36 .
it is then possible to identify during a quench three regimes by comparing the magnitude of the symmetry breaking timescale @xmath37 with the dynamical relaxation time @xmath23 . for sufficiently large values of the temperature above @xmath3 ( see fig . [ figure1 ] ) , the symmetry breaking timescale @xmath37 is larger than the dynamical relaxation time @xmath23 .
the system is in a stage where the order parameter is able to adjust to the changes implied by the quench in an essentially adiabatic fashion . in particular
, the non - equilibrium correlation length closely matches the equilibrium behavior given by eq .
( [ eq : xi ] ) . because of the critical slowing down implied by eq .
( [ eq : tau ] ) , at some time @xmath38 before the transition , the systems reaches the point at which the symmetry breaking timescale @xmath37 and the dynamical relaxation time @xmath23 are comparable ; namely @xmath39 the value of the temperature parameter at this point is @xmath40 the corresponding value of the correlation length can be directly found by substituting eq .
( [ eq : ehat ] ) into eq .
( [ eq : xi ] ) , @xmath41 within the @xmath42 $ ] interval , the system is incapable of keeping up with the changes induced by the quench .
in particular , the non - equilibrium correlation length can not longer grow at the equilibrium rate implied by eq .
( [ eq : xi ] ) . crudely speaking , the non - equilibrium correlation length
freezes out " at the value @xmath43 in eq .
( [ eq : xihat ] ) until a time @xmath44 , after the transition , when once again the dynamical relaxation time @xmath23 becomes smaller that the symmetry breaking timescale @xmath37 . that is , approximately , @xmath45 the actual behavior of @xmath46 is probably more complicated ( i.e. correlation will continue to grow even within the @xmath42 $ ] interval ) , but @xmath47 does capture the size of the correlated region of vacuum at the time when its growth falls out of step with the equilibrium scaling .
the bottom line is that during the interval [ @xmath48 , the order parameter , and therefore the correlation length , is to sluggish to keep up with eq .
( [ eq : xi ] ) . the proposal in @xcite states that the frozen out correlation length @xmath47 yields the density of topological defects .
that is , @xmath47 characterizes the size of independently selected domains of the new vacuum , so it determines the separation ( and therefore density ) of topological defects . as mentioned before , the difference with the original kibble mechanism @xcite is the role that non - equilibrium dynamics play in determining the value of @xmath47 . in that picture , @xmath47
was obtained entirely from the temperature @xmath2 ( ginzburg temperature ) at which fluctuations between the degenerate minima of the potential ceased to be possible ; that is , the temperature at which @xmath49 with @xmath50 the potential barrier " in eq .
( [ eq : dv ] ) separating degenerate minima .
the reason why @xmath2 does not appear to play a decisive role in the creation of topological defects is likely to be associated with the spatial extent of the thermally activated transitions : local thermal fluctuation can perhaps create small defects , but these topological defects will have a scale approximately equal to the size of their core ( since both are defined by the same correlation length @xmath22 ) .
such ill - defined defects are unlikely to survive after all , they represent a configurations which may be a shallow local minimum of the free energy , but which have a higher free energy than the uniform superfluid .
hence , they are not large enough to be really topologically stable change of the field configuration in a finite region of space the order of @xmath22 suffices to return to the uniform `` true vacuum '' .
moreover , such local defects can not easily lead to the creation ( or destruction ) of the large ones .
these requirements are least convincing in the case of monopoles ( which have spatial extent of order @xmath22 in every direction ) .
however even monopoles would have to be created `` in pairs '' by thermal fluctuations .
these pairs of monopoles of opposite charge will be separated by distance of order @xmath22 .
therefore , they are not likely to separate and survive . we are led to the conclusion that the dominant process in creation of topological defects will have to do with the critical slowing down and the subsequent `` freeze - out '' of the fluctuations of the order parameter at the time @xmath51 rather than with the thermal activation and ginzburg temperature .
our numerical experiment @xcite to test the mechanism described in the previous section for estimating the density of topological defects simulates the dynamics of a real field @xmath11 in a 1d system .
this system contains all the essential ingredients needed to investigate topological defect formation scenarios ( including symmetry breaking , although without an infinite range order ) , and at the same time , because of its one - dimensional nature , it allows us to perform a reasonable and accurate numerical analysis .
we assume that the system under consideration is in contact with a thermal reservoir and obeys the langevin equation @xmath52 the potential @xmath4 is given by ( [ eq:1 ] ) .
( a rescaling of @xmath53 and @xmath32 can be introduced to eliminate the parameters @xmath6 and @xmath7 . )
the noise term @xmath54 has correlation properties @xmath55 in eq .
( [ eq : noise ] ) , @xmath56 is the overall damping constant characterizing the amplitude of the noise through eq .
( [ eq : motion ] ) .
@xmath21 fixes the temperature of the reservoir .
typical values used of @xmath21 were in the range ( @xmath57 ) . since the symmetry in the system is discrete ( @xmath16 ) , our numerical experiment consisted of investigating the creation of kinks ( domain walls in 1d ) as a function of the quench timescale @xmath35 .
figure [ figure2 ] shows a sequence of `` snapshots '' of @xmath11 for a typical quench .
one can clearly see from this figure the transition from expectation value @xmath58 above the critical temperature ( @xmath59 ) to @xmath60 after the symmetry has been broken ( @xmath61 )
. we test the predictions of defect density in @xcite in the overdamped regime ( @xmath62 ) , that is , the regime for which the damping term @xmath63 in eq .
( [ eq : motion ] ) is bound to dominate over @xmath64 . for comparison
, we also show in figure [ figure2a ] a sequence of `` snapshots '' of @xmath11 for an underdamped case with @xmath65 .
overdamped evolution will take place whenever @xmath66 ( see e.g. @xcite ) .
the characteristic relaxation time is then given by @xmath67 and therefore , from eq .
( [ eq : tau ] ) , we have that @xmath68 and @xmath69 . furthermore , from eq .
( [ eq : ehat ] ) , one obtains that the freeze - out relaxation time in this case is @xmath70 one can then use ( [ eq : cond1 ] ) to estimate from ( [ eq : ehat ] ) and ( [ eq : xihat ] ) both , the freeze - out temperature parameter @xmath71 and correlation length @xmath72
. their values are @xmath73 and @xmath74 respectively , where we have adopted @xmath75 in accord with the landau - ginzburg theory .
the value of @xmath75 is also in agreement with the calculated critical exponents of the correlation length inferred from our equilibrium simulations ( see fig .
[ figure3 ] ) .
we obtained for the equilibrium correlation length a best fit yielding @xmath76 , @xmath77 above @xmath3 , and @xmath78 , @xmath79 below @xmath3 , close to the landau - ginzburg exponent of @xmath75 .
figure [ figure4 ] encapsulates the main result of our paper .
it shows the number of zeros ( kinks ) obtained in a sequence of quenches with various values of the quench timescale @xmath35 . for each value of @xmath35 ,
15 quenches were performed in a computational domain of size 2,048 units with periodic boundary conditions .
this size of computational domain is large enough to avoid boundary effects .
the number of kinks produced after the quench was obtained by counting the number of zeros of @xmath11 . above and immediately below @xmath3
, there is a significant number of zeros which have little to do with the kinks ( see fig . [ figure2 ] ) .
however , as the quench proceeds , the number of zeros quickly evolves towards an `` asymptotic '' value ( see fig .
[ figure5 ] ) .
this change of the density of zeros and its eventual stabilization " is associated with the obvious change of the character of @xmath11 and with the appearance of the clearly defined kinks . by
then , the number of kinks is nearly constant in the runs with long @xmath35 , and , correspondingly , their kink density is low . even in the runs with the smallest @xmath35 , there is still a clear break between the post - quench rates of the disappearance of zeros and the long - time , relatively small rate at which the kinks annihilate .
( kinks will eventually annihilate , but on slow timescales set by the thermal diffusion of kinks over distances of the order of their separation . )
our simulation clearly indicates @xcite a scaling relation for the number density of kinks .
we find a kink density ( no . of zeros/2,048 ) of @xmath80 , when the kinks are counted at approximately the same @xmath81 value .
by contrast , when the counting is done at the same value of absolute time @xmath32 , @xmath82 has a somewhat shallower slope ( @xmath83 ) , reflecting higher annihilation rate at high kink densities . within the estimated errors ,
this result is in agreement with the theoretical prediction of an scaling of @xmath84 .
an important point to consider is that there are of course no true " phase transitions in 1d , and that instead of vortices or strings we are counting kinks .
the remarkable agreement of our numerical results @xcite with the theoretical predictions suggests that the crucial ingredient in the scenario of ref .
@xcite is not that much the existence of a mathematically rigorous " phase transition .
the essential conditions required for this scenario are : the presence of symmetry breaking and the scalings ( [ eq : xi ] ) and ( [ eq : tau ] ) for the correlation length and relaxation time .
the central issue of the work presented in this paper was a numerical test of the mechanism for defect formation in the course of second order phase transitions involving a non - conserved order parameter .
both laboratory @xcite and computer experiments @xcite have added strong support to the proposal that the initial density of the topological defects will be set by the correlation length at the freeze - out instant @xmath38 ; that is , at the moment when the relaxation timescale of the order parameter will be comparable to the characteristic timescale of the quench . at that instant , as a consequence of critical slowing down
, perturbations of the order parameter become so sluggish that they cease to keep up with the equilibrium scalings , so that in the time interval [ @xmath48 the order parameter can not adjust to the changes of thermodynamic parameters induced by the quench . by the time @xmath85 below @xmath3
, the dynamics will `` restart '' ; however , it will be too late for the system to get rid of the defects , by then the defects will be `` set in concrete , '' that is , stabilized by the topology ( although subsequent evolution involving their interactions will ensue , and , generally alter their density ) .
strong evidence that thermal activation is not responsible for the formation of topological defects is provided by the experimental results obtained in the superfluid he@xmath0 experiment by the lancaster group @xcite and by the he@xmath1 experiments in helsinki and grenoble @xcite .
together , they demonstrate that the argument in ref .
@xcite is unaffected by the thermal activation process invoked in kibble s original discussion ( @xcite , see also @xcite ) .
ginzburg temperature that is , the temperature at which vacuum can be thermally `` flipped '' over the potential barrier in regions of size @xmath22 is far below the critical temperature in the superfluid helium .
our numerical experiments also provide a strong confirmation of the theoretical predictions of the non - equilibrium scenario @xcite .
the scaling of the kink density with @xmath35 follows closely theoretical expectations .
these results are also supported by the dependence of the number of kinks on the damping parameter @xmath56 ( compare fig .
[ figure2 ] and fig .
[ figure2a ] ) , as well as by the preliminary results of the computer experiments involving complex order parameter and/or more than 1d space that we are currently investigating .
ruutu , v.b .
eltsov , a.j .
gill , t.w.b .
kibble , m. krusius , yu.g .
makhlin , b. placais , g.e .
volovik , and w. xu . vortex formation in neutron - irradiated superfluid he@xmath1 as an analogue of cosmological defect formation . , 382:334336 , 1996 .
|
we present results of numerical studies of the landau - ginzburg dynamics of the order parameter in one - dimensional models inspired by the condensed matter analogues of cosmological phase transitions .
the main goal of our work is to show that , as proposed by one of us @xcite , the density of the frozen - out topological defects is set by the competition between the quench rate the rate at which the phase transition is taking place and the relaxation rate of the order parameter .
in other words , the characteristic domain size , which determines the typical separation of topological defects in the new broken symmetry phase , is of the order of the correlation length at the instant at which the relaxation timescale of the order parameter equals the time remaining to the phase transition
. in estimating the size of topological domains , this scenario shares with the original kibble mechanism the idea that topological defects will form along the boundaries of independently selected regions of the new broken symmetry vacuum .
however , it derives the size of such domains from non - equilibrium aspects of the transition ( quench rate ) , as opposed to kibble s original proposal in which their size was estimated from the ginzburg temperature above which thermally activated symmetry restoration can occur .
|
cool cores are regions at the centre of rich clusters where the hot thermal x - ray emitting gas ( @xmath4 @xmath2 10@xmath5 @xmath3 ) is dense enough to cool radiatively within a hubble time ( see * ? ? ?
* ; * ? ? ?
* for reviews ) .
cooling rates of the order of 100 @xmath6 @xmath7 and up to 1000 @xmath6 @xmath7 have been estimated for this hot x - ray gas ( e.g. , * ? ? ?
. however , recent _ chandra _ and _ xmm - newton _ x - ray spectra show that little or no x - ray emitting gas ( @xmath810% ) cools below one third of the virial temperature ( e.g. , * ? ? ?
* ; * ? ? ?
the solution most often invoked in the literature is that some form of reheating balances the radiative cooling of the x - ray gas .
substantial cooler gas and dust components exist in the cores of these galaxy clusters @xcite . extended @xmath9
@xmath3 emission - line nebulae are found surrounding brightest cluster galaxies ( bcg ) ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* herafter j05 ) .
these nebulae are observed to extend at least up to 50 @xmath0 from the bcg nuclei ( j05 ) .
this component at @xmath4 @xmath2 @xmath9 @xmath3 emits far more energy than can be explained by the simple cooling of the intracluster gas through this temperature i.e. , additional heating is needed @xcite .
more recently , work in the infrared and mm - wavelengths has shown that there are large quantities of molecular gas at the centre of these clusters with temperatures between 15 and 2500 @xmath3 extending at least up to 20 @xmath0 from the bcg nuclei ( e.g. , j05 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the molecular gas has a cooling time of order years @xcite . without some form of heating
one would expect this gas to collapse rapidly and form stars .
although there is strong evidence that starformation does take place at the centres of cool core galaxy clusters it is not yet observed to match the extended gas nebulae ( * ? ? ?
* ; * ? ? ?
* oonk et al . in prep . ) .
the heating and cooling of the molecular and ionised gas phases are important elements in the energetics of the cool core region .
an energy source comparable to that needed to stop the hot x - ray gas from cooling is necessary to heat these colder phases ( j05 ) .
the primary source of ionisation and heating of the h@xmath10 and hii must be local to the gas ( j05 ; * ? ? ?
* ) , consistent with a stellar photoionising source .
however , stars are unable to explain the high temperature of the ionised gas ( * ? ? ?
* hereafter vd97 ) .
the molecular hydrogen lines are much stronger relative to the ionised hydrogen lines than in other types of extragalactic sources , such as agn or starburst galaxies ( e.g. , j05 ; * ? ? ?
* ; * ? ? ?
the ratio of h@xmath10 to hii emission lines ( j05 ; * ? ? ?
* ) , as well as detailed analysis of the mid - infrared and optical line ratios ( vd97 ; * ? ? ?
* ) indicate that to explain both heating and ionisation balance , photons harder than those available from o - stars are needed . however , often very high ionisation lines are missing ( e.g. , [ ne v ] typical of agn spectra . if present , the source of these photons is elusive .
alternatively shock heating has been considered , however the characteristic [ o iii ] 4363 angstrom line is missing ( vd97 ) .
high energy particles have been invoked to explain this problem @xcite .
it is of great importance to pinpoint the nature of the heating mechanism and include it in models of cooling flows in galaxy clusters as well as models of galaxy growth and evolution .
cool core clusters are the low redshift cluster scale analogs of high redshift galaxy scale cooling flows . to understand the formation of massive galaxies at high redshift and the feeding and feedback mechanisms in agns it is important to understand the heating of the cool gas in bcgs .
all gas phases observed in the intracluster medium require reheating to avoid catastrophic cooling . a variety of heat sources to counteract this cooling gas
have been proposed over the years : agn feedback ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , low velocity shock waves @xcite , conduction ( vd97 ) , hot stars ( vd97 ) and energetic particles @xcite .
none of these heat sources have so far been able to match the detailed characteristics of cool core galaxy clusters .
whatever the heat mechanism , the cooling region extends over hundreds of @xmath0 across the cluster core , and heating is unlikely to balance the cooling exactly over such a large region .
some residual cooling will occur and presumably corresponds to the emission line nebulae and star clusters surrounding brightest cluster galaxies ( bcg ) at the centres of cool core galaxy clusters ( j05 ; * ? ? ?
* ; * ? ? ?
+ + the bcgs at centres of cooling clusters fall within a region of the bpt diagram ( * ? ? ?
* ; * ? ? ?
* vd97 ) that is occupied by liners and agn . in our previous samples ( j05 ; * ? ? ?
* ; * ? ? ?
* ) we have focused on the liner - like bcgs and we continue to do so here . these clusters were originally selected based on their high cooling rates , strong h@xmath11 , h@xmath10 emission and low ionisation radiation .
liner - like bcgs were chosen because we wish to minimise the role that their agn have on the global radiation field . in the work presented here we focus on two liner - like bcgs from our previous samples , pgc 071390 in abell 2597 ( hereafter a2597 ) and eso 0291-g009 in sersic 159 - 03 ( hereafter s159 ) .
optical ( [ o iii]/h@xmath12 and [ o i]/h@xmath11 , ( * ? ? ?
* vd97 ; ) ) and mid - infrared spectra ( [ ne iii]/[ne ii ] and [ ne v]/[ne ii ] , jaffe & bremer in prep . ) of these bcgs indicate that their ionising spectrum is very soft i.e. they are extreme liners ( vd97 ; * ? ? ?
however , these bcgs do contain radio - loud agn .
their 1.4 @xmath13 radio specific luminosity is , 29.3@xmath1410@xmath15 and 1.6@xmath1410@xmath15 @xmath16 @xmath17 @xmath18 for a2597 and s159 @xcite respectively , which are typical for bcgs in cool core galaxy clusters @xcite . in this work
we will concern ourselves with the extended molecular and ionised gas surrounding the bcgs in a2597 and s159 .
a2597 and to a lesser extent s159 have been the subject of numerous investigations and have been observed at many wavelengths from radio to x - rays ( e.g. , j05 ; vd97 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in both clusters ionised and molecular gas
was observed to at least 50 @xmath0 and 20 @xmath0 from their bcg nuclei respectively ( j05 ; * ? ? ?
previous investigations of these objects made use of narrowband imaging and longslit spectra .
using long slit observations we were only able to sample parts of the extended line emission and with narrowband imaging no information on the dynamics of the gas is obtained .
furthermore the emission sampled with longslits in previous observations is often strongly dominated by strong emission from the bcg nucleus . as we will show below
, the emission away from the nuclei has very different characteristics .
there are a number of kinematic problems concerning the cooler gas phases in cool core clusters .
in nearby clusters ionised and molecular gas is found in thin , long lived , multi-@xmath0 scale , filamentary structures surrounding the bcg ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these structures show smooth velocity gradients , but no rotation beyond the central few @xmath0 @xcite .
the molecular clouds in these structures are dense and without kinematic support should fall to the galaxy centre .
however , they show no signs of infalling .
velocities barely exceed 100 @xmath19 @xmath17 ( j05 ; * ? ? ?
* and this work ) , whereas infall velocities should exceed @xmath2600 @xmath19 @xmath17 .
magnetic support has been invoked by @xcite but there is no observational evidence yet for the required ordered magnetic fields in clusters .
there has also been speculation whether or not all the molecular and ionised gas is locked up in these dense filaments or if a more diffuse underlying component exists . here
we present the first deep k - band integral field ( ifu ) spectroscopic observations of the cluster cores in a2597 and s159 , taken with the spectrograph for integral field observations in the near - infrared ( sinfoni ) on the very large telescope ( vlt ) . with these observations we sample the molecular and ionised gas phases over a major fraction of each cluster s bcg .
our observations are able to provide information on the distribution , kinematics and temperature of this gas . using these measurements we can compare in detail the kinematic and thermal structure of the gas with the x - ray and radio structures , which represent respectively the primary source of mass in the environment and the primary source of local energy input .
similar data on three other cool core clusters has recently been presented by @xcite , but we are the first to make a detailed comparison of such data with radio and x - ray observations of cool core clusters in section 2 we describe the observations and the data reduction .
we discuss the morphology and kinematics of the molecular and ionised gas in a2597 in sections 3 and 4 and similarly for s159 in sections 5 and 6 . in section 7
we will discuss the excitation of the molecular gas and in section 8 we compare the observed gas structure to high resolution x - ray and radio maps .
we summarise our results in section 9 and present our conclusions in section 10 . throughout this paper
we will assume the following cosmology ; @xmath20=72 @xmath19 @xmath17 @xmath21 , @xmath22=0.3 and @xmath23=0.7 .
for abell 2597 at z=0.0821 ( vd97 ) this gives a luminosity distance 363 @xmath24 and angular size scale 1.5 @xmath0 @xmath25 .
for sersic159 - 03 at z=0.0564 @xcite this gives a luminosity distance 245 @xmath24 and angular size scale 1.1 @xmath0 @xmath25 .
the near infrared ( nir ) observations were performed in the k - band with the integral field spectrograph sinfoni @xcite on the very large telescope ( vlt ) .
sinfoni is an image slicer , slicing the image into strips before dispersing the light using 32 slitlets .
the instrument has a spectral resolving power of @xmath26 4000 in the k - band . opting for a 8@xmath278@xmath28 field of view ( fov ) the spatial pixels each cover an area of [email protected]@xmath28 .
each spectral pixel covers 2.45@xmath1410@xmath29 @xmath30 in wavelength , oversampling the resolution by a factor two ( i.e. nyquist sampling ) .
the total on - source exposure time for each source is listed in table [ tabl_obs_summ ] .
the observations were done in a abba pattern ( a = source , b = sky ) and each set was followed by a pointing observation to keep track of pointing drifts .
all observations were done such that the fov was oriented with north pointing up .
the spatial extent of each slitlet is then oriented east - west .
equal amounts of time were spend on the sky and on the source .
each science observation has an exposure time of 600 seconds .
each pointing observation has an exposure time of 60 seconds .
the observations were carried out in july and august of 2005 in photometric sky conditions with a typical seeing of about 0.9 arcsec in k - band .
four fields were observed for a2597 and three fields for s159 .
these fields were selected to lie within areas known to have extended h@xmath11 emission ( j05 ) .
the observed spectral and spatial resolution , as measured from telluric lines and standard star observations , is summarised in table [ tabl_smooth_resolution ] .
the reduction of the data was done using a combination of the eso sinfoni pipeline recipes ( _ sinfoni pipeline version 1.7.1 _ and _ cpl version 3.6.1 _ ) , eclipse @xcite and a set of dedicated idl procedures . from the sinfoni pipeline
we obtain a masterdark frame , masterflat frame , hot pixel map and a slit curvature model .
wavelength calibration , hot pixel removal , slit edge detection , distortion correction , sky removal and illumination correction as given by the pipeline were found to be inadequate for our purposes and therefore an additional set of reduction tasks was written in idl .
the reduction was carried out as follows .
source and sky frames are corrected for dark current and flat fielded using the masterflat and masterdark frames from the sinfoni pipeline . having estimated the slit edges ( by eye ) the different slits are defined and cut out of the science frames .
each of these slits is then treated independently in the subsequent reduction steps .
ccd artefacts are removed from the data .
hot pixels and those affected by cosmic rays are interpolated over .
we correct for the spatial curvature of the slit optics as imaged on the detector by applying the curvature model obtained by the sinfoni pipeline using the eclipse task _ warping _ @xcite .
slit columns which do not contain information across the full wavelength range are removed .
this also removes the overlap between different slits as imaged on the ccd .
the spectra are wavelength calibrated using a set of 19 identified telluric oh lines in the wavelength range 1.95 - 2.30 @xmath30 @xcite .
the output wavelength scale is set to 2.45@xmath1410@xmath29 @xmath30 per pixel thereby nyquist sampling the data .
the k - band night sky is variable on short time scales .
we have rather long exposure times , as compared to the variations in the sky background .
this means that there is a complicated relationship between the sky contribution in our source frame and the sky observed in our corresponding sky frame .
this is readily observed by subtracting two sky frames taken directly after each other , and leads to systematic residuals of up to 10% in the peak heights of telluric lines .
a scaling between the source and its corresponding sky frame thus needs to be performed in order to decrease these residuals .
the standard sky scaling performed by the eso sinfoni pipeline reduces these residuals to about 5% and the special sinfoni pipeline sky correction utility reduces the residuals further to about 4% . in both cases
it was noted that the sky removal suffered from poor wavelength matching between the sky and source frames due to flexure of the instrument . in our approach
we remove the sky emission after detailed wavelength calibration using the telluric lines .
we a apply a simple scaling function to the sky frame before subtracting it from the source frame .
this scaling function consists of a constant and a small , linear , wavelength dependent factor .
the constant is determined from the relative heights , above the continuum , of the telluric lines and assumed to hold at 2.1 @xmath30 . the small , linear , wavelength dependent factor is the slope of a linear fit to the ratio of the object spectrum and its corresponding sky spectrum .
the full wavelength dependent behaviour of the sky emission between an object and a sky frame is often more complex than the simple linear function used .
here we are only interested in line emission and as such a small residual gradient in the continuum emission does not affect the analysis performed below .
our method leads to residuals that are @xmath312% in the peak heights of telluric lines .
this is a significant improvement over the other methods mentioned above . in the final analysis of the line emission we checked our results carefully for telluric line residuals and removed wavelength regions affected by these from our analysis . after correcting for any residual distortion we collapse the sky and the sky subtracted source spectra into cubes with pixel size ( 0.125@xmath28,0.125@xmath28,0.000245 @xmath30 ) .
it is known that sinfoni , after all reduction steps described above , still has a varying illumination across its fov and that this illumination is a function of wavelength ( j. reunanen priv . comm . ) .
this is mostly due to a difference in the illumination of the various slitlets and most easily observed in the sky cubes . defining a reference slitlet in the sky cubes we determine the variation in illumination across the fov for each wavelength .
we then correct for this variation in the sky subtracted source cubes .
the correction is typically less than 10% and particularly affects wavelengths below 2.1 @xmath30 .
flux calibration is carried out using one or , if available , multiple standard star observations per night .
the standard star observations are reduced in the same way as outlined for the object observations above .
all standard stars observed were either o or b stars , and brighter than 8th magnitude in the k - band . the atmospheric transmission function was determined by dividing the spatially summed standard star spectrum with a black body spectrum of the appropriate temperature .
the absolute flux scale is set by using 2mass k - band magnitudes , these are accurate to 0.05 magnitude in k - band .
following flux calibration the source cubes are combined .
the northern and southern edges of the exposures for the different fields overlap well .
the eastern and western edges overlap less well leading to a higher noise here .
the most northern and southern slitlet have very low signal to noise and were removed from the data .
any remaining telluric emission is removed by defining off - source regions .
these are marked by the dotted lines in figs .
[ fig_area_a2597_ss44 ] and [ fig_area_sersic_ss44 ] .
continuum emission is removed by fitting the continuum in the immediate neighbourhood of the science line .
the continuum subtracted line feature is fit by a single gaussian function , using the @xmath32 routine @xcite within idl . it is observed that a single gaussian always provides a good description of the observed line profile .
the line flux , centre and width are determined from this gaussian fit . for selected regions
line profiles and gaussian fits to them are shown in appendix [ app_tb ] .
the errors quoted for the fitted line properties are based on monte - carlo simulations .
in order to visualise the surface brightness of the line emission we performed a gaussian smoothing of four pixels full width at half maximum ( fwhm ) in both the spatial ( 4 pixels=0.5@xmath28 ) and the spectral plane ( 4 pixels=9.80@xmath1410@xmath29 @xmath30 ) . to visualise the kinematics of the line emitting gas a two pixel fwhm gaussian spatial smoothing and no spectral smoothing
was found to be adequate for a2597 , whereas for s159 a two pixel fwhm gaussian smoothing in both the spatial and spectral planes was performed . the degradation of the spatial and spectral resolution as a function of the smoothing kernel used is given in table [ tabl_smooth_resolution ] .
the corresponding noise is given in tables [ tabl_smooth_sensitivity_a2597 ] and [ tabl_smooth_sensitivity_sersic ] .
surface brightness maps for all lines that could be mapped on a pixel to pixel basis are shown in appendix [ app_lmp_a2597 ] and [ app_lmp_sersic ] .
for a2597 the northern field is not shown as the signal to noise here is inadequate to show the emission on the same spatial resolution as the central and southern fields .
velocity and velocity dispersion are shown only for the strongest detected ionised and molecular gas line .
we note that the velocity structure observed in all detected emission lines agrees with that shown for these lines .
cool core clusters were first discovered by analysing their x - ray emission .
these observations lead to the still unresolved cooling flow problem for the hot x - ray gas ( e.g. * ? ? ?
* ) . in this paper
we are concerned with the cooler hii and h@xmath10 gas phases and will not focus on the cooling flow problem related to the hot x - ray gas . however ,
in section 8 we will investigate whether there is a spatial correlation between the x - ray emitting gas and cooler gas phases . in order to do so
we have obtained all available x - ray data from the _ chandra _ archive .
the a2597 image , fig . [ fig_area_a2597_ss44 ] , is a co - add of three separate observations having a combined exposure time of 153.7 @xmath33 ( project codes 7329 ; 6934 ; 922 ) . the s159 image , fig .
[ fig_area_sersic_ss44 ] , consists of only one shallow 10.1 @xmath33 observation ( project code 1668 ) .
_ chandra _ data for a2597 and s159 has previously been published by @xcite and j05 .
a very notable difference in the x - ray emission for the two clusters is that the peak emission in a2597 is well aligned with the nucleus of the bcg , whereas in s159 the peak emission is about 8@xmath28 north of the bcg nucleus .
out of the many heat sources proposed , agn feedback has received the most attention in recent years .
the observed anti - correlation between x - ray and radio emission , also referred to as _ x - ray cavities _ and _ radio bubbles _ , has led to models in which the agn outflows interact strongly with its surrounding medium ( e.g. * ? ? ? * ) .
the kinetic energy carried by these outflows has been calculated based on these x - ray cavities and recent results show that the mechanical power of the jet that created the x - ray cavities can be orders of magnitude larger than its radio inferred radiative power @xcite . in section 8 we will compare our sinfoni results with high resolution radio maps to investigate how the current agn outflows interact with the cooler gas in the cores of a2597 and s159 .
a2597 : a vla 8.4 @xmath13 map of a2597 was obtained from c. sarazin @xcite .
this map has a beam size of [email protected]@xmath28 .
the one sigma noise is 50 @xmath34 @xmath35 .
it was noted that there is a significant offset of @xmath20.1 seconds in right ascension as compared to the 2mass position of the bcg .
two more 8.4 @xmath13 maps have been published @xcite .
these have a much better agreement with the 2mass position .
we thus conclude that this offset is an error and shift the 8.4 @xmath13 map accordingly .
detailed radio maps at lower frequencies have been published by @xcite and show that there is more low level radio emission present than apparent from the 8.4 @xmath13 map .
however , the 8.4 @xmath13 map does give a good indication of the current agn outflows .
a much higher resolution radio map at 1.3 and 5.0 @xmath13 using very long baseline array ( vlba ) interferometry has been published by @xcite .
these observations show that the current jet has a position angle ( pa ) of 70 degrees .
s159 : archival vla 8.4 ghz observations of s159 ( project code : ab1190 ) were reduced with the nrao astronomical image processing system ( aips ) .
the b - configuration observations were taken in single channel continuum mode with two ifs centred at 8435 and 8485 mhz .
the total on source time was 103 min .
the data was flux calibrated using the primary calibrator 0137 + 331 .
we used the perley & taylor 1999 extension to the @xcite scale to set the absolute flux scale . amplitude and phase variations
were tracked using the secondary calibrator 2314 - 449 and applied these to the data .
the data was imaged using robust weighting set to 0.5 , giving a beam size of [email protected]@xmath28 .
the one sigma map noise is 25 @xmath34 @xmath35 .
radio maps of s159 at 1.4 , 5.0 and 8.4 @xmath13 were previously published by @xcite .
four 8@xmath278@xmath28 fields were observed on and surrounding the bcg pgc071390 in a2597 , see fig . [ fig_area_a2597_ss44 ] .
the integration time for each exposure is 600 seconds . the central field , which includes the nucleus of pgc071390 , contains 13 exposures . the south - eastern ( se ) and south - western ( sw ) field contain 8 and 15 exposures respectively .
the northern field contains 13 exposures .
the overlap region between the central and southern fields is sufficient for the line emission to be mapped without problems .
however , the se and sw fields do not completely overlap everywhere . in various locations along the overlap area
there are small gaps between the fields of one to two pixels ( 1 spatial pixel=0.125@xmath28 ) .
we interpolated these before mapping the emission .
despite this , due to the increased noise at the east , west edges of each field , this overlap region ( about 10 pixels in width ) between the southern fields has a rather poor signal to noise .
the northern field is offset by about 6@xmath28 from the central field .
a four pixel spatial and spectral smoothing was applied to the data prior to fitting the lines .
a single gaussian function provides a good fit to the observed line profile .
surface brightness maps for all other lines that could be mapped on a pixel to pixel basis are shown in appendix [ app_lmp_a2597 ] .
the northern field is not shown in these images because the signal to noise here is inadequate to show the emission on the same spatial resolution as the central and southern fields .
the integrated line fluxes for all lines detected within the observed fields for a2597 are given in table [ tabl_line_summ_a2597_ss44 ] .
all h@xmath10 1 - 0 and
h@xmath10 2 - 1 s - transitions redshifted into the k - band ( 1.95 - 2.40 @xmath36 ) are detected . a flux value for the h@xmath10 2 - 1 s(4 )
line has been omitted due to uncertain continuum subtraction .
the h@xmath10 2 - 1 s(5 ) and the br @xmath37 line are too closein wavelength to be separated by our observations .
none of the h@xmath10 3 - 2 s - transitions were detected .
as an example of the fidelity of the data we show the full k - band spectrum of the nuclear region and the south eastern filament in fig .
[ fig_fs_a2597_ss44 ] .
the h@xmath10 surface brightness maps all show the same structure . as an example of the molecular gas distribution
we show the surface brightness map for the h@xmath10 1 - 0 s(3 ) line in fig .
[ fig_sign_maps_a2597_ss44 ] .
this map clearly shows that the peak of the molecular gas emission coincides with the stellar nucleus of pgc071390 .
two extended gas structures away from the nucleus are observed .
one extends north from the nucleus and hence we will refer to this structure as the northern filament .
the second structure extends from the north - east to the south - west in the se field , just south from the nucleus and hence we will refer to this structure as the southern filament .
we observe that the surface brightness of the molecular gas varies rather smoothly within the nuclear region .
however , from higher spatial resolution hst imaging by ( * ? ? ?
* hereafter d00 ) we know , that the molecular and ionised line emission in the central 4@xmath274@xmath28 is concentrated in narrow clumpy , filamentary structures . here
we do not have the resolution to resolve these structures .
we do note some enhanced emission features , embedded within the central field , extending to the north and east away from the nucleus which are roughly coincidental with some of structures observed by d00 .
the northern filament extends at least up to the northern edge of the central field , i.e. , 6@xmath28 ( 9 @xmath0 ) from the nucleus .
this is well beyond the region in which molecular emission was detected by d00 . using deep k - band longslit spectra j05
have previously observed that the h@xmath10 emission extends at least up to 20 @xmath0 towards the north from the nucleus .
we will see below that molecular gas can still be found in the northern field observed by us , i.e. , at a distance of about 22 @xmath0 from the nucleus , thus confirming the j05 results .
the southern filament is clearly detected in the emission of the stronger lines .
this southern filament has not been observed in d00 , but j05 also find molecular gas south of the nucleus ( see their figures 8 and 11 ) .
the extent of the northern and southern filaments observed here is bounded by the edges of the observed fields , and it is likely that these continue beyond the regions mapped by us .
the pa @xmath11 line is redshifted into the k - band for both galaxy clusters studied here .
the line is redshifted into a region of poor atmospheric transmission , but it is the strongest ionised gas line by far in our spectra and unambiguously detected in both clusters . in a2597
the pa @xmath11 emission globally follows the h@xmath10 emission closely , fig .
[ fig_sign_maps_a2597_ss44 ] .
within the nuclear region enhanced emission is again observed towards the north and east .
these features are roughly coincidental with the emission line filaments observed in d00 . beyond the nuclear region
the emission extends along the northern and southern filaments , peaking in the same locations as the h@xmath10 emission .
we also detect the br @xmath38 , br @xmath37 and fe ii ( 1.8100 @xmath30 rest wavelength ) lines .
the br @xmath37 line is blended with h2 2 - 1 s(5 ) and these can not be disentangled directly by our observations . in the central field we find that br @xmath38/pa @xmath11 = 0.082 .
this ratio agrees with the dust - free case b recombination ratio of these lines for @xmath39 = 10@xmath40 @xmath41 and @xmath4 = 10@xmath42 @xmath3 @xcite .
the case b scenario then implies that br @xmath38/br @xmath37 = 1.5 , and we use this ratio to disentangle the br @xmath37 , h@xmath10 2 - 1 s(5 ) blend . in the nuclesar region
we find that br @xmath37 and h@xmath10 1 - 0 s(5 ) are of similar strength .
a small dust lane has been observed in the nuclear region of a2597 ( d00 ; * ? ? ?
we have investigated whether differential extinction in the k - band may affect our emission line ratios . from the above value for the br @xmath38/pa @xmath11 ratio we find that differential extinction is unimportant in the k - band .
this is confirmed by deep optical spectroscopy of a2597 by vd97 and @xcite .
they find a v - band extinction a@xmath43 @xmath2 1 across the nebulosity .
assuming standard galactic dust ( r@xmath43=3.1 ) an a@xmath43 @xmath2 1 translates in to a@xmath44 @xmath2 0.1 .
this amount of extinction is negligible .
the fe ii ( 1.81 @xmath30 ) line is redshifted to the short wavelength edge of our observed spectrum .
it is unambiguously detected in the nuclear region .
the decrease in the fe ii emission outwards from the nucleus , in both intensity and dispersion , appears to be much faster than for either the hii lines or the h@xmath10 lines .
the hii emission drops by a factor of 3 and the h@xmath10 flux drops by a factor 4 from the nuclear region to a region just north of the nucleus . the fe ii emission drops by a factor of 10 for the same regions .
if the fe ii emission has a different origin than hydrogen lines , for example if it is preferably coming from the agn and the associated jet instead of the gaseous filaments , this may explain the difference .
our observations do not have the spatial resolution to investigate this in detail .
we have searched our spectra for the presence of even higher ionisation lines , such as the si vi ( 1.9634 @xmath30 ) line , which one would expect from typical hard agn spectra .
none of these higher ionisation lines were detected .
this once more confirms the liner nature of pgc071390 .
it may also indicate that the active nucleus is not the main source of ionisation of the gas observed in the core of a2597 .
alternatively it would have to have an atypically soft ionising spectrum .
a single gaussian function gives a good description of the observed line profiles , see appendix [ app_tb ] . from the gaussian fits of these line profiles we obtain information about the kinematical structure of the molecular and ionised gas in a2597 .
the velocity , with respect to the systemic velocity of pgc071390 , and the velocity dispersion of the gas have been derived for all emission lines .
these all show the same global kinematical structure .
the velocity and velocity dispersion maps shown below differ from the surface brightness maps in that only a two pixel spatial smoothing has been applied , as supposed to a four pixel spatial and spectral smoothing .
this is done to preserve as much of the velocity structure as possible and provides us with a velocity resolution of 38 @xmath19 @xmath17 .
the molecular gas in a2597 shows a wealth of small scale kinematical structure .
velocity and velocity dispersion maps were made for all h@xmath10 lines .
all show the same kinematical structure on all scales observed . as an example of this structure
we have displayed the velocity and velocity dispersion maps for the h@xmath10 1 - 0 s(3 ) line in figs .
[ fig_kinm_velc_maps_a2597_ss20 ] and [ fig_kinm_disp_maps_a2597_ss20 ] .
the nuclear region contains a blueshifted and a redshifted gas component at about @xmath4580 @xmath19 @xmath17 .
this is seen more clearly if we place a pseudo slit with a width of 1@xmath28 and a pa of 105.5 degree , centred 1 @xmath0 south of the stellar nucleus .
the corresponding position - velocity diagram is shown in fig .
[ fig_a2597_pv ] .
the velocity structure observed in figs .
[ fig_kinm_velc_maps_a2597_ss20 ] and [ fig_a2597_pv ] is reminiscent of gas rotating around the nucleus and does not appear to be related to an expanding shell or agn outflows .
the average velocity of gas in the nuclear region is approximately zero with respect to the systemic velocity of pgc071390 ( z=0.0821 , vd97 ) .
this shows that the gas here is situated at or near the stellar nucleus .
the reason for placing the pseudo slit slightly south of the nucleus is because east of the nucleus there is a small strongly redshifted feature at + 150 @xmath19 @xmath17 .
whether this feature is part of the global gas flow or a single event is unclear .
it shows up prominently in all velocity maps .
projected on to the sky , the feature appears to be coincident with the north - eastern radio jet of pks2322 - 12 the radio source in pgc071390 , see fig .
[ fig_kinm_velc_maps_a2597_ss20 ] .
the filamentary structures extending towards the north and the south from the nucleus show smooth velocity gradients and these will be discussed in more detail below .
the velocity dispersion of the molecular gas also shows interesting structure . globally the dispersion of the gas decreases with distance from the nucleus .
it drops from an average of about 220 @xmath19 @xmath17 in the nuclear region to about 100 @xmath19 @xmath17 a few @xmath0 north and the south of the nucleus .
the velocity dispersion is very high in two narrow structures extending towards the east and south of the nucleus .
the two - dimensional data allows us to determine the area which is disturbed to be an elongated structure of about 2 @xmath0 by 5 @xmath0 oriented at a pa of about 45 degrees .
projected on to the sky these high dispersion structures appear to run along the inner , south east edge of the curved radio lobes of pks2322 - 12 , see fig .
[ fig_kinm_disp_maps_a2597_ss20 ] .
if we interpret the lobe morphology as a wide angle tail , caused by the relative motion of the agn and the external medium , then the dispersion map illustrates for the first time the turbulent wake expected from this motion .
alternatively , the region of maximum dispersion , at pa@xmath270 degrees from the nucleus , may represent the interaction of the current , vlbi radio jet with the surrounding medium , as has been seen in centaurus a @xcite . in this picture
we must assume that the counter - jet has been deflected near the nucleus to the south , causing the high dispersion region and radio lobe in this direction .
there is , however , no evidence for a major kinematic disturbance at the point of deflection .
the highest velocity dispersion is found for the small , strongly red - shifted feature east of the nucleus .
this high velocity and dispersion for this feature can be explained if this is gas that is entrained within the agn outflow .
the feature aligns well with the current , projected jet axis ( pa=70 degrees , * ? ? ?
velocity and the velocity dispersion maps for the ionised gas in a2597 , as traced by the pa @xmath11 line , are shown in figs .
[ fig_kinm_velc_maps_a2597_ss20 ] and [ fig_kinm_disp_maps_a2597_ss20 ] .
we observe two key features when we compare the pa @xmath11 and h@xmath10 derived kinematics .
firstly , globally we find that the pa @xmath11 derived kinematics follows the h@xmath10 derived kinematics tightly .
secondly , the velocity dispersion of the pa @xmath11 emitting gas appears on average to be slightly higher than the h@xmath10 emitting gas , especially in the nuclear region . it may be possible that on scales below the resolution of our observations the ionised gas has a different distribution than the h@xmath10 emitting gas
this may be especially true in the nuclear region where the active nucleus appears to be strongly interacting with the gas .
the position - velocity diagram shown in fig .
[ fig_a2597_pv ] also shows that the ionised gas , as traced by the pa @xmath11 and fe ii lines , reaches slightly higher velocities in the nuclear region .
d00 show that within the nuclear region the h@xmath10 and hii gas has a very complicated and disturbed morphology and it is difficult to say how well these two trace each other on small scales here .
the kinematics of the molecular and ionised gas for a2597 derived here agrees well with previous long slit investigations by j05 and @xcite .
* hereafter o94 ) detected hi in absorption against the radio continuum source pks2322 - 12 in a2597 .
the absorption observation represents a line of sight of a few arcsec along the central radio source .
they find a spatially resolved broad hi component with @xmath46174 @xmath19 @xmath17 and a narrow unresolved hi component with @xmath4693 @xmath19 @xmath17 at the position of the nucleus .
the width of the broad component is somewhat smaller than the width observed in hii and h@xmath10 .
o94 find that the widths are consistent if one takes into account that the hi absorption measurements only sample the gas in front of the radio source , whereas the hii and @xmath47 measurements sample all of the gas along the line of sight . as in our data o94 find a narow and a broad component , but the relative strength of narrow component is much stronger in their observation .
we do not see the narrow component on the nucleus .
the dominance of the narrow component in the hi observations is probably caused by the 1/@xmath48 dependence of the hi absorption , as pointed out by o94 .
@xmath48 being the spin temperature of the hi gas . in the hi absorption
spectra the cold gas at large radii in front of the nucleus is probably over - represented relative to the hii and h2 emission spectra .
we conclude , as do o94 , that there is no evidence for a kinematically distinct hi component . in fig .
[ fig_film_a2597_ss20 ] we show the surface brightness , velocity and velocity dispersion along the two filamentary structures we identified in our observations of a2597 .
the regions used for this investigation are marked by the green and red squares in fig .
[ fig_h2_area_ss44 ] .
the black points in fig .
[ fig_film_a2597_ss20 ] correspond to green squares and the red points to the red squares .
following the northern filament from slightly south of the nucleus towards the northern edge of the central field we find that the pa @xmath11/h@xmath10 1 - 0 s(3 ) is approximately equal to 0.75 in the nuclear region and rapidly increases to unity outwards .
the northern filament shows a smooth velocity gradient from south to north across the nucleus , as the velocity decreases from + 50 @xmath19 @xmath17 to -50 @xmath19 @xmath17 2@xmath28 north of the nucleus . at this point
the velocity gradient reverses and the velocity increases again to + 50 @xmath19 @xmath17 towards the northern edge of the central field .
velocity gradients and even reversals for this filament may be explained in terms of bending and stretching of the filament , perhaps due to a combination of its proper motion and gravitational forces .
however , it is more likely that we observe multiple filaments , each with its own characteristic motion , along the line of sight .
our data shows that the eastern part of the northern filament is predominantly blue shifted whereas the western part is red shifted .
higher spatial resolution images taken with hst by @xcite and oonk et al .
( in prep . )
show evidence that the northern filament observed here consists of at least two filamentary structures .
we thus favour the latter explanation for the observed velocity structure of the northern filament this interpretation also agrees with what is observed for more nearby galaxy clusters such as perseus and centaurus , where a multitude of long , thin filaments are observed along the line of sight @xcite .
the narrow spatial and velocity range observed here for the filaments however still suggest that any substructure in it will likely have a common origin .
if the gas observed in the northern filament is connected to the gas detected in the northern field its velocity continues to increase to about + 150 @xmath19 @xmath17 , as also shown by j05 . from the j05 observations
it appears that the gas in the central field is joined smoothly with that in the northern field , in terms of both surface brightness and dynamics .
the velocity dispersion along the northern filament decreases smoothly from 220 @xmath19 @xmath17 to 100 @xmath19 @xmath17 , from the nucleus to the edges of the central field .
this decrease is fastest near the nucleus and slows down beyond 3 @xmath0 north of the nucleus .
this point may mark a change in the influence of the agn upon the dynamical state of the gas .
the southern filament has a much lower surface brightness and is hence detected at a lower signal to noise .
variations along this filament are thus more difficult to detect . following this filament from the north - east ( ne ) to the south - west ( sw )
we find that the surface brightness is highest at its ne edge whereafter it decreases slightly and becomes approximately constant .
the velocity decreases from + 50 @xmath19 @xmath17 to about -40 @xmath19 @xmath17 .
the velocity dispersion remains constant at about 100 @xmath19 @xmath17 along the filament .
we will discuss the stability of the observed filaments in more detail below .
three 8@xmath278@xmath28 fields were observed on and surrounding the bcg eso291-g009 in s159 , fig . [ fig_area_sersic_ss44 ] .
the integration time for each exposure is 600s .
the south - eastern ( se ) and south - western ( sw ) fields contain 8 and 9 exposures respectively .
the northern field contains 8 exposures .
the se field contains the nucleus of eso 291-g009 .
there is no overlap between the three fields observed .
a four pixel spatial and spectral smoothing was applied to the data prior to fitting the lines .
a single gaussian function provides a good fit to the observed line profiles .
surface brightness maps for all detected emission lines that could be mapped on pixel to pixel basis are shown in appendix [ app_lmp_sersic ] .
only two out of five h@xmath10 1 - 0 s - transition lines redshifted into the k - band are unambiguously detected for s159 .
these are the h@xmath10 1 - 0 s(1 ) and h@xmath10 1 - 0 s(3 ) lines .
their resulting integrated line fluxes are given in table [ tabl_line_summ_sersic_ss44 ] .
the presence of the h@xmath10 1 - 0 s(2 ) line can also be confirmed , see also fig .
[ fig_fs_sersic_ss44 ] . this line can not be fitted reliably due to a strong telluric line residual in the red wing of the line profile and as such a flux value has been omitted .
tentative evidence is found for the presence of the h@xmath10 1 - 0 s(0 ) and h@xmath10 1 - 0 s(4 ) lines . however , both lines lie in spectral regions of poor atmospheric transmission and are close to strong telluric features , preventing us from claiming detections .
we have searched for the h@xmath10 2 - 1 s - transitions , but these and higher h@xmath10 transitions remain undetected in our spectra . except for the above mentioned two h@xmath10 lines only the pa @xmath11 line is detected .
full k - band spectra of the nuclear region and the western filament are presented in fig .
[ fig_fs_sersic_ss44 ] .
the two h@xmath10 surface brightness maps show the same structure and as an example of this we show the map for the h@xmath10 1 - 0 s(3 ) line in fig .
[ fig_sign_maps_sersic_ss44 ] .
this map clearly shows that the peak of the h@xmath10 emission is well aligned with the stellar nucleus of eso0291-g009 .
a filament of molecular emission extends north - east from the nucleus up to the edge of the se field .
we will refer to this structure as the northern filament .
clumpy emission extends towards the west and the south of the nucleus , up to the edges of the se field .
the sw field shows a strong filament of gas having an east - west elongation , originally discovered by @xcite .
we will refer to this structure as the western filament .
the northern field shows two features ( i ) low signal to noise clumpy emission in the southern and central part of the field , and ( ii ) a stronger , somewhat larger emission feature at the northern edge of the field .
both features are treated in more detail below . whether the clumpy emission observed in the northern field
is part of the northern filament observed in the se field is not clear from our observations . the spatial extent of the gas for both the western and northern filament is bounded by the edges of our observed fields and it is likely that these filaments continue beyond the regions mapped here , as seems to be the case from narrowband h@xmath11+[nii ] imaging by j05 and @xcite .
pa @xmath11 is the only ionised gas line detected in our k - band spectra for s159 .
it is redshifted into a region of poor atmospheric transmission .
strong pa @xmath11 emission is found along the northern and western filaments , fig .
[ fig_sign_maps_sersic_ss44 ] .
we detect pa @xmath11 in all places where h@xmath10 emission is detected .
a noticeable difference concerns the nuclear region .
almost no pa @xmath11 emission appears to be associated with the stellar nucleus .
as we will see below , some pa @xmath11 emission is found here , but there is a strong decrease of it relative to molecular hydrogen emission .
the pa @xmath11/h@xmath10 1 - 0 s(3 ) ratio is observed to drop from about 0.7 in the filaments to about 0.2 in the nuclear region .
we note that the nuclear region has a rather high spectral noise , due to the strong stellar continuum , which affects our ability to detect emission lines here .
the detection of the pa @xmath11 line is furthermore complicated by it being in a region of poor atmospheric transmission .
pa @xmath11 is also present within the northern field and , like the h@xmath10 gas , appears to be clumpy .
the strong emission feature observed in h@xmath10 towards the northern edge of this field is also confirmed by pa @xmath11 emission . in s159
the pa @xmath11 line is the only ionised gas line detected .
an estimate of differential extinction within the k - band can thus not be made directly from our observations .
@xcite finds finds little variation from a@xmath43 @xmath2 1 across the nebulosity .
applying the same arguments to s159 as for a2597 , we conclude that differential extinction in the k - band is negligible for this cluster .
a single gaussian function gives a good description of the observed line profiles , see appendix [ app_tb ] .
the velocity , with respect to the systemic velocity of eso 0291-g009 , and the velocity dispersion of the gas have been derived for all emission lines .
these all show the same global kinematical structure .
the velocity and velocity dispersion maps shown below differ from the surface brightness maps in that only a two pixel spatial and spectral smoothing has been applied .
this is done to preserve as much of the velocity structure as possible and provides us with a velocity resolution of 51 @xmath19 @xmath17 .
the kinematical structure observed in the h@xmath10 1 - 0 s(1 ) and h@xmath10 1 - 0 s(3 ) maps is the same . as an example of this structure
we show the velocity and velocity dispersion of the h@xmath10 gas as traced by the h@xmath10 1 - 0 s(3 ) line in figs .
[ fig_kinm_velc_maps_sersic_ss22 ] and [ fig_kinm_disp_maps_sersic_ss22 ] .
the nuclear region contains a blue- and redshifted gas component at about @xmath45120 @xmath19 @xmath17 .
this velocity structure is reminiscent of gas rotating around the nucleus .
however , the two gas filaments coming in from the north - east and west towards the nucleus may also explain the observed velocity structure .
the velocity along the filaments will be treated in more detail below .
the gas extending towards the west and south from the nucleus appears to be predominantly blueshifted .
the average velocity of the gas in the nuclear region is equal to the systemic velocity of eso 291-g009 ( @xmath49=0.0564 , * ? ? ?
* ) showing that the gas is situated at or near the stellar nucleus .
globally the dispersion of the gas is low and decreases with distance from the nucleus .
the dispersion of the gas in the nuclear region is about 230 @xmath19 @xmath17 , but drops rapidly to about 90 @xmath19 @xmath17 along the two filaments .
this is similar to what is observed in a2597 . in projection the high dispersion structure around
the nucleus appears coincidental with the radio jets of eso0291-g009 , see fig .
[ fig_kinm_disp_maps_sersic_ss22 ] .
the increase in the velocity dispersion here again indicates that the radio jets are stirring up the gas . from figs .
[ fig_kinm_velc_maps_sersic_ss22 ] and [ fig_kinm_disp_maps_sersic_ss22 ] it is difficult to draw conclusions on the average velocity and velocity dispersion of the clumpy low signal to noise emission in the northern field . from the previous investigation by j05 it appears that the gas observed here
is connected to the strong filament extending north from the nucleus .
below , we will see that the clumpy emission has velocities varying between -20 and -60 @xmath19 @xmath17 and a velocity dispersion less than 100 @xmath19 @xmath17 .
if this emission is connected to the filament extending north this implies that the line emission continues to decrease with distance to nucleus as was also shown by j05 .
velocity and velocity dispersion maps for the ionised gas in s159 as traced by the pa @xmath11 line are shown in figs .
[ fig_kinm_velc_maps_sersic_ss22 ] and [ fig_kinm_disp_maps_sersic_ss22 ] . as for a2597
we find that the pa @xmath11 derived kinematics follows that of the h@xmath10 lines closely .
the only exception in s159 being the high dispersion nuclear region observed in h@xmath10 emission , which appears to be missed by the pa @xmath11 emission . the high noise in the nuclear region of s159 combined with the poor atmospheric transmission in the wavelength range of the pa @xmath11 line makes the fit of this line here rather difficult . in fig .
[ fig_film_sersic_ss22 ] we show the surface brightness , velocity and velocity dispersion along the two filaments observed in s159 .
the regions used for this investigation are marked by the green squares in fig .
[ fig_h2_area_ss44 ] .
following the northern filament from the north - eastern edge in the se field southwards toward the nucleus and subsequently to the eastern edge of the se field , we find that the pa @xmath11/h@xmath10 1 - 0 s(3 ) decreases strongly in the nuclear region and is approximately constant outside of it .
the filament has a smooth velocity gradient .
north - east of the nucleus the velocity decreases very slightly with distance from about + 130 @xmath19 @xmath17 to about + 90 @xmath19 @xmath17 . across the nucleus the velocity changes from about + 100 @xmath19 @xmath17 to -100 @xmath19 @xmath17 .
whether this velocity reversal is due to rotation or due to the filamentary structure of the gas can not be clearly distinguished .
the velocity dispersion of the gas in the northern filament is low everywhere , except within the nuclear region .
all detected lines show an increase in the velocity dispersion near the nucleus , but the increase in the h@xmath10 lines is much higher than for the pa @xmath11 line .
the decrease in the dispersion east of the nucleus is difficult to measure due to low signal to noise here .
the surface brightness of the western filament has two prominent peaks about 11@xmath28 from the nucleus .
it has a very smooth velocity gradient along the filament , from about -100 @xmath19 @xmath17 to about + 20 @xmath19 @xmath17 .
this velocity structure agrees with the possibility that it is connected to the nuclear region .
the dispersion of the gas in this filament is low everywhere .
all three lines detected show the same flux and velocity structure along the western filament .
molecular hydrogen emission can be excited by various processes , ( 1 ) thermal excitation produced by kinetic energy injection into the gas due to for example shock heating @xcite , ( 2 ) fluorescence by soft - uv photons , i.e. photo dissociation regions ( pdr ) @xcite , ( 3 ) high energy x - ray photons , i.e. x - ray dissociation regions ( xdr ) @xcite and ( 4 ) high energy particles @xcite .
if the density of the molecular gas is above the critical density the gas is in local thermal equilibrium ( lte ) .
it is then not possible to distinguish between the various h@xmath10 excitation mechanisms and one would observe a thermal h@xmath10 spectrum where the excitation temperature is equal to the kinetic temperature of this gas .
for the photon and particle processes mentioned above the gas can be thermalised , via heating through secondary ( suprathermal ) electrons . in order to investigate the h@xmath10 excitation mechanism we have constructed h@xmath10 excitation diagrams for a number of regions in a2597 .
seven regions , a1-a7 , were selected based on having distinct physical properties in terms of either their emissivity , spatial location or kinematical structure , see fig .
[ fig_h2_area_ss44 ] .
similarly seven regions , b1-b7 , were selected in s159 , see fig .
[ fig_h2_area_ss44 ] .
excitation diagrams were not made for s159 since we only have reliable detections for two h@xmath10 lines .
all line detections for regions a1-a7 and b1-b7 are presented in appendix [ app_tb ] .
the line profiles are well described by a single gaussian function . for the lines detected in these regions
their integrated fluxes are given in tables [ tabl_line_area_a2597_ss20 ] and [ tabl_line_area_sersic_ss22 ] .
kinematical information for these regions are presented for the h@xmath10 1 - 0 s(3 ) and pa @xmath11 lines only and these are given in tables [ tabl_kinm_area_a2597_ss20 ] and [ tabl_kinm_area_sersic_ss22 ] .
gas temperatures and masses for the selected regions are given in tables [ tabl_phys_area_a2597_ss20 ] and [ tabl_phys_area_sersic_ss22 ] .
we will first shortly describe the selected regions below .
in the central field we have selected four regions , a1 , a2 , a3 and a6 .
region a1 corresponds to the nuclear region , a2 samples the region where the northern filament connects to the nucleus .
a3 , just north of a2 , samples that part of the northern filament that appears to not be influenced ( directly ) by the nucleus anymore .
lastly , in the central field a clump of strong emission is noted on the western edge of the field , which we selected as region a6 . there is tentative evidence for a narrow ridge connecting a6 to the nucleus , but we can not confirm this with the present observations .
the emission in region a6 itself is also uncertain , due to the increased noise at the edge of the field .
it is only observed significantly in two h@xmath10 lines and therefore we have not constructed an excitation diagram for this region . in the south - eastern field
we have selected one region , a4 , which captures most of the emission in the southern filament .
this region was selected such to avoid the noisy overlap region between the se and sw fields .
the strongest emission lines showed weak evidence in their surface brightness maps that the south - eastern filament may extend across the overlap region into the sw field .
region a5 was selected to test this .
significant line emission is found in this region .
the kinematical structure of the gas observed in a5 is similar to that measured in a4 and thus a connection between the two regions seems plausible .
the northern field also contains significant emission for the strongest emission lines .
a systematic search for line emission in the northern field was performed using various binnings .
region a7 was selected to show that emission does exist at a significant level in the northern field .
j05 previously showed that molecular and ionised gas existed out to 20 @xmath0 north from the nucleus using long slit spectra .
the h@xmath10 1 - 0 s(1 ) , 1 - 0 s(3 ) and pa @xmath11 lines are all detected at the 3.0 sigma level , and the h@xmath10 1 - 0 s(5 ) line is observed at the 2 sigma level .
all four detected lines show the same velocity structure .
the velocity dispersion of the gas in a7 is more difficult to constrain . of the four lines the h@xmath10 1 - 0 s(3 )
line is observed at the highest significance and has the most reliable fit .
this line has a dispersion of about 60 @xmath19 @xmath17 using a spatial smoothing of two pixels .
the other lines have higher fitted velocity dispersions ranging from 70 to 110 @xmath19 @xmath17 , however within their large errors ( 40 - 60 @xmath19 @xmath17 ) they agree with the h@xmath10 1 - 0 s(3 ) result .
the positive velocity and low velocity dispersion of the gas makes it plausible that the gas in a7 is connected with the northern filament .
this interpretation agrees with the j05 results .
we conclude that molecular and ionised gas is present at least up to region a7 , i.e. 22.5 @xmath0 from the nucleus , in good agreement with the j05 results .
there is tentative evidence from regions investigated north and south of a7 within the northern field that line emission is present there as well . in the se field
we have selected four regions , b1 , b2 , b3 and b4 .
region b1 corresponds to the nuclear region .
b2 samples the tail of the northern filament as it connects to the nucleus .
region b3 , just north of b2 , is selected to sample the filament where it is no longer ( directly ) influenced by the nucleus .
lastly , b4 is selected to contain the clump of emission just to the south of the nucleus . whether this clump is part of a filament or
not our observations can not confirm . as discussed above ,
the nuclear region b1 is significantly noisier than surrounding regions .
the spectral noise is higher here by a factor two to three .
the h@xmath10 1 - 0 s(1 ) and 1 - 0 s(3 ) lines are still easily detected here .
the h@xmath10 1 - 0 s(3 ) line is considerably stronger than the h@xmath10 1 - 0 s(1 ) line in this region as compared to any of the other regions .
the pa @xmath11 line is detected at a much lower significance and at a much lower flux level . in the sw field
we have selected one region , b5 , which captures most of the western filament .
we have selected two regions in the the northern field , b6 and b7 , to investigate the low level clumpy emission here .
region b7 was selected to capture the strong clump of emission at the northern edge of this field .
region b6 was selected to investigate the remaining emission .
the summed spectra clearly show that line emission is present in the northern field .
we can thus conclude that molecular and ionised emission is present at least up to 18.0 @xmath0 from the nucleus . whether the gas in the northern field is directly connected to that in the se field
can not be confirmed by our observations , although it seems plausible from the j05 results .
the velocity and low velocity dispersion of the gas are such that this gas can be connected smoothly to that in the northern filament . in the case of a gas in lte , assuming a ortho : para abundance ratio of 3:1 , there is a simple relation between the flux @xmath50 and the temperature @xmath51 corresponding to the energy of the upper state of a line ,
@xmath52 @xmath2 @xmath53 @xmath2 @xmath54 , ( e.g. , j05 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . normalising the flux of each h@xmath10 line flux by the flux of the corresponding h@xmath10 1 - 0 s(1 ) line , we find ; + + @xmath55 + + here @xmath50 is the flux of the line , @xmath56 its transition probability , @xmath57 its frequency and @xmath58 the statistical weight of the transition .
if the molecular gas is in lte the h@xmath10 emission lines will lie on a straight line in a diagram of @xmath59 vs. @xmath60 .
@xmath1 , the reciprocal of the slope , will then be the kinetic temperature of this gas .
we have investigated relation ( 1 ) for the all regions selected in abell 2597 in which we have detected at least 3 h@xmath10 lines .
we show in fig .
[ fig_lte_mod_a2597_ss20 ] that the h@xmath10 lines detected in these regions are well fit by a thermal excitation model , with an average temperature of about 2300 @xmath3 .
the derived excitation temperature for each region is given in table [ tabl_line_area_a2597_ss20 ] .
besides the best - fitting lte model for the h@xmath10 line fluxes we plot three additional h@xmath10 models .
these models are shown for qualitative comparison purposes only , and are not tuned exactly to our physical conditions , see fig .
[ fig_lte_mod_a2597_ss20 ] .
the best - fitting lte model is given by the black solid line .
the red dotted line is a low - density uv fluorescence model by @xcite that does not include collisions ( their model 14 ; @xmath61=3@xmath1410@xmath62 @xmath63 , a temperature @xmath4=100 @xmath3 and a uv intensity @xmath64=10@xmath62 relative to @xmath65 ) . the blue dotted line is a high - density uv fluorescence model by @xcite which does include collisions ( their model 2d ; @xmath61=1@xmath1410@xmath66 @xmath41 , a temperature @xmath4=1000 @xmath3 and @xmath64=10@xmath40 relative to @xmath65 ) .
lastly the green dotted line is the cosmic ray model by @xcite , which was developed for the gas filaments observed in the perseus cluster .
the black & van dishoeck low - density uv and the ferland cosmic ray models have distinct features that make them deviate from a thermal model .
low - density uv fluorescence models tend to boost the higher s - transitions ( 2 - 1 , 3 - 2 , ... ) relative to the 1 - 0 s - transitions as compared to a lte model .
the ferland model is observed to boost the even h@xmath10 1 - 0 s - transitions relative to uneven h@xmath10 1 - 0 s - transitions as compared to a lte model .
the high - density uv model by sternberg & dalgarno simply shows that at high densities , collisions within the gas will cause it to become thermalised and thus the line ratios also produce a straight line in our excitation diagrams .
we thus conclude , qualitatively , that out of the four models investigated here that a lte model provides the best description of the data . as the molecular line ratios in a2597
appear to be in lte this implies that the density of this gas is near its critical density , @xmath67 @xmath68 10@xmath66 @xmath69 @xcite and is dominated by collisional excitation .
information regarding the source of excitation is thus not obtainable from this data set .
there is a trend that on average we find higher lte temperatures of the molecular gas in the filaments as compared to the nuclear region .
we note though that within errors the temperatures agree for all regions , except for a6 . neither the h@xmath10 1 - 0 nor the h@xmath10 2 - 1 s - transitions lie exactly on a straight line .
if we use only pairs of lines like the h@xmath10 1 - 0 s(1 ) , 1 - 0 s(3 ) line pair or the h@xmath10 1 - 0 s(3 ) , 1 - 0 s(5 ) line pair to determine an excitation temperature , assuming lte , we find on average a temperature that is a few hundred degrees lower or higher than when all lines are used .
typically the first pair gives a lower temperature and the second pair a higher temperature .
if the h@xmath10 gas observed here is at its critical density then there is a pressure imbalance between the molecular gas @xmath70 @xmath2 @xmath71 = @xmath72 @xmath68 10@xmath73 and the ionised gas has @xmath74
@xmath2 @xmath71 = 10@xmath7510@xmath42 = 10@xmath66 . that the molecular and ionised gas are not in pressure equilibrium has previously been suggested by j05 using similar arguments . for sersic 159 - 03 only two h@xmath10 lines
were reliably detected in all regions .
if we assume that the molecular gas here is in lte , we can calculate an excitation temperature for this gas .
the resulting excitation temperatures for the selected regions are given in table [ tabl_line_area_sersic_ss22 ] .
again we find an average temperature of about 2300 k for the h@xmath10 gas in the filaments .
in the nuclear region the strong increase of the h@xmath10 1 - 0 s(3)/h@xmath10 1 - 0 s(1 ) ratio leads to a very uncertain and physically unrealistic temperature above the dissociation temperature for h@xmath10 .
this either means that our line fits overestimate this ratio or that a different excitation mechanism is at work here . in this work we have only considered the k - band h@xmath10 lines .
these h@xmath10 lines do not show strong deviations from a single temperature thermal model with @xmath4 @xmath68 2000 - 2500 @xmath3 . from recent spitzer spectroscopy ( jaffe & bremer in prep . ) we find that another h@xmath10 gas component exists in the nuclear regions of a2597 and s159 with a lte temperature of about 300 @xmath3 .
the situation in these clusters is thus similar to the situation in the perseus cluster where multiple temperature lte components have been invoked to explain the h@xmath10 line ratios ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
whether the need for multiple lte components to h@xmath10 line ratios hints at a multiphase gas , a difference in optical depth , or a different excitation mechanism is unclear currently .
a more thorough modelling of the molecular gas , including all of the measured of the h@xmath10 lines in the infrared and mid - infrared , for a2597 and s159 will be presented by us in a future paper . if we assume that the h@xmath10 gas is in lte and can be described by a single kinetic temperature of about 2000 @xmath3 , then the h@xmath10 1 - 0 s(1 ) line luminosity represents about 10 percent of the total h@xmath10 luminosity .
this is estimated using a list of 312 h@xmath10 lines , corresponding to all h@xmath10 emission lines with an intensity greater than 1 percent of h@xmath10 1 - 0 s(1 ) line flux for @xmath4 @xmath76 4000k . from the total integrated h@xmath10 1 - 0 s(1 ) line fluxes in tables [ tabl_line_summ_a2597_ss44 ] and [ tabl_line_summ_sersic_ss44 ] we thus find @xmath77(@xmath78 , a2597 ) = 1.1@xmath1410@xmath79 @xmath16 @xmath17 and @xmath77(@xmath78 , s159 ) = 1.2@xmath1410@xmath80 @xmath16 @xmath17 within the fields observed by us .
this is significantly below the total h@xmath10 luminosity estimated by j05 , i.e. @xmath77(@xmath78)@xmath210@xmath81 @xmath16 @xmath17 .
j05 calculate the total h@xmath10 luminosity in the same manner as we do here .
the difference follows from two simple arguments .
firstly j05 assumed that the h@xmath10 1 - 0 s(1 ) line luminosity represents only about 1 percent of the total h@xmath10 luminosity .
this is true for low - density uv excitation models such as the black & van dishoeck models .
however , when the gas is in lte ( see above ) or in a xdr environment @xcite a fraction of 10 percent is found .
the j05 results thus need to be corrected down by a factor of 10 as consequence of this .
the second argument that j05 made is that the h@xmath10 emission coexists with the h@xmath11 emission .
this argument follows from the hii / h@xmath10 ratio which is observed to remain rather constant over large areas ( j05 and this work ) . j05
detect h@xmath11 over a much larger area than the fields observed by us .
correcting our results upward for the area covered by the h@xmath11 map in j05 increases the h@xmath10 luminosity given above by a factor of 10 . with these corrections , we find that the h@xmath10 luminosity obtained in this work is in good agreement with the j05 results .
to conclude , upon correcting our results for the area covered by the total extent of the ionised gas nebulae , we find h@xmath10 luminosities @xmath77(@xmath78 , a2597)@xmath210@xmath82 @xmath16 @xmath17 and @xmath77(@xmath78 , s159)@xmath210@xmath79 @xmath16 @xmath17 for our clusters .
the analysis performed above relies heavily on the assumption that the h@xmath10 gas is in a single temperature phase .
if there is colder h@xmath10 gas present in these clusters , as expected from co observations by @xcite and mid - infrared spectroscopy ( jaffe & bremer in prep . ) we underestimate the total h@xmath10 luminosity here . in the previous section we found that a lte model provides a good description of the observed h@xmath10 line ratios in a2597 and assumed that the same is true for a s159 .
following @xcite and @xcite the mass of a h@xmath10 gas in lte conditions with a single kinetic temperature , can be calculated using the following equation , + + @xmath83 + + here @xmath84 is the proton mass , @xmath85 is the planck constant and @xmath86 is the distance to the cluster core .
@xmath87 is the flux of the h@xmath10 1 - 0 s(1 ) line and @xmath88 is its population fraction .
@xmath89 is the transition probability ( @xmath89 = 3.47@xmath90 @xmath17 ) and @xmath57 is the frequency of the h@xmath10 1 - 0 s(1 ) line .
the right - hand side of equation ( 2 ) is obtained by assuming a vibration temperature @xmath91 = 2000 @xmath3 for the gas , in which case @xmath88 = 1.22@xmath92 . in the above formula @xmath93
is given in units of solar masses .
the resulting h@xmath10 gas masses for the regions investigated here are given in tables [ tabl_phys_area_a2597_ss20 ] and [ tabl_phys_area_sersic_ss22 ] .
the total h@xmath10 gas mass integrated over all observed fields is 4.5@xmath1410@xmath42 @xmath45 4.2@xmath1410@xmath40 @xmath6 for a2597 and 5.2@xmath1410@xmath62 @xmath45 2.9@xmath1410@xmath62 @xmath6 for s159 .
since our observations do not cover the entire cluster core these numbers underestimate the total amount of molecular gas present in these clusters .
we furthermore note that there is evidence from mid - infrared spectroscopy from jaffe & bremer ( in prep . ) that a colder h@xmath10 phase ( @xmath4 @xmath2 300 @xmath3 ) exists in a2597 and s159 .
if this colder h@xmath10 gas coexists with the warmer h@xmath10 gas observed here this means that we strongly underestimate the h@xmath10 mass present in these clusters .
co observations by @xcite indicate that even colder h@xmath10 gas is present in a2597 at a temperature of a few tens of kelvin .
the total molecular gas mass inferred from co for a2597 is 8@xmath1410@xmath73 @xmath6 , although this value relies heavily on the co to h@xmath10 conversion factor . to our knowledge
there is no published co detection for s159 .
following @xcite the mass of the ionised hydrogen gas can be estimated as @xmath94 = @xmath95 . here
@xmath84 is the proton mass , @xmath39 is the electron density and @xmath96 is the volume of the hii emitting region .
using @xmath97 = 9.788@xmath1410@xmath98 @xmath16 @xmath17 @xmath69 and @xmath99 = 0.339 @xcite , for @xmath39 = 10@xmath40 @xmath41 and @xmath4 = 10@xmath42 @xmath3 ) , we can write @xmath100 = @xmath101 = 3.32@xmath1410@xmath102(@xmath103 ) . here
@xmath100 is the flux of the pa @xmath11 line , @xmath104 is the volume emissivity coefficient of pa @xmath11 and @xmath86 is the distance to the cluster core .
the hii gas mass can be written as : + + @xmath105 + + using the right - hand side of equation ( 3 ) the mass of the hii emitting gas , @xmath94 , is given in units of solar masses .
@xcite finds an electron density , @xmath39 = 200 @xmath41 , for the central regions of a2597 . thus assuming an electron density @xmath39 = 200 @xmath41 , the mass of the hii emitting gas in the selected regions of abell 2597 and sersic 159 - 03 can be calculated .
the resulting hii gas masses are given in tables [ tabl_phys_area_a2597_ss20 ] and [ tabl_phys_area_sersic_ss22 ] .
the total hii gas mass integrated over all observed fields is 9.7@xmath1410@xmath66 @xmath45 3.2@xmath1410@xmath106 @xmath6 for a2597 and 7.6@xmath1410@xmath106 @xmath45 8.4@xmath1410@xmath42 @xmath6 for s159 .
since our observations do not cover the entire cluster core these numbers underestimate the total amount of ionised gas present in these clusters .
the total ionised gas mass to warm molecular gas mass is similar in both clusters ; @xmath107 @xmath2 10@xmath40 .
the hii gas mass derived here for a2597 agrees well with @xcite who find @xmath94 = 5.8@xmath1410@xmath66 @xmath6 for a region with radius 2.5 @xmath0 centered on the bcg nucleus . from our observations
we see that the ionised and molecular gas is locked up in large scale filamentary structures surrounding the bcg .
if the filaments in our the clusters are similar to the ones observed in more nearby clusters , such as perseus @xcite , then we do not resolve their true physical thickness .
furthermore if we assume that the filaments observed here are connected to the global distribution of ionised gas in these clusters then its clear that these extent far beyond the distance observed here by us ( j05 ; * ? ? ?
this interpretation is strengthened by higher resolution images in for example h@xmath11 and ly @xmath11 emission ( d00 ; * ? ? ?
these show the existence of narrow gas filaments coincidental with the larger scale structures observed here . considering this ,
even if these large structures consist of many unresolved filaments it seems plausible that all of these have a similar origin .
the currently favoured picture is that these structures are related to past and current agn outflows .
another explanation may be that these structures arise in regions of these clusters where heating and cooling rates are not balanced and thus that we observe residual cooling here . not only
is the origin of these structures unclear , there are many other puzzling aspects concerning them . on large scales
these structures appear to connect all the way down to the nucleus of the bcg .
one may speculate on whether this gas falls in towards the black hole thereby feeding it and sustaining its activity . however , the velocities observed for the filaments studied here or in other clusters @xcite does not agree with gas that is freely infalling from distances far away from the potential centre .
we observe the gas to be moving at velocities of about 100 @xmath19 @xmath17 , whereas for freely infalling gas we would expect to observe velocities of about 600 @xmath19 @xmath17 .
we would expect tidal forces to rip these filaments apart as discussed in @xcite .
these authors suggest that magnetic fields may help stabilise these filaments against these gravitational forces . for a typical filament in perseus , or one of our clusters ,
the required magnetic field is around a few hundred @xmath108 g . from equipartition arguments the icm typically has magnetic field strengths less than 10 @xmath108 g .
larger magnetic field strengths , up to about 50 @xmath108 g , have been found in the centres of clusters but these are usually related to current agn outflows @xcite .
significantly higher magnetic field strengths in the filaments are necessary to stabilise them against gravitational forces . some of the gaseous filaments appear , in projection , to be related to the current agn outflows , see fig .
[ fig_sign_maps_a2597_ss44 ] and [ fig_sign_maps_sersic_ss44 ] .
these outflows maybe be responsible for pushing gas outwards , thereby in fact creating the observed filaments @xcite .
this may explain why some filaments do not show large infall velocities .
however , there are also filaments that are not related to current radio outflows .
if these filaments were deposited at their current location by previous agn outflows , then this gas has had a long time to cool and should now be falling back at high velocities towards the potential centre .
no signs of high velocity infalling gas is observed in either of our clusters .
we will discuss the relation between the filaments and the radio emission in more detail below .
in the above sections we have shown that there is a strong spatial and dynamical relation between the hii and h@xmath10 gas phases in the cores of a2597 and s159 . below we investigate how these gas phases relate to x - ray and radio emission .
we find that globally the h@xmath10 and hii line emission in a2597 appears to match the x - ray emission as imaged by _
chandra _ rather well , see fig .
[ fig_area_a2597_ss44 ] . however , upon examining the detailed distribution , fig .
[ fig_film_a2597_ss20 ] , we find that the x - ray brightness profile is considerable shallower than that of the colder gas . in s159 , see fig . [ fig_film_sersic_ss22 ] , we observe that the peak x - ray emission is significantly offset from the bcg nucleus and the peak h@xmath10 , hii emission
. one may perhaps argue for a relation between the x - ray and pa @xmath11 emission along the northern filament , but pa @xmath11 is difficult to detect in the nuclear region .
there appears to be no relation between pa @xmath11 and x - ray emission along the western filament in s159 .
this leads us to conclude that is no strong correlation between the x - ray and either the h@xmath10 or hii phase in a2597 and s159 .
the total x - ray luminosity is @xmath109(a2597 ) = 2.1@xmath1410@xmath110 @xmath16 @xmath17 ( * ? ?
* @xmath111 ) and @xmath109(s159 ) = 9.6@xmath1410@xmath82 @xmath16 @xmath17 ( * ? ? ?
* @xmath111 ) in a2597 and s159 respectively . taking into account the difference in areas between this work and that of j05 , the total warm h@xmath10 luminosity for these clusters is a factor 10 - 100 times less than their x - ray luminosities and not of the same order of magnitude as previously claimed in j05 . adding the colder h@xmath10 gas to this analysis ( jaffe & bremer in prep . ) to obtain a total warm plus cold h@xmath10 luminosity may change this conclusion . in figs .
[ fig_sign_maps_a2597_ss44 ] , [ fig_kinm_velc_maps_a2597_ss20 ] , [ fig_kinm_disp_maps_a2597_ss20 ] , [ fig_sign_maps_sersic_ss44 ] , [ fig_kinm_velc_maps_sersic_ss22 ] and [ fig_kinm_disp_maps_sersic_ss22 ] , we show vla 8.4 @xmath13 radio contours on top of the h@xmath10 1 - 0 s(3 ) and pa @xmath11 surface brightness , velocity and velocity dispersion maps .
the total radio power is 3.1@xmath1410@xmath79 @xmath16 @xmath17 for a2597 and 2.1@xmath1410@xmath80 @xmath16 @xmath17 for s159 ( * ? ? ?
* integrating the radio spectrum between 10 @xmath112 and 10 @xmath13 ) .
our observations of the molecular and ionised gas in a2597 and s159 do not have the spatial resolution to investigate the detailed sub-@xmath0 scale correspondence between the agn outflows and the gas . on @xmath0 scales we note small enhancements in the hii and h@xmath10 intensity along the lower parts of the projected radio lobes in a2597 .
most notably along the northern lobe .
d00 , using higher resolution hst imaging , have previously shown this in more detail for a2597 . similarly in s159 we observe that on @xmath0-scales some of the gas appears to lie along the radio lobes . in neither s159 and a2597
is there a clear correlation between the radio emission and the cold gas .
it has been postulated that gas observed on the edges of radio lobes is gas that was uplifted and deposited here through agn outflows @xcite .
another possibility in which the agn outflows may create the observed structures is via pressure driven compression of the in - situ thermal gas by the outward expanding non - thermal plasma . in this latter scenario the pressure driven
compression increases the local gas density leading to an increase in the cooling rate here . that the gas observed along the radio lobes is a purely projectional effect and has nothing to do with the radio outflow seems unlikely .
strong evidence that an agn interacts with the gas in the nuclear region of a bcg comes , for example , from the flux enhancement and high velocity dispersion structure observed in a2597 here and in d00 .
these structures coincide with the current radio outflows .
many other observations , mostly done in x - rays , have shown that there is a global correlation between the radio outflows in bcgs and the hot x - ray gas surrounding it ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we do caution the reader that even though this global correlation is found in many cool core clusters that there are also cases where one observes similar amounts of gas in structures that appear to have no relation at all to the current radio outflows .
one example of this is the western filament in s159 .
in a2597 we note that , in projection , there is a rough alignment between the radio axis and the axis of rotation for the hii and h@xmath10 gas .
the same seems to be true for s159 although the evidence for gas rotation is less clear here .
three other clusters , abell 1664 , pks 0745 - 19 and a2204 , have recently been observed with sinfoni by @xcite .
they show that the hii and h@xmath10 gas in pks 0745 - 19 rotates .
high resolution radio images by @xcite show us that again that the radio axis and gas rotation axis roughly agree .
for the other two clusters the situation is unclear .
the velocity fields of the gas in abell 1664 and abell 2204 do not show clear evidence for gas rotation , and radio images by @xcite and @xcite do not allow for a reliable identification of the jet axis .
higher spatial resolution spectroscopy and radio imaging will be required to investigate the possible alignment of the radio axis and the gas rotation axis in more detail .
it has been proposed in the literature that agn feedback in the form of radio outflows , jets could deliver the required heat that keeps the gas in cluster cores from cooling ( e.g. , * ? ? ?
* ; * ? ? ?
the kinetic power in the molecular and ionised phases may be a useful indicator for estimating the total energy input into the cluster medium by the current jet . here
we will estimate the kinetic power from the kinetic energy of the disturbed gas and its turbulent dissipation .
the kinetic energy @xmath113 and kinetic ( turbulent ) power @xmath114 may be written as ; + + @xmath115 + + here @xmath116 = @xmath117 with the @xmath118 the velocity dispersion of the gas and @xmath119 is the thickness of the high dispersion features .
we will perform this estimate using region a1 in a2597 .
this region contains most of the gas whose dispersion is clearly related to the agn as decribed in section [ sect_a2597_kinm ] .
the mass of the hii gas in a1 is 4.1@xmath1410@xmath66 @xmath6 and its velocity dispersion is 256 @xmath19 @xmath17 .
we use only the hii gas here as this gas component dominates in mass over the warm h@xmath10 gas .
we thus find that a total kinetic energy of @xmath113(a2597:a1 ) = 8.1@xmath1410@xmath120 @xmath16 and a kinetic energy power @xmath114(a2597:a1 ) = 5.8@xmath1410@xmath121 @xmath16 @xmath17 , where we used @xmath119 = 2 @xmath0 . for s159
this calculation is more uncertain due to the lower significance of the pa @xmath11 line in the nuclear region b1 .
if we use that the mass of the hii gas in b1 is 5.8@xmath1410@xmath42 @xmath6 and its velocity dispersion is 127 @xmath19 @xmath17 . we then find that a total kinetic energy of @xmath113(s159:b1 ) = 2.8@xmath1410@xmath122 @xmath16 and a kinetic power @xmath114(s159:b1 ) = 1.0@xmath1410@xmath123 @xmath16 @xmath17 is injected into stirring up the gas , again using @xmath119 = 2 @xmath0 .
these numbers are rather low for s159 .
if we use the h@xmath10 1 - 0 s(3 ) velocity dispersion for the pa @xmath11 line and increase the pa @xmath11 flux such that the pa @xmath11/h@xmath10 1 - 0 s(3 ) ratio is the same as in region a1 for a2597 we find that @xmath113 and @xmath114 both increase by about a factor of 10 for s159 .
we compare this mechanical power injected into the cool gas to the current radio power .
we assume that the stirring of the gas in the nuclei of a2597 and s159 is related to the most recent agn outburst and thus we look only at the recent radio emission .
this is done to avoid the older radio plasma , from previous outbursts , which will dominate the radio power due to its emissivity at low frequencies .
the 8.4 @xmath13 radio emission is a good indicator of recent agn activity .
the synchrotron lifetime of electrons emitting at 8.4 @xmath13 is about 10@xmath124 years , for a magnetic field strength of 30 @xmath125 @xcite .
the 8.4 @xmath13 specific luminosities @xmath126 in regions a1 and b1 are 1.75@xmath1410@xmath15 @xmath16 @xmath17 @xmath18 and 3.88@xmath1410@xmath127 @xmath16 @xmath17 @xmath18 .
one can then estimate the total current radio power @xmath128 in these regions by @xmath128 = @xmath129 , with @xmath57 = 8.4 @xmath13 .
we find @xmath128(a2597:a1 ) = 1.5@xmath1410@xmath80 @xmath16 @xmath17 and @xmath128(s159:b1 ) = 3.3@xmath1410@xmath130 @xmath16 @xmath17 .
the kinetic power inserted into the cool gas is thus smaller than the radio power in a2597 by a factor 2 - 3 .
for s159 we find that @xmath114 is smaller than @xmath128 by more than an order of magnitude .
if we allow for the higher dispersion and flux as mentioned above then this ratio becomes similar to that observed in a2597 .
adding the warm ( @xmath131 2300 @xmath3 ) h@xmath10 gas to the above analysis does not significantly change the derived kinetic powers due to its low mass .
we thus find that within the nuclear regions the current radio jets in a2597 and s159 are radiating efficiently with respect to the warm hii and h@xmath10 gas .
the kinetic power deduced here should though be seen as a lower limit to the total kinetic power of the cool ( @xmath13210@xmath42 @xmath3 ) gas in these clusters since the colder gas phases have not been taken into account .
o94 find a kinetic energy @xmath13310@xmath134 @xmath16 and a kinetic power @xmath13510@xmath79 @xmath136 for the hi gas in a region with radius 2.5 @xmath0 centered on the bcg nucleus in a2597 .
similarly the cold
( @xmath13210@xmath62 @xmath3 ) molecular gas may contain a significant amount of kinetic energy .
studies of x - ray cavities find evidence for inefficiently radiating jets in that the mechanical power of the jet often can be orders of magnitude larger than the radio power .
however , when looking at x - ray cavities one may be looking back in time towards an epoch in which the balance between the mechanical and radiative power of the radio source is different from the current one .
the argument that agn feedback is responsible for reheating the icm gas in cool core clusters is made purely on the global energetics of the problem and does not consider the microphysics of the energy transfer between the non - thermal radio plasma and thermal icm plasma , nor does it consider the various scales involved in this energy transfer .
fast shocks ( @xmath116 @xmath137 40 @xmath19 @xmath17 ) can be ruled out based on the absence of typical shock tracers like the [ o iii ] 4363 @xmath138 line in optical spectra ( vd97 ) and the high h@xmath10/hii ratios @xcite . however , another way of delivering the required heat for the cooler gas observed here is via low velocity shocks in a dense gas .
such a model was recently invoked to explain the strong h@xmath10 emission observed in stephan s quintet @xcite . from x - ray observations ( e.g. * ? ? ?
* ) and jet models ( * ? ? ?
* e.g. ) it is found that agn outflows may give rise to low velocity c - type shocks through a turbulent cascade of the expanding jet momentum in a inhomogeneous gas .
we have investigated low velocity c - type shocks by @xcite and @xcite to model the molecular ratios in region a1 for a2597 .
the following parameter range was allowed for , shock velocity @xmath139 = [ 15 - 30 ] @xmath19 @xmath17 , pre - shock density @xmath140 = [ 10@xmath42 - 10@xmath141 @xmath41 and magnetic field strength @xmath142 = [ 0.05 - 30 ] @xmath143 .
the best - fitting model to the molecular line ratios is @xmath139 = 22 @xmath19 @xmath17 , @xmath140 = 5@xmath1410@xmath66 @xmath41 and @xmath142 = 5.6 @xmath143 .
this model is shown by the purple line in fig .
[ fig_a2597_lte_shockonly ] .
it implies a pre - shock density above the critical density and thus that the gas is already in lte prior to the shock .
this model furthermore implies a very high magnetic field in the molecular gas , something which has not yet been observed in our objects .
warm molecular gas is observed in the central 20 @xmath0 of the a2597 and s159 galaxy clusters .
all of the h@xmath10 1 - 0 s - transitions and h@xmath10 2 - 1 s - transitions redshifted into the k - band and none of the higher ro - vibrational h@xmath10 transitions are detected in a2597 .
ionised gas is observed in a2597 through the detection of the pa @xmath11 , br @xmath38 , br @xmath37 and fe ii ( 1.81 @xmath30 ) lines . for s159 only two h@xmath10 1 - 0
s - transitions are detected .
there is tentative evidence for the presence of the other h@xmath10 1 - 0 s - transitions , but these are strongly affected by atmospheric effects .
none of the higher ro - vibrational h@xmath10 lines were detected .
ionised gas is observed in s159 through the detection of the pa @xmath11 line .
higher ionisation lines , such as [ si iv ] , are not detected in either s159 or a2597 .
molecular and ionised gas is detected out to 22.5 @xmath0 from the nucleus of pgc071390 in a2597 and 18.0 @xmath0 from the nucleus of eso291-g009 in s159 .
the fields observed by us do not cover the entire extent over which molecular and ionised emission has previously been detected in these clusters .
our observations thus miss a significant fraction of the total emission and gas mass .
based on the h@xmath11 maps in j05 we estimate that we have observed about one tenth of the total hii and h@xmath10 gas present in a2597 and s159 .
the gas morphology in a2597 is more diffuse than in s159 . in both clusters the ionised and molecular gas
is observed to be distributed in clumps and filamentary structures surrounding their bcgs .
this observation brings these higher redshift cool core clusters closer to the situation observed in more nearby clusters , such as perseus and centaurus , that show a wealth of small scale filamentary structure ( e.g. * ? ? ?
* ; * ? ? ?
in both a2597 and s159 the ionised and molecular gas phases trace each other closely on @xmath0-scales , in both extent and dynamical state . the h@xmath10 line ratios and the h@xmath10 to hii line ratios only vary slowly as a function of position throughout the investigated regions , with the exception of the nuclear regions .
this implies that there is a tight coupling between these gas phases and that there may be a single excitation mechanism responsible for the observed emission . in both clusters
the pa @xmath11 to h@xmath10 ratio is found to be lower within their nuclear regions than outside . in a2597
this decrease is small as the pa @xmath11/h@xmath10 1 - 0 s(3 ) ratio decreases from [email protected] in region a2 to [email protected] in the nuclear region a1 . in s159
the decrease is much larger and we find that the pa @xmath11/h@xmath10 1 - 0 s(3 ) ratio decreases from [email protected] in region b3 to [email protected] in the nuclear region b1 .
this may imply that the physical conditions differ within the nuclear regions as compared to further out .
the velocity structure of the ionised and molecular gas within the nuclear regions of a2597 and s159 is consistent with gas rotating around the nucleus .
the data suggests a possible alignment between the axis of rotation and the radio axis .
further high spatial resolution , spectroscopic observations are required to disentangle this rotation from velocity gradients observed along the filamentary gas structures and to confirm the axis of rotation .
the high velocity dispersion of the gas within the nuclear regions clearly shows that the agns harboured by these bcgs interact with the molecular and ionised gas situated here .
this is especially obvious in a2597 where the observed increase in the velocity dispersion is suggestive of a turbulent wake marking the passage of the radio lobes through the surrouding medium , or of direct acceleration of the dense material by the agn jet .
we estimate that the mechanical power of the agn flow is somewhat smaller than its radiative radio power within the nuclear regions of a2597 and s159 with respect to the ionised and warm molecular gas . if we were to include the mechanical energy contained within the colder atomic and molecular gas phases than this situation will reverse .
the velocities observed for the filamentary structures are less than 150 @xmath19 @xmath17 .
this does not agree with freely infalling gas which should have much higher velocities , i.e. @xmath2600 @xmath19 @xmath17 .
the filaments do show smooth velocity gradients .
possible explanations for this are ; ( i ) we see a single filament that is being stretched and bended by its proper motion and gravitational forces , ( ii ) we see a blend of a number of unresolved filaments along the line of sight . from observations of nearby clusters such as
perseus the latter explanation appears most likely ( e.g. * ? ? ?
* ; * ? ? ?
* ) . the velocity dispersion of the ionised and molecular gas decreases with distance to the nucleus .
it drops below 100 @xmath19 @xmath17 at distances of a few @xmath0 from the nucleus which implies that the gas is kinematically cold outside the nuclear region .
the decrease in velocity dispersion is strongest within the first @xmath22 @xmath0 from the nucleus and slows down after that .
this point may mark a transition in where the agn is able to strongly interact with its surrounding medium .
the low velocity and velocity dispersion of the gas in the filamentary structures indicates that the gas here resides in well defined structures .
the gas may still be clumpy but individual clumps must have a spread in velocity no larger than the derived dispersion .
it also implies that these structures need some form of support , which can not be kinematical , to keep them from plummeting towards the potential centre . a possible source of support are magnetic fields as suggested by @xcite .
this requires magnetic field strengths of a few hundred @xmath108 g .
so far there is no observational evidence for the existence of these magnetic fields in galaxy clusters .
the molecular and ionised gas emission observed in a2597 and s159 does not match the observed x - ray emission very well . in s159
there is no match in the observed brightness profiles . in a2597
there is a rough global agreement in the peak emission but the detailed x - ray brightness profiles are significantly shallower than observed for the h@xmath10 and hii gas .
there is no evidence for a strong correlation between either h@xmath10 or hii and the radio emission in a2597 and s159 .
if the agn is main source of heating and excitation for these gas phases , then we would expect to find this gas near the current agn outflows at all times .
this is especially important for the h@xmath10 gas since it has a cooling time of order years .
one prime example where we find h@xmath10 gas that appears to have no relation to the current outflows is the western filament in s159 .
the emission line spectra for all regions in s159 are similar .
this then implies that the physical proces that is heating the gas in these regions is similar and likely not directly related to the current agn outflows .
in a2597 the h@xmath10 line emitting gas is well described by an lte model with a single excitation temperature @xmath1 .
we find that @xmath1 is about 2300 k for the different regions investigated .
we find slightly higher excitation temperatures for the molecular gas in the filaments as compared to the nuclear region .
in s159 we have only been able to reliably measure two h@xmath10 lines .
the temperature and excitation of the h@xmath10 gas in this cluster thus remains largely unconstrained . assuming a single temperature lte model , we find that @xmath1 is about 2300 k for the different regions investigated . the temperature derived in this way for the nuclear region in s159 is above the dissociation temperature of h@xmath10 and considered unphysical .
no clear trends in temperature with distance to the nucleus are observed for s159 .
if the molecular gas in a2597 and s159 is in lte then this implies that the h@xmath10 gas is not in pressure equilibrium with the hii gas , as has previously been pointed out by j05 .
we find that @xmath107 @xmath2 10@xmath40 in a2597 and s159 .
we note that @xmath93 is likely to be seriously underestimated since only the k - band h@xmath10 lines tracing the warm molecular gas have been considered here . for a region with radius 2.5 @xmath0 centered on the nucleus of the a2597 bcg we can make an inventory of the cool gas .
in agreement with @xcite we find an ionised gas mass @xmath1446@xmath1410@xmath66 @xmath6 .
we find a warm molecular gas mass @xmath1453@xmath1410@xmath42 @xmath6 .
@xcite find an neutral hydrogen mass @xmath14610@xmath5 @xmath6 .
using much lower spatial resolution @xcite find a spatially unresolved cold molecular gas mass @xmath147=8@xmath1410@xmath73 @xmath6 in the central 22@xmath28 of a2597 .
we are able to fit the h@xmath10 line emitting gas by a low velocity shock model .
such a model requires high pre - shock densities and very high magnetic fields .
low velocity shocks may be a viable option for heating the h@xmath10 gas in cool core clusters , but this requires further modelling and observations .
using recent spitzer observations we find that another lower temperature h@xmath10 gas component exists at about 300 k in a2597 and s159 ( jaffe & bremer in prep . ) .
this shows that a single temperature lte model is not a good description for all of the h@xmath10 gas present in cool core clusters and again brings the situation in our clusters closer to that observed in more nearby cool core clusters @xcite .
this is also implies that we significantly underestimate the total h@xmath10 luminosity and gas mass by using only the h@xmath10 k - band spectrum .
our k - band spectra of cool core clusters do not resemble typical k - band spectra of other line emitting , extragalactic objects such as ultraluminous infrared galaxies ( ulirg ) and agns that are not in bcgs .
for comparisons we can look at the br @xmath38 and h@xmath10 1 - 0 s(1 ) line . in ulirgs theses line have a similar intensity and width and in agns that are not in bcgs the br @xmath37 line usually has a higher intensity and larger width than the h@xmath10 1 - 0 s(1 ) line @xcite . in our spectra for abell 2597
we find that these lines have similar widths , but the h@xmath10 1 - 0 s(1 ) line is more than ten times brighter than the br @xmath38 line .
above we have presented the first k - band integral - field spectroscopic observations of the extended molecular and ionised gas distributions in the core regions of the a2597 and s159 galaxy clusters .
these observations add one extra dimension to the analysis of the molecular and ionised gas in these clusters as compared to our previous investigation in j05 .
this allows us to study the distribution , kinematics and thermal structure of this gas over a much larger area than in our previous investigation .
the spatial resolution of our observations limit us to studying this gas on @xmath0-scales .
below we summarise our conclusions .
* line emission from molecular and ionised gas is found in all observed fields .
we confirm the conclusion in j05 that molecular and ionised gas are tightly coupled out to 20 @xmath0 from the bcg nucleus in a2597 and s159 . *
the molecular and ionised gas is distributed in filamentary structures surrounding the bcg .
the gas morphology is more diffuse in a2597 than in s159 .
* in all regions where we have sufficient signal to noise we find that the h@xmath10 gas can be described an by an lte model with a temperature @xmath4 @xmath2 2300 @xmath3 .
lte implies a high - density for this gas . *
kinematically there is a fairly clear separation between the central few @xmath0 , energetically dominated by the agn , and an area outside where the support of the hii and h@xmath10 gas remains to be explained . * within the nuclear region of a2597 and
s159 the kinetic luminosity of the hii and warm h@xmath10 gas is somewhat smaller than the current radio luminosity . *
the very high dispersion regions observed in h@xmath10 and hii may represent a turbulent wake of the relative motion of galaxy and surrounding medium , or the interaction of the current jet with this medium .
* a2597 shows a better correlation between the morphology of the cool gas and x - ray emission than s159 .
the detailed radial brightness profiles of the x - ray and either the h@xmath10 or hii gas do not match in either cluster .
* the high frequency radio emission and cool gas have the same overall scale size .
there is no strong correlation between either the h@xmath10 or hii gas and the radio structures , but there are weak enhancements in the surface brightness of this gas along the northern radio lobe in a2597 . * the data suggests that the hii and h@xmath10 gas situated within a few @xmath0 from the nucleus rotates around an axis parallel to the radio jet , this needs further confirmation .
a more detailed investigation into the heating and cooling of the intracluster medium will need to take into account the microphysics of the interaction between the various phases situated within the intracluster medium ; the hot x - ray gas , the cool hii , h@xmath10 gas and the radio emitting plasma .
once such a model is in place , the next step will be to compare detailed maps of local heating versus local cooling in cool core clusters .
we will follow this work up in the near future with optical integral field spectroscopy ( oonk et al . in prep . ) and spitzer irs spectroscopy ( jaffe & bremer in prep . ) .
a more detailed investigation of the excitation mechanisms for the molecular hydrogen gas will also be presented in a future paper .
jbro wishes to thank j. reunanen and l. vermaas for useful discussions regarding the reduction of the sinfoni data .
we would like to thank lars kristensen for use of his large grid of 1d shock models the sinfoni observations were taken at the very large telescope ( vlt ) facility of the european southern observatory ( eso ) as part of project 075.a-0074 .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc
. 99 baars j. w. m. , genzel r. , pauliny - toth i. i. k. , witzel a. , a&a , 1977 , 61 , 99 baker k. , thesis , 2005 , bristol university .
baldwin j. a. , phillips m. m. , terlevich r. , 1981 , pasp , 93 , 5 baum s. a. , odea c. p. , 1991 ,
mnras , 250 , 73 birzan l. , rafferty d. a. , mcnamara b. r. , wise m. w. , nulsen p. e. j. , 2004 , apj , 607 , 800 birzan l. , mcnamara b. r. , nulsen p. e. j. , carilli c. l. , wise m. w. , 2008 , apj , 686 , 859 black j. h. , van dishoeck e. f. , 1987 , apj , 322 , 412 blanton e. l. , sarazin c. l. , mcnamara , b. r. , 2003 , apj , 585 , 227 bonnet h. , abuter r. , baker a. , bornemann w. , brown a. , castillo r. , conzelmann r. , damster r. , davies r. , delabre b. , donaldson r. , dumas c. , eisenhauer f. , elswijk e. , fedrigo e. , finger g. , gemperlein h. , genzel r. , gilbert a. , gillet g. , goldbrunner a. , horrobin m. , ter horst r. , huber s. , hubin n. , iserlohe c. , kaufer a. , kissler - patig m. , kragt j. , kroes g. , lehnert m. , lieb w. , liske j. , lizon j. , lutz d. , modigliani a. , monnet g. , nesvadba n. , patig j. , pragt j. , reunanen j. , rhrle c. , rossi s. , schmutzer r. , schoenmaker t. , schreiber j. , stroebele s. , szeifert t. , tacconi l. , tecza m. , thatte n. , tordo s. , van der werf p. , weisz h. , 2004 , the eso messenger , 117 , 17 churazov e. , brueggen m. , kaiser c. r. , boehringer h. , forman w. , 2001 , apj , 554 , 261 clarke t. e .. , sarazin c. l. , blanton e. l. , neumann d. m. , kassim n. e. , 2005 , apj , 625 , 748 crawford c. s. , fabian a. c. , 1992 , mnras , 259 , 265 crawford c. s. , allen s. w. , ebeling h. , edge a. c. , fabian a. c. , 1999 , mnras , 306 , 857 crawford c. s. , hatch n. a. , fabian a. c. , sanders j. s. , 2005 , mnras , 363 , 216 dalla vecchia c. , bower r. g. , theuns t. , balogh m. l. , mazzotta p. , frenk c. s. , 2004 , mnras , 355 , 995 davies r. i. , sternberg a. , lehnert m. , tacconi - garman l. e. , 2003 , apj , 597 , 907 davies r. i. , sternberg a. , lehnert m. d. , tacconi - garman l. e. , 2005 , apj , 633 , 105 devillard n. , 1997 , the eso messenger , 87 devillard n. , 2001 , asp conf .
ser . , 238 , 525 de grandi s. , bhringer h. , guzzo l. , molendi s. , chincarini g. , collins c. , cruddace r. , neumann d. , schindler s. , schuecker p. , voges w. , 1999 , apj , 514 , 148 donahue m. , mack j. , voit g. m. , sparks w. , elston r. , maloney p. r. , 2000 , apj , 545 , 670 draine b. t. , woods d. t. , 1992 , apj , 363 , 464 dunn r. j. h. , fabian a. c. , mnras , 373 , 959 edge a. c. , 2001 , mnras , 328 , 762 edge a. c. , wilman r. j. , johnstone r. m. , crawford c. s. , fabian a. c. , allen s. w. , 2002 , et al . , mnras , 337 , 49 eisenhauer f. , abuter r. , bickert k. , biancat - marchet f. , bonnet h. , brynnel j. , conzelmann , r. d. , delabre b. , donaldson r. , farinato j. , fedrigo e. , genzel r. , hubin n. n. , iserlohe c. , kasper m. e. , kissler - patig m. , monnet g. j. , roehrle c. , schreiber j. , stroebele s. , tecza m. , thatte n. a. , weisz h. , 2003 , spie , 4841 , 1548 fabian a. c. , nulsen p. e. j. , stewart g. c. , ku w. h. m. , malin d. f. , mushotzky r. f. , 1981 , mnras , 196 , 35 fabian a. c. , crawford c. s. , edge a. c. , mushotzky r. f. , 1994 , mnras , 267 , 779 fabian a. c. , celotti a. , blundell k. m. , kassim n. e. , perley r. a. , mnras , 331 , 369 fabian a. c. , sanders j. s. , taylor g. b. , allen s. w. , crawford c. s. , johnstone r. m. , iwasawa k. , 2006 , mnras , 366 , 417 fabian a. c. , johnstone r. m. , sanders j. s. , conselice c. j. , crawford c. s. , gallagher j. s. , zweibel e. , 2008 , nat .
, 454 , 968 falcke h. , rieke m. j. , rieke g. h. , simpson c. , wilson a. s. , 1998 , apj , 494 , 155 ferland g. j. , fabian a. c. , hatch n. a. , johnstone r. m. , porter r. l. , van hoof p. a. m. , williams r. j. r. , 2009 , mnras , 392 , 1475 flower d. r. , pineau des forts g. , 2003 , mnras , 343 , 390 hansen l. , joergensen h. e. , noergaard - nielsen h. u. , pedersen k. , goudfrooij p. , linden - voernle m. j. d. , 2000 , a&a , 356 , 83 hatch n. a. , crawford c. s. , fabian a. c. , johnstone r. m. , 2005 , mnras , 358 , 765 hatch n. a. , crawford c. s. , fabian a. c. , 2007 , mnras , 380 , 33 heckman t. m. , baum s. a. , van breugel w. j. , mccarthy , p. , 1989 ,
apj , 338 , 48 hollenbach d. , mckee c. f. , 1989 , apj , 343 , 306 govoni f. , feretti l.,giovannini g. , boehringer h. , reiprich t. h. , murgia m. , 2001 , a&a , 376 , 803 govoni f. , feretti l. , 2004 , ijmpd , 8 , 1549 guillard p. , boulanger f. , pineau des forts g. , appleton p. n. , 2009 , a&a , 502 , 515 irwin j. a. , stil j. m. , bridges t. j. , 2001 , mnras , 328 , 359 jaffe w. , bremer m.n . , 1997 , mnras , 284 , 1 jaffe w. , bremer m.n .
, van der werf p.p . , 2001 ,
mnras , 324 , 443 jaffe w. , bremer m.n . , baker k. , 2005 , mnras , 360 , 748 johnstone r. m. , fabian a. c. , 1988 , mnras , 233 , 581 johnstone r. m. , hatch n. a. , ferland g. j. , fabian a. c. , crawford c. s. , wilman r. j. , 2007 , mnras , 382 , 1246 kaastra j. s. , den boggende a. j. , brinkman a. c. , ferrigno c. , den herder j. w. , mewe r. , tamura t. , de vries c. , cottam j. , kahn s. m. , paerels f. b. s. , peterson j. m. , rasmussen a. p. , branduardi - raymont g. , sakelliou i. , audard m. , gdel m. , erd c. , 2001 , aspc , 234 , 351 koekemoer a. m. , odea c. p. , sarazin c. l. , mcnamara b. r. , donahue m. , voit g. m. , baum s. a. , gallimore j. f. , 1999 , apj , 525 , 621 kristensen l. e. , ravkilde t. l. , field d. , lemaire j. l. , pineau des forts g. , 2007 , a&a , 469 , 561 lepp s. , mccray r. , , 1983 , apj , 296 , 560 maia m. a. g. , da costa l. n. , willmer c. , pellegrini p. s. , rite c. , 1987 , aj , 93 , 546 maloney p. r. , hollenbach d. j. , tielens a. g. g. m. , 1996 , apj , 466 , 561 markwardt c. b. , 2009 , aspc , 411 , 251 mcnamara b. r. , oconnell r. w. , 1992 , apj , 393 , 579 mcnamara b. r. , wise m. w. , nulsen p. e. j. , david l. p. , carilli c. l. , sarazin c. l. , odea c. p. , houck j. , donahue m. , baum s. , voit m. , oconnell r. w. , koekemoer a. , 2001 , apj , 562 , 149 neumayer n. , cappellari m. , reunanen j. , rix h. w. , van der werf p. p. , de zeeuw p. t. , davies , r. i. , 2007 , apj , 671 , 1329 odea c. p. , baum s. a. , gallimore j. f. , 1994 , apj , 436 , 669 odea c. p. , baum s. a. , mack j. , koekemoer a. m. , laor a. , 2004 , apj , 612 , 131 odea c. p. , baum s. a. , privon g. , noel - storr j. , quillen a. c. , zufelt n. , park , j. , edge a. , russell h. , fabian a , c. , donahue m. , sarazin c. l. , mcnamara b. , bregman j. n. , egami e. , 2008 , apj , 681 , 1035 osterbrock d. e. , ferland g. j. , 2006 , astrophysics of gaseous nebulae and active galactic nuclei , university science books . peterson j. r. , fabian a. c. , 2006 , phr , 427 , 1 peres c. b. , fabian a. c. , edge a. c. , allen s. w. , johnstone r. m. , white d. a. , 1998 , mnras , 298 , 416 rafferty d. a. , mcnamara b. r. , nulsen p. e. j. , wise m. w. , 2006 , apj , 652 , 216 rousselot p. , lidman c. , cuby j. g. , moreels g. , monnet g. , 1999 , a&a , 354 , 1134 salome p. , combes f. , 2003 , a&a , 412 , 657 sanders j. s. , fabian a. c. , taylor g. b. , 2009 , mnras , 393 , 71 sarazin c. l. , burns j. o. , roettiger k. , mcnamara b. , 1995 , apj , 447 , 559 scoville n. z. , hall d. n. b. , ridgway s. t. , kleinmann s. g. , apj , 253 , 136 shull j. m. , beckwith s. , a&ar , 20 , 163 sternberg a. , dalgarno a. , 1989 , apj , 338 , 197 storchi - bergmann t. , mcgregor p. j. , riffel r. a. , simoes lopes r. , beck t. , dopita m. , 2009 , mnras , 394 , 1148 sutherland r. s. , bicknell g. v. , 2007 , apjs , 173 , 37 taylor g. b. , odea c. p. , peck a. b. , koekemoer a. m. , 1999 , apj , 512 , 27 taylor g. b. , fabian a. c. , gentile g. , allen s. w. , crawford c. , sanders j. s. , 2007 , mnras , 382 , 67 voit g. m. , donahue m. , 1997 , apj , 486 , 242 wilman r. j. , edge a. c. , johnstone r. m. , fabian a. c. , allen s. w. , crawford c. s. , 2002 , mnras , 337 , 63 wilman r. j. , edge a. c. , johnstone r. m. , 2005 , mnras , 371 , 755 wilman r. j. , edge a. c. , swinbank a. m. , 2006 , mnras , 359 , 93 wilman r. j. , edge a. c. , swinbank a. m. , 2009 , mnras , 395 , 1355 quillen a. c. , zufelt n. , park j. , odea c. p. , baum s. a. , privon g. , noel - storr j. , edge a. , russell h. , fabian a. , donahue m. , bregman j. n. , mcnamara b. r. , sarazin c. l. , 2008 , apjs , 176 , 39 [ cols="<,<,<,<,<,<,<",options="header " , ]
[ lastpage ]
|
we have used the sinfoni integral field spectrograph to map the near - infrared k - band emission lines of molecular and ionised hydrogen in the central regions of two cool core galaxy clusters , abell 2597 and sersic 159 - 03 .
gas is detected out to 20 @xmath0 from the nuclei of the brightest cluster galaxies and found to be distributed in clumps and filaments around it .
the ionised and molecular gas phases trace each other closely in extent and dynamical state .
both gas phases show signs of interaction with the active nucleus . within the nuclear regions the kinetic luminosity of this gas
is found to be somewhat smaller than the current radio luminosity . outside the nuclear region
the gas has a low velocity dispersion and shows smooth velocity gradients .
there is no strong correlation between the intensity of the molecular and ionised gas emission and either the radio or x - ray emission . the molecular gas in abell 2597 and sersic 159 - 03 is well described by a gas in local thermal equilibrium ( lte ) with a single excitation temperature @xmath1 @xmath2 2300 @xmath3 .
the emission line ratios do not vary strongly as function of position , with the exception of the nuclear regions where the ionised to molecular gas ratio is found decrease .
these constant line ratios imply a single source of heating and excitation for both gas phases .
[ firstpage ] galaxies : clusters : individual : abell 2597 galaxies : clusters : individual : sersic 159 - 03 cooling flows infrared : ism .
|
at high energy , all the internal timescales of a hadron are time dilated .
therefore , more and more soft fluctuations carrying a smaller and smaller fraction @xmath0 of the hadron momentum become long - lived and become relevant in interactions with another hadron . on the contrary , on the timescales relevant for such an interaction process
, the large @xmath0 partons can be seen as completely frozen degrees of freedom , whose only role is to act as sources that radiate more small @xmath0 gluons .
moreover , the small @xmath0 modes will eventually have an occupation number larger than unity , and undergo recombinations
a process known as _ saturation _
@xcite . in the color glass condensate ( cgc )
framework @xcite , one thus divides the degrees of freedom of a hadron in static color sources
described by a density denoted @xmath1 that represent the large @xmath0 partons , and dynamical gauge fields that represent the small @xmath0 partons .
the cgc can thus be seen as an effective theory of gauge fields coupled to external sources .
the details of this separation of degrees of freedom can change with the separation scale , but this should not affect physical quantities .
this leads to a renormalization group equation known as the jimwlk equation @xcite that governs the evolution with @xmath0 of the distribution @xmath2 $ ] of hard color sources . in these proceedings
, we consider the collision at high energy of two hadrons ( or heavy ions ) described in the cgc framework .
we assume that the distributions of hard color sources that describe the two projectiles are known , and we address the question of calculating physical observables in given configurations of the two sources .
moreover , we will consider only the regime where the two projectiles are saturated , in which the two sources @xmath3 are strong both of order @xmath4 . as stated before
, one must consider an effective theory described by the following lagrangian , @xmath5 with currents given at lowest order in the sources by @xmath6 these currents must be covariantly conserved
@xmath7=0 $ ] which leads to a feedback of the gauge field on the currents . therefore , in general , the eqs .
( [ eq : j ] ) get modified by corrections of higher order in @xmath3 .
an important observation is that the strength of the sources lead to non - perturbative effects , in the sense that an infinite set of diagrams must be summed in order to calculate a quantity at a fixed order in @xmath8 .
however , the large strength of the sources has also a valuable consequence : as we shall see later , the leading order is dominated by tree diagrams only , and it can be studied by classical methods .
following @xcite , let us first consider the theory of a real scalar field coupled to strong sources .
most of the _ structural _ properties we want to discuss can indeed already be studied in this simpler framework .
some general results that are to a large extent independent of the details of the theory under consideration can be obtained by considering the generating function for the probabilities @xmath9 of producing a given number @xmath10 of particles : @xmath11 in @xcite , we have proven that , if we denote @xmath12 the sum of all the cut _ connected _ vacuum - vacuum diagrams represents the natural order of these cut diagrams when the sources are strong . ] with exactly @xmath13 cut lines , the logarithm of @xmath14 reads @xmath15 from eqs .
( [ eq : f - def ] ) and ( [ eq : f1 ] ) , one gets the following expression has recently been proposed in @xcite . ] for the probability @xmath9 : @xmath16 in this formula , the index @xmath17 is the number of connected cut subdiagrams from which the @xmath10 particles are produced : @xmath18 particles are produced by the first of these @xmath17 subdiagrams , @xmath19 by the second one , etc ... if we further expand the prefactor @xmath20 , we obtain : @xmath21 in this expression for @xmath9 , the index @xmath22 is the number of connected subdiagrams that are not cut ( this can be seen from the fact that it comes from the absorptive correction , whose sole role is to preserve unitarity ) .
the term of fixed @xmath17 and @xmath22 in this formula can therefore be interpreted as the probability of producing @xmath10 particles from @xmath23 connected subdiagrams , @xmath17 of which are cut and @xmath22 of which are not cut . by summing this probability over @xmath10 , one `` integrates out '' some degrees of freedom in order to keep only the information about the probability of having @xmath17 cut subdiagrams and @xmath22 that are not cut , regardless of the number of produced particles .
this probability reads : @xmath24 summing this expression over @xmath22 from @xmath25 to @xmath26 , one finally obtains the probability of having @xmath17 cut subdiagrams : @xmath27 one therefore sees that the number of cut subdiagrams has a poissonian distribution , with an average of @xmath28 .
( [ eq : rpq ] ) and ( [ eq : rp ] ) are the essence of the abramovsky - gribov - kancheli cancellations @xcite .
as one can see from the above derivation , they are simply a consequence of the factorization of a generic diagram in terms of its connected subdiagrams . therefore , we expect them to be much more general than the context of reggeons field theories in which they have first been discussed .
another point should also be obvious at this point : in order to obtain the eqs .
( [ eq : rpq ] ) and ( [ eq : rp ] ) , one has `` integrated out '' the number @xmath10 of produced particles . by doing this ,
the infinite sequence @xmath29 has reduced to the single combination @xmath30 .
this means that a lot of the dynamical information about the theory under consideration has been lost in this process , and that there are certain questions that can not be answered anymore by the sole knowledge of the @xmath31 s .
let us now consider the generating function @xmath14 per se , and discuss what it would take to calculate it .
the following results on this question have been established in @xcite : * diagrammatically , @xmath14 is the sum of all the cut vacuum - vacuum diagrams , where each cut propagator is weighted by a factor @xmath32 . * at * leading order * , the derivative of @xmath33 , @xmath34 can be calculated from a pair of solutions @xmath35 of the _ classical equation of motion_. for a scalar field theory with a cubic coupling and a source @xmath36 , the classical eom reads @xmath37 the expression of @xmath34 is simpler if written in terms of the fourier modes @xmath38 and @xmath39 of these classical fields , @xmath40 and reads @xmath41 * the two solutions @xmath42 of the classical eom must obey the following boundary conditions : @xmath43 * from unitarity , @xmath44 .
this property serves as the initial condition for going from @xmath45 to @xmath14 , by writing : @xmath46 unfortunately , finding the pair of solutions of the classical equation of motion that obey the boundary conditions of eqs .
( [ eq : boundary3 ] ) is a difficult numerical problem , that has not yet been studied in this context .
however , as we shall see in the next section , one can obtain from here a formula for the average multiplicity which is much easier to evaluate numerically .
the moments of the distribution of multiplicities , in particular the average multiplicity @xmath47 , enjoy a special status because extra simplifications occur in their calculation .
let us consider the multiplicity since it is the simplest one .
one can get the multiplicity from the generating function as @xmath48 ( the second equality is of course simply due to @xmath49 . ) therefore , calculating @xmath50 is a special case with @xmath51 of the calculation of @xmath34 . at @xmath51 , the third and fourth of the boundary conditions in eqs .
( [ eq : boundary3 ] ) simply reduce to @xmath52 which means that the fields @xmath53 and @xmath54 are equal at large positive times , as well as their first time derivative .
since the classical equation of motion is deterministic , this implies that these two classical fields are equal at all times .
the first two boundary conditions in eqs .
( [ eq : boundary3 ] ) then imply that @xmath42 are vanishing at large negative times , as well as their first time derivative .
therefore , at leading order , the multiplicity @xmath50 is given by eqs .
( [ eq : fourier - phi ] ) and ( [ eq : df1 ] ) with @xmath55 the _ retarded solution of the classical eom with a null initial condition at @xmath56_. one should emphasize the following : it was obvious from the beginning that the multiplicity at leading order would be somehow related to solutions of the classical equation of motion since it involves only tree diagrams at this order but it is a non - trivial result that this is with retarded boundary conditions .
the retarded nature of the boundary condition is crucial in practice , in order to solve this problem numerically .
going back to qcd , it is straightforward to generalize the previous results to the case of gluon production .
the inclusive gluon spectrum is given at leading order by @xmath57 in this formula , @xmath58 is the _ retarded solution of the classical yang - mills equations _ in the presence of the sources @xmath3 and with @xmath59 and @xmath60 at @xmath56 .
this problem was solved numerically in @xcite .
the diagrams that are resummed by eq .
( [ eq : gluon - lo ] ) are represented in figure [ fig : gluons - lo ] , as well as the resulting gluon spectrum .
similarly , the production of quarks at leading order has been considered in @xcite .
note that we use `` leading order '' somewhat abusively here , since quark production is strictly speaking a next - to - leading order effect indeed , in the saturated regime , the number of produced gluons scales like @xmath4 while the number of produced quarks scales like @xmath61 . for quark production at leading order
, one must use the following formula , @xmath62 where @xmath63 is the retarded solution of the dirac equation with the classical field obtained in the calculation of gluon production in the background with a free negative energy spinor , @xmath64 , as the initial condition .
the diagrams involved in eq .
( [ eq : q - lo ] ) are sketched in the left of figure [ fig : quarks - lo ] .
also represented in the right part of this figure is the resulting quark spectrum , for various quark masses .
in fact , the production of quarks is one of the pieces the simplest one that contribute to particle production at nlo , i.e. at order @xmath61 . in @xcite , we have detailed the principles of a full nlo calculation of the particle yield , in the case of scalar fields .
of course , things will be more complicated in qcd with gluons , but one crucial property of the result will survive : _ the calculation of particle production at nlo can be done from retarded solutions of the classical eom and retarded solutions of the eom of a small fluctuation on top of the classical field_. the crucial point again is the fact that these objects are needed with retarded boundary conditions , which means that it is a problem which is tractable numerically in a straightforward way .
two types of topologies , sketched in figure [ fig : gluons - nlo ] , contribute to the gluon multiplicity at nlo @xcite .
one of them ( left diagram ) is very similar to that already encountered in quark production
it corresponds to the production of pairs of gluons , and involves retarded solutions for the equation of motion of small gluonic fluctuations on top of the classical field .
the diagram on the right can be seen as a 1-loop virtual correction to the classical field
the field in the complex conjugate amplitude remaining the tree - level one .
it was shown in @xcite that the latter contribution can also be expressed in terms of retarded solutions of the small fluctuations equation .
therefore , the result according to which the inclusive multiplicity can be calculated from retarded solutions of some equations of motion remain true at nlo . to this diagram with a virtual gluon loop
, one must add two similar diagrams , respectively with a quark loop and a ghost loop ( only in gauges that have ghosts for the latter ) .
one additional issue arises when one considers these one - loop corrections to the gluon yield : some of the contributions have a divergence of the form @xmath65 which is reminiscent of the divergences already resummed by the jimwlk evolution of the distribution of sources @xmath66 $ ] and @xmath67 $ ] for the two projectiles . for the cgc framework to be self - consistent
, one must prove that these divergences that appear in the gluon yield for fixed @xmath68 and @xmath69 can be absorbed in the evolution of the @xmath70 $ ] @xcite .
one can also consider exclusive processes in this framework .
for instance , instead of the plain fully inclusive
probability @xmath9 of producing @xmath10 particles , one may define a probability @xmath71 of producing @xmath10 particles in a certain region @xmath72 of the phase - space _ and none outside of @xmath72 _ , and construct a generating function for these exclusive probabilities , @xmath73 one can show @xcite that this generating function can in principle be calculated at leading order by methods that are similar to those described in section [ sec : gen ] , modulo two differences . * the boundary conditions for the two classical fields in terms of which on can express @xmath74 read : @xmath75 where @xmath76 is a function which is 1 in the region @xmath72 and zero outside . * _ the `` integration constant '' @xmath77 is no longer unity_. in fact , this quantity is the total probability of not producing particles outside of @xmath72 .
it will appear as a prefactor in all the probabilities @xmath71 , and it can therefore be interpreted as a _ survival probability _ for the empty region outside @xmath72 .
multiparticle production in field theories coupled to external time - dependent sources
e.g. in the color glass condensate framework for hadronic collisions at high energy exhibits some non - trivial features when these sources are as strong as the inverse coupling : even in the weak coupling regime , one must resum at each order an infinite set of diagrams . quite generically , one recovers in this type of model the combinatoric relations among the probabilities of cut subdiagrams that lead to the agk cancellations . at leading order
, both the generating function of the probability distribution and the average multiplicity can be calculated from solutions of the classical equation of motion in the presence of the external sources . however , while these solutions must be found with complicated boundary conditions in the case of the generating function , the problem can be reduced to finding retarded solutions in the case of the multiplicity , leading to straightforward algorithms for calculating the multiplicity numerically .
this has been done at leading order in the cgc framework for the production of gluons and quarks in nucleus - nucleus collisions .
the present study can be extended in several directions .
one of these extensions is the production of gluons at nlo ( nlo particle production has in fact been studied in @xcite , but the techniques developed there must be extended to gauge theories ) .
another extension is the study of exclusive reactions , where one enforces some constraints on the phase - space of the produced particles . ,
lett . * 84 * , 4309 ( 2000 ) ; phys .
lett . * 86 * , 1717 ( 2001 ) ; a. krasnitz , y. nara , r. venugopalan , nucl . phys . * a * * 727 * , 427 ( 2003 ) ; a. krasnitz , y. nara , r. venugopalan , phys .
lett . * 87 * , 192302 ( 2001 ) ; t. lappi , phys . rev . * c * * 67 * , 054903 ( 2003 ) .
|
in this talk , we discuss some general properties of particle production in a field theory coupled to strong time dependent sources , and techniques to compute the spectrum of the produced particles in such theories .
we also discuss the application of these results to the description of hadron or heavy ion collisions in the color glass condensate framework . 1
. service de physique thorique ( ura 2306 du cnrs ) + cea / dsm / saclay , 91191 , gif - sur - yvette cedex , france 2 . nuclear theory , physics department + brookhaven national laboratory , upton , ny 11973 , usa
|
the @xmath5-cold - dark - matter ( @xmath5cdm ) model has been consistent with observational data , but the energy scale of dark energy is too low to be compatible with the cosmological constant originated from the vacuum energy in quantum field theory @xcite . since the observations allow the variation of the dark energy equation of state @xcite , many models have been proposed to explain the present accelerated expansion of the universe @xcite . for example , a light scalar field with a slowly varying potential , called quintessence , was introduced as an alternative to the cosmological constant @xcite . in general , however , it is not easy to construct viable particle physics models of quintessence because of an extremely light mass required for the cosmic acceleration today @xcite .
another approach for addressing the dark energy problem is to modify the law of gravity from general relativity at large distances @xcite . in this approach
there have been two main streams .
the first consists of introducing a lagrangian for gravity built up out of the ricci , riemann , and metric tensors , which generally leads to 4-th order differential equations .
the @xmath6 gravity @xcite and the gauss - bonnet gravity @xcite belong to this class .
the second consists of higher dimensional models that realize the cosmic acceleration through the gravitational leakage to extra dimensions .
the dvali - gabadadze - porrati ( dgp ) braneworld model
@xcite belongs to this class ( see refs .
. in general , modified gravity models of dark energy need to be constructed to recover the general relativistic behavior in the regions of high density for the consistency with local gravity experiments . in @xmath6 gravity ,
there have been a number of viable models in which a scalar - field degree of freedom ( `` scalaron '' @xcite ) has a large mass in the region where the ricci scalar @xmath4 is much larger than its cosmological value @xmath7 today @xcite .
provided that the chameleon mechanism @xcite is at work in the local regime , the gravitational coupling with non - relativistic matter can be suppressed to be compatible with solar system experiments @xcite .
there is also another mechanism called the vainshtein screen effect @xcite in which non - linear effects can effectively decouple the scalar field from gravity .
originally the vainshtein mechanism was applied to the theories of massive gravity like fierz - pauli gravity @xcite ( see also refs .
@xcite ) , but the non - linearities imply the presence of a ghost state in such theories @xcite . in the dgp model
non - linear field self - interacting lagrangians such as @xmath8 arise from a brane - bending mode ( i.e. a longitudinal graviton ) @xcite .
this allows the decoupling of @xmath0 from gravitational dynamics in the local region .
unfortunately the self - accelerating solution in the dgp model contains a ghost mode @xcite even in the absence of non - linear terms .
moreover the model is disfavored from the combined data analysis of supernovae ia and baryon acoustic oscillations @xcite . mostly inspired by the dgp model , nicolis _ et al .
@xcite derived the five lagrangians that lead to the field equations invariant under the galilean symmetry @xmath9 in the minkowski space - time [ including the term @xmath8 ] .
the scalar field that respects the galilean symmetry is dubbed `` galileon '' .
each of the five terms only leads to second - order differential equations , keeping the theory free from unstable spin-2 ghost degrees of freedom . if we extend the analysis in ref .
@xcite to the curved space - time , the lagrangians need to be promoted to the covariant forms .
@xcite derived the covariant lagrangians @xmath11 ( @xmath2 ) that keep the field equations up to second - order .
we caution that in the curved space - time the galilean symmetry is in general broken for non - linear field self - interacting terms , but in the minkowski space - time it is preserved for the covariant lagrangians @xmath11
( @xmath2 ) .
the ( modified ) galileon gravity has been extensively applied to cosmology recently @xcite .
one application is to introduce the non - linear field self - interaction of the form @xmath12 in the action of ( generalized ) brans - dicke theories @xcite , where @xmath13 is a function of @xmath0 .
although such a term breaks the galilean symmetry , the field equations remain at second - order . moreover ,
for suitable choices of the function @xmath14 , there exist de sitter ( ds ) solutions responsible for dark energy even in the absence of the field potential .
the presence of the non - linear term also allows the decoupling of the field from gravity in the regions of high density under the vainshtein mechanism .
another application of galileon gravity to cosmology is to study the expansion history of the universe in the presence of the covariant lagrangians @xmath11 ( @xmath2 ) mentioned above .
the cosmology up to the term @xmath15 has been discussed in ref .
@xcite , which showed the existence of stable ds solutions .
recently the full cosmological dynamics including the term @xmath16 have been studied in ref .
the viable model parameter space has been found by studying the conditions for the avoidance of ghosts and laplacian instabilities .
interestingly there exists a tracker solution that finally approaches a stable ds solution .
the equation of state of dark energy exhibits a peculiar phantom - like behavior along the tracker . in this paper we shall study general galileon theories in which the field @xmath0 in the covariant lagrangians @xmath11 ( @xmath2 )
is replaced by general functions @xmath1 .
since @xmath1 are scalar functions , the resulting field equations also remain at second - order .
we constrain the forms of @xmath1 from the requirement to obtain ds solutions .
this constraint gives rise to the galileon theory with @xmath17 as a specific case .
for general functions @xmath1 we also derive the conditions for the avoidance of ghosts and laplacian instabilities .
this is useful to constrain the viable parameter space of those theories
. we shall perform detailed analytic and numerical study for the cosmological dynamics of generalized galileon theory with several different choices of @xmath1 .
in the curved space - time the galilean symmetry is broken even for the lagrangian @xmath18 .
then this symmetry is not restrictive when we study the covariant generalization of the galileon field . on the other hand
, the covariant galileon formalism leads to second - order field equations .
we study a lagrangian that gives second - order equations of motion , such that the theories recover the covariant lagrangian in refs .
@xcite as a specific case .
we will consider two generalizations of the covariant galileon theory : ( i ) scalar couplings with both the ricci scalar @xmath4 and the gauss - bonnet ( gb ) term @xmath19 are introduced , ( ii ) the covariant galileon terms are extended to more general functions .
as for the first point , this step is compatible with the approach of field theory , because such scalar couplings generally exist and they are consistent with general covariance ( and even with the galileon symmetry , as in the minkowski background their contributions to the equations of motion of the field identically vanish ) . moreover the scalar couplings with @xmath4 and @xmath19
give only second - order contributions .
it is true that the gb term can change the ultra - violet behavior for the modes , but this property also holds for all the remaining terms coming from the extended galileon action . as for the second point
, we can replace the scalar field @xmath0 in each lagrangian term with a function of the field itself .
the lagrangian @xmath20 , for example , can be modified to @xmath21 .
the equations still remain at second - order because @xmath22 is a scalar quantity itself .
we will consider this generalization for all the galileon terms , introducing different functions @xmath1 ( @xmath23 ) for each of them .
the covariant action we shall discuss is given by @xmath30 where @xmath31 is a determinant of the space - time metric @xmath32 , @xmath33 is a matter lagrangian , and @xmath34 in order to control the signs of @xmath35 we have introduced the factors @xmath36 , which are either @xmath37 or @xmath38 . for the terms
@xmath39 we get the opposite signs by replacing @xmath1 with @xmath40 . for the matter
lagrangian @xmath33 we consider the contribution of two perfect fluids @xmath41 ( @xmath42 ) , described by the barotropic equations of state of the form @xmath43 ( @xmath42 ) .
note that @xmath44 and @xmath45 are the pressure and the energy density of fluids , with the energy - momentum tensor @xmath46 .
consider the flat friedmann - lemaitre - robertson - walker ( flrw ) space - time with the line element @xmath47 where @xmath48 is the scale factor with cosmic time @xmath49 .
varying the action ( [ action ] ) with respect to @xmath32 , we obtain the following equations of motion @xmath50 = -(w_{1}\rho_{1}+w_{2}\rho_{2})\ , , \label{eq : frd2}\end{aligned}\ ] ] where a dot represents a derivative with respect to @xmath49 .
the matter fluids obey the continuity equations @xmath51 differentiating eq .
( [ eq : frd1 ] ) in terms of @xmath49 and eliminating the terms @xmath52 from eq .
( [ eq : frd2 ] ) , we get the generalized klein - gordon equation for the scalar field . for the perfect fluids we consider radiation ( @xmath53 , @xmath54 ) and non - relativistic matter ( @xmath55 , @xmath56 ) .
let us restrict the functional forms of @xmath1 , @xmath3 , and @xmath14 by demanding the existence of ds solutions responsible for dark energy .
we shall focus on the theories in which the late - time cosmic acceleration can be realized by the field kinetic terms rather than the field potential , so that we set @xmath57 the absence of this term and more in general of a potential for the field can be implemented by invoking an additional shift symmetry @xmath58 for the galileon lagrangian .
the condition ( [ eq : nopotential ] ) is also important for another reason . in the minkowski space - time ( @xmath59 )
the only solution to the equations of motion without matter corresponds to @xmath60 , provided that @xmath61 .
this implies that the field is frozen , and thus implementing the vainshtein mechanism .
moreover the field perturbations would propagate with the speed of light in the minkowski background .
the friedmann equation ( [ eq : frd1 ] ) can be written in the form @xmath62 where @xmath63 , @xmath64 , and @xmath65 in order to realize the late - time ds solutions we take the power - law functions for @xmath3 , @xmath1 , and @xmath14 in terms of @xmath0 .
we can classify the theories into two classes : ( i ) @xmath66 is constant , and ( ii ) @xmath66 depends on the field @xmath0 . at the ds point ( @xmath67constant )
we require that each term in eq .
( [ omecon ] ) does not vary in time . for constant @xmath66 ,
the functions @xmath68 and @xmath13 need to have the field dependence proportional to @xmath0 .
then the ds solutions can be realized for constant @xmath69 . in order to have dimensionless couplings @xmath70 of the orders of unity , we write the functions @xmath66 , @xmath68 , and @xmath13 in the forms @xmath71 where @xmath72gev is the reduced planck mass . together with the dimensionless constants @xmath70 ( @xmath73 and @xmath74 we have introduced the mass scales @xmath75 where we have used @xmath76 .
defining the mass scales @xmath77 and @xmath78 as given above , we can express the density parameter @xmath79 in the form @xmath80 where @xmath81 we shall consider the case in which the dimensional variables @xmath82 and @xmath83 are not much different from the orders of unity today . in the asymptotic past we require that @xmath84 , @xmath85 , and @xmath86 to recover the general relativistic behavior . if the coefficients @xmath70 and @xmath74 are of the orders of unity , the dominant contribution in eq .
( [ omedef ] ) comes from the last term .
in such a case , however , the gb term has been dominant during the whole cosmological evolution by today . in order to avoid this behavior
, we set @xmath87 when we discuss the cosmological dynamics . using eqs .
( [ eq : frd1 ] ) and ( [ eq : frd2 ] ) , we obtain the following relations : @xmath88 where @xmath89 is the value of @xmath82 at the ds point , and @xmath90 we note that the theory with the functions ( [ funchoice ] ) corresponds to the covariant galileon theory discussed in refs . @xcite .
let us consider theories in which @xmath66 depends on the field @xmath0 .
if we take the power - law function of the form @xmath91 with constant @xmath92 , it follows from eq .
( [ omecon ] ) that the function @xmath93 is required to be constant at the ds point . from this demand we can restrict the functions @xmath66 , @xmath68 , @xmath13 in the forms @xmath94 where @xmath70 ( @xmath95 ) , @xmath74 are dimensionless constants , @xmath96 ( @xmath95 ) and @xmath78 are mass scales defined by @xmath97 we note that there are other possibilities to obtain ds solutions , depending on the forms of @xmath3 . if we choose the function @xmath98 , where @xmath99 is a constant , then it is possible to realize the ds solution for the choice @xmath100 , @xmath101 , @xmath102 , and @xmath103 . in the sense that @xmath104constant along the ds solution this theory is related with the constant @xmath66 theory given in ( [ funchoice ] )
since we want to consider the case in which the ds solution is realized in a different way , we shall study the cosmological dynamics for the theories with ( [ funchoice2 ] ) in which @xmath105constant at the de sitter solution . for the theories ( [ funchoice2 ] ) the galileon symmetry is explicitly broken .
however , even for the galileon action ( [ funchoice ] ) , the galileon symmetry is restored only on the minskowski background .
therefore the galileon symmetry does not restrict the form of the lagrangian on curved backgrounds .
nonetheless , these theories may represent an effective action for some more fundamental theory , e.g. , extra - dimensional models .
in fact the covariant galileon is the generalization of the decoupling limit of the dgp braneworld model .
the existence of ds solutions in this theory opens up the possibility of studying these generalizations of the original galileon field .
moreover the model is not plagued by the ostrogradski instability because the field equations remain at second order .
the situation here is not very different from other dark energy models , such as @xmath6 gravity , where the lagrangian is constructed by hands to realize the late - time cosmic acceleration but it is supposed to be originated from some fundamental theory . in this directions other papers appeared which tried to generalize the galileon lagrangian without imposing the original galileon symmetry @xcite , as it is violated , by construction , on curved backgrounds .
the density parameter @xmath79 can be expressed as @xmath106 where @xmath107 for @xmath92 and @xmath70 ( @xmath95 ) of the order of unity , we require that @xmath108 , @xmath109 , and @xmath110 in the asymptotic past . again
we shall set @xmath111 in order to avoid that the last term in eq .
( [ omede2 ] ) always dominates the cosmological dynamics . at the ds point
we have the following relations @xmath112 where @xmath113 in the above two theories we wish to study 1 ) whether different galileon - like actions have some common feature , and 2 ) how they differ .
in order to discuss the stability of theories described by the lagrangian ( [ lag ] ) in the cosmological context , it is necessary to study linear perturbation theory on the flat flrw background .
let us consider the perturbed metric @xmath114\,{\mathrm{d}}t^{2}+\partial_{i}\chi(t,\bm{x})\,{\mathrm{d}}t\ , { \mathrm{d}}x^{i}+a^{2}(t)[1 + 2\phi(t,\bm{x})]\,{\mathrm{d}}\bm{x}^{2}\,,\label{eq : perto1}\ ] ] where @xmath115 , @xmath116 , and @xmath117 are scalar metric perturbations .
we have chosen the gauge @xmath118 without a non - diagonal scalar perturbation in the spatial part of the metric , i.e. @xmath119 ( see refs .
@xcite for the details of gauge - invariant cosmological perturbation theory ) . taking into account two perfect fluids with the equations of state @xmath43 ( @xmath42 ) ,
there are three propagating scalar degrees of freedom .
the velocity potentials @xmath120 ( @xmath42 ) are related with the energy - momentum tensor @xmath121 , as @xmath122 ( @xmath42 ) .
expanding the action ( [ action ] ) at second - order in perturbations , we find that the field @xmath115 can be integrated out together with @xmath117 .
introducing the vector @xmath123 , we obtain the following second - order action for scalar perturbations ( see refs .
@xcite for the details of such analysis ) : @xmath124\,,\label{eq : perto2}\ ] ] where @xmath125 , @xmath126 and @xmath127 are @xmath128 symmetric matrices and @xmath129 is an antisymmetric matrix ( we do not write explicit forms for them ) .
let us consider tensor perturbations with @xmath130 , where @xmath131 is traceless ( @xmath132 ) and divergence - free ( @xmath133 ) .
we also expand the action ( [ action ] ) at second - order in terms of the two polarization modes , @xmath134 , where @xmath135 and @xmath136 are the polarization tensors .
for the polarization mode @xmath137 , the perturbed action is given by @xmath138\,,\label{eq : parto2gw}\ ] ] where we will show the explicit forms of @xmath139 and @xmath140 later .
the conditions for the avoidance of ghosts and laplacian instabilities of tensor perturbations correspond to @xmath141 and @xmath142 , respectively .
note that the same expression also holds for @xmath143 . in secs .
[ subghost ] and [ sublap ] we study the general theories described by the lagrangian ( [ lag ] ) without imposing that @xmath144 and @xmath13 are zero . in sec .
[ subds ] we shall apply our formula to specific theories with @xmath145 and @xmath146 . in order to avoid that a ghost mode appears for scalar perturbations , the matrix @xmath125 needs to be positive definite .
this leads to the following three no - ghost conditions @xmath147 where @xmath148 for radiation ( @xmath53 ) and non - relativistic matter ( @xmath149 ) the conditions ( [ eq : constro1 ] ) and
( [ eq : constro2 ] ) are automatically satisfied .
hence we only need to consider the condition ( [ eq : genqs ] ) to avoid the appearance of ghosts for scalar perturbations .
the no - ghost condition for tensor perturbations is given by @xmath150 the propagation speeds @xmath151 of three scalar degrees of freedom is known by solving the equation @xmath152 the standard velocities for two perfect fluids correspond to @xmath153 and @xmath154 , which are positive for both radiation and non - relativistic matter . the stability condition coming from the third solution
is given by @xmath155 - 4\dot{\gamma}_{1}\gamma_{1}\gamma_{3}+\gamma_{3}^{2}\gamma_{4}}{\gamma_{1}[4(\rho_{1}+\rho_{2}+\gamma_{2})\gamma_{1}-9\gamma_{3}^{2}]}>0\,,\label{cs2}\ ] ] where @xmath156 under the no - ghost condition ( [ eq : genqs ] ) , this translates to @xmath157 - 4\dot{\gamma}_{1}\gamma_{1}\gamma_{3}+\gamma_{3}^{2}\gamma_{4}>0\,.\label{noinst}\ ] ] the laplacian instability of tensor perturbations is absent provided that the propagation speed squared is positive : @xmath158 on the ds background there are no matter fields , so that only one scalar mode propagates .
the second - order action for scalar perturbations at the ds fixed point is @xmath159\,,\label{eq : actdes}\ ] ] where @xmath160 and @xmath161 correspond to those given in eqs .
( [ eq : genqs ] ) and ( [ cs2 ] ) with the limits @xmath162 and @xmath163 .
let us discuss the conditions for the avoidance of ghosts and instabilities on the ds solutions we have discussed in sec .
[ cosmosec ] .
we shall consider two theories described by the functions ( [ funchoice ] ) and ( [ funchoice2 ] ) , with @xmath145 and @xmath146 .
for the theory we discussed in sec .
[ confsec ] , the conditions ( [ eq : genqs ] ) , ( [ cs2 ] ) , ( [ eq : genqw ] ) , and ( [ eq : noinstgw ] ) reduce to @xmath164}>0\,,\label{csds}\\ \frac{q_{t}}{m_{{\rm pl}}^{2 } } & = & \frac{1}{4}\,(2 + 3\alpha-6\beta)>0\,,\label{eq : qcs3}\\ c_{t}^{2 } & = & \frac{2-\alpha}{2 + 3\alpha-6\beta}>0\,.\label{eq : qcs4}\end{aligned}\ ] ] in sec . [ dynamicssec ] we will show the allowed parameter space in the @xmath165 plane after deriving other conditions . from the action ( [ eq : actdes ] ) we obtain the equation for @xmath116 in fourier space : @xmath166 where @xmath167 is a comoving wavenumber .
the solution for the homogeneous perturbation ( @xmath168 ) is @xmath169 where @xmath170 and @xmath171 are integration constants . since @xmath160 is constant on the ds solution , the second term on the right hand side of eq .
( [ phiso ] ) decays with time by noting that the scale factor evolves as @xmath172 . for the same reason ,
the tensor perturbation remains always stable in the limit @xmath173 .
this means that the ds fixed point is always classically stable under homogeneous perturbations .
for the theory we discussed in sec .
[ nonconfsec ] , the conditions ( [ eq : genqw ] ) and ( [ eq : qcs4 ] ) reduce to @xmath174>0\,,\\ c_{t}^{2 } & = & 1-\frac{4[15625\tilde{\alpha}+384\tilde{\beta}(p\tilde{x}_{{\rm ds}}-5)]}{15625(3\tilde{\alpha}+512)-7680\tilde{\beta}}>0\,.\end{aligned}\ ] ] the expressions for @xmath160 and @xmath161 are more involved , but @xmath160 is proportional to @xmath175 as in the case of @xmath139 . integrating the relation @xmath176 , it follows that @xmath177 .
since @xmath178 , we find from eq .
( [ phiso ] ) that the homogenous perturbation @xmath116 evolves as @xmath179 where @xmath180 and @xmath181 are constants .
hence the ds point is classically stable for @xmath182 the stability condition ( [ dsstability ] ) is satisfied for @xmath183 . in this regime
the conditions ( [ eq : genqs ] ) and ( [ cs2 ] ) for the scalar perturbation reduce to @xmath184>0\ , , \label{eq : qc2sx}\\ c_{s}^{2 } & = & -\frac{p\tilde{x}_{{\rm ds}}}{27 } + { \cal o}(\tilde{x}_{{\rm ds}}^{2})>0\ , .
\label{eq : cc2sx}\end{aligned}\ ] ] for positive @xmath0 the no - ghost condition ( [ eq : qc2sx ] ) is satisfied . if @xmath185 , the laplacian instability of the scalar mode can be avoided .
first we study cosmological dynamics for the covariant galileon theory described by the functions ( [ funchoice ] ) with @xmath111 .
this was partially discussed in the letter @xcite , but in this paper we shall thoroughly study the cosmology in such a theory with detailed numerical simulations . in the presence of radiation ( @xmath186 ) and non - relativistic matter ( @xmath187 ) , we obtain the background equations from eqs .
( [ eq : frd1 ] ) and ( [ eq : frd2 ] ) : @xmath188 where @xmath189/(2m^{6})-3d_{5}^{5}h\dot{\phi}^{4}[5h\ddot{\phi}+2(h^{2}+\dot{h})\dot{\phi}]/m^{9}\,.\end{aligned}\ ] ] the continuity equations for radiation and non - relativistic matter are given , respectively , by @xmath190 from eqs .
( [ be1 ] ) , ( [ be2 ] ) , and ( [ be3 ] ) the dark component also obeys the continuity equation @xmath191 we define the dark energy equation of state @xmath192 and the effective equation of state @xmath193 , as @xmath194 where the latter is known by the background expansion history of the universe .
using eq .
( [ rhode ] ) together with the relation @xmath195 , it follows that @xmath196 where a prime represents a derivative with respect to @xmath197 . each term in eq .
( [ omedef ] ) has the difference of the order of @xmath198 .
the highest - order term in @xmath79 comes from the term @xmath16 , i.e. , of the order of @xmath199 . for the dynamical analysis given below ,
it is convenient to introduce the following quantities @xmath200 at the ds fixed point one has @xmath201 and @xmath202 . in terms of @xmath203 and
@xmath204 the density parameter ( [ omecon ] ) can be written as @xmath205 where @xmath206 and @xmath207 are defined in eq .
( [ alphadef ] ) .
here we have employed the relations ( [ epre ] ) and ( [ d3re ] ) to eliminate the terms @xmath208 and @xmath209 .
it is convenient to use the variables @xmath206 and @xmath207 for several reasons .
first , the coefficients of physical quantities [ such as @xmath79 in eq .
( [ omede ] ) ] , autonomous equations , quantities related with no - ghost and stability conditions can be expressed in terms of @xmath206 and @xmath207 .
second , the equations of motion , together with linear perturbation theory , are not subject to change under the following change of parameters @xmath210 and @xmath211 ( with @xmath212 ) , where @xmath213 is a real number . in this case , depending on the parameter @xmath213 , there are infinite choices for the lagrangian coefficients @xmath70 that lead to the identical physics for the same @xmath206 and @xmath207 . therefore , constraining the parameter space in terms of @xmath206 and @xmath207 allows us to remove the arbitrariness of the @xmath213 rescaling .
this also shows that one can set @xmath214 without losing generalities .
if @xmath215 at early times , the highest - order term @xmath16 gives the dominant contribution to the dark energy density @xmath79 . in this case
it is expected that the cosmological vainshtein mechanism can be at work to recover the general relativistic behavior . if @xmath216 initially , the dominant contribution to @xmath79 comes from the term @xmath20 . in this case
the field energy density decreases rapidly as in the standard massless scalar field and hence the solutions do not approach the ds fixed point at late times . the conditions ( [ eq : genqs ] ) , ( [ eq : genqw ] ) , ( [ cs2 ] ) , and ( [ eq : noinstgw ] ) for the avoidance of ghosts and instabilities of scalar and tensor perturbations reduce to @xmath217 - 4\mu_{1}'(1+\mu_{1})(1+\mu_{2})+2(1+\mu_{3})^{2}(1+\mu_{4})}{6(1+\mu_{1})(\mu_{1}+\mu_{2}+\mu_{1}\mu_{2}-2\mu_{3}-\mu_{3}^{2})}>0\,,\\ & & c_{t}^{2}=\frac{2r_{1}(2-\alpha r_{1}r_{2})-3\beta(r_{2}r_{1}'+r_{1}r_{2}')}{2r_{1}(2 + 3\alpha r_{1}r_{2}-6\beta r_{2})}>0\,,\label{st4}\end{aligned}\ ] ] where @xmath218 from eqs .
( [ eq : frd1 ] ) , ( [ eq : frd2 ] ) , and ( [ conser ] ) we obtain the following differential equations for the variables @xmath203 , @xmath204 , and @xmath219 : @xmath220\nonumber \\ & & { } \times\left[2\left(\omega_{r}+9\right)+3r_{2}\left(r_{1}^{3 } ( -3\alpha+4\beta-2)+2r_{1}^{2}(9\alpha-9\beta+2)-15r_{1}\alpha+14\beta\right)\right]\ , , \label{eq : drr1 } \\ \hspace{-0.7cm}r_{2 } ' & = & -\frac{1}{\delta}[r_{2}(6r_{1}^{2}(r_{2}(45\alpha^{2}-4(9\alpha+2)\beta+36\beta^{2})-(\omega_{r}-7)(9\alpha-9\beta+2))+r_{1}^{3}(-2(\omega_{r}+33)(3\alpha-4\beta+2)\nonumber \\ & & -3r_{2}(-2(201\alpha+89)\beta+15\alpha(9\alpha+2)+356\beta^{2}))-3r_{1}\alpha(-28\omega_{r}+123r_{2}\beta+36)+10\beta(-11\omega_{r}+21r_{2}\beta-3)\nonumber \\ & & + 3r_{1}^{4}r_{2}(9\alpha^{2}-30\alpha(4\beta+1)+2(2 - 9\beta)^{2})+3r_{1}^{6}r_{2}(3\alpha-4\beta+2)^{2}+3r_{1}^{5}r_{2}(9\alpha-9\beta+2)(3\alpha-4\beta+2 ) ) ] , \label{eq : drr2 } \\ \hspace{-0.7cm}\omega_{r } ' & = & \frac{2}{\delta}\omega_{r}[r_{1}^{2}(4(\omega_{r}-1)(9\alpha-9\beta+2)+6r_{2}(-15\alpha^{2}+36\alpha\beta+4(2 - 9\beta)\beta))-2r_{1}^{3}((\omega_{r}-1)(3\alpha-4\beta+2)\nonumber \\ & & + 9r_{2}(18(\alpha+1)\beta+\alpha(9\alpha+2)-36\beta^{2}))+12r_{1}\alpha(-3\omega_{r}+22r_{2}\beta+3)-10\beta(-4\omega_{r}+21r_{2}\beta+4)\nonumber \\ & & + r_{1}^{4}r_{2}(549\alpha^{2}+\alpha(330 - 840\beta)+2(2 - 9\beta)^{2})+3r_{1}^{6}r_{2}(3\alpha-4\beta+2)^{2}-12r_{1}^{5}r_{2}(9\alpha-9\beta+2)(3\alpha-4\beta+2 ) ] , \label{eq : drr3}\end{aligned}\ ] ] where @xmath221 + 4r_{1}^{2}[9r_{2}(5\alpha^{2}+9\alpha\beta+(2 - 9\beta)\beta)+2(9\alpha-9\beta+2)]\nonumber \\ & & + 4r_{1}^{3}[-3r_{2}\left(-2(15\alpha+1)\beta+3\alpha(9\alpha+2)+4\beta^{2}\right)-3\alpha+4\beta-2]-24r_{1}\alpha(16r_{2}\beta+3)+10\beta(21r_{2}\beta+8)\,.\end{aligned}\ ] ] the hubble parameter obeys the following equation @xmath222 where @xmath223 and @xmath224 are known from eqs .
( [ eq : drr1 ] ) and ( [ eq : drr2 ] ) . from eq .
( [ eq : drr1 ] ) we find that there is an equilibrium point characterized by @xmath225 at which the density parameter ( [ omede ] ) reduces to @xmath226 from eq .
( [ eq : r1r2 ] ) we find that @xmath227constant along the solution ( [ eq : soldr1 ] ) .
hence the field velocity evolves as @xmath228 which has the dependence @xmath229 during the radiation and matter eras .
since the field is effectively frozen at early times , this shows the implementation of the cosmological vainshtein mechanism . along the solution ( [ eq : soldr1 ] )
, the other two equations can be written as follows @xmath230 which do not depend on @xmath206 and @xmath207 .
we then have the following three fixed points @xmath231 the points ( a ) and ( b ) can be realized during the radiation and matter eras , respectively , whereas the point ( c ) corresponds to the ds solution .
the stabilities of these fixed points can be analyzed by considering linear perturbations @xmath232 , @xmath233 , and @xmath234 about them .
for example , the perturbation @xmath232 satisfies @xmath235 this shows that , in the regime @xmath236 and @xmath237 , the solution is stable in the direction of @xmath203 . defining the vector @xmath238 ,
one can write the perturbation equations in the form @xmath239 where @xmath240 is the @xmath128 matrix . the eigenvalues of the matrix @xmath240 for the points ( a ) ,
( b ) , ( c ) are given by @xmath241 this shows that ( a ) and ( b ) are saddle , while ( c ) is stable .
hence the solutions finally approach the stable ds point ( c ) .
this ds stability is consistent with the analysis in sec .
[ confsec ] based on homogeneous perturbations .
the solution ( [ eq : soldr1 ] ) can be regarded as a tracker that attracts solutions with different initial conditions to a common trajectory . along the tracker we have @xmath242 , @xmath243 , and @xmath244 during the cosmological sequence of radiation , matter , and ds eras the dark energy equation of state evolves as @xmath245 , whereas the evolution of the effective equation of state is @xmath246 .
this peculiar evolution of @xmath192 can be useful to constrain the covariant galileon theory from observations .
equations ( [ eq : dr2 ] ) and ( [ eq : domr ] ) are simple enough to be solved analytically .
in fact , combining eqs .
( [ eq : dr2 ] ) and ( [ eq : domr ] ) , it follows that @xmath247 which has the solution @xmath248 where @xmath170 is a constant of integration . substituting this solution into eq .
( [ eq : domr ] ) , we find two branches that differ from each other in the early cosmological limit .
the viable branch of solutions is given by @xmath249 where @xmath171 is another constant . since @xmath250 at early times ( @xmath251 ) , we require that @xmath252 ( provided @xmath253 )
. the coefficients @xmath170 and @xmath171 can be found by using the present density parameters of radiation and non - relativistic matter , i.e. @xmath254 and @xmath255 .
using the relation ( [ omedere ] ) as well , we find @xmath256 the density parameter of dark energy evolves as @xmath257 hence the density parameters @xmath79 , @xmath219 , and @xmath258 as well as @xmath192 and @xmath193 are analytically known in terms of the function of @xmath259 ( or the redshift @xmath260 ) . at the ds point ( c )
the conditions for the avoidance of ghosts and instabilities have been already estimated in eqs .
( [ eq : qcs])-([eq : qcs4 ] ) .
let us consider the points ( a ) and ( b ) , which are characterized by @xmath201 and @xmath261 .
in this case eqs .
( [ st1])-([st4 ] ) are simplified to give @xmath262 since @xmath261 the conditions ( [ at2 ] ) and ( [ at4 ] ) are automatically satisfied . from eq .
( [ at1 ] ) the sign change of @xmath204 means the appearance of the scalar ghost . if we choose the initial conditions with @xmath263 , then eq .
( [ at1 ] ) requires that @xmath264 plane determined by the conditions ( [ eq : qcs])-([eq : qcs4 ] ) , ( [ at3 ] ) , ( [ at1d ] ) , and ( [ ctpo ] ) along the tracker solution @xmath201 @xcite.,width=316,height=326 ] [ fig1 ] let us consider the intermediate regime between @xmath261 and @xmath202 .
as long as the conditions @xmath265 , @xmath141 , and @xmath266 are satisfied both in the regimes @xmath261 and @xmath202 , the violation of these conditions does not occur in the intermediate epoch
. however , the tensor propagation speed squared can be negative even if the conditions ( [ eq : qcs4 ] ) and ( [ at4 ] ) are satisfied . along the tracker eq .
( [ st4 ] ) gives @xmath267}\,.\label{cttr}\ ] ] the transition to the ds solution ( @xmath202 ) occurs only recently , so that the term @xmath219 can be neglected in eq .
( [ cttr ] ) .
then @xmath140 has an extremum at @xmath268 if @xmath269 and @xmath270 , for example , the physical solution corresponds to the plus sign in eq .
( [ r2so ] ) , i.e. @xmath271 , at which @xmath140 has a minimum .
as @xmath207 approaches 1 , the minimum values of @xmath140 get smaller . for @xmath207 around 1 , @xmath140 can be negative for the plus sign of eq .
( [ r2so ] ) .
this leads to the following condition for @xmath272 : @xmath273 if @xmath274 , then @xmath140 remains to be positive .
hence we do not have any additional constraint in the regime @xmath274 . in fig .
[ fig1 ] we plot the parameter space constrained by the conditions ( [ eq : qcs])-([eq : qcs4 ] ) , ( [ at3 ] ) , ( [ at1d ] ) , and ( [ ctpo ] ) . for the solutions that start from initial conditions with @xmath275 and then approach the ds attractor with @xmath201 and @xmath202 , the parameters @xmath206 and @xmath207 need to be inside the purple region in fig . [ fig1 ] .
there is another case in which both @xmath203 and @xmath204 are initially much smaller than 1 .
we shall address this case in the next subsection . from eq .
( [ eq : drr1 ] ) , it is clear that another equilibrium point exists , namely , @xmath277 .
let us now discuss this equilibrium point in more detail . in this case eqs .
( [ eq : drr2 ] ) and ( [ eq : drr3 ] ) reduce to @xmath278 which depend on @xmath207 .
the dominant contribution to the field energy density comes from the term @xmath16 , i.e. @xmath279 .
we have the following fixed points @xmath280 which represent radiation , matter , and dark energy dominated points , respectively .
perturbing eq .
( [ eq : drr1 ] ) on the @xmath277 solution leads to @xmath281 which implies that none of the fixed points ( a@xmath282)-(c@xmath282 ) can be stable .
in particular the eigenvalues of the matrix @xmath240 , where @xmath283 and @xmath238 , are given by @xmath284 this shows that the point ( a@xmath282 ) is unstable , whereas the other two are saddle . recalling that the ds fixed point ( c ) discussed in the previous subsection is stable against homogenous perturbations , the solutions finally approach ( c ) instead of ( c@xmath282 ) .
unless @xmath203 is initially very small such that the solutions reach @xmath201 only at late times , the system approaches the stable @xmath201 direction much before the ds epoch .
in the regime @xmath215 and @xmath261 it is possible to derive analytic solutions for @xmath203 and @xmath204 as well as for @xmath192 and @xmath193 . in fact , eqs .
( [ eq : drr1 ] ) , ( [ eq : drr2 ] ) , and ( [ eq : drr3 ] ) can be simplified as @xmath285 where we have assumed that @xmath286 is not very much smaller than unity . during the radiation domination ( @xmath287 ) ,
integration of eqs .
( [ r1ap ] ) and ( [ r2ap ] ) gives @xmath288 whereas during the matter era one has @xmath289 eventually the solutions approach the tracker @xmath201 . in the regime @xmath276 one
has @xmath290 this gives @xmath291 and @xmath292 during the radiation era , whereas @xmath293 and @xmath294 during the matter era .
the condition ( [ st1 ] ) reduces to @xmath295 the sign change of @xmath204 implies the appearance of ghosts . for the initial conditions with @xmath263 we require that @xmath296 if the solutions start from the regime @xmath276 and subsequently enter the regime @xmath201 , the allowed parameter space in fig .
[ fig1 ] is restricted be @xmath272 .
since @xmath297 , the no - ghost condition for the tensor mode is automatically satisfied .
the propagation speeds of scalar and tensor perturbations are given , respectively , by @xmath298 which are both positive for @xmath299 .
the scalar mode remains sub - luminal during the radiation era ( @xmath300 ) and the matter era ( @xmath301 ) . under the no - ghost condition ( [ qsre ] )
the tensor mode becomes super - luminal ( although @xmath140 is very close to 1 ) . ,
@xmath258 , @xmath219 , and @xmath193 versus the redshift @xmath260 for @xmath302 , @xmath303 , @xmath304 , @xmath305 , and @xmath214 .
we choose the initial conditions @xmath306 , @xmath307 , and @xmath308 at @xmath309.,width=316,height=307 ] [ fig2 ] versus @xmath310 for @xmath302 , @xmath303 , @xmath304 , @xmath305 , and @xmath214 [ cases ( a)-(d ) ] .
we choose four different initial conditions : ( a ) @xmath311 , @xmath312 , and @xmath313 at @xmath314 , ( b ) @xmath306 , @xmath307 , and @xmath308 at @xmath309 , ( c ) @xmath315 , @xmath316 , and @xmath317 at @xmath318 , ( d ) @xmath319 , @xmath320 , and @xmath321 at @xmath322 .
the case ( e ) corresponds to @xmath323 , @xmath324 , @xmath304 , @xmath325 , and @xmath214 with initial conditions @xmath201 , @xmath326 , and @xmath327 at @xmath328.,width=316,height=307 ] [ fig3 ] numerically we integrate eqs .
( [ eq : drr1])-([eq : drr3 ] ) to confirm the analytic estimation in the previous subsections .
let us consider the case in which the variables @xmath203 and @xmath204 are much smaller than 1 at the initial stage of cosmological evolution .
our numerical simulations show that @xmath203 and @xmath204 evolve as eq .
( [ rradiation ] ) during the radiation era , whereas their evolution during the matter era is given by eq .
( [ rmatter ] ) . depending on the initial conditions of @xmath203 and @xmath204
, the epoch at which the solutions approach the tracker ( @xmath201 ) is different .
as we increase the initial ratio @xmath329 , this epoch tends to occur earlier .
after the solutions reach the tracker , the evolution of @xmath204 , @xmath219 , and @xmath79 is given by eqs .
( [ eq : solr2 ] ) , ( [ eq : omsol ] ) , and ( [ eq : omede ] ) , respectively . in fig
. [ fig2 ] we plot one example for the evolution of density parameters @xmath79 , @xmath258 , and @xmath219 as well as the effective equation of state @xmath193 . in this case
the transition to the regime @xmath330 occurs only recently , e.g. , @xmath331 around @xmath332 with @xmath333 . after passing the present epoch ,
the solutions are attracted by the ds solution characterized by @xmath334 .
figure [ fig2 ] shows that the sequence of radiation ( @xmath287 , @xmath335 ) , matter ( @xmath336 , @xmath337 ) , and ds ( @xmath338 , @xmath339 ) epochs is in fact realized .
unlike dark energy models based on @xmath6 theories , the galileon model is not plagued by the presence of a rapidly oscillating mode associated with a heavy field mass in the early universe .
figure [ fig3 ] illustrates the variation of @xmath192 for several different initial conditions and model parameters .
the cases ( a)-(d ) correspond to @xmath302 , @xmath303 , @xmath304 , @xmath305 , and @xmath214 with different initial conditions satisfying @xmath215 and @xmath261 , whereas the case ( e ) shows the tracker solution starting from the initial condition @xmath201 and @xmath261 with the model parameters @xmath323 , @xmath324 , @xmath304 , @xmath325 , and @xmath214 .
clearly the solutions with different initial conditions converge to the tracker , depending on the epoch at which the variable @xmath203 grows to the order of 1 . in the cases ( a)-(d ) the dark energy equation of state evolves as eq .
( [ wdesm ] ) in the regime @xmath215 and @xmath261 ( @xmath291 and @xmath293 during the radiation and matter eras , respectively ) , which is followed by the evolution given in eq .
( [ wdetra ] ) after the solutions reach the tracker at @xmath201 .
as long as the tracking behavior occurs by today , the dark energy equation of state crosses the cosmological constant boundary ( @xmath340 ) .
numerically we find that for the initial conditions with @xmath341 the solutions are typically attracted by the tracker . on the other hand , if @xmath342 , the system tends to approach the matter - dominated epoch with the growth of @xmath203 . in the latter case
the dominant contribution to @xmath79 comes from the term @xmath20 , so that @xmath79 decreases as in quintessence without a potential . versus @xmath310 for the same model parameters and initial conditions as given in fig .
[ fig3].,width=316,height=297 ] [ fig4 ] in fig .
[ fig4 ] we plot the evolution of @xmath161 for the same model parameters and initial conditions as those presented in fig .
[ fig3 ] . in the regime @xmath215 and @xmath261 ,
our numerical simulations in the cases ( a)-(d ) agree with the analytic estimation of the scalar propagation speed given in eq .
( [ cssma ] ) , i.e. @xmath343 and @xmath344 during the radiation and matter eras respectively .
as the solutions reach the regime @xmath330 with @xmath261 , @xmath161 approaches the value estimated by eq .
( [ at3 ] ) .
when @xmath302 and @xmath303 the analytic estimation gives @xmath345 during the matter dominance , which agrees with the value at the plateau in the case ( d ) of fig .
finally the solutions reach the ds fixed point , at which @xmath161 shifts to the value given in eq .
( [ csds ] ) , e.g. , @xmath346 for @xmath302 and @xmath303 . for positive @xmath207
one can show that under the conditions ( [ eq : qcs ] ) , ( [ csds ] ) , and ( [ at1 ] ) the scalar propagation speed estimated by eq .
( [ at3 ] ) becomes super - luminal .
however , the scalar mode can remain sub - luminal provided the solutions reach the regime @xmath201 in the recent past .
the cases ( a ) and ( b ) in fig .
[ fig4 ] correspond to such examples in which the peak value of @xmath161 is smaller than 1 . if @xmath274 there is a parameter space in which the scalar propagation speed ( [ at3 ] ) is sub - luminal , while satisfying the conditions ( [ eq : qcs ] ) , ( [ csds ] ) , and ( [ at1 ] ) . in this case
the initial conditions of @xmath203 need to be close to 1 .
if @xmath203 is smaller than the order of unity , the scalar ghost appears for negative @xmath207 . on the other hand , if @xmath342 , the solutions do not finally approach the ds fixed point .
the case ( e ) in fig .
[ fig4 ] corresponds to an example of the sub - luminal evolution of @xmath161 for negative @xmath207 with the initial condition @xmath201 .
since the solution stays on the tracker , the scalar propagation speed is given by eq .
( [ at3 ] ) during the radiation and matter eras and by eq .
( [ csds ] ) at the ds point . versus @xmath310 for two cases : ( a ) @xmath302 , @xmath303 , @xmath304 , @xmath305 , and @xmath214 with initial conditions @xmath306 , @xmath307 , and @xmath308 at @xmath309 , ( b ) @xmath323 , @xmath324 , @xmath304 , @xmath325 , and @xmath214 with initial conditions @xmath201 , @xmath326 , and @xmath327 at @xmath328 , and ( c ) @xmath269 , @xmath270 , @xmath304 , @xmath305 , and @xmath214 with initial conditions @xmath347 , @xmath348 , and @xmath327 at @xmath328.,width=316,height=307 ] [ fig5 ] for the initial conditions with @xmath215 and @xmath261 the tensor propagation speed starts to evolve from the value estimated by eq .
( [ ctsma ] ) , which is slightly super - luminal under the no - ghost condition @xmath272 for the scalar mode .
after the solutions reach the regime @xmath330 and @xmath261 , @xmath140 is still close to 1 because it is described by eq .
( [ at4 ] ) .
the tensor propagation speed finally approaches the value ( [ eq : qcs4 ] ) at the ds point . during the transition from the regime @xmath261 to the regime @xmath349
, there is an epoch at which @xmath140 can have either the maximum or the minimum . in the case ( a ) of fig .
[ fig5 ] , the analytic formulas in eqs .
( [ cttr ] ) and ( [ r2so ] ) show that @xmath140 has a minimum value @xmath350 at @xmath351 [ plus sign of eq .
( [ r2so ] ) ] , whereas in the case ( b ) @xmath140 possesses a maximum value @xmath352 at @xmath353 [ minus sign of eq .
( [ r2so ] ) ] .
this estimation agrees well with the numerical results shown in fig .
in the case ( c ) of fig .
[ fig5 ] the condition ( [ ctpo ] ) is violated , so that @xmath140 has a negative minimum . in the region where @xmath206 and @xmath207 are positive , the condition ( [ ctpo ] ) needs to be satisfied to avoid the temporal laplacian instability of the tensor mode .
if the solutions start from the regime @xmath330 and @xmath261 , then the tensor propagation speed ( [ at4 ] ) can be sub - luminal under the condition @xmath354 for the branch @xmath263 . in this case
, however , @xmath140 exceeds 1 at the ds point , as long as the conditions ( [ eq : qcs])-([eq : qcs3 ] ) and ( [ at1])-([at3 ] ) are satisfied . since @xmath355 in the regime @xmath215 and @xmath261 as well , it is not possible to avoid the appearance of the super - luminal mode for tensor perturbations .
however , the super - luminal propagation does not necessarily imply the inconsistency of galileon theory because of the possibility for the absence of the closed causal curve @xcite .
we shall proceed to the cosmology for the theories with non - constant @xmath66 in which the functions @xmath66 and @xmath356 ( @xmath357 ) are given in eq .
( [ funchoice2 ] ) with @xmath111 .
we take into account radiation ( @xmath358 , @xmath359 ) and non - relativistic matter ( @xmath360 , @xmath361 ) , which satisfy the continuity equations ( [ be3 ] ) . taking the time - derivative of eq .
( [ eq : frd1 ] ) and combining it with eq .
( [ eq : frd2 ] ) , we obtain the equations of motion for @xmath362 and @xmath363
. then the dimensionless variables @xmath364 defined in eq .
( [ tildexy ] ) and the radiation density parameter @xmath365 obey the following equations @xmath366/(60\delta)\ , , \label{xgene } \\
\ty ' & = & \tx \ty [ 116640000000000 \tx^2 ( 24 - 24 \omega_r + p \tx ( -48 - 24 p\tx+d_2 ^ 2 \epsilon_2 ( 12 + 5p \tx ) ) -4320000000000 d_3 ^ 3 p^2 \tx^2 ( 72 - 24\omega_r \nonumber \\ & & + p \tx ( -16+(-32 + 3d_2 ^ 2 \epsilon_2 ) p\tx ) ) \ty^2 -625000000 p^3 \tx ( 2048 d_3 ^ 6 p^3 \tx^3 + 2187 d_4 ^ 4 \epsilon_4 ( 96 \omega_r+p \tx ( 108-p \tx ( -120 \nonumber \\ & & + d_2 ^ 2 \epsilon_2 ( 18+p \tx ) ) ) ) ) \ty^4 + 64800000 p^4 ( 3125 d_3 ^ 3 d_4 ^ 4 \epsilon_4 p^2 \tx^2 ( -9 + 10p \tx ) + 1728 d_5 ^ 5 ( 120 + 120 \omega_r \nonumber \\ & & + p \tx ( 128+p \tx ( 144 -d_2 ^ 2 \epsilon_2 ( 27 + 2p \tx ) ) ) ) ) \ty^6 -253125 p^7 \tx ( 3796875 d_4 ^ 8 \epsilon_4 ^ 2 p \tx + 524288 d_3 ^ 3 d_5 ^ 5 ( -3+p \tx ) ) \ty^8 \nonumber \\ & & + 5248800000 d_4 ^ 4 d_5 ^ 5 \epsilon_4 p^8 ( -9 + 28p \tx ) \ty^{10}-6019743744 d_5^{10 } p^{10 } \ty^{12}]/(60 \delta)\ , , \\
\omega_r ' & = & p\,\omega_r [ 972000000000 \tx^4 ( -48p \tx+d_2 ^ 2 \epsilon_2 ( 8 - 8\omega_r + p \tx ( 16+(-4+d_2 ^ 2 \epsilon_2 ) p \tx ) ) ) -48000000000 d_3 ^ 3 p \tx^3 ( 144 - 144 \omega_r \nonumber \\ & & + p\tx ( -72+p \tx ( -168+d_2 ^ 2 \epsilon_2 ( 36+p\tx ) ) ) ) \ty^2 + 62500000 p^2 \tx^2 ( -1024 d_3 ^ 6 p^2 \tx^2 ( -2+p \tx)+729d_4 ^ 4 \epsilon_4 ( 144 - 144 \omega_r \nonumber \\ & & + p \tx ( -192+p \tx ( -204+d_2 ^ 2 \epsilon_2 ( 43 + 3p \tx ) ) ) ) ) \ty^4 + 2160000 p^3 \tx ( 3125 d_3 ^ 3 d_4 ^ 4 \epsilon_4 p^2 \tx^2 ( -18 + 17 p \tx)+1728 d_5 ^ 5 ( -160 \nonumber \\ & & + 160 \omega_r+p \tx ( 280+p \tx ( 248-d_2 ^ 2 \epsilon_2 ( 54 + 5p \tx)))))\ty^6 - 5625 p^6 \tx^2 ( 11390625 d_4 ^ 8 \epsilon_4 ^ 2 ( 1+p \tx ) \nonumber \\ & & + 65536 d_3 ^ 3 d_5 ^ 5 ( -36 + 23p
\tx ) ) \ty^8 + 174960000 d_4 ^ 4 d_5 ^ 5 \epsilon_4 p^7 \tx ( 88 + 63p \tx ) \ty^{10}-501645312 d_5^{10 } p^8 ( 2+p \tx ) \ty^{12}]/\delta\ , , \label{omergene}\end{aligned}\ ] ] where @xmath367\,.\end{aligned}\ ] ] the ds fixed point with @xmath368 and @xmath369 exists under the conditions ( [ gened2 ] ) and ( [ gened3 ] ) . since the theory has a nonminimal coupling @xmath370 , it is possible to place constraints on the values of @xmath82 around today from the variation of the effective gravitational coupling , @xmath371^{-1}$ ] .
the lunar laser ranging experiments give the bound @xmath372yr@xmath373 @xcite , or in terms of the present hubble parameter @xmath374 , @xmath375 @xcite .
in our theory @xmath376 , which gives the constraint @xmath377 around today . since the value of @xmath378 is not much different from @xmath379 today , we employ the following criterion @xmath380 under this bound the condition ( [ dsstability ] ) is always satisfied , which means that the ds solution is classically stable . from eq .
( [ eq : cc2sx ] ) the laplacian instability of scalar perturbations at the ds point can be avoided for @xmath381 . the no - ghost condition ( [ eq : qc2sx ] ) is satisfied provided that @xmath382 . if @xmath383 in the early cosmological epoch , then the term @xmath384 dominates over the terms @xmath385 , i.e. @xmath386 . in order to avoid the dominance of dark energy during the radiation and matter eras we require that @xmath387 and @xmath388 .
in this regime the quantities @xmath389 and @xmath390 defined in eqs .
( [ eq : genqs ] ) and ( [ eq : genqw ] ) are approximately given by @xmath391 the tensor ghost is absent for @xmath382 .
since the evolution of the field is given by @xmath392 , where @xmath393 is the initial field value at @xmath394 , the condition @xmath382 is satisfied for @xmath395 . for the avoidance of the scalar ghost we require that @xmath396 in the regime @xmath383 and @xmath388 the scalar and tensor propagation speeds defined in eqs .
( [ cs2 ] ) and ( [ eq : noinstgw ] ) can be estimated as @xmath397 since @xmath398 is close to 1 , the tensor instability can be avoided . if @xmath399 , then there is no instability for the scalar perturbation ( @xmath400 ) . in the regime @xmath401
we have @xmath402 .
if @xmath403 , it can happen that the scalar perturbation is subject to the laplacian instability for negative @xmath404 . , @xmath258 , @xmath219 , and @xmath193 versus the redshift @xmath310 for the model with @xmath405 , @xmath406 , @xmath407 , @xmath408 , @xmath409 , @xmath111 , and @xmath410 .
the initial conditions are chosen to be @xmath411 , @xmath412 , and @xmath413 at @xmath414.,width=316,height=307 ] [ fig6 ] , @xmath415 , and @xmath416 versus @xmath310 for the same model parameters and the initial conditions as those given in fig . [ fig6 ] .
the solution finally approaches the de sitter attractor with @xmath417.,width=316,height=307 ] [ fig7 ] in the regime @xmath383 and @xmath388 the autonomous equations ( [ xgene])-([omergene ] ) are simplified as @xmath418\ , , \label{txap1}\\ \ty ' & \simeq & \frac{3}{8}\ty \left [ 1+\omega_r+ \frac{625}{3 } ( 1-\omega_r)\frac{\tx^2}{d_5 ^ 5 p^4 \ty^6 } \right]\ , , \label{tyap1 } \\
\omega_r ' & \simeq & -\omega_r ( 1-\omega_r)\ , .
\label{omerap}\end{aligned}\ ] ] from eq .
( [ omerap ] ) there are two fixed points characterized by @xmath419 and @xmath369 . as long as the condition @xmath420 is satisfied , the evolution of the variables @xmath379 and @xmath421 during the radiation era ( @xmath419 ) is given by @xmath422 whereas during the matter era ( @xmath369 ) one has @xmath423 in both cases @xmath379 grows faster than @xmath421 . if the quantity @xmath424 becomes larger than the order of unity , the evolution of @xmath379 and @xmath421 is subject to change . and @xmath425 versus @xmath310 for the same model parameters and the initial conditions as those given in fig .
the signs of @xmath389 and @xmath390 remain to be positive.,width=316,height=307 ] [ fig8 ] and @xmath398 versus @xmath310 for the same model parameters and the initial conditions as those given in fig .
( right ) the enlarged version for the evolution of @xmath426 in the regime @xmath427 .
the field propagation speed becomes negative in the future ( around @xmath428).,title="fig:",width=307,height=307 ] and @xmath398 versus @xmath310 for the same model parameters and the initial conditions as those given in fig .
( right ) the enlarged version for the evolution of @xmath426 in the regime @xmath427 . the field propagation speed becomes negative in the future ( around @xmath428).,title="fig:",width=307,height=307 ] [ fig9 ] for the solutions starting from the regime @xmath429 the condition ( [ genecon3 ] ) needs to be satisfied initially .
then there are two possible cases : ( i ) @xmath430 and @xmath399 , and ( ii ) @xmath431 and @xmath432 .
the avoidance of the scalar laplacian instability at the future ds fixed point requires that @xmath433 .
however , if we demand the viable cosmology _ by today _
( the redshift @xmath434 ) , the condition @xmath433 is not necessarily mandatory . in general ,
if the variable @xmath379 changes its sign during the cosmic expansion history , this signals the violation of the conditions for no ghosts and no laplacian instabilities .
for example , this can be seen in the expression of @xmath389 and @xmath426 in eqs .
( [ qsasy ] ) and ( [ csctana ] ) in the past asymptotic regime .
in fact we have numerically confirmed the violation of at least one of those conditions . in the following
we shall study the cosmological dynamics in which the sign of @xmath435 at the early epoch is same as that of @xmath436 . in the case ( i )
the condition @xmath433 is violated , but it is possible to realize cosmological trajectories in which all the required conditions are satisfied by today . in the case ( ii ) the condition @xmath433 is met , but we need to check whether there are no violations of the no - ghost and stability conditions in the cosmic expansion history .
let us first discuss the cosmological dynamics in the case ( i ) with @xmath437 . in fig .
[ fig6 ] we plot the variation of @xmath79 , @xmath258 , @xmath219 , and @xmath193 for the model with @xmath405 , @xmath406 , @xmath407 , @xmath408 , @xmath409 , @xmath111 , and @xmath410 ( in which case the condition ( [ genecon1 ] ) is satisfied ) .
the constants @xmath438 and @xmath439 are known from eqs .
( [ gened2 ] ) and ( [ gened3 ] ) .
we choose the initial conditions @xmath440 , @xmath441 , and @xmath413 at the redshift @xmath414 , in which case @xmath429 and @xmath388 initially . the background evolution in fig . [ fig6 ] shows that the sequence of radiation , matter , and ds eras is realized in this case .
figure [ fig7 ] illustrates the evolution of the variables @xmath379 and @xmath421 as well as @xmath416 .
we find that @xmath379 approaches the ds attractor with @xmath442 without changing its sign . in the regime
@xmath401 the evolution of @xmath379 and @xmath421 is well described by the analytic estimation ( [ xandyasy ] ) during the radiation era .
however , around @xmath443 , the last terms in eqs .
( [ txap1 ] ) and ( [ tyap1 ] ) starts to give rise to the contribution to the evolution of @xmath379 and @xmath421 . as we see in fig .
[ fig7 ] , @xmath379 and @xmath421 evolve differently from the analytic estimation ( [ xandyasy ] ) and ( [ xandyasy2 ] ) for @xmath443 . in figs .
[ fig8 ] and [ fig9 ] we plot the variation of the quantities @xmath444 , @xmath445 , @xmath426 , and @xmath398 for the same model parameters and initial conditions as those given in fig .
we find that @xmath446 grows rapidly , whereas @xmath447 is always close to 1 . since both @xmath446 and @xmath447 are positive , the appearance of the scalar and tensor ghosts is avoided in this case .
figure [ fig9 ] shows that @xmath426 starts to evolve from the value around @xmath448 , as estimated analytically in eq .
( [ csctana ] ) .
for @xmath443 the contribution of the second term in the expression of @xmath426 in eq .
( [ csctana ] ) becomes important , which leads to the increase of @xmath426 .
for the model parameters given in fig .
[ fig6 ] the scalar propagation speed slightly exceeds 1 during the transition from the matter era to the ds epoch . in fig . [ fig9 ] we find that @xmath426 remains positive until recently ( @xmath434 ) .
however , since the sign of @xmath379 is always positive , @xmath426 is negative at the ds point , i.e. @xmath449 . the crossing of @xmath426 at 0 occurs in future around the redshift @xmath428 .
the tensor propagation speed squared is always close to 1 ( slightly larger than 1 ) , which means that the laplacian instability of the tensor perturbation can be avoided .
versus @xmath310 for the model with @xmath405 , @xmath406 , @xmath407 , @xmath408 , @xmath450 , @xmath111 , and @xmath451 .
the initial conditions are chosen to be @xmath452 , @xmath453 , and @xmath454 at @xmath455 . in this case
the scalar perturbation is subject to the laplacian instability for many times by today.,width=316,height=307 ] [ fig10 ] let us next discuss the case with ( ii ) , i.e. @xmath431 and @xmath432 initially . in fig .
[ fig10 ] we plot one example for the evolution of @xmath426 with @xmath405 , @xmath450 , and @xmath456 . in this case
the density parameters as well as the effective equation of state evolve similarly as those in fig .
however , even if the variable @xmath379 starts from negative values , @xmath379 crosses 0 for many times before reaching the ds solution with @xmath456 . as we see in fig .
[ fig10 ] , this leads to the violation of the condition @xmath400 by today .
in addition the quantity @xmath389 also becomes negative during some periods .
we have run our numerical code for many other cases in which the condition @xmath433 is satisfied and found that in the case ( ii ) it is difficult to find a viable cosmological trajectory along which all of the no - ghost and stability conditions are satisfied . in summary
, we have shown that the cosmological solutions along which @xmath430 and @xmath399 initially and @xmath437 at the de sitter attractor can evade the problems of the ghost and instability problems for @xmath434 . in this case , although the scalar laplacian instability is present at the de sitter fixed point , the crossing of @xmath426 at 0 occurs at some time in future .
we have also run the numerical code for the initial conditions with @xmath457 and found similar properties of solutions to those discussed in this section .
finally we shall study the case in which @xmath459 in the early cosmological epoch . in this regime
the term @xmath460 is the dominant contribution to @xmath461 relative to @xmath462 , i.e. @xmath463 . the quantities @xmath389 and @xmath390 are approximately given by @xmath464 whereas both @xmath426 and @xmath398 are close to be 1 .
the tensor ghost can be avoided for @xmath382 . under this condition
the scalar ghost is absent for @xmath465 .
if @xmath466 , the absence of the scalar ghost requires that @xmath467 for @xmath468 and @xmath404 of the order of unity we find from eq .
( [ gened2 ] ) that @xmath469 , where we used the condition ( [ genecon1 ] ) .
hence the ds solution exists only for @xmath466 , in which case @xmath470 .
this is incompatible with the condition ( [ d2con ] ) .
these results show that , if the solutions start from the regime @xmath459 with @xmath466 ( i.e. negative kinetic energy ) , the requirement for the avoidance of ghosts at the initial stage is not compatible with the existence of the ds solution at late times .
in this paper we have studied the cosmology of generalized galileon theories based on the lagrangian ( [ lag ] ) . for each lagrangian @xmath11 ( @xmath2 )
the scalar field @xmath0 is replaced by general scalar functions @xmath1 .
the covariant galileon theory satisfies the galilean symmetry symmetry @xmath9 in the minkowski space - time .
the extension to scalar functions @xmath1 generally breaks this symmetry , but the equations of motion remain at second - order .
this is a welcome feature to avoid the propagation of the extra ghost degree of freedom .
we have also taken into account two terms @xmath471 and @xmath472 that give rise to second - order equations and vanish in the minkowski space - time . in the flat flrw cosmological background
we have derived the equations of motion ( [ eq : frd1])-([conser ] ) for the general lagrangian ( [ lag ] ) .
if we demand the existence of ds solutions , the functions @xmath3 , @xmath1 , and @xmath14 are restricted to be either in the form ( [ funchoice ] ) or ( [ funchoice2 ] ) .
the former corresponds to the covariant galileon theory with constant @xmath66 , respecting the galilean symmetry in the minkowski space - time .
the latter can be regarded as a kind of scalar - tensor theories in which @xmath66 is field - dependent . in the presence of two perfect fluids
we have also derived conditions for the avoidance of ghosts and laplacian instabilities associated with scalar and tensor perturbations . the no - ghost conditions ( [ eq : constro1 ] ) and ( [ eq : constro2 ] ) are automatically satisfied for the perfect fluids of radiation and non - relativistic matter .
then the no - ghost condition of the scalar mode is given by eq .
( [ eq : genqs ] ) , whereas the ghost is absent for the tensor mode under the condition ( [ eq : genqw ] ) .
the stability conditions for scalar and tensor perturbations are given , respectively , by eqs .
( [ cs2 ] ) and ( [ eq : noinstgw ] ) .
we have applied these results to two theories having ds solutions .
for the theory with constant @xmath66 the ds solutions are always classically stable against homogeneous perturbations , whereas for the theory with non - constant @xmath66 they are stable under the condition ( [ dsstability ] ) .
we have carried out detailed analysis for the cosmological dynamics of the covariant galileon theory with constant @xmath66 . introducing the dimensionless variables @xmath203 , @xmath204 , and @xmath219 together with the constants @xmath206 and @xmath207
, it is possible to express autonomous equations as well as physical quantities ( both background and perturbations ) in terms of those variables in a convenient form . in particular we showed the existence of an interesting tracker solution @xmath201 , along which the field velocity evolves as @xmath473 . on this tracker
all the non - linear field lagrangians contribute to the field energy density with the similar order , such that any of these terms can not be neglected .
moreover the cosmological dynamics along @xmath201 does not depend on the parameters @xmath206 and @xmath207 , see eqs .
( [ eq : dr2 ] ) and ( [ eq : domr ] ) .
the solutions with different initial conditions converge to a common trajectory , depending on the epoch at which they reach the regime @xmath330 . along the tracker solution
the dark energy equation of state is given by eq .
( [ wdetra ] ) , which exhibits peculiar evolution : @xmath474 ( radiation era ) , @xmath475 ( matter era ) , and @xmath340 ( ds era ) .
since we have derived analytic formulas for @xmath192 as well as @xmath204 and @xmath219 in terms of the scale factor @xmath259 , this will be convenient to confront the galileon theory with supernovae observations .
although the background dynamics on the tracker does not depend on the parameters @xmath206 and @xmath207 , the conditions for the avoidance of ghosts and laplacian instabilities do . in fig .
[ fig1 ] we showed the viable parameter space in the @xmath165 plane constrained by the no - ghost and stability conditions along @xmath201 . if the solutions start from the regime @xmath215 , we also require the condition @xmath272 to avoid the scalar ghost . in this case
the tensor mode becomes slightly super - luminal . in the minkowski space - time
the only solution to the field equation in galileon theory with @xmath476 corresponds to @xmath60 , so that the super - luminal propagation is absent .
we have also studied the cosmology based on the theories with non - constant @xmath3 having de sitter solutions at late times . for the initial conditions with @xmath477 we require that @xmath478 in the early cosmological epoch . if @xmath479 , there are some viable cosmological trajectories along which the solutions fulfill all the required conditions by today . such an example is given in figs .
[ fig6]-[fig9 ] , along which the quantity @xmath435 remains to be positive . in this case
the scalar perturbation is subject to the laplacian instability at the de sitter fixed point in future ( @xmath480 ) . if @xmath481 initially , we find that the violations of the conditions @xmath400 or @xmath482 typically occur by today . for the initial conditions with @xmath483
the condition for the avoidance of ghosts in the early cosmological epoch is not compatible with the existence of the late - time de sitter solutions .
the field - derivative couplings with the ricci scalar @xmath4 and the einstein tensor @xmath25 appearing in the terms @xmath15 and @xmath16 can lead to imprints on the dynamics of matter density perturbations through the change of the effective gravitational coupling .
it will be of interest to study the evolution of perturbations in detail in order to discriminate between the generalized galileon model and other dark energy models . the work of a.d . and s.t .
was supported by the grant - in - aid for scientific research fund of the jsps nos .
09314 and 30318802 .
also thanks financial support for the grant - in - aid for scientific research on innovative areas ( no .
21111006 ) .
we thank savvas nesseris and jiro soda for useful discussions .
a. g. riess _ et al .
_ , astron .
j. * 116 * , 1009 ( 1998 ) ; astron .
j. * 117 * , 707 ( 1999 ) ; s. perlmutter _ et al .
_ , astrophys .
j. * 517 * , 565 ( 1999 ) ; d. n. spergel _ et al .
_ , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) ; m. tegmark _ et al . _
[ sdss collaboration ] , phys . rev .
d * 69 * , 103501 ( 2004 ) ; d. j. eisenstein _ et al .
_ [ sdss collaboration ] , astrophys . j. * 633 * , 560 ( 2005 ) .
v. sahni and a. a. starobinsky , int .
j. mod .
d * 9 * , 373 ( 2000 ) ; s. m. carroll , living rev .
rel . * 4 * , 1 ( 2001 ) ; t. padmanabhan , phys .
* 380 * , 235 ( 2003 ) ; p.
j. e. peebles and b. ratra , rev .
phys . * 75 * , 559 ( 2003 ) ; e. j. copeland , m. sami and s. tsujikawa , int .
j. mod .
d * 15 * , 1753 ( 2006 ) ; p. brax , arxiv:0912.3610 [ astro-ph.co ] ; s. tsujikawa , arxiv:1004.1493 [ astro-ph.co ] . y. fujii , phys . rev .
d * 26 * , 2580 ( 1982 ) ; l. h. ford , phys .
d * 35 * , 2339 ( 1987 ) ; c. wetterich , nucl .
phys b. * 302 * , 668 ( 1988 ) ; b. ratra and j. peebles , phys .
rev d * 37 * , 321 ( 1988 ) ; r. r. caldwell , r. dave and p. j. steinhardt , phys .
lett . * 80 * , 1582 ( 1998 ) .
r. durrer and r. maartens , gen .
grav . * 40 * , 301 ( 2008 ) ; f. s. n. lobo , arxiv:0807.1640 [ gr - qc ] ; t. p. sotiriou and v. faraoni , rev .
phys . * 82 * , 451 ( 2010 ) ; a. de felice and s. tsujikawa , living rev .
* 13 * , 3 ( 2010 ) .
s. capozziello , int .
j. mod .
d * 11 * , 483 ( 2002 ) ; s. capozziello , s. carloni and a. troisi , recent res .
astrophys .
* 1 * , 625 ( 2003 ) ; s. capozziello , v. f. cardone , s. carloni and a. troisi , int .
j. mod .
d * 12 * , 1969 ( 2003 ) ; s. m. carroll , v. duvvuri , m. trodden and m. s. turner , phys .
d * 70 * , 043528 ( 2004 ) .
s. m. carroll _ et al .
, _ phys .
d * 71 * , 063513 ( 2005 ) ; s. nojiri , s. d. odintsov and m. sasaki , phys .
d * 71 * , 123509 ( 2005 ) ; s. nojiri and s. d. odintsov , phys .
b * 631 * , 1 ( 2005 ) ; g. calcagni , s. tsujikawa and m. sami , class .
* 22 * , 3977 ( 2005 ) ; o. mena , j. santiago and j. weller , phys .
* 96 * , 041103 ( 2006 ) ; a. de felice , m. hindmarsh and m. trodden , jcap * 0608 * , 005 ( 2006 ) ; t. koivisto and d. f. mota , phys .
b * 644 * , 104 ( 2007 ) ; phys .
d * 75 * , 023518 ( 2007 ) ; s. tsujikawa and m. sami , jcap * 0701 * , 006 ( 2007 ) ; a. de felice and s. tsujikawa , phys .
b * 675 * , 1 ( 2009 ) ; phys .
d * 80 * , 063516 ( 2009 ) ; a. de felice , d. f. mota and s. tsujikawa , phys .
d * 81 * , 023532 ( 2010 ) . c. deffayet , g. r. dvali and g. gabadadze , phys .
d * 65 * , 044023 ( 2002 ) ; v. sahni and y. shtanov , jcap * 0311 * , 014 ( 2003 ) ; c. de rham _ et al . _ , phys .
* 100 * , 251603 ( 2008 ) ; c. de rham , s. hofmann , j. khoury and a. j. tolley , jcap * 0802 * , 011 ( 2008 ) ; n. agarwal , r. bean , j. khoury and m. trodden , phys . rev .
d * 81 * , 084020 ( 2010 ) .
l. amendola , r. gannouji , d. polarski and s. tsujikawa , phys .
d * 75 * , 083504 ( 2007 ) ; b. li and j. d. barrow , phys .
d * 75 * , 084010 ( 2007 ) ; l. amendola and s. tsujikawa , phys .
b * 660 * , 125 ( 2008 ) ; w. hu and i. sawicki , phys .
d * 76 * , 064004 ( 2007 ) ; a. a. starobinsky , jetp lett .
* 86 * , 157 ( 2007 ) ; s. a. appleby and r. a. battye , phys .
b * 654 * , 7 ( 2007 ) ; s. tsujikawa , phys .
d * 77 * , 023507 ( 2008 ) ; e. v. linder , phys .
d * 80 * , 123528 ( 2009 ) .
j. a. r. cembranos , phys .
d * 73 * , 064029 ( 2006 ) ; i. navarro and k. van acoleyen , jcap * 0702 * , 022 ( 2007 ) ; t. faulkner , m. tegmark , e. f. bunn and y. mao , phys . rev .
d * 76 * , 063505 ( 2007 ) ; s. capozziello and s. tsujikawa , phys . rev .
d * 77 * , 107501 ( 2008 ) ; p. brax , c. van de bruck , a. c. davis and d. j. shaw , phys .
d * 78 * , 104021 ( 2008 ) . c. deffayet , g. r. dvali , g. gabadadze and a. i. vainshtein , phys .
d * 65 * , 044026 ( 2002 ) ; m. porrati , phys .
b * 534 * , 209 ( 2002 ) ; m. a. luty , m. porrati and r. rattazzi , jhep * 0309 * , 029 ( 2003 ) . m. fairbairn and a. goobar , phys . lett .
b * 642 * , 432 ( 2006 ) ; r. maartens and e. majerotto , phys .
d * 74 * , 023004 ( 2006 ) ; u. alam and v. sahni , phys .
d * 73 * , 084024 ( 2006 ) ; y. s. song , i. sawicki and w. hu , phys .
d * 75 * , 064003 ( 2007 ) ; j. q. xia , phys . rev .
d * 79 * , 103527 ( 2009 ) .
j. m. bardeen , phys .
d * 22 * , 1882 ( 1980 ) ; h. kodama and m. sasaki , prog .
suppl . * 78 * , 1 ( 1984 ) ; v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .
215 * , 203 ( 1992 ) ; j. c. hwang and h. r. noh , phys .
d * 65 * , 023512 ( 2002 ) ; b. a. bassett , s. tsujikawa and d. wands , rev .
* 78 * , 537 ( 2006 ) ; k. a. malik and d. wands , phys .
rept . * 475 * , 1 ( 2009 ) .
|
we study the cosmology of a generalized galileon field @xmath0 with five covariant lagrangians in which @xmath0 is replaced by general scalar functions @xmath1 ( @xmath2 ) . for these theories ,
the equations of motion remain at second - order in time derivatives .
we restrict the functional forms of @xmath1 from the demand to obtain de sitter solutions responsible for dark energy .
there are two possible choices for power - law functions @xmath1 , depending on whether the coupling @xmath3 with the ricci scalar @xmath4 is independent of @xmath0 or depends on @xmath0 .
the former corresponds to the covariant galileon theory that respects the galilean symmetry in the minkowski space - time . for generalized galileon theories
we derive the conditions for the avoidance of ghosts and laplacian instabilities associated with scalar and tensor perturbations as well as the condition for the stability of de sitter solutions .
we also carry out detailed analytic and numerical study for the cosmological dynamics in those theories .
|
application requirements , power , and technological constraints are driving the architectural convergence of future processors towards heterogeneous many - cores .
this development is confronted with variability challenges , mainly the susceptibility to time - dependent variations in silicon devices .
increasing guard - bands to battle variations is not scalable , due to too large worst - case cost impact for technology nodes around 10 nm .
the goal of next generation firmware is to enable next - generation embedded and high - performance heterogeneous many - cores to cost - effectively confront variations by providing dependable - performance : correct functionality and timing guarantees throughout the expected lifetime of a platform under thermal , power , and energy constraints .
an optimal solution should employ a cross - layer approach .
a middle - ware implements a control engine that steers software / hardware knobs based on information from strategically dispersed monitors .
this engine relies on technology models to identify / exploit various types of platform slack - performance , power / energy , thermal , lifetime , and structural ( hardware ) - to restore timing guarantees and ensure the expected lifetime and time - dependent variations .
dependable - performance is critical for embedded applications to provide timing correctness ; for high - performance applications , it is paramount to ensure load balancing in parallel phases and fast execution of sequential phases .
the lifetime requirement has ramifications on the manufacturing process cost and the number of field - returns .
the future firmware novelty must rely in seeking synergies in techniques that have been considered virtually exclusively in the embedded or high - performance domains ( worst - case guaranteed partly proactive techniques in embedded , and dynamic best - effort reactive techniques in high - performance ) .
this possible future solutions will demonstrate the benefits of merging concepts from these two domains by evaluating key applications from both segments running on embedded and high - performance platforms .
the intent of this paper is to describe the characteristics and the constraints of disaster management ( dm ) applications for industrial environments . when defining the requirements and their evaluation procedure , a first analysis of the dm applications modules ( hw platform , os and rt engines , monitors and knobs , reliability models ) is provided .
power consumption is a critical consideration in high performance computing systems and it is becoming the limiting factor to build and operate petascale and exascale systems .
when studying the power consumption of existing systems running hpc workloads , we find power , energy and performance are closely related leading to the possibility to optimize energy without sacrificing ( much or at all ) performance . in this paper , we propose a hpc system running with an rtrm that is aware and monitors the healthy of the platform . on the system ,
an application for disaster management is running .
the application can run with different qos depending on the situation .
we defined two main situations .
normal execution when there is no risk of disaster , even though we still have to run the system to look ahead in the near future if the situation changes suddenly . and
the second scenario where the possibilities for a disaster are very high .
then the allocation of more resources for improving the precision and human decision has to be taken into account .
the paper shows that at design time , it is possible to describe different optimal points that are going to be used at runtime by the rtrm with the application .
this environment helps to the system that must run 24/7 in saving energy with the trade - off of losing precision .
the paper shows a model execution which can improve the precision of results by 65% in average by increasing the number of iterations from 10 ^ 3^ to 10 ^ 4^. this also produces one order of magnitude longer execution time which leads to the need to use a multi - node solution .
the optimal trade - off between precision vs. execution time is computed by the rtrm with the time overhead less than 10% against a native execution .
the paper is divided into nine sections , after the introduction the state - of the art follows . in section three ,
we introduce our driving disaster management example named the uncertainty modelling of rainfall - runoff model .
section four describes the real time resource manager ( rtrm ) that monitors the sensors and knobs of the platform .
section five shows results of possible optimal computation points that are obtained at the design time .
they are going to be used by the rtrm at runtime in near future .
finally , there is a section of conclusion and future work .
one of the goals for future hpc systems is to develop automated frameworks that use power and performance to make applications - aware energy optimizations during an execution .
techniques like dynamic voltage and frequency scaling ( dvfs ) for reducing the speed ( clock frequency ) in exchange for reduced power consumption are used @xcite . and power gating technique that allows reducing power consumption by shutting off the current to blocks of the circuit that are not in use .
different computations have different power requirements . for computations where the cpu is waiting for resources the frequency
can be reduced to lower power with minimal performance impact @xcite .
system scenarios classify system behaviours that are similar from a multidimensional cost perspective , such as resource requirements , delay and energy consumption , in such a way that the system can be configured to exploit this cost . at design - time
, these scenarios are individually optimized .
mechanisms for predicting the current scenario at run - time and for switching between scenarios are also derived .
these are derived from the combination of the behaviour of the application and the applications mapping on the system platform .
these scenarios are used to reduce the system cost by exploiting information about what can happen at run - time to make better design decisions at design - time , and to exploit the time - varying behaviour at run - time . while use - case scenarios classify the application s behaviour based on the different ways the system can be used in its over - all context , system scenarios classify the behaviour based on the multi - dimensional cost trade - off during the implementation @xcite . to get precise results as fast as possible
is the main requirement for our testing application with the risk of disaster scenario .
runing applications with such requirements is called urgent computing @xcite
. there are plenty of specifics for such kind of computing , e.g. need for infrequent , but massive computational performance ( hpc ) .
since reserving computing capacity only for urgent cases is economically inefficient , the usage of cloud computing can be the solution .
on the other hand , there are limitations for communication - intensive applications and guarantee the availability of the resources may be problematic .
the framework @xcite,@xcite is the core of a highly modular and extensible run - time resource manager which provides support for an easy integration and management of multiple applications competing on the usage of one ( or more ) shared mimd many - core computation devices .
the framework design , which exposes different plug in interfaces , provides support for pluggable policies for both resource scheduling and the management of applications coordination and reconfiguration .
applications integrated with this framework get a suitable instrumentation to support design - space - exploration ( dse ) techniques , which could be used to profile application behaviours to either optimize them at design time or support the identification of optimal qos requirements goals as well as their run - time monitoring .
suitable platform abstraction layers , built on top of gnu / linux kernel interfaces , allow an easy porting of the framework to different platforms and its integration with specific execution environments .
we use a common gnu / linux with an efficient run - time resource manager ( rtrm ) @xcite,@xcite that exploits a design - time exploration ( dse ) , which performs an optimal quantization of the configuration space of run - time tunable applications , identifying a set of configurations .
the configuration of a run - time tunable application is defined by a set of parameters .
some of them could impact the application behaviour ( for example the uncertainty of rainfall - runoff ( rr ) models can produce different accuracy for different situations ) while other have direct impact on the amount of required resources .
for example , the amount of monte carlo iterations in uncertainty modelling and the time between batches of simulations lead to different requirements for allocating system resources .
the model under study is a modular part of the floreon+ project @xcite .
the main objective of the floreon+ project is to create a platform for integration and operation of monitoring , modelling , prediction and decision support for disaster management .
modularity of the floreon+ system , which is developed for this science and research project , allows for simple integration of different thematic areas , regions and data .
this system is meant to be used for decision support in operative and predictive disaster management processes .
the central thematic area of the project is hydrologic modelling and prediction .
the system focuses on acquisition and analysis of relevant data in real time and application of prediction algorithms with this data .
the results are then used for decision support in disaster management processes by providing predicted discharge volumes on measuring stations and prediction and visualization of flood lakes in the landscape .
model in this study @xcite is used for short term predictions of hydrological situation on given rivers .
rainfall - runoff ( rr ) model is a dynamic mathematical model which transforms rainfall to flow at the catchment outlet .
the main purpose of the model is to describe rainfall - runoff relations of a catchment area .
common inputs of the model are precipitations on given stations and spatial and physical properties of the river catchment area .
common outputs are surface runoff hydro - graphs which depict relations between discharge ( q ) and time ( t ) .
catchment areas for the model used in this paper for experiments are parametrized by following values . manning s roughness coefficent ( n ) which approximates physical properties of the river basin and cn curve number which approximates geological properties of the river catchment area .
rainfall - runoff models are usually used for predicting behaviour of river discharge and water level and one of the inputs of rainfall - runoff models is the information about weather conditions in the near future .
these data are provided by numerical weather prediction models @xcite and can be affected by certain inaccuracy .
such inaccuracy can be projected into output of the model by constructing confidence intervals which provide additional information about possible uncertainty of the model output .
the confidence intervals are constructed by using iterative monte carlo method .
data sets are sampled from the model input space and used as input for a large number of simulations .
quantiles selected from the simulations output are merged with the original hydrograph .
precision of modelling of the model uncertainty can be positively affected by increasing the number of performed simulations .
the precision of the simulations can be determined by estimating the nash - sutcliffe ( n - s ) model efficiency coefficient between the original simulation output and one of the quantiles selected from the monte carlo results .
the n - s coefficient is often used for estimating precision of given model by comparing its output with measured data but can be successfully used for comparison of any two model outputs .
@xmath0 where : section [ results ] presents our study about the precision that the model provides and its relation with the amount of computation .
as we explain later , at design time , it is possible to simulate situations with different trade - offs in terms of quality of services ( qos ) , execution time and power - consumption .
then the rtrm choose the optimal working point depending on the application scenarios .
application scenarios @xcite describe different triggers and states of the application that influence the system responsiveness and operation ( e.g. critical flooding level , critical state of patient s health , voice & data , etc . ) .
based on these scenarios , the system can be in different states with different service level requirements that can also be translated to the parameters of the uncertainty modelling and its required resources .
shorter response time in critical situations can for example be acquired by decreasing the number of monte carlo iterations ( also decreasing the precision of the results ) , or by allocating more computational resources ( maintaining the same precision level if the load is rebalanced appropriately ) .
we identified two main application scenarios that support the different workload of the system based on the flood emergency situation . in this scenario , weather is favourable and the flood warning level is below the critical threshold .
the computation can be relaxed ; some level of errors and deviations can be allowed .
the system should only use as much power as needed for standard operation ; one batch of rr simulations with uncertainty modelling only has to be finished before the next batch starts .
the results do not have to be available as soon as possible , so no excess use of resources is needed .
several days of continuous rain raise the water in rivers or a very heavy rainfall on a small area creates new free - flowing streams .
these conditions are signalled by the discharge volume exceeding the flood emergency thresholds or precipitation amount exceeding the flash flood emergency thresholds .
much more accurate and frequent computations are needed in this scenario and results has to be provided as soon as possible even if excess resources have to be allocated .
the development of the runtime reconfigurable application is facilitated using a real time library ( rtlib ) .
rtlib provides the abstract execution model ( aem ) .
the aem encapsulates the execution flow of the application in a way that let the real time resource manager rtrm @xcite to govern its life - cycle .
generally , we can structure an application by splitting it in more execution contexts ( exc ) . from the rtrm point of view
an exc is seen as an schedulable task .
having more exc in the same application can come from the need of schedule parts of the application with different priorities and resource usages .
this is made possible using the recipe that is associated to the execution contexts @xcite . from the application side , the mandatory requirement to exploit
the rtrm is to provide an implementation that allows it to be run - time reconfigurable .
from the rtrm side , the configuration of an application is characterized in terms of set of resource requirements , and is called application working mode ( awm ) .
the application must declare to the rtrm the list of its application working modes .
such list must be provided through a file called , in jargon , recipe which basically is an xml file .
[ h_flow ] summarizes the abstract execution model , where the application , onconfigure , workload terminated ? and qos - ok ?
boxes are related to the member methods of the derived class .
as we can see , the aem configures a sort of state diagram .
thus the methods must implement what our execution context must do whenever reach the corresponding execution state .
initialization of the framework is performed in the _ onsetup _ function . in order to allow the rtrm to perform a correct initialization of the structures supporting the collection of the runtime statistics
, this method performs the initialization of the structure named `` val '' .
it has all the input values that can modify the uncertainty simulation .
the description of the values is : * _ cn , n range and precision _ : probability distribution function and its parameters used for generating samples of the cn number and manning s rougness coefficient values . * _ input rain data _ : the data about rainfall on the land region under study received from the weather stations .
new data is received from there each ten minutes to one hour . * _ threads for uncertainty _ : the number of ( openmp ) threads equivalent to the number of monte carlo executions of the rainfall - runoff model that runs in parallel to simulate the uncertainty .
threads for rainfall - runoff model : the number of ( openmp ) threads that parallelize the main loops of the rainfall - runoff model itself . called when the rtrm is assigned for the first time , or the application working mode ( awm ) has changed , i.e. , the set of resources allocated for the application / exc .
the code of this method is related to whatever is required to reconfigure the application ( number of threads , application parameters , data structures , etc . ) to properly run according to resources assigned through the awm .
this is the entry point of the task . in our case
the uncertainty simulation .
there is a call to our wrapped main method called _ floreon - model_. it has a structure ( named _ val _ ) as an input , which provides input parameters that can vary ( modify ) the performance of the model .
@xmath1;\ ] ] where @xmath2 _ e - precision(nash - sutcliffe model efficiency coefficient ) , t - time _ after a computational run , the application may check whether the level of qos / performance / accuracy is acceptable or not . in the second case ,
some action could be taken .
this method is in charge to monitor the results received from the uncertainty model .
it decides if the current awm is enough to compute next uncertainty iteration or the awm and input data ( val struct ) has to be modified to correspond the new situation at runtime .
optional , but recommended member function .
this environment should eventually contain clean - up of the framework and applications .
this is the exit method to execute when the computation finalizes .
it is the end of the program and the use must be an user decision .
the system eventually does not have to arrive to this point since the system running 24/7 is expected .
there are two platforms that serve for development , testing and integration of the example application exposed above .
first is a 48 core smp machine hp proliant dl785 g6 @xcite .
a gnu / linux version 3.13.0 - 36-generic with a gcc version 4.8.2 is installed on this machine .
it consists of 8 sockets with six - core amd opteron(tm ) processor 8425 he 64bits 2.1@ghz clock frequency .
each core has a l1 cache memory size of 64 kb data and 64 kb for instructions and 512 kb l2 cache memory and a tlb of 1024 kb .
each cpu socket has an embedded memory 5118 kb l3 .
the machine is equipped with 256 gb ddr2 533 mhz with ecc memory .
the rtrm system uses resources of the platform , 8 sockets each with six cores in total .
one of the cores of each socket , six in total , named hosts , are allocated to monitor the system and the sensors of the platform ( i.e. power , temperature etc . ) by the rtrm system and to run all non - managed applications and processes . therefore , there are just 42 real cores for executing managed applications with the rtrm support .
second platform is hpc cluster anselm , which consists of 209 compute nodes , totaling 3344 compute cores .
each node is a powerful x86 - 64 computer , equipped with 16 cores and at least 64 gb ram .
nodes are interconnected by fully non - blocking fat - tree infiniband network and equipped with intel sandy bridge processors .
the cluster runs bullx gnu / linux operating system , which is compatible with the redhat gnu / linux family .
there is no rtrm support meantime , but it is part of future work .
after integration of the uncertainty modelling module into the rtrm on smp machine , we first computed how the coefficient to compute the precision of the simulations decrease with the number of iterations . this coefficient is computed and cumulated in each iteration and heads to the zero at infinity .
lower value of the coefficient means higher precision .
these precisions for given numbers of iterations as fig .
[ precision ] depicts are resources independent and will be used to identify optimal working modes .
[ native ] presents energy consumption of native , which means unmanaged by rtrm , executions of uncertainty simulations with a galaxy of points .
values are only estimated because of absence of hw support for power consumption measurement of the smp machine .
these points represent simulations with different precision - number of iterations ( e.g. 100 low precision , 1000 medium precision or 10000 high precision ) , different cpu frequencies and various number of cores which results in different execution times and appropriate consumed energy .
this leads to trade - offs for running the system with various amount of resources ( cores , frequencies ) .
[ bbque ] shows similar results but instead of native execution the simulations run with the rtrm .
dynamic part ( only used cores ) of estimated power consumption was computed using original formula : @xmath3 where : @xmath4 is a system load , @xmath5 is a power consumption at this load , @xmath6 is a maximum power consumption , @xmath7 is a power consumption in an idle .
estimated energy consumption @xmath8 is then computed from the formula : @xmath9 where : @xmath10 is estimated power consumption and @xmath11 is time of execution .
we have observed that the overhead of rtrm in time is below 10% in average , which is shown by fig .
we have to take into account that from each socket ( six x86 - 64 cores ) of the platform , one core is used to monitor the healthy of the system .
so , from 48 possible cores only 42 are enabled to run our uncertainty module . from these points
is possible to extract which are the optimal - pareto points that are going to be inserted in the rtmr recipe . for comparison , power consumption of native executions of uncertainty simulations on hpc cluster ( multi - node solution using mpi+openmp , 10000 iterations )
are shown on fig .
values were measured using the likwid ( likwid - powermeter ) tool for different numbers of nodes .
the graph shows steeply increasing power consumption with no significant time improvement using more than sixteen 16-cores computing nodes .
the paper presents an environment that helps to an x86 - 64 architecture platform to run 24/7 a model saving energy with the trade - off of losing precision with the different resources configurations .
the paper shows the uncertainty modelling executed with different steps of precision given by various numbers of iterations leading in appropriate execution times .
the decision of precision / executed time / consumed energy is taken by a run - time resource manager monitoring the workload of the platform and computing the best - trade off with the time overhead less than 10% in against a native execution . at design time , it is possible to describe the optimal points that are going to be used at runtime .
the environment also permits to execute the module 24/7 in nodes of a cluster and/or with accelerators .
this enables to use more resources and higher precision for the future .
this article was supported by operational programme education for competitiveness and co - financed by the european social fund within the framework of the project new creative teams in priorities of scientific research , reg .
cz.1.07/2.3.00/30.0055 , by the european regional development fund in the it4innovations centre of excellence project ( cz.1.05/1.1.00/02.0070 ) , by the project large infrastructures for research , development and innovation of ministry of education , youth and sports of czech republic with reg .
lm2011033 , and by 7^th^ eu framework programme project no . fp7 - 612069 harpa - harnessing performance variability .
theuer m. , et al .
, efficient methods of automatic calibration for rainfall - runoff modelling in the floreon+ system , international journal on non - standard computing and artificial intelligence ( accepted ) , 2014 golasowski m. et al . :
uncertainty modelling in rainfall - runoff simulations based on parallel monte carlo method , international journal on non - standard computing and artificial intelligence ( accepted ) 2014 .
|
power consumption is a critical consideration in high performance computing systems and it is becoming the limiting factor to build and operate petascale and exascale systems .
when studying the power consumption of existing systems running hpc workloads , we find that power , energy and performance are closely related which leads to the possibility to optimize energy consumption without sacrificing ( much or at all ) the performance . in this paper , we propose a hpc system running with a gnu / linux os and a real time resource manager ( rtrm ) that is aware and monitors the healthy of the platform . on the system ,
an application for disaster management runs .
the application can run with different qos depending on the situation .
we defined two main situations .
normal execution , when there is no risk of a disaster , even though we still have to run the system to look ahead in the near future if the situation changes suddenly . in the second scenario ,
the possibilities for a disaster are very high .
then the allocation of more resources for improving the precision and the human decision has to be taken into account .
the paper shows that at design time , it is possible to describe different optimal points that are going to be used at runtime by the rtrm with the application .
this environment helps to the system that must run 24/7 in saving energy with the trade - off of losing precision .
the paper shows a model execution which can improve the precision of results by 65% in average by increasing the number of iterations from 10 ^ 3^ to 10 ^ 4^. this also produces one order of magnitude longer execution time which leads to the need to use a multi - node solution . the optimal trade - off between precision vs. execution time
is computed by the rtrm with the time overhead less than 10% against a native execution .
[ pacs : 89.20.ff , 47.50.cd , 92.40.-t , 07.05.bx , 07.05.tp , 89.20.ff ]
|
a cyber - physical system ( cps ) is a mechanism controlled or monitored by computer - based algorithms
. examples of cps include smart grid , autonomous automobile systems , medical monitoring , process control systems , distributed robotics , and automatic pilot avionics , etc .
the connection between the cyber part and the physical part heavily relies on communication networks , which has been raising a major security concern , as different types of cyber attacks can tamper the data collection processes and interfere safety critical decision making processes , which may cause irreparable damadges to the physical systems being controlled and to people who depend on those systems .
+ there has been a growing number of publications addressing the cyber security issues from both the computer science community , which focuses on the computer computation related issues , and the systems control community , which focuses on issues related to the system dynamics affected by cyber attacks .
recently , more and more efforts have been made in classifying different types of malicious attacks , assuming that the attackers are sufficiently intelligent @xcite @xcite , instead of merely just generating random failures , which is well studied in the fields of reliability and fault tolerant control .
typically , an intelligent attacker requires _
system knowledge _ , and abilities for _ resource disclosure _ and _ resource disruption _ in order to carry out a successful attack , which is covert to a system user until the attacker s goal of causing a damage to the system is achieved .
so _ covertness _ and _ damage infliction _ are two major characteristics of a successful attack . by analyzing different intelligent cyber attacks
, proper countermeasures may be developed to prevent a target system from being harmed by a specific type of attacks .
+ in this paper we study a special type of data deception attacks in the discrete - event system framework , where an attacker can intercept sensor measurements ( or observations ) modeled by observable events and alter them arbitrarily but with an upper bound imposed on the length of each altered observation sequence . by sending those altered observation sequences to a given supervisor
, whose function is known to the attacker in advance , the attacker can deliberately and covertly guide the system to move into some undesirable states without making any change to the supervisor .
the key challenge is how to `` fool '' the supervisor to make it believe that the system is operating correctly , while using the supervisor s own control functions to carry out the attack , i.e. , to lead the system move into a bad state . to this end , we first propose the concept of _ attack under bounded sensor reading alterations _
( absra ) , which can be modelled as a finite - state transducer , possessing the properties of covertness , damadge infliction and control feasibility under partial observations .
then we show that the supremal ( or least restrictive ) absra exists and is computable via a specific synthesis algorithm , as long as both the plant model @xmath0 and the given supervisor @xmath1 are finitely representable . upon this novel absra synthesis algorithm
, we present a supervisor synthesis algorithm , which can ensure that a nonempty synthesized supervisor will be `` robust '' to any absra , in the sense that such an attack will either reveal itself to the supervisor due to abnormal system executions ( so that proper contingent actions can be taken by the supervisor , which is nevertheless outside the scope of this paper ) or will not be able to lead the system to a bad state ( i.e. , no damadge will be inflicted ) .
+ our construction of an absra model as a transducer is inspired by some recent work on opacity enforcement @xcite , which aims to use observable event insertions to prevent a potential attacker from correctly determining the actual state of a target system .
due to the different objectives of two works , the modeling details and synthesis algorithms are completely different .
there have been some works on cyber attack detection and prevention in the discrete - event community @xcite @xcite @xcite , mainly from an adaptive fault tolerant control point of view , which heavily rely on real - time fault diagnosis to identify the existence of an attack and then take necessary robust or adaptive supervisory control actions . in those works the intelligence of an attacker
is not considered , and an attack is treated as a fault . as a contrast , we do not rely on real time attack detection , but rely on prior knowledge of attack models , and simply build attack - robustness features into a supervisor to ensure that the supervisor will not be affected by any absra unnoticeably .
it is this robust control nature distinguishes our works from existing des - based cyber attack detection and prevention approaches , which fall in the adaptive control domain . +
the remainder of the paper is organized as follows . in section
ii we review the basic concepts and operations of discrete event systems , and formulate an absra synthesis problem , which is then solved in section iii , where we show that the supremal absra exists and computable . in section iv we present an algorithm to synthesize a supervisor , which is robust to any absra .
a simple yet realistic example runs through the entire paper to illustrate all relevant concepts and algorithms .
conclusions are drawn in section v.
in this section we first recall some standard concepts used in the ramadge - wonham supervisory control paradigm .
then we introduce the concept of absra , followed by a concrete absra synthesis algorithm , which reveals that the supremal absra is computable , as long as both the plant model and the given supervisor are regular . given an arbitrary finite alphabet @xmath2 , let @xmath3 be the free monoid with the empty string @xmath4 being the unit element and the string concatenation being the monoid operation .
given two strings @xmath5 , @xmath6 is called a _
prefix substring _ of @xmath7 , written as @xmath8 , if there exists @xmath9 such that @xmath10 , where @xmath11 denotes the concatenation of @xmath6 and @xmath12 .
any subset @xmath13 is called a _
language_. the _ prefix closure _ of @xmath14 is defined as @xmath15 . given two languages @xmath16 , let @xmath17 denote the concatenation of two sets .
let @xmath18 .
a mapping @xmath19 is called the _
natural projection _ with respect to @xmath20 , if 1 .
@xmath21 , 2 .
@xmath22 3 .
@xmath23 . given a language @xmath13 , @xmath24 .
the inverse image mapping of @xmath25 is @xmath26 given @xmath27 and @xmath28 , the _ synchronous product _ of @xmath29 and @xmath30 is defined as @xmath31 , where @xmath32 and @xmath33 are natural projections .
clearly , @xmath34 is commutative and associative . +
a given target plant is modelled as a _
deterministic finite - state automaton _ , @xmath35 , where @xmath36 stands for the state set , @xmath2 for the alphabet , @xmath37 for the ( partial ) transition function , @xmath38 for the initial state and @xmath39 for the marker state set .
we follow the notation system in @xcite , and use @xmath40 to denote that the transition @xmath41 is defined . for each state
@xmath42 , let @xmath43 be the set of events enabled at @xmath44 in @xmath0 .
the domain of @xmath45 can be extended to @xmath46 , where @xmath47 for all @xmath42 , and @xmath48 .
the _ closed _ behavior of @xmath0 is defined as @xmath49 , and the _ marked _ behavior of @xmath0 is @xmath50 .
@xmath0 is _ nonblocking _ if @xmath51 .
we assume that the marker state set @xmath52 is partitioned into two disjoint sets @xmath53 , where @xmath54 is the set of _ desirable _ states and @xmath55 denotes the set of _ bad _ states .
+ we now recall the concept of supervisors .
let @xmath56 , where disjoint @xmath57 ( @xmath58 ) and @xmath59 ( @xmath60 ) denote respectively the sets of _ controllable _ ( _ observable _ ) and _ uncontrollable _ ( _ unobservable _ ) events , respectively .
let @xmath61 be the collection of all _ control patterns_. a _
( feasible ) supervisory control map of @xmath0 under partial observation @xmath62 _ is defined as @xmath63 , where @xmath64 for each @xmath65 , @xmath66 is interpreted as the set of events allowed to be fired after @xmath6 .
thus , a supervisory control map will not disable any uncontrollable events , and will impose the same control pattern after strings , which can not be distinguished based on observations .
let @xmath67 denote the closed - loop system of @xmath0 under supervision of @xmath68 , i.e. , * @xmath69 , * @xmath70 , * @xmath71 .
the control map @xmath68 is _ finitely representable _ if @xmath67 can be denoted by a finite - state automaton , say @xmath72 such that * @xmath73 and @xmath74 , where ` @xmath34 ' is automaton product @xcite , * @xmath75 , * @xmath76 .
it has been shown that , as long as a closed - loop language @xmath77 is _ controllable _ @xcite and _ observable _ @xcite , there always exists a finitely - representable supervisory control map @xmath68 such that @xmath78 and @xmath79 . from now on we assume that @xmath67 is finitely representable by @xmath1 , which is called a _ supervisor_. we assume that @xmath1 is _ legal _ in the sense that @xmath80 , i.e. , under the supervision of @xmath1 , the plant @xmath0 should never enter any bad marker state .
we assume that an attacker can intersept each observable event generated by the plant @xmath0 , and replace it by a sequence of observable events from @xmath58 in order to `` fool '' the given supervisor @xmath1 , whose function is known to the attacker . considering that in practice any event occurance takes an unnegligible amount of time , it is impossible for an attacker to insert an arbitrarily long observable sequence to replace a received observable event . for this reason , we assume that there exists a known natural number @xmath81 such that the length of any observable sequence that the attacker can insert is no more than @xmath82 .
let @xmath83 be the set of all such bounded observable sequences , where @xmath84 denotes the length of @xmath6 , and by convention , @xmath85 .
we model a sensor attack as a finite state transducer @xmath86 , where @xmath87 is the state set , @xmath2 the input alphabet , @xmath88 the output alphabets , @xmath89 the initial state , @xmath90 the marker state set , which is specifically set as @xmath91 , and @xmath92 the ( partial ) transition map , where for all @xmath93 and @xmath94 , @xmath95 , i.e. , at each state @xmath96 all unobservable events are self - looped with @xmath4 as the output .
this is natural because an attacker can only observe observable events , thus , will not make any move upon unobservable events . we still keep unobservable events here to make it easy for us for subsequent technical development
clearly , @xmath97 .
let @xmath98 and @xmath99 be the _ input _ and _ output _ maps , respectively , where for each @xmath100 , @xmath101 and @xmath102 .
+ the basic procedure of an attack is to intercept every single observable event @xmath103 generated by the plant @xmath0 , replace it with some observable string @xmath104 , and send @xmath12 to the supervisor @xmath1 , in order to trick @xmath1 to issue a control command @xmath105 that may drive the plant @xmath0 towards a bad marker state .
this attack procedure is depicted in figure [ fig : cyber - security-0 ] .
the sequential composition of the attack @xmath106 and the supervisor @xmath1 essentially forms a new supervisor , denoted as @xmath107 , which receives an observable output @xmath103 and generates a control command @xmath105 .
the exact definition of this new supervisor reveals the nature of the attack , which is given below . the _ sequential composition _ of @xmath106 and @xmath1 is a deterministic finite state transducer @xmath108 , where @xmath109 denotes the deadlocking dump state , and for each @xmath110 , @xmath111 , @xmath112 if one of the following holds , * @xmath113 , * @xmath114 . for each @xmath115 , @xmath111 , @xmath116 if @xmath117 but @xmath118 is undefined .
thus , all transitions that go to the dumpt state @xmath109 may potentially reveal the attack , which , for an intelligent attack , should be avoided .
+ the _ impact _ of @xmath106 on the closed - loop system @xmath119 is captured by the composition of the plant @xmath0 and the new supervisor @xmath107 , i.e. , @xmath120 where for each @xmath121 , @xmath111 , @xmath122 if @xmath123 and @xmath124 . clearly , @xmath125 is also a transducer , and it is not difficult to check that @xmath126 where `` = '' is in the sense of des - isomorphism , and @xmath127 denotes a function mapping one transducer to another transducer by simply marking every state . in other words , if @xmath106 is an attack model for the system @xmath119 , then @xmath128 is also an attack model , which has the same attack effect as that of @xmath106 on @xmath119 . since @xmath129 , we call @xmath130 a _ canonical attack _ with respect to @xmath119 .
since for any attack , there exists a canonical attack , which has the same attack effect , from now on we only focus on canonical attacks . on the other hand , we will see that @xmath106 usually is stucturally simpler than its canonical one @xmath130 , whereas the latter is easier to compute .
an interesting question is how to synthesize a simplified attack model @xmath106 from a given canonical attack model @xmath130 , which bears some similarity to the problem of supervisor reduction @xcite , and will be addressed in our future works .
+ to illustrate the aforementioned concepts , let us go through a simple single - tank example depicted in figure [ fig : cyber - security-1 ] , which consists of one water supply source whose supply rate is @xmath131 , one tank , and one control valve at the bottom of the tank controlling the outgoing flow rate @xmath132 , whose value depends on the valve opening and the water level @xmath133 .
we assume that the valve can only be fully open or fully closed to simplify our illustration , and in case of a full opening , the water level @xmath133 can only go down .
the water level @xmath133 can be measured , whose value can trigger some predefined events , denoting the water levels : _ low _ ( h = l ) , _ medium _ ( h = m ) , _ high _ ( h
= h ) , and _ extremely high _ ( h = eh ) .
we construct a simple discrete - event model of the system depicted in figure [ fig : cyber - security-2 ] , , scaledwidth=90.0% ] where the alphabet @xmath2 contains all events shown in the figure .
all events are observable , i.e. , @xmath134 . only the actions of opening the valve ( @xmath135 ) and closing the valve ( @xmath136 ) are controllable , and all water level events are uncontrollable . in the model
we use a shaded oval to denote a marker state , i.e. , state 5 and state 9 in figure [ fig : cyber - security-2 ] .
assume that we do not want the water level to be extremely high , i.e. , the event h = eh should not occur .
thus , state 9 is a bad marker state , i.e. , @xmath137 and @xmath138 . to prevent state 9 from being reached ,
we compose a requirement @xmath139 shown in figure [ fig : cyber - security-3 ] , whose alphabet is @xmath140h = l , h = m , h = h , h = eh@xmath141 , but the event h = eh is never allowed in the model .
a supervisor @xmath1 can be synthesized by using the standard ramadge - wonham supervisory control paradigm , which is also depicted in figure [ fig : cyber - security-3 ] .
( left ) and the supervisor @xmath1 ( right),scaledwidth=90.0% ] it is clear that the supervisor @xmath1 only opens the valve when the water level is high , i.e. , it disables the event @xmath136 at state 6 when the event h = h occurs .
our intuition tells us that if an attack always change events of h = m , h = h , h = eh to the event h = l , then the supervisor will not prevent the water level from reaching the extreme high level , i.e. , the event h = eh will happen . for this reason , we conjecture an attack model @xmath106 shown in figure [ fig : cyber - security-4 ] , where water levels will be altered to h = l , whereas all other events will remain unchanged .
( left ) and the sequential composition @xmath107 ( right),scaledwidth=60.0% ] the sequential composition @xmath107 indicates that , no matter which water level is reached , the attack @xmath106 always sends h = l to the supervisor @xmath1 , which tricks it to believe that it is safe to allow the valve to be either closed or opened .
the impact of @xmath106 on the closed - loop system @xmath119 is depicted in figure [ fig : cyber - security-5 ] .
, scaledwidth=90.0% ] by marking every state in @xmath142 we obtain a canonical attack model @xmath143 .
+ [ prop0](1 ) @xmath144 ; ( 2 ) @xmath145 ; ( 3 ) @xmath146 , @xmath147.@xmath148 proof : by the above definition of sequential composition , the proposition follows.@xmath149 + [ prop1]given two attacks @xmath150 and @xmath151 with the same input alphabet @xmath2 and output alphabet @xmath88 , assume that @xmath152
. then we have @xmath153.@xmath148 proof : by the above definition of sequential composition , the proposition follows.@xmath149 + given two attacks @xmath150 and @xmath151 with the same input alphabet @xmath2 and output alphabet @xmath88 , let @xmath154 be their union , which is a deterministic finite - state transducers . then by the definition of transducer union , we have @xmath155 . + [ prop2]@xmath156.@xmath148 proof : since @xmath157 and @xmath158 , by prop .
[ prop1 ] we have @xmath159 . to show the other direction , for each string @xmath160 , by the definition of the sequential composition , we know that @xmath161 .
thus , either @xmath162 or @xmath163 , which means either @xmath164 or @xmath165 .
thus , @xmath166 , which concludes the proof.@xmath149 + so far we have introduced a simple sensor attack model , and explained how this attack affects the closed - loop system .
but we have not described what kind of sensor attacks can be considered intelligent .
next , we will introduce the concept of absra .
let @xmath62 be the natural projection .
an intelligent canonical attack needs to possess the following properties : 1 .
its insertions must be covert to the given supervisor , i.e. , @xmath167 namely the supervisor will not see any unexpected observable sequences from the attack .
2 . any of its insertion sequence may potentally cause damages to @xmath0 , i.e. , @xmath168 namely any sequence of insertions by the attack will cause @xmath0 to reach some bad state eventually .
a weaker version of this property is described below : @xmath169 which says that the attack @xmath106 will tamper the absolute correctness of the supervisor @xmath1 so that there exists some possibility that the system may reach some bad marker state .
3 . @xmath107 forms a standard supervisor for the plant @xmath0 that enforces normality @xcite , i.e. , @xmath170 and @xmath171 which denotes that at each state the attack will not intervene the event enablement by the supervisor because we consider only sensor attacks , not actuator attacks .
we call a nonempty model @xmath106 satisfying the aforementioned four properties ( 1)-(4 ) an _ attack with bounded sensor reading alterations _ ( absra ) of @xmath119 . + it is not difficult to check that the attack @xmath106 shown in figure [ fig : cyber - security-4 ] does not satisfy property ( 1 ) because @xmath1 can not fire @xmath136 before h = l , but @xmath106 can .
nevertheless , the sequential composition @xmath107 satisfies all three properties , thus , is an absra . by the aforementioned discussions , we know that the canonical attack model @xmath172 is also an absra . + [ thm1]given a plant @xmath0 and a legal supervisor @xmath1 , let @xmath173 be a ( possibly infinite ) collection of absra s with respect to @xmath119
. then @xmath174 satisfies properties ( 1)-(4).@xmath148 proof : by prop .
[ prop2 ] , we know that @xmath175 .
we now verify that @xmath174 satisfies all four properties .
+ ( a ) since for each @xmath176 , @xmath177 is an absra , we have that @xmath178 thus , by prop .
[ prop2 ] we have that @xmath179 ( b ) in addition , we have that for each @xmath176 , @xmath180 thus , by prop .
[ prop2 ] we have @xmath181 ( c ) since for each @xmath176 , we have @xmath182 we get @xmath183 the last property ( 4 ) can be easily checked .
thus , @xmath174 satisfies all four properties , and the theorem follows.@xmath149 + theorem [ thm1 ] only implies that the least restrictive ( or supremal ) attack language exists .
but it is not clear whether this supremal language is regular , i.e. , whether it can be recognized by a finite - state transducer .
therefore , at this moment the existance of the supremal absra is still unknown .
we now state our main problem in this paper . + given a plant @xmath0 and a legal supervisor @xmath1 , design an absra @xmath106.@xmath148
+ in the next section we will show that the supremal attack language is regular , i.e. , indeed the supremal absra exists , and is computable .
we first recall the concepts of controllability @xcite , and normality @xcite . because we deal with both finite - state automata and finite - state transducers , to make notations simple , we introduce a general purpose alphabet @xmath184 , which can be either @xmath185 or @xmath186 , depending on a specific application context .
let @xmath187 and @xmath188 be an uncontrollable alphabet and an observable alphabet respectively , where if @xmath189 then @xmath190 .
let @xmath191 be the natural projection . in case that @xmath189 , we have @xmath192 , @xmath193 if @xmath103 , or @xmath4 otherwise , and @xmath194 . when we mention a finite - state transitional structure @xmath195 , we mean that @xmath195 is either a finite - state automaton or a finite - state transducer . + [ def1]given a finite - state transitional structure @xmath195 , a sublanguage @xmath196 is _ controllable _ w.r.t . @xmath195 and
@xmath197 , if @xmath198.@xmath148 + [ def2]given a finite - state transitional structure @xmath195 , a sublanguage @xmath196 is _ normal _ w.r.t . @xmath195 and
@xmath199 , if @xmath200.@xmath148 + given a finite - state transitional structure @xmath195 , whose alphabet is @xmath184 , and a requirement @xmath201 , let @xmath202 by an argument similar to the one used in @xcite , we can derive that the supremal controllable and normal sublanguage of @xmath203 exists , denoted as @xmath204 , such that for all @xmath205 , we have @xmath206 .
+ in our setup , an attack is able to arbitrarily alter an observable event .
thus , each event @xmath111 is considered controllable , as the attack can choose not to use this alteration . under this consideration , the uncontrollable alphabet @xmath197 is actually empty .
thus , in the following attack model synthesis , we do not explicitly require controllability .
this may sound a bit unusual because we do have an uncontrollable alphabet @xmath59 for the plant @xmath0 - how those uncontrollable events affect the attack model synthesis ?
if we carefully check the properties of an absra , we can see that property ( 4 ) actually implicitly enforces controllability with respect to @xmath59 because it requires the attack not to change the event enablement of the supervisor @xmath1 at the current state , and since by default the supervisor @xmath1 ensures controllability with respect to @xmath59 , and so does the attack model .
+ assume that there exists @xmath207 , which denotes a set of _ protected _ observable events that can not be altered by an absra , i.e. , given an attack model @xmath86 , for all @xmath94 , @xmath208 , we have that @xmath209 .
we now undertake the following absra synthesis procedure . + * procedure 1 : ( absra synthesis ) * 1 .
input : a plant @xmath35 , a supervisor @xmath210 and @xmath211 .
2 . construct a single - state transducer @xmath212 , where @xmath213 and the transition map @xmath214 encodes transitions labeled by a subset of @xmath215 , denoting all observable event alterations that the attack wants to consider .
3 . let @xmath216 be a requirement .
4 . undertake the following iteration on @xmath217 1 .
compute @xmath218 .
2 . check property ( 4 ) in the definition of absra .
if it holds , then go to step 5 ) . otherwise , set @xmath219 and continue the iteration on @xmath220 .
output : @xmath221 , which recognizes @xmath222.@xmath148 + procedure 1 terminates finitely.@xmath148 proof : assume that @xmath223 is recognized by a transducer @xmath224 , whose state set is @xmath225 .
then @xmath226 is recognizable by a transducer , say @xmath227 , whose state set is a subset of @xmath228 .
it is not difficult to check that for all @xmath229 , if they hit the same state in @xmath227 , then we know that @xmath230 if and only if @xmath231 in other words , @xmath232 if and only if @xmath233 . thus , for each state in @xmath227 , either all strings hitting that state are in @xmath234 or none of them are in @xmath234 , namely @xmath234 is recognized by a sub - transducer @xmath235 of @xmath227 .
suppose the state set of @xmath235 is @xmath236 . by the property of automaton composition ,
we know that there exists a transducer @xmath237 recognizing @xmath238 such that the state set of @xmath237 is a subset of @xmath239 . since @xmath236 , we know that @xmath237 is des - isomorphic to a sub - transducer of @xmath227 . by using the same argument , we can check that each @xmath240 is recognized by a transducer , which is des - isomorphic to a sub - transducer of @xmath227 .
in addition , the state sets of those sub - transducers form a monotonic non - increasing sequence with respect to set inclusion . thus , in a finite number of iterations , a fixed sub - transducer will be reached , whose language is @xmath240 .
this means procedure 1 must terminate finitely.@xmath149 + [ lem1]let @xmath240 and @xmath221 be computed in procedure 1 . then @xmath241.@xmath148 proof : by the proof of lemma 1 we know that @xmath221 is des - isomorphic to the prefix closure of a sub - transducer of @xmath242 . then by the definitions of sequential composition and transducer product , the lemma follows .
@xmath149 + @xmath221 obtained in procedure 1 is the supremal absra of @xmath119 .
@xmath148 proof : ( a ) we first show that @xmath221 is an absra , i.e. , @xmath221 satisfies properties ( 1)-(4 ) .
it is clear that when the algorithm terminates , property ( 4 ) must hold .
so we only focus on properties ( 1)-(3 ) . by the definition of @xmath221 and prop . 1
, we know that @xmath243 by lemma 2 , we have @xmath244 . since @xmath245 and @xmath246 , by prop .
1 we have @xmath247 for the third property , by lemma 2 we know that @xmath248 .
thus , by the definition of normality and the nonblocking property associated with a sub - transducer of @xmath242 , which recognizes @xmath240 , we have @xmath249 this concludes our proof that @xmath221 is an absra . + ( b ) next , we show that @xmath221 is the supremal absra .
let @xmath130 be an absra of the system .
clearly , @xmath250 . due to the controllability of @xmath1 and the assumption that @xmath130 is an absra , i.e. , it must satisfy property ( 4 ) , it is easy to check that @xmath251 is controllable w.r.t .
@xmath252 and @xmath59 . since @xmath130 must satisfy property ( 3 ) , we know that @xmath251 must be normal w.r.t . @xmath252 and @xmath58
in addition , @xmath130 satisfies property ( 4 ) .
thus , we can easily detive that @xmath253 , namely , @xmath254 .
this means @xmath255 , which concludes the proof of the theorem .
@xmath149 + as an illustration , we apply procedure 1 to the plant @xmath0 shown in figure [ fig : cyber - security-2 ] and the supervisor @xmath1 shown in figure [ fig : cyber - security-3 ] .
we can see that the sensor attack model @xmath106 in figure [ fig : cyber - security-4 ] is actually @xmath256 in procedure 1 because all events in the model are observable . the composition @xmath257 is shown in figure [ fig : cyber - security-4 ] .
the outcome of @xmath258 is shown in figure [ fig : cyber - security-5 ] , which is isomorphic to @xmath0 .
this is not surprising because any string in @xmath259 may be potentially extended to the bad marker state .
the requirement @xmath260 is simply the same as @xmath258 . clearly , we know that @xmath240 is recognizable by a transducer shown in figure [ fig : cyber - security-6 ] , which is almost the same as @xmath258 , except that the only marker state is that bad marker state due to the requirement @xmath260 .
, scaledwidth=90.0% ] since all events are observable , from @xmath261 we can derive that @xmath262 .
thus , @xmath221 can be chosen by marking every state in @xmath258 , i.e. , @xmath263 , which means @xmath221 is actually a canonical attack of @xmath256 with respect to @xmath119 . by theorem 2 we know that @xmath221 is the supremal absra of @xmath119 .
in the previous section we discuss how to design an absra model to interrupt a given system s operations from an attacker s point of view . in this section
we present a synthesis approach to design a supervisor , which is `` robust '' to any absra in the sense that either the attack is not covert or incurs no damage to the system .
+ recall that an absra affects a target system @xmath119 by altering the sequence of observable events , which tricks @xmath1 to issue commands improperly . by protecting observable events from being altered unnoticeably can in principle effectively deter an absra .
an observable event in this framework denotes a specific set of strongly associated measurements .
for example , in the aforementioned single - tank system , the event h = h may either be associated with one simple water level measurement or possibly several sensor measurements such as the actual water level , and the corresponding pressure on the bottom of the tank - the more sensor measurements associated with the event , the harder for an attack to alter the event without being detected . when applying suitable encryption techniques , it is even more complicated for an attack to complete the job .
thus , it is indeed technically feasible to prevent observable events from being altered by either adopting new secure information transmission technologies or introducing more sensors to significantly increase the complication of altering the corresponding observable event without being detected .
nevertheless , there is always a financial consideration .
an attractive solution to a potential industrial user is to identify only critical observable events , which , when being protected from external alterations , will lead to a supervisor robust to any absra .
+ given a plant @xmath0 , a requirement @xmath139 , and a protected observable alphabet @xmath207 , synthesize a supervisor @xmath1 such that there is no arsra of the closed - loop system @xmath119.@xmath148 + with the same notations used in the previous section , let @xmath264 be the collection of all controllable and normal supervisors @xcite .
let @xmath265 , which always exists and computable , as long as @xmath266 is regular .
our goal is to design a supervisor @xmath267 such that procedure 1 returns an ampty absra @xmath221 with respect to the given protected observable alphabet @xmath211 . to this end , we present the following synthesis procedure : + * procedure 2 : ( absra - robust supervisor synthesis ) * 1 .
input : a plant @xmath0 , a requirement @xmath139 and a protected observable alphabet @xmath211 .
2 . compute @xmath268 . if @xmath269 , terminate . otherwise , assume @xmath270 is recognized by a finite - state automaton @xmath271 , and continue
compute @xmath221 by using procedure 1 , i.e. , computer @xmath272 .
4 . compute @xmath273 .
5 . output : a recognizer @xmath1 of @xmath274.@xmath148 + [ thm3]given a plant @xmath0 , a requirement @xmath266 , and a protected observable alphabet @xmath211 , let @xmath1 be computed above . if @xmath275 , then we have @xmath276 , where @xmath277 is defined in procedure 1 , i.e. , there is no absra @xmath106 of @xmath119.@xmath148 proof : assume that it is not true .
then there exists an absra @xmath106 such that @xmath278 , where @xmath279 .
since @xmath1 is controllable and normal with respect to @xmath0 , and @xmath280 , we can get that @xmath281 . since @xmath272 , we know that @xmath282 .
but on the other hand , we know that @xmath283 , i.e. , @xmath284 , which leads to a contradiction .
thus , the absra @xmath106 does not exist .
@xmath149 + we would like to emphasize here again that , although theorem [ thm3 ] indicates that there is no absra @xmath106 for the closed - loop system @xmath285 , it does not mean that a sensor reading alteration attack will not be carried out by an attacker .
but such an attack will either reveal itself to the supervisor before it achieves its attack goal due to abnormal system executions ( so that proper contingent actions such as system shutdown can be taken by the supervisor , which is outside the scope of this paper ) or will not be able to lead the system to a bad state .
+ in theorem [ thm3 ] , if @xmath286 , then with the given protected observable alphabet @xmath211 , there does not exist a supervisor @xmath1 that is absra - robust .
we face the following synthesis problem . + given a plant @xmath0 and a requirement @xmath139 ,
compute a protected observable alphabet @xmath207 of the minimum size , which allows a nonempty absra - robust supervisor @xmath1 to exist.@xmath148 + it is clear that problem 3 is solvable in the sense that it is decidable whether there exists such a @xmath211 with the minimum size because we can simply enumerate each subset @xmath287 , and apply procedure 2 on @xmath211 to compute the corresponding supervisor @xmath1 . since there is a finite number of such subsets ,
this brutal - force method will terminate , and provide a protected observable alphabet of the minimum size together with the corresponding supervisor , if it exists .
the computational complexity of this procedure is certainly high , which is exponential to @xmath288 , but polynomial to the sizes of @xmath0 and @xmath139 due to our adoption of normality to handle observability . if the size of @xmath58 is big , to find a computationally viable algorithm that can solve problem 3 becomes important , which will be addressed in our future works .
+ we now use that simple single - tank system to illustrate how to use procedure 2 to compute an absra - robust supervisor , and how to determine a minimum protected observable alphabet , which allows the existence of an absra - robust supervisor .
let @xmath289h = h@xmath141 .
the model of @xmath256 and @xmath271 are shown in figure [ fig : cyber - security-7 ] . and
@xmath271,scaledwidth=90.0% ] when we run procedure 1 , we first compute @xmath290 .
the outcome is depicted in figure [ fig : cyber - security-8 ] .
( right ) and @xmath290 ( left),scaledwidth=100.0% ] we can see that @xmath290 contains no bad marker state in @xmath55 .
thus , in procedure 1 we have @xmath291 , which returns @xmath292 .
after that , in step 4 ) of procedure 2 , we have that @xmath293 .
thus , @xmath271 is an absra - robust supervisor for @xmath0 with respect to the given @xmath211 .
clearly , it is a solution to problem 3 because we can not find any other protected observable alphabet with a size smaller than 1 , which can render an absra - robust supervisor .
in this paper we have first introduced the concept of absra , upon which we have shown that the supremal absra exists and computable , as long as the plant model @xmath0 and the supervisor @xmath1 are finitely representable , i.e. , their languages are regular . after that , we have brought in the problem of synthesizing an absra - robust supervisor , and shows that it is possible to find a minimum protected observable sub - alphabet , which may render an absra - robust supervisor . + it is interesting to point out that , if we replace the third property of an absra model with a weaker observability property , e.g. , the standard observability @xcite , the supremal absra may not exist any more . nevertheless , the existence of an absra is still decidable and computable ( with possibly a higher computational complexity ) , as this absra synthesis problem is equivalent to a synthesis problem of centralized supervisory control under partial observation , which has been shown solvable @xcite .
fortunately , the normality property can be easily satisfied in reality , as it only requires that only observable and controllable events can be disabled in online applications - in real industrial applications , it is typical that all control commands are observable . for this reason ,
the supervisor synthesis approach proposed in this paper aiming to defy absra is practically feasible .
the idea of this paper was originated from a discussion between the author and prof stephane lafortune on opacity enforcement . without such an inspiring discussion , this paper would never be formed .
for this reason , the author would like to thank prof lafortune for his contribution .
a. a. cardenas , s. amin , and s. sastry .
secure control : towards survivable cyber - physical systems . in _ proc .
28th international conference on distributed computing systems workshops _ , 2008 , pp .
495500 .
a. teixeira , d. perez , h. sandberg , and k. h. johansson .
attack models and scenarios for networked control systems . in _ proc .
1st international conference on high confidence networked systems _ , 2012 , pp .
5564 .
l. k. carvalho , y. wu , r. kwong and s. lafortune .
detection and prevention of actuator enablement attacks in supervisory control systems . in _ proc .
13th international workshop on discrete event systems _
, 2016 , pp .
|
one of the major challenges about cyber physical systems is how to prevent cyber attacks to ensure system integrity .
there has been a large number of different types of attacks discussed in the modern control and computer science communities . in this paper
we aim to investigate one special type of attacks in the discrete - event system framework , where an attacker can arbitrarily alter sensor readings after intercepting them from a target system in order to trick a given supervisor to issue control commands improperly , driving the system to an undesirable state .
we first consider the cyber attack problem from an attacker point of view , and formulate an attack with bounded sensor reading alterations ( absra ) problem .
we then show that the supremal ( or least restrictive ) absra exists and can be synthesized , as long as the plant model and the supervisor model are regular , i.e. , representable by finite - state automata . upon the synthesis of the supremal absra , we present a synthesis algorithm , which ensures that a computed supervisor will be absra - robust ,
i.e. , either an absra will be detectable or will not lead the system to an undesirable state .
discrete - event systems , supervisory control , cyber security , attack under bounded sensor reading alterations , partial observation , controllability
|
the aim of the this paper is to obtain a largeness result for the images of quantum representations of mapping class groups in genus at least @xmath7 .
the main motivation is the construction of large families of finite quotients of ( central extensions of the ) mapping class groups by using the strong approximation theorem .
this method furnishes a large supply of finite quotients of mapping class groups outside the realm of symplectic groups .
similar results were obtained independently by masbaum and reid in @xcite .
earlier , looijenga has proved in @xcite that the images of prym representations ( associated to finite abelian groups ) of suitable finite index subgroups of mapping class groups and of subgroups from the johnson filtration are arithmetic groups .
some largeness results in this direction are already known . in @xcite
we showed that the images are infinite and non - abelian ( for all but finitely many explicit cases ) using earlier results of jones who proved in @xcite that the same holds true for the braid group representations factorizing through the temperley - lieb algebra at roots of unity .
masbaum then found in @xcite explicit elements of infinite order in the image .
general arguments concerning lie groups actually show that the image should contain a free non - abelian group .
furthermore , larsen and wang proved ( see @xcite ) that the image of the quantum representations of the mapping class groups at roots of unity of the form @xmath8 , for prime @xmath1 , is dense in the projective unitary group . in a previous paper @xcite
the authors proved that the images are large in the sense that they contain explicit free non - abelian groups .
now , each image is contained into some unitary group of matrices with cyclotomic integers entries , by @xcite .
the latter group can be embedded as an irreducible higher rank lattice in a semi - simple lie group @xmath9 ( depending on the genus and the order of the roots of unity ) obtained by restriction of scalars . the main result of this paper strengthens the largeness property above by showing that , in general , the image of a quantum representation is zariski dense in the non - compact group @xmath9 .
let us introduce now some terminology . recall that in @xcite the authors defined the tqft functor @xmath10 , for every @xmath11 and a primitive root of unity @xmath12 of order @xmath13 .
these tqft should correspond to the so - called @xmath14-tqft , for even @xmath15 and to the @xmath16-tqft , for odd @xmath15 ( see also @xcite for another @xmath16-tqft ) .
[ qrep ] let @xmath17 , @xmath11 and @xmath12 be a primitive @xmath13-th root of unity .
the quantum representation @xmath18 is the projective representation of the mapping class group associated to the tqft @xmath10 at the root of unity @xmath12 .
we denote therefore by @xmath19 the linear representation of the central extension @xmath20 of the mapping class groups @xmath21 ( of the genus @xmath4 closed orientable surface ) which resolves the projective ambiguity of @xmath18 ( see @xcite ) .
furthermore , @xmath22 will denote the dimension of the space of conformal blocks associated by the tqft @xmath10 to the closed orientable surface of genus @xmath4 .
the unitary tqfts arising usually correspond to the following choices of the root of unity : @xmath23 for prime @xmath1 we denote by @xmath24 the ring of cyclotomic integers @xmath25 $ ] , if @xmath26 and @xmath27 $ ] , if @xmath28 respectively , where @xmath29 is a primitive @xmath15-th root of unity .
the main result of @xcite states that , for every prime @xmath1 , there exists a free @xmath24-lattice @xmath30 in the @xmath31-vector space of conformal blocks associated by the tqft @xmath32 to the genus @xmath4 closed orientable surface and a non - degenerate hermitian @xmath24-valued form on @xmath30 such that ( a central extension of ) the mapping class group preserves @xmath30 and keeps invariant the hermitian form
. therefore the image of the mapping class group consists of unitary matrices ( with respect to the hermitian form ) with entries in @xmath24 . let @xmath33 be the group of all such matrices , up to scalar multiplication . for the sake of simplicity of the exposition we will consider only primes @xmath15 satisfying @xmath34 , from now on .
similar results hold for the remaining primes with only minor modifications .
it is known that @xmath33 is an irreducible lattice in a semi - simple lie group @xmath35 obtained by the so - called restriction of scalars construction from the totally real cyclotomic field @xmath36 to @xmath3 .
specifically , let us denote by @xmath9 the product @xmath37 . here
@xmath38 stands for a set of representatives for the classes of complex valuations @xmath39 of @xmath40 modulo complex conjugacy .
the factor @xmath41 is the special unitary group associated to the hermitian form conjugated by @xmath39 , thus corresponding to some galois conjugate root of unity .
denote also by @xmath42 and @xmath43 the representations @xmath44 and @xmath45 , respectively .
in @xcite the authors proved that the restriction of @xmath42 to the universal central extension @xmath46 of @xmath21 which is a subgroup of @xmath20 of index 12 takes values in @xmath47 .
this implies that @xmath48 for @xmath0 and prime @xmath1 .
notice that the @xmath9 is the set of real points of a semi - simple algebraic group @xmath2 defined over @xmath3 .
our main result can be stated now as follows : [ zd ] suppose that @xmath49 and @xmath1 is prime .
then @xmath50 is a discrete zariski dense subgroup of @xmath9 whose projections onto the simple factors of @xmath9 are topologically dense .
a similar result holds for the @xmath14-tqft .
specifically let @xmath51 where @xmath52 is prime .
according to ( @xcite , 1.5 ) there is an isomorphism of tqfts between @xmath53 and @xmath54 , and hence the projection on the second factor gives us a homomorphism @xmath55 .
furthermore , the image of the tqft representation associated to @xmath56 is finite
. therefore , @xmath57 is a discrete zariski dense subgroup of @xmath58 .
notice that the result holds also for @xmath59 and prime @xmath60 using the modifications from @xcite in the constructions of free non - abelian subgroups in the image .
we skip the details .
we now consider the johnson filtration by the subgroups @xmath61 of the mapping class group @xmath21 of the closed orientable surface of genus @xmath4 , consisting of those elements having a trivial outer action on the @xmath62-th nilpotent quotient of the fundamental group of the surface , for some @xmath63 .
the main application of our density result is the following : [ fquot ] for every @xmath0 and prime @xmath1 , there exists some homomorphism @xmath64 , whose restriction to @xmath65 is surjective for all large enough primes @xmath6 and all @xmath66 .
in particular , the surjectivity holds also for @xmath20 , the torelli group @xmath67 , the johnson kernel @xmath68 , or any finite index subgroup of @xmath20 , respectively .
the pull - back of the central extension @xmath20 on the torelli subgroup @xmath67 is trivial , because the generator of @xmath69 is the pull - back of the generator of @xmath70 , when @xmath0 .
thus @xmath65 embeds into @xmath20 .
the image of the central factor is finite and @xmath71 is of finite index in @xmath72 , according to proposition [ chain3 ] .
therefore , @xmath73 is of finite index in @xmath50 and hence zariski dense in @xmath9 .
furthermore , we will use the following version of the strong approximation theorem due to nori ( see ( @xcite , thm.5.4 ) and also @xcite ) : let @xmath74 be a connected linear algebraic group @xmath74 defined over @xmath3 and @xmath75 be a zariski dense subgroup .
assume that @xmath76 is simply connected .
then the completion of @xmath77 with respect to the congruence topology induced from @xmath78 is an open subgroup in the group @xmath79 of points of @xmath74 over the pro - finite completion @xmath80 of @xmath81 .
we now consider the group @xmath82 which satisfies the assumptions of nori s theorem .
if we take @xmath77 to be a finite index subgroup of @xmath50 the strong approximation theorem implies our claim for @xmath83 .
then a classical result due to serre ( see @xcite ) for @xmath84 and extended by vasiu ( see @xcite ) to all reductive linear algebraic groups defined over @xmath3 improves the surjectivity statement to all @xmath66 .
the arithmetic group @xmath85 , for @xmath0 and prime @xmath1 has the congruence property .
this follows from results of tomanov ( see @xcite , main thm .
( a ) ) and prasad and rapinchuk ( see @xcite , thm .
2.(1 ) and thm .
3 ) on the congruence kernel for @xmath3-anisotropic algebraic groups of type @xmath86 , with @xmath87 . in this respect
it would be interesting to construct finite quotients of the residually finite ( see @xcite ) group @xmath20 , other than the quotients of groups of the form @xmath5 .
moreover , @xmath85 is cocompact in @xmath9 , since it is @xmath3-anisotropic , by a classical result of borel and harish - chandra ( see @xcite ) .
[ fquot2 ] for any prime @xmath1 and @xmath0 there exists a homomorphism @xmath88 , whose restriction to a given finite index subgroup of @xmath21 is surjective for all large enough primes @xmath89 and all @xmath66 . here
@xmath35 is the product of the projective unitary groups whose associated special unitary groups occur as factors of @xmath9 . the image of the center of @xmath20 by the homomorphism @xmath42 is contained in the centralizer of @xmath50 .
the zariski density result above implies that the centralizer is contained in the product of the centers of each simple factor of @xmath9 .
this proves the claim . 1 .
the first construction of finite quotients of mapping class group by this method was given by masbaum in @xcite .
the set of finite quotients of a particular @xmath21 ( with @xmath0 ) provided by theorem [ fquot ] is rather large .
indeed , @xmath90 are finite groups of lie type of arbitrarily large rank . in particular , the alternate group on @xmath91 elements is contained into some @xmath90 , for large enough @xmath15 and @xmath6 , and hence into some finite quotient of @xmath21 .
therefore , every finite group embeds in some finite quotient of the genus @xmath0 mapping class group .
this answers a question of u. hamenstaedt , first settled by masbaum and reid in @xcite .
3 . in @xcite
we already obtained results showing that a given mapping class group has many more finite quotients than the family of all symplectic groups , as it can be measured by the torsion of their ( essential ) @xmath92-homology groups . 4 .
the number @xmath93 of the non - compact factors in @xmath9 goes to infinity with @xmath15 .
let @xmath94 denote the vector space of quasi - homomorphisms of the group @xmath74 , namely of maps @xmath95 for which @xmath96 , where @xmath97 is the boundary 2-cocyle .
the quasi - homomorphism @xmath98 is homogeneous if @xmath99 , for every @xmath100 and @xmath101 .
we consider the quotient @xmath102 of @xmath94 by the submodule generated by the bounded quasi - homomorphisms of @xmath74 and the group homomorphisms .
it is well - known ( see @xcite ) that @xmath102 is isomorphic to the kernel of the comparison homomorphism @xmath103 , where @xmath104 denotes the second bounded cohomology group of @xmath74 .
if @xmath105 is an irreducible hermitian symmetric space of non - compact type and @xmath106 its isometry group , then denote by @xmath107 the generator of the continuous bounded second cohomology group @xmath108 of @xmath106 .
this class is defined , for instance , by the dupont cocycle @xmath109 , given by : @xmath110 where @xmath111 is the khler form on @xmath105 , @xmath112 and @xmath113 denotes an oriented smooth triangle on @xmath105 with geodesic sides .
although the interior of the triangle with geodesic sides is not uniquely defined the value of the cocycle is well - defined because @xmath111 is closed ( see @xcite for details ) .
it is also known that the class of @xmath114 in the continuous second cohomology group @xmath115 is a generator .
bestvina and fujiwara proved in @xcite that @xmath116 is infinitely generated .
quantum representations yield an explicit family of quasi - homomorphisms on the mapping class groups generalizing the rademacher function on @xmath117 ( which can be obtained for the one - holed torus ) , as follows : [ quasi ] let @xmath0 , @xmath1 be a prime number and @xmath118 be a non - compact simple factor of @xmath119 corresponding to the primitive root of unity @xmath12 . let @xmath120 be the universal central extension of @xmath21 .
then there exists a unique quasi - homomorphism @xmath121 verifying @xmath122 furthermore , the classes of those @xmath123 , for which @xmath124 , are linearly independent over @xmath3 in @xmath125 .
burger and iozzi proved in ( @xcite , thm 1.3 ) that for each set of pairwise non - equivalent ( i.e. not conjugate by an isometry of the corresponding symmetric spaces ) zariski dense representations @xmath126 , of a finitely generated group @xmath127 , for which @xmath128 , the elements @xmath129 are linearly independent over @xmath81 . on the other hand @xmath130
because @xmath131 is a universal central extension and hence it does not have any non - trivial central extension .
therefore , the cocycle @xmath132 corresponds to an element of @xmath116 and the corollary follows from the above cited result in @xcite .
if @xmath133 is some zariski dense representation and @xmath134 is the corresponding quasi - homomorphism , then @xmath135 in @xmath116 only if @xmath136 is conjugate to @xmath19 ( see @xcite ) . moreover , let @xmath137 denote the unique homogeneous quasi - homomorphism in the class of @xmath123 . then @xmath137 is a class function ( i.e. invariant on conjugacy classes ) on @xmath46 which encodes all information about the representation @xmath19 . if the groups @xmath72 were finitely presented and the number of relations in some group presentation were uniformly bounded ( for fixed @xmath4 ) , then the rank of @xmath138 would also be uniformly bounded and our density theorem would imply that @xmath139 can not be trivial for large enough @xmath15 .
for instance the uniform bound holds if we replace @xmath140 above by the subgroup generated by @xmath15-th powers of all dehn twists and one might conjecture that the two subgroups coincide in genus @xmath0 . eventually ,
if @xmath139 were infinite dimensional then @xmath72 would not be boundedly generated .
recall now the following recent result due to salehi golsefidy and varj ( see @xcite , thm . 1 and cor .
5 ) : let @xmath141 be a group generated by a symmetric set @xmath142 and denote by @xmath143 the reduction mod @xmath6 .
then the family of cayley graphs of the groups @xmath144 with generator systems @xmath145 forms a family of expanders ( see @xcite for details about expanders ) , where @xmath6 runs through square - free integers , if and only if the connected component of the zariski closure of @xmath127 is perfect .
this family of cayley graphs forms expanders iff @xmath127 has property @xmath146 with respect to the family of finite quotients @xmath144 , where @xmath6 runs through square - free integers , namely if there exists some @xmath147 ( depending on @xmath142 ) such that for every unitary representation @xmath148 of @xmath127 into some hilbert space @xmath149 factorizing through some @xmath144 without invariant nontrivial vector and every unit vector @xmath150 we have @xmath151 , for some @xmath152 . here
@xmath153 denotes the norm in @xmath149 .
observe now that the ( complex ) zariski closure of @xmath118 in @xmath154 is @xmath155 .
in particular , our density result implies that the zariski closure of @xmath50 ( within the appropiate product of copies of @xmath156 ) is the complex zariski closure of @xmath2 , namely the product of several copies of @xmath157 .
since the zariski closure is perfect then by ( @xcite , thm.1 , but cor.5 would also suffice ) we obtain the following : [ tau ] for every prime @xmath1 , the groups @xmath50 ( and @xmath158 , @xmath159 ) have property @xmath146 with respect to the family of finite quotients induced by @xmath160 , where @xmath6 runs through square - free integers .
bourgain and gamburd proved ( see @xcite ) that a dense subgroup of @xmath161 generated by matrices with algebraic entries , and in particular @xmath50 , has a spectral gap with respect to the natural action on @xmath162 . on the other hand these results can not be considered as evidence in the favor of the claim that @xmath50 has property @xmath146 with respect to the family of _ all _ its finite quotients .
in fact , by @xcite there exists a free subgroup on two generators @xmath163 which is zariski dense in @xmath9 .
although a free group has not property @xmath146 with respect to all its finite quotients , the group @xmath164 has property @xmath146 with respect to the family of finite quotients induced by @xmath160 , where @xmath6 runs through square - free integers , by the result from @xcite .
it would be very interesting to find whether @xmath20 has property @xmath146 with respect to the family of finite quotients @xmath165 , where @xmath166 are sufficiently large , possibly square - free and @xmath15 runs over the primes .
* acknowledgements .
* i am indebted to yves benoist , philippe gille , olivier guichard , greg kuperberg , toshitake kohno , greg mcshane , gregor masbaum , mahan mj , alan reid , adrian vasiu and tyakal n. venkataramana for useful discussions and advice .
the author was partially supported by the anr grant modgroup .
the starting point is the following result from @xcite , subsequently improved in @xcite : [ lw ] if @xmath167 , @xmath1 is prime and @xmath168 , then @xmath169 is ( topologically ) dense in @xmath170 , where @xmath22 is the dimension of the space of conformal blocks associated to the closed orientable surface @xmath171 of genus @xmath4 .
a cautionary remark is in order .
there are two tqfts which might reasonably be called the @xmath16-tqft and although their differences are rather minimal , they are distinct ( see the discussion in @xcite ) .
we work here with the version from @xcite , in order to have a common approach for both @xmath14-tqft and the @xmath16-tqft .
however , the proof from @xcite , although stated for one version , works also for the other version of the tqft . the hermitian form @xmath149 associated to the tqft @xmath172 and the root of unity @xmath12 has signature @xmath173 , where @xmath174 .
we prove first : [ weakzd ] let @xmath0 and @xmath12 be any primitive @xmath13-th root of unity . then @xmath175 is ( topologically ) dense in @xmath176 .
two representations into the same group @xmath74 are called equivalent if there exists an isomorphism of @xmath74 which intertwines them .
it is easy to see that @xmath177 and therefore @xmath19 and @xmath178 are equivalent .
the converse is provided by the following : [ inequivalent ] let @xmath0 , @xmath12 and @xmath179 be distinct primitive @xmath13-th roots of unity .
the the representations @xmath19 and @xmath180 are equivalent if and only if @xmath181 .
the same proof actually shows that the projective representations @xmath18 and @xmath182 are equivalent if and only if @xmath183 or @xmath181 .
let now @xmath184 , @xmath185 , be a collection of representations of the group @xmath127 .
the subgroup @xmath186 is called @xmath127-diagonal , if there exists a partition @xmath187 of @xmath188 such that : 1 .
all factors @xmath189 , with @xmath190 , @xmath191 are equivalent as representations of @xmath127 .
pick up some @xmath192 .
given some intertwining isomorphisms @xmath193 , @xmath194 , we set : @xmath195 which is the graph of the homomorphism @xmath196 .
2 . then there exist intertwining isomorphisms as above with the property that the group @xmath149 contains @xmath197 .
in particular , if all representations @xmath189 of @xmath127 are pairwise inequivalent , then @xmath198 .
we then have the following hall lemma from @xcite : let @xmath127 be a subgroup of the product @xmath199 of the adjoint simple ( i.e. connected , without center and whose lie algebra is simple ) lie groups @xmath189 .
assume that the projection of @xmath127 on each factor @xmath189 is zariski dense .
then the zariski closure of @xmath127 in @xmath199 is a @xmath127-diagonal subgroup .
therefore , the hall lemma above shows that the zariski closure of @xmath72 is all of @xmath35 .
now using ( @xcite , lemma 3.6 ) we obtain that @xmath50 is zariski dense in @xmath9 .
this proves the theorem .
recall that the ( reduced ) burau representation @xmath200)$ ] is defined on the standard generators @xmath201 of the braid group @xmath202 on @xmath203 strands by the matrices : @xmath204 @xmath205 @xmath206 squier proved that , after a suitable rescaling , @xmath209 preserves a non - degenerate hermitian form and hence it takes values in either @xmath210 or @xmath211 , depending on the signature of the invariant form .
there are several proofs of this statement .
it is a consequence of the results of deligne and mostow concerning hyper - geometric integrals ( see @xcite ) and the identification of the burau representation as such one .
alternatively , we might use a result of freedman , larsen and wang ( see @xcite ) subsequently reproved and extended by kuperberg in ( @xcite , thm.1 ) saying that the burau representation of @xmath220 at a @xmath13-th root of unity is zariski dense in the group @xmath219 .
then the curves @xmath223 , @xmath224 and @xmath225 which are surrounding two of the holes of @xmath221 the pure braid group @xmath226 embeds into @xmath227 using a non - canonical splitting of the surjection @xmath228 .
furthermore , @xmath227 embeds into @xmath21 when @xmath229 , by using the homomorphism induced by the inclusion of @xmath221 into @xmath171 as in the figure . when @xmath230 we shall use other embeddings , similar to those used in @xcite for the proof that the image of the quantum representation contains free non - abelian subgroups .
[ contain ] let @xmath1 .
the restriction of the quantum representation @xmath43 at @xmath231 has an invariant 3-dimensional subspace such that the corresponding sub - representation is equivalent to the burau representation @xmath209 . for even @xmath15 and @xmath232 instead of @xmath226
this is the content of @xcite , prop .
the odd case is similar .
the invariant subspace is the space of conformal blocks associated to the surface @xmath221 with boundary labels @xmath233 , when @xmath234 and @xmath235 , when @xmath60 respectively .
the eigenvalues of the half - twist can be computed as in @xcite .
_ end of the proof of proposition [ weakzd ] .
_ if @xmath15 is as above , then proposition [ lw ] shows that @xmath169 is topologically dense in @xmath170 which is a compact lie group .
in particular , the image is zariski dense in @xmath170 .
the zariski density is preserved by a galois conjugacy and thus , @xmath175 is zariski dense in @xmath176 for every @xmath12 .
furthermore , a zariski dense subgroup of a reductive almost simple lie group ( in particular , of @xmath176 ) not contained in the center is either dense or discrete .
now , recall that @xmath175 contains @xmath237 .
the latter group is topologically dense in @xmath215 and respectively @xmath219 .
therefore @xmath175 is in - discrete . since the image group @xmath175 is not contained within the center of @xmath176
, it follows that it should be topologically dense in @xmath176 .
this proves the claim .
assume that the representations @xmath19 and @xmath180 were equivalent as representations into the special pseudo - unitary group @xmath238 .
then @xmath18 and @xmath182 are equivalent as projective representations into @xmath239 and hence by results of walter ( see @xcite ) they are equivalent as representations into @xmath240 .
this means that there exists an isomorphism @xmath241 with the property that @xmath242 according to classical results of rickart ( improving previous results by dieudonn ) from @xcite any automorphism @xmath164 of a unitary group @xmath243 over an infinite field could be expressed under the form : @xmath244 where @xmath245 is a semi - linear isomorphism , which will be called the intertwiner .
namely , this means that there exists some field automorphism @xmath246 with the properties : @xmath247 @xmath248 @xmath249 where @xmath250 denotes the conjugate of @xmath223 and @xmath149 is the hermitian form associated to @xmath12 ( or @xmath179 ) .
furthermore , @xmath251 is some homomorphism .
it is well - known that the only automorphisms of @xmath31 commuting with the complex conjugacy are the identity and the complex conjugacy .
thus @xmath252 is either a linear or an anti - linear isomorphism preserving ( and respectively conjugating ) the hermitian form @xmath149 .
let @xmath254 denote the homomorphism defined by @xmath255 .
since @xmath256 , we obtain that @xmath257 is a character on @xmath20 . the presentation given by gervais ( see @xcite , thm.b ) shows that lantern relations lift to @xmath20 .
furthermore , @xmath20 is generated by lifts of pairwise conjugate dehn twists along non - separating curves and one central generator @xmath225 .
we can express @xmath258 using the chain relation on 2-holed subtori with non - separating boundary curves .
the lantern relation implies then that @xmath259 .
alternatively , we can use the fact that the universal central extension @xmath46 is a normal subgroup of index 12 of @xmath20 and @xmath260
. therefore @xmath261 , for any @xmath262 . on the other hand , it is known that @xmath263 is the scalar matrix @xmath264 ( see e.g. @xcite ) and hence @xmath265 . since @xmath1 is prime , it follows that @xmath266 and hence @xmath257 factors through @xmath21 .
now , @xmath21 is perfect for @xmath0 and thus the character @xmath257 is trivial . in particular , we can take @xmath253 to be trivial .
the linear representations @xmath19 and @xmath180 are either linearly or anti - linearly equivalent by means of some intertwiner isomorphism @xmath252 . now , if @xmath273 then we have : @xmath274 thus the set of eigenvalues of @xmath275 and @xmath276 should correspond to each another by means of @xmath252 .
let @xmath277 , @xmath278 , be some set of disjoint simple closed curves on the closed surface of genus @xmath4 .
then the @xmath279 lifts of dehn twists @xmath280 , @xmath281 pairwise commute and hence one could diagonalize them simultaneously .
let @xmath282 denote the space of conformal blocks associated by the tqft @xmath172 to the closed orientable surface of genus @xmath4 .
we set @xmath283 and define : @xmath284 let now @xmath285 be an essential embedding of the surface @xmath286 of genus @xmath287 with @xmath279 boundary components in @xmath171 , i.e. such that the homomorphism induced at the level of fundamental groups is injective .
we also assume that @xmath286 has a pants decomposition .
let @xmath288 be a maximal system of non null - homotopic and pairwise non - homotopic simple closed curves in @xmath289 .
then @xmath288 contains the set @xmath290 of boundary circles of @xmath286 .
we associate to each circle @xmath291 from @xmath290 some arbitrary @xmath292 .
further , to each circle @xmath293 in @xmath294 we associate @xmath295 .
we say that @xmath77 comes from a coloring of @xmath290 if there is some set of colors @xmath296 such that , for all @xmath297 the vector space @xmath298 is non - zero only if @xmath77 comes from a coloring of @xmath290 . in this case
@xmath298 can be identified with the space of conformal blocks associated to the subsurface @xmath286 and the coloring @xmath296 of the boundary components .
it is well - known that the eigenvalues @xmath299 of dehn twists in the basis given by colored trivalent graphs from ( @xcite , 4.11 ) are given by the formula from above , in terms of colors ( see also @xcite , 5.8 ) . now
the space of conformal blocks associated to a subsurface splits as a direct sum of one dimensional eigenspaces @xmath300 associated to all possible colorings of some maximal system @xmath301 of non null - homotopic and pairwise non - homotopic simple closed curves on the subsurface @xmath286 .
consider first the case ( of the @xmath16-tqft ) when @xmath303 is odd .
then the set of colors is @xmath304 and the collection of eigenvalues is ( see @xcite , 5.8 ) given by @xmath305 . by assumption
there exists a bijection between the set of colors associated to @xmath12 and @xmath179 , namely @xmath306 with the property that @xmath307 .
moreover , the bijection @xmath308 between the colors should be compatible with the conformal blocks structure
. we will prove by recurrence on @xmath309 that @xmath310 .
first we have @xmath311 . if @xmath312 , then @xmath313 .
the space of conformal blocks associated to a 4-holed sphere @xmath314 whose boundary components are colored by @xmath92 has dimension @xmath7 , corresponding to the colors @xmath315 on a separating curve ( if @xmath60 ) . the bijection @xmath308 should send this space into the space of conformal blocks associated to a 4-holed sphere whose boundary components are colored by @xmath316 . but the separating curve can be given any even color in the range @xmath317 .
thus @xmath318 would lead to a contradiction .
this implies that @xmath319 .
therefore we have @xmath320 where both @xmath12 and @xmath179 are primitive roots of unity of order @xmath13 , with odd @xmath15 , which implies that @xmath183 .
the case @xmath234 is immediate , by direct calculation .
consider now the case ( of the @xmath14-tqft ) when @xmath321 is even .
then the set of colors is @xmath322 and the collection of eigenvalues is ( see @xcite , 5.8 ) given by @xmath323 . by assumption
there exists a bijection between the set of colors associated to @xmath12 and @xmath179 , namely @xmath324 with the property that @xmath325 .
moreover , the bijection @xmath308 between the colors should be compatible with the conformal blocks structures .
we first have @xmath311 . if @xmath326 , then @xmath327 .
the space of conformal blocks associated to a 4-holed sphere @xmath314 whose boundary components are colored by @xmath328 has dimension @xmath92 , corresponding to the colors @xmath329 on a separating curve ( if @xmath330 ) .
the space of conformal blocks associated to a 4-holed sphere @xmath314 whose boundary components are colored by @xmath331 has dimension @xmath332 , corresponding to the colors @xmath333 on a separating curve .
this leads to a contradiction when @xmath334 and @xmath335 .
thus @xmath336 .
moreover , the space of conformal blocks associated to the 3-holed sphere with boundary components colored by @xmath337 and @xmath92 is 1-dimensional so this is the same for the the coloring @xmath338 and @xmath316 .
therefore @xmath339 by the clebsch - gordan admissibility conditions and hence @xmath319
. therefore @xmath340 and @xmath320 which gives us @xmath183 . for a group @xmath74
we denote by @xmath341 the lower central series defined by : @xmath342 , k\geq 1\ ] ] an interesting family of subgroups of the mapping class group is the set of higher johnson subgroups defined as follows . the @xmath62-th johnson subgroup @xmath61 is the group of mapping classes of homeomorphisms of the closed orientable surface @xmath171 of genus @xmath4 whose action by outer automorphisms on @xmath343 is trivial , where @xmath344 .
thus @xmath345 , @xmath67 is the torelli group commonly denoted @xmath346 , while @xmath68 is the group generated by the dehn twists along separating simple closed curves and considered by johnson and morita ( see e.g. @xcite ) , which is often denoted by @xmath347 .
there is a surjective homomorphism @xmath349 .
the image of the @xmath15-th power of a dehn twist in @xmath72 is trivial . on the other hand ,
the image of a dehn twist in @xmath350 is a transvection and taking all dehn twists one obtains a system of generators for @xmath350 . using the congruence subgroup property for @xmath350 , where @xmath167 , the image of @xmath15-th powers of dehn twists in @xmath350 generate the congruence subgroup @xmath351=\ker ( sp(2g,{{\mathbb{z}}})\to sp(2g,{{\mathbb{z}}}/p{{\mathbb{z}}}))$ ] .
since the mapping class group is generated by dehn twists the homomorphism @xmath352 should factor through @xmath353}=sp(2g,{{\mathbb{z}}}/p{{\mathbb{z}}})$ ] .
in particular , the image of @xmath352 is finite . by the work of johnson ( see @xcite ) one knows that when @xmath0 the quotient @xmath354 is a finitely generated abelian group @xmath12 isomorphic to @xmath355 , where @xmath149 is the homology of the surface . thus there is a surjective homomorphism @xmath356 .
the torelli group @xmath346 is generated by bp - pairs , namely elements of the form @xmath357 , where @xmath267 and @xmath358 are non - separating disjoint simple closed curves bounding a subsurface of genus 1 ( see @xcite ) .
now @xmath15-th powers of the bp - pairs @xmath359 have trivial images in @xmath360 .
but the classes @xmath357 also generate the quotient @xmath12 and hence the classes @xmath361 will generate the abelian subgroup @xmath362 of those elements of @xmath12 which are divisible by @xmath15 .
this shows that @xmath363 factors through @xmath364 , which is a finite group because @xmath12 is finitely generated . in particular , @xmath360 is finite . eventually , @xmath365 is also a finitely generated abelian group @xmath366 , namely the image of the third johnson homomorphism .
since @xmath367 is generated by the dehn twists along separating simple closed curves the previous argument shows that @xmath368 is the image of a surjective homomorphism from @xmath369 and hence is finite .
recently , dimca and papadima proved in @xcite that @xmath370 is finitely generated for @xmath0 .
the above proof implies that @xmath371)$ ] is of finite index in @xmath372 .
we have also the following alternative argument , which makes the proof independent of the result in @xcite .
the group @xmath372 is finitely generated since it is of finite index in the finitely generated group @xmath72 .
thus @xmath373)$ ] is abelian and finitely generated .
moreover @xmath373)$ ] is generated by torsion elements of order @xmath15 , since @xmath367 is generated by dehn twists along bounding simple closed curves .
now , any finitely generated abelian group generated by order @xmath15 elements must be finite .
a natural question is whether @xmath374 is of finite index in @xmath375 , for every @xmath62 .
the arguments above break down at @xmath376 since there are no products of powers of commuting dehn twists in any higher johnson subgroups .
more specifically , we have to know the image of the group @xmath377\cap i_g(k)$ ] in @xmath378 by the johnson homomorphism .
here @xmath377 $ ] denotes the normal subgroup generated by the @xmath15-th powers of dehn twists .
if the image were a lattice in @xmath378 , then we could deduce as above that @xmath379 is finite .
g. masbaum , _ an element of infinite order in tqft - representations of mapping class groups _ , low - dimensional topology ( funchal , 1998 ) , 137139 , contemp .
, 233 , amer .
soc . , providence , ri , 1999 .
g. prasad and a.s .
rapinchuk , _
developements on the congruence subgroup problem after the work of bass , milnor and serre _ , in `` collected papers of john milnor '' , vol .
( 2010 ) , 307325 .
|
our main result is that the image of the quantum representation of a central extension of the mapping class group of the genus @xmath0 closed orientable surface at a prime @xmath1 is a zariski dense discrete subgroup of some higher rank algebraic semi - simple lie group @xmath2 defined over @xmath3 . as an application we find that , for any prime @xmath1 a central extension of the genus @xmath4 mapping class group surjects onto the finite groups @xmath5 , for all but finitely many primes @xmath6 .
this method provides infinitely many finite quotients of a given mapping class group outside the realm of symplectic groups .
+ 2000 msc classification : 57 m 07 , 20 f 36 , 20 g 20 , 22 e 40 .
+ keywords : mapping class group , dehn twist , braid group , burau representation , quantum representation , discrete subgroup of semi - simple lie groups , finite quotient .
|
spin transport in semiconductor structures is a subject of much interest from the perspective of both fundamental physics and device applications .
a number of different formalisms have been used to study this problem , primary among which are a classical drift diffusion approach @xcite , a kinetic theory approach @xcite , and a microscopic semiclassical approach @xcite .
the central result of the drift diffusion approach is a differential equation that describes the spatial and temporal evolution of carriers with a certain spin polarization @xmath0 .
@xcite derived this equation for a number of special cases starting from the wigner distribution function . in a coordinate system where the x - axis coincides with the direction of electric field driving transport , this equation is of the form : @xmath1 where @xmath2 @xmath3 is the diffusion coefficient , and @xmath4 and @xmath5 are dyadics ( 9-component tensors ) that depend on @xmath3 , the mobility @xmath6 and the spin orbit interaction strength in the material
. solutions of equation ( [ spin ] ) , with appropriate boundary conditions , predict that the ensemble averaged spin @xmath7 = @xmath8 should decay exponentially with @xmath9 according to : @xmath10 where @xmath11 here @xmath12 is the strength of the driving electric field and @xmath13 is a parameter related to the spin orbit interaction strength . the quantity
@xmath14 is the characteristic length over which @xmath15 decays to @xmath16 times its original value .
therefore , it is defined as the `` spin diffusion length '' .
equation ( [ diffusion ] ) clearly shows that spin diffusion length depends on the _ sign _ of the electric field @xmath12 .
it is smaller for upstream transport ( when @xmath12 is positive ) than for downstream transport ( when @xmath12 is negative ) .
this difference assumes importance in the context of spin injection from a metallic ferromagnet into a semiconducting paramagnet .
@xcite pointed out that the spin injection efficiency across the interface between these materials depends on the difference between the quantities @xmath17 and @xmath18 where @xmath19 is the spin diffusion length in the semiconductor , @xmath20 is the conductivity of the semiconductor , @xmath21 is the conductivity of the metallic ferromagnet , and @xmath22 is the spin diffusion length in the metallic ferromagnet .
generally , @xmath23 .
however , at sufficiently high retarding electric field , @xmath24 , so that @xmath25 . when this equality is established , the spin injection efficiency is maximized .
thus , ref .
@xcite claimed that it is possible to circumvent the infamous `` conductivity mismatch '' problem @xcite , which inhibits efficient spin injection across a metal - semiconductor interface , by applying a high retarding electric field in the semiconductor close to the interface .
a tunnel barrier between the ferromagnet and semiconductor @xcite or a schottky barrier @xcite at the interface does essentially this and therefore improves spin injection .
the result of ref .
@xcite depends on the validity of the drift diffusion model and equation ( [ decay ] ) which predicts an exponential decay of spin polarization in space . without the exponential decay
, one can not even define a `` spin diffusion length '' @xmath14 .
the question then is whether one expects to see the exponential decay under all circumstances , particularly in quantum confined structures such as quantum wires .
the answer to this question is in the _ negative_. equation ( [ spin ] ) , and similar equations derived within the drift diffusion model , do not account for energy quantization in quantum confined systems and neglect the influence of subband structure on spin depolarization .
this is a serious shortcoming since in a semiconductor quantum wire , the spin orbit interaction strength is _ different _ in different subbands .
it is this difference that results in dyakonov - perel ( d - p ) spin relaxation in quantum wires . without this difference ,
the d - p relaxation will be completely absent in quantum wires and the corresponding spin diffusion length will be always infinite @xcite .
the suband structure is therefore vital to spin relaxation .
in this paper , we have studied spin relaxation using a microscopic semiclassical model that is derived from the liouville equation for the spin density matrix @xcite .
our model has been described in ref .
@xcite and wil not be repeated here .
this model allows us to study dyakonov - perel spin relaxation taking into account the detailed subband structure in the system being studied . in technologically important semiconductors , such as gaas ,
spin relaxation is dominated by the dyakonov - perel ( d - p ) mechanism @xcite .
this mechanism arises from the dresselhaus @xcite and rashba @xcite spin orbit interactions that act as momentum dependent effective magnetic fields @xmath26 .
an electron s spin polarization vector @xmath27 precesses about @xmath28 according to the equation @xmath29 where @xmath30 is the angular frequency of spin precession and is related to @xmath31 as @xmath32 , where @xmath33 is the electron s effective mass . if the direction of * b ( * k * ) * changes randomly due to carrier scattering which changes * k * , then ensemble averaging over the spins of a large number of electrons will lead to a decay of the ensemble averaged spin in space and time .
this is the physics of the d - p relaxation in bulk and quantum wells . in a quantum wire ,
the direction of * k * never changes ( it is always along the axis of the wire ) in spite of scattering .
nevertheless , there can be d - p relaxation in a _ multi - subband _ quantum wire , as we explain in the next paragraphs .
thus , * b ( * k * ) * lies in the x - z plane and subtends an angle @xmath43 with the wire axis ( x - axis ) given by @xmath44\ ] ] note from the above that in any given subband in a quantum wire , the _ direction _ of @xmath31 is fixed , irrespective of the magnitude of the wavevector @xmath40 , since @xmath43 is independent of @xmath40 . as a result , there is no d - p relaxation in any given subband , even in the presence of scattering .
however , @xmath43 is different in different subbands because the dresselhaus interaction is different in different subbands .
consequently , as electrons transition between subbands because of inter - subband scattering , the angle @xmath43 , and therefore the direction of the effective magnetic field @xmath31 , changes .
this causes d - p relaxation in a _ multi - subband _ quantum wire .
since spins precess about different axes in different subbands , ensemble averaging over electrons in all subbands results in a gradual decay of the net spin polarization .
thus , there is no d - p spin relaxation in a quantum wire if a single subband is occupied , but it is present if multiple subbands are occupied and inter - subband scattering occurs .
this was shown rigorously in ref .
@xcite .
the subband structure is therefore critical to d - p spin relaxation in a quantum wire .
in fact , if a situation arises whereby all electrons transition to a single subband and remain there , further spin relaxation due to the d - p mechanism will cease thereafter . in this case , spin no longer decays , let alone decay exponentially with distance .
hence , spin depolarization ( or spin relaxation ) can not be parameterized by a constant spin diffusion length .
the rest of this paper is organized as follows . in the next section ,
we describe our model system , followed by results and discussions in section iv .
finally , we conclude in section v.
we consider a non centro - symmetric ( e.g. gaas ) quantum wire with axis along [ 100 ] crystallographic direction .
we choose a three dimensional cartesian coordinate system with @xmath45 coinciding with the axis of the quantum wire ( refer fig . [ model ] ) .
the structure is of length @xmath46 with rectangular cross section : @xmath47 nm and @xmath48 nm . a metal gate is placed on the top ( not shown in fig .
[ model ] ) to induce the symmetry breaking electric field @xmath49 , which causes the rashba interaction . in a quantum wire defined by split schottky gates on a two - dimensional electron gas ,
@xmath49 arises naturally because of the triangular potential confining carriers near the heterointerface .
we assume @xmath50 kv / cm @xcite .
in addition , there is another electric field @xmath51 ( @xmath52 ) which drives transport along the axis of the quantum wire .
consider the case when _ spin polarized _ monochromatic electrons are constantly injected into the channel at @xmath53 with injection velocities along @xmath54 . if these electrons occupy only the lowest subband _ at all times
_ , then there will be no dyakonov - perel relaxation @xcite .
therefore , in order to study multisubband effect on spin dephasing of upstream electrons , we inject them with enough energy ( @xmath55 ) that they initially occupy multiple subbands .
we ignore any thermal broadening of injection energy @xcite since @xmath56 for the range of temperature ( @xmath57 ) considered , @xmath58 being boltzmann constant .
let @xmath59 denote the energy at the bottom of @xmath60th subband ( @xmath61 etc . ) .
we place @xmath55 between the @xmath62-th and @xmath63-th subband bottoms as shown in fig .
[ subband ] . in other words , @xmath64 .
we assume that the injected electrons , each with energy @xmath55 , are distributed uniformly over the lowest @xmath62 subbands .
in other wordes , at time @xmath65 , electron population of the @xmath66 th @xmath67 subband is given by @xmath68 where @xmath69 is the total number of injected electrons and @xmath12 denotes their energies .
at any subsequent time @xmath70 , these distributions spread out in space ( @xmath71 ) , as well as in energy , due to interaction of the injected electrons with the electric field @xmath72 and numerous scattering events .
relative population of electrons among different subbands will change as well due to intersubband scattering events . _
upstream _ electrons originally injected into , say , subband @xmath60 with velocity @xmath73 ( @xmath74 ) , gradually slow down because of scattering and the decelerating electric field .
they change their direction of motion ( i.e. become _ downstream _ ) beyond a distance @xmath75 measured from the injection point @xmath76 .
thus , no electrons will be found in the @xmath60-th subband beyond @xmath77 .
note that the value of @xmath75 depends on three factors : the initial injection velocity into subband @xmath60 , the decelerating electric field and the scattering history . on the other hand ,
the ` classical turning distance ' of monochromatic electrons injected into the @xmath60-th subband with energy @xmath78 for a given electric field @xmath72 is given by @xmath79 where @xmath59 is the energy at the bottom of the @xmath60-th subband and @xmath80 is the injection velocity in the @xmath60-th subband .
note that @xmath81 does not depend on scattering history and @xmath81 = @xmath77 in ballistic transport .
clearly @xmath82 for a given @xmath55 ( see fig .
[ subband ] ) .
thus , for a given channel electric field @xmath72 , @xmath83min@xmath84 . hence we concentrate on the region @xmath85 where , almost _ all _ injected electrons are _ upstream _ electrons . in the simulation ,
velocity of every electron is tracked and as soon as an electron alters direction and goes downstream ( i.e. its velocity becomes positive ) it is ignored by the simulator and another upstream electron is simultaneously injected from @xmath86 randomly in any of the @xmath62 lowest subbands with equal probability .
this process is continued for a sufficiently long time till electron distributions over different subbands , @xmath87 , no longer change with time . under this condition
we say that _ steady state _
is achieved for the upstream electrons .
this steady state electron distribution is extended from @xmath88 to @xmath86 and heavily skewed near the region @xmath86 .
this steady state distribution of upstream electrons _ does not _ represent the local equilibrium electron distribution because of two reasons : ( a ) upstream electrons are constantly injected into the channel at @xmath86 ; this is the reason why the distribution is skewed near @xmath86 and ( b ) we exclude any downstream electron from the distribution . at local equilibrium
, there will be of course both upstream and downstream electrons in the distribution .
the model above allows us to separate upstream electrons from downstream electrons and therefore permits us to study upstream electrons in isolation . of course
, in a real quantum wire , both upstream and downstream electrons will be present at any time , even in the presence of a strong electric field , since there will be always some non - vanishing contribution of back - scattered electrons to the upstream population .
the semiclassical model and the simulator used to simulate spin transport have been described in ref .
@xcite . based on that model , at steady state
, the magnitude of the ensemble averaged spin vector at any position @xmath9 inside the channel is given by @xmath89 here @xmath90 , @xmath91 , denotes the ensemble average of @xmath92 component of spin at position @xmath9 .
subscript @xmath60 implies that ensemble averaging is carried out over electrons _ only _ in the @xmath60 th subband .
the above equation can be simplified to @xmath93 where @xmath94 and @xmath95 is the angle between @xmath96 and @xmath97 .
note that in absence of any intersubband scattering event , @xmath98 for all @xmath9 ( i.e. initial spin polarization of the injected electrons ) @xcite .
simulation results that we present in the next section can be understood using equation ( [ avgspin1 ] ) .
we examine how ensemble averaged spin polarization of upstream electrons @xmath99 varies in space for different values of driving electric field @xmath72 and injection energy ( @xmath55 ) for a fixed lattice temperature @xmath57 .
we vary @xmath72 in the range @xmath100kv / cm for constant injection energy @xmath101 mev and @xmath57 = 30 k , where @xmath55 is measured from the bulk conduction band energy as shown in fig .
[ subband ] .
the lowest subband bottom is 351 mev above the bulk conduction band edge .
we also present results corresponding to @xmath102 mev with @xmath103kv / cm and @xmath104k . in all cases mentioned above
, injection energies lie between subband 3 and subband 4 .
injected electrons are equally distributed among the three lowest subbands initially .
obviously , this corresponds to a non - equilibrium situation .
all injected electrons are 100% spin polarized transverse to the wire axis ( i.e either @xmath105 or @xmath106 ) .
figures [ halffieldspin]-[twofieldspin ] , [ 90mev ] , and [ pos75mev - z ] show how ensemble averaged spin components @xmath107 , @xmath108 , @xmath109 and @xmath110 of upstream electrons evolve over space .
figures [ halffieldspin]-[twofieldspin ] show the influence of the driving electric field on spin relaxation , fig .
[ 90mev ] shows the influence of initial injection energy and fig .
[ pos75mev - z ] shows the influence of the intial spin polarization .
it is evident that neither the driving electric field , nor the initial injection energy , nor the intial spin polarization has any significant effect on spin relaxation .
note that @xmath111 does _ not _ decay exponentially with distance , contrary to equation ( [ decay ] ) .
spatial distribution of electrons over different subbands is shown in fig .
[ halffieldsub ] - [ twofieldsub ] , and fig .
[ sub90mev ] .
the classical turning point of electrons in the third subband ( @xmath112 ) has been indicated in each case .
[ halffieldsub ] - [ twofieldsub ] show the influence of the driving electric field and fig .
[ sub90mev ] shows the influence of initial injection energy on the spatial evolution of subband population .
as expected , @xmath113 decreases with increasing electric field in accordance with equation ( [ clas ] ) .
note that at low electric field ( figs . 7 and 8) @xmath114 since all subbands are getting nearly depopulated of `` upstream '' electrons at @xmath9 = @xmath112 . recall that @xmath115 only if transport is ballistic ; therefore we can conclude that upstream transport is nearly ballistic in the range @xmath113 when @xmath72 @xmath116 1 kv / cm . at high electric field , when @xmath72 @xmath117 1.5 kv / cm , ( fig .
10 ) @xmath118 .
this indicates that there are many upstream electrons even beyond the classical turning point .
it can only happen if there is significant scattering that drives electrons against the electric field , making them go beyond the classical turning point .
we can also deduce that most of these scattering events imparts momentum to the carriers to _ aid _ upstream motion rather than oppose it , since @xmath118 .
this behavior is a consequence of the precise nature of the scattering events and would not have been accessible in drift - diffusion models that typically treat scattering via a relaxation time approximation .
note that even though electrons are injected equally into all three subbands , most electrons end up in subband 3 the highest subband occupied initially soon after injection . beyond a certain distance ( @xmath119 )
subbands 1 and 2 become virtually depopulated .
this feature is very counter - intuitive and represents a _ population inversion _ of upstream electrons !
it can be understood as follows : scattering rate of an electron with energy @xmath12 is proportional to the density of the final state . in a quantum wire ,
density of states has @xmath120 dependence where @xmath59 is the energy at the bottom of the @xmath60th subband .
as the injected electrons move upstream they gradually cool down and their energies approach the energy at the bottom of subband 3 ( @xmath121 ) . to visualize this , imagine the horizontal line @xmath55 in fig .
[ subband ] sliding down with passage of time .
as @xmath121 is approached , electrons will increasingly scatter into subband 3 since the density of final state in subband 3 is increasing rapidly . to scatter into a final state in subband 2 or 1 that has
the same density of state as in subband 3 will require a much larger change in energy and hence a much more energetic phonon which is rare since the phonons obey bose einstein statistics .
therefore , subband 3 is the overwhelmingly preferred destination and this preference increases rapidly as electrons cool further .
consequently , beyond a certain distance , virtually all electrons are scattered to subband 3 leaving subbands 1 and 2 depleted .
this feature is a peculiarity of quasi one - dimensional system and will _ not _ be observed in bulk or quantum wells .
exact values of @xmath112 and @xmath122 depend on injection energy and electric field . in the field range @xmath123kv / cm and injection energy @xmath101 mev , @xmath124 . however , for higher values of electric field ( e.g. 2kv / cm ) or smaller values of injection energies , electrons reach classical turning point even before subbands 1 and 2 get depopulated . because of electron bunching in subband 3 , spin dephasing in the region @xmath125
is governed by equation ( 6 ) with @xmath126 .
we observe a few subdued oscillations in @xmath99 in this region because of the `` sine term '' in equation ( [ avgspin1 ] ) .
however , in the region @xmath127 subbands 1 and 2 are almost depopulated .
therefore , there is no d - p relaxation in the interval @xmath127 since only a single subband is occupied @xcite .
consequently , the ensemble averaged spin assumes a constant value @xmath128 and does not change any more .
thus in this region , one can say that spin dephasing length becomes infinite .
it should be noted that it is meaningless to study spin dephasing in the region @xmath129 because electrons do not even reach this region .
in this paper , we have used a semiclassical model to study spin dephasing of upstream electrons in a quantum wire , taking into account the subband formation .
we showed that the subband structure gives rise to rich features in the spin dephasing characteristics of upstream electrons that can not be captured in models which fail to account for the precise physics of spin dephasing and the fact that it is different in different subbands .
because spin relaxation in a multi - subband quantum wire is non - exponential ( even non - monotonic ) in space , it does not make sense to invoke a `` spin diffusion length '' , let alone use such a heuristic parameter to model spin dephasing .
finally , we have found a population inversion effect for upstream electrons .
it is possible that downstream electrons also experience a similar population inversion .
this scenario is currently being investigated .
+ + figure 1 . a quantum wire structure of length @xmath130 with rectangular cross section 30 nm @xmath131 4 nm .
a top gate ( not shown ) applies a symmetry breaking electric field @xmath34 to induce the rashba interaction .
a battery ( not shown ) applies an electric field @xmath51 ( @xmath52 ) , along the channel .
monochromatic spin polarized electrons are injected at @xmath132 with injection velocity @xmath133 .
these electrons travel along @xmath134 ( _ upstream _ electrons ) until their direction of motion is reversed due to the electric field @xmath51 .
we investigate spin dephasing of these upstream electrons .
+ figure 2 .
subband energy dispersion in the quantum wire . + figure 3 .
spatial variation of ensemble averaged spin components for driving electric field @xmath72 = 0.5kv / cm at steady state .
lattice temperature is @xmath135 k , injection energy @xmath136 mev .
electrons are injected with equal probability into the three lowest subbands .
classical turning point of subband @xmath137 electrons is denoted by @xmath112 and @xmath122 indicates the point along the channel axis where subbands @xmath138 and @xmath139 gets virtually depopulated .
injected electrons are @xmath105 polarized and @xmath53 is the point of injection .
+ figure 4 .
spatial variation of ensemble averaged spin components for driving electric field @xmath72 = 1kv / cm at steady state .
other conditions are same as in figure 3 .
+ figure 5 .
spatial variation of ensemble averaged spin components for driving electric field @xmath72 = 1.5kv / cm at steady state .
other conditions are same as in figure 3 . + figure 6 .
spatial variation of ensemble averaged spin components for driving electric field @xmath72 = 2kv / cm at steady state .
other conditions are same as in figure 3 .
+ figure 7 .
spatial variation of electron population over different subbands at steady state for driving electric field @xmath72 = 0.5 kv / cm .
other conditions are same as before .
+ figure 8 .
spatial variation of electron population over different subbands at steady state for driving electric field @xmath72 = 1 kv / cm .
other conditions are same as before
. + figure 9 .
spatial variation of electron population over different subbands at steady state for driving electric field @xmath72 = 1.5 kv / cm .
other conditions are same as before .
+ figure 10 .
spatial variation of electron population in different subbands at steady state for driving electric field @xmath72 = 2 kv / cm .
other conditions are same as before .
+ figure 11 .
spatial variation of ensemble averaged spin components for @xmath140mev , @xmath103kv / cm and lattice temperature @xmath104k .
injected electrons are @xmath105 polarized
. + figure 12 .
spatial variation of electron population over different subbands at steady state for @xmath141mev , @xmath103kv / cm and lattice temperature @xmath104k . injected electrons are @xmath105 polarized .
+ figure 13 .
spatial variation of ensemble averaged spin components for @xmath136mev , @xmath103kv / cm and lattice temperature @xmath104k . injected
electrons are @xmath142 polarized .
|
the classical drift diffusion ( dd ) model of spin transport treats spin relaxation via an empirical parameter known as the `` spin diffusion length '' . according to this model ,
the ensemble averaged spin of electrons drifting and diffusing in a solid decays exponentially with distance due to spin dephasing interactions .
the characteristic length scale associated with this decay is the spin diffusion length .
the dd model also predicts that this length is different for `` upstream '' electrons traveling in a decelerating electric field than for `` downstream '' electrons traveling in an accelerating field .
however this picture ignores energy quantization in confined systems ( e.g. quantum wires ) and therefore fails to capture the non - trivial influence of subband structure on spin relaxation .
here we highlight this influence by simulating upstream spin transport in a multi - subband quantum wire , in the presence of dyakonov - perel spin relaxation , using a semi - classical model that accounts for the subband structure rigorously .
we find that upstream spin transport has a complex dynamics that defies the simplistic definition of a `` spin diffusion length '' .
in fact , spin does not decay exponentially or even monotonically with distance , and the drift diffusion picture fails to explain the qualitative behavior , let alone predict quantitative features accurately .
unrelated to spin transport , we also find that upstream electrons undergo a `` population inversion '' as a consequence of the energy dependence of the density of states in a quasi one - dimensional structure .
|
the concepts of controllability and observability are important contribution of control theorists to the science , technology , and engineering domain . with the introduction of these structural concepts , we begin to deeply understand the relationship between the input - output description and state - space description .
the concept of controllability was first proposed for linear systems by r. e. kalman in his remarkable paper@xcite in 1960s .
controllability of nonlinear systems was further investigated by h. j. sussmann and v. jurdjevic@xcite and r. hermann and a. j. krener@xcite in 1970s .
quantum control theory has been developed ever since last century@xcite .
recently , quantum information and quantum computation is the focus of reseach@xcite .
a great progress has been made in the domain of quantum control@xcite , in which the controllability of quantum systems is a fundamental issue . in 1980s ,
controllability of quantum systems was first explored by g. m. huang , t .- j .
tarn and j. w. clark@xcite .
the different notations of controllability have been exploited in @xcite .
specially , the controllability of quantum open systems has been studied by some researchers@xcite . to manipulate a quantum system ,
it is not sufficient to just know whether or not the quantum system is controllable .
it is necessary to know how to construct permissible controls to steer the quantum system within the the specified time @xmath1 in some applications .
we need explore controllability of quantum systems under different permissible control conditions such as bounded controls and time continuous function controls , and exploit the impact of admissible control conditions on the performance indices including transition time . given a control system , whether quantum or classical , the first and most important question about its various properties is to investigate whether or not it is stable .
the most useful and general approach for studying the stability and stabilizability is the theory introduced by russian mathematician a. m. lyapunov@xcite in the late 19@xmath2 century .
lyapunov s pioneering work on stability received little attention outside russia , although it was translated into french in 1908 ( at the instigation of poincare ) , and reprinted by princeton university press in 1947 .
the publication of the work of lure and a book@xcite by la salle and lefschetz brought lyapunov s work to the attention of the larger control engineering community in the early 1960 s .
several quantitative stability concepts like finite - time stability@xcite , lipschitz stability@xcite , partial stablity@xcite and practical stability@xcite had been investigated based on lyapunov s great work in 20@xmath2 century .
set stability of dynamical systems was specifically discussed by heinen james albin@xcite in 1969 . in 2002 ,
s. k. phooi et al.@xcite further proposed the broad - sense lyapunov function and generalized the notion of stability in the sense of lyapunov . from control theory point of view , we not only need to investigate whether a dynamical system is stable or not , but also need to explore whether or not a controlled dynamical system is stabilizable by permissible control . in recent applications like quantum information storage , one of the important questions is whether or not a given state of quantum systems can be stabilizable by permissible controls . to study the stabilizability problem of quantum systems , we will have to exploit how to define the equilibrium points of schr@xmath0dinger equation .
just by investigating stabilizability problem of controlled quantum closed system , we gradually realize that the physical conditions for stabilizability of quantum systems are too strict in some concrete systems like nuclear - soin systems .
these observations indicate that we need to weaken a more general framework .
to overcome the aforementioned difficulties , we try to generalize the concepts of both controllability and stabilizability . with quantum control problems in mind , we will propose a new notion of set - stabilizability : given a pair of quantum state sets @xmath3 and @xmath4 , a quantum system is @xmath4-stabilizability from @xmath3 within the specified time @xmath1 under constrained control conditions if , for any initial state in @xmath3 , on can always find permissible controls to steer the quantum system from an arbitrary initial state in @xmath3 to another arbitrary state in the set @xmath4 in the finite time @xmath5 with @xmath6 , and to further keep the system state stay in the set @xmath4 when @xmath7 .
it is interesting to underline the following observations : ( 1 ) when @xmath4 is a one - point set , set - stabilizability notation is reduced to stabilizability .
( 2 ) when both @xmath3 and @xmath4 are the state space itself and unconstrained controls are permitted , the concept of set stabilizability is reduced to controllability proposed by kalman@xcite .
however , the set - stabilizability is not proposed to generalize the concepts of both controllability and stabilizability just for the sake of generalization , without proper motivations .
this work can be regarded as one of explorations made by many researchers who hope to investigate what kind of control goal is achievable for quantum systems by various feedback control@xcite .
stabilization of open quantum systems has been studied by @xcite . in this research , we exploit what kind of stabilizability can be expected for quantum closed systems by constrained open - loop controls , and we would like to emphasize that set stabilizability is attained by coherent control even when stabilizability itself is not an achievable control goal .
it should be also underlined that this research is very different from set value analysis@xcite and set dynamics@xcite .
the rest of this paper are organized as follows . in sect .
ii , the concept and properties of set - stabilizability are presented for general dynamical systems , and it is revealed that both controllability and stabilizability are the special case of set stabilizability . the notation of set stabilizability and stabilizability are specifically discussed for quantum closed systems . in sect .
iii , the set stabilizability and stabilizability problems of single - qubit systems are explored under different constrained controls .
we present the necessary and sufficient conditions for quantum systems and give some further discussions on the strictness of stabilizability in sect .
iv , and it is also exemplified that set - stabilizability notation can be used for exploring entanglement generation of two - qubit systems in this section .
the paper concludes with sect . v.
for the purpose of further discussions , we first recite the axiomatic definition of a dynamical system presented by kalman in 1960s@xcite . *
definition 1*. a dynamical system is a mathematical structure defined by the following axioms : ( @xmath8 ) there is given a state space @xmath9 and a set of values of time @xmath10 at which the behaviour of the system is defined ; @xmath9 is a topological space and @xmath10 is an ordered topological space which is a subset of the real numbers .
( @xmath11 ) there is given a topological space @xmath12 of functions of time defined on @xmath10 , which are the admissible inputs to the system .
( @xmath13 ) for any initial time @xmath14 in @xmath10 , any initial state @xmath15 in @xmath9 , and any input @xmath16 in @xmath12 defined for @xmath17 , the future states of the system are determined by the transition @xmath18 , which is written as @xmath19 .
this function is defined only for @xmath17 . moreover ,
any @xmath20 in @xmath10 , any @xmath15 in @xmath9 , and any fixed @xmath16 in @xmath12 defined over @xmath21\cap{t}$ ] , the following relations hold : @xmath22 @xmath23 in addition , the system must be nonanticipatory , i.e. , if @xmath24 and @xmath25 on @xmath21\cap{t}$ ] we have @xmath26 ( @xmath27 ) every output of the system is a function @xmath28reals .
( @xmath29 ) the functions @xmath30 and @xmath31 are continuous , with respect to the topologies defined for @xmath9 , @xmath32 , and @xmath12 and the induced product topologies . in other words ,
a dynamical system can be described by @xmath33 . for a dynamical system without output ,
it can be reduced to @xmath34 .
subsequently , we will recite the concepts of controllability and stabilizability in the aforementioned abstract framework of dynamical systems . * definition 2*. the dynamical system @xmath34 is controllable at time @xmath35 , if for any pair of initial state @xmath15 and target state @xmath36 in the state space @xmath9 , there always exist @xmath37 with @xmath38 and admissible control @xmath39 such that @xmath40 . * definition 3*. let @xmath41 , then the state @xmath36 of the dynamical system @xmath34 is stabilizable from @xmath9 at time @xmath35 , if for any initial state @xmath42 , there exist @xmath37 with @xmath43 and admissible control @xmath39 such that @xmath44 when @xmath45 .
suppose that both @xmath3 and @xmath4 are subsets of the state space @xmath9 , we will introduce a new concept of set stabilizability as follows .
* definition 4*. the dynamical system @xmath34 is @xmath4-stabilizable from @xmath3 at time @xmath35 , if for any initial state @xmath46 and another arbitrary target state @xmath47 , there exist @xmath37 with @xmath43 and admissible control @xmath39 such that @xmath40 and @xmath48 when @xmath49 . from the aforementioned definition
, we can easily establish the following properties of set stabilizability .
* proposition 1 * suppose that @xmath50 , @xmath3 and @xmath4 are subsets of @xmath9 ( @xmath51 ) let @xmath52 , a dynamical system @xmath34 is @xmath4-stabilizable from @xmath50 at time @xmath35 if it is @xmath4-stabilizable from @xmath3 at time @xmath35 .
( @xmath53 ) let @xmath54 , @xmath34 is stabilizable from @xmath3 at time @xmath35 if @xmath55 is @xmath4-stabilizable from @xmath3 at time @xmath35 .
( @xmath56 ) a dynamical system @xmath34 is controllable at time @xmath35 if and only if it is @xmath9-stabilizable from @xmath9 at time @xmath35 .
( @xmath57 ) let @xmath58 is a subspace of @xmath9 , a dynamical system @xmath34 is controllable on @xmath58 at time @xmath35 if and only if it is @xmath58-stabilizable from @xmath58 at time @xmath35 .
( @xmath59 ) the state @xmath36 of dynamical system
@xmath34 is stabilizable from @xmath3 at time @xmath35 if it is @xmath60-stabilizable from @xmath3 at time @xmath35 .
the proof of the aforementioned properties are quite straightforward , but the properties are of major importance : set stabilizability is regarded as the generalization of both controllability and stabilizability . because we need to steer systems within the specified finite time in many applications , we further propose the notation of set stabilizability within the specified time span @xmath1 as follows . *
definition 5*. given @xmath61 , the dynamical system @xmath34 is @xmath4-stabilizable within the specified time @xmath62 from @xmath3 at time @xmath35 , if for any initial state @xmath46 and another arbitrary target state @xmath47 , there exist @xmath37 with @xmath63 and admissible control @xmath39 such that @xmath40 and @xmath48 when @xmath49 . in this paper
, we will study a special subclass of dynamical systems , those which are quantum closed systems , to illustrate why set - stabilizable is useful . before investigating stabilizability of quantum closed systems
, we will have to discuss how to define an equilibrium point of schr@xmath0dinger equation @xmath64 where @xmath65 is a pure state in hilbert space .
for the purpose of simplicity , we set @xmath66 in the whole paper . from the mathematical point of view , it seems that zero vector is the sole equilibrium point of eq .
( [ s ] ) .
unfortunately , zero vector is nonsense from the viewpoint of physics .
we need to give some further investigation on this issue . to obtain some intuitive pictures about the solution of eq .
( [ s ] ) , let us consider a two - level quantum system given by @xmath67 where @xmath68 . from the physical point of view , both @xmath69 and @xmath70 are the equilibrium points of eq .
( [ qubit ] ) . from the mathematical point of view , the solution @xmath65 of eq .
( [ qubit ] ) with the initial state @xmath71 ( or @xmath72 ) satisfies @xmath73 ( or @xmath74 ) . in other words , @xmath75 if @xmath76 with @xmath77 .
it is also revealed that @xmath78=0 $ ] for @xmath77 , where @xmath79 $ ] is lie bracket which is specified by @xmath80=ab - ba$ ] .
based on aforementioned observations , we have the following definition : * definition 6 : * @xmath81 is called an equilibrium point of eq .
( [ s ] ) if @xmath82=0 $ ] .
* remark : * denote @xmath83 , then eq . ( [ s ] ) can be written as @xmath84 $ ] .
this implies that @xmath85 is the static solution of @xmath84 $ ] if @xmath81 is called the equilibrium point of eq .
( [ s ] ) .
* proposition 2 : * @xmath81 is an equilibrium point of eq .
( [ s ] ) if and only if @xmath81 is an eigenvector of hamiltonian @xmath86 .
consider a controlled finite - dimensional quantum system without output @xmath87|\psi(t)\rangle\ ] ] where @xmath65 is a pure state in hilbert space , and @xmath88 and @xmath89 are system hamiltonian and controlled hamiltonian , respectively .
for the controlled quantum system governed by ( [ 1 ] ) , set stabilizability can be specially defined as follows : * definition 4a * : for an arbitrary initial state @xmath90 and another arbitrary target state @xmath91 , the quantum system eq .
( [ 1 ] ) is @xmath4-stabilizable from @xmath3 at time @xmath14 if there exist a finite time @xmath92 and permissible control @xmath93 such that the system can be transferred from @xmath94 to @xmath95 and @xmath96 when @xmath45 .
to overcome decoherence , we need to steer quantum systems within the specified time span @xmath1 .
in this situation , we can present the notion of set stabilizability within the specified time span @xmath1 by modifying the definition 4a . * definition 5a : * given @xmath61 , @xmath97 @xmath90 and @xmath91 , the quantum system eq .
( [ 1 ] ) is @xmath4-stabilized within the specified time span @xmath1 from @xmath3 at time @xmath14 if there exist a finite time @xmath98 with @xmath63 and permissible control @xmath93 such that the system can be transferred from @xmath94 to @xmath95 and @xmath96 when @xmath45 .
in this section , we will focus on set stabilizability of two - level quantum systems .
let @xmath69 and @xmath70 be a basis of hilbert space of two - level quantum systems .
a two - level quantum system is governed by @xmath99|\psi(t)\rangle\ ] ] where @xmath100 , @xmath101 , @xmath102 , @xmath103 and @xmath104 are adjustable scale functions . and @xmath105 and @xmath106 . for simplicity , we assume that @xmath107 in this paper . the state space is denoted as @xmath108 and can be further parameterized as : @xmath109,\phi\in[0,2\pi)\}\ ] ] where @xmath110 and @xmath111 are bloch parameters . for the further discussions , we introduce the following notations : @xmath112 and @xmath113 where @xmath114 represents a circle on a bloch sphere , @xmath115 is regarded as a point on the circle @xmath116 , @xmath117 is a fixed value in @xmath118 $ ] and @xmath119 is a fixed value in @xmath120 . to obtain some intuitive pictures of @xmath114 and @xmath115 ,
we plot them in fig [ fig4 ] .
two kind of permission control sets are considered : bounded control with bound @xmath121 @xmath122 and bounded time - continuous controls with bound @xmath121 @xmath123 where @xmath124 is the space of time - continuous functions . in this subsection
, we will first establish the necessary and sufficient conditions for that the given target state is stabilizable by bounded controls in @xmath125 . * theorem 1 * : given @xmath126 $ ] and @xmath127 , @xmath128 , the controlled qubit system eq .
( [ 2 ] ) is @xmath115-stabilizable from @xmath9 at @xmath14 by @xmath125 if and only if @xmath129 * proof : * see in appendix .
subsequently , we will further give a theorem for circle - set stabilizability of two - level quantum systems .
* theorem 2 * : for @xmath130}$ ] , @xmath131 and @xmath132 , the controlled quantum system eq .
( [ 2 ] ) is @xmath114-stabilizable from @xmath9 by @xmath133 . *
_ proof _ * : see in appendix .
* remark 2 : * it should be underlined that @xmath134 when @xmath135 where @xmath136 .
this means that for all @xmath137 , it can be dynamically stored by permissible control in @xmath133 .
we immediately conclude that the controlled two - level quantum system eq .
( [ 2 ] ) is @xmath114-stabilizable from @xmath9 by @xmath125 for all @xmath138}$ ] since @xmath133 is the subset of @xmath125 . in this subsection
, we will establish a theorem to explore whether or not the system ( [ 2 ] ) is @xmath139-stabilized from @xmath9 within the specified time @xmath1 by admissible controls in @xmath140 . * theorem 3 * : for any @xmath138}$ ] and any @xmath141 , the controlled two - level system eq .
( [ 2 ] ) is @xmath114-stabilized within the specified time @xmath1 from @xmath9 at @xmath14 by @xmath133 if @xmath142 furthermore , eq . ( [ theorem2 ] ) can be improved if the permissible controls are chosen from @xmath125 instead of @xmath133 .
* theorem 4 * : let @xmath126 $ ] and @xmath127 , @xmath128 , the controlled two - level system eq .
( [ 2 ] ) is @xmath114-stabilizable within the specified time @xmath1 from @xmath9 at @xmath14 by @xmath125 if @xmath143 or @xmath144 in this subsection , we first exploit the sufficient conditions for set - stabilizability with both time constrain @xmath1 and energy constrain @xmath145 . for the given @xmath146 , @xmath147 , @xmath117 and @xmath119 , we will investigate whether or not there exist permissible controls such that the controlled qubit system eq .
( [ 2 ] ) is @xmath114-stabilizable with the following constrained conditions @xmath148 and @xmath149dt\leq{e_s},\ ] ] when the unbounded control are permitted , we have the following theorem : * theorem 5 * : given @xmath1 and @xmath145 , for @xmath97 @xmath146 , @xmath147 , @xmath117 and @xmath119 , there exist unbounded controls such that the controlled qubit system eq .
( [ 2 ] ) is @xmath114-stabilizable with time - energy constrains eqs .
( [ time ] ) and ( [ energy ] ) if @xmath150 and @xmath151 .
* proof * : if @xmath150 and @xmath151 , then there exists at least @xmath152 such that @xmath153 and @xmath154 and @xmath155 hold for @xmath97 @xmath146 , @xmath147 , @xmath117 and @xmath119 .
therefore , we establish the sufficient conditions for @xmath114-stabilizable with time - energy constrains .
the necessary and sufficient conditions can be established for that the state @xmath156 of quantum system eq .
( [ 1 ] ) is stabilizable from @xmath3 .
* theorem 6 * : the state @xmath156 of quantum system eq .
( [ 1 ] ) is stabilizable from @xmath3 if and only if two following conditions are satisfied : ( r1 ) @xmath156 is reachable from any state in @xmath3 by a coherent control @xmath157 ; ( e2 ) there exists a static control hamiltonian @xmath89 such that @xmath81 is an eigenvector of hamiltonian @xmath158 . _
remark : _ unfortunately , the condition ( e2 ) is very difficult to be satisfied . for two - level quantum systems
( [ 1 ] ) , the condition ( e2 ) is reduced to eq .
( [ 4 - 0 ] ) . to obtain some intuitive pictures of how eq .
( [ 4 - 0 ] ) is quite strict , we write eq .
( [ 4 - 0 ] ) as @xmath159 in fig [ fig3 ] .
+ because @xmath160 holds for nuclear spin systems@xcite , we immediately realize from fig [ fig3 ] that eq .
( [ 4 - 0 ] ) is very strict in some experimental quantum systems .
this observation about stabilizability is in remarkable contrast with that about set - stabilizability in theorem 2 : the controlled two - level quantum system eq .
( [ 2 ] ) is @xmath114-stabilizable from @xmath9 at @xmath14 by @xmath133 for @xmath130}$ ] , @xmath131 and @xmath132 . for two - level quantum system @xmath161|\psi(t)\rangle\ ] ]
the necessary condition for ( e2 ) is that @xmath162 with @xmath163 , this is also in remarkable contrast with the observation that the controlled two - level quantum system eq .
( [ 2x ] ) is @xmath114-stabilizable from @xmath9 at @xmath14 by bounded control @xmath164 for @xmath130}$ ] , @xmath131 and @xmath132 . for two - level quantum system @xmath165|\psi(t)\rangle\ ] ] the necessary condition for ( e2 ) is that @xmath162 with @xmath166 .
this is also in remarkable contrast with the observation that the controlled two - level quantum system eq .
( [ 2y ] ) is @xmath114-stabilizable from @xmath9 at @xmath14 by bounded control @xmath167 for @xmath130}$ ] , @xmath131 and @xmath132 . from the aforemention discussions , we realize that the necessary conditions are too strict for stabilizability of quantum systems but some kinds of set stabilizability notions are available .
we further present another example to illustrate that sufficient conditions are easier to be fulfilled for set stabilizability .
consider a controlled two - qubit system governed by the equation @xmath168|\psi(t)\rangle\ ] ] where @xmath169 is system hamiltonian and @xmath170 are controlled hamiltonian , respectively .
let @xmath171 be a subspace for two - qubit system , we have @xmath172,\phi\in[0,2\pi)\}\ ] ] denote a maximal entangled state subset as @xmath173 introducing @xmath174 and @xmath175 , we have @xmath176 and @xmath177 and @xmath178 from the geometric point of view , the subspace @xmath179 can be regarded as a bloch sphere of the encoded qubit @xmath180,\phi\in[0,2\pi)\}\ ] ] and the maximal entangled state subset @xmath181 can be treated as a circle on the bloch sphere .
let @xmath182 with @xmath183 , @xmath184 and @xmath185 , then eq . (
[ 5 - 1 ] ) can be rewritten as @xmath186|\psi^{l}(t)\rangle\ ] ] let @xmath187 and @xmath188 for @xmath189 , we have the following corollaries : * corollary 1 * : for @xmath131 and @xmath132 , the controlled qubit system eq .
( [ 5 - 1 ] ) is @xmath190-stabilizable from @xmath191 at time @xmath14 by @xmath192 .
* corollary 2 * : for @xmath132 , the controlled qubit system eq .
( [ 5 - 1 ] ) is @xmath190-stabilizable within @xmath1 from @xmath191 at time @xmath14 by @xmath192 if @xmath193
. * corollary 3 * : for @xmath132 , the controlled qubit system eq .
( [ 5 - 1 ] ) is @xmath190-stabilizable within @xmath1 from @xmath191 at time @xmath14 by @xmath194 if @xmath195 or @xmath196 .
the aforementioned results suggest that set - stabilizability can be used for studying entangled state generation problem .
in summery , we explored set - stabilizability by constrained open - loop controls in this research .
both controllability and stabilizability can be regarded as the special case of set - stabilizability .
the necessary and sufficient conditions are also established for stabilizability of quantum closed systems , and it is further revealed that the necessary conditions are too strict for stabilization of some concrete quantum systems like nuclear spin systems .
this further justifies the introduction of the set - stabilizability notion .
we also clarify how to define an equilibrium point of schr@xmath0dinger equation from the physical point of view .
strictly speaking , stabilizability problems for quantum systems should be discussed in terms of density operators and master equations .
it is exemplified that set - stabilizability can be used for investigating quantum information processing problems including quantum information storage and entangled state generation .
in our opinion , it should be further investigated that what kind of set - stabilizability is achieved for quantum open systems .
partially supported by the national nature science foundation of china under grant nos .
61273202 and 61134008 .
\(1 ) first , we need to prove that if eq .
( [ 4 - 0 ] ) holds , there always exist bounded controls in @xmath125 to transit the qubit system from an arbitrary initial state @xmath197 to another arbitrary target state @xmath198 with @xmath199 , and @xmath200 with @xmath201 .
choose the permissible controls as follows : @xmath202 & t\in[t_0,t_{f})\\ \omega_0\cdot\tan\theta_f\cdot\cos\phi_f & t\in[t_f,+\infty ) \end{array}\right.\ ] ] and @xmath203 & t\in[t_0,t_{f})\\ \omega_0\cdot\tan\theta_f\cdot\sin\phi_f&t\in[t_f,+\infty ) \end{array}\right.\ ] ] where @xmath204 $ ] and @xmath205 , and the designed parameters are given by @xmath206 @xmath207 @xmath208 and @xmath209 where @xmath210 and @xmath211 is such an integer that @xmath212 if there exist permissible controls in @xmath125 such that eq .
( [ 2 ] ) is @xmath115-stabilizable from @xmath9 , then there exists static hamiltonian @xmath220 such that @xmath158 satisfies @xmath221=0 $ ] .
note that @xmath221=0 $ ] if and only if there exists such a scale @xmath222 that @xmath223 $ ] .
therefore , the following equations should hold simultaneously : @xmath224 and @xmath225 and @xmath226 thus @xmath227 and @xmath228 . recall that @xmath229 , i.e. , @xmath230 and @xmath231 , we conclude that eq .
( [ 4 - 0 ] ) holds . to prove the theorem ,
it is sufficient to show there always exist bounded time - continuous functional controls in @xmath133 to transfer the qubit system from an arbitrary initial state @xmath197 to another arbitrary target state @xmath232 with @xmath199 , and @xmath233 .
we construct the following permissible controls @xmath234 and @xmath235 where @xmath236 & t\in[t_{1},\frac{t_1+t_{f}}{2})\\ g[1-(\frac{2t - t_{1}-t_{f}}{t_{f}-t_{1}})^{n } ] & t\in[\frac{t_{1}+t_f}{2},t_{f})\\ 0 & t\in[t_f,+\infty ) \end{array}\right.\ ] ] with @xmath237 and @xmath238 and @xmath239 and @xmath240 because the aforementioned analyses hold for any pair of initial and target states , this implies that one can always construct a local @xmath243order function to dynamically modulate amplitude and further steer quantum systems from an arbitrary initial state to another arbitrary target state .
s. g. schirmer , i. c. h. pullen and a. i. solomon , _ controllability of quantum systems , in hamiltonian and lagrangian methods in nonlinear control _ , proceedings of the second ifac workshop , seville , spain , 2003 , edited by a. astolfi and a. j. van der schaft ( elsevier science ltd .
, new york 2003 ) 311 - 316
|
we explore set - stabilizability by constrained controls , and both controllability and stabilizability can be regarded as the special case of set - stabilizability .
we not only clarify how to define equilibrium points of schr@xmath0dinger equations , but also establish the necessary and sufficient conditions for stabilizability of quantum systems .
unfortunately , it is revealed that the necessary conditions are quite strict for stabilizability of some concrete quantum systems like nuclear spin systems , and this further justifies the introduction of set - stabilizability notion .
it is also exemplified that set - stabilizability can be used for investigating quantum information processing problems including quantum information storage and entangled states generation . , , quantum systems , controllability , stabilizability
|
the field of network analysis helps us to look at the study of an individual component as a part of a complex social structure and its interactions @xcite .
it explains various phenomena in a wide variety of disciplines ranging from physics to psychology to economics .
the theory is adept at finding the causal relationships between network attributes such as the position of a node and the specific ties associated with it , and the fitness of the said node @xcite .
such relationships , that seemed thoroughly random to the eyes of a researcher only about a decade before , have now been vastly studied and documented @xcite .
we aim to further investigate the very interesting idea that human behavior is predictable to a fair degree @xcite using the bollywood network as a model for this purpose . making nearly one thousand feature films and fifteen hundred short films per year ,
the indian film industry is the largest in the world @xcite which has held a large global population in more spheres of its existence than just entertainment .
it mirrors a changing society capturing its peaks and valleys over time and impacts the opinions and views of the diverse populace @xcite .
an example that can be stated as a proof of this was exhibited when the number of indian tourists to spain increased by @xmath6 in the year succeeding the box office success of the movie ` zindagi na milegi dobara ' , which extensively portrayed tourist destinations in spain , and also in the fact that switzerland , depicted in various popular yesteryear indian films ( movies ) , remains a popular tourist destination for indians to date @xcite . the hollywood co - actor network is a social network that has invited a fair amount of interest in the past @xcite , studies being conducted using relational dependency network analysis , layered label propagation algorithm and pagerank algorithm @xcite . in comparison , its much larger counterpart in india has been largely ignored . flourishing with a @xmath7 growth from 2009 to 2010 @xcite and a further @xmath8 growth from 2010 to 2011 @xcite , it is an industry that sees blazingly fast growth , leading us to expect drastic changes in small time frames .
we study the bollywood industry because it provides a fair ground to capture the temporal changes in a network owing to its rapidly changing character . using data from the past 100 years ,
we construct a network for every five year period .
the nodes can be classified into the three distinct categories : 1 ) lead male actors , 2 ) lead female actors and 3 ) supporting actors .
we analyze the structural properties of this network and further study its spectral properties using the random matrix theory ( rmt ) .
though originally rooted in nuclear physics @xcite , rmt has found widespread applications in different real systems such as the stock - market indices , atmosphere , human eeg , large relay networks , biological networks and various other model networks . under the framework of rmt ,
such systems and networks follow the universal gaussian orthogonal ensemble ( goe ) statistics . though there exist other universality classes such as gaussian unitary ensemble and gaussian symplectic ensemble @xcite , which have also been extensively investigated in rmt literature
, we focus only on goe statistics as spectra of various networks have been shown to rest with this universality class @xcite .
the universality means that universal spectral behaviors , such as statistics of nearest neighbor spacing distribution ( nnsd ) are not only confined to random matrices but get extended to other systems .
a wide variety of complex systems fall under this class , i.e. their spectra follow goe statistics ( @xcite and references therein ) .
we collect all bollywood data primarily from the movie repository website _ www.bollywoodhungama.com _ and henceforth from _
www.imdb.com _ and _ www.fridayrelease.com _
( now renamed as _ www.bollywoodmdb.com_ ) and we generate no additional data .
the website _ www.bollywoodhungama.com _ previously known as _
diafm.com_ , is a reputed bollywood entertainment website , owned by hungama digital media entertainment , which acquired bollywood portal in 2000 .
we use python code to extract names of all the movies and their corresponding information for a massive period of hundred years spanning from 1913 to 2012 .
initially we document the names of all films as per their chronological sequence ( latest to oldest ) from the websites by incorporating the desired url @xcite in the code along with a built - in string function which takes the page numbers ( 932 pages in `` released before 2012 '' category and 24 pages in `` released in 2012 '' category ) as input .
each film of every page bears a unique cast i d in the website , navigating to which via `` movie info '' provides us complete information about the film . in the python code , we store the unique cast ids of films in a temporary variable and retrieve relevant information using appropriate keywords from the respective html page .
we also manually browse through other aforementioned websites in order to collect any yearwise missing data , if any .
thus we get the data in terms of names of the movies and names of the actors for 100 years .
we then merge the data from all the websites and omit repetitions .
a total of 8931 movies have been documented so far in bollywood from 1913 till 2012 . harvesting the complete data took approximately 2000 hours of work over a 4-month period , which includes manual verification , formatting , removal of typos and compilation of the data .
considering the rapidly changing nature of the bollywood network , we assort the curated massive bollywood data in to 20 datasets each containing movie data for five - year window periods , as this is an apt time frame within which the network constructed is large enough to study the important network properties , and is not too large to miss any crucial evolutionary information .
since the number of movies and their actors in the time span 1913 - 1932 were scanty and could not have yielded any significant statistics , we merge the 1913 - 1932 datasets and present as a single dataset 1928 - 1932 .
we create database of all actors who had appeared in the bollywood film industry ever since its inception in five - year window periods , as mentioned in the previous version of the manuscript , by extracting them from the movie information using python algorithm and we assign a unique i d number to each actor in every span which we preserve throughout our analysis .
we take care of ambiguities in spellings of names of actors presented in different websites by extensive thorough manual search and cross - checking to avoid overlapping of information and duplication of node identities while constructing networks .
tracking by their unique i d numbers assigned by us , we create a co - actor database for each span where every pair of actors who had co - acted in a movie within those five years are documented .
we then construct an adjacency list of all available combinations of co - actors . treating every actor as a node and every co - actor association as a connection
, we create a co - actor network of the largest connected component for every span .
we pick the actors appearing as the protagonist ( occupant of the first position ) in the movie star cast list from the movie star cast database created by us and observe that they incidentally are male actors in almost all movies with some rare exceptions . on extensive manual search based on popularity ,
award nominations we find that those male actors appear as a lead in the respective movies which made our attempt to extract lead male actors even easier .
we could very well define the lead male actor as the protagonist in the star cast of at least five films in consecutive five - year spans and extract them from the movie star cast list using python code while we were unable to find any proper definition for lead female actors as the second position of the movie star cast list is alternately occupied by either female actors or supporting actors , making it difficult to extract them only based on the network data as described .
hence we handpick the lead female actors from the movie star cast database for all the spans based on their popularity , award nominations and create their database .
we consider filmfare award nominations as the best means to assess the success rates of all lead actors of bollywood and distinguish the lead female actors from the rest .
filmfare awards were first introduced by the the times group @xcite after the central board of film certification ( cbfc ) was founded by indian central government in 1952 to secure the identity of indian culture .
the reason behind choosing filmfare awards amongst all other awards in our analysis is that it is voted both by the public and a committee of experts , thus gaining more acceptance over the years . instead of the awards bagged we rather take into account the award nominations in order to avoid the interplay of some kind of bias affecting the decision of the cbfc committee in selecting the winner . by manual navigation through every year of filmfare awards available on the web , we create a database of all categories of filmfare awards and
extract their respective nominees chronologically from the html pages using python codes .
henceforth we use c++ codes to count the number of times every actor is nominated in each five - year span .
thus we obtain a complete list of all actors in each span along with their number of filmfare nominations .
considering @xmath9 to be the fraction of vertices with the degree k , the degree distribution of the constructed networks is plotted with @xmath9 .
it has been sufficiently proven that the degree distribution of real world networks are not random , most of them having a long right tail corresponding to values that are far above the mean @xcite .
we define the betweenness centrality of a node i , as the fraction of shortest paths between node pairs that pass through the said node of interest @xcite .
@xmath10 where @xmath11 is the number of geodesic paths from @xmath12 to @xmath13 that passes through @xmath14 and @xmath15 is the total number of geodesic paths from @xmath12 to @xmath13 . in the current work
, the concept of a payoff has been borrowed from the field of management @xcite , and adapted to suit the bollywood network analysis .
payoff has elucidated the success of the center and non - center agents in a unique efficient star network @xcite .
we use an improvised version of payoff as a means to assess success rates of the nodes in bollywood . for the purpose of devising net payoff ( @xmath16 ) , we study the datasets two at a time ( accounting for ten years ) and use the following definition : @xmath17 where , @xmath18 is the change in degree of a particular node @xmath14 in two consecutive spans .
@xmath19 is its normalized degree in a particular span given as @xmath20 with @xmath21 being the degree of the node i and @xmath22 and @xmath23 being the maximum and minimum degree in that particular span , respectively .
the third term sums over all nodes j that node i has worked with where @xmath24 and @xmath25 are the number of movies that the node i and j has worked in respectively and @xmath26 the number of times the node @xmath27 has worked with the node @xmath14 in the considered time window .
the averages denoted in the net payoff ( eq . [ eq_net payoff ] ) refer to the values averaged over the two consecutive datasets . based on the values of @xmath16 ,
the actors of every set studied were ranked and lists made .
due to the absence of a unifying framework that can be used to evaluate the success of films and their actors in the years before the inception of filmfare awards in 1954 , we restrict our analysis on assessment of success to the time periods spanning from 1954 and onwards . in order to adumbrate the success of actors in the industry , we define overlap as the intersection of sets of co - actors that an actor has worked with , in two consecutive time frames .
the random matrix studies of eigenvalue spectra consider two properties : ( 1 ) global properties such as spectral distribution of eigenvalues @xmath28 , and ( 2 ) local properties such as eigenvalue fluctuations around @xmath28 .
eigenvalue fluctuations is the most popular one in rmt and is generally obtained from the nnsd of eigenvalues .
we denote the eigenvalues of a network by @xmath29 and @xmath30 . in order to get universal properties of the fluctuations of eigenvalues ,
it is customary in rmt to unfold the eigenvalues by a transformation @xmath31 , where @xmath32 is average integrated eigenvalue density .
since we do not have any analytical form for @xmath33 , we numerically unfold the spectrum by polynomial curve fitting @xcite . after unfolding ,
average spacings are unity , independent of the system . using the unfolded spectra ,
spacings are calculated as @xmath34 .
the nnsd is given by @xmath35 for intermediate cases , the spacing distribution is described by brody distribution as @xmath36 where @xmath37 and @xmath38 are determined by the parameter @xmath39 as follows : @xmath40^{\beta+1}\end{aligned}\ ] ] this is a semi - empirical formula characterized by parameter @xmath39 .
as @xmath39 goes from 0 to 1 , the brody distribution smoothly changes from poisson to goe .
fitting spacing distributions of different networks with the brody distribution @xmath41 gives an estimation of @xmath39 , and consequently identifies whether the spacing distribution of a given network is poisson , goe , or the intermediate of the two @xcite .
the nnsd accounts for the short range correlations in the eigenvalues .
we probe for the long range correlations in eigenvalues using @xmath42 statistics which measures the least - square deviation of the spectral staircase function representing average integrated eigenvalue density @xmath43 from the best fitted straight line for a finite interval of length @xmath44 of the spectrum and is given by @xmath45 ^ 2 d\overline{\lambda}\ ] ] where @xmath46 and @xmath47 are regression coefficients obtained after least square fit .
average over several choices of x gives the spectral rigidity , the @xmath42 . in case of goe statistics , the @xmath42 depends logarithmically on l , i.e. @xmath48 ) against normalized degrees ( @xmath49 ) of bollywood actors over 1953 - 2012 .
actors and their corresponding betweenness centrality are represented in same color . ]
the degree distribution of the bollywood networks follow power law , as expected based on the studies of other real world networks @xcite . but
an observation that defies intuition is that the most important nodes of the industry , acknowledged as the lead male actors , do not form the hubs of the constructed network , but instead have a moderate degree and also maintain it along sets of data that were studied ( si tables 1 - 6 ) .
considering the network on an evolutionary scale , this is a property that gains more prominence during the later sets of the data , while the network maintains power law throughout the entire timespan ( si fig .
[ deg_dist ] ) .
the prominent supporting actors of the era form the hubs of the industry in respective time frames .
this counterintuitive nature of the above observation can be explained by the fact that these actors collaborate with more nodes and take on more projects in a given time period .
hence they can be said to be instrumental in establishing connections in the network .
the scale - free behavior of the bollywood industry can be elucidated by the fact that newcomers in the industry in general aspire to act with the lead actors of the era , who intuitively form associations with high degree nodes , thus illustrating the preferential attachment property prevalent in bollywood networks @xcite . ) of top three lead male actors . in each time span
plotted against the respective time frames .
they are ranked ( as 1 , 2 and so on ) based on their number of filmfare award nominations . ` * ' denotes no filmfare award nominations .
actors and their corresponding rankings are represented in same color . ] by virtue of the sinusoidal function used in ( eq .
[ eq_net payoff ] ) , the nodes with a moderate degree lead the net payoff list with both low degree and high degree nodes trailing behind .
the inverse of the change in degree favors nodes that preserve their degree over the years hence giving a higher net - payoff to actors who preserve their degrees over the various datasets .
successful supporting actors , although bear a high degree , appear quite high in the scale of @xmath16 because they have relatively higher values of @xmath50 . though interplay of various contrasting factors influence the appearance of lead male actors in @xmath16 list , they appear high in absolute scale of @xmath16 in all the sets under consideration except the ones corresponding to 1973 - 77 and 1978 - 82 .
three of the top five filmfare award nominees in lead male actor category appear as top three lead male actors in @xmath16 list in respective time frames ( fig .
[ figure2 ] and si tables 1 - 6 ) .
this observation is more pronounced in case of the lead female actors . as observed in fig .
[ figure3 ] and si tables 7 - 12 , the three lead female actors having secured the maximum number of filmfare award nominations in a particular span of time , appear as the leading nodes in their respective @xmath16 list , a trait that is more consistent in the more recent datasets . from the above analysis based on payoff
it is supposed that possessing moderate degree and maintaining it are properties followed by the nodes that stand successful in bollywood industry and can be contemplated as keys to success . ) of top five lead female actors in each time span
plotted against the respective time frames .
they are ranked ( as 1 , 2 and so on ) based on their number of filmfare award nominations . ` * ' denotes no filmfare award nominations .
actors and their corresponding rankings are represented in same color . ] succeeding the economic liberalization in 1991 , the inclusion of diverse socio - political - economic issues in mainstream bollywood movies found favor with the audience @xcite . at around this period ,
hollywood started gaining popularity among the indian population owing to the advent of private movie channels and the internet .
these factors coupled together affected the structure of the network , which might be the underlying reason behind the observed variations in the network properties , pre , post and during liberalization . a steep rise in the bollywood network size 1993 onwards ( fig .
[ figure4 ] ) might be one of the manifestations of this shift in economic policies .
the status of an ` industry ' being conferred upon bollywood in 1998 might be a result of this increased size of the network @xcite .
the comparatively larger shift of the network properties with the advent of liberalization as opposed to that caused by the introduction of the filmfare awards in 1954 , can lead us to conclude that mainstream bollywood is largely driven by economic concerns rather than artistic ones .
the number of times an actor is nominated for the filmfare awards while they remain a lead actor , when plotted with their overlap ( as defined before ) , shows that 22 among the 25 actors exhibit an approximate direct proportionality ( fig .
[ figure5 ] ) emphasizing on the importance of winning combinations .
overlap being one of the probable factors deciding
the success of a node might explain the reason for the formation of social groups , and co - operation among them in the society @xcite .
high degree nodes indubitably have high betweenness centrality .
actors with high betweenness centrality seem to have a relatively larger span in the industry even if their popularity levels , measured as the number of filmfare award nominations , is not markedly high .
nodes with the highest betweenness centrality of all datasets are found to be male actors ( except helen ) , whether lead or supporting , adumbrating the gender disparity in bollywood .
incidentally , few of the nodes bearing moderate and low degree also exhibit high betweenness centrality and also have a long span in the bollywood industry ( fig .
[ figure1 ] ; si fig .
[ between_deg ] and table 2 ) .
this indicates that actors exhibiting mobility between diverse bollywood circles seem to have an advantage of a long span , though we are far from concluding that this is the only factor affecting the life span of a node .
there exist examples from social and biological systems which also support the importance of cooperation and mobility @xcite .
( represented by @xmath51 ) of lead male actors and their filmfare award nominations @xmath52 ( represented by @xmath53 ) against their respective time spans .
time span here represents respective individual spans of lead male actors in bollywood industry , for example dilip kumar had a long span stretching between 1943 and 1998 whereas hrithik roshan has a short spell 1998 onwards . ] the spectral density , @xmath54 ) of the connectivity matrix of bollywood networks exhibit a triangular distribution ( si fig .
[ spectral_fig ] and discussion in @xcite ) , hence providing evidence supporting its scale - free nature @xcite .
the eigenvalue distribution of the bollywood networks show a high degeneracy at @xmath55 , deviating from the commonly observed degeneracy at @xmath56 in most of the real world networks studied ( for example , biological networks @xcite ) .
this degeneracy at @xmath55 can be attributed to the presence of clique structures in the network @xcite .
presence of dead - end vertices in spectrum and motif joining or duplication have been used as plausible explanations to widespread degeneracy at @xmath56 observed in biological networks @xcite .
factors affecting a social network are vastly different from those affecting a biological network , hence making the nature of their spectra varied . owing to a relatively smaller number of nodes in the networks constructed for the periods 1913 - 17 , 1918 - 22 and 1923 - 27 , a bulk does not appear in their eigenvalue distributions .
the distributions corresponding to the datasets of 1928 - 57 , 1983 - 87 and 2003 - 12 very clearly show the presence of a few eigenvalues outside the bulk ( si fig .
[ bulk_eigen ] and fig .
[ figure6 ] ) , which is formed by the rest of the eigenvalues . while the largest eigenvalue is distinctly separated from the bulk , which is a well - known spectral feature of an undirected network @xcite , existence of other eigenvalues outside the bulk
probably indicate the existence of distinct bollywood guilds @xcite further portending an evolving network structure . the spectral data as well as the data regarding the betweenness centrality of the networks , corresponding to the time periods after 1998 - 02 , suggest that there has been a drastic change in the underlying network structure since then .
this marked change in the more recent datasets in comparison to the older ones , is clearly illustrated by the presence of several eigenvalues outside the bulk ( fig .
[ figure6 ] ) , and the presence of a lesser number of low degree nodes with a high betweenness centrality(fig .
[ figure1 ] ) .
this indicates that the community structures in the bollywood network have gotten more inter - interconnected post 1998 - 02 , leading the authors of this paper to conclude that bollywood is becoming increasingly systematic with time .
we fit the nnsd of bollywood networks by the brody distribution ( eq . [ eq_brody ] ) and find that the value of @xmath39 comes out to be close to @xmath57 for all the datasets .
this implies that the nnsd of bollywood datasets follow goe statistics of rmt ( eq . [ eq_nnsd ] and si fig .
[ nnsd ] ) bringing bollywood networks under the universality class of rmt @xcite . to examine the long range correlations , we calculate spectral rigidity via the @xmath42 statistics of rmt using eq .
[ eq_delta3 ] by taking same unfolded eigenvalues of different datasets as used for the nnsd calculations .
the value of @xmath44 for which the @xmath42 statistics follows rmt prediction ( eq . [ eq_delta3_goe ] ) is given in the table 1 and the detailed plots are deferred to @xcite as si fig .
[ fig_delta_3 ] .
the @xmath42 statistics which provides a measure of randomness in networks @xcite clearly indicate that the dataset corresponding to the 1963 - 67 timespan has the most random underlying network structure when compared with the other datasets .
this notable feature of this timespan can probably be attributed to the consecutive wars that india was a part of in the years 1962 and 1965 , which in turn lead to an extreme economic crisis in the country .
as shown by the decreasing value of l since 1933 , the networks have a trend of diminishing randomness.the dataset corresponding to 1948 - 52 witnessed a breach from this trend , probably due to the drastic political and financial changes post indian independence in 1947 .
one of the most crucial points exhibited in the analysis based on eigenvalue distribution and betweenness centrality is that , before the year 1998 the structure of the networks had either well segregated clusters or extreme random interactions , while post 1998 the structures seem to maintain a fairly consistent randomness ( randomness measured by the value of l ) .
although bollywood networks for different spans demonstrate varying amounts of randomness as suggested by the changing values of l in the @xmath42 statistics , observation of universal goe statistics of the nnsd puts forward the evidence to show that a sufficient amount of randomness is possessed by all the sets .
the efficiency of many real world systems such as the financial markets , the climatic system , neuronal systems etc , has been aided by their stochastic nature which leads to randomness @xcite .
bollywood network also provides an example to aid this relationship , as the industry has survived various valleys and crests since its inception , including in times of dire socio - economic crisis @xcite .
99 newman m , e , j ( 2001 ) scientific collaboration networks : i. network construction and fundamental results .
_ phys rev e _ 64 : e016131(1 - 8 ) .
watts d , j , strogatz s , h ( 1998 ) collective dynamics of small - world networks .
_ nature _ 393 : 440 - 442 .
newman m , e , j , strogatz s , h , watts d , j ( 2001 ) random graphs with arbitrary degree distributions and their applications .
_ phys rev e _ 64 : e026118(1 - 17 ) .
neville j , jensen d ( 2007 ) relational dependency networks . _ jmlr _ 8 : 653 - 692 .
boldi p , rosa m , santini m , vigna s ( 2011 ) layered label propagation : a multiresolution coordinate - free ordering for compressing social networks .
_ proc of the 20th international conference on world wide web _ , ( hyderabad , india ) , pp 587 - 596 .
boldi p , rosa m , vigna s ( 2011 ) robustness of social networks : comparative results based on distance distributions .
_ soc inform _ 6984 : 8 - 21 .
albert r , barabsi a - l ( 2000 ) topology of evolving networks : local events and universality .
_ phys rev lett _
85(24 ) : 5234 - 5237 .
albert r , barabsi a - l ( 2002 ) statistical mechanics of complex networks .
_ rev mod phys _
74(1 ) : 47 - 97 .
newman m , e , j ( 2003 ) the structure and function of complex networks .
_ siam rev _ 45(2 ) : 167 - 256 .
erds p , rnyi a ( 1960 ) on the evolution of random graphs .
_ publications of the mathematical institute of the hungarian academy of sciences _ 5 : 17 - 61 .
barabsi a - l , albert r ( 1999 ) emergence of scaling in random networks .
_ science _ 286 : 509 - 512 .
farkas i , j , dernyi i , barabsi a - l , vicsek t ( 2001 ) spectra of `` real - world '' graphs : beyond the semicircle law .
_ phys rev e _ 64 : e026704(1 - 12 ) .
dorogovstev s , n , goltsev a , v , mendes j , f , f , samukhin a , n ( 2003 ) spectra of complex networks .
_ phys rev e _ 68 : e046109(1 - 10 ) .
mehta m , l ( 1991 ) _ random matrices _
( academic press , new york ) , 2nd ed , pp 2 - 667 .
brody t , a ( 1973 ) statistical measure for repulsion of energy - levels .
_ lett nuovo cimento _ 7(12 ) : 482 - 484
. jalan s , bandyopadhyay j , n ( 2009 ) randomness of random networks : a random matrix analysis .
_ europhys lett _ 87 : e48010(1 - 5 ) .
|
an understanding of how individuals shape and impact the evolution of society is vastly limited due to the unavailability of large - scale reliable datasets , that can simultaneously capture information regarding individual movements as well as social interactions .
we believe that the popular indian film industry , ` bollywood ' , can provide a social network apt for such a study .
bollywood provides massive amounts of real , unbiased data that spans over 100 years and hence this network has been used as a model for the present paper .
it is seen that the nodes which maintain a moderate degree or widely cooperate with the other nodes of the network , tend to be more fit ( measured as the success of the node in the industry ) in comparison to the other nodes .
the analysis carried forth in the current work , using a conjoined framework of complex network theory and random matrix theory , aims to quantify the elements that determine the fitness of an individual node and the factors that contribute to the robustness of a network .
the authors of this paper believe that the method of study used in the current paper can be extended to study various other industries and organizations . * uncovering randomness and success in society * + * sarika jalan@xmath0 , camellia sarkar@xmath1 , anagha madhusudanan@xmath2 , sanjiv kumar dwivedi@xmath2 + _ @xmath3 complex systems lab , physics discipline , indian institute of technology indore , m - block , iet - davv campus , khandwa road , indore 452017 , india + _ @xmath4 complex systems lab , center for biosciences and biomedical engineering , indian institute of technology indore , m - block , iet - davv campus , khandwa road , indore 452017 , india + @xmath5 e - mail : [email protected]__ *
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.